final rule occupational exposure to 1,3-butadiene ... · 56746federal register/vol. 61, no....

fede

ral r

egiste

r

56745

MondayNovember 4, 1996

Part II

Department of LaborOccupational Safety and HealthAdministration

29 CFR Part 1910, et al.Occupational Exposure to 1,3-Butadiene;Final Rule

56746 Federal Register / Vol. 61, No. 214 / Monday, November 4, 1996 / Rules and Regulations

DEPARTMENT OF LABOR

Occupational Safety and HealthAdministration

29 CFR Parts 1910, 1915 and 1926

[Docket No. H–041]

RIN 1218–AA83

Occupational Exposure to 1,3-Butadiene

AGENCY: Occupational Safety and HealthAdministration (OSHA), Department ofLabor.ACTION: Final rule.

SUMMARY: This final standard amendsthe Occupational Safety and HealthAdministration’s (OSHA) occupationalstandard that regulates employeeexposure to 1,3-Butadiene (BD). Thebasis for this action is a determinationby the Assistant Secretary, based onanimal and human data, that OSHA’scurrent permissible exposure limit (PEL)which permits employees to be exposedto BD in concentrations up to 1,000parts BD per million parts of air (1,000ppm) as an eight-hour time-weightedaverage (TWA) does not adequatelyprotect employee health. OSHA’s newlimits reduce the PEL for BD to an 8-hour TWA of 1 ppm and a short termexposure limit (STEL) of 5 ppm for 15minutes. An ‘‘action level’’ of 0.5 ppmas an 8-hour TWA is included in thestandard as a mechanism for exemptingan employer from some administrativeburdens, such as employee exposuremonitoring and medical surveillance, ininstances where the employer candemonstrate that the employee’sexposures are consistently at very lowlevels. In order to reduce exposures andprotect employees, OSHA’s BD standardincludes requirements such asengineering controls, work practices andpersonal protective equipment,measurement of employee exposures,training, medical surveillance, hazardcommunication, regulated areas,emergency procedures andrecordkeeping.DATES: The effective date of theseamendments is February 3, 1997. Start-up date for engineering controls isNovember 4, 1998, and for the exposuregoal program November 4, 1999.Affected parties do not have to complywith the information collectionrequirements in § 1910.1051(d)exposure monitoring, § 1910.1051(f)methods of compliance, § 1910.1051(g)exposure goal program, § 1910.1051(h)respiratory protection, § 1910.1051(j)emergency situations, § 1910.1051(k)medical screening and surveillance,

§ 1910.1051(l) communication of BDhazards to employees; and§ 1910.1051(m) recordkeeping until theDepartment of Labor publishes aFederal Register notice informing thepublic that OMB has approved theseinformation requirements under thePaperwork Reduction Act of 1995.

Other Dates: Written comments on thepaperwork requirements of this finalrule must be submitted on or beforeJanuary 3, 1997.ADDRESSES: In accordance with 28U.S.C. 2112(a), the Agency designatesthe following party to receive petitionsfor review of this regulation: AssociateSolicitor for Occupational Safety andHealth, Office of the Solicitor, Room S–4004, U.S. Department of Labor, 200Constitution Ave., NW., Washington,DC 20210. These petitions must be filedno later than the 59th calendar dayfollowing promulgation of thisregulation; see section 6(f) of theOccupational Safety and Health Act of1970 (OSH Act), 29 CFR 1911.18(d), andUnited Mine Workers of America v.Mine Safety and Health Administration,900 F.2d 384 (D.C. Circ. 1990).

Comments regarding the paperworkburden of this regulation, which arebeing solicited by the Agency asrequired by the Paperwork ReductionAct of 1995, are to be submitted to theDocket Office, Docket No. ICR 96–13,U.S. Department of Labor, Room N–2625, 200 Constitution Ave., NW.,Washington, DC 20210, telephone (202)219–7894. Written comments limited to10 pages or less in length may also betransmitted by facsimile to (202) 219–5046.FOR FURTHER INFORMATION CONTACT: Ms.Anne Cyr, OSHA Office of PublicAffairs, United States Department ofLabor, Room N–3641, 200 ConstitutionAvenue, NW., Washington, DC. 20210,Telephone (202) 219–8151. Copies ofthe referenced information collectionrequest are available for inspection andcopying in the Docket Office and will bemailed to persons who request copies bytelephoning Vivian Allen at (202) 219–8076. For electronic copies of the 1,3–Butadiene Information CollectionRequest, contact OSHA’s WebPage onInternet at http://www.osh.gov/.

I. Collection of Information; Request forComment

This final 1,3–Butadiene standardcontains information collectionrequirements that are subject to reviewby the Office of Management andBudget (OMB) under the PaperworkReduction Act (PRA95) 44 U.S.C. 3501et seq. (see also 5 CFR part 1320).PRA95 defines collection of information

to mean, ‘‘the obtaining, causing to beobtained, soliciting, or requiring thedisclosure to third parties or the publicof facts or opinions by or for an agencyregardless of form or format.’’ (44 U.S.C.3502(3)(A))

The title, the need for and proposeduse of the information, a summary of thecollections of information, descriptionof the respondents, and frequency ofresponse required to implement therequired information collection isdescribed below with an estimate of theannual cost and reporting burden (asrequired by 5 CFR 1320.5(a)(1)(iv) and1320.8(d)(2)). Included in the estimate isthe time for reviewing instructions,gathering and maintaining the dataneeded, and completing and reviewingthe collection of information.

OSHA invites comments on whetherthe proposed collection of information:

• Ensures that the collection ofinformation is necessary for the properperformance of the functions of theagency, including whether theinformation will have practical utility;

• Estimates the projected burdenaccurately, including whether themethodology and assumptions used arevalid;

• Enhances the quality, utility, andclarity of the information to becollected; and

• Minimizes the burden of thecollection of information on those whoare to respond, including the use ofappropriate automated, electronic,mechanical, or other technologicalcollection techniques or other forms ofinformation technology, e.g., permittingelectronic submissions of responses.

Title: 1,3–Butadiene, 29 CFR1910.1051.

Description: The final 1,3-Butadiene(BD) Standard is an occupational safetyand health standard that will minimizeoccupational exposure to BD. Thestandard’s information collectionrequirements are essential componentsthat will protect employees fromoccupational exposure. The informationwill be used by employers andemployees to implement the protectionrequired by the standard. OSHA willuse some of the information todetermine compliance with thestandard.

Summary of the Collection ofInformation: The collections ofinformation contained in the standardinclude the provisions concerningobjective data; exposure monitoringrecords and employee notification ofexposure monitoring results; writtenplans for compliance, respiratoryprotection, exposure goal, emergencysituations; information to the physician;employee medical exams and medical

56747Federal Register / Vol. 61, No. 214 / Monday, November 4, 1996 / Rules and Regulations

records; respirator fit-testing records;record of training program; employeeaccess to monitoring and medicalrecords; and transfer of records toNIOSH.

Respondents: The respondents areemployers whose employees may haveoccupational exposure to BD above theaction level. The main industriesaffected are 1,3-Butadiene PolymerProduction, Monomer purification of1,3-Butadiene, Stand-Alone ButadieneTerminals, and Crude 1,3-ButadieneProducers.

Frequency of Response: Thefrequency of monitoring andnotification of monitoring results will bedependent on the results of the initialand subsequent monitoring events andthe number of different jobclassifications with BD exposure. TheCompliance Plan is required to beestablished and updated as necessaryand reviewed at least annually. TheExposure Goal Program, RespiratoryProtection Program, and EmergencyPlans are required to be established andupdated as necessary. For those usingrespirators, respirator fit testing isrequired initially, and at least annuallythereafter. The frequency of the medicalexaminations will be dependent on thenumber of employees who will beexposed at or above the action level, orin emergency situations. A record of thetraining program is required to bemaintained. Those employers usingobjective data in lieu of monitoringmust maintain records of the objectivedata relied upon. The employer mustmaintain exposure monitoring andmedical records, which includesinformation provided to the physicianor other licensed health careprofessional, in accordance with 29 CFR1910.20. Fit-Test records must bemaintained for respirator users until thenext fit test is administered.

Total Estimated Cost: First Year$820,388; Second Year $658,949; andThird and Recurring Years $75,890.

Total Burden Hours: The total burdenhours for the first year is estimated to be8,077; for the second year, the burden isestimated to be 5,342; and for the thirdand recurring years, the burden isestimated to be 1,587. The Agency hassubmitted a copy of the informationcollection request to OMB for its reviewand approval. Interested parties arerequested to send comments regardingthis information collection to the OSHADocket Office, Docket No. ICR 96–13,U.S. Department of Labor, Room N–2625, 200 Constitution Avenue, NW,Washington, DC 20210. Writtencomments limited to 10 pages or fewermay also be transmitted by facsimile to(202) 219–5046.

Comments submitted in response tothis notice will be summarized andincluded in the request for Office ofManagement and Budget approval of thefinal information collection request;they will also become a matter of publicrecord.

Copies of the referenced informationcollection request are available forinspection and copying in the OSHADocket Office and will be mailed topersons who request copies bytelephoning Vivian Allen at (202) 219–8076. Electronic copies of the 1,3-Butadiene information collectionrequest are available on the OSHAWebPage on the Internet at http://www.osha.gov/.

FederalismThis standard has been reviewed in

accordance with Executive Order 12612,52 FR 41685 (October 30, 1987),regarding Federalism. This Orderrequires that agencies, to the extentpossible, refrain from limiting Statepolicy options, consult with States priorto taking any actions only when there isclear constitutional authority and thepresence of a problem of national scope.The Order provides for preemption ofState law only if there is a clearCongressional intent for the Agency todo so. Any such preemption is to belimited to the extent possible.

Section 18 of the Occupational Safetyand Health Act (OSH Act), expressesCongress’ clear intent to preempt Statelaws with respect to which FederalOSHA has promulgated occupationalsafety or health standards. Under theOSH Act, a State can avoid preemptiononly if it submits, and obtains Federalapproval of, a plan for the developmentof such standards and theirenforcement. Occupational safety andhealth standards developed by suchState Plan-States must, among otherthings, be at least as effective inproviding safe and healthfulemployment and places of employmentas the Federal standards. Where suchstandards are applicable to productsdistributed or used in interstatecommerce, they may not unduly burdencommerce and must be justified bycompelling local conditions. (Seesection 18(c)(2).)

The final BD standard is drafted sothat employees in every State will beprotected by general, performance-oriented standards. States withoccupational safety and health plansapproved under section 18 of the OSHAct will be able to develop their ownState standards to deal with any specialproblems which might be encounteredin a particular state. Moreover, theperformance nature of this standard, of

and by itself, allows for flexibility byStates and employers to provide asmuch leeway as possible usingalternative compliance.

This final rule of BD addresses ahealth problem related to occupationalexposure to BD which is national inscope.

Those States which have elected toparticipate under section 18 of the OSHAct would not be preempted by thisregulation and will be able to deal withspecial, local conditions within theframework provided by thisperformance-oriented standard whileensuring that their standards are at leastas effective as the Federal Standard.

State PlansThe 23 States and 2 territories with

their own OSHA-approved occupationalsafety and health plans must adopt acomparable standard within 6 months ofthe publication of this final standard foroccupational exposure to 1,3-butadieneor amend their existing standards if it isnot ‘‘at least as effective’’ as the finalFederal standard. The states andterritories with occupational safety andhealth state plants are: Alaska, Arizona,California, Connecticut (for State andlocal government employees only),Hawaii, Indiana, Iowa, Kentucky,Maryland, Michigan, Minnesota,Nevada, New Mexico, New York (forState and local government employeesonly), North Carolina, Oregon, PuertoRico, South Carolina, Tennessee, Utah,Vermont, Virginia, the Virgin Islands,Washington, and Wyoming. Until suchtime as a State standard is promulgated,Federal OSHA will provide interimenforcement assistance, as appropriate,in these states and territories.

SUPPLEMENTARY INFORMATION:

I. Table of ContentsThe preamble to the final standard on

occupational exposure to BD discussesevents leading to the final rule, physicaland chemical properties of BD,manufacture and use of BD, healtheffects of exposure, degree andsignificance of the risk presented, ananalysis of the technological andeconomic feasibility, regulatory impactand regulatory flexibility analysis, andthe rationale behind the specificprovisions set forth in the proposedstandard. The discussion follows thisoutline:I. Table of ContentsII. Pertinent Legal AuthorityIII. Events Leading to the Final StandardIV. Chemical Identification, Production, and

UseA. MonomerB. Polymers

V. Health Effects


A. IntroductionB. Carcinogenicity

1. Animal Studies2. Epidemiologic Studies

C. Reproductive EffectsD. Other Relevant Studies

VI. Quantitative Risk AssessmentVII. Significance of RiskVIII. Summary of the Final Economic

AnalysisIX. Environmental ImpactX. Summary and Explanation of the Proposed

StandardA. Scope and ApplicationB. DefinitionsC. Permissible Exposure LimitsD. Exposure MonitoringE. Regulated AreasF. Methods of ComplianceG. Exposure Goal ProgramH. Respiratory ProtectionI. Personal Protective EquipmentJ. Emergency SituationsK. Medical Screening and SurveillanceL. Hazard CommunicationM. RecordkeepingN. DatesO. Appendices

XI. Final Standard and AppendicesAppendix A: Substance Safety Data Sheet for

1,3-ButadieneAppendix B: Substance Technical Guidelines

for 1,3-ButadieneAppendix C: Medical Screening and

Surveillance for 1,3-ButadieneAppendix D: Sampling and Analytical

Method for 1,3-ButadieneAppendix E: Respirator Fit Testing

ProceduresAppendix F: Medical Questionnaires

II. Pertinent Legal AuthorityThe purpose of the Occupational

Safety and Health Act, 29 U.S.C. 651 etseq. (‘‘the Act’’) is to ‘‘assure so far aspossible every working man and womanin the nation safe and healthful workingconditions and to preserve our humanresources.’’ 29 U.S.C. 651(b). To achievethis goal, Congress authorized theSecretary of Labor to promulgate andenforce occupational safety and healthstandards. U.S.C. 655(a) (authorizingsummary adoption of existingconsensus and federal standards withintwo year of Act’s enactment), 655(b)(authorizing promulgation of standardspursuant to notice and comment),654(b) (requiring employers to complywith OSHA standards.)

A safety or health standard is astandard ‘‘which requires conditions, orthe adoption or use of one or morepractices, means, methods, operations,or processes, reasonably necessary orappropriate to provide safe or healthfulemployment or places of employment.’’29 U.S.C. 652(8).

A standard is reasonably necessary orappropriate within the meaning ofSection 652(8) if it substantially reducesor eliminates significant risk, and iseconomically feasible, technologically

feasible, cost effective, consistent withprior Agency action or supported by areasoned justification for departing fromprior Agency actions, supported bysubstantial evidence, and is better ableto effectuate the Act’s purposes than anynational consensus standard itsupersedes. See 58 FR 16612–16616(March 30, 1993).

The Supreme Court has noted that areasonable person would consider afatality risk of 1/1000 over a 45-yearworking lifetime to be a significant risk.Industrial Union Dep’t v. AmericanPetroleum Institute, 448 U.S. 607, 646(1980) (benzene standard). OSHA agreesthat a fatality risk of 1/1000 over aworking lifetime is well within the rangeof risk that reasonable people wouldconsider significant. See e.g.,International Union, UAW v.Pendergrass, 878 F.2d 389 (D.C. Cir.1989) (formaldehyde standard); Buildingand Constr. Trades Dep’t, AFL–CIO v.Brock, 838 F.2d 1258, 1265 (D.C. Cir.1988) (asbestos standard).

A standard is technologically feasibleif the protective measures it requiresalready exist, can be brought intoexistence with available technology, orcan be created with technology that canreasonably be expected to be developed.American Textile Mfrs. Institute v.OSHA, 452 U.S. 490, 513 (1981)(‘‘ATMI’’), American Iron and SteelInstitute v. OSHA, 939 F.2d 975, 980(D.C. cir. 1991) (‘‘AISI’’).

A standard is economically feasible ifindustry can absorb or pass on the costof compliance without threatening itslong term profitability or competitivestructure. See ATMI, 452 U.S. at 530 n.55; AISI, 939 F. 2d at 980.

A standard is cost effective if theprotective measures it requires are theleast costly of the available alternativesthat achieve the same level ofprotection. ATMI, 453 U.S. at 514 n. 32;International Union, UAW v. OSHA, 37F. 3d 665, 668 (D.C. Cir. 1994) (‘‘LOTOIII’’).

All standards must be highlyprotective. See 58 FR 16614–16615;LOTO III, 37 F. 3d at 668. However,health standards must also meet the‘‘feasibility mandate’’ of Section 6(b)(5)of the Act, 29 U.S.C. 655(b)(5). Section6(b)(5) requires OSHA to select ‘‘themost protective standard consistentwith feasibility’’ that is needed toreduce significant risk when regulatinghealth hazards. ATMI, 452 U.S. at 509.

Section 6(b)(5) also directs OSHA tobase health standards on ‘‘the bestavailable evidence,’’ including research,demonstrations, and experiments. 29U.S.C. 655(b)(5). OSHA shall consider‘‘in addition to the attainment of thehighest degree of health and safety

protection * * * the latest scientificdata * * * feasibility and experiencegained under this and other health andsafety laws.’’ Id.

Section 6(b)(7) of the Act authorizesOSHA to include among a standard’srequirements labeling, monitoring,medical testing and other informationgathering and transmittal provisions. 29U.S.C. 655(b)(7).

Finally, whenever practical, standardsshall ‘‘be expressed in terms of objectivecriteria and of the performancedesired.’’ Id.

III. Events Leading to the FinalStandard

The standard adopted for BD byOSHA in 1971 pursuant to Section 6(a)of the OSH Act, 29 U.S.C. 655 from anexisting Walsh-Healey Federal Standardrequired employers to assure thatemployee exposure does not exceed1,000 ppm determined as an 8-hourTWA (29 CFR 1910.1000, Table Z–1).The source of the Walsh-HealeyStandard was the Threshold Limit Value(TLV) for BD developed in 1968 by theAmerican Conference of GovernmentalIndustrial Hygienists (ACGIH). ThisTLV was adopted by the ACGIH toprevent irritation and narcosis.

In 1983, the National ToxicologyProgram (NTP) released the results of ananimal study indicating that BD causescancer in rodents. (Ex. 20) Based on thestrength of the results of this animalstudy, ACGIH in 1983 classified BD asan animal carcinogen and in 1984recommended a new TLV of 10 ppm.(Ex. 2–4) Based on the same evidence,on February 9, 1984, the NationalInstitute for Occupational Safety andHealth (NIOSH) published a CurrentIntelligence Bulletin (CIB)recommending that BD be regarded as apotential occupational carcinogen,teratogen and a possible reproductivehazard. (Ex. 23–17) On January 5, 1984,OSHA published a Request forInformation (RFI) jointly with theEnvironmental Protection Agency.(EPA) (49 FR 844) EPA also announcedthe initiation of a 180 day review underthe authority of section 4(f) of the ToxicSubstance Control Act (TSCA) (49 FR845) to determine ‘‘whether to initiateappropriate action to prevent or reducethe risk from the chemical or to find thatthe risk is not unreasonable.’’ Commentswere to be submitted to OSHA by March5, 1984. On April 4, 1984, OSHAextended the comment period untilfurther notice. (49 FR 13389)

Petitions for an Emergency TemporaryStandard (ETS) of 1 ppm or less forworkers’ exposure to BD were submittedto OSHA on January 23, 1984, by theUnited Rubber, Cork, Linoleum and


Plastic Workers of America (URW), theOil, Chemical and Atomic Workers(OCAW), the International ChemicalWorkers Union (ICWU), and theAmerican Federation of Labor andCongress of Industrial Organizations(AFL–CIO). (Ex. 6–4) On March 7, 1984,OSHA denied the petitions on theground that the Agency was stillevaluating the health data to determinewhether regulatory action wasappropriate.

Based on its 180-day review of BD,EPA published, on May 15, 1984, anAdvance Notice of ProposedRulemaking (ANPR) (49 FR 20524) toannounce the initiation of a regulatoryaction by the EPA to determine andimplement the most effective means ofcontrolling exposures to the chemicalBD under the TSCA. EPA was workingwith OSHA because available evidenceindicated that exposure to BD occursprimarily within the workplace.

Information received in response tothis ANPR was used by EPA to developrisk assessments. Subsequently, EPAidentified BD as a probable humancarcinogen (Group B2) according toEPA’s classification of carcinogens, andconcluded that current exposuresduring the manufacturing of BD and itsprocessing into polymers presented anunreasonable risk of injury to humanhealth. (Ex. 17–4) Additionally, EPAdetermined that the risks associatedwith exposure to BD may be reduced toa sufficient extent by action taken underthe OSH Act. Following these findings,EPA, in accordance with section 9(a) ofTSCA, on October 10, 1985 (50 FR41393), referred BD to OSHA to give thisAgency an opportunity to regulate thechemical under the OSH Act. EPArequested that OSHA determinewhether the risks described in the EPAreport may be prevented or reduced toa sufficient extent by action taken underthe OSH Act and then if such adetermination is made, OSHA issue anorder declaring whether themanufacture and use of BD described inthe EPA report present the risk thereindescribed. EPA asked OSHA to respondwithin 180 days, by April 8, 1986. (50FR 41393)

On December 27, 1985, OSHApublished a notice soliciting publiccomments on EPA’s referral report. (50FR 52952) Based on all the availableinformation, OSHA, on April 11, 1986,responded to the EPA referral report bymaking a preliminary determination (50FR 12526) that a revised OSHA standardlimiting occupational exposure to BDcould prevent or reduce the risk ofexposure to a sufficient extent and thatsuch risks had been accuratelydescribed by EPA in the report. On

October 1, 1986, OSHA published anANPR (51 FR 35003) to initiate arulemaking within the meaning ofsection 9(a) of TSCA. The Agencyrequested that comments be submittedby December 30, 1986. Twenty-fourcomments, some of them containingnew information, were received inresponse to the ANPR. (Ex. 28–1 to 28–24) Six additional comments werereceived after the deadline. (Ex. 29–1 to29–6)

OSHA reviewed the available dataand conducted risk assessment,regulatory impact and flexibilityanalyses. These analyses demonstratethat the proposed standard wastechnologically and economicallyfeasible and substantially reduced thesignificant risk of cancers and otheradverse health effects.

On August 10, 1990, OSHA publishedits proposed rule to regulateoccupational exposure to 1,3-butadiene.(55 FR 32736) Based on the Agency’sreview of studies of exposed animalsand epidemiologic studies and takinginto account technologic and economicfeasibility considerations, OSHAproposed a permissible exposure limit(PEL) of 2 ppm as an 8-hour time-weighted average and a short termexposure limit (STEL) of 10 ppm for a15 minute sampling period. Alsoincluded in the proposal was an ‘‘actionlevel’’ of 1 ppm which triggered certainprovisions of the standard such asmedical surveillance and training.

OSHA convened public hearings inWashington, DC., on January 15–23,1991, and in New Orleans, Louisiana,on February 20–21, 1991. The post-hearing period for the submission ofbriefs, arguments and summations wasto end July 22, 1991, but was extendedby the Administrative Law Judge toDecember 13, 1991, in order to giveparticipants time to review new data onlow-dose exposures submitted by NTPand a quantitative risk assessment doneby NIOSH. The comment period closedFebruary 10, 1992.

In the Fall of 1992, the InternationalAgency for Research on Cancer (IARC)published the results of the WorkingGroup on the Evaluation ofCarcinogenic Risks to Humans, whichreviewed the carcinogenic potential ofBD and concluded that:

There is limited evidence for thecarcinogenicity in humans of 1,3-butadiene* * * There is sufficient evidence for thecarcinogenicity in experimental animals* * * (Ex. 125)

IARC stated that its overall evaluationled it to conclude that ‘‘1,3-butadiene isprobably carcinogenic to humans(Group 2A).’’ (Ex. 125)

To assist OSHA in issuing a final rulefor BD, representatives of the majorunions and industry groups involved inthe production and use of BD submittedthe outline of a voluntary agreementreached by the parties dated January 29,1996, outlining provisions that theyagreed upon and recommended beincluded in the final rule. The lettertransmitting the agreement was signedby J.L. McGraw for the InternationalInstitute of Synthetic Rubber Producers(IISRP), Michael J. Wright for the UnitedSteelworkers of America (USWA), andMichael Sprinker (CWU). Thecommittee that worked on the issuesalso included Joseph Holtshouser of theGoodyear Tire and Rubber Company,Carolyn Phillips of the Shell ChemicalCompany, representing the ChemicalManufacturers Association, RobertRichmond of the Firestone SyntheticRubber and Latex Company, and LouisBeliczky (formerly of the URW) andJames L. Frederick of the SWA.

The agreement proposed a change inthe permissible exposure limits,additional provisions for exposuremonitoring, and an exposure goalprogram designed to reduce exposuresbelow the action level. It also set forthother modifications to the scope,respiratory protection, communicationof hazards, medical surveillance, andstart-up dates sections of the final rule.

On March 8, 1996 OSHA publishedthe labor/industry jointrecommendations and re-opened therecord for 30 days to allow the publicto comment. (61 FR 9381) In responseto requests from the parties to theagreement, the comment period wasextended to April 26, 1996. (61 FR15205)

At the beginning of the commentperiod, OSHA placed in the rulemakingrecord an epidemiologic study of BDexposed workers by Delzell, et al.sponsored by IISRP, along with IARCvolume 127 ‘‘Butadiene and StyreneAssessment of Health Hazards,’’ apublished paper by Santos-Burgoa, et al.entitled ‘‘Lymphohematopoietic Cancerin Styrene-Butadiene PolymerizationWorkers,’’ and abstracts from asymposium entitled ‘‘Evaluation ofButadiene and Isoprene Health Risks.’’(Ex. 117–1; 117–2; 117–3; 117–4) Theepidemiological study had also beensubmitted to the EPA in compliancewith provisions of the Toxic SubstancesControl Act.

In response to the re-opening of theBD record, 18 sets of comments werereceived. The parties to the labor/industry agreement submitted a draftregulatory text which put theirrecommendations into specificrequirements. The outline and the


subsequent draft regulatory text aresolely the work product of thenegotiating committee. OSHA wasneither a party to nor present at thenegotiations.

While the responses to the record re-opening helped clarify the intent of thenegotiating parties, the rationalesbehind several of the changes were notfully explained.

On September 16, 1996, Judge JohnM. Vittone, for Judge George C. Piercewho presided over the BD hearings,closed the record of the public hearingon the proposed standard for 1,3-butadiene and certified it to theAssistant Secretary of Labor. (Ex. 135)

IV. Chemical Identification, Productionand Use

A. Monomer

The chemical 1,3-butadiene (BD)(Chemical Abstracts Registry Number106–99–0) is a colorless, noncorrosive,flammable gas with a mild aromaticodor at standard ambient temperatureand pressure. It has a chemical formulaof C4H6, a molecular weight of 54.1, anda boiling point of ¥4.7 °C at 760 mmHg, a lower explosive limit of 2%, andan upper explosive limit of 11.5%. Itsvapor density is almost twice that of air.It is slightly soluble in water, somewhatsoluble in methanol and ethanol, andreadily soluble in less polar organicsolvents such as hexane, benzene, andtoluene. (Ex. 17–17) It is highly reactive,dimerizes to 4-vinylcyclohexene, andpolymerizes easily. Because of its lowodor threshold, high flammability andexplosiveness, BD has been handledwith extreme care in the industry.

In the United States BD has beenproduced commercially by threeprocesses: Catalytic dehydrogenation ofn-butane and n-butene, oxidativedehydrogenation of n-butene, andrecovery as a by-product from the C4 co-product stream from the steam crackingprocess used to manufacture ethylene,which is the major product of thepetrochemical industry. For economicreasons, almost all BD currently made inthe U.S. is produced by the ethylene co-product process.

In the steam cracking process forethylene, a hydrocarbon feedstock isdiluted with steam then heated rapidlyto a high temperature by passing itthrough tubes in a furnace. The outputstream, containing a broad mixture ofhydrocarbons from the pyrolysisreactions in the cracking tubes plusunreacted components of feedstock, iscooled and then processed through aseries of distillation and otherseparation operations in which thevarious products of the cracking

operation are separated for disposal,recycling or recovery.

The cracking process producesbetween 0.02 to 0.3 pounds of BD perpound of ethylene, depending upon thecomposition of the feedstock. BD isrecovered from the C4 stream by theseparation operations. The C4 streamcontains from 30 to 50% BD plusbutane, butenes and small fractions ofother hydrocarbons. This crude BDstream from the ethylene unit may berefined in a unit on site, or transferredto another location, a monomer plant,owned by the same or a differentcompany, to produce purified BD.

Regardless of the source of the crudeBD-ethylene co-product,(dehydrogenation, or blending of C4

streams from other sources), theprocesses used by different companiesto refine BD for subsequent use inpolymer production are similar.Extractive distillation is used to effectthe basic separation of BD from butanesand butenes and fractional distillationoperations are used to accomplish otherrelated separations. A typical monomerplant process is described below.

C3 and C4 acetylene derivatives,present in the C4 co-product stream, areconverted to olefins by passing thestream through a hydrogenation reactor.The stream is then fed to an extractivedistillation column to separate the BDfrom butanes and butenes. Severaldifferent solvents have been employedfor this operation, including n-methylpyrrolidone, dimethylformamide,furfural, acetonitrile,dimethylacetamide, and cuprousammonium acetate solution. The BD,extracted by the solvent, is strippedfrom it in the solvent recovery column,then fed to another fractionationcolumn, the methylacetylene column, tohave residual acetylene stripped out.The bottom stream from themethylacetylene column, containing theBD, is fed to the BD rerun column, fromwhich the purified BD product is takenoff overhead. The solvent, recovered inthe solvent recovery column, is recycledto the extractive distillation columnwith part of it distilled to keep down thelevel of polymer. (Ex. 17–17)

A stabilizer is added to the monomerto inhibit formation of polymer duringstorage. It is stored as a liquid underpressure, sometimes refrigerated toreduce the pressure, generally stored ina tank farm in diked spheres. It isshipped to polymer manufacturers andother users by pipeline, barge, tank car,or tank truck.

BD is a major commodity product ofthe petrochemical industry. Total U.S.production of BD in 1991 was 3.0billion pounds. Although BD is a toxic

flammable gas, its simple chemicalstructure with low molecular weightand high chemical reactivity make it auseful building block for synthesizingother products. In ‘‘1,3-Butadiene Useand Substitutes Analysis,’’ EPAidentified 140 major, minor andpotential uses of BD in the chemicalindustry. (Ex. 17–15)

Over 60% of the BD consumed in theUnited States is used in the manufactureof rubber, about 12% in makingadiponitrile which in turn is used tomake hexamethylenediamine (HMDA),a component of Nylon, approximately8% in making styrene-butadienecopolymer latexes, approximately 7% inproducing polychloroprene, and about6% in producing acrylonitrile-butadiene-styrene (ABS) resins. Lesseramounts are consumed in theproduction of rocket propellants,specialty copolymer resins and latexesfor paint, coatings and adhesiveapplications, and hydrogenatedbutadiene-styrene polymers used aslubricating oil additives. Somenonpolymer applications include themanufacture of the agriculturalfungicides, Captan and Captofol, theindustrial solvent sulfolane, andanthroquinone dyes.

B. PolymersBD based synthetic elastomers are

manufactured by polymerizing BD byitself, by polymerizing BD with othermonomers to produce copolymers, andby producing mixtures of thesepolymers. The largest-volume product isthe copolymer of styrene and BD,styrene-butadiene rubber, followed involume by polybutadiene,polychloroprene, and nitrile rubber.Polybutadiene is the polymer of BDmonomer by itself. Polychloroprene ismade by polymerizing chloroprene,produced by chlorination of BD. Nitrilerubbers are copolymers of acrylonitrileand BD.

Four general types of processes areused in polymerizing BD and itscopolymers: emulsion, suspension,solution and bulk polymerization. Inemulsion and suspensionpolymerization, the monomers and themany chemicals used to control thereaction are finely dispersed ordissolved in water. In solutionpolymerization, the monomers aredissolved in an organic solvent such ashexane, pentane, toluene. In bulkpolymerization, the monomer itselfserves as solvent for the polymer. Thepolymer product, from which end-useproducts are manufactured, is producedin the form of polymer crumb (solidparticles), latex (a milky suspension inwater), or cement (a solution).


Emulsion polymerization is theprincipal process used to makesynthetic rubber. A process for themanufacture of styrene-butadiene crumbis typical of emulsion processes. Styreneand BD are piped to the process areafrom the storage area. The BD is passedthrough a caustic soda scrubber toremove the inhibitors which were addedto prevent premature polymerization.The fresh BD monomer streams aremixed with styrene, aqueousemulsifying agents, activator, catalyst,and modifier, and then fed to the firstof a train of reactors. The reactionproceeds stepwise in the series ofreactors to around 60% conversion ofmonomer to polymer. In the coldprocess, the reactants are chilled andthe reactor temperature is maintained at4°C to 7°C (40°F to 45°F) and pressureat 0 to 15 psig; in the hot rubber process,temperature and pressure are around50°C (122°F) and 40 to 60 psig,respectively.

The latex from the reactor train isflashed to evaporate unreacted BDwhich is compressed, condensed andrecycled. Uncondensed vapors areabsorbed in a kerosene absorber beforeventing and the absorbed BD is steamstripped or recovered from the keroseneby some other operation. The latexstream is passed through a steamstripper, operated under vacuum, toremove and recover unreacted styrene.The styrene and water in the condensateare separated by decanting. The styrenephase is recycled to the process.Noncondensibles from the strippingcolumn contain some BD and aredirected through the BD recoveryoperations.

Stripped latex, to which anantioxidant has been added, is pumpedto coagulation vessels where dilutesulfuric acid and sodium chloridesolution are added. The acid and brinemixture breaks the emulsion, releasingthe polymer in the form of crumb.Sometimes carbon black and oil areadded during the coagulation step sincebetter dispersion is obtained than bymixing later on.

The crumb and water slurry from thecoagulation operation is screened toseparate the crumb. The wet crumb ispressed in rotary presses to squeeze outmost of the entrained water then driedwith hot air on continuous dry beltdryers. The dried product is baled andweighed for shipment.

Production of styrene-butadiene latexby the emulsion polymerization processis similar to that for crumb but isusually carried out on a smaller scalewith fewer reactors. For some but not allproducts, the reaction is run to nearcompletion, monomer removal is

simpler and recovery may not bepracticed.

Polybutadiene rubber is usuallyproduced by solution polymerization.Inhibitor is removed from the monomerby caustic scrubbing. Both monomerand solvent are dried by fractionaldistillation, mixed in the desired ratioand dried in a desiccant column.Polymerization is conducted in a seriesof reactors using initiators and catalystsand is terminated with a shortstopsolution. The solution, called rubbercement, is pumped to storage tanks forblending. Crumb is precipitated bypumping the solution into hot waterunder violent agitation. Solvent andmonomer are recovered by stripping anddistillation similar to those previouslydescribed. The crumb is screened,dewatered, dried and baled.

Polychloroprene (neoprene)elastomers are manufactured bypolymerizing chloroprene in anemulsion polymerization processsimilar to that used for making styrene-butadiene rubber. The monomer,chloroprene (2-chloro-BD), is made bychlorination of BD to make 3,4-dichlorobutene, anddehydrochlorination of the latter.

Nitrile rubbers, copolymers ofacrylonitrile and BD, are produced byemulsion polymerization similar to thatused to make styrene-butadiene rubber.

Substantial amounts of BD are used inthe production of two other largevolume polymers: Nylon resins andABS resin. Dupont manufacturesadiponitrile from BD and uses theproduct to make hexamethylenediaminewhich is polymerized in making Nylonresins and fibers, including Nylon 6,6.Acrylonitrile, BD and styrene are themonomers used to make ABS resinwhich is a major thermoplastic resin.Chemically complex emulsion,suspension and bulk polymerizationprocesses are used by differentproducers to make ABS polymer.

V. Health Effects

A. Introduction

The toxicity of BD was longconsidered to be low and non-cumulative. Thus, the OSHA standardfor BD was 1,000 ppm on the basis ofits irritation of mucous membranes andnarcosis at high levels of exposure.However, in the 1980s, carcinogenicitystudies indicated BD is clearly acarcinogen in rodents. In 1986, theAmerican Conference of GovernmentalIndustrial Hygienists (ACGIH) wasprompted by these studies to lower theworkplace threshold limit value (TLV)from 1,000 to 10 ppm. (Ex. 2–5)

Rodent studies are now conclusivethat BD is an animal carcinogen.Further, a consistent body ofepidemiologic studies have also shownincreased mortality from hematopoieticcancers associated with BD exposureamong BD-exposed production andstyrene/BD rubber polymer workers.Complementary studies of metabolicproducts and genotoxicity support thesecancer findings. OSHA was alsoconcerned about evidence that BDaffects the germ cell as well as thesomatic cell, and what potentialreproductive toxicity might result fromexposure to BD. Since BD itself does notappear to be carcinogenic, but must bemetabolized to an active form, OSHAalso reviewed studies on the metabolismof BD to determine wether they mighthelp explain the observed differences incancer incidence among species.

The following sections discuss theeffects of BD exposure, both in humanand animal systems.

B. Carcinogenicity

1. Animal StudiesIn the proposed BD rule, OSHA

discussed the results of two lifetimeanimal bioassays, one on the Sprague-Dawley rat and one in the B6C3F1

mouse. (55 FR 32736 at 32740) Bothstudies found evidence of BDcarcinogenicity, with the greaterresponse in the mouse. The rat studyinvolved exposure levels of 0, 1000, or8000 ppm BD, starting at five weeks ofage, to groups of 100 male and 100female Sprague-Dawley rats for 6 hoursper day, five days per week, for 105weeks. Mortality was increased overcontrols in the 1,000 ppm exposedfemale rats and in both of the male ratexposure groups. Significant tumorresponse sites in the male rats includedexocrine adenomas and carcinomas(combined) of the pancreas in thehighest exposure group (3, 1, and 11tumors in the 0, 1000, and 8000 ppmgroups, respectively); and Leydig-celltumors of the testis (0, 3, and 8 in thesame groups, respectively). In thefemale rats, the significantly increasedtumor response also occurred in thehighest exposure group; cancers seenincluded follicular-cell adenomas andcarcinomas (combined) of the thyroidgland (0,4, and 11 tumors in the threeexposure groups, respectively), andbenign and malignant (combined)mammary gland tumors (50, 79, and 81in the same exposure groups). To alesser degree there were also sarcomasof the uterus (1, 4, 5 tumors in the threeexposure groups), and Zymbal gland (0,0, 4 tumors in the same exposuregroups, respectively). While only high


exposure group tumor response forsome of these sites was statisticallysignificant, trend tests were alsosignificant.

In contrast to the generally less than10% increase in tumor response seen inthe Sprague-Dawley rat at levels farabove BD metabolic saturation, thecarcinogenic response to BD in theB6C3F1 mouse in the NationalToxicology Program study (NTP I) wasextensive. (Ex. 23–1) In this study,groups of 50 male and 50 female micewere exposed via inhalation to 0, 625 or1250 ppm BD for 6 hours per day, 5days per week in a study originallydesigned to last 2 years. However, thehigh carcinogenic response includedmultiple primary cancers, with shortlatent periods, and led to early studytermination (60–61 weeks) due to highcancer mortality in both the 625 ppmand 1250 ppm exposure groups of bothsexes. This mortality was due mainly tolymphocytic lymphomas andhemangiosarcomas of the heart, both ofwhich were typically early occurringand quickly fatal. This large and rapidlyfatal carcinogenic response led to boththe NTP and industry to undertakeadditional studies to better understandthe mechanisms involved.

Some commenters have associatedqualitative or quantitative differences inmouse and rat BD carcinogenicity withthe differences in rat and mouse BDmetabolism. Many studies publishedand submitted to the BD record sincethe proposed rule have sought to bettercharacterize the metabolic,distributional, and eliminationprocesses involved, and some haveattributed species differences (at least inpart) to the metabolic differences. Thesewill be addressed separately in themetabolism section.

Another factor hypothesized toaccount for differences between mouseand rat BD carcinogenicity was the roleof activation of ecotropic retrovirus inhematopoietic tissues on tumorresponse in the B6C3F1 mouse. Thisvirus is endogenous to the B6C3F1

mouse and was hypothesized topotentiate the BD lymphoma responsein this strain. To study this hypothesisIrons and co-workers exposed both (60)B6C3F1 male (those with theendogenous virus) and (60) NIH Swissmale (those without the endogenousvirus) mice to either 0 or 1250 ppm BD,for 6 hours./day, 5 days per week for 52weeks. (Ex. 32–28D) A third group of 50B6C3F1 male mice received 1250 ppmfor 12 weeks only and was observeduntil study termination at 52 weeks. Theresults of the study showed significantlyincreased thymic lymphomas in allexposed groups but significantly greater

response in the B6C3F1 mouse—1tumor/60 (2%) in the control (zeroexposure) group, 10/48 (21%) in the 12week exposure group, and 34/60 (57%)in the 52 week exposure group—vs. theNIH Swiss mice, which developed 0tumors/60 in the control group, and 8tumors/57 (14%) in the BD exposedgroup. Hemangiosarcomas of the heartwere also observed in both strainsexposed to BD for 52 weeks—5/60 (8%)in the B6C3F1 mice vs. 1/57 in the NIHSwiss mice. (Ex. 32–28D). The B6C3F1

response was very similar to the NTP Ihigh exposure group response, verifyingthat earlier study. The qualitativelysimilar lymphoma responses of the twostrains also confirmed that the mousehematopoietic system is highlysusceptible to the carcinogenic effects ofBD, although quantitatively the strainsmay differ. The 21% 1-year lymphomaresponse in the 12-week stop-exposureB6C3F1 group also increased concernsabout high concentration, short durationexposures.

NTP II StudyConcurrent with the industry studies,

the NTP, in order to better characterizethe dose-response and lifetimeexperience, conducted a second, muchlarger research effort over a muchbroader dose range. (Ex. 90; 96) Thesetoxicology and carcinogenesis studiesincluded a 100-fold lower (6.25 ppm)low exposure group than NTP I, severalintermediate exposure groups, a studyof dose-rate effects using several high-concentration partial lifetime (stop-)exposure groups, and planned interimsacrifice groups. Other parts of the studyincluded clinical pathology studies(with the 9- and 15-month interimsacrifices, metabolism studies, andexamination of tumor bearing animalsfor activated oncogenes).

For the lifetime carcinogenesisstudies, groups of 70 B6C3F1 mice ofeach sex were exposed via inhalation toBD at levels of 0, 6.25, 20, 62.5, 200, or625 ppm (90 of each sex in this highestgroup) for 6 hours per day, 5 days perweek for up to 2 years. Up to 10randomly selected animals in eachgroup were sacrificed after 9 and 15months of exposure, and these animalswere assessed for both carcinogenicityand hematologic effects.

For the stop-exposure study, differentgroups of 50 male mice were exposed 6hours per day, 5 days per week toconcentrations of either 200 ppm for 40weeks, 625 ppm for 13 weeks, 312 ppmfor 52 weeks, or 625 ppm for 26 weeks.Following the BD exposure period, theexposed animals were then observed forthe remainder of the 2-year study. Thefirst two stop-exposure groups received

a total exposure (concentration timesduration) of 8,000 ppm-weeks, while thelatter two groups receivedapproximately 16,000 ppm-weeks ofexposure. For the analysis discussedbelow, groups are compared both witheach other for dose-rate effects and withthe lifetime (2 year) exposure groups forrecovery effects.

Methodology

Male mice were 6–8 weeks old andfemale mice were 7–8 weeks old whenthe exposures began. Animals wereexposed in individual wire mesh cageunits in stainless steel Hazelton 2000chambers (2.3 m3). The exposure phaseextended from January, 1986 to January,1988. Animals were housedindividually; water was available adlibitum; NIH–07 diet feed was alsoavailable ad libitum except duringexposure periods. Animals wereobserved twice daily for moribundityand mortality; animals were weighedweekly for the first 13 weeks andmonthly thereafter. Hematologyincluded red blood cell count (RBC),and white blood cell count (WBC). Thestudy was conducted in compliancewith the Food and Drug Administration(FDA) Good Laboratory PracticeRegulations with retrospective qualityassurance audits.

The results of the study are presentedbelow for the two-year and stop-exposure study. Between study groupcomparisons are made where it isdeemed appropriate. Emphasis is placedon the neoplastic effects.

Results

Two-Year Study

While body weight gains in bothexposed male and female mice weresimilar to those of the control groups,exposure related malignant neoplasmswere responsible for decreased survivalin all exposure groups of both sexesexposed to concentrations of 20 ppm orabove. Excluding the interim sacrificedanimals, the two-year survivaldecreased uniformly with increasingexposure for females (37/50, 33/50, 24/50, 11/50, 0/50, 0/70), and nearlyuniformly for males (35/50, 39/50, 24/50, 22/50, 4/50, 0/70). As with theearlier NTP study, all animals in the 625ppm group were dead by week 65,mostly as a result of lymphomas orhemangiosarcomas of the heart. The 200ppm exposure groups of both sexes alsohad much higher mortality, butsignificantly less than that of the 625ppm group. The survival of the lowestexposure group (6.25 ppm) was slightlybetter than controls for the male mice,slightly less for the female mice. Mean


survival for the males was an exposure-related 597, 611, 575, 558, 502, and 280days; for the females it was similarly608, 597, 573, 548, 441, and 320 days.This decreased survival with increasingexposure was almost totally due totumor lethality.

CarcinogenicityNine different sites showed primary

tumor types associated with butadieneexposures, seven in the male mice andeight in the female mice. These werelymphoma, hemangiosarcoma of theheart, combined alveolar-bronchiolaradenoma and carcinoma, combinedforestomach papilloma and carcinoma,Harderian gland adenoma andadenocarcinoma, preputial glandadenoma and carcinoma (males only),hepatocellular adenoma and carcinoma,and mammary and ovarian tumors

(females only). These are shown inTable V–1 adapted from Melnick et al.(Ex. 125) From this table it is seen thatsix of these tumor sites are statisticallysignificantly increased in the highestexposed males and five werestatistically significantly increased inthe highest exposed females. Twoadditional sites which showedsignificant increases at lower exposuresshowed decline at the highest exposuresbecause other tumors were more rapidlyfatal. At 200 ppm preputial glandadenoma and carcinoma combined weresignificantly increased in males (p<.05;0/70 (0%) control vs. 5/70 (7%) in the200 ppm group) and hepatocellularadenoma and carcinoma were increasedfor both exposed males and females. Atthe lowest exposure concentration, 6.25ppm, only female mouse lung tumors

(combined adenoma and carcinoma)showed statistical significance (p<.05;4/70 (6%) in controls vs. 15/70 (21%) inthe 6.25 ppm group); these tumors infemale mice showed a monotonicincrease with increasing exposure up to200 ppm. At 20 ppm female mouselymphomas and liver tumors alsoreached statistical significance(lymphomas, p<.05; 10/70 (15%) incontrols vs. 18/70 (26%) in the 20 ppmgroup; liver tumors, p<.05; 17/70 (24%)in controls vs. 23/70 (33%) in the 20ppm group), and at 62.5 ppm, tumors atseveral other sites were alsosignificantly increased. In general, whilethere were some differences in amountof tumor response between the male andfemale mice, there is fairly consistentpattern of tumor type in mice of bothsexes for the six non-sexual organ sites.

TABLE V–1.—TUMOR INCIDENCES (I) AND PERCENTAGE MORTALITY-ADJUSTED TUMOR RATES (R) IN MICE EXPOSED TO1,3-BUTADIENE FOR UP TO 2 YEARS.

[Adapted from Ex. 125]

Tumor Sex

Exposure concentration (ppm)

0 6.25 20 62.5 200 625

I Rc I R I R I R I R l R

Lymphoma ............................................................... M 4/70 8 3/70 6 8/70 19 11/70 a25 9/70 a27 69/90 a97F 10/70 20 14/70 30 a18/

7041 10/70 26 19/70 a58 43/90 a89

Heart—Hemangiosarcoma ...................................... M 0/70 0 0/70 0 1/70 2 5/70 a13 20/70 a57 6/90 a53F 0/70 0 0/70 0 0/70 0 1/70 3 20/70 a64 26/90 84

Lung—Alveolar-bronchiolar adenoma and car-cinoma.

M 22/70 46 23/70 48 20/70 45 33/70 a72 42/70 a87 12/90 a73

Forestomach—Papilloma and carcinoma ................ F 4/70 8 15/70 a32 19/70 a44 27/70 a61 32/70 a81 25/90 a83Harderian gland—Adenoma and adenocarcinoma M 1/70 2 0/70 0 1/70 2 5/70 13 12/70 a36 13/90 a75

F 2/70 4 2/70 4 3/70 8 4/70 12 7/70 a31 28/90 a85Preputial gland—Adenoma and carcinoma ............. M 6/70 13 7/70 15 11/70 25 24/70 a53 33/70 a77 7/90 a58

F 9/70 18 10/70 21 7/70 17 16/70 a40 22/70 a67 7/90 48Liver—Hepatocellular adenoma and carcinoma ..... M 0/70 0 0/70 0 0/70 0 0/70 0 5/70 a17 0/90 0Mammary gland—Adenocarcinoma ........................ M 31/70 55 27/70 54 35/70 68 32/70 69 40/70 a87 12/90 75Ovary—Benign and malignant granulosa-cell tu-

mors.F 17/70 35 20/70 41 23/70 a52 24/70 a60 20/70 a68 3/90 28

F 0/70 0 2/70 4 2/70 5 6/70 a16 13/70 a47 13/90 a66F 1/70 2 0/70 0 0/70 0 9/70 a24 11/70 a44 6/90 44

a Increased compared with chamber controls (0 ppm), p < 0.05, based on logistic regression analysis.b The Working Group noted that the incidence in control males and females was in the range of that in historical controls (Haseman et al.,

1985).c Mortality adjusted tumor rates are adjusted for competing causes of mortality, such as death due to other tumors, whose rates differ by expo-

sure group.

Hemangiosarcoma of the heart, withmetastases to other organs was firstobserved at 20 ppm in 1 male (thehistorical controls for this strain are 1/2373 in males and 1/2443 in females),in 5 males and 1 female at 62.5 ppm andin 20 males and 20 females at 200 ppm;at 625 ppm these tumor rates leveled offas other tumors, especially lymphomasbecame dominant. Lymphaticlymphomas increased to statisticalsignificance first in females at 20 ppmand were usually rapidly fatal, the firsttumor appearing at week 23, most likely

preempting some of the later appearingtumors in the higher exposure groups.Because of the plethora of primarytumors and the different time patternsobserved to onset of each type, severaltumor dose-response trends do notappear as strong as they wouldotherwise be.

Non-Neoplastic Effects

Several non-cancer toxic effects werenoted in the exposed groups, reflectingmany of the same target sites for which

the neoplastic effects were seen. (Ex. 90;96; 125).

Although the reported numbers differslightly in the different exhibits,generally dose-related increases inhyperplasia were observed in the heart,lung, forestomach, and Harderian gland,both in the two-year study (both sexes)and in the stop-exposure study(conducted in males only). In addition,testicular atrophy was observed in boththe two-year and stop-exposure malemice, but remained in the 6%–10%range except for the 2-year, 625 ppm


group where it was 74%. Ovariangerminal hyperplasia (2/49 (control), 3/49 (6.25 ppm), 8/48 (20 ppm), 15/50(62.5 ppm), 15/50 (200 ppm), 18/79 (625ppm), ovarian atrophy (4/49, 19/49, 32/48, 42/50, 43/50, 69/79), and uterineatrophy (1/50, 0/49, 1/50, 1/49, 8/50,41/78) were also dose related, withovarian atrophy significantly increasedat the lowest BD exposure of 6.25 ppm.These toxic effects to the reproductiveorgans are discussed in greater detail inthe reproductive effects section of thispreamble. Bone marrow atrophy wasnoted only in the highest exposuregroups, occurring in 23/73 male miceand 11/79 female mice.

Stop-Exposure StudyAs with the 2-year study, the body

weights of the four treated groups in thestop-exposure study were similar tocontrols. All exposure groups exhibitedmarkedly lower survival than controls,and only slightly better survival thanthat of the comparable full lifetimeexposure groups. Mortality appeared to

be more related to total dose than toexposure concentration. Most deathswere caused by tumors.

Neoplastic EffectsAll of these stop-exposure groups

exhibited a very similar tumor profile tothat of the lifetime high exposuregroups, with the lone exception of livertumors, which were increased only inthe lifetime exposure group; all theother multiple primary tumors wereobserved at significantly increasedlevels in both the stop- and lifetime-exposure groups, Table V–2. (Ex. 125) Inaddition, the 625 ppm, 26 weekexposure group had higher rates forseveral of the tumor types compared tothe lifetime 625 ppm group, possiblybecause of the shorter exposure group’sslightly better survival. The mostprevalent tumor type, lymphoma, alsoshowed a dose-rate effect, as the tumorincidence was greater for exposure toshort-term higher concentrationscompared with a lower long-termexposure (p=.01; 24/50 at 625 ppm for

13 weeks vs. 12/50 at 200 ppm for 40weeks: p<.0001; 37/50 at 625 ppm for 26weeks vs. 15/50 at 312 ppm for 52weeks). The same pattern was seen withforestomach tumors and preputial glandcarcinomas. Conversely, thehemangiosarcomas of the heart andalveolar-bronchiolar tumors showed anopposite trend, as lower exposures for alonger time yielded a significantlyhigher incidence of these tumors thanthe same cumulative exposures over ashorter time (survival-adjusted, asopposed to the raw incidence lungtumor rates actually suggest no dose-response trends). These inconsistenttrends with the different tumor sitesmay be the result of multiplemechanisms of carcinogenicity orpartially due to the rapid fatality causedby lymphocytic lymphomas in theshort-term high-exposure groups. Aswith the lifetime study, angiosarcomasof the heart and lymphomas presentedcompeting risks in the highly exposedmice.

TABLE V–2.—TUMOR INCIDENCES (I) AND PERCENTAGE MORTALITY-ADJUSTED TUMOR RATES (R) IN MALE MICE EX-POSED TO 1,3-BUTADIENE IN STOP-EXPOSURE STUDIES. (AFTER EXPOSURES WERE TERMINATED, ANIMALS WEREPLACED IN CONTROL CHAMBERS UNTIL THE END OF THE STUDY AT 104 WEEKS.)

[Adapted from Ex. 125]

Tumor

Exposure

0 200 ppm,40 wk

625 ppm,13 wk

312 ppm,52 wk

625 ppm,26 wk

I R c I R I R I R I R

Lymphoma ................................................................................................. 4/70 8 12/50 a 35 24/50 a 61 15/50 a 55 37/50 a 90Heart—Hemang-iosarcoma ....................................................................... 0/70 0 7/50 a47 7/50 a 31 33/50 a 87 13/50 a 76Lung—Alveolar-bronchiolar adenoma and carcinoma .............................. 22/70 46 35/50 a 88 27/50 a 87 32/50 a 88 18/50 a 89Forestomach—Squamous-cell papilloma and carcinoma ......................... 1/70 2 6/50 a 20 8/50 a 33 13/50 a 52 11/50 a 63Harderian gland—Adenoma and adenocarcinoma ................................... 6/70 13 27/50 a 72 23/50 a 82 28/50 a 86 11/50 a 70Preputial gland—Carcinoma ...................................................................... 0/70 0 1/50 3 5/50 a21 4/50 a 21 3/50 a 31Kidney—Renal tubular adenoma ............................................................... 0/70 0 5/50 a 16 1/50 5 3/50 a 15 1/50 11

From Melnick et al (1990).AAaIncreased compared with chamber controls (0ppm), p<0.05, based on logistic regression analysis.cMortality adjusted tumor rates are adjusted for competing causes of mortality, such as death due to other tumors, whose rates differ by expo-

sure group.

Activated Oncogenes

The presence of activated oncogenesin the exposed groups which differ fromthose seen in tumors in the controlgroup can help in identifying amechanistic link for BD carcinogenicity.Furthermore, certain activatedoncogenes are seen in specific humantumors and K-ras is the most commonlydetected oncogene in humans. Inindependent studies, tumors from thisstudy were evaluated for the presence ofactivated protooncogenes. (Ex. 129)Activated K-ras oncogenes were foundin 6 of 9 lung adenocarcinomas, 3 of 12hepatocellular cancers and 2 of 11lymphomas in BD exposed mice. Nine

of these 11 K-ras mutations, includingall six of those seen in lung tumors,were G to C conversions in codon 13.Activation of K-ras genes by codon 13mutations has not been detected in lungor liver tumors or lymphomas inunexposed B6C3F1 mice, but activationby codon 12 mutation was observed in1 of 10 lung tumors in unexposed mice.(Ex. 129)

ConclusionAll of the four animal bioassays (one

rat, three mouse) find a clearcarcinogenic response; together theyprovide sufficient evidence to declareBD a known animal carcinogen and aprobable human carcinogen. The three

mouse studies, all with a positivelymphoma response, further support afinding that the mouse is a good modelfor BD related lymphatic/hematopoieticand other site tumorigenicity. The mostrecent NTP II study confirms andstrengthens the previous NTP I andIrons et al. mouse studies, and presentsclear evidence that BD is a potentmultisite carcinogen in B6C3F1 mice ofboth sexes. (Ex. 23–1;32–28D, Irons) Thefinding of lung tumors at exposures aslow as 6.25 ppm, 100 fold lower thanthe lowest exposure of the NTP I studyand a level that is in the occupationalexposure range, increases concern forworkers’ health. Two other concerns


raised by both the second NTP and theIrons et al. studies are, (1) substantialcarcinogenicity is found with less-than-lifetime exposures (as low as 12 or 13weeks) for lymphomas andhemangiosarcomas, at least at higherconcentrations, and, (2) for lymphomasand at least two other sites, thereappears to be a dose-rate effect, whereexposure to higher concentrations for ashorter time yields higher tumorresponse (by a factor of as much as 2–3) than a comparable total exposurespread over a longer time. Thesefindings suggest that even short-termexposures should be as low as possible.Positive studies for genotoxicity and thedetection of activated K-ras oncogenesin several of these tumors induced inmice, including lymphomas, liver, andlung, suggest a mutagenic mechanismfor carcinogenicity, and support relianceon a linear low-dose extrapolationprocedure (on the basis of the multistagemutagenesis theory of carcinogenicity),at least for these tumor sites. Thefinding of activated K-ras oncogenes inthese mouse tumors may also berelevant to humans, because K-ras is the

most commonly detected oncogene inhumans.

The different dose-rate trends fordifferent tumor sites suggest thatdifferent mechanisms are involved atdifferent sites. The observation of ahighly nonlinear exposure-response forlymphomas at exposure levels of 625ppm and above suggests a secondaryhigh-exposure mechanism as well, notmerely a metabolic saturation, as issuspected with the high-exposuresaturation seen in Sprague-Dawley rats.(Ex. 34–6, Owen and Glaister) Thepicture emerges of BD as a potentgenotoxic multisite carcinogen in mice,far more potent in mice than in rats.

With respect to appropriate tumorsites for risk extrapolation from mouseto humans, Melnick and Huff havepresented information comparinganimal tumor response for five knownor suspected human carcinogens—BD,benzene, ethylene oxide, vinyl chloride,and acrylonitrile. (Ex. 117–2) BD,benzene, and ethylene oxide all havestrong occupational epidemiologyevidence of increased lymphatic/hematopoietic cancer (LHC) mortalityand all three cause both LHC, lung,

Harderian gland, and mammary glandtumors in mice, plus several otherprimary tumors (see Table V–3). OnlyBD and vinyl chloride cause mousehemangiosarcomas, BD in the heart andvinyl chloride in the liver. In rats, whileall five carcinogens cause tumors atmultiple sites, only brain and Zymbalgland tumors are associated with asmany as four of the compounds. Ingeneral mice and rats are affected atdifferent tumor sites by thesecarcinogens. LHC, lung, Harderiangland, mammary gland and, possiblyhemangiosarcomas are sites in micewhich correlate well with human LHC.This suggests that mice, rats andhumans may have different target sitesfor the same carcinogen, but thatcompounds which are multisitecarcinogens in the mouse and rat arelikely to be human carcinogens as well.Based on BD’s strong LHC association inhumans, and its multisitecarcinogenicity in the mouse, includingoccurrence at several of the same targetsites seen with other carcinogens, OSHAconcludes that the mouse is a goodanimal model for predicting BDcarcinogenesis in humans.

TABLE V–3.—SITES AT WHICH NEOPLASMS ARE CAUSED BY 1,3-BUTADIENE IN MICE AND RATS: COMPARISON WITHRESULTS OF STUDIES WITH BENZENE, ETHYLENE OXIDE, VINYL CHLORIDE AND ACRYLONITRILE

[From Ex. 117–2]

Site1,3–Butadiene Benzene Ethylene oxide Vinyl chloride Acrylonitrile

Mice Rats Mice Rats Mice Rats Mice Rats Mice Rats

Lymphatic/hematopoietic ................................... • • • • NS

Lung ................................................................... • • • •Heart .................................................................. f •Liver ................................................................... • • a • a •Forestomach ...................................................... • • • • •Harderian gland ................................................. • • •Ovary .................................................................. • •Mammary gland ................................................. • • • • • •Preputial gland ................................................... • •Brain ................................................................... • • • •Zymbal gland ..................................................... • • • • •Uterus ................................................................. • • •Pancreas ............................................................ •Testis .................................................................. •Thyroid gland ..................................................... •

NS, not studied.Hemangiosarcoma.

2. Epidemiologic Studies

(i) Introduction. OSHA has concludedthat the epidemiologic studiescontained in this record, as well as therelated hearing testimony and recordsubmissions, show that occupationalexposure to BD is associated with anincreased risk of death from cancers ofthe Lymphohematopoietic (LH) system.However, in contrast to the availabletoxicologic data, our understanding ofBD epidemiology is based on

observational studies, not experimentalones. In other words, the investigatorswho conducted these epidemiologicstudies did not have control over theexposure status of the individualworkers. They were, nonetheless, able toselect the worker populations and theobservational study design.

Cohort and case control studies aretwo types of observational studydesigns. Each of these designs hasstrengths and weaknesses that should beconsidered when the results are

interpreted. Cohort studies, for example,have the advantages of decreasing thechance of selection bias regardingexposure status and providing a morecomplete description of all healthoutcomes subsequent to exposure. Thedisadvantages of cohort studies includethe large number of subjects that areneeded to study rare diseases and thepotentially long duration required forfollow-up. By comparison, case controlstudies are well suited for the study ofrare diseases and they require fewer


subjects. The disadvantages of casecontrol studies, however, include thedifficulty of selecting an appropriatecontrol group(s), and the reliance onrecall or records for information on pastexposures. Regardless of the selectedobservational study design, the greatestlimitation of occupationalepidemiologic studies is their ability tomeasure and classify exposure.

In spite of the inherent limitations ofobservational epidemiologic studies,guidelines have been developed forjudging causal association betweenexposure and outcome. Criteriacommonly used to distinguish causalfrom non-causal associations include:Strength of the association as measuredby the relative risk ratio or the oddsratio; consistency of the association indifferent populations; specificity of theassociation between cause and effect;temporal relationship between exposureand disease which requires that causeprecede effect; biologic plausibility ofthe association between exposure anddisease; the presence of a dose-responserelationship between exposure anddisease; and coherence with presentknowledge of the natural history andbiology of the disease. These criteriahave been considered by OSHA in thedevelopment of its conclusion regardingthe association between BD and cancerof the LH system.

As stated previously, each type ofepidemiologic study design hasstrengths and weaknesses. Sinceepidemiologic studies are observationaland not experimental, each study willalso have inherent strengths andweaknesses; there is no perfectepidemiologic study. The mostconvincing evidence of the validity andreliability of any epidemiologic studycomes with replication of the study’sresults.

There are six major epidemiologicstudies in the record that haveexamined the relationship betweenoccupational exposure to BD andhuman cancer. These studies include: ANorth Carolina study of rubber workers(Ex. 23–41; 23–42; 23–4; 2–28; 23–27;23–3); a Texaco study of workers at a BDproduction facility in Texas (Ex. 17–33;34–4; 34–4); a NIOSH study of twoplants in the styrene-butadiene rubber(SBR) industry (Ex. 2–26; 32–25); theMatanoski cohort study of workers inSBR manufacturing (Ex. 9; 34–4); thenested case-control study of workers inSBR manufacturing conducted byMatanoski and Santos-Burgoa (Ex. 23–109); and a follow-up study of syntheticrubber workers recently completed byDelzell et al. (Ex. 117–1). Severalcomments in the record have concludedthat these studies demonstrate a positive

association between occupationalexposure to BD and LH cancers.However, OSHA has been criticized bythe Chemical Manufacturers Association(CMA) and the International Institute ofSynthetic Rubber Producers, Inc. (IISRP)for its interpretation of these studies asshowing a positive association; the chiefcriticisms will be discussed below. (Ex.112 and 113)

OSHA’s final consideration of the BDepidemiologic studies is organized andpresented according to what have beenidentified as key issues. These are theepidemiologic issues that were raisedand considered throughout therulemaking. They are also the issuesmost pertinent to OSHA’s conclusions.These key issues surrounding BDexposure and LH cancer are: Evidenceof an association; observation of a dose-response relationship; observation ofshort latency periods; the potential roleof confounding exposures and theobserved study results; the biologicalbasis for grouping related LH cancers;relevance of subgroup analyses; andappropriateness of selected referencepopulations.

(ii) Evidence of an AssociationBetween BD and LH Cancer. Each of thestudies listed above contributes to theepidemiologic knowledge upon whichOSHA’s conclusion regarding therelationship of BD exposure and LHcancer has been developed.

(a) North Carolina Studies. This seriesof studies was undertaken to examinework-related health problems of apopulation of workers in a major tiremanufacturing plant. They were notdesigned to look specifically at thehealth hazards of BD. (Tr. 1/15/91, p.117) However, in a work area thatinvolved the production of elastomers,including SBR, relative risks of 5.6 forlymphatic and hematopoieticmalignancies and 3.7 for lymphaticleukemia were found among workersemployed for more than five years. TheInternational Agency for Research onCancer (IARC) evaluation concludedthat this study suggests an associationbetween lymphatic and hematopoieticmalignancy and work in SBRmanufacturing. (Tr. 1/15/91, p. 117)However, the IISRP asserted that thesestudies do not provide ‘‘meaningfulevidence of an association betweenbutadiene and cancer.’’ (Ex. 113, p. A–23) OSHA recognizes that theresearchers who conducted thesestudies acknowledged that the workersmay have had exposures to organicsolvents, including benzene, a knownleukemogen, as pointed out by theIISRP. (Ex. 113, p. A–24)

(b) Texaco Study. The two Texacostudies examined mortality of a

population of workers in a BDmanufacturing facility in Texas. (Ex. 17–33; 34–4 Vol. III, H–2; Divine 34–4, Vol.III, H–1) A qualitative method ofexposure classification, based ondepartment codes and expert consensusjudgement, was used in the Downsstudy. (Ex. 17–33; 34–4, Vol. III, H–2)From this methodology four exposuregroups were defined: Low exposure,which included utility workers,welders, electricians, and officeworkers; routine exposure, whichincluded process workers, laboratorypersonnel, and receiving, storage andtransport workers; non-routineexposure, which included skilledmaintenance workers; and unknownexposure, which included supervisorsand engineers. It is OSHA’s opinion thatalthough this is a crude approach toexposure classification, there areimportant findings in this study thatcontribute to our understanding of BDepidemiology.

In the Downs study (Ex. 34–4, Vol. III,H–2) the standardized mortality ratio(SMR) for all causes of death in theentire study cohort was low (SMR 80; p< .05) when compared to nationalpopulation rates. However, astatistically significant excess of deathswas observed for lymphosarcoma andreticulum cell sarcoma combined (SMR235; 95% confidence interval (CI) =101,463) when compared with nationalpopulation rates. (The issue of referencepopulation selection is discussed belowin paragraph (viii).)

When analyzed by duration ofemployment, the SMR for the categoryof all LH neoplasms was higher inworkers with less than five yearsemployment (SMR = 167) than for thosewith more than five years employment(SMR = 127). (Ex. 34–4, Vol. III, H–2)However, neither of these findings wasstatistically significant. Alternatively, ithas been suggested that perhaps theshort-term workers were wartimeworkers, and that these workers wereactually exposed to higher levels of BD,albeit for a shorter time. (Tr. 1/15/91, p.119)

Analyses of the four exposure groupsalso showed elevated but notstatistically significant SMRs. Theroutine exposure group had a SMR of187 for all LH neoplasms, explainedprimarily by excesses in Hodgkin’sdisease (SMR = 197) and otherlymphomas (SMR = 282). (Ex. 34–4, Vol.III, H–2) Those workers in the non-routine exposure group also had anelevated SMR for all LH neoplasms(SMR = 167), with excess mortality forHodgkin’s disease (SMR = 130),leukemias (SMR = 201), and other


lymphomas (SMR = 150) (Ex. 34–4, Vol.III, H–2).

These data were updated by Divine byextending the period of follow-up from1979 through 1985. (Ex. 34–4, Vol. III,H–1) The SMR for all causes ofmortality remained low (SMR = 84, 95%CI = 79,90), as it did for mortality fromall cancers (SMR = 80, 95% CI = 69,94).(Ex. 34–4, Vol. III, H–1) However, theSMR for lymphosarcoma andreticulosarcoma combined was elevatedand statistically significant (SMR = 229,95% CI = 104,435). This finding wasconsistent with the previous analysesdone by Downs. (Tr. 1/15/91, p. 120).

Exposure group analyses were alsoconsistent with the previous findings byDowns. The highest levels of excessmortality from lymphatic andhematopoietic malignancy were againseen in the routine and non-routineexposure groups. The routine exposuregroup that was ‘‘ever employed’’ had astatistically significant excess oflymphosarcoma (SMR = 561, 95% CI =181,1310), that accounted for most ofthe LH excess. (Ex. 34–4, Vol. III, H–1)The cohort of workers employed before1946 (wartime workers) alsodemonstrated a statistically significantexcess of mortality due tolymphosarcoma and reticulosarcomacombined (SMR = 269, 95% CI =108,555). (Ex. 34–4, Vol. III, H–2)

In summary, the Texaco studyprovides several notable results. Thefirst of these is the consistently elevatedmortality for lymphosarcoma. Thisfinding is consistent with excesslymphomas observed in experimentalmice. (Ex. 23–92) Second, the excessrisk of mortality was found in theroutine and non-routine exposuregroups. Based on the types of jobs heldby workers in these two exposuregroups, this finding suggests that theincidence of lymphatic malignancy ishighest in the groups with the heaviestoccupational exposure to BD. (Tr. 1/15/91, p. 121) The third notable result ofthis study was the significantly elevatedrate of malignancy in workers employedfor fewer than 10 years.

(c) NIOSH Study. The NIOSH studywas undertaken in January 1976 inresponse to the report of deaths of twomale workers from leukemia. (Ex. 2–26;32–25) These workers had beenemployed in two adjacent SBR facilities(Plant A and Plant B) in Port Neches,Texas. The hypothesis tested by thisstudy is that:

Employment in the SBR productionindustry was associated, specifically, with anincreased risk of leukemia and, moregenerally, with an increased risk of othermalignancies of hematopoietic and lymphatictissue. (Ex. 2–26)

This study did not specifically examinethe association between BD and all LHcancers. Thus, OSHA agrees with thecriticism that this study by itself did notdemonstrate that occupational exposureto BD causes cancer. (Ex. 113, pp. A–13,A–19) However, the findings in thisstudy are consistent with the patternsobserved in the other epidemiologicstudies discussed in this section. InPlant A, the overall mortality wassignificantly decreased (SMR=80,p<0.05). (Ex. 2–26) The SMR for allmalignant neoplasms was alsodecreased (SMR=78), but this result wasnot statistically significant. (Ex. 2–26)The SMR for LH cancers was elevated(SMR=155), as it was forlymphosarcoma and reticulum cellsarcoma (SMR=181) and leukemia(SMR=203), but none of these resultswas statistically significant. (Ex. 2–26)

The pattern of mortality for asubgroup of wartime workers was alsoexamined for the Plant A population.For this subgroup of white males,employed at least six months betweenthe beginning of January 1943 and theend of December 1945, there was anelevated SMR for lymphatic andhematopoietic neoplasms (SMR = 212)that was statistically significant at thelevel of 0.05<p<0.1. (Ex. 2–26) Likewise,the SMR for leukemia was increased(SMR=278), also with statisticalsignificance at the level of 0.05<p<0.1.(Ex. 2–26)

At Plant B, the overall mortality waslow (SMR=66, p<0.05), as was deathfrom all malignant neoplasms (SMR=53,p<0.05). (Ex. 2–26) The SMR for LHcancer was also low (SMR=78), but thisfinding was not statistically significant.(Ex. 2–26)

When this study was updated, themortality patterns remained unchanged.(Ex. 32–25) The most remarkablefindings of the NIOSH study are theexcess mortality for malignancies of theLH system, and the excess of thesecancers in workers employed during thewartime years.

(d) Matanoski Cohort Study. Thecohort study conducted by Matanoski etal. is comprised of two follow-upperiods: In the original study,completed in June 1982, the cohort wasfollowed from 1943 to 1979; and in theupdate, completed in March 1988, thecohort follow-up period was extendedto 1982. (Ex. 9; 23–39; 34–4, Vol. III, H–3 and H–6, respectively) The originalstudy analyzed mortality data for 13,920male workers employed for more thanone year in eight SBR production plantsin the United States and Canada.Although historical quantitativeexposure data were not available,creation of a job dictionary made it

possible to designate four general workactivities as surrogates for exposure:Production; utilities; maintenance; anda combined category of all other jobs.The work activities with the highest BDexposures were production andmaintenance. (Ex. 16–39) The totalduration worked was measured by thedates of first and last employment.

The mortality experience for theoriginal study cohort, as compared withdeath rates for males in the UnitedStates, was low for all causes (SMR=81)and all cancers (SMR=84). (Ex. 9; 23–39)The SMR for all LH cancers was alsolow (SMR=85). (Ex. 9; 23–39) Themortality rate for Hodgkin’s disease wasslightly elevated (SMR=120), but it wasnot statistically significant. (Ex. 9; 23–39) In fact, there were no statisticallysignificant excesses in mortality fromcancer at any site found in this originalcohort study.

These initial data were also analyzedaccording to major work area. Therewere not any elevations of mortalityrates for the category of all LH cancers.(Ex. 9; 23–39) For production workers,the SMR for other lymphatic cancerswas elevated (SMR=202), but it was notstatistically significant. (Ex. 9; 23–39)The SMR for leukemia in the utilitieswork group was also elevated(SMR=198), but it was based on onlytwo deaths and was not statisticallysignificant. (Ex. 9; 23–39) Slightexcesses, none of which was statisticallysignificant, were seen for Hodgkin’sdisease in each of the four work groupcategories. (Ex. 9; 23–39)

OSHA has been criticized for itsopinion, expressed in the preamble ofthe BD proposed rule, that the originalMatanoski cohort study did not havesufficient power to detect a difference inthe cancer SMR if one actually existed.(Ex. 113, pp. A–10–11) Statistical powerof at least 80% is the accepted rule-of-thumb for epidemiologic research studydesign. Calculations provided byMatanoski indicate that, for theoutcomes of greatest concern to OSHA,statistical power was often below the80% level. (Ex. 9) For leukemia,statistical power to detect 25% and 50%increases in mortality was only 27%and 62%, respectively. (Ex. 9) Thepower to detect a 25% increase inmortality for all lymphohematopoieticcancers was only 49%. (Ex. 9) However,the study did have a statistical power of93% to detect a SMR of 150 for all LHcancers. (Ex. 9) Thus, for the cancers ofmost interest to OSHA, this study hadlimited statistical power to detectmortality excesses that were less thantwo-fold. OSHA does not consider thisto be an ‘‘unrealistically strict standardof acceptability,’’ as alleged by the


IISRP, but rather part of a thoroughcritique of an epidemiologic study withpurportedly ‘‘negative results.’’ (Ex. 113,p. A–11)

The update of Matanoski’s originalstudy extends the period of cohortfollow-up from 1979 to 1982, providinga full 40 years of mortality experiencefor analysis. The update study cohortdiffered from the original cohort in twoadditional ways: Canadian workers withrelatively short-term exposure wereremoved from the cohort; and theproportion of workers lost to follow-upwas reduced. The extension of follow-up resulted in findings of excessmortality from lymphatic andhematopoietic cancers that had not beenobserved in the original analyses. (Ex.34–4, Vol. III, H–6)

The SMR for all causes of mortalityremained low (SMR=81, 95% CI=78,85),as it did for death from all cancers(SMR=85, 95% CI=78,93). (Ex. 34–4,Vol. III, H–6) For lymphatic andhematopoietic cancers, the overall SMRfor white males was not increased(SMR=92, 95% CI=68,123). (Ex. 34–4,Vol. III, H–6) However, for black males,the SMR for all LH cancers was elevated(SMR=146, 95% CI=59,301). (Ex. 34–4,Vol. III, H–6) Specific increases werealso found for lymphosarcoma(SMR=132), leukemia (SMR=218, 95%CI=59,560), and other lymphaticneoplasms (SMR=116, 95% CI=14,420).(Ex. 34–4, Vol. III, H–6) These increaseswere based on small numbers ofobserved deaths.

Analyses conducted on the fourexposure groups also produced someevidence of excess mortality. For thetotal cohort of production workers, anelevated SMR was observed for alllymphopoietic cancers (SMR=146, 95%CI=88,227). (Ex. 34–4, Vol. III, H–6) Forwhite production workers, the SMR forthat category was 110, explainedprincipally by excess mortality fromother lymphatic neoplasms (SMR=230,95% CI=92,473). (Ex. 34–4, Vol. III, H–6) Although based on small numbers,the results for black production workerswere more pronounced and statisticallysignificant: The SMR for all lymphaticand hematopoietic cancers was 507(95% CI=187,1107). (Ex. 34–4, Vol. III,H–6) That overall increase in blackworkers reflected excess mortality fromlymphosarcoma (SMR=532), leukemia(SMR=656, 95% CI=135,1906), andother lymphatic cancers (SMR=482,95% CI=59,1762). (Ex. 34–4, Vol. III, H–6)

A pattern of excess mortality for allLH cancers was also seen in utilityworkers (SMR=203, 95% CI=66,474).(Ex. 34–4, Vol. III, H–6) That elevatedSMR may be explained by elevated rates

for leukemia (SMR=192, 95%CI=23,695) and other lymphatic cancers(SMR=313, 95% CI=62,695). (Ex. 34–4,Vol. III, H–6) No increases in LHmalignancy were seen in the otherexposure groups, i.e., maintenance orother workers.

From these study results Matanoski etal. concluded:

Deaths from cancers of the hematopoieticand lymphopoietic system are higher thanexpected in production workers withsignificant excesses for leukemias in blackworkers and other lymphomas in all(production) workers. (Ex. 34–4, Vol. III, H–6, p. 116)

In response to criticism from the IISRPthat OSHA placed too much emphasison the findings in the group of blackproduction workers, OSHA is aware ofthe statement offered by the researchersthat because of the potential for biasfrom misclassification of race: ‘‘* * *the total SMRs are probably the mostcorrect representation of risk.’’ (Ex. 34–4, Vol. III, H–6) However, OSHA alsoagrees with the authors that the risk ofdeath from LH cancers seems to behigher in this SBR industry populationthan in the general population, andthese causes of death seem to beassociated with different work areas.These cohort study findings stimulatedthe design and implementation of theSantos-Burgoa and Matanoski nestedcase-control study.

(e) Santos-Burgoa and MatanoskiNested Case-Control Study. To furtherinvestigate the findings of the cohortstudy, Santos-Burgoa and Matanoski etal. designed and conducted a case-control study of LH cancers in workersin the styrene-butadiene polymermanufacturing industry. (Ex. 23–109;34–4, Vol. III, H–4) The specificquestions addressed by this researchstudy are: ‘‘Is there a risk of anylymphatic or hematopoietic cancerwhich is associated with exposure to(BD) or styrene or both?’’; and ‘‘is therea risk of these cancers related toexposure to jobs within the industry?’’(Ex. 34–4, Vol. III, H–4) This is the firststudy to specifically investigate theassociation between LH cancers andindividual worker exposure to BD,which is why, contrary to the opinionof IISRP, OSHA places so much‘‘weight’’ on these results. (Ex. 113, pp.A–25–34)

The subjects in this case-control studywere ‘‘nested,’’ or contained, within thepopulation of the original cohort study.‘‘Cases’’ in this study were defined asmales who worked one year or more atany of eight synthetic rubber polymerproducing plants and who died of orwith a lymphopoietic cancer. Thesecancers included: Lymphosarcoma and

reticulum cell sarcoma, Hodgkin’slymphoma, non-Hodgkin’s lymphoma,all leukemias, multiple myeloma,polycythemia vera, and myelofibrosis.Sixty-one cases were identified, but twocases were omitted from data analyses,resulting in a total of 59 cases. One casewas omitted because he could not bematched to controls, and the other caselacked job records from which exposurecould be identified.

Eligible ‘‘controls’’ included workerswho were either alive or had died of anycause other than malignant neoplasms,who had been employed at one of theeight SBR plants, and who had not beenlost to follow-up. These controls wereindividually matched to cases on thefollowing criteria: Plant; age; hire year;employment as long or longer than thecase; and survival to the death of thecase. The study aim was to select fourcontrols per case. Even though this wasnot always possible, there were, onaverage, just over three controls per casein each group of lymphopoietic cancer.The total number of controls was 193.

Unlike the previous studies, in thisresearch study an exposuremeasurement value for BD (and also forstyrene) was determined for each caseand control. This value was determinedby a multi-step process. First, the jobrecords of each subject were reviewedand the number of months that each jobwas held was determined. Second, thelevel of BD (and styrene) associatedwith the job was estimated by a panelof five industrial experts, i.e., engineerswith long term experience in SBRproduction. The exposure level for BD(and styrene) for each job was based ona scale of zero to ten, with ten being therank given to the job with the highestexposure. The next step in thedevelopment of each individual job-exposure matrix was to add all of theexposures to the chemicals for all themonths a specific job was held and thensum the exposures over a workinglifetime. This procedure resulted in acumulative BD exposure value for eachcase and control.

The distribution of the cumulativeexposure estimates for the studypopulation was not normallydistributed, i.e., there were someextreme values. In order to approximatea normal distribution, a requiredassumption for many statisticalanalyses, a logarithmic transformationof these values was done. (Ex. 34–4, Vol.III, H–4) Exposure was analyzed as adichotomous variable, i.e., ever/neverexposed. ‘‘Exposed’’ workers weredefined as those with a log rankcumulative exposure score above themean of the scores for the entirepopulation of cases and controls within


a cancer subtype; ‘‘non-exposed’’workers were those with a score belowthe mean.

There were several important findingsin this study. First, in the unmatchedanalysis of cases and controls, theleukemia subgroup had a significantexcess risk of 6.8 fold for exposure toBD among cases compared to controls(Odds Ratio (OR)=6.82, 95%CI=1.10,42.23). (Ex. 34–4, Vol. III, H–4)The results were even stronger in thematched-pair analyses. In that analysisfor exposure to BD, the OR was 9.36(95% CI=2.05,22.94) in the leukemiasubgroup. (Ex. 34–4, Vol. III, H–4) Thisresult can be interpreted to mean thatcases with leukemia were more thannine times as likely as their controls tobe exposed to BD. Additionally, the datain this analysis indicate that BDexposure above the group mean is 2.3times (OR=2.30, 95% CI=1.13,4.71)more common among cases with alllymphopoietic cancers when comparedto a similar exposure in the controls.(Ex. 34–4, Vol. III, H–4)

This case-control study has been thesubject of criticism that has centered onboth validity and reliability. (Ex. 23–68;113) For example, the data from thisstudy have been criticized as being‘‘inconsistent’’ with the results of theMatanoski cohort study. (Ex. 23–68;113, p. A–25) Further, it has beensuggested that ‘‘the study results are notreliable and should not be relied uponby OSHA.’’ (Ex. 113, p. A–25) OSHArejects these criticisms for the reasonsdiscussed below.

First, regarding the issue ofinconsistency, a nested case controlstudy does not test the same hypothesesor make the same comparisons as acohort study. (Ex. 32–24; Tr. 1/15/91, p.161; Tr. 1/16/91, p. 347) In fact, aspresented in the above discussions ofthe studies, they ask and answerdifferent research questions. Forexample, the cohort study askedwhether all of the SBR workers have adifferent risk of leukemia from thegeneral population, and the case controlstudy asked whether workers withleukemia have different exposureswithin the industrial setting fromworkers without leukemia. (Ex. 32–24)Thus, the criticism that the results ofthese two studies are incompatible, andtherefore invalid, is not relevant. (Ex.32–24)

Second, the challenge directed at thereliability of the case-control study doesnot hold up under close scrutiny. Thiscriticism is based on four issues: Logtransformation of the exposure data;instability of the results; irregular dose-response pattern; and selection criteriafor ‘‘controls.’’ (Ex. 113, A–29–34)

Regarding the log transformation of theexposure data, the IISRP asserts thatthere is not a sound rationale for thisapproach to data analyses. (Ex. 113, A–29–30) However, Santos-Burgoa offeredthe following explanation of thisprocedure in his testimony:

For analysis, exposures were categorized inadvance above and below the mean of thecumulative exposure for the study subjects.This cutpoint was defined from the verybeginning of the analysis design as follows.The total cumulative exposures, as happensin most environmental exposures, showed askewed distribution with many observationsat the low levels and few at the high levels.Since the geometric mean is the best estimateof the central tendency point in log normaldata, such as exposure data, the cumulativeexposures were transformed by thelogarithm, and then the mean was calculated.(Ex. 40, pp. 12–13)

It is OSHA’s opinion that, given the lognormal distribution of the exposuredata, Santos-Burgoa chose the bestapproach for data analyses.

The case-control study has also beencriticized for producing ‘‘highlyunstable and therefore unreliable’’results. (Ex. 113, A–30) For example, theleukemia subgroup (matched-pairanalysis) OR of 9.36 with a 95%confidence interval of 2.05–22.94 hasbeen used to illustrate statisticalinstability of the data. (Ex. 113, A–31)However, as previously discussed, thedisease category of ‘‘all lymphopoieticcancers’’ (matched-pair analysis) had anOR of 2.30 with a confidence interval of1.13–4.71. Thus, it is OSHA’s opinionthat, while some specific odds ratiosmay have wide confidence intervals, thestudy results as a whole are not‘‘unreliable.’’

The IISRP has also criticized the case-control study for ‘‘* * * fail(ing) todemonstrate a dose-responserelationship * * *’’ (Ex. 113, A–32)However, the test for linear trend, i.e.,test for dose-response, shows astatistically significant, but irregular,trend in the odds of leukemia withincreasing levels of exposure to BD.Specifically, as exposure levels increasethe pattern of odds ratios is: 7.2; 4.9;13.0; 2.5; and 10.3. (Ex. 23–109, Table10) Although this is not a compellinglinear dose-response, in OSHA’sopinion, it is suggestive of a pattern ofincreasing disease risk at increasingexposure levels.

Inconsistent application of the controlselection criteria is the final criticismdirected at the case-control study by theIISRP. (Ex. 113, A–33) However, carefulreview of docket exhibits related to thecase-control study reveals this criticismto be unfounded. In his dissertation,Santos-Burgoa clearly states the protocolfor control selection:

All cohort subjects were arranged intogroups by plants, date of birth, date of hire,duration of work and duration of follow-up.A two and a half year period around eachtime variable was relaxed in a few instanceswhen no more controls were available. Onelymphosarcoma case was lost since no matchwas found for his date of birth, even allowingfor three and a half years around the date.This was the only case lost to analysisbecause of lack of a matched control. (Ex. 32–25, p. 80)

With only 59 cases, Santos-Burgoa wascorrectly concerned about loss ofvaluable data should any additionalcases need to be eliminated due to lackof a match. Also, regarding the potentialfor bias, abstractors were blinded to caseor control status when employment datawere being collected. (Ex. 34–4, Vol. III,App. H–5) Thus, it is most likely thatany misclassification bias would benondifferential, biasing the study resultstowards the null.

(f) Delzell et al. Follow-up Study forthe IISRP. The most recent study ofsynthetic rubber workers was conductedby Delzell et al. (Ex. 117–1) This studyupdated and expanded the research onSBR workers conducted by NIOSH,Matanoski et al., and Santos-Burgoa.More specifically, the Delzell et al.study consists of workers at seven ofeight plants previously studied by TheJohns Hopkins University (JHU)investigators, and the two plantsincluded in the NIOSH study.

This retrospective cohort studyevaluated the associations betweenoccupational exposure to BD, styrene,and benzene and mortality from cancerand other diseases among the SBRworkers. There were five studyobjectives:

(1) To evaluate the overall and cause-specific mortality experience of SBR workersrelative to that of the USA and Canadiangeneral populations;

(2) To assess the cancer incidenceexperience of Canadian synthetic rubberworkers relative to that of the generalpopulation of Ontario;

(3) To determine if overall and cause-specific mortality patterns vary by subjectcharacteristics such as age, calendar time,plant, period of hire, duration ofemployment, time since hire and payrollstatus (hourly or salaried);

(4) To examine relationships between workareas within the SBR study plants and cause-specific mortality patterns;

(5) To evaluate the relationship betweenexposure to BD and [styrene] and theoccurrence of leukemia and otherlymphopoietic cancers among SBR workers.(Ex. 117–1 p. 10)

The study population for thisinvestigation included 17,964 malesynthetic rubber workers employed inone of eight plants in either the USA orCanada. In order to be eligible for


inclusion, a worker had to be employedfor a total of at least one year before theclosing date of the study, January 1,1992. Additional eligibility criteria weredeveloped for selected plants due tolimitations in availability of plantrecords and follow-up of subjects. Theeligibility criteria in this study wereconsidered by the investigators to bemore restrictive than in either the JHUor NIOSH studies. (Ex. 117–1, p. 13)Most of the exclusions were based onless than one year of employment.During the study period of 1943 through1991, there were 4,665 deaths in thestudy population.

The methods used in this studyincluded development of work historyinformation and retrospectivequantitative exposure estimates forindividual members of the studypopulation. Complete work historyinformation was available forapproximately 97% of the study cohort.There was a total of 8,281 unique ‘‘workarea/job’’ combinations for all of theplants combined, with a range of 199 to4,850 in specific plants. Additionally,308 work area groups were definedbased on individual plant informationregarding production, maintenance, andother operations, as well as jobs andtasks within each type of operation. Five‘‘process groups’’ and seven ‘‘processsubgroups’’ were derived from the workarea groups. The process groupsinclude: Production of SBR, solutionpolymerization (SP), liquidpolymerization (LP), and latexproduction; maintenance; labor;laboratories; and other operations.

Six plants had sufficiently detailedindividual work history information foruse in development of retrospectivequantitative exposure estimates for BDand styrene. The process used toproduce these exposure estimatesincluded: In-depth walk-throughsurveys of each plant; meetings withplant management; interviews with keyplant experts, such as individuals withlong-term employment. The interviewswere used to collect informationregarding the production process,specific job tasks, and exposurepotential. Additionally, the results ofindustrial hygiene monitoring fromthese plants were obtained. The actualexposure estimation was based on:

Specification of the exposure model; theestimation of exposure intensities for specifictasks in different time periods; the estimationof exposure intensities for generic(nonspecific) job titles (e.g., ‘‘laboratoryworker’’) in different time periods; validationof exposure intensity estimates; thecomputation of job- and time period specificsummary indices; and the compilation of job-exposure matrices (JEMs) for BD, [styrene],

and [benzene] and linkage with subjects’work histories. (Ex. 117–1, pp. 27–28)

A limited validation of the quantitativeexposure estimations was conducted,which resulted in revision of theestimates used in analyses presented inthe Delzell et al. study. (Ex. 117–1)

The major findings of this study havebeen reported by Delzell et al. in fivecategories: General mortality patterns;mortality among USA subjectscompared to state populations; cancerincidence; mortality patterns by processgroup; and mortality patterns byestimated monomer exposure. Keyresults from each of these categories,especially as they relate to leukemia andother LH cancers, are briefly presented.

First, regarding general mortalitypatterns, there were deficits in both allcauses (SMR=87, 95% CI=85,90) and allcancers (SMR=93, 95% CI=87,99) for theentire cohort. (Ex. 117–1, p. 53) Of theLH cancers, excess mortality was onlyobserved for leukemia (SMR=131, 95%CI=97–174). (Ex. 117–1, p. 53) In acohort subgroup having 10 or moreyears of employment and 20 or moreyears since hire, the excess of leukemiadeaths was even greater (SMR=201, 95%CI=134,288). (Ex. 117–1, p. 54)

Analyses were also conducted toexplore the possibility of racialdifferences in the general mortalitypatterns. Regarding mortality fromleukemia, the SMRs were higher forblacks than for whites. In a subgroup of‘‘ever hourly’’ workers with 10 or moreyears of work and 20 or more yearssince hire, the SMRs for leukemia were192 (95% CI=119,294) for whites and436 (95% CI=176,901) for blacks. (Ex.117–1, p. 55)

Additionally, analyses were done byspecific groups of LH cancers:Lymphosarcoma; leukemia; and otherlymphopoietic cancer. For the overallcohort, there was an excess of mortalityfrom lymphosarcoma in those memberswho died in 1985 and beyond(SMR=215, 95% CI=59,551). (Ex. 117–1,p. 116) This excess was observed in‘‘ever hourly’’ white men; there were nolymphosarcoma deaths in blacks. (Ex.117–1, p. 119)

In the ‘‘other lymphopoietic cancer’’category, the overall cohort had a slightdeficit of mortality (SMR=97, 95%CI=70,132). (Ex. 117–1, p. 116) Whenanalyzed according to racial groups,whites were also observed to have adeficit of mortality from this group ofcancers (SMR=91, 95% CI=63,127). (Ex.117–1, p. 118) Blacks, however, had anincrease in mortality from ‘‘otherlymphopoietic’’ cancers (SMR=142,95% CI=61,279). (Ex. 117–1, p. 120)

The analyses for leukemia mortalityin the overall cohort showed a modest

increase (SMR=131, 95% CI=97,174).(Ex. 117–1, p. 116) The increase inmortality was found primarily in thesubgroups of workers who died in 1985or later, those that worked for 10 ormore years, and those with 20 or moreyears since hire. A dose-response typeof pattern was observed among ‘‘everhourly’’ subjects in the analysis of therelationship of leukemia and duration ofemployment: Less than 10 yearsworked, the SMR=95 (95% CI=53,157);10–19 years worked, the SMR=170 (95%CI=85,304); and 20 or more yearsworked, the SMR=204 (95%CI=123,318). (Ex. 117–1, p. 117)

Leukemia mortality was also analyzedfor racial difference among ‘‘everhourly’’ men. Overall, the SMR washigher for black subjects (SMR=227,95% CI=104,431) than for white(SMR=130, 95% CI=91,181). (Ex. 117–1,p. 122) In fact, there were statisticallysignificant elevations in the leukemiaSMR for black ‘‘ever hourly’’ men with20 or more years worked (SMR=417,95% CI=135,972), and 20 to 29 yearssince hire (SMR=446, 95%CI=145,1042). (Ex. 117–1, p. 122)

Second, Delzell et al. analyzed themortality data of the USA cohortsubgroup using both state generalpopulation rates and USA generalpopulation rates for comparison. Theoverall pattern of these analyses wasthat of ‘‘slightly lower’’ SMRs when thestate general population rates wereused. (Ex. 117–1, p. 60) For example, inthe analysis for leukemia mortality, theSMR using the USA rates was 131 (95%CI not provided), and it decreased to129 (95% CI=92,176) when state rateswere applied. (Ex. 117–1, pp. 61, 136)

Third, the results of the Delzell et al.study include an analysis of the cancerincidence in the Canadian plant (plant8). Regardless of whether the cancerexperience of terminated workers wasincluded or excluded, the overall cancerincidence was not elevated in thiscohort subgroup (SIR=105, 95%CI=93,117; SIR=106, 95% CI=94,119,respectively). (Ex. 117–1, pp. 61–62)However, analysis of this cohortsubgroup, with the terminated workersincluded, ‘‘revealed an excess ofleukemia cases before 1980 (overallcohort, 6 observed/3.0 expected; everhourly, 6 observed/2.9 expected)’’(further data were not provided). (Ex.117–1, p. 62)

Fourth, Delzell et al. examinedmortality patterns by work processgroup. These analyses producedelevated SMRs for both lymphosarcomaand leukemia. There was excesslymphosarcoma mortality in fieldmaintenance workers (SMR=219, 95%CI=88,451), production laborers


(SMR=263, 95% CI=32,951), andmaintenance laborers (SMR=188, 95%CI=39,548). (Ex. 117–1, pp. 65–66)However, these results were notstatistically significant, and may be dueto chance. For leukemia, the resultswere more striking: Polymerizationworkers had a SMR of 251 (95%CI=140,414); workers in coagulation hada SMR of 248 (95% CI=100,511);maintenance labor workers had a SMRof 265 (95% CI=141,453); and workersin laboratories had a SMR of 431 (95%CI=207,793). (Ex. 117–1, pp. 66,151) Itshould be noted that excess mortality bywork process group was also observedfor other cancers, i.e., lung cancer andlarynx cancer.

Fifth, the final set of analysesperformed by Delzell et al. was designedto examine mortality patterns byestimated monomer exposure, i.e., BD,styrene, and benzene. Poissonregression analyses conducted toexplore the association between ‘‘BDppm-years’’ and leukemia indicated apositive dose-response relationship,after controlling for styrene ‘‘ppm-years’’, age, years since hire, calendarperiod, and race. Specifically, in thecohort group that included all person-years and leukemia coded as eitherunderlying or contributing cause ofdeath, the rate ratios (RRs) were: 1.0, 1.1(95% CI=0.4,5.0), 1.8 (95% CI=0.6,5.4),2.1 (95% CI=0.6,7.1), and 3.6 (95%CI=1.0,13.2) for BD ppm-year exposuregroups of 0, >0–19, 20–99, 100–199, and200+, respectively. (Ex. 117–1, pp. 68–69; 158) Poisson regression analyseswere also conducted using varyingexposure categories of BD ppm-years.These analyses demonstrated a strongerand more consistent relationshipbetween BD and leukemia than betweenstyrene and leukemia. (Ex. 117–1, p. 69,159) Although a clearly positiverelationship between BD ‘‘peak-years’’and leukemia was observed fromadditional Poisson regression analyses,even after controlling for BD ppm-years,styrene ppm-years, and styrene peak-years, the dose-response relationshipwas less clear. (Ex. 117–1, pp. 71, 162)

In summary, one of the mostimportant findings of the research ofDelzell et al. was strong and consistentevidence that employment in the SBRindustry produced an excess ofleukemia. In the authors own words:

This study found a positive associationbetween employment in the SBR industryand leukemia. The internal consistency andprecision of the result indicate that theassociation is due to occupational exposure.The most likely causal agent is BD or acombination of BD and [styrene]. Exposure to[benzene] did not explain the leukemiaexcess. (Ex. 117–1, p. 85)

(g) Summary. These studies provide acurrent body of scientific evidenceregarding the association between BDand LH cancers. As previouslydiscussed, two of the criteria commonlyused to determine causal relationshipsare consistency of the association andstrength of the association. Theconsistency criterion for causality refersto the repeated observation of anassociation in different populationsunder different circumstances.Consistency is perhaps the most strikingobservation to be made from thiscollection of studies: ‘‘[E]very one ofthese studies to a greater or lesser extentfinds excess rates of deaths from tumorsof the lymphatic and hematopoieticsystem.’’ (Tr. 1/15/91, p. 129)

Strength of the association isdetermined by the magnitude andprecision of the estimate of risk. Ingeneral, the greater the risk estimate,e.g., SMR or odds ratio, and thenarrower the confidence intervalsaround that estimate, the more probablethe causal association. In the nestedcase-control study, although theconfidence intervals were wide, theodds ratios provide evidence of a strongassociation between leukemia andoccupational exposure to BD.

(iii) Observation of a Dose-ResponseRelationship. A dose-responserelationship is present when an increasein the measure of effect (response), e.g.,SMR or odds ratio, is positivelycorrelated with an increase in theexposure, i.e., estimated dose. Whensuch a relationship is observed, it isgiven serious consideration in theprocess of determining causality.However, the absence of a dose-response relationship does notnecessarily indicate the absence of acausal relationship.

OSHA has been criticized for itsconclusion that the epidemiologic datasuggest a dose-response relationship.(Ex. 113) The IISRP offers a differentinterpretation of the data. In theiropinion, the data provide a ‘‘consistentfinding of an inverse relationshipbetween duration of employment andcancer mortality.’’ (Ex. 113, A–34) Thisobservation is further described by JohnF. Acquavella, Ph.D., SeniorEpidemiology Consultant, MonsantoCompany, as ‘‘the paradox of butadieneepidemiology.’’ (Ex. 34–4, Vol. I,Appendix A) This interpretationassumes that cumulative occupationalexposure to BD will increase withduration of employment, and, thus,cancer mortality will increase withincreasing duration of employment. (Ex.113, A–35–39)

In OSHA’s opinion, this is anerroneous assumption; the

epidemiologic data for BD tell adifferent story. For the workers in theseepidemiologic studies, it is unlikely thatoccupational exposure to BD wasconstant over the duration ofemployment. According to Landrigan,BD exposures were most likely higherduring the war years than they were insubsequent years. (Tr. 1/15/91, p.146) Itis logical that exposures would beespecially intense during this timeperiod because of wartime productionpressures, the process of productionstart-up in a new industry, and thegeneral lack of industrial hygienecontrols during that phase of industrialhistory. Unfortunately, withoutquantitative industrial hygienemonitoring data, the true levels of BDexposure for wartime workers cannot beascertained. In the absence of such data,however, OSHA believes it is reasonableto consider wartime workers as a highlyexposed occupational subgroup. (Tr. 1/15/91, p. 121; Tr. 1/16/91, pp. 225–227)Thus, the excess mortality seen amongthese workers provides another piece ofthe evidence to support a dose-responserelationship between occupationalexposure to BD and LH cancers.

Additional support that excessmortality, among workers exposed toBD, is dose-related can be found in theanalyses of the work area exposuregroups. The studies by Divine,Matanoski, and Matanoski and Santos-Burgoa all provide evidence that excessmortality is greatest among productionworkers. (Ex. 34–4, Vol. III, H–1; 34–4,Vol. III, H–6; 23–109, respectively)Production workers are typically themost heavily exposed workers topotentially toxic substances. (Ex. 34–4)

The most compelling data thatsupport the existence of a dose-responserelationship for occupational exposureto BD and LH cancers are those in thestudy by Delzell et al. (Ex. 117–1)Analysis of the cumulative time-weighted BD exposure in ppm-yearsindicates a relative risk for all leukemiasthat increases positively with increasingexposure. This relationship is presenteven with statistical adjustment for age,years since hire, calendar period, race,and exposure to styrene. It is OSHA’sopinion that identification of a positivedose-response in an epidemiologicstudy is a very powerful observation interms of causality.

(iv) Observation of Short LatencyPeriods. Short latency periods, i.e., timefrom initial BD exposure to death, wereseen in two epidemiologic studies. Inthe NIOSH study, three of the sixleukemia cases had a latency periodfrom three to four years. (Ex. 2–26)Additionally, five of these six workerswere employed prior to 1945. (Ex. 2–26)


In the Texaco study update, a latency ofless than 10 years was seen in four ofthe nine non-Hodgkin’s lymphoma(lymphosarcoma) cases, and seven ofthese workers were also employedduring the wartime years. (Ex. 34–4,Vol. III, H–1)

According to OSHA’s expert witness,Dr. Dennis D. Weisenburger,these findings are contrary to the acceptedbelief that, if a carcinogen is active in anenvironment, one should expect the * * *SMRs to be higher for long-term workers thanfor short-term workers (i.e., larger cumulativedose). (Ex. 39, p. 9)

Thus, it has been argued that thesefindings appear to lack coherence withwhat is known of the natural historyand biology of LH cancers. (Ex. 113, A–40–42) Furthermore, these findings havebeen interpreted as evidence against acausal association between BD andthese LH cancers. (Ex. 113, A–42)

In OSHA’s opinion, there are otherpossible explanations for theseobservations. First, as proffered by Dr.Weisenburger, a median latency periodof about seven years has been found forleukemia in studies of atomic bombvictims, radiotherapy patients, andchemotherapy patients who havereceived high-dose, short-termexposures. (Ex. 39) In contrast, Dr.Weisenburger points out that low-doseexposure to an environmentalcarcinogen, such as benzene, has amedian latency period for leukemia ofabout 15–20 years. (Ex. 39) Heconcludes that short-term, high-doseexposures may be associated with ashort latency period, whereas long-term,low-dose exposures may be associatedwith a long latency period.

Second, the occurrence of shortlatency periods for LH cancer mortalityin these two studies was concentrated inworkers first employed during thewartime years. As previously discussed,it is possible that exposure to BD duringthe wartime years was greater than insubsequent years. (Ex. 39; Tr. 1/15/91,p. 121) Dr. Weisenburger suggests thatthe ‘‘short latency periods for LH cancerin these studies may be explained byintense exposures to BD over arelatively short time period.’’ (Ex. 39, p.10)

In his testimony, Dr. Landrigan,another OSHA expert witness, makesthe point that ‘‘duration of employmentis really only a crude surrogate for totalcumulative exposures, not itself ameasure of exposure.’’ (Tr. 1/15/91, p.121) In other words, it is possible thatshort-term workers employed during thewartime years may have actually hadheavier exposures to BD than long-termworkers. (Tr. 1/15/91, pp. 115–205) On

cross-examination, Dr. Landrigancautioned against ‘‘assuming thatduration of exposure directly relates tototal cumulative exposure.’’ (Tr. 1/15/91, p. 180) He also emphatically statedthat an increased cancer risk in short-term workers would not be inconsistentwith a causal association. (Tr. 1/15/91,p. 204)

(v) The Potential Role of ConfoundingExposures and Observed Results. Inepidemiologic studies ‘‘confounding’’may lead to invalid results.Confounding occurs when there is amixing of effects. More specifically,confounding may produce a situationwhere a measure of the effect of anexposure on risk, e.g., SMR, RR, isdistorted because of the association ofthe exposure with other factors thatinfluence the outcome under study.

For example, the IISRP has suggestedthat confounding exposures from otheremployment were responsible for theLH cancers observed in the studies ofBD epidemiology. (Ex. 113, A–43) Thisargument is based on the past practiceof using petrochemical industryworkers, who may have also beenexposed to benzene, to start up the SBRand BD production plants. The IISRPfinds support for this position in theobservation of elevated SMRs in short-term workers employed during thewartime years, precisely those mostlikely to be cross-employed. (Ex. 113,A–43)

However, there are a number ofresearch methods in occupationalepidemiology that are available tocontrol potential confounding factors.Research methods that eliminate theeffect of confounding variables include:Matching of cases and controls;adjustment of data; and regressionanalyses. In the nested case-controlstudy, for example, cases and controlswere matched on variables thatotherwise might have confounded thestudy results. In the testimony providedby Santos-Burgoa, he states that the‘‘matching scheme allowed us to controlfor potential confounders andconcentrate only on exposurevariations.’’ (Ex. 40, p. 12)

On cross-examination, Landrigan alsoaddressed the potential role ofconfounding exposures and theobserved study results. First, heobserved that Dr. Philip Cole, Professor,Department of Epidemiology, School ofPublic Health, University of Alabama atBirmingham, one of the outspokencritics of OSHA’s proposed rule, foundno evidence for confounding in hisreview of the Matanoski study. (Tr. 1/15/91, p. 178) Second, Dr. Landrigandismissed the notion of previousexposure to benzene as the causative

agent for the observed results in theshort-term workers. (Tr. 1/15/91, p.178–179)

In their analyses of mortality patternsby estimated monomer exposure,Delzell et al. used Poisson regression tocontrol for potential confoundingfactors. (Ex. 117–1) As previouslystated, the analyses conducted todetermine the association between BDppm-years and leukemia indicated apositive dose-response relationship,even after controlling for styrene ppm-years, age, years since hire, calendarperiod, and race. In the opinion of theinvestigators, benzene exposure did notexplain the excess of leukemia risk, andBD is the most likely causal agent. (Ex.117–1, p. 85)

(vi) The Biological Basis for GroupingRelated LH Cancers. The epidemiologicstudies that have examined theassociation between occupationalexposure to BD and excess mortalityhave grouped related LH cancers intheir analyses. This approach has beencriticized as evidence of a lack of‘‘consistency with respect to cell type’’which ‘‘argues against a commonetiologic agent.’’ (Ex. 113, A–45) Inother words, these critics suggest thatthe relationship between BD and excessmortality does not meet the specificityof association requirement for a causalrelationship. This requirement statesthat the likelihood of a causalrelationship is strengthened when anexposure leads to a single effect, notmultiple effects, and this finding alsooccurs in other studies.

More specifically, OSHA has beencriticized for its position that ‘‘broadcategories such as ‘leukemia’ or ‘allLHC’ should be used to evaluate theepidemiologic data.’’ (Ex. 113, A–46) Dr.Cole, for example, commented that:

It is a principle of epidemiology—and ofdisease investigation in general—that entitiesshould be divided as finely as possible inorder to maximize the prospect that one hasdelineated a homogeneous etiologic entity.Entities may be grouped for investigativepurposes only when there is substantialevidence that they share a common etiology.(Ex. 63, p. 11)

It is Dr. Cole’s opinion that LH cancersare ‘‘distinct diseases’’ with‘‘heterogeneous and multifactorial’’etiologies. (Ex. 63, p. 47)

Dr. Weisenburger, OSHA’s expert inhematopathology, provided testimony tothe contrary. (Ex. 39, pp. 7–8) Accordingto Dr. Weisenburger, ‘‘LH (cancer)cannot be readily grouped into‘etiologic’ categories, since the preciseetiologies and pathogenesis of LH(cancer) are not well understood.’’ (Ex.39, p. 7) In his opinion, because LHcancers are ‘‘closely related to one


another and arise from common stemcells and/or progenitor cells, it is validto group the various types of LH(cancer) into closely-related categoriesfor epidemiologic study.’’ (Ex. 39, p.7)

The issue of grouping related LHcancers to observe a single effect wasalso addressed by Dr. Landrigan in histestimony. (Tr. 1/15/91, pp. 131–133)The first point raised by Dr. Landriganis that the ‘‘diagnostic categories [for LHcancers] are imprecise and * * *overlapping.’’ (Tr. 1/15/91, p. 131) Forexample, he explained that in clinicalpractice transitions of lymphomas andmyelomas into leukemias may beobserved. In such a case, one physicianmay record the death as due tolymphoma and another may listleukemia as the cause of death. (Tr. 1/15/91, p. 131–132) Additionally, Dr.Landrigan testified that ‘‘some patientswith lymphomas or multiple myelomamay subsequently develop leukemia asa result of their treatments withradiation or cytotoxic drugs.’’ (Tr. 1/15/91, p. 132)

These recordings of disease transitionare further complicated by the historicalchanges that have occurred innomenclature and The InternationalClassification of Diseases (ICD) coding.According to Dr. Landrigan,certain lymphomas and * * * leukemias,such as chronic lymphatic leukemia are nowconsidered by some investigators * * * torepresent different clinical expressions of thesame neoplastic process. There have beenrecent immunologic and cytogenetic studieswhich indicate that there are stem cellswhich appear to have the capacity to developvariously into all the various sorts ofhematopoietic cells including T-lymphocytes, plasma cells, granulocytes,erythrocytes, and monocytes. (Tr. 1/15/91, p.132)

Dr. Landrigan summarized histestimony on this issue by stating that‘‘these different types of cells share acommon ancestry * * * there is goodbiologic reason to think that they wouldhave etiologic factors in common.’’ (Tr.1/15/91, pp. 132–133)

OSHA maintains the opinion, whichis well supported by the record, thatthere is a biological basis and amethodologic rationale for groupingrelated LH cancers. Furthermore, OSHArejects the criticism that the observationof different subtypes of LH cancersargues against the consistency andspecificity of the epidemiologicfindings.

(vii) Relevance of Worker SubgroupAnalyses. OSHA has been criticized forfocusing on and emphasizing the ‘‘fewpositive results’’ seen in the results ofworker subgroup analyses. (Ex. 113, A–48) It has been pointed out, for example,

that in the update of the Matanoskicohort study ‘‘there were hundreds ofSMRs computed in that study and it’snot surprising that one or two or evenmore would be found to be statisticallysignificant even when there is in factnothing going on.’’ (Tr. 1/22/91, p. 1444)Additionally, it has been suggested thatOSHA has ignored the ‘‘clearly overallnegative results’’ of the epidemiologicstudies. (Ex. 113, A–48)

OSHA agrees with the observationthat when many statistical analyses aredone on a database, it is possible thatsome positive results may be due tochance. However, OSHA rejectscriticism that the Agency hasinappropriately concentrated on thepositive results and disregarded thenegative results. It is OSHA’s opinionthat there is a compelling pattern ofresults in the epidemiologic studies.

Furthermore, a reasonableexplanation for the elevated SMR forblack production workers in the updateof the Matanoski cohort study is thatthis subset of the population actuallyhad heavy exposure to BD. Support forthis explanation can be found in theindustrial hygiene survey results ofFajen et al. (Ex. 34–4) In this case, then,the risk for excess mortality would beconcentrated in a small subset ofotherwise very healthy and unexposedworkers that would be diluted whenanalyses are based on the entire groupbeing studied. The only way to observethe risk in the most highly exposedsubset would be to analyze the data bysubgroups of the population.

(viii) Appropriateness of SelectedReference Populations. OSHA also hasbeen criticized for ‘‘ignor[ing] the factthat most of the epidemiologic studiesof butadiene-exposed workers only usedU.S. cancer mortality rates forcomparison to worker mortality.’’ (Ex.113, A–49) The significance of thiscriticism is based on the observation byDowns that ‘‘use of local (mortality)rates (for comparison) tended to bringthe SMRs closer to 100.’’ (Ex. 17–33,p.14) This finding results from cancerrates along the Texas Gulf coast that arehigher than national rates. (Ex. 17–33)In other words, it has been argued thatnational comparison rates artificiallyinflate the SMRs, while local ratesprovide a more accurate picture of themortality experience of workers withoccupational exposure to BD. (Ex. 113,A–50)

Dr. Landrigan captured the essence ofthis issue in his testimony on cross-examination,

This is a perennial debate in epidemiologyof whether to use local comparison rates orregional or national, and there’s [sic]

arguments [to] go both ways. (Tr. 1/15/91, p.154)

He presented several arguments forusing national rates. First, U.S. mortalityrates are based on the entire population,so they are more stable. Second,national rates are more commonly used,so it is easier to compare results fromdifferent studies.

On the other hand, the argument infavor of using local rates centers on thefact that people in a local area may trulybe different from the total population ora regional population(s). Thus,comparing a local subpopulation withthe entire local population may providemore accurate results. However, theweakness in this argument washighlighted by Dr. Landrigan when hesaid that,* * * if there are factors acting in the localpopulation, such as environmental pollutionthat may elevate rates in the local area so thatthey are closer to the rates in theoccupationally exposed population, thentheoretically at least one could argue that thelocal population is overmatched, too similarto the employee population and that the useof the national comparison group actuallygive [sic] a better reflection of reality. (Tr. 1/15/91, p. 155)

In fact, he went on to point out that theBD plants have been identified by theEnvironmental Protection Agency (EPA)as ‘‘major’’ polluters of the localenvironment with BD. (Tr. 1/15/91, p.155)

OSHA acknowledges that there arepros and cons to both approaches ofreference population selection.However, in the study by Delzell et al.mortality data of the USA cohortsubgroup were analyzed using bothstate, i.e., local, general population ratesand USA general population rates. (Ex.117–1) As previously stated, there waslittle difference in the overall pattern ofthese analyses. (Ex. 117–1, p. 60)Additionally, the Santos-Burgoa andMatanoski nested case control studyused the most appropriate comparisongroup of all: Those employed at thesame facilities. (Ex. 23–109 and 34–4,Vol. III, H–4) Thus, given the availabledata in the record, OSHA is of theopinion that it cannot ignore thefindings of excess mortality that arebased on national comparison rates.

(ix) Summary and Conclusions. (a)Summary. Table V–4 lists the criteriathat can be used to judge the presenceof a causal association betweenoccupational exposure to BD and cancerof the lymphohematopoietic system.When the available epidemiologic studyresults are examined in this way, thereis strong evidence for causality. Thedata fulfill all of the listed criteria:Temporal relationship; consistency;


strength of association; dose-responserelationship; specificity of association;biological plausibility; and coherence.

In his testimony, OSHA’sepidemiologist expert witness agreedthat there is ‘‘definite evidence for thefact that occupational exposure to 1,3–Butadiene can cause human cancer ofthe hematopoietic and lymphaticorgans.’’ (Tr. 1/15/91, p. 133) Dr.Weisenburger, OSHA’s expert witnessin hematopathology, also concludedthat ‘‘it would be prudent to treat BD asthough it were a human carcinogen.’’(Ex. 39, p. 11)

TABLE V–4.—EVIDENCE THAT 1,3-BUTADIENE IS A HUMAN CARCINOGEN

Criterion for causality Met byBD

Temporal relationship ...................... Yes.Consistency ..................................... Yes.Strength of association ................... Yes.Dose-response relationship ............ Yes.Specificity of association ................. Yes.Biological plausibility ....................... Yes.Coherence ....................................... Yes.

(b) Conclusion. On the basis of theforegoing analysis, OSHA concludesthat there is strong evidence thatworkplace exposure to BD poses anincreased risk of death from cancers ofthe lymphohematopoietic system. Theepidemiologic findings supplement thefindings from the animal studies thatdemonstrate a dose-response formultiple tumors and particularly forlymphomas in mice exposed to BD.

C. Reproductive Effects

In addition to the establishedcarcinogenic effects of BD exposure,various reports have led to concernabout the potential reproductive anddevelopmental effects of exposure toBD. The term reproductive effects refersto those on the male and femalereproductive systems and the termdevelopmental refers to effects on thedeveloping fetus.

Male reproductive toxicity isgenerally defined as the occurrence ofadverse effects on the male reproductivesystem that may result from exposure tochemical, biological, or physical agents.Toxicity may be expressed as alterationsto the male reproductive organs and/orrelated endocrine system. For example,toxic exposures may interfere withspermatogenesis (the production ofsperm), resulting in adverse effects onnumber, morphology, or function ofsperm. These may adversely affectfertility. Human males produce spermfrom puberty throughout life and thusthe risk of disrupted spermatogenesis is

of concern for the entire adult life of aman.

Female reproductive toxicity isgenerally defined as the occurrence ofadverse effects on the femalereproductive system that may resultfrom exposure to chemical, biological,or physical agents. This includesadverse effects in sexual behavior, onsetof puberty, ovulation, menstrualcycling, fertility, gestation, parturition(delivery of the fetus), lactation orpremature reproductive senescence(aging).

Developmental toxicity is defined asadverse effects on the developingorganism that may result from exposureprior to conception (either parent),during prenatal development, orpostnatally to the time of sexualmaturation. Developmental effectsinduced by exposures prior toconception may occur, for example,when mutations are chemically inducedin sperm. If the mutated sperm fertilizesan egg, adverse developmental effectsmay be manifested in developingfetuses. Mutations may also be inducedin the eggs. The major manifestations ofdevelopmental toxicity include death ofthe developing fetus, structuralabnormality, altered growth andfunction deficiency.

To determine whether an exposurecondition presents a developmental orreproductive hazard, there are twocategories of research studies on whichto rely: Epidemiologic, or studies ofhumans, and toxicologic, orexperimental studies of exposedanimals or other biologic systems.

Many outcomes such as earlyembryonic loss or spontaneous abortionare not easily detectable in humanpopulations. Further, some adverseeffects may be quite rare and requirevery large study populations in order tohave adequate statistical power to detectan effect, if in fact one is present. Often,these populations are not available forstudy. In addition, there are fewerendpoints which may be feasiblymeasured in humans as compared tolaboratory animals. For example, earlyembryonic loss is difficult to measure inthe study of humans, but can bemeasured easily in experimentalanimals. There are no human studiesavailable to address reproductive anddevelopmental effects of BD exposure toworkers. Thus, evidence on thereproductive and developmentaltoxicity of BD comes from toxicologicstudies performed using primarily mice.

Animal studies have proved useful forstudying reproductive/developmentaloutcomes to predict human risk. A veryimportant advantage to the toxicological

approach is the ability of theexperimenter to fully quantitate theexposure concentration and conditionsof exposure. Although extrapolation ofrisk to humans on a qualitative basis isaccepted, quantitative extrapolation ofstudy results is more complex.

In his testimony, OSHA’s witness, Dr.Marvin Legator, an internationallyrecognized genetic toxicologist from theUniversity of Texas Medical Branch inGalveston, cautioned that in assessingrisk ‘‘humans in general have proven tobe far more sensitive than animals* * * to agents characterized asdevelopmental toxicants.’’ (Ex. 72) Healso noted that ‘‘of the 21 agentsconsidered to be direct humandevelopmental toxins, in 19 * * * thehuman has been shown to be moresensitive than the animal * * *’’ Healso pointed to the possibility that sub-groups of the human population may beeven more highly sensitive than thepopulation average.

OSHA believes that the animalinhalation studies designed todetermine the effect of BD on thereproduction and development of theseanimals indicate that BD causes adverseeffects in both the male and femalereproductive systems and producesadverse developmental effects. Thesestudies are briefly summarized anddiscussed below.

Toxicity to Reproductive Organs

In the first NTP bioassay, an increasedincidence of testicular atrophy wasobserved in male mice exposed to BDatmospheric concentrations of 625 ppm.(Ex. 23–1) In female mice, an increasedincidence of ovarian atrophy wasobserved at 625 and 1,250 ppm. Theseadverse effects were confirmed inreports of the second NTP study, whichused lower exposure concentrations.The latter lifetime bioassay exposedmale and female B3C6F1 mice to 0,6.25, 20, 62.5, 200, and 625 ppm BD.(Ex. 114, p 115) See Table V–5.Testicular atrophy in males wassignificantly increased at the highestdose tested, 625 ppm, and reducedtesticular weight was observed from BDexposures of 200 ppm. (Ex. 96) Theselatter data are not shown in the Table.In female mice at terminal sacrifice, 103weeks, ovarian atrophy wassignificantly increased at all exposurelevels including the lowest dose tested,6.25 ppm, compared with controls.Evidence of ovarian toxicity was alsoseen during interim sacrifices, but inthese cases was the result of higherexposure levels. After 65 weeks ofexposure, 90% of the mice exposed to62.5 ppm experienced ovarian atrophy.


TABLE V–5.—OVARIAN AND TESTICULAR ATROPHY IN MICE EXPOSED TO BD

Lesion Weeks ofexposure

Exposure concentration (ppm)

0 6.25 20 62.5 200 625

Incidence (%)

Testicular atrophy ....................... 40 0/10(0) NE NE NE 0/10(0) 6/10(60)65 0/10(0) NE NE NE 0/10(0) 4/7(57)

103 1/50(2) 3/50(6) 4/50(8) 2/48(4) 6/49(12) 53/72(74)Ovarian atrophy .......................... 40 0/10(0) NE NE 0/10(0) 9/10(90) 8/8(100)

65 0/10(0) 0/10(0) 1/10(10) 9/10(90) 7/10(70) 2/2(100)103 4/49(8) 19/49(39) 32/48(67) 42/50(84) 43/50(86) 69/79(87)

NE, not examined microscopically.Source: Ex. 114.

Extensive comments on the BDinduced ovarian atrophy were receivedfrom Dr. Mildred Christian, atoxicologist who offered testimony onbehalf of the Chemical ManufacturersAssociation. She questioned therelevance of using the data from studiesof mice to extrapolate risk of ovarianatrophy to humans because most of theevidence was observed among theanimals who were sacrificed after thecompletion of the species reproductivelife and only after prolonged exposureto 6.25 ppm and 20 ppm (Ex. 118–13,Att 3, p. 4) On the other hand, Drs.Melnick and Huff, toxicologists from theNational Institute of EnvironmentalHealth Sciences stated that: ‘‘Eventhough ovarian atrophy in the 6.25 ppmgroup was not observed until late in thestudy when reproductive senescencelikely pertains, the dose-response dataclearly establish the ovary as a targetorgan of 1,3-butadiene toxicity atconcentrations as low as 6.25 ppm, thelowest concentration studied.’’ (Ex. 114,p. 116) In addition, it should be notedthat an elevated incidence of ovarianatrophy was observed at periods ofinterim sacrifice of female mice exposedto 20 ppm that took place at the 65 weekexposure period, a time prior to the ageswhen senescence would be expected tohave occurred. NIOSH also accepted Dr.Melnick’s view that mice exposed to6.25 ppm BD demonstrated ovarianatrophy. (Ex. 32–35) OSHA remainsconcerned about the ovarian atrophydemonstrated at low exposure levels inthe NTP study. Thus, OSHA concludesthat exposure to relatively low levels ofBD resulted in the induction of ovarianatrophy in mice.

Sperm-Head Morphology StudyNTP/Battelle investigators also

described sperm head morphologyfindings using B6C3F1 mice exposed asdescribed in the dominant lethal studymentioned below, e.g., exposures to 200,1000 and 5000 ppm BD. The mice weresacrificed in the fifth week post-

exposure and examined for gross lesionsof the reproductive system. (Ex. 23–75)The study authors chose this interval ashaving the highest probability fordetecting sperm abnormalities.Epididymal sperm suspensions wereexamined for morphology. Thepercentage of morphologically abnormalsperm heads was significantly increasedin the mice exposed at 1,000 ppm and5,000 ppm, but not for those exposed to200 ppm. The study authors concludedthat ‘‘these significant differences in thepercentage of abnormalities betweencontrol mice and males exposed to 1000and 5000 ppm [BD] indicated that theirlate spermatogonia or earlyspermatocytes were sensitive to thischemical.’’ (Ex. 23–75, p. 16)

In reviewing this study, Dr. MildredChristian stated that these results arenot necessarily correlated withdevelopmental abnormalities or reducedfertility and are ‘‘reversible in nature’’and that the observed differences are‘‘biologically insignificant.’’ (Ex. 76, p.14) In its submission, the Department ofHealth Services of California said: ‘‘Aconclusion as to the reproductiveconsequences of these abnormalitiescannot be made from this study.’’ (Ex.32–168) In reviewing Dr. Christian’scomments, OSHA is in agreement thatthe observation of a significant excess ofsperm head abnormalities as a result ofBD exposure is not necessarilycorrelated with the development ofabnormal fetuses or of reduced fertility;however, the Anderson study, whichdid evaluate fetal abnormality andreduced fertility, demonstrated asignificant excess of both fetalabnormality plus early and late fetalmortality as a result of male miceexposure to BD. (Ex. 117–1, P. 171)These observations of fetal mortalitycould only occur as a result of anadverse effect on the sperm. In responseto Dr. Christian’s comment that thesperm head abnormality observed in thestudy is reversible, the reversibilitywould be dependent upon cessation of

exposure. Since workers may beexposed to BD on a daily basis, thesignificance of reversibility may bemoot.

Developmental Toxicity

Dominant Lethal Studies

A dominant lethal study wasconducted by Battelle/NTP to assess theeffects of a 5-day exposure of male CD–1 mice to BD atmosphericconcentrations of 0, 200, 1,000 and5,000 ppm BD for 6 hours per day onthe reproductive capacity of the exposedmales during an 8-week post-exposureperiod. (Ex 23–74) If present, dominantlethal effects are expressed as either adecrease in the number of implantationsor as an increase in the incidence ofintrauterine death, or both, in femalesmated to exposed males. Dominantlethality is thought to arise from lethalmutations in the germ cell line that aredominantly expressed through mortalityto the offspring. In this study, the onlyevidence of toxicity to the adult malemouse was transient and occurred overa 20 to 30 minute period followingexposure at 5,000 ppm. Males were thenmated to a different female weekly for8 weeks. After 12 days, females werekilled and examined for reproductivestatus. Uteri were examined for number,position and status of implantation.Females mated to the BD-exposed malesduring the first 2 weeks post-exposurewere described as more likely thancontrol animals to have increasednumbers of dead implantations perpregnancy.

For week one, the percentage of deadimplantations in litters sired by malesexposed to 1,000 ppm was significantlyhigher than controls. There were smallerincreases at 200 ppm and 1000 ppm thatwere not statistically significant. Thepercentage of females with two or moredead implantations was significantlyhigher than the control value for allthree exposure groups. For week two,the numbers of dead implantations per


pregnancy in litters sired by malesexposed to 200 ppm and 1000 ppm werealso significantly increased, but not forthose exposed to 5000 ppm. Nosignificant increases in the end pointsevaluated were observed in weeks threeto eight. These results suggested to theauthors that the more mature cells(spermatozoa and spermatids) may beadversely altered by exposure to BD.(Ex. 23–74)

The State of California Department ofHealth Services concluded that theabove mentioned study showed noadverse effect from exposure to BD, withthe possible exception of the increase inintrauterine death seen as a result ofmale exposures to 1000 ppm BD at theend of one week post exposure. (Ex. 32–16) Since values for the 5000 ppmexposure group were not significantlyelevated for this same period of followup, the California Department of Healththought the biological significance ofthe results of the 1000 ppm exposurewas questionable. (Ex. 32–16) On theother hand, Dr. Marvin Legator stressedthe low sensitivity of the dominantlethal assay which, he felt was due tothe endpoint-lethality. He expressed theopinion that the studies were‘‘consistent with an effect on maturegerm cells.’’ (Ex. 72) He felt that sincean effect was observable in thisrelatively insensitive assay that only the‘‘tip of the iceberg’’ was observed, andthat ‘‘[t]ransmissible genetic damage,displaying a spectrum of abnormaloutcomes can be anticipated atconcentrations (of BD) below thoseidentified in the dominant lethal assayprocedure.’’ (Ex. 72, p. 17)

The dominant lethal effect of BDexposure was more recently confirmedby Anderson et al. in 1993. (Ex. 117–1,p. 171) They studied CD–1 mice usinga somewhat modified study design. Twoexposure regimens were used. In thefirst, ‘‘acute study,’’ male mice wereexposed to 0 (n=25), 1250 (n=25), or6250 (n=50) ppm BD for 6 hours only.Five days later they were caged with 2untreated females. One female wasallowed to deliver her litter and theother was killed on day 17 of gestationand examined for the number of livefetuses, number of early and late post-implantation deaths and the numberand type of any gross malformation. Theauthors stated that sacrifice on day 17(rather than the standard days 12through 15) allowed examination ofnear-term embryos for survival andabnormalities. The mean number ofimplants per female was reducedcompared with controls at bothconcentrations of BD, but wasstatistically significant only at 1250ppm. Neither post-implantation loss norfetal abnormalities were significantlyincreased at either concentration. Theauthors concluded that ‘‘a single 6-houracute exposure to butadiene wasinsufficient to elicit a dominant lethaleffect.’’ (Ex. 117–1, p. 171)

In the second phase of the study, the‘‘subchronic study,’’ CD–1 mice wereexposed to 0 (n=25), 12.5 (n=25), or1250 (n=50) ppm BD for 6 hours perday, 5 days per week, for 10 weeks.They were then mated. The higher 1250ppm BD exposure resulted insignificantly reduced numbers ofimplantations and in significantly

increased numbers of dominant lethalmutations expressed as both early andlate deaths. See Table V–6. Non-lethalmutations expressed as birthabnormalities were also observed in livefetuses (3/312; 1 hydrocephaly and 2runts).

The lower exposure (12.5 ppm) didnot result in decreases in the totalnumber of implants, nor in early deaths;however, the frequencies of late deathsand fetal abnormalities (7/282; 3exencephalies in 1 litter and one inanother, two runts and one with bloodin the amniotic sac) were significantlyincreased.

The authors felt that their finding ofincreased late deaths and fetalabnormalities at a subchronic, lowexposure of 12.5 ppm was the main newfinding of the study. They noted thatthese adverse health effects wereincreased 2–3 fold over historicalcontrols. In evaluating these latter twostudies OSHA notes that while therewas no demonstrable effect on dominantlethality as a result of a single exposureto 1250 ppm BD, subchronic exposureto 12.5 ppm, the lowest dose tested,resulted in the induction of dominantlethal mutations and perhaps non-lethalmutations. (Ex 117–1, p 171) OSHA hassome reservations about whether or notthe fetal abnormalities observed in theAnderson et al. ‘‘subchronic’’ studywere actually caused by non-lethalmutations or by some other mechanismbecause they were observed in only afew of the litters produced by the mice.(Ex. 117–1, p. 171)

TABLE V–6.—EFFECT OF BD ON REPRODUCTIVE OUTCOMES IN CD–1 MICE

Implantations Early deaths Late deaths Late deaths includingdead fetuses

Abnormal fetuses

No. Mean No. Mean a No. Mean aNo. Mean a No. Mean a

Control ....... 278 12.09±1.276 13 0.050±0.0597 0 2 0.007±0.0222 012.5 ............ 306 12.75±2.507 16 0.053±0.0581 7 0.23**±0.038 8 0.026±0.0424 b7 0.024*±0.0621250 ppm ... 406 10.68**±3.103 87 0.204***±0161 6 0.014***±0.0324 7 0.016±0.339 c3 0.011**±0.043≤

* Significantly different from control at: *p≤0.05; **p≤0.01; ***p≤0.001 (by analysis of variance and least significance test on arc-sine transformdata).

a Per implantation.b Four exencephalies (three in one litter), two runts (≤70% and 60% of mean body weight of others in litter; total litter sizes 7 and 9, respec-

tively one fetus with blood in amniotic sac but no obvious gross malformation (significance of difference not altered if this fetus is excluded).c One hydrocephaly, two runts (71% and 75% of mean body weight of others in litter; total litter sizes; 2 and 11, respectively).

A dominant lethal test was alsoperformed by Adler et al. (Ex. 126)Male(102/E1XC3H/E1)F1 male micewere exposed to 0 and 1300 ppm BD.They were mated 4 hours after the endof exposure with untreated virginfemales. Females were inspected for thepresence of a vaginal plug everymorning. Plugged females were replaced

by new females. The mating continuedfor four consecutive weeks. Atpregnancy day 14–16 the females werekilled and uterus contents wereevaluated for live and dead implants.Exposure of male mice to 1300 ppm BDcaused an increase of dead implantsduring the first to the third mating weekafter 5 days of exposure. The dead

implantation rate was significantlydifferent from the concurrent controlsonly during the second mating week.Adler et al. concluded that dominantlethal mutations were induced by BD inspermatozoa and late stage spermatidsand that these findings confirmed theresults of the Battelle/NTP study whichshowed effects on the same stages of


sperm development. (Ex. 23–74) Theauthors were of the opinion that BD mayinduce heritable translocations in thesegerm cell stages.

The earliest reproductive studyreported on BD was conducted byCarpenter et al. in 1944. (Ex. 23–64) Inthis study, male and female rats wereexposed by inhalation to 600, 2,300 or6,700 ppm BD, 7.5 hours per day, sixdays per week for an 8-month period.Although this study was not specificallydesigned as a reproductive study, thefertility and the number of progeny wererecorded. No significant effects due toBD exposure were noted for either thenumber of litters per female animal orfor the number of pups per litter.

In the Hazelton study, Sprague-Dawley (SD) rats were exposed byinhalation to 0, 200, 1,000 or 8,000 ppmBD on days 6 though 15 of gestation.(Ex. 2–32) There were dose-relatedeffects on maternal body weight gain,fetal mean weight and crown-to-rumplength. Post-implantation loss wasslightly higher in all BD-exposedgroups. In addition, there weresignificant increases in hematoma inpups in the 200 and 1,000 ppmexposure groups. In the 8,000 ppmexposure group, a significantlyincreased number of pups had lensopacities and there was an increasednumber of opacities per animal.According to the authors, the highestexposure groups also had a significantlyincreased number of fetuses withskeletal variants, a higher incidence ofbipartite thoracic centra, elevatedincidence of incomplete ossification ofthe sternum, higher incidence ofirregular ossification of the ribs, and‘‘other abnormalities of the skull, spine,long bones, and ribs.’’ The authorsconcluded that the fetal response wasnot indicative of a teratogenic effect, butwas the result of maternal toxicity.

In the Battelle/NTP study, pregnantSprague-Dawley (SD) rats and pregnantSwiss mice were exposed to 0, 40, 200,or 1,000 ppm BD for 6 hours per dayfrom day 6 through day 15 of gestation.(Ex. 23–72) Animals were sacrificed andexamined one day before expecteddelivery. In the rat, very little effect wasnoted; in the 1,000 ppm exposure grouponly there was evidence of maternaltoxicity, i.e., depressed body weightgains during the first 5 days of exposure.No evidence of developmental toxicitywas observed in the SD rats evaluatedin the study, e.g., the number of livefetuses per litter and the number ofintrauterine deaths were within normallimits.

In the mouse, exposure to the abovementioned concentrations did not resultin significant maternal toxicity, with the

exception of a reduction in extra-gestational weight gain for the 200 ppmand 1000 ppm BD exposed dams. In thefemale mice, there was a significantdepression of fetal body weight only atthe 200 and 1,000 ppm exposure levels.Fetal body weight for male pups wasreduced at all exposure concentrations,including the 40 ppm exposure level,even though evidence of maternaltoxicity was not observed at thisexposure concentration. No significantdifferences were noted in incidence ofmalformations among the groups.However, the incidence ofsupernumerary ribs and reducedossification of sternebrae wassignificantly increased in litters of miceexposed to 200 and 1,000 ppm BD.

In reviewing these data, Drs. Melnickand Huff noted that since maternal bodyweight gain was reduced at the 200 and1000 ppm exposure levels and bodyweights of male fetuses were reduced atthe 40, 200, and 1000 exposure levels‘‘[t]he male fetus is more susceptiblethan the dam to inhaled 1,3-butadiene.’’(Ex. 114, p. 116) They further stated that‘‘the results of the study in mice revealthat a toxic effect of 1,3-butadiene wasmanifested in the developing organismin the absence of maternal toxicity.’’ Onthe basis of this study, the authorsconcluded that ‘‘1,3-butadiene does notappear to be teratogenic in either the rator the mouse, but there is someindication of fetotoxicity in the mouse.’’(Ex. 23–72)

On the other hand, Dr. MildredChristian was of the opinion that thesignificant decrease in male mouse fetalweight gain in the 40 ppm exposuregroup was not a selective effect of BDon the conceptus, but rather was a resultof the statistical analysis used whichshe considered inappropriate. (Ex. 118–13, Att. 3, p. 6) She was also of theopinion that the larger litter sizes in the40 ppm exposure group as comparedwith the control group contributed tothe statistical finding. Dr. Christian,however, did not present any specificinformation on the type of analysis usedfor statistical testing that she thoughtmade the results inappropriate. Ingeneral, one would expect that theevaluation of data from larger litter sizeswould give one more confidence in thestatistical findings.

In reviewing the same study, the Stateof California, Department of HealthServices was more cautious. It statedthat ‘‘The increased incidence ofreduced ossifications and the fetalweight reductions in the absence ofapparent maternal toxicity in the 40-and 200-ppm groups is evidence offetotoxicity * * * in the Swiss (CD–1)mouse.’’ After reviewing the study

results and arguments about the study,OSHA concluded that the NTP studyprovides evidence of fetotoxicity in themouse. (Ex. 23–72)

Mouse spot testAdler et al. (1994) conducted a spot

test in mice. (Ex. 126) The spot test isan in vivo method for detecting somaticcell mutations. A mutation in amelanoblast is detected as a coat colorspot on the otherwise black fur of theoffspring. Pregnant females wereexposed to 0 or 500 ppm BD for 6 hoursper day on pregnancy days 8, 9, 10, 11and 12. They were allowed to come toterm and to wean their litters. Offspringwere inspected for coat color spots atages 2 and 3 weeks. Gross abnormalitieswere also recorded. Exposure to aconcentration of 500 ppm did not causeany embryotoxicity, nor were grossabnormalities observed. The BDexposure, however, significantlyincreased the frequency of coat colorspots in the offspring. This studydemonstrates that BD exposure iscapable of causing transplacentallyinduced somatic cell mutations that canresult in a teratogenic effect in mice.

Summary of Reproductive andDevelopmental Effect

OSHA has limited its discussion onreproductive and developmentalhazards to a qualitative evaluation of thedata. This approach was chosen becauseno generally accepted mathematicalmodel for estimating reproductive/developmental risk on a quantitativebasis was presented during therulemaking. For example, the CMAButadiene panel disagreed with OSHA’sfindings in the proposal regarding thepotential reproductive anddevelopmental risks presented by BDexposure using an uncertainty factorapproach. (See Ex. 112) They cited Dr.Christian’s conclusion that the mousepossessed a ‘‘special sensitivity’’ to BDand should not be used as a model onwhich to base risk estimates.

The agency has determined, however,that animal studies, taken as a whole,offer persuasive qualitative evidencethat BD exposure can adversely effectreproduction in both male and femalerodents. The Agency also notes that BDis mutagenic in both somatic and germcells. (Ex. 23–71; Ex. 114; Ex. 126)

Some evidence of maternal anddevelopmental toxicity was seen in ratsexposed to BD, but the concentrationsused were much higher than those thatelicited a response in mice. (Ex. 118–13,Att. 3, p. 2) In mice, evidence offetotoxicity was observed in either thepresence or absence of maternaltoxicity, the latter evidence being


provided by decreased fetal body weightin male mice whose dams were exposedto 40 ppm BD, the lowest dose tested inthe study. In addition, a teratogeniceffect was observed in mice (coat colorspot test) as a result of transplacentallyinduced somatic cell mutation.

OSHA is also concerned about theobservation of a significant excess ofsperm head abnormalities as a result ofBD exposure, even though thisexpression of toxicity is not necessarilycorrelated with the development ofabnormal fetuses or of reduced fertility.The Anderson study, which didevaluate reduced fertility and fetalabnormality, demonstrated a significantexcess of both early and late fetalmortality and perhaps fetal abnormalityas a result of male mice exposure to BD.(Ex. 117–1, P. 171) This observationcould only occur as a result of anadverse effect on the sperm. Twoadditional studies also provide evidenceof dominant lethality as a result of maleexposure to BD. (Ex. 23–74; Ex. 126)The observation of germ cell effects issupported by additional evidence ofgenotoxicity in somatic cells, asdemonstrated by positive results in themicronucleus test and in the mouse spottest. (Ex. 126)

Some of the adverse effects related toreproductive and developmentaltoxicity in the mouse, e.g., ovarianatrophy, testicular atrophy, reducedtesticular weight, abnormal spermheads, dominant lethal effects, wereacknowledged by Dr. Christian, but sheurged the Agency not to rely on thesefindings because of negative studyresults in other species, or becausepositive findings in other speciesrequired much higher exposure levels.(Ex. 118–13, Att. 3, p. 1)

For example, a CMA witness hasargued that the diepoxide is responsiblefor the ovarian atrophy observed inrelation to low level BD exposure (6.25ppm). (Ex. 118–13, Att. 3) However, themonoepoxide could also play a role inthe ovarian atrophy and evidenceindicates that humans can form themonoepoxide of BD and that humanshave the enzymes present that couldcause conversion to the diepoxide.Therefore on a qualitative basis, theobservation of ovarian atrophy in themouse is meaningful in OSHA’s view.In addition, the metabolic factors relatedto testicular atrophy, malformed spermand dominant lethal mutations in themouse are not known. (See section onin vitro metabolic studies.) Theseobservations further support thefindings in mice as being meaningful forhumans on a qualitative basis. Themouse spot test which demonstrates asomatic cell mutation leading to a

teratogenic effect inconsistent with datashowing the ability of BD to causeadverse effects on chromosomes andhprt mutations in humans exposed toBD.

OSHA also notes that studies ofworkers exposed to low concentrationsof BD demonstrated a significant excessof chromosomal breakage and aninability to repair DNA damage. Thus,BD exposure seems capable of inducinggenetic damage in humans as a result oflow level exposure. Therefore, themouse studies which demonstrategenetic damage (mutations) in bothsomatic and germinal cells seem to bea better model on a qualitative basisthan the rat for predicting these adverseeffects in humans.

D. Other Relevant Studies

1. Acute Hazards

At very high concentrations, BDproduces narcosis with central nervoussystem depression and respiratoryparalysis. (Ex. 2–11) LC50 values (theconcentration that produces death in 50percent of the animals exposed) werereported to be 122,170 ppm (12.2%v/v) in mice exposed for 2 hours and129,000 ppm (12.9% v/v) in ratsexposed for 4 hours. (Ex. 2–11, 23–91)These concentrations would present anexplosion hazard, thus limiting thelikelihood that humans would risk anysuch exposure except in extremeemergency situations. Oral LD50 values(oral dose that results in death of 50percent of the animals) of 5.5 g/kg bodyweight for rats and 3.2 g/kg body weightfor mice have been reported. (Ex. 23–31)These lethal effects occur at such highdoses that BD would not be considered‘‘toxic’’ for purposes of Appendix A ofOSHA’s Hazard CommunicationStandard (29 CFR 1910.1200), whichdescribes a classification scheme foracute toxicity based on lethality data.

At concentrations somewhat abovethe previous permissible exposure levelof 1,000 ppm, BD is a sensory irritant.Concentrations of several thousand ppmwere reported to cause irritation to theskin, eyes, nose, and throat. (Ex. 23–64,23–94) Two human subjects exposed toBD for 8 hours at 8000 ppm reported eyeirritation, blurred vision, coughing, anddrowsiness. (Ex. 23–64)

2. Systemic Effects

In the preamble to the proposal,OSHA reviewed the literature to discernthe systemic effects of BD exposure. (55FR 32736 at 32755) OSHA discussed anIARC review which briefly examinedseveral studies from the former SovietUnion. In these, various adverse effects,such as hematologic disorders, liver

enlargement and liver and bile-ductdiseases, kidney malfunctions,laryngotracheitis, upper respiratory tractirritation, conjunctivitis, gastritis,various skin disorders and a variety ofneurasthenic symptoms, were ascribedto occupational exposure to BD. (Ex. 23–31) OSHA and IARC have found thesestudies to be of limited use primarilydue to their lack of exposureinformation. Except for sensory irritanteffects and hematologic changes,evidence from studies of other exposedgroups have failed to confirm theseobservations.

Melnick and Huff summarized theobserved non-neoplastic effects of BDexposure in the NTP I and NTP II mousebioassays. They listed the followingeffects associated with exposure ofB6C3F 1 mice to BD for 6 hours per day5 days per week for up to 65 weeks:* * * epithelial hyperplasia of theforestomach, endothelial hyperplasia of theheart, alveolar epithelial hyperplasia,hepatocellular necrosis, testicular atrophy,ovarian atrophy and toxic lesions in nasaltissues (chronic inflammation, fibrosis,osseous and cartilaginous metaplasia, andatrophy of the olfactory epithelium.) (Ex. 114,p. 114)

They noted that the nasal lesions wereseen only in the group of male miceexposed to 1250 ppm BD and that notumors were observed at this site.Further, Melnick and Huff suggestedthat some of the proliferative lesionsobserved in the bioassay mightrepresent pre-neoplastic changes.

The findings of testicular and ovarianatrophy are discussed more fully in theReproductive Effects section of thispreamble,.

Nephropathy, or degeneration of thekidneys, was the most common non-carcinogenic effect reported for malerats in the Hazelton Laboratory Europe(HLE) study in which rats were exposedto 1000 or 8000 ppm BD for 6 hours perday, 5 days per week for up to 2 years.Nephropathy was one of the maincauses of death for the high dose males.(Ex. 2–31, 23–84) The combinedincidence of marked or severenephropathy was significantly elevatedin the high dose group over incidencein the low dose group and overincidence in the controls (p<.001).HLE’s analysis of ‘‘certainly fatal’’nephropathy shows a significant dose-related trend (p<.05), but when‘‘uncertainly fatal’’ cases were included,the trend disappeared.

The HLE study authors concludedthat the interpretation of thenephropathy incidence data wasequivocal. They stated that ‘‘an increasein the prevalence of the more severegrades of nephropathy, a common age-


related change in the kidney, wasconsidered more likely to be asecondary effect associated with otherunknown factors and not to represent adirect cytotoxic effect of the test articleon the kidney.’’

Upon reviewing the HLE rat study forthe proposed rule, OSHA expressedconcern that only 75% of the low-dosemale rats in the HLE study exhibitednephropathy, while 87% of the controlrats had some degree of nephropathy,suggesting low-dose male rats were lesssusceptible to kidney degeneration thancontrol rats, thereby decreasing thecomparability between rats in the low-dose and control groups. (55 FR 32736at 32744) Dr. Robert K. Hinderer, intestifying for the CMA BD Panel,countered that the NTP I mouse studyalso had ‘‘selected instances where theresponse in the test group (was) lowerthan that in the controls’’ and that‘‘* * * (o)ne cannot look at single or afew individual site responses toevaluate the health status or overalleffect of the chemical.’’ (Ex. 51) OSHAagrees that there may be somevariability in background response ratesfor specific outcomes. However, theAgency believes that it is important toassess the impact of the variability inbackground response rates whendrawing conclusions about dose-relatedtrends in the data. This was not done inthe HLE study nephropathy analysis.

Other non-carcinogenic effectsobserved in the HLE rat study wereelevated incidence of metaplasia in thelung of high dose male rats at terminalsacrifice as compared with incidence inmale controls at terminal sacrifice, anda significant increase in high dose malerat kidney, heart, lung, and spleenweights over the organ weights incontrol male rats.

3. Bone Marrow EffectsThere was a single study of BD-

exposed humans discussed in theproposal—a study by Checkoway andWilliams that examined 163 hourlyproduction workers who were employedat the SBR facility studied byMcMichael et al.. (described more fullyin the Epidemiology Section of thisPreamble.) (Ex. 23–4, 2–28).

Exposure to BD, styrene, benzene, andtoluene was measured in all areas of theplant. BD and styrene concentrations, 20(0.5–65) ppm and 13.7 (0.14–53) ppm,respectively, were considerably higherin the Tank Farm than in otherdepartments. In contrast, benzeneexposures, averaging 0.03 ppm, andtoluene concentrations, averaging 0.53ppm, were low in the Tank Farm. Theauthors compared the hematologicprofiles of Tank Farm workers (n=8)

with those of the other workersexamined.

The investigation focused on twopotential effects, bone marrowdepression and cellular immaturity.Bone marrow depression was suspectedif there were lower levels oferythrocytes, hemoglobin, neutrophils,and platelets. Cellular immaturity wassuggested by increases in reticulocyteand neutrophil band form values.

Although the differences were small,adjusted for age and medical status,hematologic parameters in the TankFarm workers differed from those of theother workers. Except for total leukocytecount, the hematologic profiles of theTank Farm workers were consistentwith an indication of bone marrowdepression. The Tank Farm workers alsohad increases in band neutrophils, apossible sign of cellular immaturity, butno evidence that increased destructionof reticulocytes was the cause.

While acknowledging the limitationsof the cross-sectional design of thestudy, the authors felt, nevertheless, thattheir results were ‘‘suggestive ofpossible biological effects, the ultimateclinical consequences of which are notreadily apparent.’’ OSHA finds anyevidence of hematological changes inworkers exposed at BD levels wellbelow the existing permissible limit(1000 ppm) to be of concern since suchinformation suggests the inadequacy ofthe present exposure limit. However,this cross-sectional study involved only8 workers with relatively high levels ofexposure to BD and low levels ofexposure to benzene, so it is quiteinsensitive to minor changes inhematologic parameters.

In a review of BD-related studies,published in 1986, an IARC WorkingGroup felt the study of Checkoway andWilliams could not be consideredindicative of an effect of BD on the bonemarrow (Ex. 2–28). In 1992, IARCconcluded that the ‘‘changes cannot beinterpreted as an effect of 1,3-butadieneon the bone marrow particularly asalcohol intake was not evaluated.’’ (Ex.125, p. 262)

In light of the more recent animalstudies that were not available to IARC,however, OSHA believes that the bonemarrow is a target of BD toxicity.Furthermore, the fact that changes inhematologic parameters could bedistinguished in workers exposed to BDat 20 ppm indicates that suchmeasurements may prove a sensitiveindicator of excessive exposure to BD.

In testimony for the CMA BD Panel,Dr. Michael Bird stated his conclusionthat the hematological differencesbetween the 8 tank farm workers andthe lesser exposed group of workers was

not ‘‘statistically significant by the usualconventional statistics.’’ (Tr. 1/18/1991,p. 1078) He believed that although theraw data were not available, thereported means were within thehistorical and expected range for theseparameters. (Tr. 1/18/1991, p., 1078) Incontrast, OSHA concludes from thisstudy that the hematologic differencesobserved in BD-exposed workers,although small, are suggestive of aneffect of BD on human bone marrowunder occupational exposureconditions.

Thus OSHA considers the Checkowayand Williams study to be suggestive ofhematologic effects in humans, but doesnot regard it as definitive. No otherpotential systemic effects of BDexposure on this population wereaddressed in the Checkoway andWilliams study.

In 1992, Melnick and Huff reviewedthe toxicologic studies of BD exposurein laboratory animals. (Ex. 114) Onlyslight to no systemic effects wereobserved in an early study of rats,guinea pigs, rabbits and a dog exposedto BD up to 6,700 ppm daily for 8months. (Ex. 23–64) The study ofSprague Dawley rats exposed to doses ofBD up to 8,000 ppm daily for 13 weeksalso did not result in hematologic,biochemical, neuromuscular, norurinary effects. However, there weremarked effects seen in exposed mice.

Epidemiologic studies of the styrene-butadiene rubber (SBR) industry suggestthat workers exposed to BD are atincreased risk of developing leukemia orlymphoma, two forms of hematologicmalignancy (see preamble section onepidemiology). Consequently,investigators have looked for evidenceof hematopoietic toxicity resulting fromBD exposure in animals and in workers.For example, Irons and co-workers atCIIT found that exposure of maleB6C3F1 mice to 1,250 ppm of BD for 6–24 weeks resulted in macrocytic-megaloblastic anemia, an increase inerythrocyte micronuclei andleukopenia, principally due toneutropenia. Bone marrow cell typesoverall were not altered, but there wasan increase in the number of cells in thebone marrow of exposed mice due to anincrease in DNA synthesis. (Ex. 23–12)

Melnick and Huff also reviewed theavailable information on bone marrowtoxicity. (Ex. 114, p. 114) Table V–7represents the reported findings of astudy of 10 B6C3F1 mice sacrificed after6.25–625 ppm exposure to BD for 40weeks. The authors concluded thatthese data demonstrated aconcentration-dependent decrease inred blood cell number, hemoglobinconcentration, and packed red cell


volume at BD exposure levels from 62.5to 625 ppm. The effects were notobserved at 6.25 and 20 ppm exposurelevels. Melnick and Kohn also noted theincrease in mean corpuscular volume inmice exposed at 625 ppm, andsuggested that this and otherobservations (such as those of Tice (Ex.

32–38D)) who observed a decrease inthe number of dividing cells in miceand decreased rate of their division),suggested that BD exposure led to asuppression of hematopoiesis in bonemarrow. Melnick and Huff concludedthat this, in turn, led to release of largeimmature cells from sites such as the

spleen, which was consideredindicative of macrocytic megaloblasticanemia by Irons. They concluded thatthese findings ‘‘(establish) the bonemarrow as a target of 1,3-butadienetoxicity in mice.’’ (Ex. 114, p. 115)

TABLE I.—HEMATOLOGIC CHANGES IN MALE B6C3F1 MICE EXPOSED FOR 6 HOURS/DAY, 5 DAYS/WEEK FOR 40 WEEKS

BD exposure (ppm)Red blood cell

count(×10 6/ul)

Hemoglobinconc.(g/dl)

Volumepacked RBC

(ml/dl)

Mean corpus-cular vol

0 ........................................................................................................................ 10.4±0.3 16.5±0.4 48.1±1.5 46.3±0.86.25 ................................................................................................................... 10.3±0.3 16.4±0.5 47.8±1.7 46.4±1.020 ...................................................................................................................... 10.4±0.4 16.7±0.7 48.2±2.2 46.3±0.862.5 ................................................................................................................... a 9.9±0.4 a 15.9±0.6 a 45.9±2.1 46.7±1.2200 .................................................................................................................... a 9.6±0.5 a 15.6±0.9 a 45.4±2.7 47.2±1.0625 .................................................................................................................... a 7.6±1.2 a 13.5±1.8 a 39.9±5.3 a 53.2±2.9

Adapted from Melnick and Huff, Exhibit 114.a Different from chamber control (0 ppm), P<0.05. Results of treated groups were compared to those of control groups using Dunnett’s t-test.

4. Mutagenicity and Other GenotoxicEffects

OSHA discussed the genotoxic effectsof BD exposure in some detail in theproposal. (55 FR 32736 at 32760)Briefly, BD is mutagenic to Salmonellatyphimurium strains TA 1530 and TA1535 when activated with S9 liverfraction of Wistar rats treated withphenobarbital or Arochlor 1254. Thesebacterial strains are sensitive to base-pair substitution mutagens. Since theliver fraction is required to elicit thepositive mutagenic response, BD is nota direct-acting mutagen and likely mustbe metabolized to an active form beforebecoming mutagenic in this test system.IARC published an extensive list of‘‘genetic and related effects of 1,3-butadiene.’’ (Ex. 125) They noted insummarizing the data that BD wasnegative in tests for somatic mutationand recombination in Drosophila, andthat neither mouse nor rat liver fromanimals exposed to 10,000 ppm BDshowed evidence of unscheduled DNAsynthesis.

As OSHA described in the proposedrule, and Tice et al. reported in 1987,BD is a potent in vivo genotoxic agentin mouse bone marrow cells thatinduced chromosomal aberrations andsister chromatid exchange in marrowcells and micronuclei in peripheral redblood cells. (55 FR 52736 at 52760)Some of these effects were evident atexposures as low as 6.25 ppm (6 hours/day, 10 days). However, similar effectswere not observed in rat cells exposedto higher levels of BD (10,000 ppm for2 days).

Sister chromatid exchange is arecombinational event in which nucleicacid is exchanged between the two

sister chromatids in each chromosome.It is thought to result from breaks ornicks in the DNA. Irons et al. describedmicronuclei as ‘‘* * * chromosomefragments or chromosomes remaining asthe result of non-dysjunctional event.Their presence in the circulation isfrequently associated with megaloblasticanemia.’’ (Ex. 23–12).

In a subsequent study, Filser and Boltexposed B6C3F1 mice to the same 3concentrations of BD, 6.25, 62.5 or 625for 6 hours/day, 5 days/week, for 13weeks. (Ex. 23–10) Peripheral bloodsamples were taken from 10 animals pergroup and scored for polychromaticerythrocytes (PCE) and micronucleatednormochromatic erythrocytes (MN–NCE). The MN–NCE response, whichreflects an accumulated response, wassignificantly increased in both sexes atall concentrations of BD, including 6.25ppm.

Certain metabolites of BD alsoproduce genotoxic effects. These aredetailed in a number of reviews (see forexample, Ex. 114, 125). Briefly,epoxybutene (the monoepoxide) ismutagenic in bacterial systems in theabsence of exogenous metabolicactivation. Epoxybutene also reacts withDNA, producing two structurallyidentical adducts and has been shownto induce sister chromatid exchanges inChinese hamster ovary cells and inmouse bone marrow in vivo.

IARC in its review concluded that thediepoxide, 1,2,:3,4-diepoxybutane,induced DNA crosslinks in mousehepatocytes and, like epoxybutene, ismutagenic without metabolic activation.As discussed below, BD diepoxide alsoinduced SCE and chromosomalaberrations in cultured cells.

A human cross-sectional studyinvolving a limited number of workersin a Texas BD plant indicated genotoxiceffects. (Ex. 118–2D) Peripherallymphocytes were cultured from 10non-smoking workers and from age- andgender-matched controls who worked inan area of very low BD exposure (0.03ppm). Production areas in the plant hada mean exposure of 3.5 ppm BD, withmost exposed workers in this sampleexperiencing exposure of approximately1 ppm BD.

Standard assays for chromosomalaberrations and a gamma irradiationchallenge assay that was designed todetect DNA repair deficiencies wereperformed. The results of the standardassay indicated that the exposed grouphad a higher frequency of cells withchromosome aberrations and higherchromatid breaks compared with thecontrol group. This difference was notstatistically significant. In the challengeassay, the exposed group had astatistically significant increasedfrequency of aberrant cells, chromatidbreaks, dicentrics (chromosomes having2 centromeres) and a marginallysignificant higher frequency ofchromosomal deletions than controls.Au and co-workers concluded that cellsexposed to BD are likely to have moredifficulty in repairing radiation induceddamage. (Ex. 118–2D)

To determine the mutagenic potentialof both BD and its three metaboliteepoxides, Cochrane and Skopek studiedeffects in human lymphoblastoid cells(TK6) and in splenic T cells fromexposed B6C3F1 mice. (Ex. 117–2, p.195) TK6 cells were exposed for 24hours to epoxybutene (0–400 uM), 3,4-epoxy-1,2-butanediol (0–800 uM), ordiepoxybutane (0–6 uM). All


metabolites were mutagenic at both thehprt (hypoxanthine-guaninephosphoribosyl transferase) and tk(thymidine kinase) loci, withdiepoxybutane being active atconcentrations 100 times lower thanepoxybutane or epoxybutanediol.

They also studied mice exposed to625 ppm BD for 2 weeks and found a3-fold increase in hprt mutationfrequency in splenic T cells comparedwith controls. They also intended togive daily IP doses of epoxybutene (60,80 or 100 mg/kg) or diepoxybutane (7,14, or 21 mg/kg) every other day forthree days. However, only animals giventhe lowest dose of the diepoxidereceived three doses because oflethality. After two weeks of expressiontime, cells were isolated fordetermination of mutation frequency.Both exposure regimens resulted inincreased mutation frequency. Forexample, at the highest exposure toepoxybutene, the average mutationfrequency was 8.6×106, while thediepoxide exposed group had afrequency of 13×106, compared to acontrol mutation frequency of 1.2×106.

Cochrane and Skopek useddenaturing gradient gel electrophoresisto study the nature of the splenic T cellhprt mutants in the DNA. They foundabout half were frameshift mutations. Apotential ‘‘hotspot’’ was also describedin which a plus one (+1) frameshiftmutation in a run of six guanine baseswas observed in four BD-exposed mice,in four expoxybutene-exposed mice andin two mice exposed to the diepoxide.They observed both G:C and A:T basepair substitutions in the epoxide treatedgroup; however, similar to the findingsof Recio, et al. (described below), A:Tsubstitutions were observed only in theBD-treated group. The authors offeredno hypothesis for this observation.These researchers also noted asignificant correlation of dicentrics withthe presence of a BD metabolite, (1,2-dihydroxy-4-(N-acetyl-cysteinyl-S)butane) in the urine of exposedworkers. They further concluded that:

This study indicates that the workers hadexposure-induced mutagenic effects.Together with the observation of genemutation in a subset of the population, thisstudy indicates that the current occupationalexposure to butadiene may not be safe toworkers. (Ex. 118–2D)

An abstract by Hallberg submitted tothe Environmental Mutagenesis Societydescribes a host-cell reaction assay inwhich lymphocytes transfected with aplasmid with an inactivechloramphenicol acetyl transferase(CAT) reporter gene were challenged torepair the damaged plasmid andreactivate the CAT gene. No effect was

noted among cells of workers exposed to0.3 ppm benzene; however, BD-exposedworkers (mean exposure 3 ppm) hadsignificantly reduced DNA repaircapacity (p=0.001). The authorsbelieved that this finding confirmed theDNA repair defect due to BD exposureobserved in the Au et al. study’schallenge assay. (Ex. 118–2D)

Ward and co-workers reported theresults of a preliminary study todetermine whether a biomarker for BDexposure and a biomarker for thegenetic effect of BD exposure could bedetected in BD-exposed workers. (Ex.118–12A) The biomarker for exposurewas excretion of a urinary metabolite ofBD, (1,2-dihydroxy-4-(n-acetylcysteinyl-S)butane. The geneticbiomarker was the frequency oflymphocytes containing mutations atthe hypoxanthine-guaninephosphoribosyl transferase (hprt) locus.Study subjects included 20 subjectsfrom a BD production plant and 9 fromthe authors’ university; all were verifiednon-smokers. Seven workers were inareas or at jobs that were ‘‘consideredlikely to expose them to higher levels ofbutadiene than in other parts of theplant.’’ Ten worked in areas where thelikelihood of BD exposure was low.Three ‘‘variable’’ employees worked inboth types of jobs or areas. hprt assaysof 6 of the 7 high exposure group and5 of the 6 non-exposed groups werecompleted at the time of the report. Airsampling was used to estimateexposure. In the production area, themean was approximately 3.5 ppm, withmost samples below 1 ppm. In thecentral control area (lower exposure) themean was 0.03 ppm. The frequency ofmutant lymphocytes in the high-exposure group compared with eitherthe low- or no-exposure group wassignificantly increased. The low- andnon-exposed groups were notsignificantly different from each other inmutant frequencies.

Similarly, the concentration of the BDmetabolite in urine was significantlygreater in the high exposure group thanin the lower- or non-exposed groups.There was a strong correlation amongexposed subjects between the level ofmetabolite in urine and the frequency ofthe hprt mutants (r=0.85). (Ex. 118–2A)

Another study of humans for potentialcytogenetic effects of BD exposure wasreported recently by Sorsa et al. inwhich peripheral blood was drawn from40 BD production facility workers andfrom 30 controls chosen from otherdepartments of the same plants, roughlymatched for age and smoking habits.(Ex. 124) Chromosome aberrations,micronuclei and sister-chromatidexchanges were analyzed. No exposure

related effects were seen in any of thecytogenetic endpoints. The typicalexposure was reported as less than 3ppm. (Ex. 124)

Among the limited number of humanstudies involving BD exposed workers isthat of Osterman-Golker who evaluatedpost-exposure adduct formation in thehemoglobin of mice, rats, and a smallnumber of workers. (Ex. 117–2, p. 127)Mice and rats were exposed at 0, 2, 10,or 100 ppm for 6 hours per day, 5 daysper week for 4 weeks and their bloodtested for the presence and quantity ofthe BD metabolite, 1,2-epoxybutene,forming an adduct with the N-terminalvaline of hemoglobin. The result was alinear response for mice at 2, 10 and 100ppm; and, for rats at 2 and 10 ppm, withthe 100 ppm dose group deviating fromlinearity. In addition, while the adductlevel per gram of globin in the 100 ppmrats was about 4 times lower than thelevel observed in mice exposed to 100ppm BD, at lower exposures, the adductlevels were similar.

In the portion of the study dealingwith effects on humans, blood wastaken from four workers in two areas ofa chemical production plant withknown BD exposure, and five workersfrom two non-production areas whereBD concentrations were low. In thehigher exposure area, the mean BDexposure was about 3.5 ppm, asdetermined by environmental sampling.The lower exposure areas had a meanBD level of about 0.03 ppm. On a moleof adduct per gram of hemoglobin level,the adduct levels in the higher BDexposed workers were 70 to 100 timeslower than those of either the rat ormice exposed at the 2 ppm leveldiscussed above. Production workershad adduct levels ranging from 1.1 to2.6 pmol/g globin. Most controls in thestudy were below the level of detectionof the assay (0.5 pmol adduct/ g globin).(Two heavy smokers reported from aprevious study had higher adduct levelsthan non-smokers; their levelsapproached those observed in BDexposed workers and were consistentwith the amount of BD in mainstreamsmoke.)

Similar results for mice and ratsexposed to BD were reported byAlbrecht et al. (Ex. 117–2, p. 135) In thisstudy which exposed the rodents to 0,50, 200, 500 or 1300 ppm for 6 hours/day, for 5 consecutive days, BDmonoepoxide adduct levels in thehemoglobin of mice were about fivetimes that of the rat at most BD exposureconcentrations. Humans were notstudied in this report.

Another observation pertaining tohuman cytogenetics with potentiallyimportant implications for BD-induced


1 For example, in the 58 newspaper workerstested, 24% had greater than 95 SCE/cell, while theremaining 76% had fewer than 80 SCE/cell.

2 Cytochrome is defined as any of a class ofhemoproteins whose principal biologic function iselectron transport by virtue of a reversible valencychange of its heme iron. Cytochromes are widelydistributed in animal and plant tissues.

human disease is contained in a reportby Wiencke and Kelsey. (Ex. 117–2, p.265) These researchers studied theimpact of the BD metabolite,diepoxybutane, exposure on sisterchromatid exchange (SCE) frequenciesin several groups of human blood cellcultures (n=173 healthy workers). Theydiscovered that the study populationswere bimodally distributed according totheir sensitivity to induction of SCEswhen cell cultures were exposed to 6uM diepoxybutane. Wiencke and Kelseyreported that they had observed inearlier studies that ‘‘genetic deficiencyof glutathione S-transferase type u leadsto bimodal induction of SCEs byepoxide substrates of the isozyme’’ andthat cells from individuals with thedeficiency had SCE induction scoresthat were significantly higher than thoseobserved in the general population. (Ex.117–2, p. 271) Approximately 20% ofthe tested groups were sensitive toinduction of SCE and the remaining80% were relatively insensitive.1Subsequent testing indicated that thesensitive population was also sensitiveto induction of chromosomalaberrations by diepoxybutane withsignificant increases in the frequenciesof chromatid deletions, isochromatiddeletions, chromatid exchanges andtotal aberrations. The relevance of thesefindings in not yet clear; however, theymay indicate that certain subsets of thepopulation are more highly susceptibleto the effects of this mutagenicmetabolite of BD.

Recio et al. used transgenic micecontaining a shuttle vector with arecoverable lac 1 gene to study in vivomutagenicity of BD and the spectrum ofmutations produced in various tissues.(Ex. 118–7D) Mice were exposed to 62.5,625 or 1250 ppm BD for 4 weeks (5days/week, 6 hours/day). Theinvestigators extracted DNA from bonemarrow and determined mutagenicity atthe lac 1 transgene.

The mutant DNA was sequenced.Dose-dependent mutagenicity—up to a3-fold increase over air controls—wasobserved among mice exposed at 625 or1250 ppm. Although a number ofdifferences in patterns were noted, themost striking was that sequence analysisindicated an increased frequency of invivo point mutations induced by BDexposure at adenine and thymine (A:T)base pairs following inhalation.

In further studies of BD-exposedtransgenic mice, Sisk and co-workersexposed male B6C3F1 mice to 0, 62.5,625, or 1250 ppm, BD for 4 weeks (6

hour/day, 6 days/week). (Ex. 118–7Q)Bone marrow cells were isolated andmutation frequency and spectrumevaluated. Lac 1 mutation frequencieswere significantly increased at all 3exposure levels and were dose-responsive in the 62.5 and 625 ppm BD-exposed mice, compared to controls. Aplateau in mutation frequencies wasobserved at 1250 ppm BD-exposed mice,perhaps indicating saturation or mutantloss due to the effects of high levelexposure.

When the mutants were sequenced,several from the same animal werefound to have identical mutations.Although they might have arisenindependently, Sisk et al. felt that thiswas likely due to clonal expansion of abone marrow cell with a mutated lac 1gene.

As had Recio et al., Sisk et al.observed a higher frequency ofmutations at A:T sites in the exposedmice DNA, compared with controls. A:Tto G:C transitions comprised only 2% ofthe background mutations, but made up15% of those in the exposed mice.

Sisk et al. concluded that theirobservation coupled with in vitrostudies ‘‘ * * * suggest that BD maymutate hematopoietic stem cells.’’ (Ex.118–7Q, p. 476)

As discussed in the animalcarcinogenicity section in this preamble,BD-induced mouse tumors have beenfound to have activated proto-oncogenes. Specifically, the K-rasoncogene is activated and is the mostcommonly detected oncogene inhumans. (Ex. 129)

OSHA concludes that BD ismutagenic in a host of tests which showpoint and frameshift mutations, hprtmutations, chromosome breakage, andSCEs in both animals and humans. Thedata suggest that mice are moresusceptible than rats to these alterations.In addition, certain subsets of thehuman population may be moresusceptible to the effects of BD exposurethan others (based on the Wiencke andKelsey study of human blood cellcultures, Ex. 117–2, p. 265). OSHAfurther notes with concern the fact thatthe data suggest that BD exposure atrelatively low levels adversely affectsDNA repair mechanisms in humans andis associated with mutational effects.

5. MetabolismIn vitro genotoxicity studies have

shown that BD is mutagenic only afterit is metabolically activated.Biotransformation is probably alsoimportant to the carcinogenicity of thisgas. It is thought that the formation ofepoxides, specifically epoxybutene, alsotermed the ‘‘monoepoxide’’ and 1,2:3,4-

diepoxybutane, termed the ‘‘diepoxide,’’is required for activity and that thereaction is cytochrome P450 mediated 2.Both the mono- and diepoxide aremutagenic in the Salmonella assay, withthe diepoxide being more active. Thereactive epoxides can bind to DNA, andformation of DNA adducts ishypothesized to initiate a series ofevents leading to malignancy.

As described earlier, for most cancersites, mice are more sensitive than ratsto the carcinogenic effects of BDexposure. Studies of the metabolism ofBD have been undertaken in an attemptto elucidate the contributions of dose-metric factors for the observeddifferences in carcinogenicity betweenthe species.

Much of the research in this area hasbeen performed at the ChemicalIndustry Institute of Toxicology and inGerman laboratories. Work onmetabolism of BD was described byOSHA in the 1990 proposal. (55 FR32736 at 32756) OSHA reviewed thecurrent literature in the record andconcluded:

1. The rate of metabolism of BD inmice is approximately twice that in rats;

2. Mice accumulate moreradiolabelled BD equivalents in a 6 hourexposure than do rats at the sameconcentration;

3. Mice have about twice theconcentration of the metabolite (1,2-epoxy-3-butene) (BMO) in blood as ratsexposed at similar concentrations;

4. Over a wide range of exposures,mice received a larger amount ofinhaled BD per unit body weight thanrats, and had a higher concentration ofBMO in the blood than rats (Asexpected, because of body sizedifferences and breathing rates, andsome enzymology);

5. BD is readily absorbed and widelydistributed in tissues of both mice andrats, with tissue concentrations perumole BD inhaled higher in mice thanin rats, by factors of 15-fold or more;

6. While there are species differencesin the amount of BD metabolism atvarious sites, both mice and ratsmetabolize BD to the same reactivemetabolites suspected of being ultimatecarcinogens.

In comments on OSHA’s proposal, Dr.Michael Bird of Exxon testified onbehalf of the CMA BD Task Group thatthe mouse ‘‘will attain a significantlyhigher amount of the epoxides over alonger period of time than the rat. . . orprimate when exposed to butadiene.’’


3 A microsome is defined as one of the finelygranular elements of protoplasm, resulting fromfragmentation (homogenization) of the endoplasmicreticulum.

(Ex. 52, p. 27) Dr. Bird concluded thatthe differences in metabolism of BD inthe species help ‘‘explain the greatersensitivity of the mouse to BDcarcinogenic activity.’’ He furtherconcluded that the differences in ratesof enzyme mediated processes indicatenon-human primates have lowerinternal concentrations of BD or BMO,and ‘‘man is more similar to the primatewith respect to 1,2-epoxy-3-buteneformation than the rat or mouse.’’ (Ex.52, p. 22) He argued that the mouse maybe ‘‘uniquely sensitive ‘‘ to BDcarcinogenicity due to its greater uptake,faster BD metabolism and ‘‘eliminationof the epoxide 1,2-epoxy-3-butene issaturable in mice but not in rats.’’ (Ex.52, p. 21) He felt this observationcorrelated well with the observedcytogenetic and bone marrow response(seen in mouse, but not rats.)

Others hold an opposing view, e.g.,Melnick and Kohn argued that‘‘[b]ecause the rat appears to beexceptionally insensitive to leukemia/lymphoma induction, the mouse mustbe considered as the more appropriatemodel for assessing human risk forlymphatic and hematopoietic cancers.’’(Ex. 130, p. 160)

Dr. Bird urged OSHA to use themonkey data of Dahl, et al. whichindicated that the retention rate for BDin primates is over 6 times lower thanthat for the mouse, in ‘‘drawing any firmconclusions about the cancer risk tohumans.’’ (Ex. 52, p. 36) During thepublic hearing, the work of Dahl waspresented as a preliminary report. (Ex.44) Dahl exposed 3 cynomolgusmonkeys to BD and measured uptakeand metabolism. Each animal wasexposed to three concentrations of C14-labeled BD, progressing from 10,300 to8000 ppm with at least 3 monthsseparating the re-exposure of eachmonkey. Post-exposure blood was taken.Each animal’s breathing frequency andtidal volume was measured.

Dahl and co-workers found BD uptaketo be lower in monkeys than in rats. Thereported blood levels of the epoxideswere also lower in the monkey than thelevels reported by Bond et al. in rats andmice.

Dahl et al. attempted to quantitatetotal BD metabolites through collectionof feces, urine and exhaled materialthough use of cryogenic traps.Measurement of residual labeledmaterial retained in the animals at theend of the 96 hour post exposure periodwas not determined. HPLC (high-performance liquid chromatography)identification of the trapped material (at95 C) indicated that only 5 to 15% of theradioactivity was present asmonoepoxide.

Melnick and Huff, in reviewing thisstudy, found its significance ‘‘clouded’’because only three animals of unknownage were studied and there wasuncertainty about the ability of vacuumline cryogenic distillation alone toidentify and quantitate BD metabolites.(Ex. 114, p. 133) In testimony at thepublic hearing, Dr. James Bond of CIITacknowledged the limitations of the useof vacuum-line cryogenic distillation asfollows:

* * * there will be some material nomatter what kind of vacuum you apply toit * * * simply will not move into the traps.That’s referred to as non-volatile material.

We don’t know what that material is andI think that’s an important component of thisstudy, because, in fact, in many cases it canrepresent 70 to 80 percent of the material thatactually distills out. (Tr. 1/22/91, p. 1553)

Melnick and Huff were alsoconcerned that only the monkeys, notthe mice or rats, were anesthetizedduring exposure and question whatimpact that might have had onrespiratory rates and cardiac output andwhat the influence might be oninhalation pharmacokinetics of BD. (Ex.114, p. 133) In their 1992 review,Melnick and Huff concluded thatstudies to date have not revealed speciespharmacokinetic differences ofsufficient magnitude ‘‘to account for thereported different toxic or carcinogenicresponses in one strain of rats comparedto two strains of mice.’’ (Ex. 114, p. 134)In post hearing comments Dr. David A.Dankovic of NIOSH reviewed this topicand concluded ‘‘* * * the mostprudent course is to base 1,3-butadienerisk assessments on the externalexposure concentration, unlesssubstantial improvements are made inthe methodology used for obtaining‘internal’ dose estimates.’’ (Ex. 101, Att.2, p. 5)

Recent StudiesRecent studies have focused on the

metabolism of BD to the epoxides,epoxybutene and diepoxybutane, andtheir detoxification by epoxidehydrolase and glutathione. Bond et al.recently reviewed BD toxicologic data.(Ex. 118–7G) Epoxybutene anddiepoxybutane were reported to becarcinogenic to mice and rats via skinapplication and/or subcutaneousinjection, with the diepoxide havingmore carcinogenic potency. Bond et al.also concluded that the diepoxide ismore mutagenic than the monoepoxideby a factor of nearly 100 on a molarbasis. The diepoxide also inducesgenetic damage in vitro mammaliancells (Chinese hamster ovary cells andhuman peripheral blood lymphocytes).These studies are summarized in this

preamble discussion of reproductiveeffects.

In vitro metabolic studies

In 1992 Csanady et al. reported use ofmicrosomal and cytosolic preparationsfrom livers and lungs of Sprague-Dawley rats, B6C3F1 mice and humansto examine cytochrome P450-dependentmetabolism of BD. (Ex. 118–7AA) Thepreparations were placed in sealed vialsand BD was injected by use of a gas-tight syringe. Air samples were takenfrom the head space at 5 minuteintervals and analyzed by gaschromatography for epoxybutene.

Cytochrome P450-dependentmetabolism of the monoepoxide to thediepoxide was examined. Enzymemediated hydrolysis of BMO by epoxidehydrolase was measured. (Non-enzymemediated hydrolysis was determinedusing heat-inactivated tissue and nonewas observed.) Second order rateconstants were determined using 100mM monoepoxide and 10 mM GSH. Thehuman samples were quite variable,with rates ranging from 14 to 98 nmol/min/mg protein.

The maximum rates for BD oxidationto monoepoxide (Vmax) weredetermined to be highest for mouse livermicrosomes 3 (2.6 nmol/mg protein/min); the Vmax values for humans wereintermediate, at 1.2 nmol/mg protein/min; the Vmax values for rats was 0.6nmol/mg protein/min. For lungmicrosomes, the Vmax in the mousewas found to be similar to the mouseliver rate, but over 10-fold greater thanthat of either humans or rats.

From these data Csanady et al.calculated a ratio of activation todetoxification for each species tested.These values, expressed as mg cytosolicprotein/gm liver [glutathione-S-transferase is a cytosolic enzyme],resulted in the determination of anoverall activation:detoxification ratio of12.3 for the mouse, 1.3 for the rat, and4.4 for the human samples.

If these in vitro liver microsomalstudies can be extrapolated to the wholeanimal in vivo, then this implies, aspointed out by Kohn and Melnick, thatthe mouse produces 2.8 times as muchBMO per mol of BD as the human andthat the human activation:detoxificationratio is 3.4 times that of the rat.However, the Csanady et al. studydemonstrated a wide variability in BDmetabolic activity among the 3 humanliver microsomes, and a 60-foldvariation was found in 10 human liver


samples by Seaton et al. (Ex. 118–7N)Kohn and Melnick noted that thishuman variability in CYP2E1, the P450enzyme primarily responsible for theactivity, suggests that a ‘‘* * * fractionof the human population may be assensitive to butadiene as mice are.’’ (Ex.131, p. 620).

A study similar to that of Csanady etal., reported by Duescher and Elfarra in1994, determined that the Vmax/Kmratios for BD metabolism in human andmouse liver microsome were similarand were nearly 3 to 3.5 fold higherthan the ratio obtained with rat livermicrosomes. (Ex. 128) Duescher andElfarra suggest that differences betweentheir results and those of Csanady et al.may have been due in part toexperimental methodology differences,such as incubation and assay methods.Duescher and Elfarra found that twoP450 isozymes, 2A6 and 2EI, were mostactive in forming BMO of the 7isozymes tested. They concluded thatsince human liver microsomes oxidizedBD at least as efficiently as mouse livermicrosomes (and much more so than ratliver microsomes), this ‘‘suggests that if[BMO] formation rate is the primaryfactor which leads to toxicity, humansmay be at higher risk of expressing BDtoxicity than mice or rats, and that themouse may be the more appropriateanimal model for assessing toxicity.’’Duescher and Elfarra felt that sinceP450/2A6 appears to play a major rolein BD oxidation in human livermicrosomes, and that it is more similarto that of mouse P450/2A5 than to ratP450/2A1, the mouse may be a bettermodel to use in assessing human risk.

In 1994 Himmelstein et al.hypothesized that ‘‘[S]pecies differencesin metabolic activation anddetoxification most likely contribute tothe difference in carcinogenic potencyof BD by modulating the circulatingblood levels of the epoxides.’’ (Ex. 118–13, Att 3) To address this, Himmelsteinand colleagues looked at the levels ofBD, BMO, and BDE in blood of rats andmice exposed at 62.5, 625, or 1250 ppmBD. Samples were collected at 2, 3, 4,and 6 hours of exposure for BD andBMO and at 3 and 6 h for the BDE.Blood was collected from mice bycardiac puncture and from rats throughan in-dwelling jugular cannula. Melnickand Huff criticized earlier studies whichfailed to use in-dwelling cannulae.

Because steady state levels of[monoepoxide] are lower in rats than in miceand because the metabolic elimination ratefor this compound is 5 times faster in ratsthan in mice, any delay in obtainingimmediate blood samples would have amuch greater effect on analyses in blood

samples obtained from rats than thoseobtained from mice. (Ex. 114, p. 133)

Himmelstein et al. found that theconcentration of BD in blood was notdirectly proportional to the inhaledconcentration of BD, suggesting that theuptake of BD was saturable at thehighest inhaled concentration. In bothrats and mice BD and the BMO bloodlevels were at steady state at 2, 3, 4 and6 hours of exposure and declinedrapidly when exposure ceased. This isconsistent with exhalation being theprimary route of elimination of BD. (Ex.118–7B)

Genter and Recio used Western blotand immunohistochemical analyses todetect P450/2E1 in bone marrow ofB6C3F1 mice. (Ex. 118–7T) Althoughboth methods detected the presence ofthe protein in livers of both male andfemale mice, non was seen in the bonemarrow. The limits of detection werenot stated in the report. The authorhypothesized the BD might be convertedto the monoepoxide in the liver prior touptake by the bone marrow or thatanother pathway (e.g., myeloperoxidase)is responsible for BD oxidation in themarrow. Recio and Genter suggest thatthe greater sensitivity of mice to BD-induced carcinogenicity can beexplained in part by the higher levels ofboth epoxides in the blood of micecompared with that of rats.

Himmelstein et al. furthered this workin 1995 in a report in which theydetermined levels of the epoxides inlivers and lungs of mice and ratsexposed to BD. (Ex. 118–7/O) Animalswere exposed at 625 or 1250 ppm of BDfor 3 or 6 hours. Himmelstein et al.found that in mice exposed to thisregimen, the monoepoxide levels werehigher in lungs than in livers. Rats at625 and 1250 ppm had lowerconcentrations of BMO in lungs andlivers than mice. When rats wereexposed to 8000 ppm BD, the maximumconcentration of BMO in the lung andliver was nearly the same. Thediepoxide levels in lungs of miceexposed at 625 and 1250 ppm were 0.71and 1.5 nmol/g respectively. Thediepoxide was not detected in livers orlungs of rats exposed at any tested level.

Himmelstein et al. also observeddepletion of glutathione in liver andlung samples from both rodent species.Following 6 hours of exposure, thelungs of mice exhibited greaterdepletion of GSH than mouse liver, ratliver or rat lung at all concentrations ofBD tested. The conclusion reached bythe study authors was that their dataindicate that GSH depletion isassociated with tissue burden of theepoxides and that this target organ

dosimetry might help explain some ofthe non-concordance of cancer sitesobserved between the species. OSHAnotes, however, that while % GSHdepletion was highest in the mouselung, the major increase in depletionwas at 1250 ppm BD, while lung tumorincidence was increased in the femalemice at 6.25 ppm and in male mice at62.5 ppm. Depletion of glutathione wasdependent on concentration andduration of BD exposure.

Himmelstein et al. stressed theimportance of the fact that thediepoxide was detected in the mouselung but was not quantifiable in themouse liver, and stated that if thediepoxide was formed in the liver, it israpidly detoxified or otherwise movedout of the liver. They also found thatdepletion of glutathione was greater inmouse than rat tissues for similarinhaled concentrations of BD andconcluded that conjugation of themonoepoxide with glutathione byglutathione S-transferase is an importantdetoxification step.

In contrast to rats and mice, lungs andlivers from humans had much fasterrates of microsomal monoepoxidehydrolysis by epoxide hydrolasecompared to cytosolic conjugation withglutathione by the transferase. (Ex. 118–7AA)

Thornton-Manning et al. in 1995examined the production anddisposition of monoepoxide anddiepoxide in tissues of rats and miceexposed at 62.5 ppm BD. (Ex. 118–13,Att. 3) They found monoepoxide wasabove background in blood, bonemarrow, heart, lung, fat, spleen andthymus tissues of mice after 2 or 4 hoursof exposures to BD. In rats, levels ofmonoepoxide were increased in blood,fat, spleen and thymus tissues. Noincrease in monoepoxide in rat lung wasobserved. The more mutagenicdiepoxide was detected in all tissues ofthe mice examined immediatelyfollowing 4 hours of exposure. It wasdetected in heart, lung, fat, spleen andthymus of rats, but at levels 40- to 160-fold lower than those seen in mice.

In mice, the level of diepoxideexceeded the monoepoxide levelsimmediately after exposure in suchtarget organs as the heart and lungs.Thornton-Manning et al. concluded thatthe high concentrations of diepoxide inheart and lungs they observed suggestedto them that this compound may beparticularly important in BD-inducedcarcinogenesis.

The study authors noted that neitherepoxide was detected in rats’ liver andwas present only in quite lowconcentrations in the livers of mice.Thornton-Manning et al. found this


surprising since epoxides present inblood in the liver should have yieldedvalues greater than those observed inthe liver samples. They hypothesizedthat it might be due to prior metabolismof the epoxides before reaching the liveror it might be an artifact due to post-exposure metabolism of the epoxides inthe liver.

Thornton-Manning et al. did notdetect the monoepoxide in rat lungs,

and found the diepoxide level to bequite low. In contrast, in the mice theyfound both epoxides present in lungtissue, with the monoepoxide levelpresent at a concentration less thanexpected using blood volume values,and the diepoxide level agreeing withthat expected as a function of bloodvolume. Thornton-Manning et al.concluded that these results ‘‘* * *suggest that the lung is capable of

metabolizing BDO, but perhaps is lessactive in metabolizing BDO2. (Ex. 118–13, Att. 3) Moreover, Thornton-Manninget al. believed that although BD isoxidatively metabolized by similarmetabolic pathways in the rats andmice, the quantitative differences intissue levels between species may beresponsible for the increasedcarcinogenicity of BD in mice.

TABLE V–8.—TISSUE LEVELS [PMOL/GM TISSUE, MEAN±S.E.] OF EPOXYBUTENE AND DIEPOXYBUTANE IN RATS AND MICEFOLLOWING A 4-HOUR EXPOSURE TO 62.5 PPM BD BY INHALATION

TissueEpoxybutene Diepoxybutane

Rats Mice Rats Mice

Blood ................................................................................................................................. 36±7 295±27 5±1 204±15Heart ................................................................................................................................. 40±16 120±15 3±0.4 144±16Lung .................................................................................................................................. ND 33±9 0.7±0.2 114±37Liver .................................................................................................................................. ND 8±4 ND 20±4Fat ..................................................................................................................................... 267±14 1302±213 2.6±0.4 98±15Spleen ............................................................................................................................... 7±6 40±19 1.7±0.5 95±12Thymus ............................................................................................................................. 12.5±3.2 104±55 2.7±0.7 109±19Bone marrow 1 .................................................................................................................. 0.2±0.1 2.3±1.5 ND 1.4±0.3

ND=Not Detected.1 Bone marrow data are presented as mean pmol/mg protein ±; n=3 or 4 for each determination. Adapted from Ex. 118–13, Att. 3.

These data are shown in Table V–8.Seaton et al. examined the activities

of cDNA-expressed human cytochromeP450 (CYP) isozymes for their ability tooxidize epoxybutene to diepoxybutane.(Ex. 118–7N) They also determined therate of formation of the diepoxide bysamples of human liver microsomes(n=10) and in mice and rat livermicrosomes. Seaton et al. found thattwo of the cytochrome P450 isozymes,CYP2E1 and CYP3A4, catalyzedoxidation of 80 uM of monoepoxide todetectable levels of diepoxide, and thatCYP2E1 catalyzed the reaction at higherlevels of monoepoxide (5mM),suggesting the predominance of 2E1activity at low substrate concentrations.Hepatic microsomes from all 3 speciesformed the diepoxide when incubatedwith the monoepoxide. Seaton et al.hypothesized that the differencebetween these results and those ofCsanady et al. (who did not detect thediepoxide when the monoepoxide wassubstrate in a similar microsomal assay)was due to differences in experimentalmethodology.

Seaton et al. noted a 25-foldvariability in Vmax/Km among the 4human livers. They reported that Vmax/Km for oxidation of the monoepoxide tothe diepoxide for the 4 human sampleswas 3.8, 1.2, 1.3 and 0.15, while that ofthe pooled rat samples was 2.8, and themouse ratio was 9.2.

The authors, using available data,calculated an overall activation/detoxification ratio (Vmax/Km for

oxidation of BD to the monoepoxide)taking into account hydrolysis of themonoepoxide by epoxide hydrolase andconjugation with glutathione. Theactivation/detoxification ratio wasestimated at 1295 for the mouse, 157 forrats and 230 for humans. However,Melnick and Kohn point out that ‘‘whenyields of microsomal and cytosolicprotein content and liver size wereconsidered, the activation todetoxification ratio was only 2.8 timesgreater in mice than in humans and 3.4times greater in humans than in rats.These ratios do not take into accountinter-individual variability in theactivities of the enzymes involved.’’ (Ex.131)

Recently, Seaton et al. studiedproduction of the monoepoxide inwhole airways isolated from mouse andrat lung. (Ex. 118–7C) They explainedthe impetus to use fresh intact tissue bystating that lung subcellular fractions, asemployed in experiments by Csanady etal., described above, contained mixturesof cell type ‘‘so that the metabolizingcapacities of certain cell populationsmay have been masked.’’ Theyanticipated that use of airway tissuewould allow more precise quantitationof differences in lung metabolism of BD.

Whole airways or bronchioles isolatedfrom both male B6C3F1 mice and maleSprague-Dawley rats were incubated for60 min with 34 um BD. Levels of10.4±5.6 nmol epoxybutene/mg proteinwere detected in mouse lungs, while 2–3 nmol/mg protein was observed in rat

lung airway regions. Seaton et al. notedthat while the species differences ‘‘arenot dramatic,’’ they may in partcontribute to the differences incarcinogenicity observed in mice andrats.

To characterize conjugation of BDmetabolites with glutathione (GSH),Boogard et al. prepared cytosol fromlungs and livers of rats and mice andfrom 6 human donor livers andincubated them with 0.1 to 100 mMdiepoxide and labeled glutathione(GSH). (Ex. 118–7J) NMR (nuclear massresonance) and HPLC techniques wereused to characterize and quantitateconjugate formation.

Non-enzymatic reaction wasconcluded to be negligible. Theconjugation rates (Vmax) in mouse andrat livers were similar and 10-foldgreater than those observed in thehuman samples. The initial rate ofconjugation (Vmax) was much higher inmouse than rat lung. Both rodentspecies exhibited higher initial rates ofconjugation than human. This ledBoogard et al. to conclude that thehigher diepoxide levels observed in BD-exposed mice compared with rats ‘‘arenot due to differences in hepatic orpulmonary GSH conjugation of BDE (thediepoxide),’’ and further that sincehumans oxidize BD to the epoxides ata low rate, the low activity of GSHconjugation of the diepoxide in humanliver cytosol demonstrated in this study‘‘will not necessarily lead to increasedBDE (diepoxide) levels in humans


4 A preliminary study on the human populationof this study is described in the section of thispreamble dealing with the genetic toxicology of BDexposure.

potentially exposed to BD.’’ They alsopointed out the need to determine therate of BDE detoxification by othermeans, specifically by epoxidehydrolase in all three species.

Studies of Urinary Metabolites of BD

Two metabolites of BD have beenidentified in urine of exposed animalsby Sabourin et al. (Ex. 118–13 Att. 3)These are 1,2-dihydroxy-4-N-acetylcysteinyl-S-)-butane, designatedMI, and MII, which is 1-hydroxy-2-N-acetylcysteinyl-S-)-3-butene. (Ex. 118–13–Att. 3)

These mercapturic acids are formedby addition of glutathione (GSH) ateither the double bond (MI) or theepoxide (MII). MI is thought to form byconjugation of GSH with butenediol, thehydrolysis product of the monoepoxide,while MII is thought to form fromconjugation of the monoepoxide withGSH.

Sabourin et al. measured MI and MIIin urine from rats, mice, hamster andmonkeys. Mice were observed to excrete3 to 4 times as much MII as MI, whilethe hamsters and rats produced about1.5 times as much MII as MI. Themonkeys produced primarily MI.

The ratio of formation of metabolite Ito the total formation of the twomercapturic acids, MI and MII,correlated well with the known hepaticepoxide hydrolase activity in thedifferent species, suggesting that themonoepoxide undergoes more rapidconjugation with glutathione in themouse than in the hamster or rats, andthat the least rapid conjugation occursin the monkey. The epoxide availabilityis inversely related to the hepaticactivity of epoxide hydrolase, whichremoved the epoxide by hydrolysis.

In 1994, Bechtold et al. published apaper describing a comparison of thesemetabolites between mice, rats, andhumans.4 In workers exposed tohistorical atmospheric concentrations of3 to 4 ppm BD, Bechtold measuredurine levels of MI and MII by use ofisotope-dilution gas chromatography,and found MI, but not MII, to be readilydetectable. Bechtold et al. found thatemployees who worked in productionareas (having 3–4 ppm BD exposure)could be distinguished by this assayfrom outside controls and that low levelhuman exposure to BD resulted information of epoxide.

Bechtold et al. stated in their abstractthat since monkeys displayed a higherratio of MI to MI + MII than mice did,

and ‘‘because humans are known tohave epoxide hydrolase activities moresimilar to those of monkeys than mice,we postulated that after inhalation ofbutadiene, humans would excretepredominantly MI and little MII.’’ (Ex.118–13 Att. 3) Their observationssuggested that the predominant pathwayfor clearance of the monoepoxide inhumans is by hydrolysis rather thanconjugation with glutathione.

Bechtold et al. found when mice andrats were exposed to 11.7 ppm BD for4 hours and the ratio of the twometabolites was then measured, formice, the ratio of MI to MI ± MII (or the% of total which is MI) was 20%, thatof rats was 52%, while humansexhibited more than 97% MI. Thesedata also indicate the predominance ofclearance by hydrolysis pathways ratherthan GSH conjugation in the human.

Nauhaus et al. used NMR techniquesto study urinary metabolites of rats andmice exposed to ([(1,2,3,4)-13C]-butadiene). (Ex. 118–7I) Theycharacterized metabolites in mouse andrat urine following exposure byinhalation to approximately 800 ppmBD for 5 hours. Urine was collected over20 hours from exposed and controlanimals, centrifuged and frozen.

The findings of this study are quiteextensive and are briefly summarized asfollows. Nine metabolites were detectedand chemically identified in mouseurine and 5 in that of rats. Five weresimilar in the 2 species, though differingmarkedly in concentration. One wasunique to the rat and four to the mouse.Nauhaus et al. observed that ‘‘whennormalized to body weight (umol/kgbody weight), the amount of diepoxide-derived metabolites was four timesgreater in mouse urine than in raturine.’’ They further hypothesized that‘‘the greater body burden of (diepoxide)in the mouse and the ability of rats todetoxify [it] though hydrolysis may berelated to the greater toxicity of BD inthe mouse.’’ Nauhaus et al. found thatboth mice and rats conjugated themonoepoxide with glutathione, but therat preferentially conjugated at the twocarbon, while the mouse preferentiallyconjugated at the one carbon.Additionally, the finding of a metaboliteof 3-butenal, a proposed intermediate inthe oxidation of BD to crotonaldehyde,an animal carcinogen, is suggestive ofan alternative carcinogenic pathway forBD. In general, this study supports thein vitro findings of Csanady et al. whoreported similar rates for BMOconjugation with glutathione betweenrats and mice. (Ex. 118–7AA)

Interaction of Butadiene With OtherChemicals

Bond et al. described use of availabledata to simulate the potentialinteraction of BD with other workplacechemicals. (Ex. 118–7V) Specificallythey modeled potential interactionassuming competitive inhibition of BDmetabolism by styrene, benzene andethanol. The model predicted that co-exposure to styrene would reduce theamount of BD metabolized, but thatbecause of its relative insolubility, BDwould not effectively inhibit styrenemetabolism. Benzene, which, like BD, ismetabolized by P450/2E1, was alsopredicted to be a highly effectiveinhibitor of BD metabolism because ofits solubility in tissues. The modelspredicted that ethanol would have onlya marginal effect on BD metabolism atconcentrations of BD ‘‘relevant tohuman exposure.’’

BD and styrene co-exposures oftenoccur in the SBR industry and both aremetabolized by oxidation to activemetabolites, in major part, bycytochrome P450/2E1. To determine themetabolic effect of joint exposure to BDand styrene, Levans and Bonddeveloped and compared two PBPKmodels, one with one oxidative pathwayand competition between BD andstyrene and the other with twooxidation pathways for both BD andstyrene. (Ex. 118–7E) For modelvalidation, Levans and Bond exposedmale mice to mixtures of BD and styreneof 100 or 1000 ppm BD and 50, 100 or250 ppm styrene for 8 hours. They usedchamber inlet and outlet concentrationsto calculate uptake and, when steady-state was reached, calculated the rate ofmetabolism. They analyzed blood forstyrene, styrene oxide, epoxybutene anddiepoxybutane by GC–MS.

Leavens and Bond found BDmetabolism was inhibited when micewere co-exposed to styrene. Theinhibition approached maximum valueat co-exposure concentrations of styreneabove 100 ppm.

The report also described thepreliminary development ofpharmacokinetic models to simulate theobserved rate of BD metabolism in co-exposed mice. Their results supportedthe hypothesis that ‘‘more than oneisozyme of P450 metabolized BD andstyrene and competition does not occurbetween BD and styrene for allisozymes.’’ They were unable toaccurately predict blood concentrationsof styrene following exposure, and feltthat ‘‘ perhaps the diepoxide mayinhibit metabolism of styrene bycompeting for the same P450 enzyme.’’


Although preliminary in nature andreflecting effects of relatively highexposures, these observations ofinteractions between styrene and BDexposure may have implications for theobserved pattern of BD-induced effectsin human populations jointly exposed.Specifically, the cancer effects seen inSBR production workers mayunderestimate the effects of BD with nostyrene or benzene exposure.

Pharmacokinetic Modeling of BDMetabolism

In a recent publication, Bond et al.reviewed the results of application of anumber of physiologically-basedpharmacokinetic (PBPK) dosimetrymodels. (Ex. 118–7M) They noted thatthree of the models which includedmonoepoxide disposition (Kohn andMelnick, Johanson and Filser,Medinsky) predicted that, for any BDexposure concentration, steady-statemonoepoxide levels will be higher formice than for rats. Bond et al. furtherobserved that ‘‘while the three modelsaccurately predict BD uptake in rats andmice, they overestimate the circulatingblood concentrations of (monoepoxide)in these species compared to thoseexperimentally measured byHimmelstein.’’ Their results also ledBond et al. to conclude that thedisagreement between modelpredictions for the monoepoxide andexperimental data suggests that thestructure and/or parameter valuesemployed in these models are notaccurate for predicting blood levels ofBD epoxides, and conclusions based onmodel predictions of BD epoxide levelsin blood or tissue may be wrong.’’ (Ex.118–7M, p. 168) OSHA agrees withthese authors that BD epoxide levelsshould not be used in assessing risk. Inthe discussion, the authors pointed tothe need for inclusion of diepoxidetoxicokinetics (as well as that of themonoepoxide) in future modelingexercises, since they believe thediepoxide to be the ultimatecarcinogenic metabolite of BD.

Kohn and Melnick, in a recentpublication, used available data andattempted to apply a PBPK model to seewhether it was consistent with observedin vivo uptake and metabolism. (Ex.131) The model included compartmentsfor rapidly and for slowly perfusedtissues. Rate equations for monoepoxideformation, its hydrolysis, and forconjugation with glutathione wereincluded.

Kohn and Melnick acknowledgednumerous sources of uncertainty inapplying the model to the data (inwhich there are many gaps),necessitating various assumptions.

Their calculations led them to concludethat the ‘‘model reproduces whole-bodyobservations for the mouse and rat’’ andthat it predicts that ‘‘inhalation uptakeof butadiene and formation andretention of epoxybutene are controlledto a much greater extent byphysiological parameters than bybiochemical parameters. . . ‘‘ (Ex. 131)

When Kohn and Melnickinterchanged the biochemicalparameters in the mouse and humanmodels to see if ‘‘the differences incalculated net uptake of butadieneamong the three species were due todifferences in metabolic activity,’’ theyfound that use of human parameters inthe mouse model decreased the level ofabsorption of BD, but not to a level aslow as that of the human. Kohn andMelnick noted that the modelpredictions of epoxybutene levels in theheart and lung of mice and rats failed toaccount for the observation that mice,but not rats, develop tumors at thesesites. Kohn and Melnick suggested thatfactors other than epoxybutene levels,not accounted for in the model, areprobably crucial to induction ofcarcinogenesis.

Conclusions

Many metabolism studies have beenconducted both in vitro and in vivo,mostly in mice and rats, to determinethe BD metabolic, distribution, andelimination processes, and these studieshave been extended in attempts toexplain, at least in part, the greatercarcinogenic potency of BD in themouse, whether the mouse or the rat isa better surrogate for human cancer andreproductive risk assessment, and whatis the proper dose-metric to use in dose-response assessments. The question ofwhether the mouse or the rat is a bettermodel for the human on the basis oftumor response is partly addressed inthe risk assessment section of thispreamble. This section more specificallyconsiders whether these metabolicstudies in total can explain the differentcancer responses and potenciesobserved in the mouse, rat, and human.What is clear throughout the record isthat most scientists who study the topicconsider not BD itself, but the majorepoxide metabolites of BD, BMO andBDE and 1, 2-epoxybutane-3,4-diol, tobe the putative carcinogenic agents.Most of this research has focused on therelative species production of BMO andBDE. Both BMO and BDO have beenreported in early studies to becarcinogenic to mice and rats via skinapplication and/or subcutaneousinjection, with BDO being somewhatmore potent. (Ex. 23–88, Ex. 125).

Metabolism of BD to BMO in both theliver and lung of mice, rats and humansis by the P450 oxidation pathway, withCYP2E1 and CYP1A6 being the majorenzymes. Based on the studies reviewedby OSHA, overall the mousemetabolizes BD to the monoepoxide andthe diepoxide in these organs at a fasterrate than do the rat and human. This issupported by the following evidence: (1)The mouse has higher BMO and BDElevels in blood, lung, and liver (i.e., seeEx. 118–7S, Ex. 118–7D, and Ex. 118–13), which are the target organs forcancer in the mouse but not the rat; (2)the mouse has higher in vitro lung andliver microsome Vmax/Km ratios forboth BD and BMO metabolism than dorats or humans (Ex. 118–7AA); and (3)the mouse has higher hemoglobin-BMOadduct levels than rats and much higherlevels than humans. (Ex. 118–7Y) Amajor exception to the findings of thesestudies is the study by Duescher andElfarra, who found the in vitro BDVmax/km ratios to be the same in miceand human liver microsomes and 3–4times higher than they were in rats,suggesting that mice and humans havesimilar BD metabolic potential, at leastin the liver. (Ex. 128) Large variations,about 60 fold, were found among 10human liver microsome BD metabolicactivities. (Ex. 118–7N) A recent BD invitro metabolism study by Seaton et al.on whole rat and mouse lung airwayisolates found that the mouse producedabout twice the amount of BMO as therat (this difference could not explain thedifference between mouse and rat tumorincidence). (Ex. 118–7C)

BMO and BDE were also measured inheart, spleen, thymus, and bone marrow(target sites for mouse but not rattumors) following 4 hour BD inhalationexposure (62.5 ppm) to mice and rats.(Ex. 118–13) In these tissues, mouseBMO and BDE levels were 3 to 55 foldhigher than rat levels for the samemetabolites, although the mice organlevels of these metabolites correlatedpoorly with the mouse target organcancer response at this exposure level.Only high BDE levels in the mouse lungwere consistent with the mortalityadjusted cancer incidence (see hazardidentification—animal studies section,Ex. 114). This suggests that BDmetabolite tissue levels can, at best,only partly explain differences incarcinogenic response. Differences inboth species and tissue sensitivity mustalso be accounted for.

The Thornton-Manning and otherstudies also provided information aboutBD elimination. (Ex. 118–7I) Withhigher experimental exposure levels, themajor route of elimination of BD is viaexpiration. Elimination of BMO occurs


5 One exception: Seaton et al. found evidence‘‘that in mouse airways hydrolysis of BMO byepoxide hydrolase (EH) contributes to BMOdetoxification to a greater extent than doesglutathione conjugation.’’ (Ex. 118–7C)

by different pathways in differentspecies and different organs. At higherBD exposure concentrations, some BMOis expired. The mouse liver and lungappear to eliminate BMO predominantlyby direct conjugation with GSH 5. Forthe rat there is approximately equalelimination by the GSH and EHmediated pathways, while for thehuman and monkey hydrolysis tobutanediol is the major pathway forexcretion. ( Ex. 118–13 Att. 3) Thisspecies elimination pathway differenceis a partial explanation for the higherlevels of both BMO and BDE seen in themouse, assuming that most of the BDmetabolism takes place in the liver.With respect to the bone marrow BDdistribution and metabolism, mouselevels of the BD metabolites in the bonemarrow were lower than at any of theother target organs studied. (Ex. 118–13)In vitro studies by Gentler and Reciohave found no detectable P4502E1 inthe bone marrow of B6C3F1 mice. (Ex.118–7T) These authors conclude thatthis ‘‘suggests that BD is converted toBMO outside of bone marrow and issubsequently concentrated in bonemarrow, or that the conversion of BD toBMO occurs by an alternate enzymaticpathway within the bone marrow.’’ Thelatter appears to be the more likely sinceManiglier-Poulet and co-workersshowed that in vitro BD metabolism toBMO in both B6C3F1 mouse and humanbone marrow occur by a peroxidase-mediated process and not via the P450cytochrome system. (Ex. L–133) Since intheir system both human and mousebone marrow generated about the sameamount of BMO/cell, this suggests thatboth BD distribution to bone marrowand local metabolic reactions should beconsidered in species-to-speciesextrapolations and in PBPK modeling.

Inclusion of bone marrow localreactions becomes even more importantwhen considering the animal species touse for modeling human cancer. BD isgenotoxic in the bone marrow of mice,but not in rats. (Tice et al. 1987;Cunningham et al. 1986, reported in Ex.131) BD and BMO have been implicatedas affecting primitive hematopoieticbone marrow stem and progenitor cellsrelated to both T-cell leukemia andanemia in the mouse. (Irons et al., 1993,in Ex. 117–2) BD causes lymphoma inmice, but no lymphoma or leukemia inrats even at 8,000 ppm. Furthermore,the body of epidemiologic evidencestrongly indicates that BD exposure

poses an increased risk of humanleukemia (see the epidemiologic sectionand especially Ex. 117–1).

Fat storage of BD during exposure,and release following cessation ofexposure, is also a major concern, bothin estimating target organ levels and indetermining species differences. Thereis little in the record on the effect of fatstorage and release. In the Thornton-Manning study discussed above, bothmouse and rat fat levels of both BMOand BDE declined rapidly followingcessation of exposure, suggesting littlelingering effect. However, Kohn andMelnick present a model in which post-exposure release of BD from the fatwould result in extended epoxideproduction in humans in contrast withthe mouse. (Ex. 131)

Bond et al. suggest that the more rapidmetabolism of BD to BMO in the mouse,and the more rapid EH BMOelimination pathways in the rat andhuman may be an explanation for lower,if any, BDE levels seen in rat and humanliver microsomes and why BD will notbe carcinogenic to humans at exposurelevels seen in the environment or theworkplace. (Ex. 130) They also concludethat ‘‘Since significant tumor inductionin male rats occurs only at 8000 ppmBD, BMO levels are probably notpredictive of a carcinogenic response.’’Thornton-Manning et al. characterizethe peak levels of BDE in the mouselung and heart as being either greaterthan or equivalent to peak levels ofBMO, and suggest ‘‘that the formation ofBDE may be more important than theformation of BMO in the ultimatecarcinogenicity of BD.’’ (Ex. 118–13)However, BMO levels in these organswere also quite high, and were higherthan BDE levels in blood and bonemarrow, target organs for hematopoieticsystem cancers. OSHA believes that theevidence is not sufficient to dismiss thepotential contribution of BMO to mouse,rat or human carcinogenicity; toconclude that BDE should be consideredmore actively carcinogenic than BMO;or to find that BDE levels aresufficiently characterized in eithermouse or human tissue to be used as thedose metric for BD human riskassessment.

Thus, OSHA concludes, based on thebody of metabolic and other evidencepresented, and the above discussion,that the mouse is a suitable animalmodel for the human for BD cancer riskassessment purposes, and thatmetabolism of BD to active metabolitesis probably necessary forcarcinogenicity. However, while theuptake, distribution, and metabolism ofBD to active carcinogenic agents areimportant, local BD metabolic reactions

and specific species sensitivities appearto have at least as large an impact on BDpotency in the various species. This islikely to be especially true in thehuman, whose metabolic processesappear to be much more variable withrespect to BD. Thus, although themetabolism studies provide insight intoBD’s metabolic processes in variousspecies and organs (with the possibleexception of mouse lung tumorigenicityrelated to lung BDE levels and proteincross linking), OSHA finds that toomany questions remain unanswered,both with PBPK modeling efforts andwith actual in vivo measurements (andthe lack of such measurements inhumans) to base a quantitative riskassessment on BD metabolite levelequivalence between mice and humans.(Ex. L–132)

VI. Quantitative Risk Assessment

A. IntroductionIn 1980, the United States Supreme

Court ruled on the necessity of a riskassessment in the case of IndustrialUnion Department, AFL–CIO v.American Petroleum Institute, 448 U.S.(607), the ‘‘Benzene Decision.’’ TheUnited States Supreme Court concludedthat the Occupational Safety and Health(OSH) Act requires, prior to issuance ofa standard, that the new standard bebased on substantial evidence in therecord considered as a whole, that thereis a significant risk of health impairmentat existing permissible exposure limits(PELs) and that issuance of the standardwill significantly reduce or eliminatethat risk. The Court stated that, beforethe Secretary of Labor can promulgateany permanent health or safetystandard, he is required to make athreshold finding that a place ofemployment is unsafe in the sense thatsignificant risks are present and can beeliminated or lessened by a change inpractices. (448 U.S. 642)

In 1981, the Court’s ruling on theOSHA’s Cotton Dust Standard(American Textile ManufacturersInstitute v. Donovan, 452 U.S. 490(1981)) reaffirmed its previous positionin the Benzene Decision, that a riskassessment is not only appropriate, butthat OSHA is required to identifysignificant health risk to workers and todetermine if a proposed standard willachieve a reduction in that risk, andOSHA as a matter of policy agrees thatassessments should be put intoquantitative terms to the extent possible.

For this rulemaking, OSHA hasconducted a quantitative riskassessment to estimate the excess riskfor cancer and consequently forpremature deaths associated with


exposure to an 8-hour time-weighted-average (TWA), 5 days/week, 50 weeks/year, 45-year exposure to BD atconcentrations ranging from 0.1 to 5ppm, the range of permissible exposurelimits (PELs) considered by OSHA inthis rulemaking. The data used in thequantitative risk assessment were froma National Toxicology Program (NTP)chronic inhalation study in whichB6C3F1 mice of both sexes were exposedto either ambient air or BD exposureconcentrations ranging from 6.25 to 200ppm, known as NTP II. (Ex. 90) Forseven gender-tumor site combinations,multistage Weibull time-to-tumormodels were fit to these NTP II data.The best fitting models were chosen viaa log-likelihood ratio test.

OSHA’s maximum likelihoodestimate (MLE) of the excess risk ofdeveloping cancer and subsequentpremature death as a result of an 8-hourTWA occupational lifetime exposure to2 ppm BD, the PEL proposed by OSHAin 1990, was 16.2 per 1,000 workers,based on the most sensitive gender-tumor site combination, female mouselung tumors. If the occupational lifetime8-hour time-weighted-average (TWA)exposure level is lowered to 1 ppm BD,based on female mouse lung tumors, theestimate of excess cancer and prematuredeath drops to 8.1 per 1,000 workers. Inother words, an 8-hour TWA lifetimeoccupational exposure reduction from 2ppm to 1 ppm BD would be expected toprevent, on average, 8 additional casesof cancer and probable prematuredeaths per 1,000 exposed workers.Based on the individual tumor site dose-response data, which were bestcharacterized by a 1-stage Weibull time-to-tumor model, (male-lymphoma, male-lung, female-lymphoma and ovarian),on average, one would expect there tobe between 1 and 6 fewer excess casesof cancer per 1,000 workers based on a8-hour TWA occupational lifetimeexposure to BD at 1 ppm versus BD at2 ppm. Estimates of leukemia deaths atthe former 8-hour TWA PEL of 1,000ppm of BD, for an occupational lifetime,are not presented because contemporaryBD exposures are generally far lowerthan this level.

B. Assessment of Carcinogenic Risk

1. Choice of Data Base for QuantitativeRisk Assessment

The choice of data provides theplatform for a quantitative riskassessment (QRA). Either animal studieswhich evaluate the dose-responserelationship between BD exposure andtumorigenesis or epidemiological dose-response data may be suitable sources ofdata.

Estimates of the quantitative risks tohumans can be based on the experienceof animals from a chronic lifetimeexposure study. Chronic lifetimeinhalation bioassays with rats and micegenerally last 2 years or two-thirds ofthe lifespan of the animal. (Ex. 114)These types of studies provide insightinto the nature of the relationshipbetween exposure concentration,duration and resulting carcinogenicresponse under a controlledenvironment. Furthermore, someresearchers have estimated a variety ofmeasures of dose of BD, includinginhaled and absorbed dose as well as BDmetabolites, to estimate human risksbased on the observed dose-responserelationship of animals in a bioassay;the form of the dose used in a dose-response analyses is called the dose-metric.

The carcinogenicity of lifetimeinhalation of BD was studied inSprague-Dawley rats by theInternational Institute of SyntheticRubber Producers (IISRP) and in B6C3F1

mice by the National ToxicologyProgram. The IISRP sponsored a two-year inhalation bioassay of Sprague-Dawley rats performed at HazeltonLaboratories Europe (HLE). (Ex. 2–31)Groups of 110 male and female Sprague-Dawley rats were exposed for 6-hoursper day, 5 days per week to 0, 1,000, or8,000 parts per million (ppm) of BD.The males were exposed for 111 weeksand the females for 105 weeks.Statistically significant increased ratesof tumors were found in both male andfemale rats. Among exposed male rats,there were increased occurrences ofpancreatic and testicular tumors andamong the exposed female rats therewere higher incidence rates of uterine,zymbal gland, mammary and thyroidtumors than in the control groups.

The National Toxicology Program(NTP) has performed two chronicinhalation bioassays using B6C3F1 mice.(Ex. 23–1; 90; 96) The first study, NTPI, was intended to be a two-yearbioassay, exposing groups of 50 maleand female mice to 0, 625, or 1,250 ppmof BD for a 6-hour day, 5 days/week.The study was prematurely curtailed at60 weeks for the males 61 weeks for thefemales caused by an unusually highcancer mortality rate due to malignantneoplasms in multiple organs. Despitesome weaknesses in the way the studywas conducted, the results of this studyshow that BD is clearly carcinogenic inthese mice, with statistically significantincreases in malignant lymphomas,heart hemangiosarcomas, lung tumors,and forestomach tumors in comparisonto the controls for exposed male andfemale mice. (Ex. 90)

The second NTP BD chronicinhalation bioassay, NTP II, had groupsof 70 (except for the group exposed tothe highest concentration, whichcontained 90) male and female miceexposed to concentrations of 0, 6.25, 20,62.5, 200 and 625 ppm for 6 hours/day,5 days/week for up to 104 weeks. TheNTP II bioassay provided lowerexposures, closer to prevailingoccupational exposure levels, than theNTP I and HLE chronic inhalationstudies. The NTP II supported thepattern of carcinogenic response foundin NTP I. Both male and female miceexposed to BD developed tumors atmultiple sites including: lymphomas,heart hemangiosarcomas, and tumors ofthe lung, liver, forestomach, andHarderian gland (an accessory lacrimalgland at the inner corner of the eye inanimals; they are rudimentary in man).Reproductive tissues were alsoadversely affected. Among the exposedmales there were significant increases intumors of the preputial gland; amongfemales there were significant increasesin the incidence of ovarian andmammary tumors.

In 1996, a retrospective cohort studyby Delzell and co-workers of about18,000 men who worked in NorthAmerican synthetic rubber plants wassubmitted to OSHA. (Ex. 117–1) In thisstudy researchers derived estimates ofoccupational exposure to BD using avariety of resources, such as workhistories, engineering data, productionnotes, and employees’ institutionalmemories. In their October 2, 1995report Dr. Delzell et al., characterizedtheir effort as follows:

Retrospective quantitative exposureestimation was done to increase the power ofthe study to detect associations and to assistwith the assessment of the impact of specificexposure levels on mortality from leukemiaand other lymphopoietic cancers. (Ex.117–1)

In April 1996, Dr. Delzell expressedconcern with possible discrepanciesbetween estimated cumulativeexposures and actual measurements.(Ex. 118–2) OSHA believes that in awell-conducted study, retrospectiveexposure estimates can be reasonablesurrogates for true exposures;misclassifications or uncertainty candecrease the precision of the riskestimates derived from such a study, butthe problem must be severe andwidespread to invalidate the basicfindings.

At the time of publication of theproposed standard on occupationalexposure to BD (August 1990), only theNTP I mouse and HLE rat bioassayswere available for quantitative riskassessments (QRA). Presented in Table


6 Competing tumors refers to the lack ofopportunity of a later developing tumor to expressitself due to the occurrence of early developinglethal tumor; Among the 625 ppm exposure grouplymphocytic lymphomas were mortal earlydeveloping tumors which prevented laterdeveloping disease such as heart hemangiosarcomasfrom possibly developing.

V–9 is an overview of authorship anddata sets used in the various QRAssubmitted to the OSHA docket. Withone exception, the rest of the QRA’s inthe BD Docket have relied on animalchronic exposure lifetime bioassays.Each of the five risk assessmentsdiscussed in the proposal based its

quantitative risk assessment on one orboth of the higher-exposure chronicbioassays (exposure groups exposed toBD concentrations ranging between625–8,000 ppm). (Exs. 17–5; 17–21; 23–19; 28–14; 29–3; 32–27) The three QRAsconducted using bioassay datasubsequent to the publication of the

NTP II study used NTP II data withexposures of 6.25–625 ppm BD, closerto actual occupational exposures, forcalculating their best estimates of risk.(Exs. 90; 118–1b; 32–16)

A summary of each of the ten QRA’sfollows:

TABLE V–9.—SUMMARY TABLE OF QUANTITATIVE RISK ASSESSMENTS (QRAS) IN ORDER OF THEIR REVIEW IN THEOSHA BD STANDARD

Exhibit Author Data-set

90 .................... National Institute for Occupational Safety and Health (NIOSH)(Preliminary).

NTP II a bioassay (preliminary).

118–1b ............ NIOSH ........................................................................................ NTP II bioassay.118–1 .............. NIOSH ........................................................................................ Delzell et al. epidemiological study.17–21 .............. United States EPA Carcinogen Assessment Group (CAG) ...... NTP I b and HLE c bioassays; Epidemiological based on Fajen

Exposure Data.32–27 .............. California Occupational Health Program (COHP) of the Cali-

fornia Department of Health services (CDHS).NTP I; HLE bioassays Epidemiological based on Fajen Expo-

sure Data32–16 .............. Shell Oil Corporation ................................................................. NTP I, NTP II and HLE bioassays.17–5 ................ United States EPA Office of Toxic Substances (OTS) ............. NTP I bioassay.23–19 .............. ICF/Clement Inc ......................................................................... NTP I bioassay.29–3 ................ Center for Technology, Policy, and Industrial Development at

the Massachusetts Institute of Technology.NTP I and HLE bioassays.

28–14 .............. Environ Inc ................................................................................. HLE bioassay.

a NTP II, The National Toxicology Program, Technical Report 434, 2-year bioassay of B6C3F1 mice to 5 exposure groups receiving between6.25 and 625 parts per million (ppm) of BD

b NTP I, The National Toxicology Program, prematurely terminated longtime bioassay of B6C3F1 mice to 2 exposure groups receiving either625 or 1,200 ppm of BD

c HLE, Hazelton Laboratories Europe’s, lifetime bioassay of Sprague Dawley rats, exposed groups received 1,000 ppm of BD or 8,000 ppm ofBD

NIOSH-Quantitative Risk Assessmentsbased on NTP II

In the early 1990’s, two QRAs wereconducted sequentially by the NationalInstitutes for Occupational Safety andHealth (NIOSH). One was a preliminaryand the other a final, with the latterusing final pathology data for histiocyticsarcomas and one particular type oflymphoma from NTP II. In 1991, NIOSHsubmitted a preliminary QRA using thethen preliminary NTP II tumorpathology data for various individualorgan sites (8 from the female mice and6 from the male mice) to estimate excesscancer risk at different BD exposuresover an occupational lifetime. (Ex. 90)For all gender-tumor site analyses,NIOSH excluded the 625 ppm exposuregroup in its best estimate of risk sincethe plethora of competing tumors 6 inthis high exposure group provide lessinformation for a dose-response analysisof individual tumor sites than do datafrom some of the lower exposuregroups. Another reason for theexclusion was that the dose-time-

response relationship in mice issaturated for exposures above 500 ppmand the data would thus provide verylittle additional information for lowdose extrapolation. NIOSH’s QRA reliedon an allometric conversion of bodyweight to the three-quarters power, (mg/kg)3⁄4, and equated a 900-day-old mouseto a 74-year old human. To avoidduplication of risks, NIOSH presentedonly maximum likelihood estimatesbased on the aggregate of all types oflymphomas even though dose-responsedata were also available for thelymphocytic lymphoma subset.

Of the fourteen gender-tumor site datasets NIOSH modeled to extrapolateanimal data to humans, 12 (86%)yielded excess risks greater than 2cancer deaths per 1,000 workers, givenan 8-hour TWA lifetime occupationalexposure of 1 ppm BD. Estimates ofexcess risks to workers based on the bestfitting models for each of the six dose-time-response relationships for maletumor sites were between 0.4 and 15.0per 1,000 workers assuming an 8-hourTWA, 45 year occupational exposure to1 ppm BD. Among estimates based onmale mice’s dose-response data, thelowest and highest excess risk estimateswere from the heart hemangiosarcomaand Harderian gland dose-responserelationships, respectively. Forestimates of excess risk based on either

gender’s set of individual tumor dose-response relationships, only the hearthemangiosarcoma data predicted a riskof less than 1 per 1,000 workers with anoccupational lifetime exposure of 1ppm: these data predicted 0.4 and3×10¥3 excess cancer cases per 1,000workers based on the best fitting modelsfor male and female mice, respectively.

Based on tissue sites in females, theexcess risk estimates for 8-hour TWAoccupational lifetime exposure to 1 ppmBD range between 4 and 31 per 1,000workers.

NIOSH presented its findings forlifetime exposure to 2 ppm as follows:

Based on tumors at the most sensitive site,the female mouse lung [assuming (mg/kg)3⁄4conversion], our maximum likelihoodestimates of the projected human increasedrisk of cancer due to a lifetime occupationalexposure to BD at a TWA PEL of 2 ppm isapproximately 60 in 1,000 (workers). (Ex. 90)

For the linear models, if scaling wereon a (mg/kg) basis rather than the (mg/kg)3⁄4 used by NIOSH for allometricconversion, the revised estimate ofexcess cancer risk for an 8-hour TWAoccupational lifetime exposure to 2 ppmBD would decrease approximately 6fold to 9.2 per 1,000 workers based onthe same female mouse lung tumor data.

In 1993, NIOSH finalized its estimatesof excess risk caused by occupationalexposure based on the tumorigenesis


experience of mice in the NTP II study.(Ex. 118–1B) The rounded maximumlikelihood estimates (MLE) from thefinal QRA are presented in Table V–10.NIOSH expanded the gender-tumor sitesto include histiocytic sarcoma for bothmale and female mice. NIOSH chose topresent only its risk estimate based onlymphocytic lymphoma, rather than anassessment based on the aggregate oflymphomas. In the preliminary andfinal NIOSH QRAs, 1-stage time-to-

tumor models’’ rounded estimates ofrisk associated with lifetime exposure to1 ppm BD ranged from 1 to 30 excesscancer cases per 1,000 workers, withestimates based on the male-lymphocytic lymphoma and the female-lung dose-response data providing thelower and upper ends of the range ofrisk, respectively.

As part of its sensitivity analyses,NIOSH derived the estimates of riskbased on (1) equating a human lifespan

to a mouse equivalent age of 784 days,a figure OSHA has used, and (2)equating a human lifespan to a mouselifespan of 900 days (a figure more oftenused by NIOSH.) The best estimates ofrisk equating human lifespan to a mouselifespan of 784 days were lower, byabout one-third, than those assuming ahuman lifespan equivalency to 900 daysfor the mouse, all else held constant.

TABLE V–10.—NIOSH’S a FINAL QUANTITATIVE RISK ASSESSMENT’S (QRA) MAXIMUM LIKELIHOOD ESTIMATES (M.L.E.S) b

PER 1,000 WORKERS OF LIFETIME EXCESS RISK DUE TO AN OCCUPATIONAL c EXPOSURE TO 1 PPM OF BD USINGBEST FITTING MODELS, AS DESIGNATED BY NUMBER OF STAGES OF THE WEIBULL TIME-TO-TUMOR MODEL

Gender-tumor site MLE, Final QRA(Stages)

Male mouse: ............................Forestomach ............................................................................................................................................................................. 0.03 (2)Harderian gland ........................................................................................................................................................................ 10 (1)Heart hemangiosarcoma .......................................................................................................................................................... 0.5 (2)Histiocytic sarcoma ................................................................................................................................................................... 8 (1)Liver .......................................................................................................................................................................................... 4 (1)All Lymphoma ........................................................................................................................................................................... NALymphocytic lymphoma ............................................................................................................................................................ 0.9 (1)Lung .......................................................................................................................................................................................... 10 (1)

Female mouse: ............................Forestomach ............................................................................................................................................................................. 5 (1)Harderian Gland ....................................................................................................................................................................... 7 (1)Heart hemangiosarcoma .......................................................................................................................................................... 3×10¥3 (3)Histiocytic sarcoma ................................................................................................................................................................... 10 (1)Liver .......................................................................................................................................................................................... 7 (1)All lymphoma ............................................................................................................................................................................ NALymphocytic lymphoma ............................................................................................................................................................ 9 (1)Lung .......................................................................................................................................................................................... 30 (1)Mammary .................................................................................................................................................................................. 4 (1)Ovarian ..................................................................................................................................................................................... 9 (1)

a Based on NTP II, excluding the 625 ppm exposure category, equating a 900-day-old mouse to a 74-year old human and assuming anallometric conversion of (mg/kg)3/4.

b Rounded to one significant figure.c Occupational lifetime is an 8-hour time-weighted-average, 40-hours per week, 50-weeks per year, time-weighted-average (TWA) for 45-years.

The Carcinogen Assessment Group QRA

The Carcinogen Assessment Group(CAG) and the Reproductive EffectsAssessment Group of the Office ofHealth and Environmental Assessmentat the United States EnvironmentalProtection Agency (EPA) also conductedan assessment of the mutagenicity andcarcinogenicity of BD. (Ex. 17–21) In itsquantitative risk assessment, CAG usedboth male and female response datafrom the two chronic bioassays availableat the time, NTP I with B6C3F1 mice andthe HLE Sprague Dawley rat study. TheCAG analysis is based on EPA’sestablished procedures for quantitativerisk analyses, which fit the total numberof animals with significantly increasedor highly unusual tumors with thelinearized multistage model and use theupper 95% confidence interval. Micedying before week 20 and rats dyingduring the first year of the study (beforethe observation of the first tumor) were

eliminated from the analysis to adjustfor non-tumor differential mortality.

The dose-metric was based on apreliminary report by the LovelaceInhalation Toxicology Research Instituteof its six-hour exposure study in B6C3F1

mice and Sprague Dawley rats atdifferent concentrations of BD, roughlycorresponding to the concentrationsused in NTP I and HLE, with totalinternal BD equivalent dose expressedas a function of inhalation exposureconcentration. Then CAG estimated theamount and percent of BD retained forvarious exposure concentrations inthese bioassays. These internal dose-estimates were then extrapolated tohumans based on animal-to-human ppmair concentration equivalence.

CAG adjusted risk estimates from themouse study by a factor of (studyduration/lifetime) 3 to account for less-than-lifetime observations, since theNTP I study was prematurelyterminated at 60 weeks for males and 61

weeks for females due to predominatingcancer mortality. CAG extrapolated theshort lifespan mouse data to anexpected mouse lifetime, 104 weeks, inorder to estimate lifetime risk tohumans.

CAG estimated all risks based oncontinuous exposure to BD, 24 hoursper day, 365 days per year, for a 70-yearlifetime. The incremental unit riskestimates for the female mouse wereabout eight times as high as those for thefemale rat; for the males, theincremental unit risk estimate for micewas about 200 times as high as for rats.The CAG final incremental unit riskestimate of 0.64 (ppm)¥1 is based on thegeometric mean of the upper-limit slopeestimates for male and female mice andwould predict an upper limit of 640excess cancers per 1,000 people exposedto 1 ppm continuously throughout theirlifetime, 70 years. Extrapolating thissame estimate to an equivalent 45-yearworking lifetime of 240 work days per


year at an 8-hour TWA exposure to 1ppm BD would yield an upper-limit riskestimate of 90 excess cancers per 1,000workers. If the working day is assumedto require one-half (10m 3) the dailytidal volume, the total amount of airinhale, the excess would be 135 cancersper 1,000 workers.

California Occupational Health Program(COHP) QRA

In 1990, five years after the CAGconducted its quantitative riskassessment, the California OccupationalHealth Program (COHP) produced itsestimates of risk with a similarassessment of the carcinogenicity of BD,using the same available bioassays, withmore recent information on BD risk inhumans, pharmacokinetic (PK)modeling, and animal low exposureabsorption efficiency. (Ex. 32–16) Usingthree separate dose-metrics for eachbioassay and multistage models tocharacterize the basic dose-responserelationship, CAG presented severalquantitative estimates of incrementallifetime unit risks. Quantal lifetimeresponse multistage models were fit tothe data. COHP, like NIOSH, used theindividual data with a multistageWeibull time-to-tumor model tocharacterize the dose responserelationship. COHP stated that it also fitMantel-Bryan and log-normal models tothe data, and that the multistage modelsgave a better fit; the results obtainedwith these other models were notreported.

COHP performed calculations on eachprimary tumor site separately, and alsodid calculations on the pool of primarytumors that showed significantlyincreased tumor incidences. For theirmain dose-metric, COHP refined theCAG approach, using a revised estimateof low-exposure absorption viainhalation. COHP also included anestimate of the PK model derived BDmonoepoxide metabolites, but de-emphasized their use by stating thatthese were ‘‘presented for comparativepurposes only.’’ The third dose-metricwas straight ppm for animal-to-humanspecies conversion (adjusting forduration of exposure). COHP stated:(COHP) followed standard EPA practice andassumed that a certain exposureconcentration in ppm or mg/m 3 inexperimental animals was equivalent to thesame exposure concentration in humans. (Ex.32–16)

Like CAG, COHP also adjusted for lessthan lifetime survival in the NTP Imouse study, by using a cubic power oftime, (study duration/lifetime) 3.COHP’s potency estimate adjustment forthe male mouse study with 60-weeksurvival was 5.21; for the 61-week

female mouse survival the adjustmentwas 4.96.

With all the combinations of sites,species, sexes, models, and dose-metrics, COHP presented over 60potency estimates for the rat and over100 for the mouse. As with the CAG andother analyses, the estimates based onNTP I were typically one to two ordersof magnitude greater than those basedon the rat for similar dose-metrics,models and total tumors. COHP chosethe estimates based on the male mouseas final indicators of human risk basedon the ‘‘superior quality of the mousestudy.’’ From these estimates, using thequantal form of the multistage model,COHP chose ‘‘the upper bound forplausible excess cancer risk to humans.’’COHP’s final cancer potency estimate of0.32 (ppm)¥1 presented in units ofcontinuous lifetime exposure, is basedon all significant tumors in the malemouse and uses the internal BDequivalent dose conversion factor of0.54 mg/kg-d/ppm for the mouse andanimal-to-human ppm equivalency.COHP’s final potency estimate was one-half the value of 0.64 (ppm)¥1

calculated by the CAG; the difference isdue mainly to a low exposureabsorption modification by COHP. Thecontinuous lifetime exposure potencyfactor converts to a working lifetime riskof 45 to 67 excess cancers per 1,000workers, exposed to 1 ppm of BD at an8-hour TWA over a 45 year workinglifetime.

COHP, like CAG, attempted todetermine whether its animal-based riskextrapolation could predict theleukemia mortality observed inepidemiology studies. Following theapproach employed by CAG in itsanalyses of the Meinhardt (1982) study,the COHP compared its estimates of riskfrom bioassays to the then most recentepidemiological studies of Downs et al.(1987) and Matanoski and Schwartz(1987). Both COHP and CAG used MLEsbased on mouse lymphoma forcomparing the animal-derived potencyestimates with the occupationalresponse. In addition, neither COHP norCAG used the upward adjustment factorof approximately 5 to correct for theless-than-lifetime duration of NTP I.Because neither of these epidemiologystudies (Downs et al. (1987) orMatanoski and Schwartz (1987)) hadrecorded exposure estimates, the COHPrelied on 8-hr TWA estimates of 1 and10 ppm taken at different but similarplants reported by Fajen et al. (1986).For lifetime unit risk estimates, COHPused the initial MLE of 0.0168 (ppm)¥1

derived from the male mouse lymphomaanalysis, unadjusted for less-than-lifetime survival. This part of the

analysis also assumed that alymphocytic outcome in the animalswould equate to leukemia death inhumans. These assumptions yielded arange of 6 to 21 predicted lymphocyticcancer deaths (for 1 and 10 ppmexposures) versus the 8 observed byDowns et al.

Office of Toxic Substances (OTS) QRAThe Office of Toxic Substances (OTS),

U.S. Environmental Protection Agency(EPA) conducted a quantitative riskassessment using only the NTP I data.(Ex. 17–5) The reasons cited for thischoice include: (1) The mouse is a moresensitive test species for BD than the rat;(2) a quality control review had beendone for the mouse bioassay at the timeOTS wrote its risk assessment whereasnone was available for the rat bioassay;(3) greater amount of histopathologicaldata was available for the NTP I studythan for the HLE rat study; and (4) thetype of BD feedstock used by NTP I hada much lower dimer concentration thanthe BD used by HLE (increased dimerconcentration results in the lowering ofavailability of BD for metabolism to themono- and di-epoxides, which arethought to be the carcinogenic agents).To compensate for early termination ofthe NTP I study, OTS adjusted dose bya factor of (study duration/lifetime).3Butadiene ppm exposure concentrationwas used as the measure of dose andmouse-to-human species extrapolationwas also on a ppm equivalence basis.OTS estimated cancer risks based onheart hemangiosarcoma and pooledtumors (grouping of sites showingstatistically significant elevatedincidence rates) tumors using a 1-stagequantal model. Workplace exposures toBD were converted to estimated lifetimeaverage daily doses. Since the NTP Istudy was curtailed at 61 weeks, tumorincidence rates were adjusted forsurvival by life-table methods. Cancerrisks were based on administered doseof BD and not delivered dose to varioustarget organs. (Ex. 17–5) Estimated 95%upper confidence-limits for the excessrisk of cancer from an occupationallifetime exposure to an 8-hour TWA of1 ppm BD, for 240 days/year for 40years, ranged between 10 and 30 per1,000 workers, based on pooled tumorincidence for female and male animals,respectively.

ICF/Clement EstimatesIn 1986, ICF/Clement (ICF) estimated

the risk of cancer associated withoccupational exposure to BD. (Ex. 23–19) ICF determined that only the NTP Idata were suitable for a risk assessmentbased on animal data, (NTP II data werenot available at that time) based on ICF/


Clement’s concern over thediscrepancies between HLE’s summarystatistics and individual counts. ICFchose to use individual tumor type datafor some of its analyses. ICF fitted alinearized multistage quantal model tothe NTP I data. Based on a preliminarystudy by Bond (a senior toxicologist atthe Chemical Industry Institute ofToxicology), ICF adjusted the NTP Iexposure concentrations for percentretention which varied inversely from100% at 1 ppm to 5% at 1,000 ppm.

ICF assumed ppm as the proper dose-metric and ppm to ppm for the mouse-to-human species extrapolation factor.(Exs. 23–86; 23–19) The 95% upperconfidence limit estimates of risk basedon pooled female tumor data with alifetime occupational exposure was 200per 1,000 workers at 1 ppm BD, and 400per 1,000 workers at 5 ppm BD; the non-proportionality reflects the assumptionof lower percentage retentions at higherconcentrations.

Massachusetts Institute of Technology(MIT) QRA

Hattis and Wasson at the Center forTechnology, Policy, and IndustrialDevelopment at MIT conductedpharmacokinetic/mechanism-basedanalyses of the carcinogenic riskassociated with BD. (Ex. 29–3) Theanalyses include both HLE and NTP Idata. Key elements, such as partitioncoefficients for blood/air and tissue/blood, were not available to bemeasured and had to be estimated. Thebest estimate of excess risk of cancergiven a lifetime occupational exposureof 1 ppm BD 8-hr TWA was 5 per 1,000workers based on the NTP I femalemouse data set, incorporatingpharmacokinetic models which set theblood/air partition coefficient to 0.2552.Based on the HLE female rat data witha blood/air partition coefficient of0.2552, an excess risk was estimated tobe 0.4 additional cases of cancer forevery 1,000 workers at an 8-hour TWA,occupational lifetime exposure to 1 ppmBD.

Environ QRAEnviron conducted a quantitative risk

assessment based on the HLE ratbioassay data. (Ex. 28–14) Environ notedthat the relatively high BDconcentrations of the earlier bioassays(HLE with groups exposed to 8,000 and1,000 ppm BD and NTP I withexposures of 1,250 and 625 ppm BD)made it difficult to extrapolate risks tothe relevant, lower exposure levels ofBD in occupational settings. Environstated that among B6C3F1 mice,metabolic saturation occurs with 8-hourTWA BD concentrations greater than

500 ppm; thus, the time-dose-responserelationship is different at higher dosesthan at lower doses. Environ stated thatthe methodological problems and thehigh early mortality shown in the NTPI data contributed to the uncertainty ofits relevance to human risks andtherefore chose to use the HLE ratbioassay data instead. Environ believesthat human metabolism of BD is moresimilar to that in the Sprague-Dawley ratthan in the B6C3F1 mouse. Extrapolatedrisks were based on estimates ofabsorbed dose, expressed in mg/kg, asdefined in the Bond et al. (1986)absorption study. (Ex. 23–86)

Environ used the HLE female rats toestimate the extra lifetime risk ofdeveloping cancer given anoccupational lifetime 8-hr TWAexposure to 1 ppm BD. Using MLEsfrom multistage, Weibull, and Mantel-Bryan models, based on the totalnumber of female rats with significantlyincreased tumors, Environ’s predictedoccupational lifetime risks were 0.575(Multistage), 0.576 (Weibull), and 0.277(Mantel-Bryan) per 1,000 workers.

Shell Oil Company QRAShell Oil Company estimated excess

cancer risks by the multistage quantaland the Weibull time-to-tumor modelsbased on female hearthemangiosarcomas and pooledmalignant tumors from the NTP II study.Shell estimated human risks based onvarious assumptions, correcting for BDretention and/or relative human epoxidedose. Shell stated that the Weibull time-to-tumor model better characterizedrisks since it was able to fully utilizeavailable dose-response data, includingtime until onset of tumors and latency(time from initiation until detection oftumor). (Ex. 32–27) Shell used

* * * crude time-to-tumor data consistingof early deaths to 40- weeks, 40-week interimsacrifices, deaths to 65- weeks, 65-weekinterim sacrifices, death to 104- weeks andterminal sacrifices * * * in-lieu ofindividual animal data [for NTP II data]. (Ex.32–27)

OSHA believes that the true dose-response relationship is obscured byShell’s use of crude time-to-tumor dataand its grouping of early deaths to 40weeks, deaths to 65 weeks and deaths to104 weeks; instead, dose-time-tumorresponse data for each individual mouseshould have been used.

Shell did not explain why it choseone model over the other. For example,without explanation, Shell dropped thehighest exposure group, 625 ppm, whenestimating lifetime occupational risk forall of its Weibull time-to-tumor modelsand dropped additional dose groupswhen using some multistage quantal

models. Moreover, estimates of excessrisk were presented only for 5-stageWeibull time-to-tumor models, althoughthere is no discussion of correct modelspecifications. For example, no reasonsare given for choosing the 5-stage modelrather than another. Also, Shell does notsupport its estimation that the latencybetween the induction of a tumor andits observation is for the pooled femalemice malignant tumors and 40-weeksfor the female mice hearthemangiosarcomas.

Based on the Shell analyses,extrapolating from pooled malignantfemale mice tumors, assuming 10%human BD retention efficiency at 2ppm, and on a 5-stage Weibull time-to-tumor model, one would expect 18excess cancers per 1,000 workers givenan 8-hour TWA occupational lifetimeexposure of 2 ppm BD. Based on thesame data set, but assuming a mouse-to-human species conversion factor basedon an epoxide ratio of 590 (mouse-to-monkey) in addition to a 10% BDretention efficiency factor, the estimateof excess risk of cancer drops to 0.3cases per 1,000 workers with an 8-hourTWA occupational lifetime exposure of2 ppm. Using the same pooledmalignant female mice tumors, butassuming the blood epoxide estimates ofthe Dahl et al. study and an 8-hour TWAlifetime occupational BD exposure of 2ppm, the estimate of excess risk ofcancer is slightly lower, 0.24 per 1,000workers. The excess risk estimates basedon female hemangiosarcomas and a 5-stage Weibull time-to-tumor model andoccupational lifetime exposure to 2 ppmof BD were: (a) 6.4×10¥8 (assuming a10% BD retention factor); (B) 6.2×10¥15

(assuming a 10% BD retention factorand an epoxide ration of 590); and (c)1.3×10¥11 (assuming the blood epoxideestimates of the Dahl et al. study).

Shell also presented the Environ Inc.QRA based on the HLE Sprague-Dawleyrat bioassay and made similaradjustments for BD retention and bloodepoxide to those it made for the NTP IIB6C3F1 mice data. As had Environ, Shellstated that the dose- response of the ratis more relevant than that of the mice inpredicting risk in humans. Shellconcluded that the risk estimatesderived from HLE Sprague Dawley ratdata should be given greater weight thanthose based on the B6C3F1 mouse data.

NIOSH’s QRA Based on the Delzell et al.Study

NIOSH estimated the excess risk ofworkers developing leukemia based onthe Delzell et al. preliminary estimatesof occupational exposure categories of aretrospective cohort study. (Exs. 117–1;118–1) NIOSH derived excess risks from


the best fitting relative risk (RR) model,the square root model, as fit by Delzellet al. who adjusted for age, years sincehire, and calendar period. The preferredfinal model specified by Delzell et al.was:Relative Risk=1+0.17×(BD ppm-years)0.5

Under this model the age-cause specificleukemia death rates (ACSDR) are afunction of cumulative occupationalexposure up to that age. Theoccupational ACSDRs are amultiplicative function of backgroundACSDR times the BD-caused relativeincrease (0.17 * BD ppm-years) inleukemia. These total ACSDRs werethen applied to an actuarial programwhich adjusted for competing risks toestimate lifetime excess risk of leukemiaassociated with 45-year 8-hour TWAoccupational exposures for a number ofPELs for BD. Estimates of backgroundrates of leukemia and all causes of deathwere taken from the mortality rates forall males, 20 to 65 years of age, from the1989 Vital Statistics of the UnitedStates. This model estimates the excessrisk of leukemia death, given anoccupational lifetime exposure of 2 ppmof BD, as 11 per 1,000 workers.Lowering the 8-hour TWA occupationallifetime BD PEL to 1 ppm, on average,one would expect there to be 8 excessleukemia deaths per 1,000 workers overa working lifetime.

In most animal bioassays, exposure tochemical carcinogens is usuallyassociated with an elevated tumorincidence at only one or two targettissues. BD is of great concern becausesignificantly increased incidences oftumors at multiple sites and doses wereobserved in both rats and mice.

OSHA’s final risk assessment is basedupon the NTP II bioassay. (Exs. 90; 96)In NTP II, the following tumor sites’incidence rates were elevated: Heart,lymph nodes, lung, forestomach,Harderian gland, preputial gland, liver,ovaries and mammary gland. The NTPII bioassay was preferred over the NTPI mouse and the HLE rat bioassay forseveral reasons. First, most of theexposure levels for NTP II (6.25, 20, 62.5and 200 ppm) were closer to currentoccupational exposure levels than werethose in the other bioassays (625; 1,000and 8,000 ppm); studies with higherthan typical occupational exposureconcentrations may lead to difficultiesin extrapolating the effects to the lowerconcentrations of BD which typicallyoccur in current occupational settings.Furthermore, for doses (625 to 8,000ppm) above the metabolic saturationlevel of 500 ppm, the biologicallyeffective doses are not proportional toppm exposure concentrations. Second,

the NTP II mice were successfullyrandomized to exposure groups andtheir individual pathology reports wereconsistently coded. The randomizationof the bioassay mouse population lendsto the internal validity of the studythrough the similar composition ofexperimental and control groups. Third,Good Laboratory Practices werefollowed, as verified by audits. Fourth,there was a clear dose-responserelationship for several cancer sites.Fifth, since the carcinogenic mechanismis still unknown, OSHA conservativelyestimates excess risk to humans basedon the experience of the more sensitiveanimal species unless there is specificevidence indicating that the choice ofthat species is inappropriate. Sixth, riskassessment results based on thepreliminary findings from the mostrecent epidemiologic study suggest thatthe B6C3F1 mouse is a reasonablespecies to use for quantitative riskassessment. (Ex. 118–1)

For its risk assessment, OSHA hasfocused exclusively on those tumor sitesthat are scientifically pertinent. Fromthe NTP II study, the range of excesscancer risk associated with a lifetimeoccupational exposure to BD isestimated based on the dose-responserelationships of four target tissues, threecommon to both genders: Heart(hemangiosarcoma), lung, andlymphoma, and one, ovarian tumors,observed in one gender only. OSHA’sfocus on these four individual targettissues is based not on an objection tothe use of other tissue tumors and sitesbut rather on the judgment that thechosen animal sites are appropriatebecause they include both rare (e.g.,heart hemangiosarcoma) and commontumors (e.g., lung) and those sites withthe lowest (heart hemangiosarcoma) andhighest incidence rates (lymphatic).

Three of the target organs chosen forthe QRA demonstrated a significantlyelevated tumor incidence in both maleand female animals; ovarian tumorincidence was also significantlyelevated in female animals. For bothmale and female mice, hearthemangiosarcomas were selected formodeling because there is virtually nobackground incidence of hearthemangiosarcoma among untreatedmice in the NTP control population;only 0.04% of unexposed B6C3F1 micedevelop heart hemangiosarcoma, andthus any observed increase in theincidence of heart hemangiosarcomacould be attributed to BD exposure. (Ex.114, p. 121) The earlier developinglymphocytic lymphoma caused asignificant number of mice to die.Therefore, leaving mice are left at riskfor the later developing tumor, heart

hemangiosarcoma. (Ex. 114, p. 123) Thissituation is known as competing risk(the lack of opportunity for laterdeveloping tumors to expressthemselves because an earlierdeveloping tumor has already causedthe death of the animal. The occurrenceof heart hemangiosarcomas in the NTPstudy is even more notable because ofthese competing risks.

In the absence of definitive,pharmacokinetic information, OSHAhas estimated excess risks to humansbased on the most sensitive species-sex-tumor site. Lung tumors are the mostsensitive sites for both male and femaleB6C3F1 mice and, as such, wereincluded in OSHA’s final riskassessment.

Ovarian tumors are an example of thegroup of reproductive tumors whichalso had significantly increasedincidence rates among the animals inthe NTP II bioassay. Other significantlyincreased incidence rates were seen intesticular, preputial and mammarytumors.

The increased risk of developingleukemia that has been observed in theepidemiological studies suggests thatlymphomas might be the most relevanttumor site in animals for estimating thequantitative cancer risk to workers.Some have suggested that the high rateof lymphoma among B6C3F1 mice mighthave been due to the presence of themurine retro virus (MuLV) and haveasserted that the presence of this virusin B6C3F1 mice may be partiallyresponsible for the incidence of thymiclymphoma. For example, in 1990, Dr.Richard Irons reported,

A major difference between NIH Swiss andB6C3F1 mice is their respective exotropicretro viral background (MuLV) * * *Chronic exposure to BD (at 1250 ppm) for upto a year resulted in a fourfold difference inthe incidence of thymic lymphoma betweenB6C3F1 mice and NIH Swiss mice * * * Therole of endogenous retro virus (MuLV) in theetiology of chemically induced murineleukemogenesis is presently not understood.(Ex. 23–104)

Dr. Melnick of the National ToxicologyProgram testified during his publichearing statement,

In terms of the difference in responsebetween the B6C3F1 mouse or the NIH SwissMouse, you must be aware that the study isnot a complete cancer study. It’s a one-yearexposure. We do not know the full responsein the NIH Swiss mouse if it were conductedas a cancer study (about 2-years). (Tr. 1/16/91, p. 382)

Furthermore, NIOSH stated: ‘‘It is notknown whether the retro virusactivation mechanism is operative at thelower exposure concentrations of 1,3-butadiene [below 1250 ppm].’’ (Ex. 90)


There is no information in the record toshow that retrovirus insertion into theB6C3F1 mice of the NTP II study led tothe induction of lymphoma. Nor is thereinformation indicating that the murineretro virus may have led to anenhancement of butadiene-inducedlymphomas in B6C3F1 mice. Thedevelopment of thymic lymphoma inBD-exposed NIH Swiss mice that do nothave this endogenous virus arguesagainst the virus alone inducing thelymphomas observed in the BD-exposedB6C3F1 mice. (Ex. 23–104)

Tables V–11 and V–12 show thebreakdown of microscopically examinedtissues included in OSHA’s QRA, byexposure concentration and deathdisposition of female and male mice. Asillustrated in the tables, microscopic

examination varied by tissue type,exposure group, means of death, andgender. Microscopic examinations of alltissues were made for all natural deaths,and moribund and terminal sacrifices,irrespective of exposure group.

For each gender-exposure-group, 10animals were sacrificed at 40 and 65weeks. Microscopic evaluations werenot made for all tissue types amonginterim sacrifices (40 and 65 weeks).Among early sacrifices (40 weeks) forthe 6.25 and 20 ppm exposure groups,there were no microscopic examinationsof the relevant tissues. For the 65-weekfemale sacrifices at the 6.25 and 20 ppmdose levels only lung and ovariantissues were examined microscopically.No microscopic evaluations were madefor male 65-week sacrifices at the 6.25

ppm exposure level, but at the 20 ppmexposure level, animals weremicroscopically examined for hearthemangiosarcoma and lung cancer.Male and female interim sacrificesexposed to 62.5 ppm of BD were notmicroscopically examined for hearthemangiosarcoma.

Only observations confirmed bymicroscopic examination were includedin the analyses. Among natural deathsfor some gender-tissue combinations,there were a few animals for whichtissues were not available. Tissueunavailability was due to autolysis (celldestruction post death) and missingtissues due to the delay betweenaccident and discovery.

TABLE V–11.—TYPES OF TISSUES MICROSCOPICALLY EXAMINED BY CONCENTRATION DOSE AND DISPOSITION GROUPSAMONG FEMALE MICE FROM NTPa

Concentrationppm

Natural death and moribundsacrifice Week 40 sacrifice Week 65 sacrifice Terminal sacrifice

0 ...................... lymphoma, heartb, lung, ova-ries.

lymphoma, heart, lung, ova-ries.



6.25 ................. lymphoma, heart, lung, ova-ries.

nonec ...................................... lung, ovaries .......................... lymphoma, heart, lung, ova-ries.

20 .................... lymphoma, heart, lung, ova-ries.

none ....................................... lung, ovaries .......................... lymphoma, heart, lung, ova-ries.

62.5 ................. lymphoma, heart, lung, ova-ries.

lymphoma, lung, ovaries ........ lymphoma, heart, lung, ova-ries.


200 .................. lymphoma, heart, lung, ova-ries.




a These organs and tissue types are those contained in the OSHA risk assessment and do not reflect all of the types of tissues which were mi-croscopically examined.

b Heart, specifically Heart hemangiosarcoma.c None of the four tissue types used in the OSHA quantitative risk assessment were microscopically examined.

TABLE V–12.—TYPES OF TISSUES MICROSCOPICALLY EXAMINED BY CONCENTRATION DOSE AND DISPOSITION GROUPSAMONG MALE MICE FROM NTPa

Concentration ppm Natural death and mori-bund sacrifice

Week 40sacrifice

Week 65sacrifice

Terminalsacrifice

0 ......................................... lymphoma, heart b, lung, lymphoma, heart, lung ...... lymphoma, heart, lung ...... lymphoma, heart, lung.6.25 .................................... lymphoma, heart, lung, none c ................................ none .................................. lymphoma, heart, lung.20 ....................................... lymphoma, heart, lung, none .................................. heart, lung ......................... lymphoma, heart, lung.62.5 .................................... lymphoma, heart, lung ...... lymphoma, lung, ............... lymphoma, heart, lung ...... lymphoma, heart, lung.200 ..................................... lymphoma, heart, lung ...... lymphoma, heart, lung ...... lymphoma, heart, lung ...... lymphoma, heart, lung.

a These organs and tissue types are those contained in the OSHA risk assessment and do not reflect all of the types of tissues which were mi-croscopically examined.

b Heart, specifically heart, hemangiosarcomac None of the four tissue types used in the OSHA quantitative risk assessment were microscopically examined.

2. Measure of DoseThe mechanism of cancer induction

by BD is unknown for both rodents andhumans. One or more of the metabolitesof BD, epoxybutene, diolepoxybutaneand diepoxybutane, are suspected asbeing responsible for the carcinogenicresponse in at least some of the cancers.However, which of the metabolites maybe responsible for how much of thecarcinogenic response has yet to bedetermined. Bond suggests that

epoxybutene and diepoxybutane may beresponsible for carcinogenic responses.(Ex. 32–28) Dr. Bond wrote:

If carcinogenic response is elicited by ametabolite, as has been suggested, micebecause of their higher rate of metabolism,might be expected to yield a greater(carcinogenic) response than rats. (Ex. 17–21)

Because there are different theoriesabout which metabolites of BD areresponsible for the various carcinogenicresponses, some risk assessments have

characterized carcinogenic risk as aresult of type of dose: External,absorbed, or retained. In the BDproposal (55 FR 32736), OSHAcalculated the 14C–BD equivalents thatwere retained in mice at the conclusionof a 6-hour exposure period andincorrectly labeled the level as‘‘absorbed dose.’’ This does notnecessarily represent all the BDabsorbed through inhalation exposure.(Ex. 34–1)


The metabolic and pharmacokineticproperties of BD have not been fullycharacterized for either humans oranimals. Despite the absence of agenerally accepted pharmacokineticmodel, some metabolic information canstill be applied to OSHA’s QRA. Theoverall rate of BD metabolism in B6C3F1

mice is approximately linear at externalconcentrations up to 200 ppm; BDmetabolism increases sublinearly asconcentrations increase until it issaturated at 625 ppm. (Ex. 90) Bondreported that epoxybutene is one of theputative carcinogenic metabolites forwhich metabolism in the B6C3F1 mousebecomes saturated at 500 ppm; thus, theB6C3F1 mouse is unable to eliminateepoxybutene as quickly above 500 ppm.Bond suggests that above 500 ppmdirect quantitative extrapolation of riskfrom mouse studies may not be justified.(Ex. 23–86) Therefore, the 625 ppmexposure group was excluded fromOSHA’s risk assessment. Similarly,NIOSH and Shell did not include the625 ppm exposure group in their bestestimates of risks using NTP II data.However, NIOSH did include the 625ppm dose group in its sensitivityanalyses to see how the inclusion of thedata would affect the specification (theform and number of dose explanatoryvariables e.g., d, d2, d3, etc.) of themodel and the estimates of risk. (Ex. 90)

3. Animal-to-Human ExtrapolationA QRA based on a mouse bioassay

requires setting values for some mouseand human variables, including thoseused in animal-to-humanextrapolations. The values of thesevariables were chosen before conductingthe analyses. In OSHA’s quantitativerisk assessment, a mouse’s life span wasassumed to be 113 weeks. Mice were 8weeks old at the beginning of the studyand were exposed for up to 105 weeks.OSHA assumes workers will have anaverage lifespan of 74 years and anoccupational lifetime, working 5 days/week, 50 weeks/year, of 45 years. In theNTP II study, the average male mouseweighed 40.8 grams and female mouseweighed 38.8 grams. (Ex. 90) Mice wereassigned breathing rates of 0.0245 l/min.Breathing rates of workers (for an 8-hourworkday) were set at 10 m3/8-hr.

OSHA has chosen to use a straightmg/kg, body weight to the first power,(BW)1, intake as the animal-to-humanspecies extrapolation factor for doseequivalence. Other BD QRAs employedvarious extrapolation factors such asppm equivalence, (mg/kg)3/4

equivalence, BD mono-epoxide bloodlevels between mice and monkeyequivalence, and BD total bodyequivalence in (mg/kg)2/3. OSHAbelieves that the evidence for the use ofany of the alternative extrapolationfactors is persuasive, although theAgency believes that body weightextrapolation is appropriate in this case

because of the systemic nature of thetumors observed in both animalbioassays. This conversion of bodyweight, (BW)1 , produces estimates ofrisk which are lower than those derivedusing (BW)3/4, everything else heldconstant. For example, with a linear, 1-stage model, if OSHA used the (BW)3/4

conversion, holding all other elementsconstant, one would expect theestimates of excess risk to humans to beabout 6.5 times higher than if the (BW)extrapolation factor had been usedbecause of the weight of theexperimental species (between 38.8 and40.8 grams), and their breathing rate.For the quadratic (2-stage) and cubic (3-stage) models, the effect of relying onthe (BW)3/4 conversion rather than the(BW)1, holding all else constant, wouldbe to increase the predicted excesshuman risk more than 6.5 fold. (Ex. 90)

4. Estimation of Occupational Dose

It is necessary to estimate thedevelopment of cancer at a variety ofoccupational doses. This requiresoccupational doses to be converted intounits comparable to those used tomeasure the animal experimental dose.As discussed earlier, OSHA firstconverted animal experimentalexposures measured in ppm intooccupational intake dose measured in(mg/kg).

An exposure of 1 ppm BD isconverted into an equivalent exposuremeasured in mg/m3 using the equation:

1

154 1

242 213

3

ppm BDMolecular Weight BD

Molecular Weight of Airdensity of air

ppm BDmg mole

mole mBD mg m

= ×

= =. /

.45 /. /

Given a worker weighing 70 kg, breathing 10 m3 of air per 8-hour day, and exposed to air containing Y ppmBD, the inhaled dose of BD in mg/kg is given by:

Y mg kg BD inhaled Y ppm BDmg m

ppm

m

kg/ .

/( ) = ( ) × ×2 2110

70

3 3

Using the above formula, one cancalculate the estimated equivalentinhaled BD exposure among workersbased on the exposure concentrationsfor animals (See Table V–13).

TABLE V–13.—ESTIMATE OF TOTALHUMAN INHALED DOSE OVER AWORKDAY FOR VARIOUS EXPOSURELEVELS OF BD

Exposure con-centrations (ppm)

Estimate of total humaninhaled BD over a work-

day (mg/kg/8-hours)

200 ........................ 63.262.5 ....................... 19.820 .......................... 6.35 ............................ 1.62 ............................ 0.61 ............................ 0.3

5. Selection of Model for QuantitativeRisk Assessment

In the proposal (55 FR 32736), OSHAestimated excess risk using a quantalform of the multistage model (in areparameterized form as calculated byGLOBAL83), which based estimates ofrisk to humans on the experience of thegroup rather than the individual. Threeof the later risk assessments, Shell,NIOSH, and COHP, used a Weibulltime-to-tumor form of the multistagemodel to fit the mouse bioassays. (Exs.32–27; 90; 32–16) Time-to-tumor


models use more of the availableinformation than quantal multistagemodels to characterize time until thedevelopment of each observable tumor,and extrapolate risks, based on anoccupational dosing pattern. Sincesignificant increases in tumor incidenceoccurred at multiple sites in the NTP IIbioassay and a time-to-tumor modeltakes these competing risks intoaccount, a time-to-tumor method ispreferred over a quantal model. (Ex.118–1B)

Therefore OSHA used a Weibull time-to-tumor form of the multistage modelto characterize the risks of developmentof observable tumors, using the softwarepackage, TOXlRISK Version 3.5 by ICFKaiser. The model predicts theprobability, P(t,d), of tumor onset withdose pattern d by time t. It adjusts forcompeting causes of death prior totime t.

The Weibull time-to-tumor model is amultistage model based on the theory ofcarcinogenesis developed by Armitageand Doll. This theory of carcinogenesisis based on the assumption that a singleline of stem cells must pass through acertain number of stages sequentially forthe development of a single tumor cell.In the reparameterized form of themodel used here, a k stage model isdescribed by a polynomial of degree k,with all dose parameters greater than orequal to zero. The number of stagesnecessary for a model to be correctlyspecified varies by type of tumor,animal, and exposure agent, or anycombination of the three.

Both the MLE and the 95% upperlimit of the risk of developing cancer invarious tissues per 1,000 workers bytime t are calculated. The 95% upperbound is the largest value of excess riskthat is consistent with the observed datawith two-sided 95% confidenceintervals. The 95% upper bound iscomputed based on the Weibull time-to-tumor model for which the parameterssatisfy:¥2 (Log likelihood¥Log

likelihoodmax)≤2.70554Where: Log likelihoodmax is the

maximum value of the log-likelihood

A 1-stage model is linear in dose; a 2-stage model is quadratic in dose; a 3-stage Weibull model is cubic in dose.Below is a mathematical representationof a 3-stage Weibull time-to-tumormodel:P(t,d)=1-exp [¥(q0 + q1d + q2d2 + q3d3)

(t¥t0z)]where: t0 designates the time of onset ofthe tumor, t is the variable for time thetumor was observed and is assumed tofollow a Weibull distribution; d is the

dose-metric and is multistage; z is aparameter to be estimated, constrainedbetween 1 and 10; the backgroundparameter qo and the dose parameters,q1, q2, q3, are constrained to be non-negative. Constraining the doseparameters to zero or greater isbiologically based, since the doseparameters are proportional to themutation rates of the successive stagesin the development of a tumor cell. TheWeibull time-to-tumor model providesreasonable fits for about 75% of thetissues in the NTP historical controldata base, but the precision of the fit tothe dose-response data depends on thespecific agent. (Ex. 90)

Four forms of the model, one less thanthe number of exposure groups, for eachgender-outcome were fit to the data. Thecorrect specification of the model, thenumber of stages, is determined by thefit of the model to the data. Thelikelihood ratio test identifies whichmodel is a better fit by determining ifthe log-likelihood of a model issignificantly greater than anothermodel’s value. The 1-, 2-, 3- and 4-stageWeibull time-to-tumor models for eachgender-outcome combination wereordered according to the value of theirlog-likelihood. If the log-likelihood ofthe higher stage model is significantlygreater than that of the next lower stagemodel’s log-likelihood, one would rejectthe null hypothesis (the additional stagedoes not create a model that bettercharacterizes the data) and concludethat the higher stage model is asignificantly better predictor of theestimates of risk in the observed rangethan is the lower stage model.

The steps of the likelihood ratio testare as follows:

For example, assuming an alpha of0.05, and 1 degree of freedom (thedifference in the number of parametersfrom 1-stage and 2-stage models), thecritical value would be 3.84.

Fail to Reject H0 if:2 (log likelihood1-stage¥log

likelihood2-stage)<3.84Reject H0 if:

2 (log likelihood1-stage¥loglikelihood2-stage)≥3.84

If two times the difference of the loglikelihood values of the nth stage modeland the nth + 1 stage model was lessthan 3.84, then the additional stagewould be deemed unnecessary forgoodness of fit; on the grounds ofparsimony, the lower stage modelwould be used for the risk assessment.Otherwise, the higher stage modelwould be judged a better fit than thelower stage one and the process wouldcontinue.

While the likelihood ratio test issuitable for testing the significance ofthe next higher degree dose parameter,the biologically reasonable constraint onthe background incidence parameter q0

and dose parameters that they be non-negative q1, q2, q3>=0,—may impair thelog-likelihood ratio test’s power todetermine statistical significance.

The incidences of lymphoma, hearthemangiosarcoma, lung and ovariantumors are shown in Tables V–14 andV–15 for males and females,respectively. The TOXRISK Weibulltime-to-tumor model requires that thetumor context be described for eachobservation. Outcomes can be put intothree context categories: (1) Censored,no tumor; (2) rapidly fatal tumor; and(3) observed, tumor incidental to theanimal’s survival. Since OSHA waspredicting the time until onset of tumor,assuming no lag time between onset anddetection of tumor, t0 was set to zero.Therefore, estimates of risk to humansbased on the contribution to thelikelihood of either a rapidly fatal orincidental tumor are mathematically thesame.

Tables V–16 and V–17 show theWeibull time-to-tumor model estimatesof log-likelihoods, the shape parameters,intercept and dose coefficients forrelevant target tissues for male andfemale mice, respectively. The relativeperformance of various staged modelsfor a specific target tissue-gender areenumerated in the log-likelihood values.It should be noted that some of thetissue-gender combination’s log-likelihood values do not vary eventhough there is a change in the numberof the stages in the model. For example,the log-likelihood values for models ofall lymphoma for males and lungtumors for males and females are¥6.986 E+1, ¥1.763 E+2, ¥1.626 E+2,respectively, regardless of thespecification, number of stages, in themodel. OSHA concluded that the 1-stage models were preferred.

As identified in Tables V–16 and V–17, only heart hemangiosarcoma modelsare non-linear. This is consistent withNIOSH’s results when fitting Weibulltime-to-tumor models to these gender-tumor combinations. The quadratic (2-stage) model for males and the cubic (3-stage) model for females bettercharacterized the dose-responserelationship in modeling time todetection of heart hemangiosarcomathan did the linear models. The higherstage model necessary to fit the hearthemangiosarcoma data is driven by theabsence of cases in the two lowerexposure groups, shown in Tables V–14and V–15. Unlike the other tissuesstudied, there were no cases of heart


hemangiosarcoma in the control andlowest exposure groups for both maleand female mice. Both male and femalemice had similar hearthemangiosarcoma tumor rates, almost

30%, among the 200 ppm exposuregroups. The intercepts, q0, were zero formodels of both male and female micebased on the dose-response of hearthemangiosarcomas. This is consistent

with what one would expect, given theabsence of background incidence ratesof heart hemangiosarcomas.

TABLE V–14.—UNIVARIATE ANALYSIS OF HEART, LUNG, AND ALL LYMPHOMA NEOPLASMS BY EXPOSURE LEVEL OF 1,3-BUTADIENE AMONG NTP II MALE MICE ANALYZED IN THE TIME-TO-TUMOR MODELS

Neoplasm

Outcome

Tumor n a

(%N b)Censored c

n (%N) Total N

All lymphoma, 0 ppm ....................................................................................................................................... 4 (5.7) 66 (94.3) 70All lymphoma, 6.25 ppm .................................................................................................................................. 3 (6.0) 47 (94.0) 50All lymphoma, 20 ppm ..................................................................................................................................... 8 (16.0) 42 (84.0) 50All lymphoma, 62.5 ppm .................................................................................................................................. 11 (15.9) 58 (84.1) 69All lymphoma, 200 ppm ................................................................................................................................... 9 (12.9) 61 (87.1) 70Heart hemangiosarcoma, 0 ppm ...................................................................................................................... 0 (0) 70 (100) 70Heart hemangiosarcoma, 6.25 ppm ................................................................................................................. 0 (0) 49 (100) 49Heart hemangiosarcoma, 20 ppm .................................................................................................................... 1 (1.7) 59 (98.3) 60Heart hemangiosarcoma, 62.5 ppm ................................................................................................................. 5 (8.6) 53 (91.4) 58Heart hemangiosarcoma, 200 ppm .................................................................................................................. 20 (29.4) 48 (70.6) 68Lung tumor, 0 ppm ........................................................................................................................................... 22 (31.4) 48 (68.6) 70Lung tumor, 6.25 ppm ...................................................................................................................................... 23 (46.9) 26 (53.1) 49Lung tumor, 20 ppm ......................................................................................................................................... 20 (33.3) 40 (66.7) 60Lung tumor, 62.5 ppm ...................................................................................................................................... 33 (47.8) 36 (52.2) 69Lung tumor, 200 ppm ....................................................................................................................................... 42 (60.0) 28 (40.0) 70

a n is number of microscopically determined outcomes per tumor-context, gender, exposure-group outcome site combination.b N is the total number of gender, exposure-group, outcome site combination which were microscopically examined.c Tumor’s context is C (censored); animals were microscopically examined and no tumor was found at this site.

TABLE V–15.—UNIVARIATE ANALYSIS OF HEART, LUNG, ALL LYMPHOMA AND OVARIAN NEOPLASMS BY EXPOSURE LEVELOF 1,3-BUTADIENE AMONG NTP II FEMALE MICE ANALYZED IN THE TIME-TO-TUMOR MODELS

Neoplasm

Outcome

Tumorna (%Nb)

Censored c

n (%N) Total N

All lymphoma, 0 ppm ....................................................................................................................................... 10 (14.3) 60 (85.7) 70All lymphoma, 6.25 ppm .................................................................................................................................. 14 (28.0) 36 (72.0) 50All lymphoma, 20 ppm ..................................................................................................................................... 18 (36.0) 32 (64.0) 50All lymphoma, 62.5 ppm .................................................................................................................................. 10 (14.3) 60 (85.7) 70All lymphoma, 200 ppm ................................................................................................................................... 19 (27.1) 51 (72.9) 70Heart hemangiosarcoma, 0 ppm ...................................................................................................................... 0 (0) 70 (100) 70Heart hemangiosarcoma, 6.25 ppm ................................................................................................................. 0 (0) 50 (100) 50Heart hemangiosarcoma, 20 ppm .................................................................................................................... 0 (0) 50 (100) 50Heart hemangiosarcoma, 62.5 ppm ................................................................................................................. 1 (1.7) 58 (98.3) 59Heart hemangiosarcoma, 200 ppm .................................................................................................................. 20 (28.6) 50 (71.4) 70Lung tumor, 0 ppm ........................................................................................................................................... 4 (5.7) 66 (94.3) 70Lung tumor, 6.25 ppm ...................................................................................................................................... 15 (25.0) 45 (75.0) 60Lung tumor, 20 ppm ......................................................................................................................................... 19 (31.7) 41 (68.3) 60Lung tumor, 62.5 ppm ...................................................................................................................................... 27 (38.6) 43 (61.4) 70Lung tumor, 200 ppm ....................................................................................................................................... 32 (45.7) 38 (54.3) 70Ovarian tumor, 0 ppm ...................................................................................................................................... 1 (1.4) 68 (98.6) 69Ovarian tumor, 6.25 ppm ................................................................................................................................. 0 (0) 59 (100) 59Ovarian tumor, 20 ppm .................................................................................................................................... 0 (0) 59 (100) 59Ovarian tumor, 62.5 ppm ................................................................................................................................. 9 (12.9) 61 (87.1) 70Ovarian tumor, 200 ppm .................................................................................................................................. 11 (15.7) 59 (84.3) 70

a n is number of microscopically determined outcomes per tumor-context, gender, exposure-group outcome site combination.b N is the total number of gender, exposure-group, outcome site combination which were microscopically examined.c Tumor’s context is C (censored); animals were microscopically examined and no tumor was found at this site.


TABLE V–16.—MAXIMUM LIKELIHOOD ESTIMATES OF MODEL COEFFICIENTS FROM VARIOUS STAGES OF WEIBULL TIME-TO-TUMOR MODELS USING THREE TUMOR RESPONSES OF MALE MICE IN THE NTP II STUDY, EXCLUDING 625 PPMEXPOSURE GROUP; SELECTION OF SPECIFICATION OF MODEL IS BASED ON LIKELIHOOD RATIO TEST

Neoplasm Stage a Log-likeli-hood Z b q0 q1 q2 q3 q4

Heart hemangiosarcoma .................... W1 ¥7.061 9.810 0.00 8.306 E–23Heart hemangiosarcoma .................... c W2 ¥2.783 E–1 10 0.00 0.00 3.071 E–25Heart hemangiosarcoma .................... W3 ¥2.712 E–1 10 0.00 1.058 E–24 2.636 E–25 2.057 E–28Heart hemangiosarcoma .................... W4 ¥2.659 E–1 10 0.00 1.119 E–24 2.664 E–25 0.00 9.626 E–31All lymphoma ...................................... c W1 ¥6.986 E+1 4.743 2.709 E–11 6.136 E–13All lymphoma ...................................... W2 ¥6.986 E+1 4.743 2.709 E–11 6.136 E–13 0.00All lymphoma ...................................... W3 ¥6.986 E+1 4.743 2.709 E–11 6.136 E–13 0.00 6.540 E–33All lymphoma ...................................... W4 ¥6.986 E+1 4.743 2.709 E–11 6.136 E–13 0.00 0.00 0.00Lung tumor .......................................... c W1 ¥1.763 E+2 3.318 1.132 E–7 2.636 E–9Lung tumor .......................................... W2 ¥1.760 E+2 3.413 7.674 E–8 1.253 E–9 3.134 E–12Lung tumor .......................................... W3 ¥1.760 E+2 3.143 7.674 E–8 1.253 E–9 3.134 E–12 0.00Lung tumor .......................................... W4 ¥1.760 E+2 3.413 7.674 E–8 1.253 E–9 3.139 E–12 0.00 0.00

a Stage of time-to-tumor model; W1, Weibull 1-stage time-to-tumor model; W2, Weibull 2-stage time-to-tumor model; W3, Weibull 3-stage time-to-tumor model; W4, Weibull 4-stage time-to-tumor model.

b Z is the shape parameter; it is bounded, (1<=z<=10).c Selected Model.

TABLE V–17.—MAXIMUM LIKELIHOOD ESTIMATES OF MODEL COEFFICIENTS FROM VARIOUS STAGES OF WEIBULL TIME-TO-TUMOR MODELS USING FOUR TUMOR RESPONSES OF FEMALE MICE IN THE NTP II STUDY, EXCLUDING 625 PPMEXPOSURE GROUP; SELECTION OF SPECIFICATION OF MODEL IS BASED ON LIKELIHOOD RATIO TEST

Neoplasm Stagea Log-likeli-hood Zb q0 q1 q2 q3 q4

Heart hemangiosarcoma .................... W1 ¥2.097 E+1 4.957 0.00 4.356 E–13Heart hemangiosarcoma .................... W2 –8.745 6.126 0.00 0.00 2.222 E–17Heart hemangiosarcoma .................... W3c ¥4.866 6.770 0.00 0.00 0.00 8.088 E–21Heart hemangiosarcoma .................... W4 ¥4.267 7.011 0.00 0.00 0.00 2.637 E–22 1.368 E–3Ovarian tumor ..................................... W1c ¥6.140 E+1 2.857 1.407 E–8 .031 E–9Ovarian tumor ..................................... W2 ¥6.069 E+1 4.079 5.397 E–11 7.075 E–12 1.399 E–13Ovarian tumor ..................................... W3 ¥6.069 E+1 4.079 5.397 E–11 7.075 E–12 1.399 E–13 0.00Ovarian tumor ..................................... W4 ¥6.069 E+1 4.079 5.397 E–11 7.075 E–12 1.399 E–13 0.00 0.00All lymphoma ...................................... W1c ¥5.724 E+1 6.857 3.453 E–15 1.338 E–16All lymphoma ...................................... W2 ¥5.501 E+1 7.143 1.18 E–15 2.577 E–18 2.453 E–19All lymphoma ...................................... W3 ¥5.426 E+1 7.230 7.758 E–16 6.847 E–18 0.00 7.809 E–22All lymphoma ...................................... W4 ¥5.401 E+1 7.258 7.360 E–18 7.359 E–18 0.00 0.00 3.387 E–24Lung tumor .......................................... W1c ¥1.626 E+2 3.416 2.096 E–8 2.096 E–9Lung tumor .......................................... W2 ¥1.626 E+2 3.416 2.090 E–8 2.090 E—9 0.00Lung tumor .......................................... W3 ¥1.626 E+2 3.416 2.090 E–8 2.096 E–9 0.00 0.00Lung tumor .......................................... W4 ¥1.626 E+2 3.416 2.090 E–8 2.096 E–9 0.00 0.00 0.00

a Stage of time-to-tumor model; W1, Weibull 1-stage time-to-tumor model; W2, Weibull 2-stage time-to-tumor model; W3, Weibull 3-stage time-to-tumor model; W4, Weibull 4-stage time-to-tumor model.

b Z is the shape parameter; it is bounded, (1<=z<=10).c Selected Model.

OSHA’s Estimates of Risk

The estimates from OSHA’squantitative risk assessment based an 8-hour TWA, occupational lifetime,working 5 days/week, 50 weeks/year,for 45 years, at various BD PELS areshown in Table V–18. The MLEs ofexcess risk of material impairment of

health per 1,000 workers for cancer,based on tumors of various tissue sitesand the 95% upper bounds, arepresented. Various 8-hour TWA PELS,ranging from 0.1 to 5 ppm, are presentedto provide a context in which toevaluate the OSHA final rule PEL of 1ppm and to explore the feasibility ofother PELS, including the proposed PEL

of 2 ppm. Risks at the former BD 8-hourTWA PEL, 1,000 ppm, are not presentedin Table V–18. Although risks could beestimated for an occupational lifetimeexposure to an 8-hour TWA of 1,000ppm of BD from the linear models, thereis little relevancy to estimating the truerisk at an 8-hour PEL for BD at 1,000ppm for an occupational lifetime, since


8-hour TWA BD exposures have beengenerally far lower than 1,000 ppm.

Although the estimates ofcarcinogenic outcomes differ, excessrisks derived from tumor sites commonto both male and female B6C3F1 micehad the same relative ranking fromlowest to highest risk estimates by targettissues (heart hemangiosarcomas <lymphomas < lungs) within each gendergroup. After a lifetime occupationalexposure to BD at the proposed 8-hourTWA PEL of 2 ppm based on the abovemodel fits to these three individualtumor sites, one would expect between2.7×10¥4 to 16.2 excess cancer cases per1,000 workers, depending on whichgender-tumor site dose-responserelationship is used as the basis for theextrapolation to human occupationalexcess risks. Decreasing the BD 8-hourTWA PEL from 2 to 1 ppm, results ina reduction of the range of estimates ofexcess risk of cancer to between3.4×10¥5 to 8.1 cases per 1,000 workers.

The estimate of excess cancer riskbased on male mouse lymphoma is 1.3per 1,000 workers at an 8-hour TWA foran occupational lifetime exposure to 1ppm BD. Extrapolating from femalemouse lymphoma data results in an

estimate of 6.0 extra cancer deaths per1,000 workers at a BD 8-hour TWA PELof 1 ppm for an occupational lifetime ofexposure.

Extrapolating from the most sensitivesite, the female mouse lung, based onthe 1-stage Weibull time-to-tumormodel, with an 8-hour TWA PEL of 2ppm of BD for an occupational lifetime,one would expect 16 excess cancercases per 1,000 workers. Lowering thePEL to 1 ppm would cut the expectednumber of excess cancers in half to 8cases, based on the same gender-tumorsite. Based on male lung tumors, theestimate of excess cancer deaths for an8-hour TWA exposure to 2 ppm BD overan occupational lifetime was 12.8 per1,000 workers; lowering the 8-hourTWA occupational lifetime exposurelevel to 1 ppm BD decreases theestimate of excess cancer risk to 6.4 per1,000 workers, a reduction of 6 cancercases per 1,000 workers.

OSHA’s estimates of prematureoccupational leukemia deaths based onthe 1-stage Weibull time-to-tumormodels for the following outcome sites:All lymphoma, lung tumors, andovarian tumors, ranged between 1.3 and8.1 per 1,000 workers. Similarly,

NIOSH’s 14 estimates of the excess riskof death due to leukemia, based on 1-stage Weibull time-to-tumor models, asa consequence of exposure to an 8-hourTWA of 1 ppm BD over an occupationallifetime, ranged between 0.9 and 30cases per 1,000 workers. Thepreliminary estimate of 8 per 1,000 fromthe Delzell et al. study is concordantwith this range of animal-basedestimates. OSHA acknowledges thatthere is uncertainty in the Delzell et al.estimate, perhaps due to the naturalsampling variability present in anyepidemiologic study plus the possibilityof extra-binomial uncertainty stemmingfrom exposure misclassification. Whilethis uncertainty makes it difficult to saywhether quantitative risk estimateswould be adjusted up or down relativeto animal-based estimates, thissuggestion is far less important than thebasic conclusion that the Delzell et al.study reinforces earlier estimates. Evenif refinement of exposures caused theDelzell et al. estimate to move up ordown by even as much as a factor of 5or more, it would not change thisqualitative, and roughly quantitative,agreement.

TABLE V–18.—MAXIMUM LIKELIHOOD ESTIMATES (MLE) AND NINETY-FIVE PERCENT UPPER BOUNDS OF LIFETIME EXTRARISK TO DEVELOP AN OBSERVABLE TUMOR PER 1,000 WORKERS DUE TO AN 8-HOUR TWA FOR AN OCCUPATIONALLIFETIME a OF EXPOSURE TO 1,3-BUTADIENE, USING NTP II BIOASSAY b AND THE BEST FITTING WEIBULL TIME-TO-TUMOR MODELS

Neoplasms Stages

8-hour time-weighted average concentration c

0.1 ppm 0.2 ppm 0.5 ppm 1 ppm 2 ppm 5 ppm

MLE 95%U.B.d MLE 95%

U.B. MLE 95%U.B. MLE 95%

U.B. MLE 95%MLE MLE 95%

U.B.

Male mice:Heart Hemangiosarcoma ..... 2 e<0.1 0.2 e<0.1 0.4 e<0.1 0.9 e<0.1 1.8 e< 0.1 3.6 0.4 9.1All lymphoma ........................ 1 0.1 0.2 0.3 0.5 0.6 1.1 1.3 2.3 2.5 4.5 6.3 11.2Lung tumor ........................... 1 0.7 0.1 1.3 2.0 3.2 4.9 6.4 9.8 12.8 19.4 31.7 47.9

Female mice:Heart Hemangiosarcoma ..... 3 f<0.1 f<0.1 f<0.1 < 0.1 f<0.1 0.2 f< 0.1 0.5 f<0.1 1.0 f<0.1 2.4Ovarian tumor ...................... 1 0.1 0.3 0.3 0.5 0.7 1.3 1.4 2.6 2.8 5.2 6.9 13.0All lymphoma ........................ 1 0.6 0.9 1.2 1.8 3.0 4.6 6.0 9.2 12.0 18.3 29.7 45.0Lung tumor ........................... 1 0.8 1.2 1.6 2.4 4.1 6.1 8.1 12.2 16.2 24.1 40.00 59.4

a Occupational lifetime, working 5 days/week, 50 weeks/year, for 45 years.b Using data from NTP II for the following exposure groups: 0, 6.25, 20, 62.5 and 200 ppm; the 625 ppm exposure group was excluded.c Estimated lifetime excess risk for cancer assuming: mouse life-span of 113 weeks, male mouse body weight of 40.8g; female mouse body

weight of 38.8 g; worker’s breathing rate is 1.25 m3/hr; mouse to human risk extrapolated in mg/kg-day equivalent units.d 95% U.B., 95% Upper Bounds is the largest value of excess risk that is compatible with the animal response data at a confidence level of

95%.e MLEs ranged from 1.5µ10¥4 to 6.0µ10¥2

f MLEs ranged from 3.4µ10¥8 to 4.3µ10¥3

VII. Significance of Risk

A. IntroductionIn the 1980 ‘‘Benzene Decision,’’ the

Supreme Court, in its discussion of thelevel of risk that Congress authorizedOSHA to regulate, indicated its view ofthe boundaries of acceptable andunacceptable risk. The Court stated:

It is the Agency’s responsibility todetermine in the first instance what itconsiders to be a ‘‘significant’’ risk. Somerisks are plainly acceptable and others areplainly unacceptable. If for example, theodds are one in a billion that a person willdie from cancer by taking a drink ofchlorinated water, the risk clearly could notbe considered significant. On the other hand,

if the odds are one in a thousand that regularinhalation of gasoline vapors that are 2percent benzene will be fatal, a reasonableperson might well consider the risksignificant and take the appropriate steps todecrease or eliminate it. (I.U.D. v. A.P.I., 448U.S. 607, 655).

So a risk of 1⁄1000 (10¥3) is clearlysignificant. It represents the uppermost


end of the million-fold range suggestedby the Court, somewhere below whichthe boundary of acceptable versusunacceptable risk must fall.

The Court further stated that ‘‘whilethe Agency must support its findingsthat a certain level of risk exists withsubstantial evidence, we recognize thatits determination that a particular levelof risk is significant will be basedlargely on policy considerations.’’ Withregard to the methods used to determinethe risk level present (as opposed to thepolicy choice of whether that level is‘‘significant’’ or not), the Court addedthat assessment under the OSH Act is‘‘not a mathematical straitjacket,’’ andthat ‘‘OSHA is not required to supportits findings with anything approachingscientific certainty.’’ The Court ruledthat ‘‘a reviewing court [is] to giveOSHA some leeway where its findingsmust be made on the frontiers ofscientific knowledge [and that] * * *the Agency is free to use conservativeassumptions in interpreting the datawith respect to carcinogens, riskingerror on the side of overprotectionrather than underprotection’’ (448 U.S.at 655, 656).

Nonetheless, OSHA has taken varioussteps that make it fairly confident itsrisk assessment methodology is notdesigned to be overly ‘‘conservative’’ (inthe sense of erring on the side ofoverprotection). For example, there areseveral options for extrapolating humanrisks from animal data via interspeciesscaling factors. The plausible factorsrange at least as widely as from bodyweight extrapolation at one extreme(risks equivalent at equivalent bodyweights, (mg/kg) 1) to (body weight) 2/3

(risks equivalent at equivalent surfaceareas) at the other. Intermediate valueshave also been used, and the value of(body weight) 3/4, which is supported byphysiological theory and empiricalevidence, is generally considered to bethe midpoint of the plausible values.(Body weight) 2/3 is the mostconservative value in this series, whilebody weight extrapolation is the leastconservative. OSHA has generally usedbody weight extrapolation in assessingrisks from animal data, an approach thattends to be significantly less riskconservative than the othermethodologies and is likely to be lessconservative even than the centraltendency of the plausible values.

Other steps in OSHA’s riskassessment methodology where theAgency does not use the mostconservative approach are selection ofthe maximum likelihood estimate (MLE)of the parameterized dose-responsefunction rather than selection of theupper 95% confidence limit, and the

use of site-specific tumor incidence,rather than pooled tumor response, indetermining the dose-response functionfor a chemical agent.

Other aspects of OSHA’s riskassessment methodology reflect moreconservative choices, including: basingthe risk estimate on the more sensitivespecies tested (the mouse); includinglung tumors in the range of riskspresented in the quantitative analysis,even though excess deaths from lungcancer have not been observed in any ofthe human studies; and, assumingworkers will be exposed to butadiene atthe maximum permissible level for 45years. As discussed below, if workersare exposed to BD for fewer years, theirestimated risks from BD will be lessthan indicated. This caveat, of course,does not address lifetime risks takinginto account occupational exposure toother substances encountered at otherjobs. For reasons already explained,OSHA believes these choices areappropriate for the BD risk assessment.OSHA also recognizes that use of themost conservative approach at everystep of the risk assessment analysiscould produce mathematical riskestimates which, because of the additiveeffect of multiple conservativeassumptions, may overstate the likelyrisk. OSHA believes its quantitative riskassessment for BD strikes an appropriatebalance.

Risk assessment is only one part ofthe process OSHA uses to regulate toxicsubstances in the workplace. OSHA’soverall analytic approach to regulatingoccupational exposure to particularsubstances is a four-step processconsistent with judicial interpretationsof the OSH Act, such as the BenzeneDecision, and rational policyformulation. In the first step, OSHAquantifies the pertinent health risks, tothe extent possible, performingquantitative risk assessments. TheAgency considers a number of factors todetermine whether the substance to beregulated currently poses a significantrisk to workers. These factors includethe type of risk posed, the quality of theunderlying data, the plausibility andprecision of the risk assessment, thestatistical significance of the findingsand the magnitude of risk. (48 FR 1864,January 14, 1983) In the second step,OSHA considers which, if any, of theregulatory options being considered willsubstantially reduce the identified risks.In the third step, OSHA looks at the bestavailable data to set permissibleexposure limits that, to the extentpossible, both protect employees fromsignificant risks and are alsotechnologically and economicallyfeasible. In the fourth and final step,

OSHA considers the most cost-effectiveway to fulfill its statutory mandate bycrafting regulations that allowemployers to reach the feasible PEL asefficiently as possible.

B. Review of Data Quality andStatistical Significance

As discussed in the Health Effectssection, OSHA has concluded thatbutadiene is a probable humancarcinogen. This conclusion is based ona body of evidence comprised of animalbioassays, human epidemiologicalinvestigations, and other experimentalstudies that together are both consistentin their findings and biologicallyplausible. First, OSHA has reviewedfour rodent inhalation bioassays, twomouse bioassays conducted under theNational Toxicology Program(designated NTP I and NTP II), a mousestudy by Irons et al. in 1989, and a ratstudy sponsored by the IISRP. (Exs. 2–32, 23–1, 32–28D, 90, 96) All threemouse studies found a consistently hightumor response in BD-exposed mice,relative to control animals. Severaltarget organs were identified,particularly by the NTP II study;however, all three studies found dose-related increases in the incidences oflymphocytic lymphoma and hearthemangiosarcomas associated withexposure to BD. Most significantly, theNTP II study reported statisticallysignificant increases in tumor incidenceamong mice exposed to BD well belowOSHA’s current PEL of 1,000 ppm(exposure to as low as 6.25 ppm wasassociated with a statistically significantincrease in tumors, e.g., lung tumors infemale mice). There was also evidencefor a dose-rate effect, meaning that theobserved tumor incidence in miceexposed to high concentrations overshort periods of time was higher thanthat observed in mice administered anequivalent cumulative concentrationover a long period of time. The studyemploying BD-exposed rats also foundincreased incidences of several types ofcancer, albeit at lower response ratesthan were observed in the mousestudies. The two major epoxidemetabolites of BD have also been shownto be carcinogenic in rats and mice.

OSHA has also reviewed a number ofhuman epidemiological studies thathave examined the mortality experienceof styrene-butadiene rubber (SBR)workers. These studies haveconsistently reported an elevatedrelative risk of leukemia-or lymphoma-related death among BD-exposedworkers. The most recent of these, thestudy by Delzell et al., updated andexpanded previous SBR workermortality studies and found a positive


and statistically significant dose-response relationship betweencumulative exposure to BD andincreased leukemia mortality, whichremained statistically significant evenafter controlling for the potentialconfounder of concurrent styreneexposure. (Ex. 117–1) The Delzell et al.study thus provides further and moredirectly relevant evidence that anincreased risk of leukemia-related deathis associated with exposure to BD.Furthermore, other epidemiologicstudies have reported finding anunusually short latency period (as littleas 3 to 4 years from time of initialexposure to death) for exposure-relatedhematologic malignancies amongworkers who experienced exposures toBD in the past that were higher thanexposures that prevail today. (Ex. 2–26,3–34 Vol III H–1)

Evidence for the carcinogenicity of BDis further strengthened by a collection ofstudies showing that the epoxidemetabolites of BD are mutagenic in awide variety of in vitro and in vivo testsystems. Examination of culturedlymphocytes from BD-exposed workershas revealed the presence ofchromosome aberrations, an elevatedfrequency of chromatid breaks, andvarious mutations, thereby providingdirect evidence of genotoxicity inoccupationally-exposed humans. (Exs.118–2A, 118–2D) Furthermore, thefinding of activated K-ras oncogenes intumors of BD-exposed mice providesadditional support for a mutagenicmode of action; this finding hasparticular relevance to human risk inthat K-ras is the most commonlydetected oncogene in human cancer.(Ex. 129)

The findings from the animalbioassays and human epidemiologicstudies identify the hematopoieticsystem as a primary target organ for BD-related carcinogenesis. Target organs fortoxicity are not necessarily those forcarcinogenicity. Other experimentalfindings are consistent with theseobservations. Studies in BD-exposedrodents have found concentration-dependent decreases in red blood cellcounts, hemoglobin concentration, andother indicators of hematopoieticsuppression. (Exs. 114, 32–38D, 23–12)There is also some suggestive evidencethat workers exposed to BD at levelswell below the current 1,000 ppm PELexhibit hematological changesindicative of bone marrow depression.(Exs. 23–4, 2–28) Finally, many of thetumor types found in BD-exposed mice,including lymphocytic/hematopoieticcancer, lung cancer, mammary glandtumors, and possiblyhemangiosarcomas, are tumors that are

often found in association withexposure to other industrial chemicalsknown to cause lymphocytic/hematopoietic cancer in humans. Thus,OSHA finds that the body of scientificstudies contained in the BD record,which includes well-conducted animalbioassays, human epidemiologicstudies, and other experimentalinvestigations, provides convincingevidence that BD is a probable humancarcinogen.

This view is also held by otherscientific organizations that haveexamined some or all of the sameevidence. EPA considers BD to be aprobable human carcinogen, and NIOSHregards BD as a potential occupationalcarcinogen and recommends controllingexposures to the lowest feasible level. In1983, based on the findings of the firstNTP bioassay alone, ACGIH classifiedBD as an animal carcinogen and, in thefollowing year, recommended a newTLV of 10 ppm. In 1992, before theDelzell et al. study was released, IARCclassified BD as a probable humancarcinogen (Group 2A).

As discussed in the Quantitative RiskAssessment section, OSHA has selectedthe NTP II mouse bioassay forquantitative assessment of cancer risksfor several reasons. Chief among these isthat the NTP II study was conducted atBD concentrations that arerepresentative of current exposureconditions and that the resultsdemonstrated a strong dose-responserelationship for several cancer sites. Inaddition, the study is of very highquality and pathology results fromindividual animals were available to theAgency, enabling OSHA to use a time-to-tumor model that could account forthe early cancer-related deaths thatoccurred among the test animals(competing risks). OSHA also chose tobase its risk estimates on the dose-response relationships for three cancertypes: lung, ovarian, and lymphoma.The incidence of each was significantlyelevated. It should be noted that poolingthe total number of animals having anyof these tumor types would haveyielded risk estimates higher thanOSHA’s final values.

Because data were available onindividual animals, including time ofdeath, OSHA chose to use a Weibulltime-to tumor form of the multistagemodel based on the biologicalassumption that cancer is induced bycarcinogens through a series of events.This model has the advantage ofaccounting for competing risks.

The multistage model is mostfrequently used by OSHA; it is also amechanistic model based on thebiological assumption that cancer is

induced by carcinogens through a seriesof independent stages. The model maybe conservative, because it assumes nothreshold for carcinogenesis andbecause it is approximately linear at lowdoses, although there are other plausiblemodels of carcinogenesis which aremore conservative. The Agency believesthat the multistage model conformsmost closely to what we know about theetiology of cancer, including the factthat linear-at-low-dose behavior isexpected for exogenous agents, whichincreases the risk of cancer alreadyposed by similar ‘‘background’’processes. There is no evidence that themultistage model is biologicallyincorrect and abundant evidencesupports its use, especially for genotoxiccarcinogens, a category that most likelyincludes BD. OSHA’s preference isconsistent with the position of theOffice of Science and Technology Policyof the Executive Office of the President,which recommends that ‘‘when dataand information are limited, and whenmuch uncertainty exists regarding themechanisms of carcinogenic action,models or procedures that incorporatelow-dose linearity are preferred whencompatible with limited information.’’(OSTP, Chemical Carcinogens: AReview of the Science and ItsAssociated Principles. Federal Register,March 14, 1985, p. 10379)

The BD record contained a great dealof commentary on the possible role ofthe principal epoxide metabolites of BDon the development of cancer in testanimals, and on whether differences inBD metabolism, distribution, andexcretion can explain the observeddifferences in cancer responses betweenBD-exposed mice and rats. In evaluatingthis information, OSHA explored thepossibility of using a physiologically-based pharmacokinetic (PBPK)approach to estimate cancer risk amongBD-exposed workers. In considering theuse of PBPK modeling for estimatingequivalent human dose in its final riskassessment for BD, OSHA consideredseveral preselected criteria for judgingwhether the available data was adequateto permit OSHA to rely on a PBPKanalysis in place of administeredexposure levels. These are the samecriteria that OSHA has recently used torely on a PBPK-based analysis in its riskassessment of methylene chloride. Thecriteria included the following:

1. The predominant and all relevantminor metabolic pathways must be welldescribed in several species, includinghumans.

2. The metabolism must be adequatelymodeled.


7 A dose metric is the way in which dose isexpressed in describing a dose-responserelationship. A dose metric may be expressed as anapplied dose, such as ppm concentration or mg ofintake per kg body weight, or as an internal dose,such as mg per gram wet weight of an organ or mgof total metabolite formed per kg body weight.

3. There must be strong empiricalsupport for the putative mechanism ofcarcinogenesis.

4. The kinetics for the putativecarcinogenic metabolic pathway musthave been measured in test animals invivo and in vitro and in correspondinghuman tissues at least in vitro.

5. The putative carcinogenicmetabolic pathway must containmetabolites that are plausible proximatecarcinogens.

6. The contribution to carcinogenesisvia other pathways must be adequatelymodeled or ruled out as a factor.

7. The dose surrogate in target tissuesused in PBPK modeling must correlatewith tumor responses experienced bytest animals.

8. All biochemical parameters specificto the compound, such as blood:airpartition coefficients, must have beenexperimentally and reproduciblymeasured. This must especially be truefor those parameters to which the PBPKmodel is sensitive.

9. The model must adequatelydescribe experimentally measuredphysiological and biochemicalphenomena.

10. The PBPK models must have beenvalidated with other data (includinghuman data) that were not used toconstruct the models.

11. There must be sufficient data,especially data from a broadlyrepresentative sample of humans, toassess uncertainty and variability in thePBPK modeling.

For the BD risk assessment, OSHA haschosen to use for animal-to-human doseequivalency mg/kg-day uptake based onthe ppm exposure levels in the NTP IImouse study as the dose-metric.7 Whilethe body of data in the record leadsOSHA to conclude that metabolism ofBD to active metabolites is probablynecessary for carcinogenicity, OSHA haschosen total body uptake rather thanorgan metabolic levels because theAgency was unable to determine fromthe record (a) which of the activemetabolites are responsible for whichobserved tumors in the mice, (b) whatthe mouse and human metabolicequivalent doses were, (c) whether anyof the PBPK models can successfullycorrelate with the tumor responsesobserved in mice and rats, and (d)whether local reactions in the mouseand human bone marrow were moreimportant than total body burden.

OSHA would have considered using BDmetabolite body burden based on totalhuman BD metabolites if the humanchamber concentration data had beenavailable, which would supportestimating total human BD metabolism.Data of this type were available andused in OSHA’s PBPK modeling formethylene chloride. In the absence ofhuman chamber data or some betterestimate of human equivalent dose,OSHA has chosen to use mg/kg-day BDuptake from the ppm inhalationexposure levels in the NTP II mousebioassay as suitable for animal-to-human equivalency.

C. Material Impairment of Health

The 1 ppm 8-hour TWA PEL isdesigned to reduce cancer risks amongexposed workers. As mentioned aboveand in the Health Effects section, someepidemiological studies indicate thatthe increased risk of leukemia posed byBD exposure may occur within a shortperiod after initial exposure. (This issupported by the NTP mouse bioassays,in which there was high early mortalityresulting from the development of BD-induced cancers, especiallylymphomas.) Therefore, OSHA believesthese hematopoietic cancers are likely tobe fatal, will result in substantiallyshortened worker lifespans, and clearlyrepresent ‘‘material impairment ofhealth’’ as defined in the OSH Act andcase law.

OSHA has also concluded thatexposure to BD is associated with apotential risk of adverse reproductiveeffects in both males and females. Thisconclusion is based on the two NTPanimal bioassays, which foundtesticular atrophy in male mice exposedto 625 ppm BD and ovarian atrophy infemale mice exposed to BDconcentrations as low as 6.25 ppm, aswell as other animal studies that havereported dominant lethal effects(indicating a genotoxic effect on germcells) and abnormal sperm morphologyin BD-exposed male mice. (Exs. 23–74,23–75, 117–1) There is also evidencethat BD exposure is associated withfetotoxicity in mice, and a teratogeniceffect indicative of a transplacentallyinduced somatic cell mutation wasobserved in one mouse study. (Exs. 2–32, 23–72, 126) OSHA believes thatteratogenic effects and gonadal atrophywould also unambiguously constitute‘‘material impairment of health.’’Furthermore, although OSHA did notquantify reproductive risks that may beassociated with exposure to BD, OSHAbelieves that reducing the 8-hour TWAPEL from 1,000 ppm to 1 ppm is likelyto substantially reduce this risk.

D. Risk Estimates

OSHA’s final estimate of excesscancer risks associated with exposure to5 ppm BD (8-hour TWA) ranges from11.2 to 59.4 per 1000, based onlymphomas, lung tumors and ovariantumors seen in the NTP II mouse study(OSHA did not estimate the risksassociated with exposure to the currentPEL of 1,000 ppm, since workers arerarely, if ever, exposed to BD levels ofthat magnitude). Based on linear modelsthe estimated risks at the new PEL of 1ppm range from 1.3 to 8.1 per 1000,which represents a substantial reductionin risk from those associated withexposures to 5 ppm or greater.

OSHA’s risk estimates for the 1 ppmPEL are similar in magnitude to, orlower than, most of the estimatescontained in several risk assessmentssubmitted to the BD record, whichutilized a variety of models and dosemetrics. Furthermore, NIOSH’squantitative assessment based on theDelzell et al. epidemiologic study ofSBR workers yielded an estimate of 8cancer deaths per 1,000 workersexposed to 1 ppm BD, a figure that is inclose agreement with the upper end ofthe range of risks predicted by OSHA.

Risks greater than or equal to 10¥3 (1per 1,000) are clearly significant and theAgency deems them unacceptably high.OSHA concludes that the new BDstandard substantially lowers risk butdoes not reduce risk below the level ofinsignificance. The estimated levels ofrisk at 1 ppm are 1.3 to 8.1 per 1000.The ancillary provisions including theexposure goal program will furtherreduce risk from exposure to BD.

E. ‘‘Significant Risk’’ Policy Issues

Further guidance for the Agency inevaluating significant risk andnarrowing the million-fold rangedescribed in the ‘‘Benzene Decision’’ isprovided by an examination ofoccupational risk rates, legislativeintent, and the academic literature on‘‘acceptable risk’’ issues. For example,in the high risk occupations of miningand quarrying, the average risk of deathfrom an occupational injury or an acuteoccupationally-related illness over alifetime of employment (45 years) is15.1 per 1,000 workers. The typicaloccupational risk of deaths for allmanufacturing industries is 1.98 per1,000. Typical lifetime occupational riskof death in an occupation of relativelylow risk, like retail trade, is 0.82 per1,000. (These rates are averages derivedfrom 1984–1986 Bureau of LaborStatistics data for employers with 11 ormore employees, adjusted to 45 years ofemployment, for 50 weeks per year).


Congress passed the OccupationalSafety and Health Act of 1970 becauseof a determination that occupationalsafety and health risks were too high.Congress therefore gave OSHA authorityto reduce significant risks when it isfeasible to do so. Within this context,OSHA’s final estimate of risk fromoccupational exposure to BD at levels of2 ppm (2.5 to 16.2 deaths per 1,000workers) or higher is substantiallyhigher than other risks that OSHA hasconcluded are significant, issubstantially higher than the risk offatality in some high-risk occupations,and is substantially higher than theexample presented by the SupremeCourt in the benzene case. Moreover, arisk in the range of 1.3 to 8.1 per 1000at 1 ppm is also clearly significant;therefore, the PEL must be set at least aslow as the level of 1 ppm documentedas feasible across all industries.

Because of technologic feasibilityconsiderations, OSHA could notsupport promulgating a PEL below 1ppm. However OSHA has integratedother protective provisions into the finalstandard to further reduce the risk ofdeveloping cancer among employeesexposed to BD.

Based on OSHA’s QRA, employeesexposed to BD at the 8-hour TWA PELlimit, without the benefit of thesupplementary provisions, wouldremain at significant risk of developingadverse health effects, so that inclusionof other protective provisions, such asmedical surveillance and employeetraining, is both necessary andappropriate. The exposure goal programand action level trigger incorporatedinto the standard will encourageemployers to lower exposures below 0.5ppm to further reduce significant risk ifit is feasible to do so in theirworkplaces. Consequently, the programstriggered by the action level will furtherdecrease the incidence of diseasebeyond the predicted reductionsattributable merely to a lower PEL.

As OSHA has explained, numerousissues arise in quantifying estimatedrisk to workers from BD. Such estimatesare thus inherently uncertain; and, asmore information becomes available,some of that uncertainty may beaddressed and may substantially alterthe risk estimate. Although OSHAbelieves the estimates fulfill its legalobligation to provide substantialevidence of significant risk the estimatesshould not be interpreted as a precisequantification of the cancer riskassociated with the new PEL, or asdemonstrated evidence of actual workerdisease caused by BD.

OSHA’s determination of significantrisk is predicated, consistent with

empirical evidence and the legalmandates of the OSHA Act, ondetermining the risk to a workerexposed to BD for a working lifetime (45years) at the PEL. To the extent thatfuture exposures to BD are(substantially) lower than 1 ppm, theestimated risks associated with thoseexposures will be (substantially) lowerthan the range presented in OSHA’sQRA.

OSHA believes the final standard willreduce the risks of BD below thoseestimated using the mathematicalmodel. The estimates of risk consideronly exposures at the PEL, and do nottake fully into account the otherprotective provisions of the standardsuch as medical surveillance, hazardcommunication, training, monitoring,and the exposure goal program. Thedecrease in risk to be achieved byadditional provisions cannot beadequately quantified beyond adetermination that they will add to theprotection provided by the lower PELalone. OSHA has determined thatemployers who fulfill the provisions ofthe standard as promulgated willprovide protection for their employeesfrom the hazards presented byoccupational exposure to BD wellbeyond those which would be indicatedsolely by reduction of the PEL.

Furthermore, as discussed above andin the Health Effects section, there isevidence from the NTP bioassays thatexposure to periodic highconcentrations of BD may be associatedwith a higher cancer risk compared toan equivalent cumulative exposureadministered over a longer time frame.OSHA has included a 5 ppm short-termexposure limit (STEL), averaged over 15minutes, to provide protection toemployees who are exposed to elevatedBD concentrations during brief periods,such as in maintenance work.

As a result, OSHA concludes that its8-hour TWA PEL of 1 ppm andassociated action level (0.5 ppm) andSTEL (5 ppm) will reduce significantrisk and that employers who complywith the other provisions of thestandard will be taking feasible,reasonable, and necessary steps to helpprotect their employees from thehazards of BD.

VIII. Summary of the Final EconomicAnalysis

As required by Executive Order 12866and the Regulatory Flexibility Act of1980 (as amended 1996), OSHA hasprepared a Final Economic Analysis toaccompany the final standard foroccupational exposure to 1,3-butadiene(BD). (The entire analysis, withsupporting appendix material, has been

placed in the BD rulemaking docket. SeeExhibit 137.) The purpose of the finaleconomic analysis is to:

• Describe the need for a standardgoverning occupational exposure to 1,3-butadiene;

• Identify the establishments andindustries potentially affected by thestandard;

• Evaluate the costs, benefits,economic impacts and small businessimpacts of the standard on affectedfirms;

• Assess the technological andeconomic feasibility of the standard foraffected establishments, industries, andsmall businesses; and

• Evaluate the availability of effectivenon-regulatory approaches to theproblem of occupational exposure to1,3-butadiene.

Need for the StandardOSHA’s final BD standard covers

occupational exposures to thissubstance, a high-volume chemical usedprincipally as a monomer in themanufacture of a wide range ofsynthetic rubber and plastic polymersand copolymers. In all, about 9,700employees are estimated to be exposedto BD. However, for 2,100 of theseemployees in the petroleum refiningindustry, BD exposures are below theaction level. The largest group ofexposed workers is found in the BDend-product industry. Other BDoperations in which workers areexposed are crude BD production, BDmonomer production, andtransportation terminals handling BDmonomers (stand-alone terminals).

There is strong evidence thatworkplace exposure to BD poses anincreased risk of cancer. Animalbioassays have shown BD to be a sourceof significant risk for tumors at multiplesites (i.e. lung tumors, hearthemangiosarcomas, lymphomas andovarian tumors). BD may alsopotentially cause both male and femalereproductive effects. To protect all BD-exposed workers from these adversehealth effects, the final standard lowersthe airborne concentration of BD towhich workers may be exposed from thecurrent permissible exposure limit (PEL)of 1,000 ppm as an 8-hour time-weighted average (8-hour TWA) to 1ppm, and adds a short term exposurelimit (STEL) of 5 ppm, measured over15 minutes. (For a detailed discussion ofthe risks posed to workers fromexposure to BD, see the QuantitativeRisk Assessment and Significance ofRisk sections of the preamble, above.)

OSHA’s final BD standard is similarin format and content to other healthstandards issued under Section 6 (b)(5)


of the Act. In addition to PELs, thestandard requires employers to monitorthe exposures of workers; establishregulated areas when exposures mayexceed one of these PELs; implementengineering and work practice controlsto reduce employee exposures to BD;develop an exposure goal program;provide respiratory protection tosupplement engineering controls wheresuch controls are not feasible, areinsufficient to meet the PELs, arenecessary for short infrequent jobs, or inemergencies; provide medical screening;train workers about the hazards of BD(also required by OSHA’s HazardCommunication Standard); and keeprecords relating to the BD standard.Recognizing that workers exposed to BDare at significant risk, an industry-laborworking group joined together todevelop joint recommendations for thefinal standard for BD. This group’srecommendations form the basis forOSHA’s final rule. The contents of thestandard are explained briefly inChapter I of the final economic analysisand in detail in the Summary andExplanation (Section X of the preamble,below).

Chapter II of the economic analysisdescribes the uses of BD and theindustries in which such use occurs.Exposure to 1,3-butadiene occurs as aresult of exposure to the monomer.Once BD is in polymer form, theexposure is minimal to non-existent. Inall, OSHA analyzed 5 types of processesin which BD exposure occurs: crude BDproduction, where the feedstock for BDmonomer is produced; BD monomerproduction, in which BD is refined fromcrude BD to a 99 percent pure monomer;BD product manufacture, where BDmonomer is converted to variouspolymer products; stand-aloneterminals, which receive, store anddistribute BD monomer; and petroleumrefineries, where BD may occur as anunwanted byproduct in some types ofrefining units. Table VIII–1 shows theseindustry operations and the number ofworkers affected by the final rule. Atotal of 255 facilities are estimated to bepotentially affected by the standard.These establishments employ 9,700

workers who are estimated to beexposed to BD in the course of theirwork. The industry operation with thelargest number of directly exposedemployees is BD product manufacture,which has 6,500 exposed employees(over two-thirds of the total).

TABLE VIII–1.—INDUSTRY OPERATIONSAND NUMBER OF WORKERS AF-FECTED BY THE FINAL RULE FOR1,3-BUTADIENE

Number ofaffectedworkers

Number offacilities inindustry a

Crude 1,3-Buta-diene Produc-tion ................. 540 27

1,3-ButadieneMonomer Pro-duction ........... 552 12

1,3-ButadienePolymer Prod-uct Manufac-ture ................ 6,461 c 71

Standard-AloneTerminals ....... 50 5

Subtotal ...... 7,603 115

Petroleum Refin-ing Sector ...... b 2,100 140

Total ........... 9,703 255

Source: U.S. Department of Labor, OSHA,Office of Regulatory Analysis, 1996.

a Some facilities may fall under several in-dustry sectors. For example, 9 monomer facili-ties are also crude producing facilities.

b Potential exposures to 1,3-butadiene arelow and of extremely short duration in refining.

c Represents number of processes and notnecessarily plants.

Chapter III of the analysis assesses thetechnological feasibility of the finalstandard’s requirements, andparticularly its PELs, for firms in the 5industry operations with employeeexposure identified in the IndustryProfile. OSHA finds, based on ananalysis of exposure data taken onworkers performing the BD-related tasksidentified for each operation, thatcompliance with the standard istechnologically feasible forestablishments in the industries studied.With few exceptions, employers will be

able to achieve compliance with bothPELs through the use of engineeringcontrols and work practices. The fewexceptions are maintenance activities,such as vessel cleaning, which havetraditionally often involved the use ofrespiratory protection.

The exposure data relied on by OSHAin making its technological feasibilitydeterminations were gathered by NIOSHin a series of site visits to plants in theaffected industries. These data showthat many facilities in the affectedindustries have already achieved thereductions in employee exposuresrequired by the final rule. At least someworkers in every job category work infacilities that have already achieved thePEL requirements. OSHA’s analysis oftechnological feasibility evaluatesemployee exposures at the operation ortask level to the extent that such dataare available. In other words, theanalysis identifies relevant exposuredata on a job category-by-job categorybasis to permit the Agency to pinpointthose BD-exposed workers and joboperations that are not yet under goodprocess control and will thus needadditional controls (including improvedhousekeeping, maintenance procedures,and employee work practices) toachieve compliance. Costs are thendeveloped (in Chapter V of theeconomic analysis) for the improvedcontrols needed to reach the new levels.

The benefits that will accrue to BD-exposed employees and theiremployers, and thus to society at large,are substantial and take a number offorms. Chapter IV of the analysisdescribes these benefits, both inquantitative and qualitative form. At thecurrent baseline exposure levels to BD,the risk model estimates that 76 cancerdeaths will be averted over a 45-yearperiod. By reducing the total number ofBD-related cancer deaths from 76 deathsto 17 deaths over 45 years, the standardis projected to save an average of 1.3cancer deaths per year. Table VIII–2shows these risk estimates. In additionto cancer deaths, the standard mayprevent male and female reproductiveeffects.

TABLE VIII–2.—WORKER EXPOSURE TO BD AND LUNG CANCER RISK OVER 45 YEARS AT CURRENT EXPOSURE LEVELSAND LEVELS EXPECTED UNDER THE STANDARD

8-hour time weighted average (ppm)

0–0.5 0.5–1.0 1 1.0–2.0 2.0–5.0 5.0–10.0 10+c Total

Lifetime Excess Cancer Risk (per thousand workers)a ................ 2.05 6.1 8.1 12.15 28.1 60 480 ............Baseline Number of Workers Exposed ......................................... 5697 354 156 598 320 440 38 7603Estimated Excess Deaths in Baseline (Existing PEL)b ................. 12 2 1 7 9 27 18 76Predicted Number of Workers Exposed at New PEL ................... 7177 426 0 0 0 0 0 7603

final rule occupational exposure to 1,3-butadiene ... · 56746federal register/vol. 61, no....

Documents