Appropriate Levels of Cytotoxicity for Genotoxicity TestsUsing Mammalian Cells In Vitro
Lutz Muller1* and Toshio Sofuni21Federal Institute for Drugs and Medical Devices, Berlin, Germany
2National Institute of Health Sciences, Tokyo, Japan
Among standard battery genotoxicity assays, thein vitro chromosome aberration test and the mouselymphoma tk assay (MLA) yield about fourfoldhigher incidences of positive test results than thebacterial reverse mutation test or in vivo bone mar-row tests. This is a result of experience with sub-missions of 335 new pharmaceuticals to the Ger-man Federal Institute for Drugs and MedicalDevices. While all of the standard systems havetheir value in detecting relevant genotoxins, thereis no supportive evidence for DNA reactivity for aconsiderable number of in vitro clastogens andMLA positives. In particular the clastogenic re-sponse of such compounds is often associated withhigh cytotoxicity. This may invoke the need to
change the approach to test for clastogenicity invitro. A combination of measures such as (1) achange in the upper limits of cytotoxicity that arecurrently given in International Conference on Har-monisation (ICH) and Organization for EconomicCo-Operation & Economic Development (OECD)guidelines, (2) the creation of a common ground ofunderstanding for interpretation of in vitro (posi-tive) test results, and (3) lowering the upper limits oftest compound concentration irrespective of cyto-toxicity may prove useful to ensure a sufficientreliability of genotoxicity testing with mammaliancells in vitro. Environ. Mol. Mutagen. 35:202–205,2000. © 2000 Wiley-Liss, Inc.
Key words: cytotoxicity in vitro; mammalian cells; threshold; chromosome aberrations
INTRODUCTION
The analysis of standard genotoxicity test data that havebeen conducted for safety evaluation of chemical com-pounds usually reveals a fairly high percentage of positiveresults, in particular in the in vitro chromosome aberrationtest and in the mouse lymphoma tk assay (MLA). Table Iillustrates this on the basis of experience with submissionsfor registration of 335 new pharmaceutical entities to theGerman Federal Institute for Drugs and Medical Devices.Both the in vitro chromosome aberration test and the MLAyielded positive results for more than 25% of the pharma-ceuticals tested. The percentage of positives in this databaseis somewhat higher than the 18% positives/equivocals re-ported for the aberration test by Galloway (2000) for a totalof 1,420 compounds tested. One reason may be differentcriteria for a “positive” call.
The positive results in mammalian cells in vitro forpharmaceutical compounds are reflecting a variety of com-pound groups with different mechanisms of action. Tradi-tional alkylating agents or other compounds with knowngenotoxic potential, such as nucleoside analogues or gyraseinhibitors are only a fraction (approximately 25%) of thosethat yielded positive test results, in particular in mammaliancells in vitro. For quite a substantial percentage of com-pounds, there was no immediate obvious and convincingrationale that would have helped to reason for the positivetest results. Yet, distinct cytotoxicity was frequently noted
at concentrations that showed an increased incidence ofchromosome aberrations or an increase in mutation fre-quency for the MLA. The overwhelming majority of thesecompounds yielded negative results in the in vivo part of thegenotoxicity battery and was negative regarding genotoxi-city related tumor findings in rodent 2-year bioassays. Afull review of the data is provided in Mu¨ller and Kasper[2000].
Although genotoxicity is usually associated with cytotox-icity, most known genotoxic carcinogens will yield positiveeffects in the in vitro assays already at low levels of cyto-toxicity but proceeding well into the cytotoxic range [Gal-loway et al., 1998; Galloway, 2000]. On the other hand, aclose association of genotoxicity, in particular chromosomedamage, with high toxicity has in some cases been evaluatedas nonrelevant or thresholded genotoxicity, although theexact mechanisms of action have not been found in mostcases [Mu¨ller et al., 1991; Armstrong et al., 1992; Kirkland,1992; Galloway et al., 1998; Hilliard et al., 1998; Kirklandand Muller, 2000; Muller and Kasper, 2000].
Among the various groups of new pharmaceutical com-pounds with structural and/or pharmacological similaritiesthat clearly display positive effects in the in vitro chromo-
T. Sofuni is currently at Olympus Optical Co. Ltd., Tokyo, Japan.
*Correspondence to: Lutz Mu¨ller, Federal Institute for Drugs and MedicalDevices, Seestr. 10, 13353 Berlin, Germany. E-mail: [email protected]
Environmental and Molecular Mutagenesis 35:202–205 (2000)
© 2000 Wiley-Liss, Inc.
some aberration test, two may be of particular interestbecause of their widespread medical use in recent years: (1)angiotensin-converting enzyme (ACE) inhibitors and (2)angiotensin II receptor (AIIR) antagonists. In Germany,there are nine ACE inhibitors and seven AIIR antagonistsmarketed. Among them, five ACE inhibitors and five AIIRantagonists induced chromosome aberrations in vitro inhuman lymphocytes or established cell lines with differentlevels of potency; one ACE inhibitor yielded up to 91%aberrant metaphases. All compounds showed marked reduc-tion in cell count, mitotic index, culture confluency, or BrdUuptake at concentrations yielding chromosome damage.There was no evidence for genotoxicity in bacteria or invivo. Furthermore, rodent carcinogenicity studies did notyield tumors that might be genotoxicity related. DNA syn-thesis inhibition has been suggested as one possible mech-anism supporting an indirect mechanism of aberration in-duction for AIIR antagonists [Galloway et al., 1998]. Inaddition to these examples, an extensive review of geno-toxicity data of paracetamol (synonymous to acetamino-phen) provides evidence that widely used compounds canbe convincingly genotoxic, i.e., inducers of chromosomedamage in vitro and in vivo, yet with convincing data for athreshold-related mechanism, their safe use can be advo-cated [Bergman et al., 1996].
Experience with regulation of such cases as describedabove shows that there appears to be not enough commonunderstanding on how to evaluate the relevance of each case[Kirkland and Muller, 2000; Muller and Kasper, 2000]. Thishas been a cause for concern for some time, and problemsof nonrelevant genotoxicity in vitro have been discussed byan ICPEMC group [Scott et al., 1991] and has been a topicat a Japanese Environmental Mutagen Society (JEMS)meeting as early as in 1984 [Tazima et al., 1984]. Inprinciple, guidelines such as those given by OECD or ICH52A or 52B offer the possibility to describe and refineprotocols that avoid an excess number of nonrelevant pos-itives and allow for the inclusion of appropriate criteria forjudgement. Table II provides criteria mentioned in the ICH
S2A guidance regarding interpretation of test results andmakes clear that in the pharmaceutical area there is regula-tory acceptance of conditions that may account for nonrel-evant genotoxicity.
The definition of upper limits for cytotoxicity has been animportant issue in the update of the relevant OECD guide-lines and in the creation of new internationally harmonisedguidances for the testing of pharmaceuticals for genotoxi-city [ICH guidances 52A and 52B, Mu¨ller et al., 1999].Within the OECD and ICH expert working groups workingon these documents, it was acknowledged that irrelevantresults may be produced under conditions of high toxicityand that this problem is most often encountered with the invitro chromosome aberration test and in the MLA [Mu¨ller etal., 1999]. However, during the ICH Expert Working Groupdeliberations: (1) exact concentration response informationon cytotoxicity and clastogenicity on a large variety ofcompounds were not available; (2) lack of data did notallow for a sufficient comparison of various measures ofcytotoxicity such as mitotic index, cell proliferation kinet-ics, inhibition of DNA synthesis, cloning efficiency, cultureconfluency; and (3) ideal means of cytotoxicity determina-tion for the various cell types in use could not be delineated.Regarding the MLA, the data that were reviewed under ICHauspicies indicated that below a level of 20% survival manynoncarcinogenic and Salmonella negative compounds havebeen found to produce elevated mutation frequencies [Gar-riott et al., 1995]. However, lack of reliable data led to theconclusion that “caution is due with positive results ob-tained at levels of survival lower than 10%” in the ICH S2Aguidance.
PROPOSALS
Within the IWGTP, several options to address and pos-sibly solve the problem of appropriate levels of cytotoxicityhave been presented and discussed.
TABLE I. Summary of Genotoxicity Results for StandardBattery Tests Submitted for Registration of 335 NewPharmaceutical Entities to the German Federal Institute forDrugs and Medical Devices Between 1990 and 1997
System
No. ofcompounds
tested
No. ofcompounds with
positive testresults
Percent ofcompounds with
positive testresults
Bacterial reversemutation 298 23 7.7%
Mouse lymphoma tkassay 104 28 26.9%
HPRT test 162 6 3.7%Chromosome
aberrations in vitro 266 77 28.9%In vivo bone marrow 283 19 6.7%
TABLE II. Criteria Identified in the ICH S2A Guidance onSpecific Aspects of Regulatory Genotoxicity Tests forPharmaceuticals to Judge the Relevance of an In Vitro TestResult
For in vitro positive results• Attributable to a contaminant?• Due to in vitro specific metabolic pathway or metabolite?• Response not concentration related?• For weak/equivocal responses, is effect reproducible?• Is effect due to extreme culture conditions (pH, osmolality, heavyprecipitate especially in cell suspensions) that will not occur in vivo?• For mammalian cells, is effect only seen at extremely low survival?• Do results agree with others for compounds of same chemical class?
For in vitro negative results• Are standard techniques of metabolic activation (e.g., rodent liver)appropriate?• Does the structure/reactivity of the compound indicate that other testmethods/systems may be appropriate?
Genotoxicity and Cytotoxicity 203
A Change in the Upper Limits of Cytotoxicity thatare Currently Given in ICH Guidances and OECDGuidelines
The upper limits of cytotoxicity for in vitro chromosomalaberration tests according to ICH guidances are, “greaterthan 50% reduction in cell number or culture confluency,”or for human lymphocytes, “an inhibition of mitotic indexby greater than 50%” (ICH S2A). The OECD guideline No.473 has similar wording [OECD, 1998]. A lowering of theseupper limits to, e.g., more than 40% cytotoxicity instead ofmore than 50% cytotoxicity (measured by cell counts ormonolayer confluence) would result in about 25% fewerpositives from a given set of chemicals and the strict en-forcement of a 50% limit would give a reduction by 36%(Galloway, 2000), both apparently without losing accuracy,i.e., the detection of “true positives.” It has not yet beenproven that such a lowering of cytotoxicity limits wouldapply to all different measures of cytotoxicity, would beappropriate for all cell types, and, most importantly, wouldstill permit the detection of most relevant genotoxins of awide selection of chemicals.
The Creation of a Common Ground ofUnderstanding for Interpretation of In Vitro(Positive) Test Results
Nonrelevant positive results from in vitro systems are acommon phenomenon in pharmacology and toxicology.What makes genetic toxicology unique in this regard is thegenerally perceived lack of thresholds, which implies highsignificance for every positive finding. Historically, this isbecause of the roots of genetic toxicology in studies ofdirect damage to DNA by strand breakage, formation ofdimers or crosslinks, or covalent binding to DNA. While athreshold was possible because of DNA repair, this wouldbe at very low exposure levels. Now we know more aboutindirect effects at high/toxic doses and should have theoption to appropriately handle even a high percentage ofpositives via a change in the attitude of genotoxicity inter-pretation to an acceptance of different thresholded andnonthresholded mechanisms.
The criteria given in the ICH guidances for the evaluationof in vitro test results (Table II) are a significant step towardthe creation of common ground for understanding. Nowa-days, the idea of threshold-like mechanisms in genetic toxi-cology seems to be more widely acknowledged [Bergman etal., 1996; Sofuni et al., 2000; Kirkland and Mu¨ller, 2000;Muller and Kasper, 2000] and the problem of “too manypositives” more often becomes “the many different facesand mechanisms of genotoxicity.” A continuation of thediscussion of the issues of test result interpretation andtesting strategy in workshops like the IWGTP ensures com-munication among all scientists involved.
Lowering the Upper Limits of Test CompoundConcentration Irrespective of Cytotoxicity
Currently, the upper limit of testing in mammalian cellsin vitro for relatively nontoxic compounds is 5 mg/ml or 10mM (whichever is lower) (OECD and ICH). This high-exposure practice can be questioned because cellular func-tions and integrity may be severely disturbed and exposurewill generally be dramatically higher than human exposure.An alternative approach might be to calculate an upperconcentration limit on the basis of a sufficient safety marginover human exposure. The author (T.S.) suggested a lowerlimit of about 1–2 mg/ml, which gives a safety margin of atleast 100 over human peak plasma levels for pharmaceuti-cals, because these limits seldom exceed 10–20mg/ml.Normally, the safety margin will be much higher. Althoughthis proposal does not specifically address the problem ofnonrelevant positives, a lower concentration limit will prob-ably alter the overall incidence of positives significantly. Asdescribed above, five of nine ACE inhibitors and five out ofseven AIIR antagonists produced chromosome aberrationsin vitro in the absence of S9 without any further evidencefor DNA damage. Under an upper limit of 1 mg/ml, threecompounds each from both groups would not have pro-duced chromosome aberrations.
The proposal seems reasonable for areas where humanexposure data are available and predictable (e.g., for phar-maceuticals) but must be widely discussed on the basis ofpast experience before it can be further considered forimplementation. The validity of this approach also has to beinvestigated for compounds needing metabolic activationsystems such as S9 because the limited duration of treat-ment may invoke the need for high concentrations in somecases.
REFERENCES
Armstrong MJ, Bean CL, Galloway SM. 1992. A quantitative assessmentof cytotoxicity associated with chromosome aberration detection inChinese hamster ovary cells. Mutat Res 265:45–60.
Bergman K, Muller L, Weberg Teigen S. 1996. The genotoxicity andcarcinogenicity of paracetamol: a regulatory (re)view. Mutat Res349:263–288.
Galloway SM, Miller JE, Armstrong MJ, Bean CL, Skopek TR, NicholsWW. 1998. DNA synthesis inhibition as an indirect mechanism ofchromosome aberrations: comparison of DNA-reactive and non-DNA-reactive clastogens. Mutat Res 400:169–186.
Galloway SM. 2000. Cytotoxicity and chromosome aberrations in vitro:Experience in industry and the case for an upper limit on toxicity inthe aberration assay. Environ Mol Mutagen 35:191–201.
Garriott ML, Casciano DA, Schechtman LM, Probst GS. 1995. Interna-tional workshop on mouse lymphoma assay testing practices anddata interpretations: Portland, Oregon, May 7, 1994. Environ MolMutagen 25:162–164.
Hilliard CA, Armstrong MJ, Bradt CI, Hill RB, Greenwood SK, GallowaySM. 1998. Chromosome aberrations in vitro related to cytotoxicityof non-mutagenic chemicals and metabolic poisons. Environ MolMutagen 31:316–326.
204 Muller and Sofuni
ICH S2A Guidance on Specific Aspects of Regulatory Genotoxicity Testsfor Pharmaceuticals. http://www.ifpma.org/ich1.html.
ICH S2B A Standard Battery for Genotoxicity Testing of Pharmaceuticals.http://www.ifpma.org/ich1.html.
Kirkland DJ. 1992. Chromosomal aberration tests in vitro: problems withprotocol design and interpretation of results. Mutagenesis 7:95–106.
Kirkland DJ, Muller L. 2000. Interpretation of the biological relevance ofgenotoxicity test results: the importance of thresholds. Mutat Res464:137–147.
Muller L, Kasper P, Madle S. 1991. The quality of genotoxicity testing ofdrugs. Experiences of a regulatory agency with new and old com-pounds. Mutagenesis 6:143–149.
Muller L, Kasper P. 2000. Human biological relevance and the use ofthreshold-arguments in regulatory genotoxicity assessment: expe-rience with pharmaceuticals. Mutat Res 464:19–34.
Muller L, Probst G, Schechtman L, Shimada H, Sofuni T, Tweats D. 1999.ICH-harmonised guidances on genotoxicity testing of pharmaceu-ticals: evolution, reasoning and impact. Mutat Res 436:195–225.
OECD. 1998. Ninth Addendum to the OECD Guidelines for the Testing ofChemicals. Paris: OECD.
Scott D, Galloway SM, Marshall RR, Ishidate M Jr, M, Brusick D, AshbyJ, Myhr BC. 1991. Genotoxicity under extreme culture conditions.Mutat Res 257:147–204.
Sofuni T, Hayashi M, Nohmi T, Matsuoka A, Yamada M, Kamata E. 2000.Semi-quantitative evaluation of genotoxic activity of chemical sub-stances and evidence for a biological threshold of genotoxic activ-ity. Mutat Res 464:97–104.
Tazima Y, Kondo S, Kuroda Y. 1984. Problems of threshold in chemicalmutagenesis. The Environmental Mutagen Society of Japan. To-kyo: Kokusai-bunken Printing Co. Ltd. 190 p.
Genotoxicity and Cytotoxicity 205
