rapid metabolic phenotypes for acetyltransferase and ... · vol. 3, 675- 682, december 1994 cancer...

Vol. 3, 675- 682, December 1994 Cancer Epidemiology, Biomarkers & Prevention 675

Rapid Metabolic Phenotypes for Acetyltransferase and CytochromeP4501A2 and Putative Exposure to Food-borne Heterocyclic Amines

Increase the Risk for Colorectal Cancer or Polyps

Nicholas P. Lang,1 Mary A. Butler,2 Joyce Massengill,Mary Lawson, R. Craig Stotts, Martin Hauer-Jensen,and Fred F. Kadlubar

University of Arkansas for Medical Sciences, Arkansas Cancer Research

Center and I. L. McClellan Veterans Affairs Medical Center, Little Rock,Arkansas 72205 [N. P. L., M. L., M. H-J.]; National Center forToxicological Research, Jefferson, Arkansas 72079 EM. A. B., I. M.,

F. F. K.); and University of Arkansas for Medical Sciences, College ofNursing and Arkansas Cancer Research Center, Little Rock, Arkansas

72205 ER. C. S.)

AbstractThe metabolic adivation of food-borne heterocyclicamines to colon carcinogens in humans is hypothesizedto occur via N-oxidation followed by O-acetylation toform the N-acetoxy arylamine that binds to DNA to givecarcinogen-DNA adducts. These steps are catalyzed byhepatic cytochrome P4501 A2 (CYP1 A2) andacetyltransferase-2 (NAT-2), respedively, which areknown to be polymorphic in humans. On the basis ofthis proposed metabolic adivation pathway, patients atgreatest risk to develop colorectal cancer or nonfamilialpolyps should be those who possess both the rapid NAT-2 and rapid CYP1 A2 phenotypes and are exposed tohigh dietary levels of carcinogenic heterocyclic amines.Using a method that involves caffeine administrationand high pressure liquid chromatographic analysis ofurinary metabolites, we have determined the CYP1 A2and NAT-2 phenotypes of 205 controls and 75 cancer!polyp cases. Exposure information was obtained using adietary and health habits questionnaire.

Both the rapid CYP1 A2 and rapid NAT2 phenotypeswere each slightly more prevalent in cases versuscontrols (57% and 52% versus 41 % and 45%,respedively). However, the combined rapid CYP1 A2-rapid NAT-2 phenotype was found in 35% of cases andonly 1 6% of the controls, giving an odds ratio of 2.79(P = 0.002). Univariate analysis of the questionnaireindicated that age, rapid-rapid phenotype, andconsumption of well done red meat were associatedwith increased risk of coloredal neoplasia. Furthermore,a logistic regression model that included age (as acontinuous variable), consumption of well done redmeat, and rapid-rapid phenotype as independent

covariates gave odds ratios of 1 .08, 2.08, and 2.91,respedively. The odds ratios for cooked meat preferenceand phenotype combinations ranged from 1 .00 for rare!medium preference with slow/slow phenotype to 6.45for well done preference with rapid-rapidNAT-2/CYP1 A2 phenotype. This study indicates thatmetabolic phenotype combined with a measure ofexposure to food-borne heterocyclic amine carcinogensare important risk factors in human colorectal cancer.

Introduction

In 1 994, 1 59,000 Americans will be diagnosed with cancer

ofthe colon/rectum and as many as 56,000 persons may dieof the disease (1 ). Although dietary factors are recognized asimportant risk determinants (2), no clear etiology for colorectalcancer has been established, even though the molecularpathogenesis of this disease is at least partially understood(reviewed in Ref. 3). The current hypotheses regarding the

etiology of colorectal cancer involve dietary factors (4, 5),intestinal tract physiology (6, 7), and other determinants such

as sex hormones (8, 9).In addition to variations in exposure to pro- or anticar-

cinogens that are mediated by diet and gender, some in-vestigators have further suggested that individual suscepti-

bility may be due to variations in a metabolic response,namely a polymorphism of NAT-23 (1 0, 1 1 ), which is in-volved in the metabolic activation of carcinogenic hetero-

cyclic amines that are formed during the cooking of food.Recently, a second polymorphism of this activation path-

way for heterocyclic amine carcinogens has been con-firmed (12) and involves CYP1A2. A method for determin-ing both NAT-2 and CYP1A2 phenotypes was alsodeveloped (1 2). NAT-2 and CYP1 A2 are each genetic poly-morphisms exhibiting autosomal codominant transmission.

For NAT-2, these are due to point mutations in the codingregions of an intronless gene; while for CYP1A2, the poly-morphism appears to reside in gene(s) controlling CYP1A2

expression or mnducibility (reviewed in Refs. 13 and 14).While the identification ofdifferences in DNA sequences

between rapid and poor CYP1A2 metabolizers has not oc-curred, there are four reports of family studies that clearlyindicate the poor metabolizer CYP1 A2 phenotype is inheritedas an autosomal recessive trait. These studies have evaluated

CYP1A2 by caffeine metabolism (14), phenacetin O-deethy-lation (1 5), and theophylline 1 -demethylation activity (1 6, 1 7).

Received 3/10/94; revised 7/22/94; accepted 7/28/94.

1 To whom requests for reprints should be addressed, at Arkansas Cancer

Research Center, Department of Surgery, 4301 West Markham, Slot 725,

Little Rock, AR 72205.2 Present address: National Institute for Occupational Safety and Health,

Cincinnati, OH 45226.

3 The abbreviations used are: NAT-2, N-acetyltransferase-2; CYP1A2, cyto-

chrome P4501A2; 137X, caffeine; 17X, 1,7-dimethylxanthine; 17U, 1,7-dimethyluracil; AFMU, 5-acetylarnino-6-formylamino-3-methyluracil; 1 X,

1 -methylxanthine; Cl, confidence interval.

Association for Cancer Research. by guest on September 24, 2020. Copyright 1994 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

https://bloodcancerdiscov.aacrjournals.org

676 Metabolic Phenotype, colon Neoplasia, and Heterocyclic Amines

__\ Ouc�toride

-‘I OH

1’H

�___N( CYP1A2 H

�H OH

Blood I

y

I�c�t1on I Covalent BIndIng

to DNA

/NAT2

�NAT2

Fig. 1. Proposed pathway for heterocyclic amine co-Ion cancinogenesis. UDPGT, U DP-glucuronyl trans-

ferase.

Neoplasla � -* Covalent BIndIngto DNA

Currently, it is hypothesized that the food-borne het-�erocyclic amine carcinogens undergo N-oxidation by

CYP1A2 to an N-hydroxy metabolite, which is then con-verted through O-acetylation by NAT-2 to a reactiveN-acetoxy derivative that forms DNA adducts in the gutepitheliurn (18). Evidence has been presented that directO-acetylation of the N-hydroxy metabolite by NAT-2 canoccur in the colon mucosa and lead to the generation ofDNA adducts (19). However, NAT-2 is present at muchhigher levels in human liver (20), and, as suggested by morerecent data (14), this may result in formation of theN-acetoxy derivative in the liver with transport through the

circulation to the colon mucosa where subsequent bindingto DNA can occur. In either model, the combination ofrapid N-oxidation with rapid O-acetylation proficienciesshould give the highest levels of the ultimate carcinogen in

the colon epithebium (See Fig. 1). Accordingly, we havehypothesized that the individuals at greatest risk of devel-

oping colorectal cancer or colorectal polyps would bethose who are exposed to hetenocyclic amine carcinogens

and who possess both a rapid N-oxidation phenotype thatproduces high levels of the N-hydroxy hetenocyclic aminesand a rapid O-acetylation phenotype that converts thesecompounds to their reactive N-acetoxy forms.

By using a method (1 2) to evaluate NAT-2 and CYP1 A2phenotypes that uses a single caffeine dose (1 00 mg) and a

single urine specimen, we have examined acetylation andN-oxidation by two enzyme systems that not only have aknown genetic polymorphism but are known to be impor-tant in hetenocyclic amine metabolic activation. By com-bmning phenotype results with lifestyle variables collected ina patient questionnaire, we have also examined the rolesthat phenotype and diet exposure may play in coborectalcancer risk. Combining phenotype results with question-name information has further enabled us to assess whetheror not phenotypes play a greater role in this risk whencombined with putative heterocyclic amine exposures.

Materials and Methods

Study Design. A case-control study was designed to eval-uate the roles of metabolic phenotype and dietary exposureto heterocyclic amines in the risk to develop colorectalpobyps or cancer. The case patients were those persons with

colon cancer or colon polyps who were admitted to the

University Hospital of Al kansas and to the John L. McClel-

ban Memorial Veterans Hospital in Little Rock, Arkansas,

and who agreed to participate in this study. Approximately

one-half ofall patients who met the inclusion criteria agreed to

participate in the study. The control participants were selectedby random digit dialing from all households in central Ankan-

sas with a telephone (95% of households have a telephone).Some of the participants in this report have appeared in pre-

viously published reports of our studies on colon neoplasia.

Inclusion Criteria. Inclusion criteria for the case-controlstudy were: (a) newly diagnosed with colonectal cancer

Dukes’ stage A, B, or C or colorectal adenomatous or vilbous

polyp between January 1, 1988 and June 30, 1992; (b)diagnosis based on pathological examination of tissue re-

moved by endoscopic biopsy or via another surgical pro-cedure; (c) age 1 8-85; and (d) informed consent as ap-

proved by the University Hospital of Arkansas and the

Veterans Administration Hospital Institutional Review

Board protocol and procedures.

Exclusion Criteria. Exclusion criteria for the case-controlstudy were: (a) evidence of malignant disease other thanskin cancer or coborectal cancer; (b) current, or a history of,

abnormal liven function on alcoholism, as well as any pa-

tient with a serum creatinine greater than 1 .8 mg/dl, biliru-

bin greater than 1 .5 mg/dl, alkaline phosphatase greaten

than 1 40 units/liter, or SCOT greater than 40 units/mI (these

exclusions were necessary because the phenotypingmethod is dependent on normal hepatic and renal func-tion); and (c) known caffeine intolerance.

Exclusion criteria for cases and controls were identical

except for diagnosis of colorectal cancer or cobonectab polypsin cases; skin cancer was the only cancer allowed in controls.

Dietary and Health Habits Questionnaire. A modifiedBlock questionnaire was used with the changes being more

detailed questions regarding cooking practices (21). The

smoking status assigned was based on self-reporting as

current smoker on nonsmoker for this analysis. After ques-

tionnaines were screened for completeness and clarity, the

data were entered by keypunching directly from the inter-

view form. An audit was performed to verify keypunchingand as an additional validation on editing and coding.



Nonsmoking Caucasian Contro)s

14

12

10

14

12

10

0

Unnary Molar Ratio of (1 7X+17U)/1 37X

Fig. 2. Frequency histogram ofurinary molar ratio of [1 7X + 1 7U1/1 37X fornonsmoking Caucasian controls.

0 10 20 30 40 50

Cancer Epidemiology, Biomarkers & Prevention 677

Patient Instructions and Sampling. Patients were instructed

to abstain from food after midnight or fluids after 7 a.m. of

the day of the test. At 8 a.m., the subject was given approx-imately 100 mg of caffeine (instant coffee diluted in 9

ounces of hot water). After 1 1 a.rn., food and drink (non-

caffeinated) was permitted as desired. At 12 p.m., each

subject was directed to empty their bladder of urine and to

refrain from further urination until 1 :00 p.m., when urinesamples were taken for caffeine metabolite determinations.

These urine samples were collected in coded containers

and a 1 5-mI aliquot was removed to another coded tube for

analysis and for storage of unused urine. The samples werefrozen immediately. Within 1 week the samples were

thawed, the urine pH was immediately adjusted to 3.5 with0.1 N HCI, and the samples were then stored at -20#{176}Cuntil

analysis by high pressure liquid chromatography. Sample

analyses for the patients and controls were carried out at the

same time (batch processing).

Laboratory Procedures. The use of caffeine for phenotyp-

ing both NAT-2 (12, 22) and CYP1A2 (12, 23, 24) has beendescribed in detail and found to be highly reproduciblewithin individuals. Briefly, 1 37X and four of its metabolites,

1 7X, 1 7U, AFMU, and 1 X, are quantified by computerized

high pressure liquid chromatography with photo-diode an-

ray detection and spectral validation. The urinary molar

ratio of AFMU/1 X was selected to represent NAT-2 activity(12). The urinary molar ratio of [17X + 17U]/137X was

selected for the CYP1 A2 activity because this ratio wasfound to represent caffeine 3-demethylation activity better

than the other proposed ratios of 1 7X/1 37X (24), [AFMU +

1XJ/17U (25), and [AAMU + 1X + 1U]/17U (26) wheninvestigated by Butler et a!. (1 2).

Statistical Analysis. The analyses were performed using thecomputer programs iMP (Ver. 3.0; SAS Institute, Inc., Cary,NC) for the Macintosh and EGRET (Ver. 0.25.6; Statistics

and Epidemiology Research Corporation, Seattle, WA) forthe IBM personal computers. Because of the known patho-

genesis from polyps to cancers (3, 27) and the similarities of

the groups, these two groups have been combined for anal-ysis. For these reasons, we chose to report our results as asingle group of cases (cancers and polyps) versus controls.

Univaniate statistics were used to compare cancer andpolyp patients to normal controls (Student’s t test for con-tinuous variables and 2 x 2 contingency table analysis fordichotomous variables). Additional analyses were per-formed to evaluate the influence of age on the selection ofcovaniates for inclusion in the model. In a comparison of casesto controls for those less than 60 years of age and for those 60years of age and older, as well as a comparison of cases to

controls with all controls less than 35 years of age deleted,both demonstrated similar associations to those reported here.

The probit plot ofthe molar ratio of AFMU/1 X provided

a clear breakpoint at 0.6, which was used as the cutoff point

to separate rapid (�0.6) from slow (<0.6) NAT-2 pheno-types (12, 22). Probit plots of the molar ratio of [17X +1 7U]/1 37X showed apparent breakpoints at 4 and 1 0 fornonsmoking controls (Caucasian) and at 6 and 20 for smok-ing controls (Caucasian), allowing designation of slow, in-

termediate and rapid CYP1A2 phenotypes (12, 24). A sim-ilar breakpoint is seen in the frequency distribution of theurinary molar ratio [17X + 17UJ/137X for nonsmoking

Caucasian controls provided in Fig. 2. For the purposes ofthese analyses, the slow and intermediate phenotypes were

combined and compared to the rapid phenotype.

The selection of �ovaniates for initial inclusion in alogistic regression model was based on the results of theunivaniate statistics. Biologically plausible interactions such

as between race and phenotype were also included. The

model was fitted by backward manual elimination. TheWald statistic was used to select covariates for elimination,

and the reduced and full models were compared with the

likelihood ratio test at each step (28).

Results

Group Characteristics. Two hundred eighty individualswere recruited for this study; 205 participants served as

controls and 75 were patients with colorectal polyps (n =

41 ) or cancers (n = 34). The control group consisted of

persons ages 20-80; 63% were male, 88% were Caucasian(n = 180), and 12% were African-American (n = 25). Thesmoking rate among controls was 36% for African-Amen-

can (9 of 25) and 23% for Caucasians (41 of 1 80). Patientswith colorectal polyps or cancers ranged in age from 36 to

84 years; 56% were male, 83% were Caucasian, and 1 7%were African-American (Table 1).

NAT-2 and CYP1A2 Phenotypes. NAT-2 phenotype distni-bution in the control subjects was influenced slightly by ageand by race, but not by gender or smoking status [age t test= 2.49, P = 0.01 (each 1 year increase in age increasedNAT-2 by 0.01 ); race t test = 1 .95, P = 0.05 (African-American mean = 1 .46, Caucasian mean = 1 .05); sex ttest= 0.74, P = 0.46; smoking t test = 0.47, P = 0.641.

Although the NAT-2 values differed between African-Americans and Caucasians, the NAT-2 phenotype assign-

ment did not differ significantly with a slow phenotype rate

in African-Americans of 44% and a slow NAT-2 phenotype

of 57% in Caucasians (P = 0.28, Fisher’s exact).The CYP1 A2 phenotype was clearly influenced by

smoking status, demonstrating a mean CYP1A2 value of10.6 for all nonsmoking control participants, compared to18.6 for all smoking control participants. Phenotypic clas-sification, adjusted for smoking status, gave 41% rapid

CYP1A2 for the controls and 57% rapid CYP1A2 for pa-tients with colorectal polyps or cancers. Neither age nor

gender showed significant influence (t test = -1 .44, P =

0.15; ttest = 0.33, P= 0.74, respectively).



x

�. 20

:� 15

� 10

I

0

678 Metabolic Phenotype, Colon Neoplasia, and Heterocyclic Amines

Table 1 Pilot study results comparing cob rectal polyp/cancer patients I o healthy controls

VariableHealthy

volunteers

X±SD

Polyp/cancerpatients

X±SD

P values

Age )yr) continuous variable

Age

Categorical variable contingency table

46.9 ± 1 3.8

80%;<60

20#{176}!,,;�60

60.3 ± 1 1 .9

45%; <60

55%; �60

�0.0001

�0.0001

Gender

Male

Female

63%

370/,,

56%

44%, 0.28

Ethnic group

Cau asian

African-American

88%

1 2”/,,

83%

1 7”/,, 0.26

Smoking status

No 76% 77% 0.78

Acetylalion rate (urinary molar ratio of AFMU,’lX( 1 .10 ± 0.99 1 .44 ± 1 .64 0.04

Rapid NAT-2 phenotype contingency table 45% 52% 0.31

N-Oxidation rate (Urinary molar ratio of 17X + 17U)/137X)

All

Smokers

Nonsmokers

Cauasian

Smokers

Nonsmokers

Alrk an-American

Smokers

Nonsmokers

Rapid CYP1A2 phenotype contingency table

12.5 ± 8.4

18.6 ± 10.0

10.6 ± 6.0

19.9 ± 9.96

10.7 ± 6.02

10.7 ± 6.29

9.77 ± 5.55

41%

14.0 ± 9.06

18.2 ± 10.4

1 2.4 ± 8.1

Smokers vs.

nonsmokers

Smokers vs.

nonsmokers

57%

0.21

0.90

0.11

�0.00001

0.74

0.03

Rapid-rapid phenotype contingency table 1 6/, 35% 0.002

Frequvni use of cooking oil contingency table 20/ 41% 0.003

Frequent (raw) vegetable consumption contingency table 93% 77% 0.0002

Cooked meal preference (well done) contingency table 26% 45/,, 0.002

Frequ’nI cheese consumption contingency table 97#{176}!,, 84% 0.0002

Controls (N-205)

. Mean Vaiue±S.E.

ILCaucasian Caucasian Alrican�Amercan African-American

No Yes No Yes

Ethnic Group and Smoking Status

Fig. I. Influence of smoking on Caucasian but not on African-AmericanCYP1 A2 phenotype.

However, ethnic variation approached significancewith a mean value for CYP1A2 of 10.5 for all African-Americans and 1 3.2 for all Caucasians (t test = 1 .81 ; P =

0.07). Moreover, the influence of smoking on CYP1 A2 wasvery different in the two ethnic groups (Fig. 3). Caucasian

control patients had a mean value of 10.7 for nonsmokersand of 19.9 for smokers (ttest = 6.83; P< 0.00001), while

25 African-American control patients had a value of 9.77 fornonsmokers and 10.7 for smokers (ttest = 0.34; P = 0.74).The rate of smoking (cigarettes/day) was evaluated as a

20 possible explanation for the CYP1A2 differences observed.

Among control cases the African-Americans averaged 1 715 cigarettes/day while the Caucasians smoked 21/day. This is

insufficient to explain the differences seen in CYP1A2 ac-

10 tivity between smoking Caucasians and African-Americans.Analysis of Dietary and Health Habits Questionnaire.Analysis of patient questionnaires was aimed at comparing

5 dietary and other health practices of enrollees. Foodchoices did not vary notably between groups; however, the

0 preference for eating meat that was cooked to well donewas much greater among the polyp/cancer patients than

among control participants (45% versus 26%). Neither thequantity non the type (beef, pork, chicken, on fish) of meat

consumption varied significantly between control patientsand case patients. Use of cooking oil occurred more fre-quently among case patients than among controls (41%versus 20%). Frequent raw vegetable consumption wasgreater in the control patients than in the cases (93% versus77%). Finally, the regular consumption of cheese occurredmore frequently in the control patients than in the case

patients (97% versus 84%). Two other questions regardingcheese consumption (mixed dishes with cheese and pizzaconsumption) showed significant increased cheese con-sumption by the control group compared to the case pa-



Cancer Epidemiology, Biomarkers & Prevention 679

Table 2 Comparison of colorectal polyp patients to col orectal cancer patients

.Variable

Polyp patients-

(n=41(X±SDCancer patients

-(n=34)X±SD

P values

Age (yr)” 60.6 ± 1 1 .4 59.9 ± 1 2.6 >0.80

Age (categorical variable) 46% <60

54% �60

44% <60

56% �60

>0.85

Gender

Male

Female

68%

32#{176}k

41%

59%

0.02”

Ethnic group

Caucasian

African-American

83%

1 7’Y0

82%

18/,

>0.95

Smoking status

No 67% 91/, 0.01”

Acetylation rate (urinary molar ratio of

AFMU/1 X)1 .80 ± 2.0 1 .01 ± 0.92 0.04”

Rapid NAT-2 phenotype 62% 41% >0.08

N-Oxidation rate (urinary molar ratio of

)17X + 17U}/137X)

All

Smokers

Nonsmokers

Caucasian

Smokers

Nonsmokers

African-American

Smokers

Nonsmokers

14.7 ±7.1

17.4 ±9.3

1 2.95 ± 4.7

18.1 ±9.3

1 2.98 ± 4.6

8.71 (Only 1)

12.9 ± 5.5

13.0 ±11.1

21.9 ± 16.7

1 1 .95 ± 1 0.2

21.9 ±16.7

1 2.41 ± 1 1 .2

(None)

10.14 ± 4.1

>0.46

>0.51

>0.69

>0.58

>0.85

ND

>0.40

Rapid CYP1A2 phenotype 63% 50% 0.32

Rapid-rapid phenotypes” 40% 29”,’, 0.43

Frequent use cooking oil 36% 46% 0.58

Frequent raw vegetable consumption 82#{176}/, 73% 0.41

Cooked meat preference (well-done)” 37% 56’Y 0.1 1

Frequent cheese consumption 78% 91% 0.21

.‘ Variables in final model.

1, Variables with significant differences.

tients. Seventy other questions regarding food prefer-ences showed no differences or smaller differences

between the case and control patients than these. Ethnicgroup, gender and smoking status showed no between-group differences.

Combined Phenotype and Questionnaire Analyses. Strati-fied analyses of colon cases and polyp cases are shown inTable 2. Only three factors (gender, smoking status, andacetylator rate) varied significantly between polyp and can-

cer cases. In addition to the assignment of subjects as rapidor slow NAT-2 and as rapid and slow CYP1 A2 phenotypes,another variable (“rapid-rapid”) was created to classify in-

dividuals who were rapid for both of these phenotypes.Remarkably, 35% of the cancer/polyp cases were rapid-rapid, compared to only 1 6% of the controls (x2 = 9.96;

P = 0.002) (Table 1).Logistic regression analysis identified the following co-

variates as independent predictors of colon cancer or pol-yps: age; well done meat preference; and rapid-rapid phe-notype (Table 3). These covaniates showed significant

positive association with colon neoplasia (age, OR = 1 .08;

preference for well done meat, OR = 2.08; and rapid-rapidphenotype, OR = 2.91 ). An alternative way to consider the

Table 3 Logisti c regression analysi s of final multiva nate model

CovariateOdds ratio

lower95”/,, Cl Uoper

95”,’,� Cl

Age

(OR/i yr)

1.08 1.05 1.12

Well done 2.08 1.05 4.12meat

Rapid-rapid

phenotypes

2.91 1 .38 6.1 1

interaction of phenotype and meat cooking preference is

provided in Table 4, where the odds ratios for different

combinations of phenotypes and cooked meat preferenceare provided. The ratios range from 1 .00 for the NAT-2/

CYP1 A2 phenotype slow/slow and cooked meat preference

of rare/medium to 6.45 for the rapid-rapid phenotype corn-bined with well done cooked meat preference. Cooking oil

use, raw vegetable consumption, and cheese consumptionwere unsuitable for use in the logistic regression because of

small cell numbers. None of the interactions proved to be

significant.




Discussion

While colorectal cancer and polyp disease are commonproblems in the United States population, there is no model

that would allow clear risk assessment on an individual

basis. While the evidence to increase fiber and vegetableconsumption is reasonable, these factors probably influ-

ence exposure and do not address the question of individ-

ual susceptibility based on genetic variations. Two previousreports (1 0, 1 1 ) suggested that part of the variation in I isk

might be explained by the NAT-2 polymorphism and its

probable involvement in the metabolic activation of het-

erocyclic amine carcinogens. Accompanying dietary ques-

tionnaires also suggested that environmental or dietary ex-

posures to such carcinogens were risk factors (29). In thestudy reported herein, we sought to demonstrate that bycombining information regarding exposure (questionnaire

data) with information on metabolic phenotype, a betterassessment of specific risks could be made.

Prior to final analysis, the colon cancer and colon

polyp groups were compared for those factors planned for

univaniate analysis. This comparison (Table 2) indicated the

two groups differed significantly in two factors: gender and

smoking status. It is possible the gender difference is due to

the sample size. Since there is little evidence for a sex bias

in colon neoplasia, these differences should not influence

the outcome. The effect of combining the two groups brings

the male:female ratio closer to 50:50.When the two groups are combined, gender had no

influence on colon cancer or colon polyp risk. This finding

fits prior studies that find similar rates in both sexes. Themale/female percentage was 63/37% for controls and 56/

44% for polyp/cancer cases (x2 = 1 .29; P = 0.26). The

mean NAT-2 and CYP1 A2 values did not appear to beinfluenced by gender status.

The greaten smoking rate combined with the greater fre-

quency of males in the polyp group is compatible with re-cently published reports (30, 31) ofthe association of cigarettesmoking and colorectal adenoma and colorectal cancer. Thelower rate of smoking among the cancer patients could be due

to a recent change brought on by the diagnosis of cancer,could indicate that other agents in addition to those evaluatedin this study are involved in colon cancer, or that the transitionfrom polyp to cancer involves carcinogens in addition to thosethat produced the initial polyp. Because these two groups are

similar in so many variables, the two are combined for thepurposes of final analysis.

Ilett and associates have reported a change in acetyla-

tion rate with age (1 1 ). By using a control group that rangedfrom 20 to 80 years, we also found that the acetylation rate

showed an increase with increasing age. While this in-crease has statistical significance (P = 0.01), there is little

biological significance since the rate changes only 0.01/year of age. The effect of age on CYP1 A2 was not statisti-

cally significant. Other investigators have reported age to beone of the important risk factors for the development ofcolon cancer and polyps (32). This study neconfirms this byfinding the continuous variable “age” associated with in-creased risk of colon cancer or polyp disease. Clearly themodel derived in this study (Table 4) suggests that age is stilla very important risk factor. This is probably due to thelength of exposure required to agents such as the hetero-

cyclic amines for the development of the genetic changesnecessary to induce a polyp or a cancer (33).

Ethnic origin had no influence on the risk of polyp or

cancer development. The proportion of African-Americanswas not statistically different between cases (1 7%) and con-trols (1 2%). However, NAT-2 function varied with ethnicitywith the African-American mean value being 39% greater

than the mean for Caucasian controls (1 .46/1 .05, t = 1 .95,P = 0.05). This gave a rapid/slow proportion of 56/44% forAfrican-American controls and 43/57% for Caucasians con-

trols, which is similar to the results reported by Weben (34).This compares favorably to a recent genotyping studywhere investigators found a rapid/slow proportion of 59/

41% for African-Americans and 45/55% for Caucasians(35). The difference seen for African-Americans betweenthe two studies is probably due to the smaller numbers ofAfrican-Americans in the present study.

To evaluate the influence of race on the odds ratio forNAT-2, CYP1A2, and the rapid-rapid phenotype, direct

comparison was made between African-American and Cau-casian controls. These calculations demonstrated no signif-icant differences (P = 0.28, P = 0.49, and P = 0.75) for

NAT-2, CYP1A2 or rapid-rapid phenotype, respectively.Two additional tests were performed to determine

whether race had a “meaningful” confounding effect on themodel: (a) the logistic regression was performed with andwithout race as a covaniate. This produced only a 2%

change in the coefficient for rapid-rapid phenotype (1 .064

to 1 .040); and (b) the logistic regression was performed onAfrican-American plus Caucasians and compared to the

analysis performed on Caucasians only. This produced onlya 6% change in the coefficient for rapid-rapid phenotype(1 .064 to 0.997). Neither of these are sufficiently large tojustify inclusion of race in the final model.

This study extends previously reported data regarding

colon cancer and acetylation phenotype. A weakness of thestudy reported by our laboratory in 1 986 was that the ratioof rapid to slow acetylators in the control group was un-usually low (28% rapid, 72% slow). The current study usesrandomly selected controls from the general population to

address this and further demonstrates a ratio of 45% napid/55% slow, similar to other reports for American populations(34). This lowered the statistical significance of acetylationphenotype as a dichotomous variable (� = 1 .1 2; P = 0.29),

although the acetylation rate remained significantly differ-ent between the two groups (1 .1 0 ± 0.99 for controls versus1 .44 ± 1 .64 for cases, t = 2.09, P = 0.04). Comparison ofthe NAT-2 phenotype designation for controls versus casesin a 2 x 2 contingency table gave an odds ratio of 1.35

(P = 0.38, 95% Cl = 0.90-2.02). This result differs from twoprevious reports (1 0, 1 1) in the size of odds ratio and P value.

The power of the phenotype to predict risk may be

related not only to the level of exposure to the putativecarcinogens (hetenocyclic amines) but also the presence ofother risk factors in the group studied (such as familialpolyposis or Lynch syndrome). The presence of other riskfactors was determined by self-reporting of the family his-tory for the presence of benign as well as malignant dis-eases. There were no cases of familial polyposis or Lynchsyndrome recognized among the cases.

Since smoking is a known inducer of CYP1A2, theinteraction of ethnicity, smoking, and CYP1A2 was inves-tigated by comparing the mean values for smoking andnonsmoking Caucasian and African-American control

cases. This influence of smoking on CYP1 A2 function wasmodulated by ethnicity. Caucasians who smoke had a

mean [1 7X + 1 7U]/1 37X ratio that was nearly double thatseen in nonsmoking Caucasians, while smoking and non-



Cancer Epidemiology, Biomarkers & Prevention 68!

1 . Boring, C. C., Squires, T. S., Tong, T., and Montgomery, S. CancerStatistics, 1994. CA Cancer J. Clin., 44: 7-26, 1994.

Table 4 Inter actions betw een phenotype and d ietary exposure

Covariate

OddsRatio

Phenotypes)NAT-2 and

CYP1 A2)

.Meat cooking

preference

Slow-slow

Rapid-slow

Slow-rapid

Rapid-rapid

Rare/medium 1.00

0.91

1 .39

3.13

Slow-slow

Rapid-slow

Slow-rapid

Rapid-rapid

Well done 2.06

1 .87

2.86

6.45

smoking African-Americans had mean values similar tononsmoking Caucasians (Fig. 2, Table 2). Apparently, Cau-

casian CYP1A2 levels are inducible, while African-Amen-can CYP1A2 expression is either maximally induced oruninducible by smoking. The latter seems more likely sincethe values are similar to those seen in nonsmoking Cauca-sian and the rate of cigarette consumption did not differ

significantly between the two groups.Comparison of the rapid and slow CYP1 A2 phenotype

for cases and controls produced an odds ratio of 1 .91 (P =

0.03, 95% CI = 1 .20-2.87) for the rapid phenotype. Thissuggests that rapid CYP1 A2 phenotype alone is associatedwith increased risk of colon cancer or polyp disease. The

significant odds ratio (2.79, P = 0.002, 95% CI = 1 .69-4.47 for cases versus controls) for the combined rapid-rapidphenotype strongly indicates that the proposed metabolic

pathway and the level of exposure to the putative carcinogenscould play a significant role in the risk for colon neoplasia.

Several studies have shown an association betweenincreased beef consumption and colon cancer risks (36-

38). The importance of cooking techniques is suggested by

a study by Schiffman and Felton that found a preference forwell done meat among colon cancer patients compared tocontrols (39) and by a recent study from Sweden showingincreased colon cancer risk associated with overcooking of

meat (40). Prolonged cooking at high temperatures in-creases the production ofthe heterocyclic amines as shownby Felton et a!. (41 ) and Sugimuna et a!. (42). These corn-pounds are known colon carcinogens in animals (43).

The importance offood preparation techniques is further

stressed by a recent report by Tannenbaum et a!. (44), who

found thatthe highest levels ofa heterocyclic amine (2-amino-3,8-dimethylimidazo-[4,5-flquinoxaline) in human urine sam-pIes was associated with the consumption ofvery crispy bacon

and the grease sediment from the bacon. Additional supportfor the significance of preparation is provided by Cross et a!.(33), who found 10-1 00-fold higher hetenocyclic amine levelsin pan scrapings than in the meat (beef, bacon on fish) being

cooked. While our study included questions regarding meatcooking preference (rare, medium, or well-done) as well astype (beef, pork, chicken, or fish) and quantity (frequency andportion size), there were no questions regarding bacon cook-ing techniques. There were no differences between the groups(cases versus controls) regarding the quantity of beef, pork,

chicken, or fish consumed. The use of oil in cooking wassignificantly increased in the cases compared to controls.Since cooking in oil increases the temperature of the food

compared to cooking in water, this association fits the hypoth-

esis of increased heterocyclic amine exposure leading to in-creased risk of polyp or cancer development in the colon.

The interactions between phenotype and meat cookingpreference shown in Table 4 suggest there are sources forhetenocyclic amines outside of cooked meat since the oddsratio increases from 1 .0 to 3.1 as the phenotype goes from

slow-slow to rapid-rapid, while the meat cooking preferenceremai ns rare/medium . The importance of phenotype-exposure

interaction is further illustrated by the change in the odds ratiofor slow-slow from 1 .0 to 2.06, while the rapid-rapid odds

ratio goes from 3.1 to 6.5 when meat cooking preference is

changed from rare/medium to well done.The pattern of consumption of raw vegetables and

cheese noted in this study is compatible with prior studieson the epidemiology of colon cancer (reviewed in Ref. 45).Vegetable consumption has been recognized as beneficial

in reducing colon cancer risk (reviewed in Ref. 45). Oneproposed mechanism is through increasing the intake of

dithiolthiones (found in cruciferous vegetables), which in-duce glutathione S-transferases known to catalyze the de-toxification of N-acetoxy 2-amino-i -methyl-6-phenylimi-dazo-[4,5-b]pyridine (46). While no difference inconsumption of crucifenous vegetables (mostly cooked) be-tween case patients and controls was seen, consumption of

raw vegetables was significantly higher in control patients.It is possible that the raw vegetables noted here represent

crucifenous vegetables since this is the form frequentlyeaten uncooked. Our questionnaire did not indicate thetype of vegetable consumed raw. The beneficial relation-ship noted with cheese may be due to its dietary value as asource of calcium, whose mechanism of action is thought to

occur through reduction of cell proliferation in the gastro-intestinal tract (47). The possibility exists that these are chancefindings because ofthe large number offood items considered.

In the present study, very few subjects did not use oil, cheese,

or raw vegetables. This produced a problem in the logisticregression analysis that could be overcome in a larger study.

While the risk of colon cancer or polyp disease in-

creases as a person ages, this risk is increased further forthose who are rapid-rapid phenotype and prefer meat welldone. Should the associations identified in this study proveto be causally related to colon neoplasia, the risk could beinfluenced by a change in the type and preparation of

certain components of the diet. The simple expedient ofeating rare meat, premicrowaving meat (41 ), or avoidingcooking methods that result in burning meat juices couldsubstantially reduce the risk of colorectal neoplasia.

Note Added in Proof

Recently, pharmacokinetic studies by Tang et a!. (48) mdi-

cated that the caffeine metabolite ratio in our studies reflectsa polymorphism in renal flow rate and clearance of caffeinerather than a polymorphism in hepatic CYP1 A2. We exam-ned this possibility here and in our previous report (12) and

could find no evidence of this phenomenon under our exper-imental conditions and methods of analysis. Moreover, fam-ily studies and in vitro metabolism data (reviewed in Refs. 12,1 4) have consistently shown a genetic polymorphism forCYP1 A2.

References




2. Willett, W. C. Overview of nutritional epidemiology. In: NutritionalEpidemiology, pp. 3-19, New York: Oxford University Press, 1990.

3. Peinado, M. A., Fernandez-Renart, M., Capella, G., Wilson, L., andPerucho, M. Mutations in the p53 suppressor gene do not correlate withc-K-ras oncogene mutations in colorectal cancer. nt. J. Oncol., 2: 1 23-1 34,

1993.

4. Burkitt, D. Related disease-related cause? Lancet, 2: 1229-1231, 1969.

5. Steinmetz, K. A., and Potter, J. D. Vegetable, fruit, and cancer. I.

Epidemiology. Cancer Causes Control., 2: 325-357, 1991.

6. Stephen, A., and Cummings, j. Mechanism of action of dietary fibre in the

human colon. Nature (Lond.), 284: 283-284, 1980.

7. Cummings, I. Fermentation in the human large intestine: evidence and

implications for health. Lancet, 1: 1206-1209, 1983.

8. McMichael, A. J., and Potter, J. D. Reproduction, endogenous and exog-enous sex hormones, and colon cancer: a review and hypothesis. J. NatI.

Cancer Inst., 65: 1201-1207, 1980.

9. McMichael, A. J., and Potter, J. D. Diet and colon cancer: integration of

the descriptive, analytic, and metabolic epidemiology. NatI. Cancer Inst.Monogr., 69:223-228, 1985.

10. Lang, N. P., Chu, D. Z., Hunter, C. F., Kendall, D. C., Flammang, T. J.,

and Kadlubar, F. F. Role of aromatic amine acetyltransferase in humancolorectal cancer. Arch. Surg., !21: 1259-1261, 1986.

1 1 . llett, K. F., David, B. M., Detchon, P., Castleden, W. M., and Kwa, R.Acetylation phenotype in colorectal carcinoma. Cancer Res., 47:

1466-1469, 1987.

12. Butler, M., Lang, N., Young, J., Caporaso, N., Vineis, P., Hayes, R.,Teitel, C., Massengill, J., Lawsen, M., and Kadlubar, F. Determination of

CYP1A2 and acetylator phenotypes in several human populations by anal-ysis of caffeine urinary metabolites. Pharmacogenetics, 2: 1 1 6-1 27, 1992.

1 3. Kadlubar, F. F. Biochemical individuality and its implications for drugand carcinogen metabolism: recent insights from acetyltransferase and cy-

tochrome P4501A2 phenotyping and genotyping in humans. Drug Metab.Rev., 26:37-46, 1994.

14. Nakajima, M., Yokoi, T., Mizutani, M., Shin, S., Kadlubar, F. F., andKamataki, T. Phenotyping of CYP1 A2 in Japanese populations by analysis of

caffeine urinary metabolites: absence of mutation prescribing the phenotypein the CYPA2 gene. Cancer Epidemiol., Biomarkers & Prey., 3: 41 3-421,1994.

1 5. Shahidi, N. T. Acetophenetidin-induced methemoglobinemia. Ann. NYAcad. Sci., 151:822-832, 1968.

16. Miller, C. A., Slusher, L. B., and Vesell, E. S. Polymorphism of theophyl.line metabolism in man. J. Clin. Invest., 75: 1415-1425, 1985.

1 7. Teunissen, M. W. E., De Leede, L. G. J., Boeijinga, J. K., and Breimer, D.

D. Correlation between antipyrine metabolite formation and theophyllinemetabolism in humans after simultaneous single-dose administration and at

steady state. J. Pharmcol. Exp. Ther., 233: 770-775, 1985.

1 8. Kadlubar, F. F., Kaderlik, R. K., Mulder, G. J., Lin, D., Butler, M. A.,Minchin, R. F., IIett, K. F., Friesen, M. D., Bartsch, H., Nagao, M., Esumi, H.,

Sugimura, T., and Lang, N. P. Metabolic activation and DNA adduct detec-lion of PhIP in dogs, rats, and humans in relation to urinary bladder and

colon carcinogenesis. In: R. Adamson, J-A. Gustafsson, N. Ito, M. Nagao, T.Sugimura, K. Wakabayashi, and Y. Yamazoe (eds.), Heterocyclic Amines and

Human Cancer. Princeton, NJ: Princeton Scientific Publishers, 1994.

19. Turesky, R. J., Lang, N. P., Butler, M. A., Teitel, C. H., and Kadlubar, F.F. Metabolic activation of carcinogenic heterocyclic aromatic amines by

human liver and colon. Carcinogenesis (Lond.), 12: 1839-1845, 1991.

20. Ilett, K. F., Ingram, D. M., Carpenter, D. S., Teilel, C. H., Lang, N. P.,Kadlubar, F. F., and Minchin, R. F. Expression of monomorphic and poly-morphic N-acetyltransferase in human colon. Biochem. Pharmacol., 47:914-917, 1994.

21. Block, G., Hartman, A. M., Dresser, C. M., Carroll, M. D., Gannon, J.,and Gardner, L. A data-based approach to diet questionnaire design andtesting. Am. J. Epidemiol., (24:453-469, 1986.

22. Tang, B. K., Grant, D. M., and Kalow, W. Isolation and identification of

5-acetylamino-6-formylamino-3-methyluracil as a major metabolite of

caffeine in man. Drug Melab. Dispos., 11: 218-220, 1983.

23. Kaderlik, K. R., Minchin, R. F., Mulder, G. J., Ilett, K. F., Daugaard-

Jenson, M., Teitel, C. H., and Kadlubar, F. F. Metabolic activation pathway

for the formation of DNA adducts of the carcinogen, 2-amino-i -methyl-6-phenylimidazol4,5-blpyridine (PhIP), in rat extrahepatic tissues. Carcinogen-

esis (Lond.), !5: 1703-1709, 1994.

24. Kadlubar, F. F., Talaska, G., Butler, M. A., Teitel, C. H., Massengill, J.,and Lang, N. P. Determination of carcinogenic arylamine N-oxidation phe-notype in humans by analysis of caffeine urinary metabolites. In: M. L.

Mendelsohn and R. J. Albertini (eds.), Mutation and the Environment, pp.

107-1 14. New York: Wiley-Liss, 1990.

25. Campbell, M. E., Spielberg, S. P., and Kalow, W. A urinary metaboliteratio that reflects systemic caffeine clearance. Clin. Pharmacol. Ther., 42:

157-165, 1987.

26. Kalow, W., and Tang, B-K. Use of caffeine metabolite ratios to explore

CYP1A2 and xanthine oxidase activities. Clin Pharmacol. Ther., 50:508-519, 1991.

27. Corman, M. L. Polyp diseases. In: M. L. Corman (ed), Colon and RectalSurgery, pp. 436-486. Philadelphia: J. B. Lippincott, 1993.

28. Hosmer, D. W., and Lemeshow, S. Applied Logistic Regression, p. 307.New York: John Wiley & Sons, 1989.

29. Wohlleb, J. C., Hunter, C. F., Blass, B., Kadlubar, F. F., Chu, D. Z. J., andLang, N. P. Aromatic amine acetyltransferase as a marker for colorectal

cancer: environmental and demographic associations. Int. J. Cancer, 46:22-30, 1990.

30. Zahm, S. H., Cocco, P., and Blair, A. Tobacco smoking as a risk factor

for colon polyps. Am. J. Public Health, 8!: 846-849, 1991.

31 . Giovannucci, E., Colditz, G. A., Stampfer, M. J., Hunter, D., Rosner, B.A., Willett, W. C., and Speizer, F. E. A prospective study of cigarette smoking

and riskofcolorectal adenoma and colorectal cancer in U. S. women. J. NatI.Cancer Inst., 86: 192-199, 1994.

32. Corman, M. L., Veidenheimer, M. C., and Coller, J. A. Colorectal car-cinoma: a decade of experience at the Lahey Clinic. Dis. Colon. Rectum, 22:477-479, 1979.

33. Gross, G. A., Turesky, R. J., Fay, L. B., Stillwell, W. G., Skipper, P. L., and

Tannenbaum, S. R. Heterocyclic aromatic amine formation in grilled bacon,beefand fish and in grill scrapings. Carcinogenesis (Lond.), 14: 2313-2318,1993.

34. Weber, W. W., and Hem, D. W. N-acetylation pharmacogenetics.

Pharmacol. Rev., 37: 25-79, 1985.

35. Bell, D. A., Taylor, J. A., Butler, M. A., Stephens, E. A., Wiest, J.,Brubaker, L. H., Kadlubar, F. F., and Lucier, G. W. Genotype/phenotype

discordance for human arylamine N-acetyltransferase (NAT-2) reveals a newslow-acetylator allele common in African-Americans. Carcinogenesis

(Lond.), 14:1689-1692, 1993.

36. Haenszel, W., Berg, J. W., Segi, M., Kurihara, M., and Locke, F. B. Large

bowel cancer in Hawaiian Japanese. J. NatI. Cancer Inst., 51: 1 765-1 799,1973.

37. Potter, J. D., and McMichael, A. J. Diet and cancer of the colon andrectum: a case-control study. J. NatI. Cancer Inst., 76: 557-569, 1986.

38. WiIlett, W. C., Stampfer, M. J., Colditz, G. A., Rosner, B. A., and Speizer,

F. E. Relation of meat, fat, and fiber intake to the risk of colon cancer in aprospective study among women. N. EngI. J. Med., 323: 1 664-1 672, 1990.

39. Schiffman, M. H., and Felton, J. S. Fried foods and the risk of coloncancer (Letter). Am. J. Epidemiol., 13!: 376-378, 1990.

40. de Verdier Gerhardsson, M., Hagman, U., Peters, R. K., Steineck, G.,

and Overvik, E. Meat, cooking methods and colorectal cancer: a case-referent study in Stockholm. Int. J. Cancer, 49: 520-525, 1991.

41. Felton, J. S., Knize, M. G., Roper, M., Fultz, E., Shen, N. H., and

Turteltaub, K. W. Chemical analysis, prevention, and low-level dosimetry ofheterocyclic amines from cooked food. Cancer Res., 7 (Suppl.):

21035-21075, 1992.

42. Sugimura, T., and Sato, S. Mutagens-carcinogens in foods. Cancer Res.,43 (Suppl.): 2415s-2421s, 1983.

43. Sugimura, 1. Carcinogenicity of mutagenic heterocyclic amines formedduring the cooking process. Mutat. Res., 150: 33-41, 1985.

44. Tannenbaum, S. R., Stillwell, S., Ji, H., Skipper, P., Yu, M., Ross, R.,

Henderson, B., Turesky, R. J., and Gross, G. A. MeIQx (2-amino-3,8-dim-

ethylimidazol4,5-f]quinoxaline): metabolism in humans and urinary metab-olites in human populations. Environ. Health Perspect., in press, 1994.

45. Potter, J. D. Reconciling the epidemiology, physiology, and molecularbiology of colon cancer. JAMA, 268: 1 573-1 577, 1992.

46. Lin, D., Meyer, D., Ketterer, B., Lang, N. P., and Kadlubar, F. F. Effectsof human and rat glutathione S-transferases on the covalent DNA binding ofthe N-acetoxy derivative of heterocyclic amine carcinogens in vitro: a pos-sible mechanism oforgan specificity in their carcinogenesis. Cancer Res., 54:4920-4926, 1994.

47. Barsoum, C. H., Hendrickse, C., Winslet, M. C., Youngs, D., Donovan,I. A., Neoptolemos, J. P., and Keighley, M. R. Reduction of mucosal crypt cell

proliferation in patients with colorectal adenomatous polyps by dietary

calcium supplementation. Br. J. Surg., 79: 581-583, 1992.

48. Tang, B-K., Zhou, Y., Kadar, D., and Kalow, W. Caffeine as a probe for

CYP1 A2 activity: potential influence of renal factors on urinary phenotypictrait measurements. Pharmacogenetics, 4: 1 1 7-1 24, 1994.



1994;3:675-682. Cancer Epidemiol Biomarkers Prev N P Lang, M A Butler, J Massengill, et al. polyps.heterocyclic amines increase the risk for colorectal cancer orcytochrome P4501A2 and putative exposure to food-borne Rapid metabolic phenotypes for acetyltransferase and

Updated version

http://cebp.aacrjournals.org/content/3/8/675

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/3/8/675To request permission to re-use all or part of this article, use this link



https://bloodcancerdiscov.aacrjournals.orgcgi/alerts

mailto:[email protected]



rapid metabolic phenotypes for acetyltransferase and ... · vol. 3, 675- 682, december 1994 cancer...

Documents