glutathione transferase null genotype, broccoli, and lower ...glutathione transferase 51 i. ......

Vol. 7, 64 7-652. August 1998 Cancer Epidemiology, Biomarkers & Prevention 647

3 The abbreviations used are: OR. odds ratio: CI. contidence interval: GSTM I.

glutathione transferase 51 I.

Glutathione Transferase Null Genotype, Broccoli, and Lower Prevalence

of Colorectal Adenomas’

Henry J. Lin,2 Nicole M. Probst-Hensch,Andrew D. Louie, Irving H. Kau, John S. Witte,Sue A. Ingles, Harold D. Frankl, Eric R. Lee, andRobert W. Haile

Division of Medical Genetics. Department of Pediatrics, Harbor-UCLA

Medical Center, Torrance. Califomia 90502 IH. J. L.. A. D. L.. I. H. K.l;

Department of Preventive Medicine. USC/Norris Comprehensive Cancer

Center. Los Angeles. California 90033 IN. M. P-H.. S. A. I.. R. W. HI:

Department of Epidemiology and Biostatistics, Case Westem Reserve

University. Cleveland, Ohio 44109 [J. S. WI: Department of Gastroenterology.

Kaiser Permanente. Los Angeles. Califomia 9�27 IH. D. Fl; and Department

of Gastroenterology. Kaiser Permanente. Bellflower. Califomia 90706

IE.R.L.I

Abstract

Cruciferous vegetables, especially broccoli, may preventcancer through anticarcinogenic compounds. Forexample, broccoli contains isothiocyanates that inducecarcinogen-detoxifying enzymes. Glutathione transferaseenzymes conjugate isothiocyanates, leading to excretion.We hypothesized that broccoli consumption incombination with the glutathione transferase Ml(GSTMI) null genotype would be associated with a lowerprevalence of colorectal adenomas because of higherisothiocyanate levels. We used a case-control study ofmainly asymptomatic subjects aged 50-74 years whounderwent a screening sigmoidoscopy at either of twoSouthern California Kaiser Permanente Medical Centersduring 1991-1993. Cases (n 459) had a first-time

diagnosis of histologically confirmed adenomas detectedby flexible sigmoidoscopy. Controls (n 507) had nopolyp detected. Subjects had a 45-mm in-person interviewfor information on various risk factors and basicdemographic data and completed a 126-item,semiquantitative food frequency questionnaire. Bloodsamples were used for GSTMJ genotyping. Subjects with

the highest quartile of broccoli intake (an average of 3.7servings per week) had an odds ratio of 0.47 (95%confidence interval, 0.30-0.73) for colorectal adenomas,compared with subjects who reportedly never atebroccoli. When stratified by GSTMJ genotype, aprotective effect of broccoli was observed only among

Received 12/19/97: revised 5/6/98: accepted 5/12/98.

The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertiseuzent in

accordance with I 8 U.S.C. Section I 734 solely to indicate this fact.

I H. J. L. was supported by the UCLA Clinical Nutrition Research Unit (NIH

Grant 2P��lCA427l0): the Peg Liddle Fund at Eisenhower Medical Center.

Rancho Mirage. CA: and NIH Grants 1R01CA66782 and IROICA734O3.

R. W. H. was supported by NIH Grants IR01CA51923 and 1R01CA63699.

J. S. W. was supported by NIH Grant R29CA73270.

2 To whom requests for reprints should be addressed. at Hartxr-UCLA Medical

Center. E4. 1124 West Carson Street. Torrance. CA 90502. Phone: (310) 222-

3673: Fax: (310) 328-9921: E-mail: [email protected].

subjects with the GSTMI null genotype (P for trend,0.001; P for interaction, 0.01). The observed broccoli-GSTMJ interaction is compatible with an isothiocyanatemechanism.

Introduction

Cruciferous vegetables are viewed as potentially protective

against cancer ( 1 , 2). These vegetables include broccoli. brus-sels sprouts, cabbage, cauliflower, collards, kale, and kohlrabi,which are varieties of Brassica oleracea (3). Of early epide-

miological evidence for a protective effect against colorectalcancer, Graham et a!. (4) found an OR3 of 0.3 comparing

people who ate cabbage at least once per week to those whonever ate cabbage. Other studies gave similar results, based ondata for: cruciferous vegetables as a group (5. 6): broccoli.

cauliflower, brussels sprouts, and turnips (7): and cabbage,cauliflower, and sprouts (8). However, not all studies have

found inverse associations between cruciferous vegetables andcolorectal cancer (9-12).

Isothiocyanate compounds are derived from glucosino-lates in various plants. Medicinal properties of glucosinolates

are historically important. as seen in writings by Pythagoras and

Hippocrates. At least 20 such compounds were identified by thel980s (13). All cruciferous vegetables are believed to contain

glucosinolates, but brussels sprouts (6()0-3900 �.tg/g), broccoli(450-1480 ;.Lg/g), and Savoy cabbage (470-1 290 �.tg/g) havethe highest content among Brassica oleracea.

Anticarcinogens in cruciferous vegetables induce enzymes

that detoxify environmental mutagens. Broccoli extract was themost potent enzyme inducer in a murine hepatoma cell assay

(14), probably due to sulforaphane, an isothiocyanate corn-pound ( I 4, 15). Sulforaphane blocked 7, 12-dirnethylbenz(a)-

anthracene-induced mammary tumors in rats. supporting a con-clusion that isothiocyanates protect against cancer ( I 6).

Zhang et a!. ( 1 7) and Kolm ci a!. ( 18) found that GSTMIrapidly conjugates isothiocyanates to glutathione, leading to

excretion. Approximately 50% of people completely lackGSTM I enzyme. due to homozygous deletion of the gene (i.e.,the GSTMI null genotype; Ref. 19-21 ). We hypothesized thata cancer preventive effect of broccoli would be stronger with

the GSTMJ null genotype. due to potentially slower excretionof isothiocyanates. Here we show effects of broccoli consump-

tion and the GSTMJ null genotype on prevalence of colorectal

adenomas, which are precursors to cancer.

Materials and Methods

Cases and Controls. Subjects were from one of two SouthernCalifornia Kaiser Permanente Medical Centers (Belltlower or

on June 6, 2020. © 1998 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

http://cebp.aacrjournals.org/

Table 1 Characteristics of subjects with colorectal adenomas and controls

Colorectal adenomas Control subjects

(n = 459) (n = 507)

n �/c ,j

648 GSTMJ, Broccoli, and Colorectal Adenomas

We used unconditional logistic regression to estimate ORs

Sunset) and had a sigmoidoscopy during the period fromJanuary 1, 1991, through August 25, 1993. Eligible men andwomen were ages 50-74 years, residents of Los Angeles or

Orange counties, and fluent in English. None had a history ofinvasive cancer, inflammatory bowel disease, familial polypo-sis, previous bowel surgery, symptoms suggestive of gastroin-

testinal disease, or a physical or mental problem that wouldpreclude an interview. Cases had a first-time diagnosis of one

or more histologically confirmed colorectal adenomas. Controlswere selected from subjects with no polyps and were individ-

ually matched to cases by gender, age (within a 5-year cate-gory), sigmoidoscopy date (within a 3-month category), and

center where examined.There were 628 potentially eligible cases and 689 poten-

tially eligible controls during the accrual period. Seventy cases

and 94 controls refused an interview, and 29 cases and 32

controls could not be contacted. Response rates (number inter-viewedlnumber eligible) were 84% for cases and 82% for

controls. A replacement was identified when the control subjectoriginally matched to a case could not be interviewed.

Indications for sigmoidoscopy among interviewed sub-jects were: “routine” for 45% of cases and 44% of controls;“minor symptoms” for I 6% of cases and I 3% of controls; and“not specified” for 39% of cases and 43% of controls. Average

depth of the flexible sigmoidoscope was 55 ± 1 1 (SD) cm for

cases and 59 ± 5 cm for controls. Fifteen cases had carcinomain situ. Size and number of polyps were recorded by thesigmoidoscopist.

Subjects gave information on smoking, therapeutic drug

use, physical activity, height, weight, family history of cancer,

and other factors during a 45-mm, in-person interview. Theinterview was S months after sigmoidoscopy, on average, and

exposure data referred to the period up to the sigmoidoscopy.Interviewers were unaware of case or control status for 70% of

cases and 87% of controls.

Diet Data. A 126-item, semiquantitative food frequency ques-

tionnaire (22), asking about diet during the year before sig-moidoscopy, was filled-out by 5 19 cases and 556 controls. The

questionnaire provides information on intake of the followingcruciferous vegetables: broccoli; cabbage or cole slaw; cauli-flower; brussels sprouts; and kale, mustard, or chard greens.Standard methods were used to compute nutrient intakes. Spe-

cific foods corresponding to items on the questionnaire wereselected based on the relative frequency of consumption among

participants of the 1988-1989 Nationwide Food ConsumptionSurvey, southwest region. The Nutrient Data System (NDB

version 2.4) was used as a nutrient database for foods. Data on

nutrient content of supplements were from the Harvard School

of Public Health.

Genotyping. GSTMJ genotyping was done on 977 subjectswho provided a blood specimen. Dried blood spots on blotterpaper (Schleicher and Schuell no. 903) were used for DNA

templates for 936 specimens. Liquid DNA was used for theremaining 41 specimens. PCR assays were as described (23).

Duplicate assays were done for all specimens, without knowl-edge of case or control status.

Data Analysis. The analysis included 966 subjects, because 1 1of the 977 genotyped subjects did not complete a food fre-quency questionnaire and had to be excluded. We assumed thatpersons who did not answer a specific item on the questionnaire

(including cruciferous vegetables) never ate that item, havingfound that missing responses most commonly turn out to be

zero.

Gender

MaleFemale

Ethnicity

White

Hispanic (not black)

Black

Asian/Pacific Islander

Smoking

Never

Ex-smoker

Current

297 63.9 338 67.0

162 36.1 169 33.0

263 55.1 278 54.6

77 16.6 91 17.5

74 15.9 80 15.3

45 9.8 58 11.0

168 36.7 217 42.9

205 43.1 231 44.3

86 20.2 59 12.8

Average age, years 62.0 62.0

Smoking, average pack-years 22. 1 16.7

Fruits and vegetables, servings”/day 5.5 6.1

Cruciferous vegetables, servings/week 2.9 3.5

Broccoli, servings/week 1 .2 1.5

Cabbage or cole slaw, servings/week 0.85 0.88

Cauliflower, servings/week 0.5 1 0.66

Brussels sprouts, servings/week 0.24 0.24

Kale, mustard, or chard, servings/week 0. 14 0.23

Saturated fat, g/day 25.2 22.6

Energy intake. kcal/day 2027 1924

“ Serving size = /‘ cup of each vegetable (broccoli: cabbage or cole slaw:

cauliflower: brussels sprouts; kale, mustard, or chard greens).

and controlled for the matching factors-date of sigmoidoscopy

(6-month intervals), age (5-year intervals), gender, and centerattended-as indicator variables in the model. This allowed us toinclude information on unmatched subjects in the analysis.

Unmatched controls occurred when, for example, their matchedcases did not speak English or were found to have invasivecancer at follow-up colonoscopy. Unmatched cases occurredwhen we were unable to interview an eligible control. Un-matched subjects also occurred if blood, and therefore GSTMJ

genotypes, was obtainable from only one subject of a pair.

Categories of vegetable intake as presented in the tables wereentered as indicator variables in the model.

Covariates considered for possible inclusion in the multi-variate models were: race; body mass index; vigorous leisure

time activity; intake of total energy, saturated fat, and fruits andvegetables; family history of colorectal cancer; intake of non-

steroidal anti-inflammatory drugs; and smoking. None of the

covariates changed the size of the point estimates of interest bymore than 10%. We therefore adjusted only for matching vari-ables, smoking, and intake of total energy, saturated fat, and all

other fruits and vegetables (i.e., vegetables not specified by themodel).

Subjects homozygous for the GSTMJ gene deletion were

coded as having the GSTMJ null genotype. Interaction betweenvegetable intake and GSTMJ genotype was assessed by models

containing interaction terms of vegetable intake with GSTMJ

genotype. One hypothesis is that the null genotype may beassociated with increased risk of colorectal cancer in smokers,

so that smoking might modify a vegetable-GSTMJ interaction.We did not stratify analyses by smoking status, however, be-cause no interaction between smoking and GSTMJ was found

in our original analyses (23). We did restrict the analysis to

people who never smoked to corroborate that any vegetable-



Cancer Epidemiology, Biomarkers & Prevention 649

Table 2 ORs for colorectal adenomas by intake of crucifer ous vegetables”

Intake frequency (servings)

Never 1-3/month 1/week >1/week

Broccoli

OR l.0() 0.62 0.65 0.47 0.007 1)0395C/� CI

- 0.424)93 0.43-0.99 0.3()-0.73

Controls/Cases 72/105 152/134 136/123 147/97

Mean servings/week 0 0.5 1 3.7

Cabbage/Cole slaw

OR 1.00 1.34 1.1 1 1.40 1)33 0.7495C/� CI

- 0.96-1.86 0.75-1.66 0.87-2.27

Controls/Cases I 34/ 108 204/203 106/89 63/59


Cauliflower

OR I .00 0.79 1 .0 1 0.7 I (1.30 0.2(1

95(7� CI - 0.58-1.07 0.68-1.51 0.41-1.23

Controls/Cases 2()4/2l2 186/149 75/70 42/28


Brussels sprouts

OR 1.00 0.76 1.30 0.96 0.91 (1.9795C/� CI

- 0.55-1.06 0.75-2.24 0.38-2.40

Controls/Cases 357/332 1 1 2/84 26/34 I 2/9


Kale, mustard. chard greens

OR 1.00 0.62 0.99 0.41 0(14 0.1)495C/� CI

- 0.4()-().97 0.52-1.88 0.16-1.06

Controls/Cases 408/393 62/38 2 1/2 1 16/7

Mean servings/week 0 0.5 I 3.8

“ The effect of single vegetables (as indicator variables for categories) were adjusted for smoking (indicator variables for current smoking and ex-stiloking): matching

variables (age in 5-year categories: date of sigmoidoscopy in five categories total): clinic: gender: saturated fat (indicator variables for quartiles of saturated fat intake in

the total population): energy intake (calories as continuous variable): intake of fruits and vegetables indicator variables for quartiles of fruit and vegetable intake minus

the specific vegetable(s) of interest in the modell.I, p for trend was calculated by entering the mean number of servings/week in each group as a categorical variable in the model.

‘ P for trend was calculated by entering the intake of the respective vegetable as a continuous (ordered) variable in the model.

GSTMJ interaction was not different in this subgroup. Cigarettesmoking data were abstracted from in-person interviews, and

results of the main effect of smoking are presented elsewhere(24). Individuals who had never smoked 100 cigarettes in theirlife were coded as never smokers. Subjects who had smokedmore than 100 cigarettes were coded as ex-smokers if they quit

smoking before the sigmoidoscopy and as current smokers if

they were smoking at the time of sigmoidoscopy.Restriction of analyses to subjects with an energy intake

between 800 and 4200 kcal did not substantially alter results

and was thus omitted.

Results

Cases and controls are described in Table 1 . Male:female dis-tribution, ethnic mix, and average ages were fairly comparablefor the two groups. There was a lower percentage of smokersamong controls. On average. controls ate more fruits andvegetables per week, including cruciferous vegetables and

broccoli.

Main effects of broccoli and of cruciferous vegetables onthe prevalence of colorectal adenomas are shown in Table 2.

Earlier we presented protective effects of cruciferous vegeta-

bles and broccoli in our study population (25). Because we now

focus on interaction between cruciferous vegetables andGSTMI, we again show the main effects of cruciferous vege-tables and broccoli, but among subjects with known GSTMJ

genotype.We estimated ORs for single cruciferous vegetables across

four levels of serving intake: never; 1-3/month; 1/week;

and > I/week. Increased consumption of broccoli and kale, but

not of other cruciferous vegetables, were associated with lower

prevalence ofcolorectal adenornas (broccoli: P for trend, 0.007;kale: P for trend, 0.04). The OR comparing the highest to

lowest category of broccoli intake was 0.47 (95% CI, 0.30-

0.73). The corresponding OR for kale was 0.41 (95C/ CI,

0.16-1.06).Results from a model with vegetable intake as a continu-

ous variable (instead of indicator variables) agreed well with

the results from the model shown in Table 2. The ORs (and

95% CIs) per additional serving/week and the Ps for trend

were: broccoli, 0.91 (0.83-0.99), P 0.03; cabbage, 1.02

(0.91-I . 15), P = 0.74; cauliflower, 0.91 (0.79-1 .05). P = 0.20:

brussels sprouts, 1.00 (0.80-1.26), P = 0.97: and kale. 0.78

(0.62-0.99), P = 0.04.A main effect of the GSTMI null genotype was reported

earlier (23). The overall OR for adenomas with the GSTM/ null

genotype was 0.85 (95% CI, 0.65-1.10), consistent with no

GSTMI effect.

We used quartiles of vegetable intake to assess interaction

between GSTMI and vegetables, because there were too few

subjects in the reference category of “never” for all cruciferous

vegetables combined. For broccoli, quartiles produced the same

groupings as in Table 2. The overall risk of adenomas decreased

with increasing intake of cruciferous vegetables (Table 3). but

the effect was somewhat weaker than for broccoli (Table 2).

Analysis of quartiles was not possible for brussels sprouts andkale, because they were rarely eaten.

A protective effect of broccoli and cruciferous vegetables



procarcinogens ______

I enzymes

DNA

mutagens ‘�z:::I:II:damage � neoplasms

GSTs, mEH. detoxified� NQO1. UGTs � mutagens *excretion

inhibit : � induce

broccoli * isothiocyanates ______� conjugated � excretionGSTM1 isothiocyanatesGSTP1

650 GSTMJ, Broccoli, and Colorectal Adenomas

T able 3 ORs for cob rectal a denomas by GSTMI genotype and intake of cruci ferou s vegetab les and broccoli”

GSTMI null GSTMI non-null P’ Without GSTMI stratification

QI Q2 Q3 Q4 ‘�ir�nd QI Q2 Q3 Q4 ‘�trcnd Ql Q2 Q3 Q4 �t,c,d” �

Broccoli

OR 1(8) 0.65 0.99 0.36 0.0()l 1.30 0.80 0.61 0.74 0.43 0.01 (see Table 2 for broccoli)95C/� CI - 0.37-1.1 0.54-1.8 0.19-0.68 0.70-2.4 0.50-l.() 0.24-0.88 0.40-0.99

Controls/ 36/44 85/67 5 1/6)) 77/33 36/61 67/67 85/63 70/64

Cases

Servings/ 0 0.5 I 3.7 0 0.5 1 3.7

week

Cruciferous vegetables

OR 1(X) 1.08 0.85 0.52 0.02 1.24 1.04 0.77 0.95 0.71 0.26 1(8) 0.95 0.72 0.66 0.04 0.04

95(7� Cl - 0.64-1.8 0.5()-l.4 0.29-0.93 0.76-2.0 0.84-1.2 0.51-0.98 0.72-1.1 - 0.66-1.4 0.50-1.0 0.44-1.0

Controls/ 68/65 56/58 59/49 66/32 58/75 58/58 76/59 66/63 1 26/140 1 I 4/1 16 135/1 08 132/95

Cases

Servings/ 0.6 1.8 3.2 7.3 0.6 1.7 3.3 8.1 0.6 1.7 3.3 7.8week

‘, Effects of broccoli and cruciferous vegetables Iservings/week: sum of broccoli, cabbage/cole slaw, cauliflower, brussels sprouts. kale/mustard/chard greens (as indicator

variables for quartiles)l were adjusted for smoking (indicator variables for current smoking and ex-smoking): matching variables (age in 5-year categories: date of

sigmoidoscopy in live categories total): clinic: gender: saturated fat (indicator variables for quartiles of saturated fat intake in the total population): energy intake (calories

as a continuous variable): intake of fruits and vegetables indicator variables for quartiles of fruit and vegetable intake minus the specific vegetable(s) of interest in the

model I.

1’ p for trend was calculated from entering the mean number of servings/week in each quartile (calculated separately for GSTMI null and non-null genotypes) as a categorical

variable in the model. If the P for trend was calculated from entering broccoli and cruciferous vegetable intake as an ordered (continuous) variable into the model, we

obtained values of 0.04 (null genotype) and 0.26 (non-null genotype) for broccoli and 0.02 (null genotype) and 0.38 (non-null genotype) for cruciferous vegetables.‘ P for interaction was calculated as the likelihood ratio test comparing the model with three vegetable intake � GSTMI interaction terms to the model not containing any

interaction terms. We also investigated the interaction by entering vegetable intake as a continuous variable into the model. so that we had to also enter only one interaction

term between GSTMI and vegetable intake into the model. We obtained: P for interaction, 0.30 for broccoli: and P for interaction, 0.16 for cruciferous vegetables..1 p for trend was calculated by entering the mean number of servings/week in each category as a categorical variable in the model.

(. P for trend was calculated by entering the intake of the respective vegetable as a continuous (ordered) variable into the model.

Fig. I. Proposed relationships among broccoli, isothio-

cyanates, and mutagen metabolism. Isothiocyanates in-

hibit P450 enzymes (phase I enzymes) that transform

procarcinogens to mutagens (32). Isothiocyanates also

induce phase 2 enzymes (GSTs. NQOI. ,nEH. and UGTs),

which detoxify mutagens ( 14). Glutathione transferases

GSTMI and GSTPI conjugate isothiocyanates and divert

them from the enzyme induction pathway to excretion

( 17, 18). Lack of GSTM I is hypothesized to favor the

pathway that leads to enzyme induction, thereby protect-

ing against neoplasms. GSTs, glutathione transferases;

NQOJ, NAD(P)H:quinone oxidoreductase: mEH. micro-somal epoxide hydrolase: UGTs. UDP-glucuronosyl-

transferases.

was observed only among subjects with the GSTMJ null gen-otype, as judged from statistically significant Ps for trend(based on using mean intake of vegetable in each quartile as thecategorical variable). For broccoli, the P for trend was 0.001 forsubjects with the GSTMI null genotype, compared with 0.43 forsubjects with non-null genotypes. For cruciferous vegetables,

values of P for trend were 0.02 and 0.71, respectively. Theinteraction between GSTMI genotype and vegetable intake

reached statistical significance for broccoli (P = 0.01) but notfor cruciferous vegetables combined (P 0.26). The lowest

adenoma prevalence was among subjects with broccoli con-sumption in the highest quartile and with the GSTMI null

genotype (OR, 0.36; 95% CI, 0.19-0.68).We also assessed interaction by using vegetable intake as

a continuous variable in the model. For broccoli, the Ps fortrend were 0.04 and 0.26 for the GSTMJ null and non-null

genotypes, respectively. For cruciferous vegetables, corre-sponding Ps for trend were 0.02 and 0.38. Ps for interactionbased on the continuous models were 0.30 for broccoli and 0.16for cruciferous vegetables.

Discussion

Our aim is to identify mechanisms by which cruciferous veg-

etables protect against colorectal cancer. Although fiber and

folate may contribute (26-29), these vegetables also contain

nonnutrient compounds that block cancers in laboratory rats.

Examples are 3,3’-diindolylmethane, indole-3-carbinol, and

brassinin, found in brussels sprouts, cabbage, and cauliflower

(30, 31). Here we focused mainly on broccoli, because it

contains the isothiocyanate sulforaphane, a potent inducer of

carcinogen-detoxifying enzymes (14).

People consume several milligrams of isothiocyanates or

glucosinolates per day. Protection against cancer by these corn-

pounds is attributed to inhibition of phase 1 enzymes (32) and

induction of phase 2 enzymes (Fig. I ). Induction of detoxifying,

phase 2 enzymes has been called the electrophile counterattack

response (33, 34). However, glutathione transferases GSTM1

and GSTP1 conjugate isothiocyanates, leading to excretion (17,

18, 35). Roughly half the population lack GSTM 1 due to gene

deletion. This common genetic variant led to our strategy of



Cancer Epidemiology, Biomarkers & Prevention 651

indirectly assessing isothiocyanate levels in relation to colorec-

tal adenoma prevalence.

Subjects with high broccoli intake and the GSTMI null

genotype had the lowest adenoma prevalence. These subjectscould have higher isothiocyanate levels, due to high intake and

slow excretion. A GSTMJ interaction was observed only for

broccoli, perhaps due to the larger range of broccoli intakes

compared with other vegetables. Average broccoli intake was

1 .7 servings per week, compared with 0.97 for cabbage or cole

slaw; 0.70 for cauliflower; 0.27 for brussels sprouts; and 0.24

for kale, mustard, or chard greens.

The protective effect of broccoli could be due to other

compounds whose levels are higher in broccoli than in other

cruciferous vegetables. For example, broccoli has a higher

content of carotenoids. However, adjustment for carotenoids,

folate, vitamin C, and fiber in our previous analysis did not

substantially change protective effects of broccoli or crucifer-

ous vegetables (25).The GSTMJ null genotype conceivably could have raised

the risk of adenomas, instead of lowering the risk, through

production of less GSTM 1 for mutagen removal. However,

GSTM 1 is only one of several enzymes that detoxify electro-

philes. Lack of GSTMI may not matter for mutagen removal if

these compounds can be quickly cleared by other enzymes.

Also, GSTM1 is not inducible in the colon (36) or in hepato-

cytes (32).

Earlier reports showed increased risks of bladder and lung

cancer in smokers with the GSTMJ null genotype (37-39).

However, our previous work did not find a higher risk for

colorectal adenomas in smokers with the null genotype (23).

Analyses of broccoli and GSTMJ effects in the subgroup of

people who never smoked did not change results, again sup-

porting an interpretation of no interaction between GSTMI andsmoking for colorectal adenomas.

Strengths of the study were screening asymptomatic sub-jects for a first-time diagnosis of adenomas, a fairly large

sample size, and a response rate greater than 80% for both casesand controls. Diet effects from our study population (24, 29, 40,

4 1) agree with other published studies, supporting the validity

of our results. Weaknesses of the study were reliance on diet

questionnaires, the potential for recall bias, detection of prey-

alent rather than incident adenomas, and assessment of adeno-

mas in only the left colon. Some 50% of colon cancers occur in

the proximal segment (42). Up to 15-17% ofcontrols who have

no polyp by sigmoidoscopy may have a polyp in the right colon

(43, 44). If etiologies of left- versus right-sided adenomas are

different (45-47), then our study is an unbiased analysis of

left-sided adenomas. If etiologies are the same, our estimates of

effect would be biased toward the null.

In summary, prevalence of colorectal adenomas was low-

est among people with high broccoli intake and the GSTMI null

genotype. Slower conjugation and excretion of protective iso-

thiocyanates, due to lack of GSTM 1 enzyme, is a plausible

mechanism. Studies on enzyme variants and cancer risk usually

focus on carcinogens, like those in cigarette smoke. In contrast,

our work suggests a novel gene-environment interaction in-

volving an anticarcinogen. If confirmed, the result suggests that

people with the GSTMI null genotype may benefit more from

broccoli.

Acknowledgments

Chun-Ya Han. Bruce K. Lin, and David B. Lee provided technical assistance.

References

I. Steinmetz. K. A.. and Potter. J. D. Vegetables. fruit. and cancer. I. Epidemi.ology. Cancer Causes Control. 2: 325-357. 1991.

2. Steinmetz, K. A.. and Potter. J. D. Vegetables. fruit. and cancer. II. Mecha.

nisms. Cancer Causes Control. 2: 427-442. 1991.

3. Beecher. C. W. W. Cancer preventive properties of varieties of Brassico

oleracea: a review. Am. J. Clin. Nutr.. 59 (Suppl.): I 1665-I 170S. 1994.

4. Graham. S.. Dayal. H.. Swanson. M.. Mittelman. A.. and Wilkinson. G. Diet

in the epidemiology of cancer of the colon and rectum. J. NatI. Cancer. Inst.. 61:

709-714. 1978.

5. Kune. S., Kune, G. A.. and Watson, L. F. Case-control study of dietary

etiological factors: the Melboume colorectal cancer study. Nutr. Cancer. 9:

21-42, 1987.

6. Lee, H. P., Gourley. L.. Duffy. S. W.. Est#{232}ve.J.. Lee. J.. and Day. N. E.

Colorectal cancer and diet in an Asian population-a case-control study among

Singapore Chinese. Int. J. Cancer. 43: l(X)7-1016. 1989.

7. Young. T. B., and Wolf. D. A. Case-control study of proximal and distal colon

cancer and diet in Wisconsin. Int. J. Cancer. 42: 167-175. 1988.

8. Benito. E., Obrador, A.. Stiggelboul. A., Bosch. F. X.. Mulct. M.. Mu#{241}ot.N..

and Kaldor, J. A population-based case-control study of colorectal cancer in

Majorca. I. Dietary factors. Int. J. Cancer. 45: 69-76. 199().

9. Miller. A. B., Howe. G. R.. Jam, M.. Craib. K. J. P.. and Harrison. L. Fooditems and food groups as risk factors in a case-control study of diet and cob-rectal

cancer. Int. J. Cancer. 32: 155-161. 1983.

10. Colditz. G. A., Branch. L. G.. Lipnick. R. J.. Willett. W. C.. Rosner. B..

Posner, B. M., and Hennekens, C. H. Increased green and yellow vegetable intake

and lowered cancer deaths in an elderly population. Am. J. Clin. Nutr.. 41: 32-36.

1985.

I I . LaVecchia, C.. Negri. E.. Decarli. A.. D’Avanzo. B.. Galbotti. L.. Gentile. A..

and Franceschi, S. A case-control study of diet and cob-rectal cancer in northem

Italy. Int. J. Cancer. 41: 492-498. 1988.

12. Bidoli. E.. Franceschi, S.. Talamini. R.. Barra. S.. and LaVecchia. C. Food

consumption and cancer of the colon and rectum in North-eastem Italy. Int. J.

Cancer, 50: 223-229, 1992.

13. Fenwick, G. R.. Heaney, R. K.. and Mullin, W. J. Glucosinolates and their

breakdown products in food and food plants. CRC Cr0. Rev. Food Sci. Nutr.. /8:

123-201. 1983.

14. Prochaska. H. J.. Santamaria. A. B.. and Talalay. P. Rapid detection ofinducers ofenzymes that protect against carcinogens. Proc. Nail. Acad. Sci. USA.

89: 2394-2398. 1992.

15. Zhang. Y.. Talalay, P.. Cho. C-G.. and Posner. G. H. A major inducer of

anticarcinogenic protective enzymes from broccoli: isolation and elucidation of

structure. Proc. NatI. Acad. Sci. USA. 89: 2399-2403. 1992.

16. Zhang, Y., Kensler. T. W., Cho. C-G., Posner. G. H.. and Talalay. P.

Anticarcinogenic activities of sulforaphane and structurally related synthetic

norbomyl isothiocyanates. Proc. NatI. Acad. Sci. USA. 9/: 3147-3150. 1994.

17. Zhang. Y.. KoIm. R. H.. Mannervik. B.. and Talalay. P. Reversible conju-

gation of isothiocyanates with glutathione catalyzed by human glutathione trans-

ferases. Biochem. Biophys. Res. Commun.. 206: 748-755. 1995.

18. KoIm, R. H.. Danielson. U. H.. Zhang. Y.. Talalay. P.. and Mannervik. B.

Isothiocyanates as substrates for human glutathione transferases: structure-activity

studies. Biochem. J.. 3/1: 453-459. 1995.

19. Seideg#{227}rd. J.. Vorachek. W. R.. Pero, R. W.. and Pearson. W. R. Hereditary

differences in the expression of the human glutathione transferase active on

trans-stilbene oxide are due to a gene deletion. Proc. NatI. Acad. Sci. USA. 85:

7293-7297. 1988.

20. Lin. H. J.. Han. C-Y.. Bemstein. D. A.. Hsiao. W., Lin. B. K.. and Hardy. S.Ethnic distribution ofthe glutathione transferasc Mu I-I (GSTM/) null genotype

in 1473 individuals and application to bladder cancer susceptibility. Carcinogen-

esis (Lond.), /5: 1077-1081, 1994.

21. Wild. C. P.. Fortuin. M.. Donato. F.. Whittle. H. C.. Hall. A. J.. Wolf. C. R..

and Montesano. R. Aflatoxin. liver enzymes. and hepatitis B virus infection in

Gambian children. Cancer Epidemiol. Biomark. Prey.. 2: 555-561. 1993.

22. Rimm. E. B.. Giovannucci, E. L.. Stampfer. M. J.. Colditz. G. A.. LiOn. L. B..

and Willett. W. C. Reproducibility and validity of an expanded self-administered

semiquantitative food frequency questionnaire among male health professionals.

Am. J. Epidemiol.. /35: 1114-1 126. 1992.

23. Lin. H. J., Probst-Hensch, N. M.. Ingles. S. A.. Han. C-Y.. Lin. B. K.. Lee.

D. B., Frankl, H. D.. Lee, E. R.. Longnecker. M. P.. and Haile. R. W. Glutathionetransferase (GSTM/) null genotype. smoking. and prevalence of colorectal ade-

nomas. Cancer Res.. 55: 1224-1226. 1995.

24. Longnecker. M. P.. Chen, M-J.. Probst-Hensch, N. M.. Harper. J. M.. Lee.

E. R., Frankl. H. D., and Haile. R. W. Alcohol and smoking in relation to the



652 GSTMI, Broccoli, and Colorectal Adenomas

prevalence of adenomatous colorectal polyps detected at sigmoidoscopy. Epide-

miology. 7: 275-280. 1996.

25. Write. J. S.. Longnecker. M. P.. Bird. C. L.. Lee. E. R., Frankl. H. D.. and

Haile. R. W. Relation of vegetable. fruit. and grain consumption to colorectal

adenomatous p()lyps. Am. J. Epidemiol.. /44: 1015-1025. 1996.

26. Giovannucci. E., Stampfer. M. J.. Colditz, G.. Rimm, E. B.. and Willett,W. C. Relationship of diet to risk of colorectal adenoma in men. J. NatI. Cancer

Inst.. 84: 91-98. 1992.

27. SandIer. R. S.. Lyles. C. M.. Peipins. L. A.. McAuliffe. C. A., Woosley, J. T.,and Kupper. L. L. Diet and risk of colorectal adenomas: macronutrients, choles-

terol, and fiber. J. NatI. Cancer Inst., 85: 884-891, 1993.

28. Giovannucci. E.. Stampfer. M. J.. Colditz. G. A.. Rimm. E. B.. Trichopoulos,

D.. Rosner. B. A.. Speizer. F. E., and Willett. W. C. Folate, methionine. and

alcohol intake and risk of colorectal adenoma. J. NatI. Cancer Inst.. 85: 875-884.

I 993.

29. Bird, C. L.. Swenseid, M. E.. Write, J. S.. Shikany. J. M.. Hunt, I. F., Frankl,

H. D.. Lee, E. R.. Longnecker, M. P.. and Haile, R. W. Red cell and plasma folate,

folate consumption. and the risk of colorectal adenomatous polyps. Cancer

Epidemiol. Biomark. Prey.. 4: 7(83-714. 1995.

30. Loub, W. D.. Wattenberg. L. W.. and Davis. D. W. Aryl hydrocarbon

hydroxyla.se induction in rat tissues by naturally occurring indoles of cruciferous

plants. J. NatI. Cancer Inst.. 54: 985-988. 1975.

3 1 . Mehta, R. G.. Liu. J.. Constantinou. A., Thomas, C. F.. Hawthorne, M., You.M.. Gerhiiuser. C.. Pezzuto. J. M., Moon, R. C.. and Moriarty, R. M. Cancer

chemopreventive activity of bra.ssinin, a phytoalexin from cabbage. Carcinogen-

esis (Lond.). /6: 399-4()4. 1995.

32. Mah#{233}o,K., Morel. F.. Langouet, S.. Kramer, H.. Fence. E. L., Ketterer, B.,

and Guilbouzo. A. Inhibition of cytochromes P-450 and induction of glutathione

S-transferases by sulforaphane in primary human and rat hepatocytes. Cancer

Rca.. 57: 3649-3652, 1997.

33. Spamins. V. L.. Venegas. P. L.. and Wattenberg. L. W. Glutathione S-

transferase activity: enhancement by compounds inhibiting chemical carcinogen-

esis and by dietary constituents. J. NatI. Cancer. Inst.. 68: 493-496, 1982.

34. Prestera. T.. Zhang. Y.. Spencer, S. R., Wilczak, C. A.. and Talalay. P. The

electrophile counterattack response: protection against neoplasia and toxicity.

Ads. Enzyme Regul.. 33: 281-296. 1993.

35. Zhang, Y., and Talalay. P. Anticarcinogenic activities of organic isothiocya-

nates: chemistry and mechanisms. Cancer Res., 54 (Suppl.): 1976s-1981s, 1994.

36. Nijhoff. W. A.. Grubben, M. J. A. L.. Nagengast, F. M., Jansen, J. B. M. J..

Verhagen. H.. van Poppel. G.. and Peters. W. H. M. Effects of consumption of

Brussels sprouts on intestinal and lymphocytic glutathione S-transferases in

humans. Carcinogenesis (Lond.). /6: 2125-2128, 1995.

37. Bell, D. A., Taylor. J. A., Paulson, D. F., Robertson, C. N., Mohler, J. L., and

Lucier. G. W. Genetic risk and carcinogen exposure: a common inherited defect

of the carcinogen-metabolism gene glutathione S-transferase M I (GSTM I) that

increases susceptibility to bladder cancer. J. NatI. Cancer Inst., 85: I 159-I 164,

1993.

38. BrockmOller, J.. Kerb, R., Drakoulis, N., Staffeldt, B., and Roots, I. Gluta-

thione S-transferase M 1 and its variants A and B as host factors of bladder cancersusceptibility: a case-control study. Cancer Res., 54: 4103-4111. 1994.

39. Nazar-Stewart, V., Motulsky. A. G., Eaton, D. L., White, E., Homung, S. K.,

Leng, Z-T., Stapleton. P.. and Weiss, N. S. The glutathione S-transferasc �z

polymorphism as a marker for susceptibility to lung carcinoma. Cancer Res., 53:

2313-2318, 1993.

40. Bird, C. L., Witte, J. S., Swenseid, M. E., Shikany, J. M., Hunt, I. F., Frankl,

H. D., Lee, E. R., Longnecker, M. P., and Haile, R. W. Plasma ferritin, iron intake,

and the risk of colorectal polyps. Am. J. Epidemiol.. 144: 34-41, 1996.

41. Enger, S. M., Longnecker. M. P., Chen, M-J., Harper, J. M., Lee, E. R.,

Frankl, H. D., and Haile, R. W. Dietary intake of specific carotenoids and

vitamins A. C, and E, and prevalence of colorectal adenoma.s. Cancer Epidemiol.

Biomark. Prey., 5: 147-153. 1996.

42. Slattery, M. L., Friedman, G. D., Potter, J. D., Edwards, S., Caan, B. J., and

Samowitz, W. A description of age. sex, and site distributions of colon carcinoma

in three geographic areas. Cancer (Phila.), 78: 1666-1700, 1996.

43. Rex, D. K., Lehman, G. A., Ulbright, T. M.. Smith, J. J., Pound, D. C.,

Hawes. R. H., Helper, D. J.. Wiersema, M. J., Langefeld, C. D., and Li, W.

Colonic neopla.sia in a.symptomatic persons with negative fecal occult blood tests:influence ofage, gender. and family history. Am. J. Ga.stroenterol., 88: 825-831,

I 993.

44. Foutch, P. G.. Mai, H.. Pardy. K., Disario, J. A.. Manne, R. K., and Kerr, D.Flexible sigmoidoscopy may be ineffective for secondary prevention of colorectal

cancer in asymptomatic, average-risk men. Dig. Dis. Sci., 36: 924-928, 1991.

45. Potter, J. D., Slattery, M. L., Bostick, R. M., and Gapstur, S. M. Colon cancer:

a review of the epidemiology. Epidemiol. Rev., 15: 499-545, 1993.

46. Watatani, M., Yoshida, T., Kuroda, K., leda, S.. and Yasutomi. M. Allelic

loss of chromosome lip, mutation of the p53 gene, and microsatellite instability

in right- and left-sided colorectal cancer. Cancer (Phila.), 77: 1688-1693, 1996.

47. Bufill, J. A. Colorectal cancer: evidence for distinct genetic categories based

on proximal or distal tumor location. Ann. Intem. Med., 113: 779-788, 1990.



1998;7:647-652. Cancer Epidemiol Biomarkers Prev H J Lin, N M Probst-Hensch, A D Louie, et al. prevalence of colorectal adenomas.Glutathione transferase null genotype, broccoli, and lower

Updated version

http://cebp.aacrjournals.org/content/7/8/647

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/7/8/647To request permission to re-use all or part of this article, use this link



http://cebp.aacrjournals.org/cgi/alerts

mailto:[email protected]



glutathione transferase null genotype, broccoli, and lower ...glutathione transferase 51 i. ......

Documents