REPORTS

Individuals homozygous for an allele of (called the c2 variant) contains polymor-

the CYP2E1 gene that is detected by phic base substitution sites in a region of
CYP2E1 Genetic Rsa I digestion (c2 allele) were found to the gene that is not transcribed but that
Polymorphisms and Risk of have an increased risk of nasopharyn- appears to be involved in the transcrip-
geal carcinoma (relative risk [RR] = tional regulation of CYP2E1 expression.
Nasopharyngeal Carcinoma
2.6; 95% confidence interval [CI] = A study (12) has suggested that the vari-
in Taiwan 1.2–5.7); this effect was limited to non- ant form of the gene detectable by Rsa I

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

smokers (RR = 9.3; 95% CI = 2.7–32) digestion is expressed at higher levels
Allan Hildesheim, Lucy M. and was not affected by alcohol con- than the wild-type form of the gene. Fur-
Anderson, Chien-Jen Chen, sumption. Conclusions: Our findings thermore, individuals with the variant
Yu-Juen Cheng, Louise A. suggest that the CYP2E1 genotype is a form of the gene have been shown in
determinant of nasopharyngeal carci- some studies to have higher hepatic
Brinton, Ann K. Daly, C. D. CYP2E1 messenger RNA and protein
noma risk. [J Natl Cancer Inst 1997;89:
Reed, I-How Chen, Neil E. levels and a greater ability to metabolize
1207–12]
Caporaso, Mow-Ming Hsu, acetaminophen, a drug metabolized in
Jen-Yang Chen, Jeffrey R. Idle, part by CYP2E1 (13–17). Interestingly,
Nasopharyngeal carcinoma is a rare tu- the c2 variant of the CYP2E1 gene has
Robert N. Hoover, Czau-Siung mor in most parts of the world, but occurs been shown to be present in approxi-
Yang, Saranjit K. Chhabra* at relatively high rates in some geo- mately 20%–25% of Asians compared
graphic regions and among certain ethnic with fewer than 10% of Caucasians
groups, with the highest incidence world- (3,18–22). Furthermore, detection of the
Background: Nasopharyngeal carci- wide being reported from southeast Asia CYP2E1 c2 allele has been shown to cor-
noma occurs disproportionately among (1). Numerous environmental factors relate strongly with the presence of the
individuals of Chinese descent. The cy- have been associated with risk of devel- rare allele detectable by Dra I digestion
tochrome P450 2E1 enzyme (CYP2E1) oping nasopharyngeal carcinoma, includ- (C allele) (3,22).
is known to activate nitrosamines and ing infection with the Epstein–Barr virus, Recently, we reported an association
other carcinogens that are possibly in- occupational exposure to formaldehyde, between the presence of the Rsa I c2 and
volved in the development of this dis- cigarette smoking, and various dietary Dra I C alleles of the CYP2E1 gene and
ease. Certain alleles of the CYP2E1 factors (2). Host factors, including human risk of nasopharyngeal carcinoma in a
gene are thought to be more highly ex- leukocyte antigens and cytochrome P450 study of 50 nasopharyngeal carcinoma
pressed than others, and their distribu- 2E1 (CYP2E1), have also been postulated
tion varies between Asian and Cauca- to be important in nasopharyngeal carci-
sian populations. We conducted a case– noma development (2,3).
control study to investigate whether CYP2E1 is an enzyme involved in the *Affiliations of authors: A. Hildesheim, L. A.
such variations affect the risk of devel- metabolic activation of various procar- Brinton, N. E. Caporaso, R. N. Hoover, Division of
oping nasopharyngeal cancer. Methods: cinogens, including various low-molecu- Cancer Epidemiology and Genetics, National Can-
Three hundred sixty-four patients with cer Institute, Bethesda, MD; L. M. Anderson, C. D.
lar-weight nitrosamines, such as N-
Reed, S. K. Chhabra, Laboratory of Comparative
nasopharyngeal carcinoma (96% of nitrosodimethylamine (NDMA) and the Carcinogenesis, National Cancer Institute-Frederick
378 eligible patients) and 320 control tobacco-specific nitrosamine N-nitro- Cancer Research and Development Center, Freder-
subjects (86% of 374 eligible subjects) sonornicotine (NNN) (4,5). The CYP2E1 ick, MD; C.-J. Chen, Y.-J. Cheng (Institute of Epi-
were studied. A risk factor question- gene is expressed and is inducible in he- demiology, College of Public Health), J.-Y. Chen,
C.-S. Yang (Institute of Microbiology, College of
naire was administered to participants patic tissue (6). Studies (7–11) have also
Medicine), National Taiwan University, Taipei;
to assess factors postulated to be linked demonstrated the expression of CYP en- A. K. Daly, J. R. Idle, GenoType Limited, New-
to nasopharyngeal carcinoma. Periph- zymes in the nasal epithelium of numer- castle upon Tyne, U.K.; I.-H. Chen, Department of
eral blood was obtained from all sub- ous animal species and humans. Otolaryngology, MacKay Memorial Hospital, Tai-
jects and DNA was purified from The CYP2E1 gene is present in the pei; M.-H. Hsu, Department of Otolaryngology, Na-
tional Taiwan University Hospital, Taipei.
nucleated cells. A polymerase chain re- population in various polymorphic forms.
Correspondence to: Allan Hildesheim, Ph.D., Na-
action-based restriction fragment Polymorphisms detectable by Dra I and tional Institutes of Health, Executive Plaza North,
length polymorphism assay that used Taq I digestion are not thought to affect Rm. 443, Bethesda, MD 20892. E-mail:
the restriction enzymes Rsa I and Dra I transcription or function of the enzyme Hildesha@epndce.nci.nih.gov
was used to detect wild-type and vari- coded for by the gene. In contrast, the See ‘‘Notes’’ following ‘‘References.’’
ant forms of the CYP2E1 gene. Results: variant detectable by Rsa I digestion © Oxford University Press

Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997 REPORTS 1207
case subjects and 50 control subjects (3). National Cancer Institute and the National Taiwan ses. The median age of the case subjects
In this study, we have expanded our initial University. was 44.5 years (mean, 45.5 years; range,
For the 100 subjects reported on previously, DNA
study to include 364 patients with naso- was isolated from peripheral blood mononuclear
15–74 years); 69.8% of the case subjects
pharyngeal carcinoma and 320 control cells (PBMCs) obtained from whole blood and the were male. Comparable statistics for con-
subjects. This was done to 1) examine our polymerase chain reaction (PCR)-based restriction trol subjects were 45 years (mean, 46.0
initial findings with greater statistical fragment length polymorphism assay was performed years; range, 19–74 years) and 69.3%
power, 2) determine the independent ef- as described (3). The remaining subjects were pro- were male. Ethnically, 81.6% of the case
cessed and tested as follows. Genomic DNA was
fects of polymorphisms detectable by Rsa isolated from PBMCs as previously described (23).
subjects and 73.0% of the control subjects
I and Dra I digestion, 3) explore the effect Amplification of the desired polymorphic sequences were of Fukienese origin; 13.2% of the
of CYP2E1 polymorphisms on risk asso- was carried out by use of PCR with the Perkin- case subjects and 11.0% of the control
ciated with exposures to nitrosamine- Elmer PCR kit (The Perkin-Elmer Corp., Branch- subjects were of Hakka, Cantonese, or
containing cigarette smoke, and 4) con- burg, NJ). The total volume of 50 mL consisted of 10 Taiwanese (Aboriginal) origin; while the
mM Tris (pH 4 8.3), 50 mM KCl, 1.5 mM MgCl2,
sider the effect of ethanol, a known 125 mM deoxynucleoside triphosphate, 0.2 mM
remaining 5.2% of the case subjects and
modulator of hepatic CYP2E1, on the primers, 2 U Taq polymerase, 0.4 mg TaqStart an- 16.0% of the control subjects were of
other Han origin (P 4 .001). Of the case

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

CYP2E1/disease association. tibody (CLONTECH Laboratories, Inc., Palo Alto,
CA), and 50–100 ng of template DNA. Primers were subjects, 42.6% reported less than a junior
Methods identical to those used previously (3). Amplification high school education compared with
was carried out for 30 cycles at 92 °C for 1 minute,
Patients with histologically confirmed incident 60 °C for 1 minute, and 72 °C for 2 minutes in a
31.4% of the control subjects; conversely,
nasopharyngeal carcinoma were recruited over a 31⁄2 Perkin-Elmer Thermocycler. Twenty microliters of 40.7% of the case subjects and 50.2% of
year period from July 15, 1991, through December the 410-base-pair (bp) and 995-bp products were the control subjects reported a high school
31, 1994, at two large referral hospitals in Taipei, digested with 10 U Rsa I (410-bp product) or Dra I education or higher (P for trend 4 .003).
Taiwan: the National Taiwan University and (995-bp product) (Boehringer Mannheim Corp., In-
MacKay Memorial Hospitals. Case subjects were
Other factors found to be associated with
dianapolis, IN) restriction enzyme overnight at
eligible for study if they were less than 75 years of 37 °C. The restricted product was run on a 10% pre-
disease in our study or previous studies
age, had had no prior diagnosis or treatment for cast Tris–Borate-EDTA nondenaturing polyacryl- and considered as potential confounders
nasopharyngeal carcinoma, and had resided in Tai- amide gel (Novex, San Diego, CA) at 100 V for 1.5 included consumption of Guangdong
pei city or county for 6 months or longer. Three hours. Bands were visualized with an ultraviolet (Cantonese) salted fish during childhood
hundred seventy-eight eligible case subjects were
identified.
transilluminator after ethidium bromide staining. (RR 4 1.4; 95% CI 4 0.47–4.5), a fam-
Successful genotyping was obtained for 364 of
For each eligible case subject, we attempted to the 369 case subjects and all 320 control subjects
ily history of nasopharyngeal carcinoma
match one population control subject. Three hun- tested. A 6% masked, random sample (n 4 42) of (RR 4 7.7; 95% CI 4 2.3–26), and ciga-
dred seventy-four eligible control subjects were subjects were tested twice. The results were concor- rette smoking (RR 4 2.1; 95% CI 4 1.2–
matched to eligible case patients. Matching was ac- dant for all masked duplicate sets. An additional 3.7 for more than 30 years of use relative
complished by use of listings available through the group of 10 samples were tested twice, the first time
National Household Registration System. Each con- to nonsmokers).
using the method described in our previous publica-
trol subject was selected by visiting the registry of- tion (3) and the second time using refinements de-
The Rsa I c2 form of the CYP2E1 gene
fice located in the district (if in Taipei city) or town- scribed herein to assure comparability. Results for was observed among our control popula-
ship (if in Taipei county) in which the case subject all 10 samples were concordant. tion with an allele frequency of 20.5%
resided. Within these large geographic units, a ran- Allele and genotype frequencies were computed. compared with 22.3% among case sub-
dom Li (communities of 1000–1500 households
within each district/township) was selected. From
The Pearson correlation coefficient was used to de- jects (P 4 .40). Similarly, the Dra I C
termine the correlation between genotypic variables
within the selected Li, a random Lin (neighborhoods (24). The relative risk (RR), as estimated by the odds
form of the CYP2E1 gene was observed
of 50–100 households) was selected. Individuals ratio, was used to estimate the association between with an allele frequency of 23.6% among
within the selected Lin were then enumerated and a genotype or environmental exposures and disease control subjects and 25.4% among case
control who matched the eligible nasopharyngeal (25); 95% confidence intervals (CIs) were calculated subjects (P 4 .64). Both Rsa I and Dra I
carcinoma case subject on age (5-year groupings) to determine the statistical significance of findings
and sex was randomly selected. Control subjects alleles were found to be in Hardy–
(25). Multivariate analyses were performed by use
were required to have no history of nasopharyngeal of logistic regression methods (26). Stratified analy-
Weinberg equilibrium. The distribution of
carcinoma before identification for our study. Ineli- ses were conducted to examine the joint effect of the Rsa I and Dra I polymorphisms was
gible control subjects were replaced, but no replace- CYP2E1, smoking, and alcohol consumption. The observed to be similar in the different
ment was attempted for eligible control subjects who significance of trends across multiple levels of an Chinese ethnic subgroups included in our
refused participation. exposure was assessed by categorizing the variable
All subjects were requested to respond to a per- study (Fukienese, Hakka, Cantonese, Ab-
of interest and treating the scored variable as a con-
sonal interview that elicited detailed information on tinuous variable in the regression model. Statistical
original, and other Han).
potential risk factors for nasopharyngeal carcinoma, significance of the observed interactions was deter- Case and control subjects are com-
including sociodemographic characteristics, ciga- mined by comparing the fit of the logistic model that pared with respect to CYP2E1 genotypic
rette smoking, alcohol consumption, dietary history, included only the main effects with that of the model variants in Table 1. Twenty-seven case
occupational history, and a family history of cancer. that included separate estimates of risk for CYP2E1
Case subjects were interviewed prior to treatment, at subjects (7.4%) and nine control subjects
genotype for each level of the effect modifier (e.g.,
the time of their biopsies for confirmation of disease. smoking). All P values resulted from two-sided sta-
(2.8%) were homozygous for the Rsa I c2
In addition to the interview, subjects were asked to tistical tests. form of the CYP2E1 gene; 30 case sub-
consent to the collection of approximately 30 mL jects (8.2%) and 14 control subjects
blood. Three hundred sixty-nine case subjects (98%)
Results (4.4%) were homozygous for the Dra I C
and 320 control subjects (86%) gave verbal or writ-
ten informed consent for both the interview and form of the CYP2E1 gene. Those homo-
blood collection. This study was reviewed and ap- A total of 364 case subjects and 320 zygous for the c2 variant were at a 2.6-
proved by the Institutional Review Boards at the control subjects are included in the analy- fold increased risk of developing naso-

1208 REPORTS Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997
Table 1. Frequency distribution among case and control subjects and relative risk associated with genotypic variants of CYP2E1 detected by restriction enzyme
digestion with Rsa I or Dra I

Frequency

Case subjects Control subjects Relative 95% confidence Relative 95% confidence
CYP2E1 genotype (n 4 364) (n 4 320) risk (1)* interval risk (2)† interval
Rsa I
c1-c1 229 198 1.0 1.0
c1-c2 108 113 0.83 0.60–1.2 0.79 0.44–1.4
c2-c2 27 9 2.6 1.2–5.7 3.2 0.69–15
Dra I
D-D 209 183 1.0 1.0
D-C 125 123 0.89 0.65–1.2 1.1 0.61–1.9
C-C 30 14 1.9 0.98–3.7 0.81 0.20–3.3

*Controlled for age and sex.

†Controlled for age, sex, and the variables listed in the table (i.e., Rsa I adjusted for Dra I and vice versa).

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

pharyngeal carcinoma (95% CI 4 1.2– risk of disease (95% CI 4 0.69–15). Con- 1.2–3.7). Smokers (defined as subjects
5.7), while those homozygous for the C versely, adjustment of the Dra I effect for who reported smoking regularly for a pe-
variant were at 1.9-fold increased risk of Rsa I polymorphisms resulted in an RR of riod exceeding 6 months) who reported
developing disease (95% CI 4 0.98–3.7). 0.81 associated with homozygosity to the 30 or fewer years of smoking were at
Adjustment for education, ethnicity, a C form of CYP2E1 (95% CI 4 0.20–3.3). similar risk to never smokers (RR 4 1.2;
family history of nasopharyngeal carci- Given these findings, further analyses fo- 95% CI 4 0.80–1.8). No effect on dis-
noma, consumption of Guangdong salted cused on investigating the role of the c2 ease risk was observed for alcohol con-
fish at a young age, or smoking did not variant of the CYP2E1 gene on nasopha- sumption. Compared with nondrinkers,
materially affect our estimates of risk. ryngeal carcinoma risk. those who reported drinking two times
Subjects heterozygous at the CYP2E1 lo- Since cigarette smoking is an impor- per week or less and more than two times
cus (by either Rsa I or Dra I digestion) tant source of nitrosamine exposure and weekly had an RR of disease of 0.89
were not found to be at increased risk of alcohol consumption is known to modu- (95% CI 4 0.56–1.4) and 0.92 (95% CI
disease relative to those homozygous for late CYP2E1 activity, we examined the 4 0.58–1.4), respectively. No association
the more common allele (RR 4 0.83 for effect of these exogenous factors on dis- between alcohol consumption and disease
Rsa I c1-c2 and 0.89 for Dra I D-C). Pres- ease risk and the joint effect of these fac- was observed when the level of use was
ence of the c2 and C variants of the tors and CYP2E1 polymorphisms on na- examined in finer strata.
CYP2E1 gene was highly correlated (sp sopharyngeal carcinoma risk. Smoking We observed a strong and statistically
4 0.85 among control subjects and 0.89 was positively associated with risk of na- significant modification of the CYP2E1
among case subjects). After adjustment sopharyngeal carcinoma. Those reporting effect by smoking status (Table 2).
for Dra I polymorphisms, individuals ho- more than 30 years of smoking were at a Among never smokers, individuals homo-
mozygous for the c2 variant of CYP2E1 2.1-fold increased risk of disease com- zygous for the CYP2E1 c2 variant were at
were found to be at a 3.2-fold increased pared with never smokers (95% CI 4 a 9.3-fold increased risk of disease rela-

Table 2. Joint effect of CYP2E1 genotypic variants detected by Rsa I digestion and cigarette smoking/alcohol consumption on nasopharyngeal carcinoma risk

c1-c1 c1-c2 c2-c2

No. of case/No. of control Relative No. of case/No. of control Relative No. of case/No. of control Relative
Smoking subjects* risk† subjects* risk† subjects* risk†

Nonsmokers 91/113 1.0 61/53 1.4 22/3 9.3‡

Smokers, y 138/84 2.3‡ 47/59 1.1 5/6 1.1
ø15 32/27 1.6 14/13 1.3 2/1 2.4
16-30 62/37 2.2‡ 17/31 0.70 2/3 0.84
>30 44/20 3.6‡ 16/15 1.9 1/2 0.96
Test for interaction: chi-squared 4 23.44 (6 df); P<.001
Nondrinkers 162/139 1.0 78/78 0.85 22/7 2.7‡
Drinkers 67/58 0.96 30/34 0.73 5/2 2.1
ø2 times/wk 34/28 0.98 14/17 0.68 2/1 1.7
>2 times/wk 33/30 0.94 16/17 0.78 3/1 2.4
Test for interaction: chi-squared 4 0.25 (4 df); P>.25

*Two control subjects were excluded from analysis because of missing information on smoking history.
†Controlled for age and sex.
‡95% confidence interval excludes 1.0 (i.e., P<.05).

Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997 REPORTS 1209
tive to nonsmokers who were homozy- tion was examined (P>.25; Table 2). Ad- carcinoma pathogenesis of procarcino-
gous for the CYP2E1 c1 allele (95% CI justment for cigarette smoking did not al- gens known to be activated by CYP2E1.
4 2.7–32). Conversely, among smokers, ter the alcohol/CYP2E1 findings. Our One such group of procarcinogens, nitro-
there was no evidence that being homo- study size did not allow us to examine the samines, have previously been hypoth-
zygous for the CYP2E1 c2 variant in- joint effect on disease risk of consump- esized to be linked to nasopharyngeal
creased disease risk, relative to smokers tion of salted fish or family history of na- carcinoma (2). Studies (27–39) have dem-
who were homozygous for the wild-type sopharyngeal carcinoma and CYP2E1 ge- onstrated an association between factors
gene, although this was based on only five notype. known to contain nitrosamines and naso-
case subjects and six control subjects who We also examined the effect of pharyngeal carcinoma risk, an increased
were both smokers and c2 homozygotes. CYP2E1 genotype separately for males (n ability of individuals residing in high-risk
When we restricted our analysis to never 4 476) and females (n 4 208). A stron- areas for nasopharyngeal carcinoma to
smokers who reported no passive smok- ger association between homozygosity for form nitrosamines endogenously, and the
ing exposure in their household during CYP2E1 c2 and nasopharyngeal carci- ability of nitrosamine-containing foods to
childhood and adult life (n 4 91), eight noma was observed among females (RR induce nasal cavity tumors in animals.
4 17; 95% CI 4 2.1–130) compared

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

case subjects and no control subjects were In addition to nitrosamines, CYP2E1 is
observed to be homozygous for the with males (RR 4 1.2; 95% CI 4 0.50– also involved in the activation of other
CYP2E1 c2 variant. Furthermore, among 3.1), but this effect was largely explained procarcinogenic compounds (5). The pos-
these nonsmokers who reported no house- by the fact that the majority of males re- sibility that the CYP2E1/nasopharyngeal
hold passive smoke exposure, a signifi- ported a history of smoking (68.6%) carcinoma association results from the
cant increase in risk among individuals while very few women reported having ability of CYP2E1 to activate compounds
heterozygous at the CYP2E1 locus com- ever smoked (6.3%). When sex-specific other than nitrosamines should therefore
pared with individuals homozygous for analyses were performed separately for
not be overlooked. For example, CYP2E1
the CYP2E1 c2 allele was observed (RR smokers and nonsmokers, the results were
is involved in the metabolism of haloge-
4 3.9; 95% CI 4 1.4–11). When the RR as follows. Excesses in risk of disease as-
nated hydrocarbons, including organic
of disease associated with cigarette smok- sociated with being homozygous for the
solvents linked to the development of nu-
ing was investigated within strata of CYP2E1 c2 variant compared with being
merous tumors (40). In fact, results from a
CYP2E1 genotype, a dose-related effect homozygous for the wild-type allele were
recent study (31) suggested that exposure
of years of cigarette smoking was ob- observed for both sexes among nonsmok-
to solvents may be associated with naso-
served only among individuals who were ers; the risk estimate associated with ho-
pharyngeal carcinoma risk, although ex-
homozygous for the wild-type CYP2E1 mozygosity to the CYP2E1 variant was
gene. Among these subjects, risk in- 16-fold among females (95% CI 4 2.0– amination of the independent effect of
creased from 1.6 for those reporting 15 or 130) and sixfold among males (95% CI solvents in that study was hampered by
fewer years of cigarette smoking (95% CI 4 1.2–30). No effect of the CYP2E1 c2- high correlations observed between sol-
4 0.82–2.9), to 2.2 for those reporting c2 genotype was observed among male vent exposure and exposure to other oc-
16–30 years of use (95% CI 4 1.3–3.8), smokers (RR 4 0.39; 95% CI 4 0.11– cupational factors.
to 3.6 for those who reported smoking for 1.4). Too few women reported a history A provocative and unexpected finding
more than 30 years (95% CI 4 1.8–6.9), of cigarette smoking (n 4 13) to enable of our study was the observation that the
relative to nonsmokers who were homo- us to examine the effect of CYP2E1 ge- association between the CYP2E1 c2 allele
zygous for the wild-type form of notype on risk of disease among female and nasopharyngeal carcinoma risk was
CYP2E1. Among subjects homozygous smokers. limited to nonsmokers and that, among
for the CYP2E1 c2 variant, a trend of de- nonsmokers who reported no household
creasing risk with increasing years of Discussion exposure to passive smoke, an effect was
smoking was observed, but this trend was observed among individuals heterozygous
not statistically significant (P 4 .13). Results from this study confirm our at the CYP2E1 locus. Interpretation of
Similarly, no significant trend of increas- preliminary report suggesting an associa- this finding should be made with caution
ing or decreasing risk with increasing tion between CYP2E1 genetic polymor- and requires confirmation. If confirmed,
years of smoking was observed among in- phisms and risk of nasopharyngeal carci- however, several explanations exist. One
dividuals who were heterozygous at the noma (3). In the present study of 364 explanation for our finding is that tobacco
CYP2E1 locus (P 4 .59). Parallel results patients diagnosed with nasopharyngeal products have inhibiting effects that coun-
were observed when intensity of cigarette carcinoma and 320 control subjects, an teract the effect of CYP2E1-induced acti-
smoking was examined, although the ef- RR of 2.6 was observed among individu- vation. In fact, a recent study (41) has
fect was not as striking, probably because als homozygous for the CYP2E1 c2 al- suggested that tobacco smoke constitu-
intensity of cigarette smoking was not ob- lele. ents, including nicotine and cotinine, in-
served in our study population to be as Since CYP2E1 is involved in the meta- hibit the activation of two nitrosamines
strongly predictive of nasopharyngeal bolic activation of numerous procarcino- present in cigarette smoke, NDMA and
carcinoma risk as duration of use (data gens (5), our finding of an association be- NNK, in a dose-dependent manner. It is
not shown). No effect modification was tween CYP2E1 genetic polymorphisms also possible that the effect of polymor-
observed when the joint effect of and nasopharyngeal carcinoma also indi- phisms in the CYP2E1 gene is more pro-
CYP2E1 genotype and alcohol consump- rectly supports a role in nasopharyngeal nounced at low levels of exposure to ex-

1210 REPORTS Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997
ogenous factors. Finally, it is possible that trosodimethylamine and formation of adducts (20) Ingelman-Sundberg M, Johansson I, Yin H,
CYP2E1 might not be important for the with cellular DNA. Cancer Res 1992;52: Terelius Y, Eliasson E, Clot P, et al. Ethanol-
1796–800. inducible cytochrome P4502E1: genetic poly-
activation of tobacco-specific nitrosa-
(5) Koop DR. Oxidative and reductive metabolism morphism, regulation, and possible role in the
mines and that exposure to other procar- by cytochrome P450 2E1. FASEB J 1992;6: etiology of alcohol-induced liver disease. Al-
cinogens, including other nitrosamines, 724–30. cohol 1993;10:447–52.
that are activated primarily by CYP2E1 (6) Anderson LM. Modulation of nitrosamine me- (21) Hirvonen A, Husgafvel-Pursiainen K, Anttila
and that are associated with cigarette tabolism by ethanol: implications for cancer S, Karjalainen A, Vainio H. The human
smoking might explain why the CYP2E1 risk. In: Watson RR, editor. Alcohol and can- CYP2E1 gene and lung cancer: DraI and RsaI
cer. Boca Raton (FL): CRC Press, 1992:17–54. restriction fragment length polymorphisms in a
effect is limited to nonsmokers in our Finnish study population. Carcinogenesis
(7) Gervasi PG, Longo V, Naldi F, Panattoni G,
study. In fact, laboratory data suggest Ursino F. Xenobiotic-metabolizing enzymes in 1993;14:85–8.
that, in humans, liver metabolism of the human respiratory nasal mucosa. Biochem (22) Persson I, Johansson I, Bergling H, Dahl ML,
tobacco-specific nitrosamine NNN is Pharmacol 1991;41:177–84. Seidegard J, Rylander R, et al. Genetic poly-
catalyzed mainly by CYP3A4 and (8) Longo V, Pacifici GM, Panattoni G, Ursino F, morphism of cytochrome P4502E1 in a Swed-
CYP2A6, not CYP2E1, and epidemio- Gervasi PG. Metabolism of diethylnitrosamine ish population. Relationship to incidence of
by microsomes of human respiratory nasal mu- lung cancer. FEBS Lett 1993;3:207–11.

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

logic data suggest that dietary factors are cosa and liver. Biochem Pharmacol 1989;38: (23) Hixson JE, Vernier DT. Restriction isotyping
more important determinants of nasopha- 1867–9. of human apolipoprotein E by gene amplifica-
ryngeal carcinoma risk than cigarette (9) Smith TJ, Guo Z, Hong JY, Ning SM, Thomas tion and cleavage with HhaI. J Lipid Res 1990;
smoking (2,42). As such, perhaps dietary PE, Yang CS. Kinetics and enzyme involve- 31:545–8.
nitrosamines, nitrosamine precursors, ment in the metabolism of 4-(methylnitrosa- (24) Kleinbaum DG, Kupper LL, Morgenstern H.
mino)-1-(3-pyridyl)-1-butanone (NNK) in Epidemiologic research: principles and quan-
and/or other procarcinogens present in the
microsomes of rat lung and nasal mucosa. Car- titative methods. New York: Van Nostrand
diet might be important (43–47). The low cinogenesis 1992;13:1409–14. Reinhold, 1982:152–3.
frequency with which the subjects in our (10) Ding XX, Coon MJ. Induction of cytochrome (25) Schlesselman JJ. Case-control studies. Design,
study reported consumption of individual P-450 isozyme 3a (P-450IIE1) in rabbit olfac- conduct, and analysis. New York: Oxford Uni-
indicator food items in our questionnaire tory mucosa by ethanol and acetone. Drug versity Press, 1982:33-4; 227–90.
did not allow us to address this issue in Metab Dispos 1990;18:742–5. (26) Breslow NE, Day NE. Statistical methods in
(11) Porter TD, Khani SC, Coon MJ. Induction and cancer research. Volume I—The analysis of
the present report. However, efforts are
tissue-specific expression of rabbit cytochrome case-control studies. IARC Sci Publ 1980:
under way to create composite food P450IIE1 and IIE2 genes. Mol Pharmacol 5–338.
groups and nitrate, nitrite, and nitrosa- 1992;36:61–5. (27) Yu MC. Diet and nasopharyngeal carcinoma.
mine indices for subjects in our study us- (12) Hayashi SI, Watanabe J, Kawajiri K. Genetic FEMS Microbiol Immunol 1990;64:235–42.
ing food-frequency questionnaire data. polymorphism in the 58-flanking region change (28) Yu MC. Nasopharyngeal carcinoma: epidemi-
transcriptional regulation of the human cyto- ology and dietary factors. IARC Sci Publ
In summary, our results demonstrate
chrome P450IIE1 gene. J Biochem 1991;110: 1991;(105):39–47.
an association between CYP2E1 genetic 559–65. (29) Ning JP, Yu MC, Wang QS, Henderson BE.
polymorphism and nasopharyngeal carci- (13) Watanabe J, Hayashi SI, Kawajiri K. Different Consumption of salted fish and other risk fac-
noma risk. The association was restricted regulation and expression of the human tors for nasopharyngeal carcinoma (NPC) in
to never smokers and was strongest CYP2E1 gene due to the RsaI polymorphism in Tianjin, a low-risk region for NPC in the Peo-
the 58-flanking region. J Biochem 1994;116: ple’s Republic of China. J Natl Cancer Inst
among never smokers who also reported
321–6. 1990;82:291–6.
no passive smoking exposure in their (14) Tsutsumi M, Wang JS, Takada A. Hepatic (30) Lin TM, Chen KP, Lin CC, Hsu MM, Tu SM,
household. Independent confirmation of messenger RNA contents of cytochrome Chiang TC, et al. Retrospective study on naso-
this finding is required, and additional ex- P4502E1 in patients with different P4502E1 pharyngeal carcinoma. J Natl Cancer Inst
amination of the joint effect of CYP2E1 genotypes. Int Hepatol Commun 1994;2:135–8. 1973;51:1403–8.
(15) Tsutsumi M, Wang JS, Takase S, Takada A. (31) West S, Hildesheim A, Dosemeci M. Non-viral
genotype and other non-tobacco-related
Hepatic P4502E1 expression in human livers risk factor for nasopharyngeal carcinoma in
exposures is needed before more conclu- with different CYP2E1 genotypes. Hepatology The Philippines: results from a case-control
sive interpretation of our results can be 1994;20(4, Part 2):311A. study. Int J Cancer 1993;55:722–7.
made. (16) Carriere V, Berthou F, Baird S, Belloc C, (32) Mabuchi K, Bross DS, Kessler II. Cigarette
Beaune P, de Waziers I. Human cytochrome smoking and nasopharyngeal carcinoma. Can-
References P450 2E1 (CYP2E1): from genotype to pheno- cer 1985;55:2874–6.
type. Pharmacogenetics 1996;6:203–11. (33) Zhu K, Levine RS, Brann EA, Gnepp DR,
(1) Parkin DM, Muir CS, Whelan SL, Gao YT, (17) Dai Y, Cederbaum AI. Cytotoxicity of acet- Baum MK. A population-based case-control
Ferlay J, Powell J. Cancer Incidence in Five aminophen in human cytochrome P4502E1- study of the relationship between cigarette
Continents. IARC Sci Publ 1992;VI. transfected HepG2 cells. J Pharmacol Exp Ther smoking and nasopharyngeal cancer (United
(2) Hildesheim A, Levine PH. Etiology of naso- 1995;273:1497–505. States). Cancer Causes Control 1995;6:507–12.
pharyngeal carcinoma: a review. Epidemiol (18) Kato S, Shields PG, Caporaso NE, Hoover RN, (34) Chen CJ, Liang KY, Chang YS, Wang YF,
Rev 1993;15:466–85. Trump BF, Sugimura H, et al. Cytochrome Hsieh T, Hsu MM, et al. Multiple risk factors
(3) Hildesheim A, Chen CJ, Caporaso NE, Cheng P450IIE1 genetic polymorphisms, racial varia- of nasopharyngeal carcinoma: Epstein-Barr vi-
YJ, Hoover RN, Hsu MM, et al. Cytochrome tion, and lung cancer risk. Cancer Res 1992; rus, malarial infection, cigarette smoking, and
P4502E1 genetic polymorphisms and risk of 52:6712–5. familial tendency. Anticancer Res 1990;10:
nasopharyngeal carcinoma: results from a case- (19) Uematsu F, Kikuchi H, Motomiya M, Abe T, 547–53.
control study conducted in Taiwan. Cancer Sagami I, Ohmachi T, et al. Association be- (35) Nam JM, McLaughlin JK, Blot WJ. Cigarette
Epidemiol Biomarkers Prev 1995;4:607–10. tween restriction fragment length polymor- smoking, alcohol, and nasopharyngeal carci-
(4) Nouso K, Thorgeirsson SS, Battula N. Stable phism of the human cytochrome P450IIE1 noma: a case-control study among U.S. whites.
expression of human cytochrome P450IIE1 in gene and susceptibility to lung cancer. Jpn J J Natl Cancer Inst 1992;84:619–22.
mammalian cells: metabolic activation of ni- Cancer Res 1991;82:254–6. (36) Chow WH, McLaughlin JK, Hrubec Z, Nam

Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997 REPORTS 1211
 JM, Blot WJ. Tobacco use and nasopharyngeal 0.41; 95% CI = 0.20–0.60; P<.0001).
carcinoma in a cohort of US veterans. Int J Conclusions: Our results are consistent
Cancer 1993;55:538–40.
(37) Yi Z, Ohshima H, Bouvier G, Roy P, Zhong J, Correlation of Vascular with the idea that progression of cervi-
Li B, et al. Urinary excretion of nitrosamino Endothelial Growth Factor cal dysplasia is dependent on angio-
acids and nitrate by inhabitants of high- and genesis. [J Natl Cancer Inst 1997;89:
Expression and Microvessel
low-risk areas for nasopharyngeal carcinoma in 1212–7]
southern China. Cancer Epidemiol Biomarkers Density in Cervical
Prev 1993;2:195–200. Intraepithelial Neoplasia
(38) Yu MC, Nichols PW, Zou XN, Estes J, Hen- The role of angiogenesis in tumor
derson BE. Induction of malignant nasal cavity growth is well defined for a variety of
tumors in Wistar rats fed Chinese salted fish. Andreas Obermair, Dagmar solid malignant tumors (1). Once a tumor
Br J Cancer 1989;60:198–201. Bancher-Todesca, Selcuk Bilgi, has become established, a prevascular
(39) Huang DP, Ho JH, Saw D, Teoh TB. Carci-
noma of the nasal and paranasal regions in rats Alexandra Kaider, Petra phase of uncertain length, during which
fed Cantonese salted marine fish. IARC Sci Kohlberger, Silvia Müllauer-Ertl, the tumor is dormant, is followed by a
Publ 1978;(20):315–28. phase of intense endothelial cell prolifera-
Sepp Leodolter, Gerald Gitsch*

Downloaded from https://academic.oup.com/jnci/article-abstract/89/16/1207/942771 by guest on 11 July 2019

(40) Raucy JL, Kraner JC, Lasker JM. Bioactiva- tion and subsequent tumor growth (2).
tion of halogenated hydrocarbons by cyto- The angiogenic response is mediated by
chrome P4502E1. Crit Rev Toxicol 1993;23:
diffusible signals, so-called angiogenic
1–20. Background: Angiogenesis (the forma-
(41) Lee CK, Fulp C, Bombick BR, Doolittle DJ. factors, which can be secreted by malig-
tion of new blood vessels) appears to be
Inhibition of mutagenicity of N-nitrosamines nant tumor cells (2). Observations in vivo
required for the growth of invasive tu-
by tobacco smoke and its constituents. Mutat suggest that the inhibition of blood vessel
Res 1996;367:83–92. mors, but little information exists about
formation results in a reduction in tumor
(42) Patten CJ, Smith TJ, Tynes RW, Freseu M, Lee its role in the development of preinva-
size and a reduction in metastasis (3–5).
T, Yang CS, et al. Evidence for cytochrome sive lesions. We examined microvessel
P450 2A6 and 3A4 as catalysts for N- The vascular endothelial growth factor
density and expression of vascular en-
nitrosonornicotine alpha hydroxylation in hu- (VEGF), which is also known as the vas-
dothelial growth factor in specimens of
man liver microsomes. Carcinogenesis. In cular permeability factor (VPF), is a mul-
press. cervical intraepithelial neoplasia (CIN),
tifunctional cytokine that acts on endothe-
(43) Poirier S, Bouvier G, Malaveille C, Ohshima a preinvasive lesion of the uterine cer-
lial cells as a highly specific mitogen
H, Shao YM, Hubert A, et al. Volatile nitrosa- vix, to determine whether a connection
mine levels and genotoxicity of food samples (6,7), binding to specific class III receptor
could be established between these pa-
from high-risk areas for nasopharyngeal carci- tyrosine kinases (flt-1 and KDR) and
rameters of angiogenesis and the grade
noma before and after nitrosation. Int J Cancer opening calcium channels to increase in-
1989;44:1088–94. of dysplasia (i.e., tissue abnormality).
tracellular calcium levels (8). VEGF also
(44) Poirier S, Ohshima H, de The G, Hubert A, Methods: Sections of biopsy specimens
stimulates angiogenesis by increasing
Bourgade MC, Bartsch H. Volatile nitrosamine from 83 patients with grades I–III CIN
levels in common food from Tunisia, south vascular permeability (7,9). Four VEGF
were examined retrospectively. Mi-
China and Greenland, high-risk areas for naso- isoforms (polypeptides containing 206,
crovessels were localized by use of a
pharyngeal carcinoma (NPC). Int J Cancer 189, 165, and 121 amino acids) have been
1987;39:293–6. polyclonal antibody directed against
isolated, but they apparently express iden-
(45) Shao YM, Poirier S, Ohshima H, Malaveille C, factor VIII-related antigen; vascular
tical biologic activities (10).
Zeng Y, de The G, et al. Epstein-Barr virus endothelial growth factor was detected
activation in Raji cells by extracts of preserved Although extensive data are available
by means of a monoclonal antibody.
food from high risk areas for nasopharyngeal on the role of angiogenesis in the devel-
Reported P values are two-sided. Re-
carcinoma. Carcinogenesis 1988;9:1455–7. opment of a variety of invasive malignant
(46) Bouvier G, Hergenhahn M, Polack A, sults: Highest microvessel densities and
tumors, little information exists concern-
Bornkamm GW, de The G, Bartsch H. Char- highest expression of vascular endothe-
ing preinvasive lesions. Cervical intraep-
acterization of macromolecular lignins as Ep- lial growth factor were found in a nar-
stein-Barr virus inducer in foodstuff associ- ithelial neoplasia (CIN) is a preinvasive
row border region between CIN lesions
ated with nasopharyngeal carcinoma risk. lesion of the uterine cervix that is charac-
Carcinogenesis 1995;16:1879–85.
and the underlying stroma. A signifi-
(47) Chen W, Weisburger JH, Fiala ES, Spratt TE, cant correlation was observed between
Carmella SG, Chen D, et al. Gastric carcino- high vascular endothelial growth factor
genesis: 2-chloro-4-methylthiobutanoic acid, a expression and high microvessel den- *Affiliations of authors: A. Obermair, D.
novel mutagen in salted, pickled Sanma hiraki sity (Kendall’s t = 0.27; 95% confi- Bancher-Todesca, P. Kohlberger, S. Leodolter, G.
fish, or similarly treated methionine. Chem Res Gitsch (Department of Gynecology and Obstetrics),
Toxicol 1996;9:58–66.
dence interval [CI] = 0.03–0.50; P = S. Bilgi, S. Müllauer-Ertl (Department of Pathology,
.018). Mean microvessel density values Gynecopathology Unit), A. Kaider (Department of
± standard deviations for CIN I, CIN Medical Computer Sciences/Biometrics), University
Notes II, and CIN III lesions were 19.4 ± 5.8, Hospital of Vienna, Austria.
Correspondence to: Andreas Obermair, M. D.,
21.9 ± 7.0, and 34.1 ± 14.8, respectively
Department of Gynecology and Obstetrics, Univer-
Present address: J. R. Idle, Institute of Cancer (Kendall’s t = 0.46; 95% CI = 0.30– sity Hospital of Vienna, Währinger Gürtel 18-20,
Research and Molecular Biology, Norwegian Uni-
0.61; P<.0001). Corresponding values A-1090 Vienna, Austria. E-mail: andreas.
versity of Science and Technology, Trondheim,
Norway. for vascular endothelial growth factor obermair@akh-wien.ac.at
expression were 8.3 ± 3.5, 8.4 ± 2.0, and See ‘‘Notes’’ following ‘‘References.’’
Manuscript received January 13, 1997; revised
May 30, 1997; accepted June 13, 1997. 12.2 ± 3.6, respectively (Kendall’s t = © Oxford University Press

1212 REPORTS Journal of the National Cancer Institute, Vol. 89, No. 16, August 20, 1997