| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Division of Cancer Epidemiology and Genetics [A. J. D., N. R., P. D. I., M. S. L.] and Neuro-oncology Branch [H. A. F.], National Cancer Institute, Bethesda, Maryland 20892-7240; St. Joseph’s Hospital and Medical Center, Phoenix, Arizona [W. R. S.]; Western Pennsylvania Hospital, Pittsburgh, Pennsylvania [R. G. S.]; Brigham and Women’s Hospital, Boston, Massachusetts [P. M. B.]; and Laboratory of Computational Biology and Risk Analysis, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina [G. S. P., D. A. B.]
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
GST2 and CYP genes encode enzymes involved in the activation and detoxification of a wide variety of chemicals. GST genes, including mu (M), pi (P), and theta (T) GSTs, produce enzymes that catalyze reduced glutathione-dependent reactions with compounds containing an electrophilic center (10) . The range of potential GST substrates is very large, including occupational and environmental carcinogens such as solvents, pesticides, and polycyclic aromatic hydrocarbons. Polymorphisms in GSTM1 and GSTT1 result in absence of a functional gene product (11 , 12) . Two polymorphisms in GSTP1 have been discovered (I105V, A114V) for which effects on function are not known (13) , although there is some indication of decreased enzyme activity among individuals with genotypes containing the 105 valine allele (14) . Another gene hypothesized to play a role in human brain tumors is CYP2E1, which metabolizes and activates solvents that also act to induce its expression, including benzene, styrene, carbon tetrachloride, ethylene glycol, and ethanol (15) . The frequencies of CYP2E1 variant sequences, including RsaI and Ins96, differ considerably between ethnic groups (15 , 16) .
Persons with variant alleles for GST and CYP2E1 genes may differ in their ability to metabolize carcinogenic compounds and, thus, may have an altered risk of cancer. Associations between GSTT1 null genotype and increased incidence of meningioma (n = 50) and glioma subtype astrocytoma (n = 112) were observed in a hospital-based case-control study (17) . A second case-control study reported no association between GSTT1 and glioma (n = 118), although an association with the glioma subtype oligodendroglioma (n = 16) was observed (18) . There was no difference in the distribution of GSTT1 genotypes between glioma cases (n = 90) and controls in a third study (19) . All three studies reported no overall association between GSTM1 null genotype and glioma (17 , 19 , 20) . Other tumor types, such as acoustic neuroma, have not been extensively studied, nor have associations between brain tumors and GSTP1 or CYP2E1 variants. To evaluate relationships between adult brain tumor incidence and genes involved in the metabolism of major categories of chemicals previously associated with the risk of brain tumors in adults, we examined the effects of polymorphisms in GST and CYP2E1 genes in a side-by-side comparison of three major categories of malignant and benign brain tumors, namely the gliomas, meningiomas, and acoustic neuromas.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Controls were patients admitted to the same hospitals and treated for a variety of nonneoplastic conditions. They were frequency-matched to the total case series by hospital, age, sex, race, and proximity of residence to hospital. Of the eligible controls identified and asked to participate, 799 control subjects were recruited, representing 86% of those contacted. Discharge diagnoses of the control subjects were trauma, injury, or poisoning (24.7%), circulatory disease (22.4%), musculoskeletal disease (21.5%), disease of the digestive system (11.5%), and other (19.9%).
Trained nurses administered a structured, computerized questionnaire that included detailed questions on: lifetime job history; specific occupational exposures, processes, and tasks; cellular telephones and other forms of communication devices; exposure to diagnostic and therapeutic radiation and other aspects of medical history; reproductive characteristics and history; use of hair dyes; and family history of cancer and selected other conditions.
Laboratory Analyses.
DNA was extracted from peripheral WBCs (buffy coat or granulocytes) from blood samples collected from 1277 subjects [80% of all subjects; 422 gliomas (86%), 172 meningiomas (87%), 79 acoustic neuromas (86%), and 604 controls (76%)], by GenoType, Ltd. in the United Kingdom, using a phenol-chloroform method as described by Daly et al. (22)
. GSTT1 genotyping was conducted by GenoType, Ltd. Genotyping assays for GSTM1, GSTP1, and CYP2E1 polymorphisms were conducted by the NIEHS.
GenoType, Ltd. determined GSTT1 genotype using an allele-specific PCR-based method described previously (11) . At the NIEHS, genotyping was performed using 50 ng of genomic DNA and PCR-based methods. For GSTM1, an allele-specific PCR method was used (23) . Analysis for GSTP1 I105V variant genotypes used a restriction fragment length polymorphism-PCR method (14) . To detect the GSTP1 A114V single nucleotide polymorphism, the NIEHS laboratory used a melting temperature-shift genotyping method using two allele-specific forward primers of different lengths and melting temperatures and a common reverse primer (24) . The NIEHS laboratory developed a novel PCR multiplex method for the CYP2E1 RsaI (restriction fragment length polymorphism-PCR) and Ins96 (allele specific-PCR) genotypes using published sequence information (16 , 25) .
Quality control measures included the addition of replicates [68 samples from 3 individuals who were not study subjects (QC-A, n = 34; QC-B, n = 19; QC-C, n = 15), collected and processed in identical fashion as study subjects] interspersed throughout the batches for all six genotyping assays and duplicates (two samples for each of 92 individuals who were study subjects) interspersed throughout the batches for all assays, except GSTT1.
Statistical Analyses.
SAS software versions 6.12 and 8.0 (SAS Institute Inc., Cary, NC), and Epicure for Windows (1998; Hirosoft International Corporation, Seattle, WA) were used for statistical analyses. We computed Hardy-Weinberg equilibrium for GSTP1 and CYP2E1 genotypes among the control group, to determine whether the distribution of alleles was as expected (the GSTM1 and -T1 genotypes were coded as wild type or null, making direct calculation of Hardy-Weinberg equilibrium impossible).
The effect of each gene variant on the incidence of each brain tumor type, using the nonvariant genotype as the referent, was estimated by conventional maximum likelihood using unconditional logistic regression to calculate ORs and 95% CLs. All effect estimates for gene variants were adjusted for the study matching factors of age (coded in years: 18–29; 30–39 as the referent; 40–49; 50–59; 60–69; 70–79; 80–99), race (non-Hispanic white as the referent; Hispanic white; black; other), sex (male; female as the referent), hospital (Phoenix, AZ, as the referent; Boston, MA; Pittsburgh, PA), and proximity of patient’s residence to the hospital (coded in miles: 0–4 as the referent; 5–14; 15–29; 30–49; 50 or more). We checked the influence of the control series composition on results by examining the main effect of each genotype on the three tumor types, while excluding one major category of control discharge diagnoses at a time.
In addition to models in which each of the variant genotypes was treated separately, one model for each tumor type analyzed all six genotypes simultaneously, using penalized quasi-likelihood hierarchical regression modeling (26
, 27)
in SAS/GLIMMIX (28)
. We chose to use hierarchical regression modeling because simulation studies have indicated that estimates from this approach are generally more accurate and stable than those calculated using conventional likelihood methods, especially when considering multiple exposures and sparse data (26)
. Because previous results on GSTT1 and GSTM1 genotypes and brain tumors were few and were conflicting for GSTT1, we assumed that we did not have prior knowledge indicating any one variant genotype as more likely to be associated with brain tumors than any other genotype. Therefore, the true effects for the six gene variants were assumed to be exchangeable, random effects, arising from a common prior normal distribution with an unknown mean and a variance of 0.35; thus assuming, with 95% certainty, that the true rate ratio for each gene variant would fall within a 10-fold range (note: for a 10-fold range, residual variance = (1n(10))/3.92)2 
0.35) (26)
. These values for the prior distribution were chosen to insure that the expected value of the betas for the gene variants would be the mean of all the betas, and that large magnitude values for effects of the gene variants would rarely occur, consistent with a prior assumption of probably modest effects of gene variants in the context of background exposures.
We examined each gene variant-disease association separately for three age groups (
40 years, 40–60 years, >60 years), and for each sex. Other factors of interest, such as race or family history of nervous system tumors, did not have sufficient numbers in subgroups to allow stratified analyses. We also examined the association of each gene variant with high- and low-grade tumors, and with specific glioma subtypes including glioblastoma, anaplastic astrocytoma, other astrocytoma, oligodendroglioma, and mixed oligoastrocytoma. 
2 statistics and corresponding Ps (based on 4 degrees of freedom) were calculated to test whether the distribution of each gene variant differed between the five glioma subtypes.
We evaluated associations of brain tumor incidence with several combinations of GSTP1 variants. To evaluate the risk associated with increasing numbers of variant GSTP1 valine alleles, we estimated the effects of GSTP1 I105V heterozygous (Ile/Val) and homozygous (Val/Val) variant genotypes, compared with the wild type (Ile/Ile). Because there are demonstrated differences in structure and stability of GSTP1 proteins expressed from alleles with different combined I104V and A114V variants (13) , we estimated effects of GSTP1 alleles with different variant combinations (GSTP1*A, wild type for both; GSTP1*B, 105 Val/Val and 114 Ala/Ala; GSTP1*C, 105 Val/Val and 114 Ala/Val or Val/Val). Trend tests were conducted where a monotonic trend by increasing number of variant alleles was observed, by calculating a P for the beta coefficient in a logistic regression model with the exposure coded as an ordinal categorical variable.
Data were analyzed for potential interactions of gene variants that were associated with any of the brain tumor types in our study (namely, CYP2E1 RsaI, GSTP1 I105V, and GSTT1 null). Individual and joint effects of each gene variant combination were estimated to assess potential interaction, and likelihood ratio tests were used to test gene-gene interactions on the multiplicative and additive scales using Epicure software. Because estimates of individual and joint effects from logistic regression models were very imprecise due to small numbers of subjects with combined variant genotypes, we also used hierarchical regression models to shrink unstable estimates toward a common mean. These models treated the parameters for the joint exposure and two independent exposures as arising from the same prior distribution with an unknown mean and a variance of 0.35.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
. Cases, on average, were older and more highly educated than controls. Meningioma cases were more often female compared with controls or the other tumor types. Genotyping was successfully conducted for GSTM1 null (97.5% of all samples analyzed), GSTP1 I105V (96.5%), GSTP1 A114V (97.9%), GSTT1 null (90.8%), CYP2E1 RsaI (97.3%), and CYP2E1 Ins96 (97.8%), and genotyping of all six variants was successful for 89% of the samples analyzed for all six genotypes. Missing values, primarily the result of insufficient quantity of DNA or poor amplification for a specific locus or overall, were equally likely to be from case or control samples. We achieved 99–100% agreement between duplicate samples, and among replicates for GSTM1, GSTP1, and CYP2E1 assays, and 95% agreement among replicate samples analyzed for GSTT1.
|
were GSTM1 null (55.8%), GSTP1 105 Val/Val (10.4%), GSTP1 114 Ala/Val and Val/Val (13.5%), GSTT1 null (18.4%), CYP2E1 RsaI CYP2E1*1A/CYP2E1*5 and CYP2E1*5/CYP2E1*5 (5.8%), and CYP2E1 Ins96 CYP2E1*1C/CYP2E1*1D and CYP2E1*1D/CYP2E1*1D (7.0%), similar to published values (10
, 15
, 16
, 29)
. There was no evidence of departure from Hardy-Weinberg equilibrium for the GSTP1 or CYP2E1 genotypes. There was significant linkage disequilibrium between the GSTP1 I105V and A114V genotypes among controls (2 = 159.2, P = 0.001), largely caused by the absence of the combined 105 (Ile/Ile)/114 (Ala/Val or Val/Val) genotypes. GSTP1 I105V and CYP2E1 Ins96 genotypes were also statistically associated among controls (2 = 5.0, P = 0.02).
|
. The GSTP1 105 Val/Val genotype was associated with an 80% increased glioma incidence. Meningioma was not strongly associated with any of the genotypes examined, but there was a weak association with GSTT1 (Table 3
; OR, 1.5; 95% CL, 1.0, 2.3). The CYP2E1 RsaI variant was weakly associated with increased incidence of all three tumor types, whereas CYP2E1 Ins96 showed inverse associations; however, the small numbers of subjects with CYP2E1 variant genotypes made estimates imprecise. There was no association of the GSTM1 or GSTP1 I114V variants with the risk of any tumor type. The results of conventional maximum likelihood estimation including each gene variant in a separate model were generally similar to those calculated using hierarchical regression models adjusting for all genotypes simultaneously, indicating little confounding between different genotypes. Differences in results between the two modeling strategies occurred where data were sparse; for example, the association between CYP2E1 RsaI and acoustic neuroma appeared moderate in logistic regression modeling (OR, 2.3), but was shrunk considerably in hierarchical modeling (OR, 1.3), indicating a potentially weak effect. Analyses to check the sensitivity of our results to the control series composition did not indicate any major bias resulting from inclusion of any of the control discharge diagnoses. One exception was noted for the association between meningioma and GSTT1 null genotype, for which the OR dropped from 1.5 to 1.2 when subjects with circulatory disease were excluded from the control group.
|
. Although the number of subjects in each age group did not support a formal assessment of heterogeneity of effect, the positive association of the GSTP1 105 Val/Val genotype with glioma decreased progressively, albeit modestly, across increasing age groups (age, 40 years; OR, 2.2; age, >60; OR, 1.3), as did associations for GSTT1 null with meningioma (age, 40 years; OR, 2.1; age, >60; OR, 1.4). Whereas GSTP1 105 Val/Val was not strongly associated with acoustic neuroma overall incidence, there was an almost 3-fold increased risk among younger subjects (age, <50 years; OR, 2.9; 95% CL, 1.1, 7.8; results not shown in Table 4
). Associations between the CYP2E1 RsaI variant with all three tumor types followed the same pattern, with the strongest associations among those less than age 40. Similarly, inverse associations of the CYP2E1 Ins96 variant were only observed among those aged 60 years or younger. There were no meaningful differences in associations between gene variants and brain tumors by sex (results not shown).
|
. 2 tests indicated that the distribution of some genotypes differed between glioma subtypes, namely, GSTM1, GSTP1 A114V, and CYP2E1 RsaI. Some differences were observed for associations of several gene variants with oligodendroglioma, compared with the other subtypes, including an inverse association with GSTM1 null, no association with GSTP1 105 Val/Val, and a positive association with GSTP1 114 Ala/Val or Val/Val. No apparent confounding factor accounted for these findings. There were no important differences between associations of gene variants with high- versus low-grade tumors (results not shown).
|
. However, there was little evidence that the combined effect of both GSTP1 I105V and A114V variants on the risk of any brain tumor type differed from that for the GSTP1 I105V variant alone (Table 6)
.
|
. Even in hierarchical regression analyses in which estimates were shrunk toward a common prior mean, the joint associations were greater than multiplicative. Other combinations of GSTT1, GSTP1 I105V, and CYP2E1 RsaI genotypes demonstrated no suggestion of interaction.
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Previously reported associations between GSTT1 null genotype and the risk of overall glioma incidence (17) were not observed in this study population, nor in an earlier study (19) . In our analyses of glioma subtypes, a positive association was observed with oligodendroglioma, similar to a previous report (18) . We identified some problems with the GSTT1 null genotyping assay conducted at Genotype, Ltd., including failure of about 9% of samples to amplify. The frequency of GSTT1 null genotype among our controls was very similar to other populations (10) , providing some indication that failure to amplify was not associated with GSTT1 null genotype. To fill in some missing data on GSTT1 genotype, the samples that did not amplify were sent to additional laboratories (NIEHS and Thetagen, Inc.) for further attempts. However, statistical analyses based on GSTT1 null genotyping results only from Genotype, Ltd. did not differ meaningfully from the more full dataset from the three laboratories combined. Another concern relates to the somewhat lower reproducibility of the GSTT1 assay compared with other genotyping assays conducted in this study. To explore the possibility that a positive association was masked by nondifferential misclassification of the genotype (30) , we conducted simple sensitivity analyses (31) assuming imperfect sensitivity (e.g., 94%) and specificity (e.g., 97%) of the genotyping assay (sensitivity and specificity were estimated from the results of replicate samples). Under these assumptions, the most likely "corrected" OR for GSTT1 null and glioma would still be negligible (corrected OR, 1.1). However, even a small amount of misclassification could have importantly affected results for glioma subtypes or brain tumor types in which GSTT1 null genotype was more prevalent; for example, such sensitivity analyses indicated that the most likely corrected OR for oligodendroglioma would be further elevated (observed OR, 1.5; corrected OR, 1.8).
The observed positive association in our study of the GSTP1 I105V variant and glioma incidence has not been reported previously. GSTP1 is thought to be the most strongly expressed of the GST isoenzymes in the human brain (32 , 33) , with increased expression in tumors (13 , 33 , 34) . We observed a moderate association between glioma incidence and the I105V variant that followed a trend of increasing magnitude by number of variant alleles. Immunohistochemical screening has shown that expression of GSTP1 in the adult brain is high in astrocytes and consistently absent in oligodendrocytes (32) ; these observations would support our results in which the GSTP1 I105V variant was positively associated with glioblastoma, astrocytoma, and mixed oligoastrocytoma incidence, but not with oligodendroglioma. Analysis of GSTP1 proteins has shown that three active, structurally different encoded proteins are expressed from three possible combinations of wild-type and variant GSTP1 I105V or A114V alleles, resulting in functional differences in stability and half-life (13 , 35) . It is reasonable to hypothesize that such functional differences between GSTP1 proteins may result in differing capabilities for metabolizing carcinogens. Furthermore, combined GSTP1 I105V/A114V variants have been observed to be 4-fold more prevalent in malignant gliomas than in normal brain tissues (13) . These GSTP1 variants were in linkage disequilibrium in our study population, similar to previous reports (29) , and the A114V variant did not occur independently of the I105V variant. However, no association between brain tumors and the GSTP1 A114V variant genotype were observed in these data, nor was there evidence that any effect of combined GSTP1 I105V/A114V variants differed from the effect of the I105V variant alone.
CYP2E1 RsaI and CYP2E1 Ins96 polymorphisms may result in differing functional activity that could potentially explain the divergent results we observed. In oral clearance studies of chlorzoxazone, the RsaI variant has been associated with decreased metabolic activity (36) , whereas Ins96 variant was associated with increased activity (37) . However, increased activity of the Ins96 variant has primarily been observed among the obese or those who have recently consumed alcohol (37) . Further research into the expression of these two variant CYP2E1 genotypes in the human brain and among different population subgroups may clarify interpretation of the associations we observed in our study.
Where genotypes were associated with brain tumor incidence, we observed stronger associations among younger age groups. Because the OR is a relative measure, these differences are due, in part, to the higher incidence of brain tumors at older rather than younger adult ages. Thus, even when the absolute magnitude of effect is similar in both age groups, the relative measure can appear stronger in the stratum with lower incidence. For comparison of an absolute measure of effect across age strata, we used incidence rates from the Surveillance, Epidemiology, and End Results program of the National Cancer Institute (http://seer.cancer.gov/) to estimate the rate difference associated with the gene variant in each age category, using the formula RD = (RR-1)*R0 [where RD = rate difference, RR = relative rate as estimated by the OR, R0 = underlying incidence in the unexposed population (from Surveillance, Epidemiology, and End Results program data for brain and other nervous system tumors, all races, 1994–1998)]. Estimation of the absolute effect of gene variants on glioma incidence indicates that differences in effect by age are supported for the CYP2E1 RsaI variant, but not for the GSTP1 I105V variant for which the estimated absolute effect was actually higher for the older age group than for the younger age group (results not shown).
It is possible that selection bias could have produced spurious associations in our hospital-based case-control study, because controls were selected from patients with differing discharge diagnoses. One or more of the gene variants evaluated could be associated with one of the diseases constituting the control series, thereby creating a control group that is not representative of the general source population. Some reassurance was provided by the similarity of genotype frequencies of our controls with those described in the literature. GSTT1 null genotype has been associated with an increased risk of coronary heart disease in a recent study (38) . If this relationship were true, then inclusion of circulatory disease patients in the control group could be masking any potentially real associations between brain tumors and GSTT1; however, removal of circulatory disease patients did not change most ORs and changed the OR for meningioma in the opposite direction than would be expected. Because there is not a clear association between GSTT1 and circulatory disease, we can only speculate on the possibility that selection bias affected our results for GSTT1, although it seems unlikely based on these results.
Potential effects of GST and CYP2E1 gene variants on risk of three major categories of adult brain tumors as shown in our data merit further investigation. Replication of these analyses in other study populations will help to resolve the possibility of chance findings. If these findings are confirmed, then pooling of multiple epidemiological studies will provide the statistical power necessary to examine interactions of genotypes with specific occupational and environmental exposures.
|
Footnotes |
|---|
1 To whom requests for reprints should be addressed, at National Cancer Institute, 6120 Executive Boulevard, EPS 8109, Bethesda, MD 20892-7240. Phone: (301) 594-7901; Fax: (301) 402-1819; E-mail: deroosa{at}mail.nih.gov 
2 The abbreviations used are: GST, glutathione S-transferase; CYP, cytochrome P450; NIEHS, National Institute of Environmental Health Sciences; OR, odds ratio; CL, confidence limit.
Received 11/30/01; revised 10/ 9/02; accepted 10/29/02.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. E. Liu, P.-Y. Hsiao, C.-C. Lee, J.-A. Lee, and H.-Y. Chen NAT2*7 Allele Is a Potential Risk Factor for Adult Brain Tumors in Taiwanese Population Cancer Epidemiol. Biomarkers Prev., March 1, 2008; 17(3): 661 - 665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A. Schwartzbaum, A. Ahlbom, S. Lonn, M. Warholm, A. Rannug, A. Auvinen, H. C. Christensen, R. Henriksson, C. Johansen, C. Lindholm, et al. An International Case-Control Study of Glutathione Transferase and Functionally Related Polymorphisms and Risk of Primary Adult Brain Tumors Cancer Epidemiol. Biomarkers Prev., March 1, 2007; 16(3): 559 - 565. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Semmler, M. Simon, S. Moskau, and M. Linnebank The Methionine Synthase Polymorphism c.2756A>G Alters Susceptibility to Glioblastoma Multiforme. Cancer Epidemiol. Biomarkers Prev., November 1, 2006; 15(11): 2314 - 2316. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. J. De Roos, N. Rothman, M. Brown, D. A. Bell, G. S. Pittman, W. R. Shapiro, R. G. Selker, H. A. Fine, P. M. Black, and P. D. Inskip Variation in genes relevant to aromatic hydrocarbon metabolism and the risk of adult brain tumors. Neuro-oncol, April 1, 2006; 8(2): 145 - 155. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W J Lee, J S Colt, E F Heineman, R McComb, D D Weisenburger, W Lijinsky, and M H Ward Agricultural pesticide use and risk of glioma in Nebraska, United States Occup. Environ. Med., November 1, 2005; 62(11): 786 - 792. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Malmer, M. Feychting, S. Lonn, A. Ahlbom, and R. Henriksson p53 Genotypes and Risk of Glioma and Meningioma Cancer Epidemiol. Biomarkers Prev., September 1, 2005; 14(9): 2220 - 2223. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Tang, W. Shao, M. T. Dorak, Y. Li, R. Miike, E. Lobashevsky, J. K. Wiencke, M. Wrensch, R. A. Kaslow, and C. S. Cobbs Positive and Negative Associations of Human Leukocyte Antigen Variants with the Onset and Prognosis of Adult Glioblastoma Multiforme Cancer Epidemiol. Biomarkers Prev., August 1, 2005; 14(8): 2040 - 2044. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Lai, L. Crevier, and L. Thabane Genetic Polymorphisms of Glutathione S-Transferases and the Risk of Adult Brain Tumors: A Meta-analysis Cancer Epidemiol. Biomarkers Prev., July 1, 2005; 14(7): 1784 - 1790. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Sadetzki, P. Flint-Richter, S. Starinsky, I. Novikov, Y. Lerman, B. Goldman, and E. Friedman Genotyping of Patients with Sporadic and Radiation-Associated Meningiomas Cancer Epidemiol. Biomarkers Prev., April 1, 2005; 14(4): 969 - 976. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. Hung, P. Brennan, C. Malaveille, S. Porru, F. Donato, P. Boffetta, and J. S. Witte Using Hierarchical Modeling in Genetic Association Studies with Multiple Markers: Application to a Case-Control Study of Bladder Cancer Cancer Epidemiol. Biomarkers Prev., June 1, 2004; 13(6): 1013 - 1021. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wrensch, K. T. Kelsey, M. Liu, R. Miike, M. Moghadassi, K. Aldape, A. McMillan, and J. K. Wiencke Glutathione-S-Transferase Variants and Adult Glioma Cancer Epidemiol. Biomarkers Prev., March 1, 2004; 13(3): 461 - 467. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Cell Growth & Differentiation |
