Search tips
Search criteria 


Logo of carcinLink to Publisher's site
Carcinogenesis. 2010 April; 31(4): 614–620.
Published online 2010 January 4. doi:  10.1093/carcin/bgp321
PMCID: PMC2847086

Variants in blood pressure genes and the risk of renal cell carcinoma


Hypertension is a known risk factor for renal cell carcinoma (RCC), although the underlying biological mechanisms of its action are unknown. To clarify the role of hypertension in RCC, we examined the risk of RCC in relation to 142 single-nucleotide polymorphisms (SNPs) in eight genes having a role in blood pressure control. We analyzed 777 incident and histologically confirmed RCC cases and 1035 controls who completed an in-person interview as part of a multi-center, hospital-based case–control study in Central Europe. Genotyping was conducted with an Illumina® GoldenGate® Oligo Pool All assay using germ line DNA. Of the eight genes examined, AGT (angiotensinogen) was most strongly associated with RCC (minimum P-value permutation test = 0.02). Of the 17 AGT tagging SNPs considered, associations were strongest for rs1326889 [odds ratio (OR) = 1.35, 95% confidence interval (CI) = 1.15–1.58] and rs2493137 (OR = 1.31, 95% CI = 1.12–1.54), which are located in the promoter. Stratified analysis revealed that the effects of the AGT SNPs were statistically significant in participants with hypertension or high body mass index (BMI) (≥25 kg/m2), but not in subjects without hypertension and with a normal BMI (<25 kg/m2). Also, haplotypes with risk-conferring alleles of markers located in the promoter and intron 1 regions of AGT were significantly associated with RCC compared with the common haplotype in subjects with hypertension or high BMI (global P = 0.003). Our findings suggest that common genetic variants of AGT, particularly those in the promoter, increase RCC risk among subjects who are hypertensive or overweight.


The kidney is an important regulator of blood pressure given its primary role to maintain the homeostatic balance of bodily fluids by filtering and secreting metabolites and minerals from the blood prior to excreting them in the urine. Hypertension and obesity are known risk factors for renal cell carcinoma (RCC), the predominant type of kidney cancer (14). The mechanism by which hypertension contributes to RCC is unclear but may be associated with related functional or metabolic changes, such as decreased blood flow to the kidney and cell proliferation in the renal tubule that can subsequently induce hypoxia and promote angiogenesis (5). Hypertension is also closely linked to obesity, both of which are associated with oxidative damage and lipid peroxidation, which may play an important unifying mechanism in kidney carcinogenesis (69).

A high incidence of RCC in certain ethnic groups (10), as well as a familial susceptibility (1112), suggest that this cancer has a genetic component, which has not been well studied. Given the function of the kidney and the association between hypertension and RCC, it is biologically plausible that variants in genes involved in blood pressure control may contribute to RCC susceptibility. In a large, multi-center, hospital-based case–control study that was specifically designed to assess genetic risk factors for RCC, we examined the association of 142 tagging single-nucleotide polymorphisms (SNPs) in eight candidate genes (ACE, ADD1, AGT, AGTR1, AVP, HTR2A, SCNN1B and SLC2A1) having a role in blood pressure control and vasoconstriction on RCC risk. Briefly, ACE is involved in the conversion of angiotensin, ADD1 in the regulation of cellular signal transduction, AGT encodes angiotensinogen precursor, AGRTR1 encodes the type 1 receptor of angiotensinogen, AVP encodes arginine vasopressin, HTR2A is involved in platelet aggregation, SCNN1B encodes the epithelial sodium channel blocker and SLC2A1 is a major glucose transporter. Since hypertension and obesity are strong risk factors for RCC and may play a role in the pathway linking these genetic variants and RCC, we also examined the interaction between these variants and hypertension and body mass index (BMI).

Materials and methods

Study subjects

Details of the Central and Eastern European Renal Cancer Study have been previously reported (1315). Briefly, participants for this hospital-based case–control study were selected from hospitals in seven study centers: Moscow, Russia; Bucharest, Romania; Lodz, Poland and Prague, Olomouc, Ceske Budejovice and Brno, Czech Republic. Cancer cases were recruited between August 1999 and January 2003 by trained medical staff who reviewed hospital medical records for newly diagnosed and histologically confirmed RCC (ICD-O-2 codes C64) patients between 20 and 79 years of age. Date and method of diagnosis, histological classification, tumor location, stage and grade were also abstracted. RCC cases included clear cell, clear cell with papillary or sarcomatoid features, papillary type I, type II or non-type I and hybrid tumors. Controls were patients admitted to the same hospitals as cases for conditions unrelated to smoking or genitourinary disorders (except for benign prostatic hyperplasia) and were frequency matched to cases on age (±3 years), sex and study center. Some controls had been also included in concurrent case–control studies of lung and head and neck cancer (9). No one disease made up >20% of the control group. Demographic, medical and lifestyle information was obtained through in-person interviews by trained personnel using standardized questionnaires. In total, 1097 cases and 1476 controls were interviewed. Blood samples were collected from all participants who gave consent, 987 cases (90%) and 1298 controls (88%). Response rates at each center ranged from 90.0 to 98.6% for cases and from 90.3 to 96.1% for controls.


We selected eight candidate genes known for their role in blood pressure control and vasoconstriction (ACE, ADD1, AGT, AGTR1, AVP, HTR2A, SCNN1B and SLC2A1) for genotyping. SNPs were selected using a tag SNP method with an r2 of at least 0.80 with preference given for their putative functional significance (16). A total of 142 tagging SNPs spanning regions 20 kb upstream and 10 kb downstream of the eight genes and having a variant allele frequency of at least 5% in Caucasians were analyzed (17). Germ line DNA was extracted from whole blood buffy coat by the standard phenol chloroform method. Genotyping was conducted with an Illumina® GoldenGate® Oligo Pool All (OPA) assay at the National Cancer Institute Core Genotyping Facility. The completion rate for all SNPs examined was >98%, and the genotype frequencies among controls showed no deviation from the expected Hardy–Weinberg equilibrium proportions (P > 0.05). Replicate quality control samples (5% samples) were interspersed among genotyping plates, and the concordance rates between duplicate samples were >98%.

Statistical analysis

This analysis included subjects who completed the interview and had sufficient quality and suitable quantity of DNA for the Illumina GoldenGate® platform (777 cancer cases and 1035 controls). The distributions of selected characteristics including, sex, age (<45, 45–54, 55–64, 65–74 and ≥75), country of residence (Czech Republic, Russia, Poland and Romania), cigarette smoking (never and ever), alcohol intake (none, low, moderate and high), BMI (<18.5, 18.5–24, 25–29, 30–34 and ≥35), self-reported hypertension (no and yes), occupational lead exposure (no and yes) and family history of cancer (no and yes), were compared between cases and controls using the chi-square test. Characteristics having statistically significant associations (P < 0.05) were evaluated as potential confounders of RCC risk.

For each of the eight genes examined, the minimum-P value test (min-P), which corrects for multiple testing while accounting for correlations between SNPs using a resampling permutation procedure, was performed to identify genes for further analysis (min-P < 0.05) (18). Subsequently, we conducted a haplotype-based sliding window method called HaploWalk using Matlab that consecutively examines a 3-SNP region across a gene. The 3-SNP regions that were statistically significant at a False Discovery Rate (FDR) (19) of at least 10% were further analyzed. For these initial ‘gene-based’ tests, we only adjusted for sex, age and country in order to use the most parsimonious model.

Associations between individual SNPs and cancer risk were estimated by calculating odds ratios (ORs) and 95% confidence intervals (CI) using unconditional logistic regression. Risk estimates were calculated relative to the common homozygous genotype assuming no genetic model, a dominant model and an additive model. Models were initially adjusted for age, sex and country and then further adjusted for smoking status, BMI, hypertension, lead exposure and family history of cancer; factors resulting in at least a 10% change in risk estimates were retained in the final model. To account for multiple testing of SNPs, the P-trend was adjusted using the FDR based on the number of SNPs within the gene. Statistical interactions between the above-mentioned characteristics and SNPs were evaluated using a likelihood ratio test, and stratified analyses were also conducted. All SNP tests were conducted using SAS version 9.1 (SAS Institute, Cary, NC).

Linkage disequilibrium (LD) between SNPs was assessed by calculating pairwise Lewontin's D′ and r2 using Haploview version 3.32 (20). Haplotypes were reconstructed using Haplostats in R version 2.0.1 (21,22). Associations between common haplotypes (>5% frequency) and RCC risk were evaluated by computing ORs and 95% CI adjusting for potential confounders, with the most common haplotype as the referent category. Global differences in haplotype frequencies between cases and controls were assessed for each gene using a global score test (22).


The largest proportion of RCC cases were male (60.7%), from the Czech Republic (49.8%), and clear cell in histology (90%). Compared with controls, RCC cases had a higher BMI, were more probably to be exposed to lead, have self-reported hypertension and have a first-degree family history of cancer (Table I). The difference in smoking status was no longer significant after adjusting for the other lifestyle and medical factors. Of the eight genes examined, only AGT had a statistically significant association with RCC risk (minimum P-value permutation test = 0.02) (Table II). Similar statistical significance was achieved in association with three consecutive SNPs in the promoter region of the AGT gene using a 3-SNP sliding Haplowalk analysis (Haplowalk P = 0.02) (Figure 1). Individual SNP analysis showed that at least one marker in the AGT (rs2493137: P = 0.001), AGTR1 (rs385338: P = 0.04) and SCNN1B (rs2303153: P = 0.04) genes was associated with RCC based on the test of trend adjusting for gender, age and country (Table II). However, only the SNP in AGT remained statistically significant after taking multiple testing into account using the FDR (rs2493137: FDR-adjusted P = 0.02) (Table II). Based on these gene-based findings, we focused our additional analyses and presented results for AGT.

Table I.
Selected characteristics of participants
Table II.
Hypertension-related genes and their association with RCC
Fig. 1.
HaploWalk and Haploview of the 17 SNPs examined in the AGT gene.

Of the 17 AGT SNPs examined, five (29%) were significantly associated with RCC under an additive model adjusting for age, gender, country, smoking status, BMI, hypertension and lead exposure: rs1326889 (OR = 1.35, 95% CI = 1.15–1.58, P-trend = 0.0002), rs2493137 (OR = 1.31, 95% CI = 1.12–2.54, P-trend = 0.001), rs7539020 (OR = 1.22, 95% CI = 1.04–1.43, P-trend = 0.02), rs3889728 (OR = 1.27, 95% CI = 1.07–1.51, P-trend = 0.01) and rs3789662 (OR = 1.24, 95% CI = 1.03–1.49, P-trend = 0.02) (Table III). To determine which of these five AGT SNPs was most strongly associated with cancer risk, we performed a backward elimination procedure including the five markers and the covariates listed above. Iteratively, the SNP with the highest P-value was removed from the model. Using this approach, we determined that the kidney cancer association was largely driven by rs1326889 and secondly by rs2493137. The risk estimates in relation to each of these SNPs while adjusting for the effect of the other SNP were as follows: rs1326889 (OR = 1.26, 95% CI = 1.03–1.54, P-trend = 0.02) and rs2493137 (OR = 1.12, 95% CI = 0.99–1.38, P-trend = 0.05). Both of these SNPs are located in the promoter region of AGT and are in high LD but only moderately correlated (D′ = 0.94, r2 = 0.40) (20).

Table III.
ORs and 95% CIs for RCC in relation to AGT SNPs

Stratified analysis suggested a modifying effect of hypertension and overweight/obesity on the genetic susceptibility of RCC risk in relation to AGT, although tests for interaction were not statistically significant (Table IV). Significant increased risks of RCC were seen in relation to the five AGT SNPs mentioned above (rs1326889, rs2493137, rs7539020, rs3889728 and rs3789662) in participants with either hypertension or a high BMI (≥25 kg/m2) or both of these factors, but there was no significant increased risk in participants without hypertension or with a normal BMI (<25 kg/m2). Among subjects with hypertension or a high BMI, the excess risks of RCC were again largely driven by rs1326889 (OR = 1.42, 95% CI = 1.20–1.70, P-trend = 0.001, P-interaction = 0.08) and rs2493137 (OR = 1.36, 95% CI = 1.14–1.63, P-trend = 0.001, P-interaction = 0.24). We also examined the associations stratified by hypertension status and BMI (<25 and ≥25 kg/m2) separately, but excess risk were not higher in just hypertensive or just overweight/obese subjects than what was observed in the main effect analysis. None of the five AGT SNPs associated with RCC were associated with either hypertension or high BMI among the controls. Approximately 93% of self-reported hypertensive subjects also reported using hypertension medication. Therefore, we did not have enough cancer cases to evaluate the heterogeneity of effect between subjects who took and did not take hypertension medication, but none of our findings changed measurably when subjects were limited to those who used medication. Associations between the AGT SNPs and RCC were not significantly modified by country, smoking status, lead exposure or histologic cell type (clear cell versus other subtypes).

Table IV.
ORs and 95% CIs for RCC in relation to AGT SNPs by hypertension and BMI

Strong LD was present among most of the 17 AGT SNPs examined (Figure 1). Among the five AGT SNPs significantly associated with RCC, D′ ranged between 0.61 and 0.99 and r2 between 0.23 and 0.67 (20). Based on the results of the Haplowalk analysis for AGT, a region including five SNPs (rs1326889, rs7515609, rs2493137, rs7539020 and rs3889728) upstream of AGT spanning intron 1 showed a strong association with RCC at an FDR of 0.05 (Figure 1). We inferred five major haplotypes from this block with frequencies ranging between 7.3 and 45.7% among controls (Table V). The global Score test for this block of haplotypes showed a statistically significant difference in the haplotype frequencies between cases and controls (P = 0.02). Consistent with our single marker results, the haplotypes, T-T-T-T-G and T-C-T-T-G, that contain the risk-conferring alleles of rs1326889, rs2493137, rs7539020 and rs3889728, were significantly associated with 1.37- and 1.51-fold risks of RCC compared with the common haplotype (C-T-C-C-A) (Table V). These two haplotypes were significantly associated with RCC among participants with either hypertension or a high BMI (T-T-T-T-G: OR = 1.48, 95% CI = 1.17–1.88; T-C-T-T-G: OR = 1.46, 95% CI = 1.01–2.09) (global P = 0.003), but not among those without both of these factors compared with the common haplotype (global P = 0.49) (Table V). As with the individual SNPs, these haplotypes were not associated with hypertension or BMI in the controls.

Table V.
ORs and 95% CIs for RCC in Relation to AGT haplotypes


In this study, 142 tagging SNPs of eight genes involved in blood pressure control were comprehensively evaluated for their association with RCC using several analytic methods that controlled for Type I error. Five of 17 tagging SNPs targeting the AGT gene, and several of the haplotypes containing the risk-conferring markers, were significantly associated with RCC. Although hypertension or BMI did not significantly modify these associations, increased risks of cancer were limited to subjects with either hypertension or high BMI, suggesting that the AGT gene may play a role in the genetic susceptibility of RCC only among subjects with either hypertension or those who are overweight/obese.

The AGT gene is located on chromosome 1q42.3 and contains five exons and four introns that span ~13 kb (23). The protein encoded by this gene, preangiotensinogen or angiotensinogen precursor, is expressed mainly in the liver. It is cleaved by the enzyme renin resulting in angiotensin I that interacts with the angiotensin-converting enzyme (ACE) to produce angiotensin II, which is primarily involved in blood pressure regulation (23). Mutations in the AGT gene are associated with susceptibility to essential hypertension, pre-eclampsia and renal tubular dysgenesis (23).

Our findings suggest that the promoter and region upstream of AGT may play a role in RCC susceptibility. The promoter is located between the TATA box and transcription initiation site that contains a cis-acting DNA element and is a critical regulatory region of the AGT gene (24). Thus, it is biologically plausible that variants in the promoter region of AGT could influence gene expression by modulating transcription (24,25) or they could be in LD with other nearby functional variants. RCC risk in relation to genetic variants in AGT has not been previously studied; however, AGT has been linked with other cancers. Specifically, an increased risk of postmenopausal breast cancer was found in a Dutch population among homozygous carriers of the M allele of rs699 (M235T) (26); a significantly higher number of heterozygote carriers of rs5051 (−6G>A) were found among Czech men with colorectal cancer compared with healthy men of similar age (27), and an increased risk of H. pylori-related gastric cancer was observed in a Japanese population among carriers of the C allele of rs5050 (−20A>C) (28). Rs699 is located in exon 2 and encodes a methionine to threonine amino acid change, and rs5051 and rs5050 are both located in the promoter (Figure 1). Rs699 is in complete LD but moderate correlation with rs5051 and rs5050 (e.g. rs5050 and rs699: D′ = 1.0, r2 = 0.23), and rs5051 and rs5050 and are in very high LD but moderate correlation with the markers associated with RCC in our study (e.g. rs5050 and rs1326889: D′ = 1.0, r2 = 0.15; rs5050 and rs2493137: D′ = 1.0, r2 = 0.36). Our findings, together with the results from these previous studies, suggest the AGT promoter may play a role in carcinogenesis.

Since the associations between AGT and cancer risk were seen in subjects with either hypertension or overweight/obesity, but not in subjects without both of these factors, the risk-conferring effects of AGT appear to be mediated by hypertension and overweight/obesity. Both hypertension and obesity are linked to cell proliferation and lipid peroxidation in the renal tubule that can lead to inflammation, angiogenesis and oxidative damage, which are known factors in kidney carcinogenesis (57,9). Previously reported associations between other AGT variants (e.g. rs5051, rs5050 and rs699) and hypertension have varied widely (29). The AGT SNPs we examined were not associated with hypertension or overweight/obesity in the controls in our study population and have not been examined in other studies. Thus, it appears that these variants act on the carcinogenic process by modulating the effects of hypertension and overweight rather than by causing these risk factors.

Strengths of the study include the large-scale evaluation of candidate genes and tagSNPs involved in blood pressure control in a large sample of RCC cases that provided sufficient statistical power to identify small associations and to evaluate potential confounding and effect modification. We ensured high gene coverage (r2 > 0.80) by tagging the whole-genomic region upstream and downstream to reduce chance of false-negative findings. Also, we were able to statistically evaluate chance findings due to multiple testing. Misclassification of cancer and genotype was minimal due to the use of histologically confirmed cases and the high genotyping completion and concordance rates, respectively. Any selection bias was minimal given the high response rate of both cases and controls.

Our study also had some limitations. Although unlikely, the use of hospital-based versus population-based controls may introduce selection bias if certain SNPs are associated with particular diseases that commonly lead to hospitalization; however, no one disease comprised >20% of the control group in our study. Since our subjects were from different countries in Eastern Europe, population stratification is a possibility; there was no evidence of population stratification in a genome-wide association study conducted in this population (13), and the likelihood of this was found to be small among European populations (30). Hypertension status was based on a one-time self-report that was not clinically confirmed; however, blood pressure measurements at interview may not reflect historical levels due to the potential effects of disease and medication use. Also, use of hypertension medication was based on a one-time self-report with no information on type. Since 93% of self-reported hypertensive subjects reported using medication, we did not have enough cancer cases to evaluate subjects who did not take medication separately; however, we found no change in our results when subjects who did not use medication were excluded from the analysis.

In conclusion, we identified several genetic variants of the AGT gene that increased the risk of RCC, particularly in subjects with hypertension, overweight/obese or both. The identified variants are located in the promoter and upstream region of the gene and are in LD with other variants that may effect gene transcription or function. Since the role of common genetic variants in RCC is not well studied, our findings are in need of replication but do highlight a potential genetic region of interest.


Federal funds from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services.

Supplementary Material

[Supplementary Data]



body mass index
confidence interval
False Discovery Rate
Linkage disequilibrium
odds ratio
renal cell carcinoma
single-nucleotide polymorphisms


1. McLaughlin JK, et al. International renal-cell cancer study VIII. Role of diuretics, other anti-hypertensive medications and hypertension. Int. J. Cancer. 63:216–221. [PubMed]
2. Yuan JM, et al. Hypertension, obesity and their medications in relation to renal cell carcinoma. Br. J. Cancer. 1998;77:1508–1513. [PMC free article] [PubMed]
3. Chow WH, et al. Obesity, hypertension, and the risk of kidney cancer in men. N. Engl. J. Med. 2000;343:1305–1311. [PubMed]
4. Chow WH, et al. Epidemiology of renal cell carcinoma. In: Vogelzang N, Scardino P, Shipley W, Debruyne F, Linehan W, editors. Comprehensive Textbook of Genitourinary Oncology. 3rd edn. Philadelphia: Lippincott Williams & Wilkins; 2006. pp. 669–679.
5. McLaughlin JK, et al., editors. Cancer Epidemiology and Prevention (Edition III) Oxford University Press; 2006. pp. 787–800.
6. Schlehofer B, et al. International renal-cell-cancer study. VI. The role of medical and family history. Int. J. Cancer. 1996;66:723–726. [PubMed]
7. Lindblad P, et al. The role of diabetes mellitus in the aetiology of renal cell cancer. Diabetologia. 1999;42:107–112. [PubMed]
8. Gago-Domiguez M, et al. Lipid peroxidation: a novel and unifying concept of the etiology of renal cell carcinoma (United States) Cancer Causes Control. 2002;13:287–293. [PubMed]
9. Moore LE, et al. Apolipoprotein E/CI locus variants modify renal cell carcinoma risk. Cancer Res. 2009;69:8001–8008. [PMC free article] [PubMed]
10. Parkin CM, et al. Cancer Incidence in Rive Continents, Volume VIII. IARC Scientific Publications NO 155. Lyon: International Agency for Research on Cancer;
11. Zbar B. Inherited epithelial tumors of the kidney: old and new diseases. Semin. Cancer Biol. 2000;10:313–318. [PubMed]
12. Gago-Dominguez M, et al. Family history and risk of renal cell carcinoma. Cancer Epidemiol. Biomarkers Prev. 2001;10:1001–1004. [PubMed]
13. Moore LE, et al. Folate metabolism genes, vegetable intake and renal cancer risk in central Europe. Int. J. Cancer. 2008;122:1710–1715. [PubMed]
14. Moore LE, et al. Glutathione S-transferase polymorphisms, cruciferous vegetable intake and cancer risk in the Central and Eastern European Kidney Cancer Study. Carcinogenesis. 2007;28:1960–1964. [PubMed]
15. Brennan P, et al. Tobacco smoking, body mass index, hypertension, and kidney cancer risk in central and eastern Europe. Br. J. Cancer. 2008;99:1912–1915. [PMC free article] [PubMed]
16. Carlson CS, et al. Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium. Am. J. Hum. Genet. 2004;74:106–120. [PubMed]
17. Packer BR, et al. SNP500Cancer: a public resource for sequence validation, assay development, and frequency analysis for genetic variation in candidate genes. Nucleic Acids Res. 2006;34:D617–D621. [PMC free article] [PubMed]
18. Chen BE, et al. Resampling-based multiple hypothesis testing procedures for genetic case-control association studies. Genet. Epidemiol. 2006;30:495–507. [PubMed]
19. Benjamini Y, et al. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B Methodol. 1995;57:289–300.
20. Barrett JC, et al. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–265. [PubMed]
21. Excoffier L, et al. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol. Biol. Evol. 1995;12:921–927. [PubMed]
22. Schaid DJ, et al. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am. J. Hum. Genet. 2002;70:425–434. [PubMed]
23. Corvol P, et al. Seven lessons from two candidate genes in human essential hypertension: angiotensinogen and epithelial sodium channel. Hypertension. 1999;33:1324–1331. [PubMed]
24. Yanai K, et al. A cis-acting DNA element located between TATA box and transcription initiation site is critical in response to regulatory sequences in human angiotensinogen gene. J. Biol. Chem. 1996;271:15981–15986. [PubMed]
25. Sarzani R, et al. Angiotensinogen promoter variants influence gene expression in human kidney and visceral adipose tissue. J. Hum. Hypertens. 2009:1–7. [PubMed]
26. González-Zuloeta Ladd AM, et al. Differential roles of Angiotensinogen and Angiotensin Receptor type 1 polymorphisms in breast cancer risk. Breast Cancer Res. Treat. 2007;101:299–304. [PubMed]
27. Vašků A, et al. Obesity-related genes variability in Czech patients with sporadic colorectal cancer: preliminary results. Int. J. Colorectal. Dis. 2008;24:289–294. [PubMed]
28. Sugimoto M, et al. Role of angiotensinogen gene polymorphism on Helicobacter pylori infection-related gastric cancer risk in Japanese. Carcinogenesis. 2007;28:2036–2040. [PubMed]
29. Staessen JA, et al. Genetic variability in the renin-angiotensin system: prevalence of alleles and genotypes. J. Cardiovasc. Risk. 1997;4:401–422. [PubMed]
30. Wacholder S, et al. Counterpoint: bias from population stratfication is not a major threat to the validity of conclusions from epidemiological studies of common polymorphisms and cancer. Cancer Epidemiol. Biomarkers Prev. 2002;11:513–520. [PubMed]

Articles from Carcinogenesis are provided here courtesy of Oxford University Press