The clinical use of genetic variation in the evaluation of cancer risk is expanding, and thus understanding how determinants of cancer susceptibility identified in one population can be applied to another is of growing importance. However there is considerable debate on the relevance of ethnic background in clinical genetics, reflecting both the significance and complexity of genetic heritage. We address this via a systematic review of reported associations with cancer risk for 82 markers in 68 studies across six different cancer types, comparing association results between ethnic groups and examining linkage disequilibrium between risk alleles and nearby genetic loci. We find that the relevance of ethnic background depends on the question. If asked whether the association of variants with disease risk is conserved across ethnic boundaries, we find that the answer is yes, the majority of markers show insignificant variability in association with cancer risk across ethnic groups. However if the question is whether a significant association between a variant and cancer risk is likely to reproduce, the answer is no, most markers do not validate in an ethnic group other than the discovery cohort’s ancestry. This lack of reproducibility is not attributable to studies being inadequately populated due to low allele frequency in other ethnic groups. Instead, differences in local genomic structure between ethnic groups are associated with the strength of association with cancer risk and therefore confound interpretation of the implied physiologic association tracked by the disease allele. This suggest that a biological association for cancer risk alleles may be broadly consistent across ethnic boundaries, but reproduction of a clinical study in another ethnic group is uncommon, in part due to confounding genomic architecture. As clinical studies are increasingly performed globally this has important implications for how cancer risk stratifiers should be studied and employed.
Citation: Jing L, Su L, Ring BZ (2014) Ethnic Background and Genetic Variation in the Evaluation of Cancer Risk: A Systematic Review. PLoS ONE 9(6): e97522. https://doi.org/10.1371/journal.pone.0097522
Editor: Devin C. Koestler, University of Kansas Medical Center, United States of America
Received: December 17, 2013; Accepted: April 21, 2014; Published: June 5, 2014
Copyright: © 2014 Jing et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by research grants from Huazhong University of Science and Technology (0124170068). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The incidence, prevalence and mortality of many cancers among different ethnic populations are often very distinct –. For example, African-American men have among the highest incidence of prostate cancer, while Japanese men living in Japan have the lowest incidence . Strong ethnic differences have also been observed in breast cancer risk; Hispanic and Native American women have a markedly lower incidence of breast cancer compared with non-Hispanic women of European descent . The causes of these disparities are manifold, including intrinsic differences, i.e., genetic variation, and extrinsic differences, which include dissimilarities in social, economic, and geographical environments. Understanding these differences in cancer risk and the underlying causes of these differences is crucial for creating research and health care practices that can span ethnic boundaries.
Genetic variation is an important contributor of cancer risk; and recently genome-wide association studies (GWAS) in several cancers have elucidated the roles of many common risk alleles in affecting disease susceptibility. BRCA1 and BRCA2 are the most well-known genes whose mutations are linked to breast cancer risk, and the list of known risk alleles is rapidly expanding –. It is becoming increasingly apparent that ethnic background can play an important role in determining how different alleles are associated with risk of cancer , . Furthermore, several studies examining factors contributing towards cancer susceptibility across multiple ethnic groups, such as the Multi-Ethnic Cohort (MEC) Study, have shown that the tested non-genetic factors did not account for all differences in cancer susceptibility among ethnic groups . In a large prospective study of colon cancer, one MEC study found that ethnic variation in the incidence of colon cancer was not fully explained by differences in the prevalence of the tested extrinsic risk factors: Japanese Americans of both sexes and African American women remained at increased risk of cancer relative to those of European descent after accounting for differences in tested extrinsic risk factors . Similarly, another MEC study found significant differences in the association between cigarette smoking and the risk of lung cancer among five ethnic groups. The findings could not be explained by differences between populations in the tested risk factors, including diet, occupation, and socioeconomic status . These studies suggest that unexplained genetic factors may be important for understanding differences in cancer risk between ethnic groups.
Genetic variation among ethnic groups impacts cancer risk in multiple manners: there may be different frequencies for a risk allele between populations, an allele may have dissimilar associations with risk in different populations, and an allele may interact with other genetic or environmental factors that vary among populations. The HapMap project has made great advances in elucidating the varying prevalence of alleles among ethnic groups . However, information about the other ways in which differences among ethnic groups can affect cancer susceptibility is less well systematically studied. While the Multiethnic Cohort Study is an example of how this research can occur, and individual studies highlight the importance of an understanding of ethnic variation, there is a pressing need for more thorough surveys of the interplay of genetics and ethnicity in determining cancer susceptibility.
As the potential clinical utility of risk alleles for patient stratification are increasingly considered –, the need to understand how these variations may differentially affect members of diverse ethnic groups is growing. Concomitantly, the accurate translation of clinical studies from one ethnic group to another becomes more important as economic factors drive an increasing number of clinical studies to be performed as multiregional trials, with global results used in support of applications in the sponsoring country , . To date, well-populated studies for the identification of associations between gene variants, and the validation of these associations, has been conducted primarily in populations of European ancestry; however the utilization of these findings in other populations may not be straightforward. A study by Ioannidis et al., which examined published meta-analyses of gene association studies involving several complex diseases (including four cancer types) wherein the polymorphism was seen to be significant in at least one ethnic group, found low heterogeneity among ethnic groups in the majority of the studied loci . This study, which focused on validation studies of candidate markers, possibly contained many causative genetic variants, and suggests that basic biology is conserved across ethnic boundaries. However, many apparent differences between ethnic populations in how alleles are associated with cancer risk have been identified (for examples, Table S2–S7). A related study to the 2004 study of Ioannidis et al. showed that when loci identified from genome wide association studies of several complex diseases were assessed the majority of studied loci did not show consistency of disease association across ethnic backgrounds . This second study, focusing on GWAS nominated variants, likely includes many markers only in linkage disequilibrium with the causative variant. Similarly, a study of several GWAS identified prostate cancer risk loci showed that most of the assessed loci did not replicate in a Japanese population . The results of these study suggest that GWAS identified loci, as compared to those identified from family studies (such as BRCA1) or candidate gene approaches, are less likely to be tightly linked to the true functional loci, leading to relatively weaker strengths of association. Further clarification of the role of ethnic background in affecting the association of variants with cancer risk is needed. As cancer risk profiling becomes increasingly common, and as an increasing number of treatment decisions are linked to genotyping results, e.g., erlotinib used for the treatment of lung cancer patients with EGFR mutations , or cetuximab therapy for colon patients lacking KRAS mutations  elucidating the roles ethnic differences have in the clinical management of cancer will entail a better understanding the relationship between ethnicity and predictive markers.
Here we present a survey and systematic analysis of association studies conducted in multiple ethnic groups for the primary known risk alleles in lung, stomach, liver, colon, breast and prostate cancer. These cancers were chosen based on incidence rates; lung, stomach, liver, colon and breast cancer are the cause of most cancer-related deaths each year in both sexes, and in men the second most frequent cause of cancer-related mortality is prostate cancer . We find that most of the associations between gene variants and cancer risk that we surveyed did not validate in new ethnic populations, consistent with other studies that have examined the reproducibility of complex disease risk variants. As low prevalence of the risk alleles in some populations may lead to studies being inadequately populated to validate associations found to be significant in another population, some of the disparate associations among ethnic groups may be attributable simply to low powered studies. However we found that, though many studies were inadequately powered, low allele frequency did not explain the inability to reproduce significant findings between ethnic groups. Instead, we show that differences in linkage disequilibrium appear to be associated with differences in the odds ratio (OR) between ethnic groups. Despite the infrequent validation of significant associations, we find that variability in the odds ratios for the studied variants among ethnic groups are usually not significant. This suggest that the basic biological role, or at least their association, of genetic variants are broadly consistent across ethnic boundaries, but that most well-studied risk loci may be poorly linked to the probable true functional loci in many populations. Therefore great attention needs to be paid when attempting to translate cancer risk associations between ethnic groups. Identification of more tightly linked risk markers is important, as well as validation within the ethnic group in question, for understanding the potential role of ethnic background in affecting cancer susceptibility and to allow proper utilization of potentially clinically relevant findings between ethnic groups.
Materials and Methods
We systematically searched PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) and Web of Science electronic databases (http://apps.webofknowledge.com) for meta-analyses published prior to December 2013 that reported the association between alleles and cancer risks within ethnic groups in six cancer types: lung, stomach, liver, colon, breast and prostate. We also searched for SNPs currently used by major popular genome profiling services for the risk stratification of the six cancers, including 23&Me (https://www.23andme.com/), Navigenics (http://www.navigenics.com/), and United Gene (http://www.ugi.hk/), for these alleles we broadened the search to any study (not limited to meta-analysis) that provided information on ethnic background. This was to ensure that variations already used in commercial assays were in this study; however the records found in these specific searches were all identified by the searches open to all variants. When multiple reports were available for a single study, only the most recent report was included.
For inclusion, the studies must have met all the following criteria: (1) included information for at least two ethnic groups; (2) were meta-analyses of case-control or cohort studies that had original data of a quantitative assessment of the relationship of one gene or SNP and risk of one of the six specified cancer; (3) results were expressed as an odds ratio; and (4) with a 95% confidence interval (CI) for the OR. In addition, variants included in commercial personal genomics assays offered by 23&Me, Navigenics, and United Gene to estimate the risk of the six cancers were used as search criteria, the requirement of the study being a meta-analysis was not used for these variants.
The following exclusion criteria were used: (1) case-only studies, case reports, editorials and abstracts; (2) studies that were missing case and control numbers or an OR; and (3) studies reporting only results in only one ethnic group. No language or publication date restrictions were imposed.
Data from all included papers was tabulated (Tables S1–S7). When data for multiple genetic models are presented, the model with the largest population that had a significant association between allele and risk was selected for further analysis. If no significant association existed then the model with the largest total population was selected. In tabulating all pairwise comparisons between ethnic groups for each SNP, the ethnic group with the largest population giving a significant result was selected as a reference population. When significance was not found for any ethic group the largest population was used as the reference. For alleles where one ethnic group exhibited a significant association with cancer occurrence and another group did not, a power analysis was performed for the non-significant studies using the Genetic Power Calculator , based on the cancer’s prevalence in the relevant ethnic groups, the number of cases and controls, and an estimated relative risk. Prevalence was derived from World Health Organization statistics (http://globocan.iarc.fr/). Relative risk was estimated using the power calculator using the known prevalence and the given odds ratio as an initial approximation of the relative risk. For this study, “well powered” is a power greater than or equal to 80%. To assess heterogeneity among ethnic groups for the associations with risk a Breslow-Day test with Tarone’s adjustment  was employed, as implemented in the R metafor package . Loci were excluded if incomplete case and control numbers for each ethnic group were not reported. Pairwise linkage disequilibrium was measured using Haploview 4.2 software . All r2 values for SNP pairs with the assessed variant within a region 50 kilobase on each side of the locus of interest were evaluated with a one-way permutation test based on Monte-Carlo resampling (replications = 10,000) to compare LD patterns between ethnic groups, as implemented in the R coin package . Only SNPs which had at least 20 SNP pairs available for the LD analysis within this region were assessed. Agreement between odds ratios was compared with a z test on the difference of the odds ratios only the odds ratios for which the association study was significant or were at least 80% powered to validate the significant finding in the reference population were evaluated. A mixed model with the SNP as a grouping variable and cancer type as a random effect were used to evaluate significance of the association of agreement in LD between ethnic groups with significance of the difference in OR. Potential publishing bias was assessed using funnel plots and Egger’s regression test . Results were considered significant for p-values (two tailed) less than 0.05.
Genetic Variant Selection
We searched for studies comparing association of cancer risk with allelic variations in breast, colon, lung, liver, gastric and prostate cancer in different ethnic groups. This analysis was open any genetic variant affecting the six cancer types but we also specifically included SNPs currently used by major popular genome profiling services for the risk stratification of the six cancers. Based on this strategy, 68 publications met our inclusion criteria for further analysis (Figure 1A). We obtained data for 96 assessed associations between cancer risk and genetic variants across the six cancers (82 unique variants) (Tables S1) from these papers. In total, 50 loci were associated with breast cancer, the other SNPs were distributed as: colon: 23 SNPs, liver: 8 SNPs, gastric: 4 SNPs, lung: 6 SNPs and prostate 5 SNPs. The ancestral allele and frequency of ancestral allele are summarized (see Methods, Table S1). Study process is shown in Figure 1B. An assessment of potential publication bias for the included studies (using funnel plots and Egger’s regression test) showed no significant bias for all cancers except breast (Figure S3). When assessed within each ethnic group, no bias was observed in the included breast cancer studies either.
B) Associations between markers and cancer risk were compared between ethnic groups. Among the 86 SNPs assessed in this study, 123 pairwise comparisons of association results between ethnic groups were made. The association results were assessed to determine if each ethnic group was sufficiently populated to find significant results found in other groups. Where differences were found between groups, linkage disequilibrium analysis was performed. The Breslow-Day test for heterogeneity with Tarone’s adjustment was used on all studies with sufficient data. *Both groups had significant results, but with opposite signs.
Association with Cancer Risk among Ethnic Groups
To estimate the importance of genetic contributions among ethnic groups in the evaluation of cancer risk, we surveyed the OR’s primarily in populations of European, Asian and African descent. To reduce the amount of heterogeneity within North American studies (where populations may have diverse ancestry), if the ethnic background was not stated then participants were not assumed to have European heritage. The odds ratio values and numbers of studies, number of cases and controls for each model, and the type of genetic models tested were collected (Tables S2–S7). To clarify the possible causes of the dissimilarities in OR observed between ethnic groups, we calculated the power of all studies within each ethnic group that gave non-significant results where another ethnic group had a significant association for the same allele in the same study. Results are tabulated in table 1 for all pairwise comparisons between ethnic groups within each SNP; the available data for the 82 unique variants assessed in this analysis allowed 123 pairwise comparisons. Disagreement between populations on the presence of a significant association is potentially due to the non-significant study being underpowered. Indeed, within the 80 comparisons between ethnic groups where a significant association was found in at least one population, 39 comparisons (49%) were underpowered to validate the significant result. However, in the 41 comparisons between ethnic groups that were adequately powered to validate the significant result, only 12% (5/41) of allele associations replicated, 85% of comparisons in well-powered studies (35/41) showed no significance for the association in the validation population. The results were similar across all studied cancers and between ethnic groups. In general, the association of genetic loci with cancer risk usually do not replicate in different ethnic groups
Heterogeneity of Association with Cancer Occurrence among Ethnic Groups
Though the most of the associations between genetic variants and cancer risk that were assessed in this study do not replicate between ethnic groups, this does not establish that there is no consistency of association for these variants across ethnic groups. Indeed, a survey of the odds ratios and confidence intervals in the studied loci suggests that the effect on cancer risk associated with the studied alleles may often be consistent across ethnic boundaries (Fig. 2, lung, gastric, liver, and prostate cancer; Fig. S1, breast cancer; Fig. S2, colorectal carcinoma). Though considerable variation is apparent among the ethnic groups, the direction of the association is often conserved. To more rigorously evaluate this, the differences of the odds ratio between ethnic groups was assessed using the Breslow-Day test with Tarone’s adjustment  to determine whether there was significant heterogeneity among ethnic groups. The Breslow-Day test assesses the homogeneity of the odds ratio across contingency tables and has an approximate chi-squared distribution. Loci were excluded if incomplete case and control numbers for each ethnic group were not reported. Only a minority of loci showed significant heterogeneity among ethnic groups (25%, 15/60 SNPs, Table 2). There were some differences between the cancer types, with two out of the four tested loci in gastric cancer showing significant heterogeneity, but the number of loci is too small to statistically determine if there is a meaningful difference in heterogeneity between the different cancers. Excluding gastric cancer, loci exhibiting significant heterogeneity were in the minority, ranging from 8% (colon cancer) to 40% (prostate cancer). If the analysis is restricted to only include data from populations where a significant result was found or the study was well powered, similar results are found, with 67% (28/42) of loci showing non-significant heterogeneity among ethnic groups (data not shown). As discoveries of significant risk associations in small populations could skew the results, the analysis was also performed excluding discovery populations whose number of total participants were less than the 10th percentile of this entire study (N<548). The results were not appreciably changed, five SNPs were affected and the number of loci showing significant heterogeneity was 26%. Therefore, by this measure, association with cancer risk is broadly consistent across ethnic boundaries; a finding of an association with risk in one population predicts the direction of that risk association in another ethnic group. However, as our results in table 1 demonstrate, this does not mean that one should expect a significant association in one ethnic group to lead to a significant result in another ethnic group.
The results within liver, gastric, lung and prostate cancer are shown. OR’s from European populations are shown in black, Asian in red, African in green, and other groups in blue. Though considerably heterogeneity is apparent, the association with risk for a marker in one ethnic group appears to predict the direction of the association in the other ethnic groups, as supported by the test for heterogeneity. Similar plots for breast and colon cancer are given in Figure S1 and S2, respectively.
Linkage Disequilibrium Analysis
For sites of variation with disagreement between ethnic groups (as defined by significant results predicting increased or decreased risk for the same allele, or significant results in one group but non-significant yet powered analysis in another group), linkage disequilibrium (LD) analysis was performed. LD patterns between ethnic groups within a region 50 kilobase on each side of the locus of interest were compared. Looking at cases which had at least 20 SNP pairs available for LD analysis within this region, 62% of loci showed significant differences in the r2 of SNPs compared with the tested variant between ethnic groups and 23% of loci showed significant disagreement between ethnic groups in odds ratio, as assessed by a z test on the difference in the odds ratios (Table 3). A linear mixed model for the agreement of these tests showed significant association (p = 0.013). This result suggests that the agreement in OR between different ethnic groups is associated with the comparative variation in the surrounding genome structure. This likely reflects that in conserved regions the link between the tested marker and actual risk allele remain tight.
Our results demonstrate that ethnic background usually plays an important role in affecting the association between a putative risk marker and cancer risk. In a survey of studies encompassing 96 risk:variant associations (82 unique alleles) in six cancers assessing the association between cancer susceptibility and allelic variations, we found that a significant result in one ethnic group was usually not reproducible on other ethnicities in well-powered studies. This is consistent with other studies , , , though this is the first large review to focus on cancer risk associations. Whether clinical studies are to be expected to validate has been a subject of interest of late , and there are many reasons why a result may fail to replicate. One hypothesis we initially entertained was that insufficient case numbers for rare alleles would account for the majority of disparate results. However this hypothesis was not supported, limiting analysis to well-powered studies still saw that most associations between variants and cancer risk did not replicate in different ethnic groups. However, we also saw that most loci exhibited consistency in their association with risk, most loci did not have statistically significant heterogeneity in the OR’s among the studied ethnic groups. These results are not contradictory, but the distinction is important in understanding the complicated manners that ethnic variation can affect clinical studies. The test for heterogeneity suggests that the basic biologic effect of a site of genetic variation may often be shared across ethnic boundaries. On the other hand, the power analyses suggests that, despite this putatively shared biology, reproducing a result found in one ethnic group may be difficult to achieve in another group. Therefore, although a basic biological effect may be conserved, the tested alleles’ contributions to cancer risk appear to include factors intrinsically distinct between ethnic groups. These factors are likely to confound efforts to translate utility of a marker from one ethnic group to another unless adequately accommodated.
The cause of the different association for a marker among ethnic groups could be due to either the risk alleles being linked to the real causative allele with differing strengths between the groups, the allele acting in different manners across ethnic boundaries in how it affects cancer risk, or differing interactions between the risk allele with environmental or other genetic elements that vary among populations. We present evidence that genetic linkage appears to be a strong factor in explaining the differing association between marker and risk for many of the tested alleles, consistent with findings in other studies . This does not mean that environmental and higher level genetic interactions do not contribute to inter-ethnic diversity. These results do suggest that, when trying to translate genetic association results from one ethnic group to another, validation within all ethnic groups of interest is vitally important and efforts to identify causal genetic loci and or closely linked loci will improve conservation across ethnic boundaries.
The results reported here suggest that the linkage between commonly utilized or studied cancer risks markers at defined risk loci are often poorly linked to the actual risk alleles. Using this genetic diversity among populations may therefore allow better mapping of these true risk alleles. As LD structure has been demonstrated to vary among ethnic groups, studies assessing multiple ethnic groups can greatly aid these types of efforts . Allelic variation in high LD with a marker linked to risk in a studied population serves as candidates of possible risk alleles to be assessed in the index ethnic group and yet untested populations. For example, fine-mapping in Asian, European, and African-Americans in a FGFR2 associated allele in breast cancer led to better definition of the risk region . In this regard, the linkage differences among ethnic populations may be useful for the nomination of SNPs that are more closely linked to the true functional SNP. Therefore, SNPs in high LD to the tested risk marker in the ethnic group with the significant association, but more loosely linked in the group with a non-significant association may indicate regions where the true functional SNP resides. As an example, in figure 3 we show examples from breast (Fig. 3A) and colon (Fig. 3B) cancers. In each case, SNPs with a more consistent association with cancer across ethnic boundaries were found in regions nearby the initial tested markers. Continued fine mapping of variants, and increased reporting of all results from GWAS (not just the markers that meet the corrected significance levels required in the identification of novel markers) will greatly speed up the ability to use such information to identify risk markers that translate across ethnic boundaries.
Black triangles represent a SNP which exhibited significant association with risk in one population and non-significant association in a different ethnic group. Gray triangles are SNPs that are tightly linked to this marker in the population with a significant association but more loosely linked in the non-significant population. White triangles are SNPs very close to this new candidate region with a measured association with outcome. The OR’s shown below the black marker are from this study, those under the gray and white markers from the referenced studies. Significant results are marked with asterisks. A) rs1137101 failed to validate in a European breast cancer population, however the nearby rs3828034 has a higher OR that nears significance . B) rs6983267 failed to replicate in studies of European (US) populations, however the nearby rs7837328 has a more consistent association –. The odds ratio for rs6983267 as reported in this study (Table S2) is based on the ancestral allele, which is also the rare allele in European populations, the odds ratio for the nearby SNPs were reported in relation to the most common allele, therefore for consistency we have also given the OR for rs6983267 in this figure in relation to the common allele.
As understanding of the genetic variation among disparate population groups is of clear importance in assessing cancer risk, the risks of using self-reported ethnic designations as surrogates for complete genetic information must be considered as a limitation of this study, and of any study that uses self-reported ethnicity. Potential problems in the use of ancestry identifiers (such as race and ethnicity) in medical studies have been addressed in several reports –. These ethnic labels are surrogates, with significant short-comings, for the shared genetic variation and shared genetic history that explain the differences in allele frequencies observed between population groups , . However the factors that comprise self-reported ethnicity may encompass elements common to genetically distinct groups, such as shared cultural and historical experiences, beliefs and rituals, and other customs. While these elements may also be important in creating a complete risk model for an individual, distinguishing these different types of factors is important, and the use of self-reported ethnic labels may not contribute to their differentiation. Nonetheless, until fine scale mapping or sequencing of individuals becomes the norm in medical diagnostic and therapeutic decision making, the use of ethnic group labels appears necessary.
Forest plot of odds ratios for breast cancer. OR’s from European populations are shown in black, Asian in red, African in green, and other groups in blue.
Forest plot of odds ratios for colon cancer. OR’s from European populations are shown in black, Asian in red, African in green, and other groups in blue.
Funnel plots for assessment of publication bias. Plots are shown for each cancer, and within breast, each ethnic group. Egger’s regression test is used to assess the significance of deviation from symmetry; the P value for this test is shown. A) breast cancer, all populations; B) breast cancer, European populations; C) breast cancer, Asian populations; D) breast cancer, African populations; E) colon cancer, all populations; F) lung cancer, all populations; G) Gastric cancer, all populations; H) Liver cancer, all populations; I) Prostate cancer, all populations.
Variations in this study and their known prevalence in the studied ethnic categories.
Conceived and designed the experiments: BZR. Performed the experiments: LJ LS. Analyzed the data: BZR LJ. Wrote the paper: BZR LJ.
- 1. Garte S (1998) The role of ethnicity in cancer susceptibility gene polymorphisms: the example of CYP1A1. Carcinogenesis 19: 1329–1332.
- 2. Parkin DM (2004) International variation. Oncogene 23: 6329–6340.
- 3. Kheirandish P, Chinegwundoh F (2011) Ethnic differences in prostate cancer. Br J Cancer 105: 481–485.
- 4. Brawley OW (2003) Population categorization and cancer statistics. Cancer Metastasis Rev 22: 11–19.
- 5. Shavers VL, Brown ML (2002) Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst 94: 334–357.
- 6. Fejerman L, Ziv E (2008) Population differences in breast cancer severity. Pharmacogenomics 9: 323–333.
- 7. Ford D, Easton DF, Stratton M, Narod S, Goldgar D, et al. (1998) Genetic heterogeneity and penetrance analysis of the BRCA1 and BRCA2 genes in breast cancer families. The Breast Cancer Linkage Consortium. Am J Hum Genet 62: 676–689.
- 8. Mitchell G, Antoniou AC, Warren R, Peock S, Brown J, et al. (2006) Mammographic density and breast cancer risk in BRCA1 and BRCA2 mutation carriers. Cancer Res 66: 1866–1872.
- 9. Rennert G, Bisland-Naggan S, Barnett-Griness O, Bar-Joseph N, Zhang S, et al. (2007) Clinical outcomes of breast cancer in carriers of BRCA1 and BRCA2 mutations. N Engl J Med 357: 115–123.
- 10. O’Donovan PJ, Livingston DM (2010) BRCA1 and BRCA2: breast/ovarian cancer susceptibility gene products and participants in DNA double-strand break repair. Carcinogenesis 31: 961–967.
- 11. Chinegwundoh F, Enver M, Lee A, Nargund V, Oliver T, et al. (2006) Risk and presenting features of prostate cancer amongst African-Caribbean, South Asian and European men in North-east London. BJU Int 98: 1216–1220.
- 12. Economopoulos K, Sergentanis T (2010) Does race modify the association between CYP1B1 Val432Leu polymorphism and breast cancer risk? A critical appraisal of a recent meta-analysis. Breast Cancer Research and Treatment 124: 293–294.
- 13. Kolonel LN, Altshuler D, Henderson BE (2004) The multiethnic cohort study: exploring genes, lifestyle and cancer risk. Nat Rev Cancer 4: 519–527.
- 14. Ollberding NJ, Nomura AM, Wilkens LR, Henderson BE, Kolonel LN (2011) Racial/ethnic differences in colorectal cancer risk: the multiethnic cohort study. Int J Cancer 129: 1899–1906.
- 15. Haiman CA, Stram DO, Wilkens LR, Pike MC, Kolonel LN, et al. (2006) Ethnic and racial differences in the smoking-related risk of lung cancer. N Engl J Med 354: 333–342.
- 16. International HapMap Consortium (2003) The International HapMap Project. Nature 426: 789–796.
- 17. Robson M, Offit K (2010) Inherited predisposition to cancer: introduction and overview. Hematol Oncol Clin North Am 24: 793–797.
- 18. Plon SE, Cooper HP, Parks B, Dhar SU, Kelly PA, et al. (2011) Genetic testing and cancer risk management recommendations by physicians for at-risk relatives. Genet Med 13: 148–154.
- 19. Green VL (2013) Breast cancer risk assessment, prevention, and the future. Obstet Gynecol Clin North Am 40: 525–549.
- 20. Lang T, Siribaddana S (2012) Clinical trials have gone global: is this a good thing? PLoS Med 9: e1001228.
- 21. Glickman SW, McHutchison JG, Peterson ED, Cairns CB, Harrington RA, et al. (2009) Ethical and scientific implications of the globalization of clinical research. N Engl J Med 360: 816–823.
- 22. Ioannidis JP, Ntzani EE, Trikalinos TA (2004) ‘Racial’ differences in genetic effects for complex diseases. Nat Genet 36: 1312–1318.
- 23. Ntzani EE, Liberopoulos G, Manolio TA, Ioannidis JP (2012) Consistency of genome-wide associations across major ancestral groups. Hum Genet 131: 1057–1071.
- 24. Yamada H, Penney KL, Takahashi H, Katoh T, Yamano Y, et al. (2009) Replication of prostate cancer risk loci in a Japanese case-control association study. J Natl Cancer Inst 101: 1330–1336.
- 25. Zhou C, Wu YL, Chen G, Feng J, Liu XQ, et al. (2011) Erlotinib versus chemotherapy as first-line treatment for patients with advanced EGFR mutation-positive non-small-cell lung cancer (OPTIMAL, CTONG-0802): a multicentre, open-label, randomised, phase 3 study. Lancet Oncol 12: 735–742.
- 26. Wilson PM, Labonte MJ, Lenz HJ (2010) Molecular markers in the treatment of metastatic colorectal cancer. Cancer J 16: 262–272.
- 27. Ferlay JSH, Bray F, Forman D, Mathers C, Parkin DM (2008) GLOBOCAN 2008 v2.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 10 [Internet]. Lyon, France: International Agency for Research on Cancer; 2010.
- 28. Purcell S, Cherny SS, Sham PC (2003) Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 19: 149–150.
- 29. Bray F, Ren JS, Masuyer E, Ferlay J (2013) Global estimates of cancer prevalence for 27 sites in the adult population in 2008. Int J Cancer 132: 1133–1145.
- 30. Tarone R (1985) On Heterogeneity Tests Based on Efficient Scores. Biometrika 72: 91–95.
- 31. Viechtbauer W (2010) Conducting meta-analyses in R with the metafor package. Journal of Statistical Software 36: 1–48.
- 32. Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21: 263–265.
- 33. Hothorn T, Hornik K, van de Wiel MA, Zeileis A (2008) Implementing a Class of Permutation Tests: The coin Package. Journal of Statistical Software 28: 1–23.
- 34. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629–634.
- 35. Rosenberg NA, Huang L, Jewett EM, Szpiech ZA, Jankovic I, et al. (2010) Genome-wide association studies in diverse populations. Nat Rev Genet 11: 356–366.
- 36. Ioannidis JP (2005) Why most published research findings are false. PLoS Med 2: e124.
- 37. Shifman S, Kuypers J, Kokoris M, Yakir B, Darvasi A (2003) Linkage disequilibrium patterns of the human genome across populations. Hum Mol Genet 12: 771–776.
- 38. Udler MS, Meyer KB, Pooley KA, Karlins E, Struewing JP, et al. (2009) FGFR2 variants and breast cancer risk: fine-scale mapping using African American studies and analysis of chromatin conformation. Hum Mol Genet 18: 1692–1703.
- 39. Thomas DC, Witte JS (2002) Point: population stratification: a problem for case-control studies of candidate-gene associations? Cancer Epidemiol Biomarkers Prev 11: 505–512.
- 40. Wacholder S, Rothman N, Caporaso N (2002) Counterpoint: Bias from Population Stratification Is Not a Major Threat to the Validity of Conclusions from Epidemiological Studies of Common Polymorphisms and Cancer. Cancer Epidemiology Biomarkers & Prevention 11: 513–520.
- 41. Rebbeck TR, Sankar P (2005) Ethnicity, ancestry, and race in molecular epidemiologic research. Cancer Epidemiol Biomarkers Prev 14: 2467–2471.
- 42. Kaufman JS, Cooper RS (2001) Commentary: Considerations for Use of Racial/Ethnic Classification in Etiologic Research. Am J Epidemiol 154: 291–298.
- 43. Cooper RS, Kaufman JS, Ward R (2003) Race and genomics. N Engl J Med 348: 1166–1170.
- 44. Jorde LB, Wooding SP (2004) Genetic variation, classification and ‘race’. Nat Genet 36: S28–33.
- 45. Manica A, Prugnolle F, Balloux F (2005) Geography is a better determinant of human genetic differentiation than ethnicity. Hum Genet 118: 366–371.
- 46. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, et al. (2007) A genome-wide association study identifies alleles in FGFR2 associated with risk of sporadic postmenopausal breast cancer. Nat Genet 39: 870–874.
- 47. Kupfer SS, Anderson JR, Hooker S, Skol A, Kittles RA, et al.. (2010) Genetic heterogeneity in colorectal cancer associations between African and European americans. Gastroenterology 139: 1677–1685, 1685 e1671–1678.
- 48. Cui R, Okada Y, Jang SG, Ku JL, Park JG, et al. (2011) Common variant in 6q26-q27 is associated with distal colon cancer in an Asian population. Gut 60: 799–805.
- 49. Berndt SI, Potter JD, Hazra A, Yeager M, Thomas G, et al. (2008) Pooled analysis of genetic variation at chromosome 8q24 and colorectal neoplasia risk. Hum Mol Genet 17: 2665–2672.
- 50. Lu PH, Wei MX, Yang J, Liu X, Tao GQ, et al. (2011) Association between two polymorphisms of ABCB1 and breast cancer risk in the current studies: a meta-analysis. Breast Cancer Research and Treatment 125: 537–543.
- 51. Mao C, Chung VH, He BF, Luo RC, Tang JL (2012) Association between ATM 5557G>A polymorphism and breast cancer risk: a meta-analysis. Molecular Biology Reports 39: 1113–1118.
- 52. Sun H, Bai J, Chen F, Jin Y, Yu Y, et al. (2011) Lack of an association between AURKA T91A polymorphisms and breast cancer: a meta-analysis involving 32,141 subjects. Breast Cancer Research and Treatment 125: 175–179.
- 53. Qiu LX, Yao L, Xue K, Zhang J, Mao C, et al. (2010) BRCA2 N372H polymorphism and breast cancer susceptibility: a meta-analysis involving 44,903 subjects. Breast Cancer Research and Treatment 123: 487–490.
- 54. Sergentanis T, Economopoulos K (2011) Cyclin D1 G870A polymorphism and breast cancer risk: a meta-analysis comprising 9,911 cases and 11,171 controls. Molecular Biology Reports 38: 4955–4963.
- 55. Mao C, Wang XW, Qiu LX, Liao RY, Ding H, et al. (2010) Lack of association between catechol-O-methyltransferase Val108/158Met polymorphism and breast cancer risk: a meta-analysis of 25,627 cases and 34,222 controls. Breast Cancer Research and Treatment 121: 719–725.
- 56. Sergentanis T, Economopoulos K (2010) Four polymorphisms in cytochrome P450 1A1 (CYP1A1) gene and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 122: 459–469.
- 57. Yao L, Yu X, Yu L (2010) Lack of significant association between CYP1A1 T3801C polymorphism and breast cancer risk: a meta-analysis involving 25,087 subjects. Breast Cancer Research and Treatment 122: 503–507.
- 58. Qiu LX, Yao L, Mao C, Yu KD, Zhan P, et al. (2010) Lack of association of CYP1A2–164 A/C polymorphism with breast cancer susceptibility: a meta-analysis involving 17,600 subjects. Breast Cancer Research and Treatment 122: 521–525.
- 59. Yao L, Fang F, Wu Q, Zhong Y, Yu L (2010) No association between CYP1B1 Val432Leu polymorphism and breast cancer risk: a meta-analysis involving 40,303 subjects. Breast Cancer Research and Treatment 122: 237–242.
- 60. Pabalan N, Francisco-Pabalan O, Sung L, Jarjanazi H, Ozcelik H (2010) Meta-analysis of two ERCC2 (XPD) polymorphisms, Asp312Asn and Lys751Gln, in breast cancer. Breast Cancer Research and Treatment 124: 531–541.
- 61. Zhang J, Qiu LX, Wang ZH, Leaw SJ, Wang BY, et al. (2010) Current evidence on the relationship between three polymorphisms in the FGFR2 gene and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 124: 419–424.
- 62. Hu J, Zhou GW, Wang N, Wang Y-J (2010) GPX1 Pro198Leu polymorphism and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 124: 425–431.
- 63. Lu S, Wang Z, Cui D, Liu H, Hao X (2011) Glutathione S-transferase P1 Ile105Val polymorphism and breast cancer risk: a meta-analysis involving 34,658 subjects. Breast Cancer Research and Treatment 125: 253–259.
- 64. Ma Y, Yang J, Zhang P, Liu Z, Yang Z, et al. (2011) Lack of association between HER2 codon 655 polymorphism and breast cancer susceptibility: meta-analysis of 22 studies involving 19,341 subjects. Breast Cancer Research and Treatment 125: 237–241.
- 65. Gu D, Wang M, Zhang Z, Chen J (2010) Lack of association between the hOGG1 Ser326Cys polymorphism and breast cancer risk: evidence from 11 case–control studies. Breast Cancer Research and Treatment 122: 527–531.
- 66. Yao L, Cao LH, Qiu LX, Yu L (2010) The association between HSD17B1 Ser312Gly polymorphism and breast cancer risk: a meta-analysis including 31,053 subjects. Breast Cancer Research and Treatment 123: 577–580.
- 67. Qiu LX, Yao L, Yuan H, Mao C, Chen B, et al. (2010) IGFBP3 A-202C polymorphism and breast cancer susceptibility: a meta-analysis involving 33,557 cases and 45,254 controls. Breast Cancer Research and Treatment 122: 867–871.
- 68. Huang Q, Wang C, Qiu LJ, Shao F, Yu JH (2011) IL-8–251A>T polymorphism is associated with breast cancer risk: a meta-analysis. Journal of Cancer Research and Clinical Oncology 137: 1147–1150.
- 69. He BS, Pan YQ, Zhang Y, Xu YQ, Wang SK (2012) Effect of LEPR Gln223Arg polymorphism on breast cancer risk in different ethnic populations: a meta-analysis. Molecular Biology Reports 39: 3117–3122.
- 70. Chen MB, Li C, Shen WX, Guo YJ, Shen W, et al. (2011) Association of a LSP1 gene rs3817198T>C polymorphism with breast cancer risk: evidence from 33,920 cases and 35,671 controls. Molecular Biology Reports 38: 4687–4695.
- 71. Zhao E, Cui D, Yuan L, Lu W (2012) MDM2 SNP309 polymorphism and breast cancer risk: a meta-analysis. Molecular Biology Reports 39: 3471–3477.
- 72. Qiu LX, Yao L, Mao C, Chen B, Zhan P, et al. (2010) Lack of association between MnSOD Val16Ala polymorphism and breast cancer risk: a meta-analysis involving 58,448 subjects. Breast Cancer Research and Treatment 123: 543–547.
- 73. Qi X, Ma X, Yang X, Fan L, Zhang Y, et al. (2010) Methylenetetrahydrofolate reductase polymorphisms and breast cancer risk: a meta-analysis from 41 studies with 16,480 cases and 22,388 controls. Breast Cancer Research and Treatment 123: 499–506.
- 74. Hu J, Zhou GW, Wang N, Wang YJ (2010) MTRR A66G polymorphism and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 124: 779–784.
- 75. Wang Z, Cui D, Lu W (2010) NBS1 8360G > C polymorphism is associated with breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 123: 557–561.
- 76. Yao L, Fang F, Zhong Y, Yu L (2010) The association between two polymorphisms of eNOS and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 124: 223–227.
- 77. Qiu LX, Zhang J, Zhu XD, Zheng CL, Sun S, et al. (2010) The p21 Ser31Arg polymorphism and breast cancer risk: a meta-analysis involving 51,236 subjects. Breast Cancer Research and Treatment 124: 475–479.
- 78. Zhou GW, Hu J, Peng XD, Li Q (2011) RAD51 135G>C polymorphism and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 125: 529–535.
- 79. Xi B, Zeng T, Liu L, Liang Y, Liu W, et al. (2011) Association between polymorphisms of the renin–angiotensin system genes and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 130: 561–568.
- 80. Ma X, Chen C, Xiong H, Fan J, Li Y, et al. (2010) No association between SOD2 Val16Ala polymorphism and breast cancer susceptibility: a meta-analysis based on 9,710 cases and 11,041 controls. Breast Cancer Research and Treatment 122: 509–514.
- 81. Wang Z, Fu Y, Tang C, Lu S, Chu WM (2010) SULT1A1 R213H polymorphism and breast cancer risk: a meta-analysis based on 8,454 cases and 11,800 controls. Breast Cancer Research and Treatment 122: 193–198.
- 82. Huang Y, Li B, Qian J, Xie J, Yu L (2010) TGF-β1 29T/C polymorphism and breast cancer risk: a meta-analysis involving 25,996 subjects. Breast Cancer Research and Treatment 123: 863–868.
- 83. Ma X, Chen C, Xiong H, Li Y (2010) Transforming growth factorβ1 L10P variant plays an active role on the breast cancer susceptibility in Caucasian: evidence from 10,392 cases and 11,697 controls. Breast Cancer Research and Treatment 124: 453–457.
- 84. Shen C, Sun H, Sun D, Xu L, Zhang X, et al. (2011) Polymorphisms of tumor necrosis factor-alpha and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 126: 763–770.
- 85. He XF, Su J, Zhang Y, Huang X, Liu Y, et al. (2011) Association between the p53 polymorphisms and breast cancer risk: meta-analysis based on case–control study. Breast Cancer Research and Treatment 130: 517–529.
- 86. Wang J, Wang B, Bi J, Di J (2011) The association between two polymorphisms in the TYMS gene and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 128: 203–209.
- 87. Yao L, Qiu LX, Yu L, Yang Z, Yu XJ, et al. (2010) The association between TA-repeat polymorphism in the promoter region of UGT1A1 and breast cancer risk: a meta-analysis. Breast Cancer Research and Treatment 122: 879–882.
- 88. Liu L, Liu L, Zeng F, Wang K, Huang J, et al. (2011) Meta-analysis of the association between VEGF-634 G>C and risk of malignancy based on 23 case–control studies. Journal of Cancer Research and Clinical Oncology 137: 1027–1036.
- 89. Gu D, Wang M (2011) VEGF 936C>T polymorphism and breast cancer risk: evidence from 5,729 cases and 5,868 controls. Breast Cancer Research and Treatment 125: 489–493.
- 90. Ding DP, He XF, Zhang Y (2011) Lack of association between XPG Asp1104His and XPF Arg415Gln polymorphism and breast cancer risk: a meta-analysis of case–control studies. Breast Cancer Research and Treatment 129: 203–209.
- 91. Zhang LQ, Wang J, Shang JQ, Bai Jl, Liu FY, et al. (2011) Cyclin D1 G870A polymorphism and colorectal cancer susceptibility: a meta-analysis of 20 populations. International Journal of Colorectal Disease 26: 1249–1255.
- 92. Haerian MS, Baum L, Haerian BS (2011) Association of 8q24.21 loci with the risk of colorectal cancer: a systematic review and meta-analysis. J Gastroenterol Hepatol 26: 1475–1484.
- 93. Dong J, Dai J, Zhang M, Hu Z, Shen H (2010) Potentially functional COX-2–1195G>A polymorphism increases the risk of digestive system cancers: a meta-analysis. J Gastroenterol Hepatol 25: 1042–1050.
- 94. Zheng Y, Wang JJ, Sun L, Li HL (2012) Association between CYP1A1 polymorphism and colorectal cancer risk: a meta-analysis. Molecular Biology Reports 39: 3533–3540.
- 95. Nock NL, Plummer SJ, Thompson CL, Casey G, Li L (2011) FTO polymorphisms are associated with adult body mass index (BMI) and colorectal adenomas in African-Americans. Carcinogenesis 32: 748–756.
- 96. Liu D, Duan W, Guo H, Xu X, Bai Y (2011) Meta-analysis of associations between polymorphisms in the promoter regions of matrix metalloproteinases and the risk of colorectal cancer. International Journal of Colorectal Disease 26: 1099–1105.
- 97. Keku T, Millikan R, Worley K, Winkel S, Eaton A, et al. (2002) 5,10-Methylenetetrahydrofolate reductase codon 677 and 1298 polymorphisms and colon cancer in African Americans and whites. Cancer Epidemiol Biomarkers Prev 11: 1611–1621.
- 98. Taioli E, Garza MA, Ahn YO, Bishop DT, Bost J, et al. (2009) Meta- and pooled analyses of the methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism and colorectal cancer: a HuGE-GSEC review. Am J Epidemiol 170: 1207–1221.
- 99. Yu K, Zhang J, Zhang J, Dou C, Gu S, et al. (2010) Methionine synthase A2756G polymorphism and cancer risk: a meta-analysis. Eur J Hum Genet 18(3): 370–378.
- 100. Fang F, Yu L, Zhong Y, Yao L (2010) TGFB1 509 C/T polymorphism and colorectal cancer risk: a meta-analysis. Medical Oncology 27: 1324–1328.
- 101. Wang JJ, Zheng Y, Sun L, Wang L, Yu PB, et al. (2011) TP53 codon 72 polymorphism and colorectal cancer susceptibility: a meta-analysis. Molecular Biology Reports 38: 4847–4853.
- 102. Jiang Z, Li C, Xu Y, Cai S (2010) A meta-analysis on XRCC1 and XRCC3 polymorphisms and colorectal cancer risk. International Journal of Colorectal Disease 25: 169–180.
- 103. Wang B, Huang G, Wang D, Li A, Xu Z, et al. (2010) Null genotypes of GSTM1 and GSTT1 contribute to hepatocellular carcinoma risk: evidence from an updated meta-analysis. J Hepatol 53: 508–518.
- 104. Lin ZH, Xin YN, Dong QJ, Wang Q, Jiang XJ, et al. (2010) Association between HLA-DRB1 alleles polymorphism and hepatocellular carcinoma: a meta-analysis. BMC Gastroenterol 10: 145.
- 105. Wei Y, Liu F, Li B, Chen X, Ma Y, et al. (2011) Polymorphisms of Tumor Necrosis Factor-Alpha and Hepatocellular Carcinoma Risk: A HuGE Systematic Review and Meta-Analysis. Digestive Diseases and Sciences 56: 2227–2236.
- 106. Liu F, Li B, Wei Y, Yan L, Wen T, et al. (2011) XRCC1 genetic polymorphism Arg399Gln and hepatocellular carcinoma risk: a meta-analysis. Liver International 31: 802–809.
- 107. Chen B, Zhou Y, Yang P, Liu L, Qin XP, et al. (2011) CDH1 −160C>A gene polymorphism is an ethnicity-dependent risk factor for gastric cancer. Cytokine 55: 266–273.
- 108. Vincenzi B, Patti G, Galluzzo S, Pantano F, Venditti O, et al. (2008) Interleukin 1beta-511T gene (IL1beta) polymorphism is correlated with gastric cancer in the Caucasian population: results from a meta-analysis. Oncol Rep 20: 1213–1220.
- 109. Chen B, Zhou Y, Yang P, Wu XT (2012) Polymorphisms of XRCC1 and gastric cancer susceptibility: a meta-analysis. Molecular Biology Reports 39: 1305–1313.
- 110. Fang F, Wang J, Yao L, Yu XJ, Yu L, et al. (2011) Relationship between XRCC3 T241M polymorphism and gastric cancer risk: a meta-analysis. Medical Oncology 28: 999–1003.
- 111. Liu F, He Y, Peng X, Wang W, Yang X (2010) Association of the 8473T>C cyclooxygenase-2 (COX-2) gene polymorphism with lung cancer risk in Asians. Asian Pac J Cancer Prev 11: 1257–1262.
- 112. Xu W, Zhou Y, Hang X, Shen D (2012) Current evidence on the relationship between CYP1B1 polymorphisms and lung cancer risk: a meta-analysis. Molecular Biology Reports 39: 2821–2829.
- 113. Zhang J, Gu SY, Zhang P, Jia Z, Chang JH (2010) ERCC2 Lys751Gln polymorphism is associated with lung cancer among Caucasians. Eur J Cancer 46: 2479–2484.
- 114. Qian Q, Liu R, Lei Z, You J, Zhou Q, et al. (2011) [Meta analysis of association between Ser326Cys polymorphism of hOGG1 gene and risk of lung cancer]. Zhongguo Fei Ai Za Zhi 14: 205–210.
- 115. Gui XH, Qiu LX, Zhang HF, Zhang DP, Zhong WZ, et al. (2009) MDM2 309 T/G polymorphism is associated with lung cancer risk among Asians. European Journal of Cancer 45: 2023–2026.
- 116. Li Y, Qiu LX, Shen XK, Lv XJ, Qian XP, et al. (2009) A meta-analysis of TP53 codon 72 polymorphism and lung cancer risk: Evidence from 15,857 subjects. Lung Cancer 66: 15–21.
- 117. Ntais C, Polycarpou A, Ioannidis JP (2003) Association of the CYP17 gene polymorphism with the risk of prostate cancer: a meta-analysis. Cancer Epidemiol Biomarkers Prev 12: 120–126.
- 118. Xu B, Tong N, Chen SQ, Hua LX, Wang ZJ, et al. (2011) FGFR4 Gly388Arg polymorphism contributes to prostate cancer development and progression: a meta-analysis of 2618 cases and 2305 controls. BMC Cancer 11: 84.
- 119. Mao C, Qiu LX, Zhan P, Xue K, Ding H, et al. (2010) MnSOD Val16Ala polymorphism and prostate cancer susceptibility: a meta-analysis involving 8,962 Subjects. Journal of Cancer Research and Clinical Oncology 136: 975–979.
- 120. Wang C, Tao W, Chen Q, Hu H, Wen XY, et al. (2010) SRD5A2 V89L polymorphism and prostate cancer risk: a meta-analysis. Prostate 70: 170–178.
- 121. Geng J, Zhang Q, Zhu C, Wang J, Chen L (2009) XRCC1 genetic polymorphism Arg399Gln and prostate cancer risk: a meta-analysis. Urology 74: 648–653.