The genome-wide single-nucleotide polymorphisms (SNPs) profiles can be used as diagnostic markers for human cancers. The associations between mitogen-activated protein kinase kinase kinase 1 (MAP3K1) SNPs rs889312 A>C, rs16886165 T>G and breast cancer risk have been widely evaluated, but the results were inconsistent. To derive a conclusive assessment of the associations, we performed a meta-analysis by combining data from all eligible case-control studies up to date.
By searching PubMed, ISI web of knowledge, Embase and Cochrane databases, we identified all eligible studies published before September 2013. Odds ratios (ORs) with 95% confidence intervals (CIs) were used to assess the strength of associations in fixed-effect or random-effect model. False-positive report probability (FPRP) was calculated to confirm the significance of the results.
A total of 59670 cases in 20 case-control studies were included in this meta-analysis. Significant associations with breast cancer risk were observed for SNPs rs889312 and rs16886165 polymorphisms with a per-allele OR of 1.11 (95% CI: 1.09–1.13) and 1.14 (95% CI: 1.09–1.20) respectively. For rs889312, in subgroup analysis by ethnicity, significant associations were identified in Europeans and Asians, but not in Africans. When stratified by estrogen receptor (ER) expression status, rs889312 was associated with both ER-positive and ER-negative breast cancers. Results from the FPRP analyses were consistent with and supportive to the above results.
Citation: Zheng Q, Ye J, Wu H, Yu Q, Cao J (2014) Association between Mitogen-Activated Protein Kinase Kinase Kinase 1 Polymorphisms and Breast Cancer Susceptibility: A Meta-Analysis of 20 Case-Control Studies. PLoS ONE 9(3): e90771. https://doi.org/10.1371/journal.pone.0090771
Editor: Balraj Mittal, Sanjay Gandhi Medical Institute, India
Received: November 28, 2013; Accepted: February 4, 2014; Published: March 4, 2014
Copyright: © 2014 Zheng 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: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Globally breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death in females, with an estimation of 1 million new cases and over 400,000 deaths per year . Breast cancer is a genetically heterogeneous disease following a polygenic mode of inheritance. A better understanding of the genomic variations pertaining to the disease will eventually lead to improved diagnostic and treatment strategies for breast cancer patients. Evidence has been accumulated that mutations in several high and moderate penetrance genes, such as BRCA1, BRCA2, ATM, CHEK2, BRIP1 and PALB2, contribute to breast cancer susceptibility. However, these mutations explain only a small portion of the disease risk and the majority of genetic variations that contribute to the breast cancer risk remains unclarified, especially those of low penetrance genes. Genome-Wide Association Study (GWAS), an examination of disease-related genetic variations across the human genome, provides a powerful tool in identifying such mutations/polymorphisms genome-wide. Some common low penetrance genetic variants that might confer increased risk to breast cancer have been identified recently , .
Mitogen-activated protein kinase kinase kinase 1 (MAP3K1) gene, which is located in the chromosome 5q11.2, encodes a serine/threonine kinase which is involved in the mitogen-activated protein kinase (MAPK) signaling pathway. MAPK signal transduction regulates the transcription of important cancer genes including c-Myc, c-Elk1, c-Jun and c-Fos , . The single-nucleotide polymorphism (SNP) rs889312 in MAP3K1 was identified to be associated with breast cancer risk by GWAS , with confirmation of the association in European ancestry population by another study . A GWAS conducted in subjects of European descendant reported that the minor allele of MAP3K1 rs16886165 was associated with increased risk of breast cancer under heterozygote codominant and homozygote codominant genetic models .
Up to date, a number of studies have investigated the relationship between polymorphisms in MAP3K1 and breast cancer or its malignant phenotypes, whereas the results were inconsistent and inconclusive in populations of diverse ethnicities, especially among Asians and Africans , . The current GWASs were mainly conducted in people of European ancestry, and most single studies with insufficient sample sizes were not powerful in detecting the minor effects of low penetrance alleles on breast cancer risk. Therefore, the meta-analysis presented here aims to summarize the available evidence for the genomic variants in MAP3K1 and intends to provide the highest level of evidence for the association between breast cancer risk and MAP3K1 (rs889312 and rs16886165) polymorphisms among diverse ancestry populations.
Materials and Methods
Literature search strategy
We conducted a comprehensive literature search in Pubmed, ISI web of knowledge, Embase and Cochrane databases up to September 2013 using the following terms: “breast cancer” or “breast carcinoma” and “MAP3K1” or “mitogen-activated protein kinase kinase kinase 1” or “rs889312” or “rs16886165” and “polymorphism” or “variation”. All potentially relevant studies were retrieved, and the abstracts were screened to exclude clearly irrelevant studies. The remaining articles were read to determine whether they contained information on the topic of interest. We also examined the reference lists from the main reports and reviews to identify additional relevant studies.
Eligible studies had to meet the following criteria: (1) original papers containing independent data which have been published in peer-reviewed journals, (2) case-control studies which evaluated the association between breast cancer risk and MAP3K1 (rs889312 and rs16886165) polymorphisms, (3) had odds ratio (OR) with its 95% confidence interval (CI) or genotype distribution information for estimating OR (95% CI).
Two investigators independently reviewed and extracted data with any discrepancies resolved by consensus. For each eligible study, the following data were extracted: first author's surname, year of publication, country of origin, population ethnicity, source of control subjects, genotyping method, age, the values of OR with its 95% CI, total numbers of cases and controls, the genotype counts in cases and controls, matching criteria, whether genotype distribution of control group was consistent with Hardy-Weinberg equilibrium (HWE). Ethnic groups were categorized as European, Asian and African (i.e. people of European, Asian and African ancestry). Different case-control groups in one study were considered as independent studies if the data was available.
Methodological quality was assessed using a classification method known as “extended-quality score”. The criteria cover diagnostic criteria, the degree of matching of controls, genotyping examination, Hardy-Weinberg equilibrium in the control population, and bias in data processing. Each paper was scored of ‘high’, ‘median’ or ‘poor’ quality. Detailed procedure of the quality assessment was performed according to literature .
The meta-analysis evaluated the association between rs889312 and the risk of breast cancer, for the: (1) allele contrast model (C vs. A), (2) the heterozygote codominant model (AC vs. AA), and (3) the homozygote codominant model (CC vs. AA) model . For rs16886165 polymorphism, pooled OR was obtained by an allele contrast model (G vs. T). The I2-test was performed to assess possible heterogeneity among studies . Hardy-Weinberg equilibrium (HWE) was tested by the x2 test. Pooled odds ratios (ORs) with 95% confidence intervals (95% CIs) were calculated using fixed-effects (Mantel-Haenszel method)  or random-effects (DerSimonian-Laird method)  models. When the effects were assumed to be homogeneous, the fixed-effects model was then used; otherwise, the random-effects model was more appropriate. Subgroup analyses were performed by ethnicity, sample size (<1000 cases and >1000 cases) and estrogen receptor (ER) status (ER-positive and ER-negative). Sensitivity analyses were performed to evaluate the stability of the results by sequential omission of individual studies in the meta-analysis to show the influence of the individual data set to the pooled OR. The Egger's test and funnel plots were utilized to provide diagnosis of publication bias . All of the above analyses were performed using STATA version 12.0 (StataCorp, College Station, TX).
For each statistically significant association, the false-positive report probability (FPRP) analysis was performed using the method reported by Wacholder et al. . A prior probability of 0.001 was set to detect an OR of 1.20. An FPRP cutoff value of 0.2 was used, and only the results with FPRP values less than 0.2 were referred as noteworthy. The Excel spreadsheet provided by Wacholder et al. was used to calculate statistical power and FPRP values (http://jnci.oxfordjournals.org/content/96/6/434/suppl/DC1).
Based on the above criteria, a total of 20 eligible studies involving 59,670 cases and 66,862 controls met all selection criteria for the pooled analyses , , , –. Figure 1 illustrates the study selection process and Table 1 describes the main characteristics of these studies. For the rs889312 polymorphism, 18 studies including a total of 54674 cases and 64542 controls were included in the meta-analysis. For the rs16886165 polymorphism, 4 studies involved a total of 7949 cases and 7660 controls. Of the cases, 67% were Europeans, 21% were Asians, and 10% were of African descendants. The extended-quality scores ranged from 5 to 8, and 19 were given high quality, while the remaining 1 study was given median quality. The distributions of genotypes in the controls in 18 studies were consistent with the Hardy-Weinberg equilibrium, but those in 2 other studies remained unknown.
Overall, the rs889312 polymorphism was significantly associated with increased breast cancer risk under all genetic models (C vs. A: OR = 1.11, 95% CI: 1.09–1.13, P = 0.000, I2 = 27.0%; AC vs. AA: OR = 1.09, 95% CI: 1.05–1.13, P = 0.000, I2 = 20.0%; CC vs. AA: OR = 1.21, 95% CI: 1.15–1.28, P = 0.000, I2 = 31.9%; Figure 2). Additionally, the meta-analysis resulted in a statistically significant association between rs16886165 and breast cancer. The pooled OR for risk G allele was 1.14 under allele contrast model (95% CI: 1.09–1.20, P = 0.000, I2 = 0.0%; Figure 3).
Fixed-effect model was used for the analysis (allele contrast model C vs. A).
Fixed-effect model was used for the analysis (allele contrast model G vs. T).
In the subgroup analysis by ethnicity, our results indicated a significant association between the SNP rs889312 and breast cancer incidence in European ancestry populations (C vs. A: OR = 1.12, 95% CI: 1.10–1.14, P = 0.000, I2 = 0.0%; AC vs. AA: OR = 1.12, 95% CI: 1.08–1.18, P = 0.000, I2 = 0.0%; CC vs. AA: OR = 1.24, 95% CI: 1.15–1.33, P = 0.000, I2 = 0.0%). For Asians, two genetic comparisons produced significantly increased risks (C vs. A: OR = 1.09, 95% CI: 1.05–1.14, P = 0.000, I2 = 42.0%; CC vs. AA: OR = 1.15, 95% CI: 1.05–1.26, P = 0.002, I2 = 17.8%), but the heterozygote codominant model did not reach statistically significance. Among Africans, we did not detect any significant association under all genetic models (Table 2). According to sample size, the rs889312 polymorphism presented significantly increased risks of breast cancer both in small and large studies. The data on rs889312 polymorphism stratified by ER status were available in 6 studies involving 29,200 cases and 44,104 controls. Subsidiary analyses of ER status yielded a per-allele OR for ER-positive tumors of 1.12 (95% CI: 1.10–1.15, P = 0.000, I2 = 0.0%) and ER-negative tumors of 1.08 (95% CI: 1.04–1.12, P = 0.000, I2 = 29.6%; Figure 4). FPRP values at the pre-specified prior probability of 0.001 were lesser than 0.2, except associations among Asians under the homozygote codominant model and small group under the heterozygote codominant model (Table 2).
Fixed-effect model was used for the analysis (allele contrast model C vs. A).
Sensitivity analyses and publication bias
A sensitivity analysis was performed by sequential removal of individual studies. The results suggested that no individual study significantly altered the pooled ORs. The publication bias of the studies was evaluated by funnel plot and Egger's test. As shown in Figure 5 and 6, the shape of the funnel plots did not reveal obvious evidence of asymmetry for both polymorphisms. All the p values of Egger's tests were more than 0.05, suggesting that no publication bias was detected in these studies.
The common variants rs889312 and rs16886165 lie in a linkage disequilibrium (LD) block of approximately 280 kb which includes MAP3K1 gene . The MAP3K1 gene, which is also known as MEKK1 (Mek kinase 1), encodes a 196-kDa serine/threonine protein kinase that activates the ERK (extracellular signal-regulated kinase), JNK (c-Jun NH2-terminal kinase) and NF-κB (nuclear factor-κB) pathways . The downstream signal transductions regulate the survival, differentiation, proliferation and apoptosis of cell, and appear to be involved in tumor development and tumor progression –. As rs16886165 and rs889312 are 88.3 and 79.5 kb upstream of MAP3K1 respectively, the causal variant may be closer to rs16886165 than rs889312 . Although these two SNPs do not change the structure and therefore the biochemical function of MAP3K1, they may have effects on the modulation of MAP3K1 expression and therefore the tuning of MAPK signal transductions.
This meta-analysis demonstrated that the rs889312 and rs16886165 SNPs in MAP3K1 were associated with increased breast cancer susceptibility. There are a number of meta-analyses that have been performed to elucidate the associations between SNPs and breast cancer risk in recent years; however, only 21.7% of those significant associations were noteworthy . Therefore, we performed FPRP analyses in this work to confirm robustness of the linkages. Although a previous meta-analysis directed by Lu et al.  has demonstrated that the rs889312 polymorphism was significantly correlated with breast cancer risk, it only included seven case-control studies without subgroup analysis. In the current meta-analysis, we have made much more powerful and detailed analysis to support our results: (1) more studies were included; (2) subgroup analysis was conducted and stratified by ethnicity and sample size; (3) the association between ER status and breast cancer was considered.
When stratified by ethnicity, the rs889312-C allele showed to be a risk factor for the development of breast cancer in European and Asian ancestry populations, but not in Africans. Different frequencies of mutant alleles in different ethnic groups may contribute to different susceptibilities to cancer . Different life styles and environmental factors among the different populations lead to diverse gene-environment interactions, and may therefore also account for different cancer susceptibilities . In addition, the result may be partly biased because of the limited number of studies in Africans, which had insufficient power to detect a different linkage disequilibrium (LD) pattern or minor effect of the SNP among African populations. Large population studies, which significantly reduce the evidence from smaller studies, play dominant roles in the meta-analysis. When considering sample size, we found rs889312 was associated with increased breast cancer risk both in large and small studies under all genetic models. For rs16886165, the available studies were insufficient to stratified analysis. The pooled result of the current meta-analysis indicated that the rs16886165-G allele conferred breast cancer risk in Asian and African ancestry population. Additionally, rs16886165 have been identified as a low-penetrance risk factor for breast cancer in European ancestry population by GWAS .
The prognosis of breast cancer is affected by ER status. Previous studies suggested that rs889312 was related to ER-positive breast cancers, which was confirmed by our stratified analysis by ER status. For ER-negative tumors, the results of previous studies were contradictory, but this meta-analysis showed rs889312 might also confer risk. Furthermore, the frequencies of the rs889312-C allele in MAP3K1 are similar in ER-positive and ER-negative tumors.
Despite the strengths of this study, such as the large sample size, no significant heterogeneity and high quality of the qualified studies, there are several limitations that should be addressed. First, the sample size of Africans was relatively small, thus the analyses might have insufficient statistical power to detect an association. The trial sequential analysis (TSA) can be used to reveal insufficient information size and potentially false positive results in meta-analyses . However, we are unable to do TSA due to the lack of detailed information which is needed for TSA. Future larger sample studies are necessary to clarify more exact associations between these SNPs in MAP3K1 and breast cancer in African descendants. Second, we performed the analyses under heterozygote codominant model, homozygote codominant model and the subgroup analysis by ER status on a fraction of all data, the selection bias might therefore have occurred. Third, the controls were not uniformly defined, for those based on hospital population might have had benign disease and had different risks for developing breast cancer. Therefore, non-differential misclassification bias might be possible. Finally, due to the lack of individual-level data, the results were based on unadjusted published estimates. We were unable to examine the interactions of possible confounders including age, menopausal status, obesity, smoking, alcohol consumption and environmental factors.
In summary, this meta-analysis indicated that the polymorphism rs889312 was associated with breast cancer risk in Europeans and Asians, while rs16886165 was a risk factor for breast cancer in Asian and African women, and both could be served as markers predisposition to breast cancer. Larger sample studies by using homogeneous patients, unbiased genotyping methods, as well as well-matched controls will provide further supporting information on association of MAP3K1 polymorphisms rs889312 and rs16886165 in breast cancer predisposition.
Conceived and designed the experiments: QZ JY JC. Performed the experiments: QZ HW. Analyzed the data: QZ HW QY. Contributed reagents/materials/analysis tools: QZ QY JC. Wrote the paper: QZ JC JY.
- 1. Jemal A, Bray F (2011) Center MM, Ferlay J, Ward E, et al (2011) Global cancer statistics. CA Cancer J Clin 61: 69–90.
- 2. Pharoah PD, Antoniou AC, Easton DF, Ponder BA (2008) Polygenes, risk prediction, and targeted prevention of breast cancer. N Engl J Med 358: 2796–2803.
- 3. Foulkes WD (2008) Inherited susceptibility to common cancers. N Engl J Med 359: 2143–2153.
- 4. Vinik BS, Kay ES, Fiedorek FT Jr (1995) Mapping of the MEK kinase gene (Mekk) to mouse chromosome 13 and human chromosome 5. Mamm Genome 6: 782–783.
- 5. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, et al. (2007) Genome-wide association study identifies novel breast cancer susceptibility loci. Nature 447: 1087–1093.
- 6. Garcia-Closas M, Hall P, Nevanlinna H, Pooley K, Morrison J, et al. (2008) Heterogeneity of breast cancer associations with five susceptibility loci by clinical and pathological characteristics. PLoS Genet 4: e1000054.
- 7. Thomas G, Jacobs KB, Kraft P, Yeager M, Wacholder S, et al. (2009) A multistage genome-wide association study in breast cancer identifies two new risk alleles at 1p11.2 and 14q24.1 (RAD51L1). Nat Genet 41: 579–584.
- 8. Jiang Y, Han J, Liu J, Zhang G, Wang L, et al. (2011) Risk of genome-wide association study newly identified genetic variants for breast cancer in Chinese women of Heilongjiang Province. Breast Cancer Res Treat 128: 251–257.
- 9. Chan M, Ji SM, Liaw CS, Yap YS, Law HY, et al. (2012) Association of common genetic variants with breast cancer risk and clinicopathological characteristics in a Chinese population. Breast Cancer Res Treat 136: 209–220.
- 10. Li D, Collier DA, He L (2006) Meta-analysis shows strong positive association of the neuregulin 1 (NRG1) gene with schizophrenia. Hum Mol Genet 15: 1995–2002.
- 11. Lu PH, Yang J, Li C, Wei MX, Shen W, et al. (2011) Association between mitogen-activated protein kinase kinase kinase 1 rs889312 polymorphism and breast cancer risk: evidence from 59,977 subjects. Breast Cancer Res Treat 126: 663–670.
- 12. Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21: 1539–1558.
- 13. Mantel N, Haenszel W (1959) Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 22: 719–748.
- 14. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7: 177–188.
- 15. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629–634.
- 16. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N (2004) Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst 96: 434–442.
- 17. Rebbeck TR, DeMichele A, Tran TV, Panossian S, Bunin GR, et al. (2009) Hormone-dependent effects of FGFR2 and MAP3K1 in breast cancer susceptibility in a population-based sample of post-menopausal African-American and European-American women. Carcinogenesis 30: 269–274.
- 18. Barnholtz-Sloan JS, Shetty PB, Guan X, Nyante SJ, Luo J, et al. (2010) FGFR2 and other loci identified in genome-wide association studies are associated with breast cancer in African-American and younger women. Carcinogenesis 31: 1417–1423.
- 19. Gorodnova TV, Kuligina E, Yanus GA, Katanugina AS, Abysheva SN, et al. (2010) Distribution of FGFR2, TNRC9, MAP3K1, LSP1, and 8q24 alleles in genetically enriched breast cancer patients versus elderly tumor-free women. Cancer Genet Cytogenet 199: 69–72.
- 20. Latif A, Hadfield KD, Roberts SA, Shenton A, Lalloo F, et al. (2010) Breast cancer susceptibility variants alter risks in familial disease. J Med Genet 47: 126–131.
- 21. Tamimi RM, Lagiou P, Czene K, Liu J, Ekbom A, et al. (2010) Birth weight, breast cancer susceptibility loci, and breast cancer risk. Cancer Causes Control 21: 689–696.
- 22. Zheng W, Wen W, Gao YT, Shyr Y, Zheng Y, et al. (2010) Genetic and clinical predictors for breast cancer risk assessment and stratification among Chinese women. J Natl Cancer Inst 102: 972–981.
- 23. Campa D, Kaaks R, Le Marchand L, Haiman CA, Travis RC, et al. (2011) Interactions between genetic variants and breast cancer risk factors in the breast and prostate cancer cohort consortium. J Natl Cancer Inst 103: 1252–1263.
- 24. Chen F, Chen GK, Millikan RC, John EM, Ambrosone CB, et al. (2011) Fine-mapping of breast cancer susceptibility loci characterizes genetic risk in African Americans. Hum Mol Genet 20: 4491–4503.
- 25. Han W, Woo JH, Yu JH, Lee MJ, Moon HG, et al. (2011) Common genetic variants associated with breast cancer in Korean women and differential susceptibility according to intrinsic subtype. Cancer Epidemiol Biomarkers Prev 20: 793–798.
- 26. Slattery ML, Baumgartner KB, Giuliano AR, Byers T, Herrick JS, et al. (2011) Replication of five GWAS-identified loci and breast cancer risk among Hispanic and non-Hispanic white women living in the Southwestern United States. Breast Cancer Res Treat 129: 531–539.
- 27. Butt S, Harlid S, Borgquist S, Ivarsson M, Landberg G, et al. (2012) Genetic predisposition, parity, age at first childbirth and risk for breast cancer. BMC Res Notes 5: 414.
- 28. Harlid S, Ivarsson MI, Butt S, Grzybowska E, Eyfjord JE, et al. (2012) Combined effect of low-penetrant SNPs on breast cancer risk. Br J Cancer 106: 389–396.
- 29. Kim HC, Lee JY, Sung H, Choi JY, Park SK, et al. (2012) A genome-wide association study identifies a breast cancer risk variant in ERBB4 at 2q34: results from the Seoul Breast Cancer Study. Breast Cancer Res 14: R56.
- 30. Shan J, Mahfoudh W, Dsouza SP, Hassen E, Bouaouina N, et al. (2012) Genome-Wide Association Studies (GWAS) breast cancer susceptibility loci in Arabs: susceptibility and prognostic implications in Tunisians. Breast Cancer Res Treat 135: 715–724.
- 31. Jara L, Gonzalez-Hormazabal P, Cerceno K, Di Capua GA, Reyes JM, et al. (2013) Genetic variants in FGFR2 and MAP3K1 are associated with the risk of familial and early-onset breast cancer in a South-American population. Breast Cancer Res Treat 137: 559–569.
- 32. Resler AJ, Malone KE, Johnson LG, Malkki M, Petersdorf EW, et al. (2013) Genetic variation in TLR or NFkappaB pathways and the risk of breast cancer: a case-control study. BMC Cancer 13: 219.
- 33. Zheng Y, Ogundiran TO, Falusi AG, Nathanson KL, John EM, et al. (2013) Fine mapping of breast cancer genome-wide association studies loci in women of African ancestry identifies novel susceptibility markers. Carcinogenesis 34: 1520–1528.
- 34. Mokhtari D, Myers JW, Welsh N (2008) MAPK kinase kinase-1 is essential for cytokine-induced c-Jun NH2-terminal kinase and nuclear factor-kappaB activation in human pancreatic islet cells. Diabetes 57: 1896–1904.
- 35. Dent P (2013) ERK plays the baddie (again). Cancer Biol Ther 14..
- 36. Seki E, Brenner DA, Karin M (2012) A liver full of JNK: signaling in regulation of cell function and disease pathogenesis, and clinical approaches. Gastroenterology 143: 307–320.
- 37. Fan Y, Mao R, Yang J (2013) NF-kappaB and STAT3 signaling pathways collaboratively link inflammation to cancer. Protein Cell 4: 176–185.
- 38. Peng S, Lu B, Ruan W, Zhu Y, Sheng H, et al. (2011) Genetic polymorphisms and breast cancer risk: evidence from meta-analyses, pooled analyses, and genome-wide association studies. Breast Cancer Res Treat 127: 309–324.
- 39. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K (2002) A comprehensive review of genetic association studies. Genet Med 4: 45–61.
- 40. Dick DM (2011) Gene-environment interaction in psychological traits and disorders. Annu Rev Clin Psychol 7: 383–409.
- 41. Brok J, Thorlund K, Gluud C, Wetterslev J (2008) Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol 61: 763–769.