Breast Cancer DOI 10.1007/s12282-015-0609-8
SPECIAL FEATURE
The way to the next generation molecular diagnostics
Molecular epidemiology, and possible real-world applications in breast cancer Hidemi Ito1 • Keitaro Matsuo1,2
Received: 26 February 2015 / Accepted: 2 April 2015 Ó The Japanese Breast Cancer Society 2015
Abstract Gene-environment interaction, a key idea in molecular epidemiology, has enabled the development of personalized medicine. This concept includes personalized prevention. While genome-wide association studies have identified a number of genetic susceptibility loci in breast cancer risk, however, the application of this knowledge to practical prevention is still underway. Here, we briefly review the history of molecular epidemiology and its progress in breast cancer epidemiology. We then introduce our experience with the trial combination of GWAS-identified loci and well-established lifestyle and reproductive risk factors in the risk prediction of breast cancer. Finally, we report our exploration of the cumulative risk of breast cancer based on this risk prediction model as a potential tool for individual risk communication, including genetic risk factors and gene-environment interaction with obesity. Keywords Molecular epidemiology Cancer Breast cancer Genome-wide association study GWAS
Introduction Allowing for several putatively hereditary types of cancer, cancer is generally considered a multi-factorial disease which has its genesis in a mixture of environmental factors
& Keitaro Matsuo [email protected]; [email protected] 1
Division of Epidemiology and Prevention, Aichi Cancer Center Research Institute, 1-1 Kanokoden, Chikusa-ku, Nagoya 464-8681, Japan
2
Department of Preventive Medicine, Kyushu University Faculty of Medical Sciences, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan
and host, or in other words genetic factors. Lichtenstein et al. [1] attempted to quantify the contribution of these two factors in a cohort study of twins, and reported that genetic factors accounted for 20–40 % of cancer risk in various organs while environmental factors accounted for the rest. Although this study did not consider interaction between environmental and genetic factors, its clear message is that the search for causes in cancer epidemiology must consider both. Aims of this review are to give (1) a brief overview of molecular epidemiology for cancer overall and (2) applications of molecular epidemiology specifically to breast cancer with perspective. Molecular epidemiology The question of why not all smokers get lung cancer has fascinated researchers ever since the association between smoking and lung cancer risk was first identified. Molecular epidemiology, a branch of epidemiology, sheds light on this type of question, in the case of cancer by employing molecular markers to evaluate the impact of exposure on cancer risk [2, 3]. A simple but crucial idea behind this new epidemiologic approach is ‘‘gene-environment interaction,’’ which can be rephrased as ‘‘those with genetic susceptibility to harm by certain types of environmental factors.’’ This concept has broken new ground in the field of cancer epidemiology: it has allowed the linkage of biologic phenomenon observed in animal models to human populations and has extended conventional cancer prevention through the modification of environmental factors to a more individualized model. It is no overstatement to say that the basic concept of order-made, tailor-made, or precision medicine came from this idea. Recent advances in genome analysis have allowed lower cost and larger
123
Breast Cancer
sized studies, and resulted in the explosive enrichment of genetic information, including the Human Genome Project [4], International Hap-Map Project [5], and ENCODE projects [6]. The emergence of molecular epidemiology has caused the reorganization of cancer epidemiology into a multi-disciplinary field. Gene-environment interaction Numerous examples of gene-environment interaction are available. One case we worked on, which is very familiar to Japanese, is the association between alcohol consumption, ALDH2 polymorphism, and esophageal cancer risk [7, 8]. ALDH2 is an enzyme which plays a crucial role in the oxidation of acetaldehyde, an oxidative product of ethanol and potential mutagen, into acetate. The coding gene of ALDH2, ALDH2, is polymorphic in East Asians. A single nucleotide polymorphism in Codon 504 (dbSNP id: rs671) causes substitution of an amino acid (Glu- [ Lys), which results in a functional change in enzyme activity [9–12]. Those with the Lys allele have less than 6–8 % of the enzyme activity of homozygotes of the Glu allele [11]. We have elucidated that this polymorphism has substantial impact on the amount of alcohol consumption [13] and on the cancer risk of various types of alcohol-related cancer [14–18]. In this series of studies, we reported a significant interaction between alcohol drinking and ALDH2 rs671 polymorphism. As shown in Fig. 1, the impact of heavy drinking on the risk of esophageal cancer was markedly higher in Lys allele carriers than in those with the Glu/Glu genotype. This finding elegantly demonstrates the substantial involvement of acetaldehyde, and at the same time
Fig. 1 Example of gene-environment interaction. Adjusted odds ratios (ORs) for the risk of esophageal cancer are presented separately for ALDH2 Glu/Glu and the Lys allele carriers (Matsuo et al. [14]). The OR of heavy drinking among Lys allele carriers was markedly higher than that with the Glu/Glu genotype, indicating geneenvironment interaction between alcohol drinking and ALDH2 polymorphism. This supports the hypothesis that acetaldehyde, a major substrate of ALDH2 enzyme, is an important carcinogen for esophageal cancer
123
sheds light on a newer type of prevention measure: lifestyle modification based on genetic information. Apart from ALDH2 polymorphism and alcohol consumption, several other discoveries of gene-environment interaction in the risk of cancer have been reported [19]. Inferences based on these gene-environment interactions have advanced our understanding of the mechanism of carcinogenesis. Nevertheless, not all molecular epidemiologic studies have been successful. For breast cancer, polymorphisms in the genes involved in estrogen metabolism (CYP17, COMT, CYP1A1, CYP19 etc.,) had been extensively evaluated. Thompson and Ambrosone reviewed earlier works on this topic and pointed out the inconsistency of their results [20]. They also noted other issues in these earlier studies, including ambiguity in the extent of individual variation in estrogen exposure explained by allelic variations in estrogen-metabolizing genes, and methodological problems with sample size, population selection, and approach to analysis. While acknowledging the role of smaller but conflicting studies in promoting the field, they emphasized the importance of large cohort studies. Advances in molecular epidemiology Advances in genotyping technologies have enabled the measurement of multiple polymorphisms in very large numbers of samples. Application of knowledge in linkage disequilibrium has allowed the use of haplotype tagging polymorphisms (tagSNPs) to evaluate associations. One striking advance occurred with the introduction of array technologies for the simultaneous examination of very large numbers of polymorphisms. This has brought a new paradigm to the field, the genome-wide association study (GWAS) [21]. In contrast to studies based on certain hypotheses about candidate genes, GWAS is basically a hypothesis-free way of comprehensively measuring polymorphisms in an array. The first GWAS discovery in breast cancer was the FGFR2 loci [22, 23], and many more riskassociated loci have been discovered since. Introduction of the next-generation sequencer has accelerated the identification of pathogenic loci around risk-associated loci and further clarified the mechanisms of carcinogenesis. There is no room to doubt the major role of international consortia in the advancement of GWASs. The effect size of loci identified in the many GWAS is not large, and discovery of significant loci requires tremendous statistical power. Accordingly, many international consortia have been organized for various types of cancer, including several for breast cancer alone. The Breast Cancer Association Consortium (BCAC: http://apps.ccge.medschl.cam. ac.uk/consortia/bcac//index.html) includes authors who were involved in the initial discovery of the FGFR2 loci
Breast Cancer
The ‘‘Gail model’’ was the first of an ongoing series of efforts to predict breast cancer risk. This model incorporates personal medical and reproductive history and family history of breast cancer in first-degree relatives. It has been validated in Caucasian women in the US [25, 26] and is currently under validation in other populations [27, 28]. The BOADICEA risk prediction was developed for those with a known mutation in either BRCA1 or 2 [29].
Given the progressive increase in the number of riskassociated loci identified in many GWASs, it is natural to consider the establishment of risk prediction models which apply risk loci information. Several studies have undertaken this challenge in several populations [30–32]. Here we introduce the results of our own study. We conducted a hospital-based case–control study based on the framework of HERPACC (Hospital-based Epidemiologic Research Program at Aichi Cancer Center, [30, 33, 34]). Using 23 GWAS-identified risk loci in conjunction with known epidemiologic risk factors, we identified a genetic predictor which uses seven selected variants, namely rs2981579 in FGFR2, rs3803662 in TOX3/TNRC9, rs2046210 in C6orf97, rs3817198 in LSP1, rs13281615 on 8q24, rs10931936 in CASP8, and rs4973768 in SLC4A7 [30]. As shown in Fig. 2a, when the study subjects were divided into five groups according to the number of risk alleles they have, we could identify high and super-high risk groups with 8 or more risk alleles. The receiver-operating characteristic curves of the three models (genetic predictor, lifestyle and reproductive risk, and inclusive risk) demonstrated that the inclusive risk model, which
Fig. 2 Genetic risk predictor for breast cancer using a combination of seven selected breast cancer risk-related polymorphisms (a), and receiver-operating characteristic (ROC) curves in the three risk models (b). a Adjusted odds ratios (ORs) of breast cancer by genetic risk factor, including seven selected GWAS-identified polymorphisms for breast cancer: rs2981579 in FGFR2, rs3803662 in TOX3/TNRC9, rs2046210 in C6orf97, rs3817198 in LSP1, rs13281615 on 8q24, rs10931936 in CASP8, and rs4973768 in SLC4A7. Subjects were divided into five groups according to the number of risk alleles they had: 0–3, 4–5, 6–7, 8–9, and 10 or more. ORs for the high and superhigh risk groups with 8 or more risk alleles were 3.01 (95 %
confidence interval, 1.97–4.58) and 8.69 (2.75–27.5). P for trend; 1.86 9 10-9. * P \ 1.0 9 10-3. b In the ROC, the thin straight line with an AUC of 50 % is reference. AUC of the upper black curved line, which represents the genetic risk model in addition to the epidemiological and reproductive risk factors, is 69.33 %, whereas that of the gray line, representing the epidemiological and reproductive risk model only, is 66.52 % (P = 1.3 9 10-4) and that of the lowest curved line, representing the genetic risk model only, is 59.65 %. The epidemiological and reproductive risk factors included age, menopausal status, age at first live delivery, family history of breast cancer, BMI, and regular exercise
[23]. This trend has been accelerated by cross-consortium collaborations, such as the Collaborative Oncological Gene-environment Study (COGS: http://www.cogseu.org/, [24]), a collaborative study of four consortia which introduced iCOGS array into their collaborative evaluation. This series of studies has discovered numerous risk-associated loci in breast, ovarian, and prostate cancers, all of which are hormone-related. The overlapping of genetic susceptibility loci in COGS supports the existence of pleiotropy in the carcinogenic process [24]. Application of GWAS findings in breast cancer risk prediction
123
Breast Cancer
Fig. 3 Cumulative risk (CR) of breast cancer incidence by age 80 years, stratified by genetic risk group. Genetic risk was determined by the number of risk alleles identified in women: 1–3 (low risk), 4–5 (intermediate risk), 6–7 (high risk), and 8–11 (very high). CR was calculated using the odds ratios stratified by the number of risk alleles and the distribution of the number of risk alleles in our case–control study, and the Japanese female population and 5-year age-specific
incidence rates in females in 2003 from the Monitoring of Cancer Incidence in Japan (modified methods in Refs. [36, 37]). CR by age 75 for the low, intermediate, high, and very high-risk groups was 20.6, 10.9, 8.8, and 6.5 %, respectively. In the highest risk group, cumulative risk at age 35 exceeded that for the other genetic risk groups at age 40
combines the genetic predictor and lifestyle and reproductive risk factors, had the strongest discrimination power (0.6933 of the area under the curve, Fig. 2b). These results suggest that this genetic predictor, in combination with lifestyle and reproductive risk factors, can distinguish women at high and low risk and might accordingly be useful in targeted breast cancer prevention. Expanding this risk prediction to practical-targeted breast cancer prevention requires the identification of geneenvironment interactions between genetic and environmental factors. We have reported a significant gene-environment interaction between body-mass index (BMI) and FGFR2 polymorphism for breast cancer risk among Japanese postmenopausal women [35], albeit that this interaction was not replicated in a large study in the UK [36]. Gene-environment interactions appear to be critical to the individualized feedback of risk information. Further methodologically valid studies are thus essential.
Peto et al. [37, 38], as a tool for risk communication in breast cancer prevention. One of the main aims of communicating risk information is to motivate people to modify their behavior in primary and secondary prevention. Feedback of cumulative risk appears more intuitive than that of relative risk and helps provide an accurate understanding of risk information [39]. The cumulative risk of breast cancer estimated by the genetic predictor in our study is shown in Fig. 3. Those with the highest genetic risk had a greater than 20 % cumulative risk by age 75, a markedly higher rate than in other groups. Cumulative risk at age 35 in this highest risk group exceeds that for other genetic risk groups at age 40. Thus, the age of initiation of breast cancer screening in Japanese, presently 40, might be better lowered to 35 for the highest risk group. Figure 4 shows cumulative risk with a combination of BMI and genetic risk predictor. This greater impact of a high BMI in women with greater genetic risk than in those with lower genetic risk indicates the potential application of weightloss intervention in risk communication for primary prevention. Several issues in clinical risk communication remain, such as genetic counseling, and more effective ways to communicate risk information are required, including genetic factors. Nevertheless, we consider that this is a promising approach to the issue of targeted cancer prevention.
Cumulative risk applying genetic predictor and its potential in the future breast cancer prevention The next step in risk prediction using genetic and environmental risk factors involves its application to cancer prevention. We are now attempting to estimate cumulative risk stratified by genetic risk using modified method by
123
Breast Cancer
Fig. 4 Cumulative risk (CR) of breast cancer incidence by age 80 years, stratified by genetic risk and BMI. Genetic risk was determined according to the number of risk alleles identified in women: 1–3 (low risk), 4–5 (intermediate risk), 6–7 (high risk), and 8–11 (very high). BMI was categorized into the three groups: less than 20, 20–23, more than 23. CR was calculated using the odds ratios stratified by genetic risk and
BMI and the distribution of the number of risk alleles in our case–control study and in the Japanese female population and 5-year age-specific incidence rates in females in 2003 from the Monitoring of Cancer Incidence in Japan (modified methods in Refs. [36, 37]). While the impact of BMI was low in the low or intermediate genetic risk group, it was high in the high or very high genetic risk group
Conclusion
2. Harris CC, Weston A, Willey JC, Trivers GE, Mann DL. Biochemical and molecular epidemiology of human cancer: indicators of carcinogen exposure, DNA damage, and genetic predisposition. Environ Health Perspect. 1987;75:109–19. 3. Perera FP, Poirier MC, Yuspa SH, Nakayama J, Jaretzki A, Curnen MM, et al. A pilot project in molecular cancer epidemiology: determination of benzo[a]pyrene–DNA adducts in animal and human tissues by immunoassays. Carcinogenesis. 1982;3: 1405–10. 4. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, et al. Initial sequencing and analysis of the human genome. Nature. 2001;409:860–921. 5. International HapMap C. The International HapMap Project. Nature. 2003;426:789–96. 6. Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489:57–74. 7. IARC. Alcoholic beverage consumption and ethyl carbamate (urethane). IARC monographs on the evaluation of carcinogenic risks to humans 96. Lyon: International Agency for Research on Cancer; 2010. 8. WCRF/AICR. Alcoholic drinks. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: American Institute for Cancer Research; 2007. 9. Yoshida A, Huang IY, Ikawa M. Molecular abnormality of an inactive aldehyde dehydrogenase variant commonly found in Orientals. Proc Natl Acad Sci USA. 1984;81:258–61. 10. Bosron WF, Crabb DW, Li TK. Relationship between kinetics of liver alcohol dehydrogenase and alcohol metabolism. Pharmacol Biochem Behav. 1983;18(Suppl 1):223–7. 11. Crabb DW, Edenberg HJ, Bosron WF, Li TK. Genotypes for aldehyde dehydrogenase deficiency and alcohol sensitivity. The inactive ALDH2(2) allele is dominant. J Clin Invest. 1989;83: 314–6.
In this article, we reviewed the history of the molecular epidemiology of cancer and it’s application in breast cancer epidemiology with several perspectives. Extending current knowledge into actual prevention awaits further studies from a range of aspects including methodology, individual and collaborative research environment, and ethics. Acknowledgments The authors appreciate the efforts of the many contributors to the HERPACC study. This study was supported by Grants-in-Aid for Scientific Research on Priority Areas and Grant-inAid for Scientific Research (A) and (C) from the Ministry of Education, Science, Sports, Culture and Technology of Japan; by a Grant-in-Aid for the Third Term Comprehensive 10-year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare of Japan; and by research grant from Takeda Science Foundation. These grantors were not involved in the study design, subject enrollment, study analysis or interpretation, or submission of the manuscript for this study. Conflict of interest
Authors declare no conflict of interest.
References 1. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M, et al. Environmental and heritable factors in the causation of cancer–analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med. 2000;343:78–85.
123
Breast Cancer 12. Li Y, Zhang D, Jin W, Shao C, Yan P, Xu C, et al. Mitochondrial aldehyde dehydrogenase-2 (ALDH2) Glu504Lys polymorphism contributes to the variation in efficacy of sublingual nitroglycerin. J Clin Invest. 2006;116:506–11. 13. Matsuo K, Wakai K, Hirose K, Ito H, Saito T, Tajima K. Alcohol dehydrogenase 2 His47Arg polymorphism influences drinking habit independently of aldehyde dehydrogenase 2 Glu487Lys polymorphism: analysis of 2,299 Japanese subjects. Cancer Epidemiol Biomarkers Prev. 2006;15:1009–13. 14. Matsuo K, Hamajima N, Shinoda M, Hatooka S, Inoue M, Takezaki T, et al. Gene-environment interaction between an aldehyde dehydrogenase-2 (ALDH2) polymorphism and alcohol consumption for the risk of esophageal cancer. Carcinogenesis. 2001;22:913–6. 15. Yang C, Wang H, Wang Z, Du H, Tao D, Mu X, et al. Risk factors for esophageal cancer: a case-control study in Southwestern China. Asian Pac J Cancer Prev. 2005;6:48–53. 16. Hiraki A, Matsuo K, Wakai K, Suzuki T, Hasegawa Y, Tajima K. Gene-gene and gene-environment interactions between alcohol drinking habit and polymorphisms in alcohol-metabolizing enzyme genes and the risk of head and neck cancer in Japan. Cancer Sci. 2007;98:1087–91. 17. Oze I, Matsuo K, Hosono S, Ito H, Kawase T, Watanabe M, et al. Comparison between self-reported facial flushing after alcohol consumption and ALDH2 Glu504Lys polymorphism for risk of upper aerodigestive tract cancer in a Japanese population. Cancer Sci. 2010;101:1875–80. 18. Matsuo K, Oze I, Hosono S, Ito H, Watanabe M, Ishioka K, et al. The aldehyde dehydrogenase 2 (ALDH2) Glu504Lys polymorphism interacts with alcohol drinking in the risk of stomach cancer. Carcinogenesis. 2013;34:1510–5. 19. Spitz MR, Bondy ML. The evolving discipline of molecular epidemiology of cancer. Carcinogenesis. 2010;31:127–34. 20. Thompson PA, Ambrosone C. Molecular epidemiology of genetic polymorphisms in estrogen metabolizing enzymes in human breast cancer. J Natl Cancer Inst Monogr. 2000;27:125–34. 21. Manolio TA. Genomewide association studies and assessment of the risk of disease. N Engl J Med. 2010;363:166–76. 22. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, Hankinson SE, et al. A genome-wide association study identifies alleles in FGFR2 associated with risk of sporadic postmenopausal breast cancer. Nat Genet. 2007;39:870–4. 23. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, Ballinger DG, et al. Genome-wide association study identifies novel breast cancer susceptibility loci. Nature. 2007;447: 1087–93. 24. Sakoda LC, Jorgenson E, Witte JS. Turning of COGS moves forward findings for hormonally mediated cancers. Nat Genet. 2013;45:345–8. 25. Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81:1879–86. 26. Costantino JP, Gail MH, Pee D, Anderson S, Redmond CK, Benichou J, et al. Validation studies for models projecting the
123
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
risk of invasive and total breast cancer incidence. J Natl Cancer Inst. 1999;91:1541–8. Gail MH, Costantino JP, Pee D, Bondy M, Newman L, Selvan M, et al. Projecting individualized absolute invasive breast cancer risk in African American women. J Natl Cancer Inst. 2007;99: 1782–92. Matsuno RK, Costantino JP, Ziegler RG, Anderson GL, Li H, Pee D, et al. Projecting individualized absolute invasive breast cancer risk in Asian and Pacific Islander American women. J Natl Cancer Inst. 2011;103:951–61. Lee AJ, Cunningham AP, Kuchenbaecker KB, Mavaddat N, Easton DF, Antoniou AC, et al. BOADICEA breast cancer risk prediction model: updates to cancer incidences, tumour pathology and web interface. Br J Cancer. 2014;110:535–45. Sueta A, Ito H, Kawase T, Hirose K, Hosono S, Yatabe Y, et al. A genetic risk predictor for breast cancer using a combination of low-penetrance polymorphisms in a Japanese population. Breast Cancer Res Treat. 2012;132:711–21. Wacholder S, Hartge P, Prentice R, Garcia-Closas M, Feigelson HS, Diver WR, et al. Performance of common genetic variants in breast-cancer risk models. N Engl J Med. 2010;362:986–93. Zheng W, Wen W, Gao YT, Shyr Y, Zheng Y, Long J, et al. Genetic and clinical predictors for breast cancer risk assessment and stratification among Chinese women. J Natl Cancer Inst. 2010;102:972–81. Tajima K, Hirose K, Inoue M, Takezaki T, Hamajima N, Kuroishi T. A Model of practical cancer prevention for out-patients visiting a hospital: the Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC). Asian Pac J Cancer Prev. 2000;1:35–47. Hamajima N, Matsuo K, Saito T, Hirose K, Inoue M, Takezaki T, et al. Gene-environment interactions and polymorphism studies of cancer risk in the Hospital-based Epidemiologic Research Program at Aichi Cancer Center II (HERPACC-II). Asian Pac J Cancer Prev. 2001;2:99–107. Kawase T, Matsuo K, Suzuki T, Hiraki A, Watanabe M, Iwata H, et al. FGFR2 intronic polymorphisms interact with reproductive risk factors of breast cancer: results of a case control study in Japan. Int J Cancer. 2009;125:1946–52. Travis RC, Reeves GK, Green J, Bull D, Tipper SJ, Baker K, et al. Gene-environment interactions in 7610 women with breast cancer: prospective evidence from the Million Women Study. Lancet. 2010;375:2143–51. Brennan P, Crispo A, Zaridze D, Szeszenia-Dabrowska N, Rudnai P, Lissowska J, et al. High cumulative risk of lung cancer death among smokers and nonsmokers in Central and Eastern Europe. Am J Epidemiol. 2006;164:1233–41. Peto R, Darby S, Deo H, Silcocks P, Whitley E, Doll R. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-control studies. BMJ. 2000;321:323–9. Zipkin DA, Umscheid CA, Keating NL, Allen E, Aung K, Beyth R, et al. Evidence-based risk communication: a systematic review. Ann Intern Med. 2014;161:270–80.
SPECIAL FEATURE
The way to the next generation molecular diagnostics
Molecular epidemiology, and possible real-world applications in breast cancer Hidemi Ito1 • Keitaro Matsuo1,2
Received: 26 February 2015 / Accepted: 2 April 2015 Ó The Japanese Breast Cancer Society 2015
Abstract Gene-environment interaction, a key idea in molecular epidemiology, has enabled the development of personalized medicine. This concept includes personalized prevention. While genome-wide association studies have identified a number of genetic susceptibility loci in breast cancer risk, however, the application of this knowledge to practical prevention is still underway. Here, we briefly review the history of molecular epidemiology and its progress in breast cancer epidemiology. We then introduce our experience with the trial combination of GWAS-identified loci and well-established lifestyle and reproductive risk factors in the risk prediction of breast cancer. Finally, we report our exploration of the cumulative risk of breast cancer based on this risk prediction model as a potential tool for individual risk communication, including genetic risk factors and gene-environment interaction with obesity. Keywords Molecular epidemiology Cancer Breast cancer Genome-wide association study GWAS
Introduction Allowing for several putatively hereditary types of cancer, cancer is generally considered a multi-factorial disease which has its genesis in a mixture of environmental factors
& Keitaro Matsuo [email protected]; [email protected] 1
Division of Epidemiology and Prevention, Aichi Cancer Center Research Institute, 1-1 Kanokoden, Chikusa-ku, Nagoya 464-8681, Japan
2
Department of Preventive Medicine, Kyushu University Faculty of Medical Sciences, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan
and host, or in other words genetic factors. Lichtenstein et al. [1] attempted to quantify the contribution of these two factors in a cohort study of twins, and reported that genetic factors accounted for 20–40 % of cancer risk in various organs while environmental factors accounted for the rest. Although this study did not consider interaction between environmental and genetic factors, its clear message is that the search for causes in cancer epidemiology must consider both. Aims of this review are to give (1) a brief overview of molecular epidemiology for cancer overall and (2) applications of molecular epidemiology specifically to breast cancer with perspective. Molecular epidemiology The question of why not all smokers get lung cancer has fascinated researchers ever since the association between smoking and lung cancer risk was first identified. Molecular epidemiology, a branch of epidemiology, sheds light on this type of question, in the case of cancer by employing molecular markers to evaluate the impact of exposure on cancer risk [2, 3]. A simple but crucial idea behind this new epidemiologic approach is ‘‘gene-environment interaction,’’ which can be rephrased as ‘‘those with genetic susceptibility to harm by certain types of environmental factors.’’ This concept has broken new ground in the field of cancer epidemiology: it has allowed the linkage of biologic phenomenon observed in animal models to human populations and has extended conventional cancer prevention through the modification of environmental factors to a more individualized model. It is no overstatement to say that the basic concept of order-made, tailor-made, or precision medicine came from this idea. Recent advances in genome analysis have allowed lower cost and larger
123
Breast Cancer
sized studies, and resulted in the explosive enrichment of genetic information, including the Human Genome Project [4], International Hap-Map Project [5], and ENCODE projects [6]. The emergence of molecular epidemiology has caused the reorganization of cancer epidemiology into a multi-disciplinary field. Gene-environment interaction Numerous examples of gene-environment interaction are available. One case we worked on, which is very familiar to Japanese, is the association between alcohol consumption, ALDH2 polymorphism, and esophageal cancer risk [7, 8]. ALDH2 is an enzyme which plays a crucial role in the oxidation of acetaldehyde, an oxidative product of ethanol and potential mutagen, into acetate. The coding gene of ALDH2, ALDH2, is polymorphic in East Asians. A single nucleotide polymorphism in Codon 504 (dbSNP id: rs671) causes substitution of an amino acid (Glu- [ Lys), which results in a functional change in enzyme activity [9–12]. Those with the Lys allele have less than 6–8 % of the enzyme activity of homozygotes of the Glu allele [11]. We have elucidated that this polymorphism has substantial impact on the amount of alcohol consumption [13] and on the cancer risk of various types of alcohol-related cancer [14–18]. In this series of studies, we reported a significant interaction between alcohol drinking and ALDH2 rs671 polymorphism. As shown in Fig. 1, the impact of heavy drinking on the risk of esophageal cancer was markedly higher in Lys allele carriers than in those with the Glu/Glu genotype. This finding elegantly demonstrates the substantial involvement of acetaldehyde, and at the same time
Fig. 1 Example of gene-environment interaction. Adjusted odds ratios (ORs) for the risk of esophageal cancer are presented separately for ALDH2 Glu/Glu and the Lys allele carriers (Matsuo et al. [14]). The OR of heavy drinking among Lys allele carriers was markedly higher than that with the Glu/Glu genotype, indicating geneenvironment interaction between alcohol drinking and ALDH2 polymorphism. This supports the hypothesis that acetaldehyde, a major substrate of ALDH2 enzyme, is an important carcinogen for esophageal cancer
123
sheds light on a newer type of prevention measure: lifestyle modification based on genetic information. Apart from ALDH2 polymorphism and alcohol consumption, several other discoveries of gene-environment interaction in the risk of cancer have been reported [19]. Inferences based on these gene-environment interactions have advanced our understanding of the mechanism of carcinogenesis. Nevertheless, not all molecular epidemiologic studies have been successful. For breast cancer, polymorphisms in the genes involved in estrogen metabolism (CYP17, COMT, CYP1A1, CYP19 etc.,) had been extensively evaluated. Thompson and Ambrosone reviewed earlier works on this topic and pointed out the inconsistency of their results [20]. They also noted other issues in these earlier studies, including ambiguity in the extent of individual variation in estrogen exposure explained by allelic variations in estrogen-metabolizing genes, and methodological problems with sample size, population selection, and approach to analysis. While acknowledging the role of smaller but conflicting studies in promoting the field, they emphasized the importance of large cohort studies. Advances in molecular epidemiology Advances in genotyping technologies have enabled the measurement of multiple polymorphisms in very large numbers of samples. Application of knowledge in linkage disequilibrium has allowed the use of haplotype tagging polymorphisms (tagSNPs) to evaluate associations. One striking advance occurred with the introduction of array technologies for the simultaneous examination of very large numbers of polymorphisms. This has brought a new paradigm to the field, the genome-wide association study (GWAS) [21]. In contrast to studies based on certain hypotheses about candidate genes, GWAS is basically a hypothesis-free way of comprehensively measuring polymorphisms in an array. The first GWAS discovery in breast cancer was the FGFR2 loci [22, 23], and many more riskassociated loci have been discovered since. Introduction of the next-generation sequencer has accelerated the identification of pathogenic loci around risk-associated loci and further clarified the mechanisms of carcinogenesis. There is no room to doubt the major role of international consortia in the advancement of GWASs. The effect size of loci identified in the many GWAS is not large, and discovery of significant loci requires tremendous statistical power. Accordingly, many international consortia have been organized for various types of cancer, including several for breast cancer alone. The Breast Cancer Association Consortium (BCAC: http://apps.ccge.medschl.cam. ac.uk/consortia/bcac//index.html) includes authors who were involved in the initial discovery of the FGFR2 loci
Breast Cancer
The ‘‘Gail model’’ was the first of an ongoing series of efforts to predict breast cancer risk. This model incorporates personal medical and reproductive history and family history of breast cancer in first-degree relatives. It has been validated in Caucasian women in the US [25, 26] and is currently under validation in other populations [27, 28]. The BOADICEA risk prediction was developed for those with a known mutation in either BRCA1 or 2 [29].
Given the progressive increase in the number of riskassociated loci identified in many GWASs, it is natural to consider the establishment of risk prediction models which apply risk loci information. Several studies have undertaken this challenge in several populations [30–32]. Here we introduce the results of our own study. We conducted a hospital-based case–control study based on the framework of HERPACC (Hospital-based Epidemiologic Research Program at Aichi Cancer Center, [30, 33, 34]). Using 23 GWAS-identified risk loci in conjunction with known epidemiologic risk factors, we identified a genetic predictor which uses seven selected variants, namely rs2981579 in FGFR2, rs3803662 in TOX3/TNRC9, rs2046210 in C6orf97, rs3817198 in LSP1, rs13281615 on 8q24, rs10931936 in CASP8, and rs4973768 in SLC4A7 [30]. As shown in Fig. 2a, when the study subjects were divided into five groups according to the number of risk alleles they have, we could identify high and super-high risk groups with 8 or more risk alleles. The receiver-operating characteristic curves of the three models (genetic predictor, lifestyle and reproductive risk, and inclusive risk) demonstrated that the inclusive risk model, which
Fig. 2 Genetic risk predictor for breast cancer using a combination of seven selected breast cancer risk-related polymorphisms (a), and receiver-operating characteristic (ROC) curves in the three risk models (b). a Adjusted odds ratios (ORs) of breast cancer by genetic risk factor, including seven selected GWAS-identified polymorphisms for breast cancer: rs2981579 in FGFR2, rs3803662 in TOX3/TNRC9, rs2046210 in C6orf97, rs3817198 in LSP1, rs13281615 on 8q24, rs10931936 in CASP8, and rs4973768 in SLC4A7. Subjects were divided into five groups according to the number of risk alleles they had: 0–3, 4–5, 6–7, 8–9, and 10 or more. ORs for the high and superhigh risk groups with 8 or more risk alleles were 3.01 (95 %
confidence interval, 1.97–4.58) and 8.69 (2.75–27.5). P for trend; 1.86 9 10-9. * P \ 1.0 9 10-3. b In the ROC, the thin straight line with an AUC of 50 % is reference. AUC of the upper black curved line, which represents the genetic risk model in addition to the epidemiological and reproductive risk factors, is 69.33 %, whereas that of the gray line, representing the epidemiological and reproductive risk model only, is 66.52 % (P = 1.3 9 10-4) and that of the lowest curved line, representing the genetic risk model only, is 59.65 %. The epidemiological and reproductive risk factors included age, menopausal status, age at first live delivery, family history of breast cancer, BMI, and regular exercise
[23]. This trend has been accelerated by cross-consortium collaborations, such as the Collaborative Oncological Gene-environment Study (COGS: http://www.cogseu.org/, [24]), a collaborative study of four consortia which introduced iCOGS array into their collaborative evaluation. This series of studies has discovered numerous risk-associated loci in breast, ovarian, and prostate cancers, all of which are hormone-related. The overlapping of genetic susceptibility loci in COGS supports the existence of pleiotropy in the carcinogenic process [24]. Application of GWAS findings in breast cancer risk prediction
123
Breast Cancer
Fig. 3 Cumulative risk (CR) of breast cancer incidence by age 80 years, stratified by genetic risk group. Genetic risk was determined by the number of risk alleles identified in women: 1–3 (low risk), 4–5 (intermediate risk), 6–7 (high risk), and 8–11 (very high). CR was calculated using the odds ratios stratified by the number of risk alleles and the distribution of the number of risk alleles in our case–control study, and the Japanese female population and 5-year age-specific
incidence rates in females in 2003 from the Monitoring of Cancer Incidence in Japan (modified methods in Refs. [36, 37]). CR by age 75 for the low, intermediate, high, and very high-risk groups was 20.6, 10.9, 8.8, and 6.5 %, respectively. In the highest risk group, cumulative risk at age 35 exceeded that for the other genetic risk groups at age 40
combines the genetic predictor and lifestyle and reproductive risk factors, had the strongest discrimination power (0.6933 of the area under the curve, Fig. 2b). These results suggest that this genetic predictor, in combination with lifestyle and reproductive risk factors, can distinguish women at high and low risk and might accordingly be useful in targeted breast cancer prevention. Expanding this risk prediction to practical-targeted breast cancer prevention requires the identification of geneenvironment interactions between genetic and environmental factors. We have reported a significant gene-environment interaction between body-mass index (BMI) and FGFR2 polymorphism for breast cancer risk among Japanese postmenopausal women [35], albeit that this interaction was not replicated in a large study in the UK [36]. Gene-environment interactions appear to be critical to the individualized feedback of risk information. Further methodologically valid studies are thus essential.
Peto et al. [37, 38], as a tool for risk communication in breast cancer prevention. One of the main aims of communicating risk information is to motivate people to modify their behavior in primary and secondary prevention. Feedback of cumulative risk appears more intuitive than that of relative risk and helps provide an accurate understanding of risk information [39]. The cumulative risk of breast cancer estimated by the genetic predictor in our study is shown in Fig. 3. Those with the highest genetic risk had a greater than 20 % cumulative risk by age 75, a markedly higher rate than in other groups. Cumulative risk at age 35 in this highest risk group exceeds that for other genetic risk groups at age 40. Thus, the age of initiation of breast cancer screening in Japanese, presently 40, might be better lowered to 35 for the highest risk group. Figure 4 shows cumulative risk with a combination of BMI and genetic risk predictor. This greater impact of a high BMI in women with greater genetic risk than in those with lower genetic risk indicates the potential application of weightloss intervention in risk communication for primary prevention. Several issues in clinical risk communication remain, such as genetic counseling, and more effective ways to communicate risk information are required, including genetic factors. Nevertheless, we consider that this is a promising approach to the issue of targeted cancer prevention.
Cumulative risk applying genetic predictor and its potential in the future breast cancer prevention The next step in risk prediction using genetic and environmental risk factors involves its application to cancer prevention. We are now attempting to estimate cumulative risk stratified by genetic risk using modified method by
123
Breast Cancer
Fig. 4 Cumulative risk (CR) of breast cancer incidence by age 80 years, stratified by genetic risk and BMI. Genetic risk was determined according to the number of risk alleles identified in women: 1–3 (low risk), 4–5 (intermediate risk), 6–7 (high risk), and 8–11 (very high). BMI was categorized into the three groups: less than 20, 20–23, more than 23. CR was calculated using the odds ratios stratified by genetic risk and
BMI and the distribution of the number of risk alleles in our case–control study and in the Japanese female population and 5-year age-specific incidence rates in females in 2003 from the Monitoring of Cancer Incidence in Japan (modified methods in Refs. [36, 37]). While the impact of BMI was low in the low or intermediate genetic risk group, it was high in the high or very high genetic risk group
Conclusion
2. Harris CC, Weston A, Willey JC, Trivers GE, Mann DL. Biochemical and molecular epidemiology of human cancer: indicators of carcinogen exposure, DNA damage, and genetic predisposition. Environ Health Perspect. 1987;75:109–19. 3. Perera FP, Poirier MC, Yuspa SH, Nakayama J, Jaretzki A, Curnen MM, et al. A pilot project in molecular cancer epidemiology: determination of benzo[a]pyrene–DNA adducts in animal and human tissues by immunoassays. Carcinogenesis. 1982;3: 1405–10. 4. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, et al. Initial sequencing and analysis of the human genome. Nature. 2001;409:860–921. 5. International HapMap C. The International HapMap Project. Nature. 2003;426:789–96. 6. Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489:57–74. 7. IARC. Alcoholic beverage consumption and ethyl carbamate (urethane). IARC monographs on the evaluation of carcinogenic risks to humans 96. Lyon: International Agency for Research on Cancer; 2010. 8. WCRF/AICR. Alcoholic drinks. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: American Institute for Cancer Research; 2007. 9. Yoshida A, Huang IY, Ikawa M. Molecular abnormality of an inactive aldehyde dehydrogenase variant commonly found in Orientals. Proc Natl Acad Sci USA. 1984;81:258–61. 10. Bosron WF, Crabb DW, Li TK. Relationship between kinetics of liver alcohol dehydrogenase and alcohol metabolism. Pharmacol Biochem Behav. 1983;18(Suppl 1):223–7. 11. Crabb DW, Edenberg HJ, Bosron WF, Li TK. Genotypes for aldehyde dehydrogenase deficiency and alcohol sensitivity. The inactive ALDH2(2) allele is dominant. J Clin Invest. 1989;83: 314–6.
In this article, we reviewed the history of the molecular epidemiology of cancer and it’s application in breast cancer epidemiology with several perspectives. Extending current knowledge into actual prevention awaits further studies from a range of aspects including methodology, individual and collaborative research environment, and ethics. Acknowledgments The authors appreciate the efforts of the many contributors to the HERPACC study. This study was supported by Grants-in-Aid for Scientific Research on Priority Areas and Grant-inAid for Scientific Research (A) and (C) from the Ministry of Education, Science, Sports, Culture and Technology of Japan; by a Grant-in-Aid for the Third Term Comprehensive 10-year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare of Japan; and by research grant from Takeda Science Foundation. These grantors were not involved in the study design, subject enrollment, study analysis or interpretation, or submission of the manuscript for this study. Conflict of interest
Authors declare no conflict of interest.
References 1. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M, et al. Environmental and heritable factors in the causation of cancer–analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med. 2000;343:78–85.
123
Breast Cancer 12. Li Y, Zhang D, Jin W, Shao C, Yan P, Xu C, et al. Mitochondrial aldehyde dehydrogenase-2 (ALDH2) Glu504Lys polymorphism contributes to the variation in efficacy of sublingual nitroglycerin. J Clin Invest. 2006;116:506–11. 13. Matsuo K, Wakai K, Hirose K, Ito H, Saito T, Tajima K. Alcohol dehydrogenase 2 His47Arg polymorphism influences drinking habit independently of aldehyde dehydrogenase 2 Glu487Lys polymorphism: analysis of 2,299 Japanese subjects. Cancer Epidemiol Biomarkers Prev. 2006;15:1009–13. 14. Matsuo K, Hamajima N, Shinoda M, Hatooka S, Inoue M, Takezaki T, et al. Gene-environment interaction between an aldehyde dehydrogenase-2 (ALDH2) polymorphism and alcohol consumption for the risk of esophageal cancer. Carcinogenesis. 2001;22:913–6. 15. Yang C, Wang H, Wang Z, Du H, Tao D, Mu X, et al. Risk factors for esophageal cancer: a case-control study in Southwestern China. Asian Pac J Cancer Prev. 2005;6:48–53. 16. Hiraki A, Matsuo K, Wakai K, Suzuki T, Hasegawa Y, Tajima K. Gene-gene and gene-environment interactions between alcohol drinking habit and polymorphisms in alcohol-metabolizing enzyme genes and the risk of head and neck cancer in Japan. Cancer Sci. 2007;98:1087–91. 17. Oze I, Matsuo K, Hosono S, Ito H, Kawase T, Watanabe M, et al. Comparison between self-reported facial flushing after alcohol consumption and ALDH2 Glu504Lys polymorphism for risk of upper aerodigestive tract cancer in a Japanese population. Cancer Sci. 2010;101:1875–80. 18. Matsuo K, Oze I, Hosono S, Ito H, Watanabe M, Ishioka K, et al. The aldehyde dehydrogenase 2 (ALDH2) Glu504Lys polymorphism interacts with alcohol drinking in the risk of stomach cancer. Carcinogenesis. 2013;34:1510–5. 19. Spitz MR, Bondy ML. The evolving discipline of molecular epidemiology of cancer. Carcinogenesis. 2010;31:127–34. 20. Thompson PA, Ambrosone C. Molecular epidemiology of genetic polymorphisms in estrogen metabolizing enzymes in human breast cancer. J Natl Cancer Inst Monogr. 2000;27:125–34. 21. Manolio TA. Genomewide association studies and assessment of the risk of disease. N Engl J Med. 2010;363:166–76. 22. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, Hankinson SE, et al. A genome-wide association study identifies alleles in FGFR2 associated with risk of sporadic postmenopausal breast cancer. Nat Genet. 2007;39:870–4. 23. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, Ballinger DG, et al. Genome-wide association study identifies novel breast cancer susceptibility loci. Nature. 2007;447: 1087–93. 24. Sakoda LC, Jorgenson E, Witte JS. Turning of COGS moves forward findings for hormonally mediated cancers. Nat Genet. 2013;45:345–8. 25. Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81:1879–86. 26. Costantino JP, Gail MH, Pee D, Anderson S, Redmond CK, Benichou J, et al. Validation studies for models projecting the
123
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
risk of invasive and total breast cancer incidence. J Natl Cancer Inst. 1999;91:1541–8. Gail MH, Costantino JP, Pee D, Bondy M, Newman L, Selvan M, et al. Projecting individualized absolute invasive breast cancer risk in African American women. J Natl Cancer Inst. 2007;99: 1782–92. Matsuno RK, Costantino JP, Ziegler RG, Anderson GL, Li H, Pee D, et al. Projecting individualized absolute invasive breast cancer risk in Asian and Pacific Islander American women. J Natl Cancer Inst. 2011;103:951–61. Lee AJ, Cunningham AP, Kuchenbaecker KB, Mavaddat N, Easton DF, Antoniou AC, et al. BOADICEA breast cancer risk prediction model: updates to cancer incidences, tumour pathology and web interface. Br J Cancer. 2014;110:535–45. Sueta A, Ito H, Kawase T, Hirose K, Hosono S, Yatabe Y, et al. A genetic risk predictor for breast cancer using a combination of low-penetrance polymorphisms in a Japanese population. Breast Cancer Res Treat. 2012;132:711–21. Wacholder S, Hartge P, Prentice R, Garcia-Closas M, Feigelson HS, Diver WR, et al. Performance of common genetic variants in breast-cancer risk models. N Engl J Med. 2010;362:986–93. Zheng W, Wen W, Gao YT, Shyr Y, Zheng Y, Long J, et al. Genetic and clinical predictors for breast cancer risk assessment and stratification among Chinese women. J Natl Cancer Inst. 2010;102:972–81. Tajima K, Hirose K, Inoue M, Takezaki T, Hamajima N, Kuroishi T. A Model of practical cancer prevention for out-patients visiting a hospital: the Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC). Asian Pac J Cancer Prev. 2000;1:35–47. Hamajima N, Matsuo K, Saito T, Hirose K, Inoue M, Takezaki T, et al. Gene-environment interactions and polymorphism studies of cancer risk in the Hospital-based Epidemiologic Research Program at Aichi Cancer Center II (HERPACC-II). Asian Pac J Cancer Prev. 2001;2:99–107. Kawase T, Matsuo K, Suzuki T, Hiraki A, Watanabe M, Iwata H, et al. FGFR2 intronic polymorphisms interact with reproductive risk factors of breast cancer: results of a case control study in Japan. Int J Cancer. 2009;125:1946–52. Travis RC, Reeves GK, Green J, Bull D, Tipper SJ, Baker K, et al. Gene-environment interactions in 7610 women with breast cancer: prospective evidence from the Million Women Study. Lancet. 2010;375:2143–51. Brennan P, Crispo A, Zaridze D, Szeszenia-Dabrowska N, Rudnai P, Lissowska J, et al. High cumulative risk of lung cancer death among smokers and nonsmokers in Central and Eastern Europe. Am J Epidemiol. 2006;164:1233–41. Peto R, Darby S, Deo H, Silcocks P, Whitley E, Doll R. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-control studies. BMJ. 2000;321:323–9. Zipkin DA, Umscheid CA, Keating NL, Allen E, Aung K, Beyth R, et al. Evidence-based risk communication: a systematic review. Ann Intern Med. 2014;161:270–80.
