Easton, Douglas F.

Abstract Background: Prospective validation of breast cancer risk models integrating classical risk factors and genetic variants is required for risk-stratified prevention and screening strategies. The objective of this study was to validate a breast cancer risk model integrating classical risk factors and a 313-variant polygenic risk score (PRS) in multiple prospective cohort studies, and to project five-year risk of breast cancer in six different countries. Methods: The study population included 7,529 cases and 230,103 controls from 14 prospective cohort studies in Australia, Germany, the Netherlands, Sweden, UK, and USA. We used the Individualized Coherent Absolute Risk Estimator (iCARE) tool for risk model building, validation, and risk projection. Expected five-year risk of invasive or in situ breast cancer was compared to observed risk, overall and within deciles of expected risk using goodness of fit statistics. We evaluated calibration of the relative risk through meta-analysis across cohorts, and of the absolute risk within each cohort. Model discrimination was evaluated using the area under the curve (AUC), and percentages of women crossing risk thresholds. Projections of five-year risk distributions were estimated for women of European ancestry aged 50-70 years in the general populations of these six countries. Results: Analysis showed overall good calibration of the integrated iCARE-based model relative risk for both women younger than 50 years (χ2=14.9, P=0.09) and aged 50 years or older (χ2=14.6, P=0.10), with a small overestimation of risk for women in the highest decile of expected risk (RR = 3.5 expected vs 2.3 (95% CI 1.6 to 3.2) observed for women <50 years; and 2.8 expected vs 2.3 (95% CI 2.0 to 2.7) observed for women 50+ years). The age-adjusted AUCs for the integrated model were 63.1 (95% CI 60.9 to 65.3) and 62.9 (95% CI 61.8 to 64.0), for the two age groups respectively. The calibration of absolute risk showed substantial variation across cohorts, particularly for the older group, but had no systematic bias. Model based projections in the general populations showed that compared to the population average, women in the 1st and 99th percentiles of the integrated risk score had relative risks 0.19 and 3.56 respectively. The proportion of women of European ancestry aged 50-70 years with a five-year risk greater than 3% (threshold for consideration of risk-lowering drugs by U.S. Preventive Services Task Force) ranged from 7.1% in Germany to 18.2% in the US, which corresponds to ~5.5 million women in the US. Conclusions: Five-year risk predictions from a model with classical risk factors and PRS are well calibrated and provide substantial risk stratification across multiple cohorts in six different countries. Further studies are needed to evaluate the clinical utility of the validated model for risk stratified screening and prevention of breast cancer. Citation Format: Parichoy Pal Choudhury, Amber Wilcox, Chi Gao, Brian Carter, Anika Husing, Mark Brook, Mikael Eriksson, Kara Martin, Chris Scott, Min Shi, Thomas Ahearn, Michael Jones, Nick Orr, Minouk Schoemaker, Kamila Czene, Jenny Chang-Claude, Jacques Simard, Doug Easton, Marjanka K. Schmidt, Dale Sandler, Clarice R. Weinberg, Celine Vachon, Roger Milne, Per Hall, Anthony Swerdlow, Rudolph Kaaks, Myrto Barrdahl, Mia Gaudet, Antonis Antoniou, Peter Kraft, Montserrat Garcia-Closas, Nilanjan Chatterjee. Validation of breast cancer risk model incorporating classical risk factors and polygenic risk scores in 14 prospective cohort studies in 6 countries [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 962.

AbstractFemale Hodgkin lymphoma (HL) patients treated with chest radiotherapy (RT) have a very high risk of breast cancer. The contribution of genetic factors to this risk is unclear. We therefore examined 211 155 germline single-nucleotide polymorphisms (SNPs) for gene-radiation interaction on breast cancer risk in a case-only analysis including 327 breast cancer patients after chest RT for HL and 4671 first primary breast cancer patients. Nine SNPs showed statistically significant interaction with RT on breast cancer risk (false discovery rate, <20%), of which 1 SNP in the PVT1 oncogene attained the Bonferroni threshold for statistical significance. A polygenic risk score (PRS) composed of these SNPs (RT-interaction-PRS) and a previously published breast cancer PRS (BC-PRS) derived in the general population were evaluated in a case-control analysis comprising the 327 chest-irradiated HL patients with breast cancer and 491 chest-irradiated HL patients without breast cancer. Patients in the highest tertile of the RT-interaction-PRS had a 1.6-fold higher breast cancer risk than those in the lowest tertile. Remarkably, we observed a fourfold increased RT-induced breast cancer risk in the highest compared with the lowest decile of the BC-PRS. On a continuous scale, breast cancer risk increased 1.4-fold per standard deviation of the BC-PRS, similar to the effect size found in the general population. This study demonstrates that genetic factors influence breast cancer risk after chest RT for HL. Given the high absolute breast cancer risk in radiation-exposed women, these results can have important implications for the management of current HL survivors and future patients.

Abstract Ovarian cancer is the most lethal gynecologic malignancy and the sixth most common cause of death from all cancer in women, with an estimated 140,000 deaths per year worldwide. Endometrial cancer (cancer of the uterine lining) is the most commonly diagnosed gynecologic cancer, ranking sixth in incident cancer in women. Ovarian and endometrial cancer share many epidemiologic, histopathologic, and tumor genetic characteristics. Meta-analyses of genome-wide association study (GWAS) datasets across etiologically related diseases have successfully been used to increase statistical power and identify novel genetic risk regions. We hypothesised that joint meta-analysis of ovarian and endometrial cancer GWAS datasets would identify novel genetic loci for both cancers. Following quality control, summary statistics for >11 million genetic variants were available from the largest genome-wide association studies performed by the Ovarian Cancer Association Consortium (OCAC, Phelan et al., Nature Genetics 2017) and the Endometrial Cancer Association Consortium (ECAC, unpublished). A total of 35,312 cancer cases from all histologic subtypes (22,406 ovarian and 12,906 endometrial cancer cases) and 149,920 controls were included in the analysis. Summary statistics were combined using an inverse-variance, fixed-effects model in METAL and identified ten loci at genome-wide significance (P < 5 x 10-8), of which one locus at 14q23.3 (rs4072776 combined OR 0.94; 95% CI 0.92-0.96; P = 8.0 x 10-9) had not been previously associated with the risk of either cancer. Integration with chromatin conformation data (RNAPolII HiChIP) from the Ishikawa endometrial cancer cell line suggests interaction between the risk signal and the promoter of FUT8. FUT8 encodes a fucosyltransferase and its expression has been implicated in breast, stomach, and melanoma and lung cancer. Interestingly, FUT8 is a downstream target regulated by the loss of PAX2 and mutated p53, which is the earliest known molecular aberration in the progression of the fallopian tissue epithelial to serous ovarian cancer. Further functional evaluation of the 14q23.3 risk region is required to determine the regulatory effect of genetic risk variants on FUT8. Future meta-analyses of datasets, stratifying results by histologic subtypes, will be undertaken. Citation Format: Tracy A. O'Mara, Dylan M. Glubb, Daniel D. Buchanan, Diether Lambrechts, Per Hall, Emma Tham, Jone Trovik, Ellen L. Goode, Peter Fasching, Thilo Dörk, Rodney J. Scott, Paul L. Auer, Roger L. Milne, Graham G. Giles, John Perry, Immaculata de Vivo, Ian Tomlinson, Douglas F. Easton, Deborah J. Thompson, Amanda B. Spurdle, E2C2, BCAC, OCAC, ECAC. Joint genome-wide association study of endometrial cancer and ovarian cancer identifies a novel genetic risk region at 14q23.3 [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 230.

Abstract Introduction: Genome-wide association studies over the past 10 years have identified over 100 common variants associated with breast cancer risk. The association of these polymorphisms varies according to estrogen receptor status of the tumor, with some associated primarily with ER+ or ER- disease, or both subtypes. However, ER expression in breast cancer cells is continuous. The aim of this study was to establish whether the association between polygenic risk scores and ER expression in breast cancer is continuous in nature. Methods: Data were taken from the international Breast Cancer Association Consortium (BCAC). Germline genotype data for 75 SNPs previously reported to be associated with breast cancer were available for 4,999 breast cancer cases for which ER expression data were also available. Automated ER expression on a continuous scale was measured using two image analysis algorithms: Astrogrid and Ariol. A third form of ER expression score, the manual Allred ordinal, was used. Genotype data were used to calculate polygenic risk scores using the equation: PRS = ∑ βx where β is the log-odds of the odds ratio (OR) of breast cancer associated with that single nucleotide polymorphism (SNP), and x is the number of variant alleles at a SNP locus. Five different PRSs were calculated: i) SNPs associated with both ER+ and ER- disease (PRS-A, n= 74) ii) SNPs associated with ER+ disease at P<0.01 (PRS-B, n= 65) iii) SNPs associated with ER- disease at P<0.01 (PRS-C, n= 43) iv) SNPs associated with ER+ but not ER- disease (PRS-D, n= 30) v) SNPs associated with ER- but not ER+ disease (PRS-E, n = 8) The ER-specific weights for the PRS were taken from previous OR estimates. Linear regression models adjusted for study were used to test for a relationship between the five PRSs and ER expression measured on the Ariol, Astrogrid and Allred scales. Results: There was a strong positive correlation found between PRS-B against Astrogrid (β=0.0176, P=4.06x10-4), a weak negative correlation found with PRS-C (β=-0.012, P=0.019) and a weak positive correlation for PRS-A (β=0.012, P=0.0144). PRS-D and E showed far stronger association with Astrogrid than PRS-B and C, respectively (β=0.0186 and -0.017, P=1.97x10-4 and 8.1x10-4). There was a qualitatively similar association when the PRSs were correlated with Ariol and Allred, apart from the stronger correlations with PRS-B (β=91.2 and 0.0853, P=5.52x10-4 and 1.33x10-3). Conclusion: We have shown a robust association between PRS and ER expression in the directions we predicted. This suggests that the continuous variation in ER expression is biologically plausible and linked to genetic polymorphism. The weakening of association with PRS-A and the stronger associations with PRS-D and E show that having SNPs that are associated with both ER subtypes in the same PRS weakens significance, and suggests that subtype-specific PRSs may be improved on by using SNPs exclusive to the respective subtype. Citation Format: Anamay H. Shetty, Raza Ali, Fiona Blows, Carlos Caldas, Nicholas Inard, Javier Benítez, Jenny Chang-Claude, Georgia Chenevix-Trench, Fergus Couch, Angela Cox, Peter Devilee, Douglas Easton, Montserrat Garcia-Closas, Arto Mannermaa, Heli Nevanlinna, Marjanka Schmidt, Paul Pharoah. Patient level polygenic risk scores and continuous estrogen receptor expression in breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2211.

Abstract Large-scale genome-wide association studies (GWAS) have identified approximately 35 loci associated with epithelial ovarian cancer (EOC) risk. The majority of GWAS-identified disease susceptibility variants are located in noncoding regions, and causal genes underlying these associations remain largely unknown. Here, we performed a transcriptome-wide association study to search for novel genetic loci and plausible causal genes at known GWAS loci. We used RNA sequencing data (68 normal ovarian tissue samples from 68 individuals and 6,124 cross-tissue samples from 369 individuals) and high-density genotyping data from European descendants of the Genotype-Tissue Expression (GTEx V6) project to build ovarian and cross-tissue models of genetically regulated expression using elastic net methods. We evaluated 17,121 genes for their cis-predicted gene expression in relation to EOC risk using summary statistics data from GWAS of 97,898 women, including 29,396 EOC cases. With a Bonferroni-corrected significance level of P < 2.2 × 10−6, we identified 35 genes, including FZD4 at 11q14.2 (Z = 5.08, P = 3.83 × 10−7, the cross-tissue model; 1 Mb away from any GWAS-identified EOC risk variant), a potential novel locus for EOC risk. All other 34 significantly associated genes were located within 1 Mb of known GWAS-identified loci, including 23 genes at 6 loci not previously linked to EOC risk. Upon conditioning on nearby known EOC GWAS-identified variants, the associations for 31 genes disappeared and three genes remained (P < 1.47 × 10−3). These data identify one novel locus (FZD4) and 34 genes at 13 known EOC risk loci associated with EOC risk, providing new insights into EOC carcinogenesis. Significance: Transcriptomic analysis of a large cohort confirms earlier GWAS loci and reveals FZD4 as a novel locus associated with EOC risk. Cancer Res; 78(18); 5419–30. ©2018 AACR.

AbstractMotivationThe use of single nucleotide polymorphism (SNP) interactions to predict complex diseases is getting more attention during the past decade, but related statistical methods are still immature. We previously proposed the SNP Interaction Pattern Identifier (SIPI) approach to evaluate 45 SNP interaction patterns/patterns. SIPI is statistically powerful but suffers from a large computation burden. For large-scale studies, it is necessary to use a powerful and computation-efficient method. The objective of this study is to develop an evidence-based mini-version of SIPI as the screening tool or solitary use and to evaluate the impact of inheritance mode and model structure on detecting SNP–SNP interactions.ResultsWe tested two candidate approaches: the ‘Five-Full’ and ‘AA9int’ method. The Five-Full approach is composed of the five full interaction models considering three inheritance modes (additive, dominant and recessive). The AA9int approach is composed of nine interaction models by considering non-hierarchical model structure and the additive mode. Our simulation results show that AA9int has similar statistical power compared to SIPI and is superior to the Five-Full approach, and the impact of the non-hierarchical model structure is greater than that of the inheritance mode in detecting SNP–SNP interactions. In summary, it is recommended that AA9int is a powerful tool to be used either alone or as the screening stage of a two-stage approach (AA9int+SIPI) for detecting SNP–SNP interactions in large-scale studies.Availability and implementationThe ‘AA9int’ and ‘parAA9int’ functions (standard and parallel computing version) are added in the SIPI R package, which is freely available at https://linhuiyi.github.io/LinHY_Software/.Supplementary informationSupplementary data are available at Bioinformatics online.

Abstract Background Previous genome-wide association studies (GWAS) have identified 42 loci (P < 5 × 10−8) associated with risk of colorectal cancer (CRC). Expanded consortium efforts facilitating the discovery of additional susceptibility loci may capture unexplained familial risk. Methods We conducted a GWAS in European descent CRC cases and control subjects using a discovery–replication design, followed by examination of novel findings in a multiethnic sample (cumulative n = 163 315). In the discovery stage (36 948 case subjects/30 864 control subjects), we identified genetic variants with a minor allele frequency of 1% or greater associated with risk of CRC using logistic regression followed by a fixed-effects inverse variance weighted meta-analysis. All novel independent variants reaching genome-wide statistical significance (two-sided P < 5 × 10−8) were tested for replication in separate European ancestry samples (12 952 case subjects/48 383 control subjects). Next, we examined the generalizability of discovered variants in East Asians, African Americans, and Hispanics (12 085 case subjects/22 083 control subjects). Finally, we examined the contributions of novel risk variants to familial relative risk and examined the prediction capabilities of a polygenic risk score. All statistical tests were two-sided. Results The discovery GWAS identified 11 variants associated with CRC at P < 5 × 10−8, of which nine (at 4q22.2/5p15.33/5p13.1/6p21.31/6p12.1/10q11.23/12q24.21/16q24.1/20q13.13) independently replicated at a P value of less than .05. Multiethnic follow-up supported the generalizability of discovery findings. These results demonstrated a 14.7% increase in familial relative risk explained by common risk alleles from 10.3% (95% confidence interval [CI] = 7.9% to 13.7%; known variants) to 11.9% (95% CI = 9.2% to 15.5%; known and novel variants). A polygenic risk score identified 4.3% of the population at an odds ratio for developing CRC of at least 2.0. Conclusions This study provides insight into the architecture of common genetic variation contributing to CRC etiology and improves risk prediction for individualized screening.

AbstractEpidemiological, biological, and molecular data suggest links between endometriosis and endometrial cancer, with recent epidemiological studies providing evidence for an association between a previous diagnosis of endometriosis and risk of endometrial cancer. We used genetic data as an alternative approach to investigate shared biological etiology of these two diseases. Genetic correlation analysis of summary level statistics from genomewide association studies (GWAS) using LD Score regression revealed moderate but significant genetic correlation (rg = 0.23, P = 9.3 × 10−3), and SNP effect concordance analysis provided evidence for significant SNP pleiotropy (P = 6.0 × 10−3) and concordance in effect direction (P = 2.0 × 10−3) between the two diseases. Cross‐disease GWAS meta‐analysis highlighted 13 distinct loci associated at P ≤ 10−5 with both endometriosis and endometrial cancer, with one locus (SNP rs2475335) located within PTPRD associated at a genomewide significant level (P = 4.9 × 10−8, OR = 1.11, 95% CI = 1.07–1.15). PTPRD acts in the STAT3 pathway, which has been implicated in both endometriosis and endometrial cancer. This study demonstrates the value of cross‐disease genetic analysis to support epidemiological observations and to identify biological pathways of relevance to multiple diseases.