Polley, Eric C.

Abstract Background: Polygenic risk scores (PRS) based on a large number of SNPs associated with breast cancer have recently been made available in combination with multigene panel testing for clinical management of breast cancer risk. Breast cancer risk assessment has historically been performed using various non-genetic risk models, which predict lifetime risk of breast cancer in unaffected women based on personal and family history information. While clinical use of PRS in combination with these models continues to increase, the predictive performance of a PRS combined with clinical information in real-world patient populations has not been thoroughly investigated. Methods: We assessed a 100-SNP breast cancer PRS in combination with the Tyrer-Cuzick model-estimated lifetime breast cancer risk in 5,374 Caucasian women from the case-control Mayo Clinic Breast Cancer Study (MCBCS), comprised of 3,373 breast cancer cases and 2,001 unaffected controls seen at Mayo Clinic from 2004 to 2016. Clinical, family history, and demographic information was collected on all MCBCS participants via questionnaires and medical record review. SNPs were selected for the PRS if they exceeded genome-wide significance in ≥2 large-scale breast cancer genome wide association studies (GWAS) in women of European descent. Selected SNPs were then used to construct a population standardized PRS, accounting for per-allele odds ratios reported in the largest GWAS to date and population-specific allele frequencies. PRS and Tyrer-Cuzick model associations with breast cancer and related clinical features were tested using multivariate logistic regression. Results: Among cases and controls, mean±SD age at diagnosis and study enrollment was 56.4±12.4 and 56.2±14.1, respectively. Approximately 36.9% of cases and 16.9% of controls had ≥1 first- or second-degree relative with breast or ovarian cancer, respectively. Among unaffected women, the median (IQR) estimated lifetime risk of breast cancer to age 85 based on Tyrer-Cuzick alone was 8.3% (6.9%), and 2.6% had a clinically actionable lifetime risk of ≥20%. The mean±SD sum of the risk alleles across 100 SNPs was significantly higher for cases vs. controls (94.8±6.3 vs. 92.7±6.2, p<0.001). Compared to women in the lowest quartile of PRS, those in the 2nd, 3rd and 4th quartiles were 1.47 (95% CI: 1.26-1.72), 1.87 (95% CI: 1.60-2.19), and 2.22 (95% CI: 1.89-2.61) times more likely to have breast cancer after adjustment for age and family history. The AUC was 0.59 (95% CI: 0.58-0.61) for PRS and 0.60 (95% CI: 0.58-0.61) for Tyrer-Cuzick model estimated risk alone; but when combined, the model achieved an AUC of 0.63 (95% CI: 0.62-0.65). Among unaffected women, 6.7% had ≥20% lifetime risk of breast cancer based on the combined PRS-clinical model. Conclusions: These findings suggest that the 100-SNP PRS significantly improves estimation of breast cancer risk based on non-genetic models, and can be used to further identify women at increased lifetime risk who would otherwise not be identified by clinical assessment alone. Citation Format: Mary Helen Black, Shuwei Li, Holly LaDuca, Sebastian Armasu, Eric C. Polley, Patrick Fitz-Gibbon, Chris G. Scott, Stacey J. Winham, Janet E. Olson, Kathryn J. Ruddy, Celine M. Vachon, Fergus J. Couch. Performance of a polygenic risk score combined with clinical assessment for breast cancer risk [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P5-07-06.

Abstract Background: BRCA1/2 carriers have higher scores in the 21-gene RS assay compared to non-carriers. Little is known about the 21-gene RS patterns in women harboring germline mutations in genes outside BRCA1/2. Methods: We utilized the Mayo Clinic Breast Cancer Study Registry to identify individuals with non-metastatic estrogen receptor positive breast cancer and clinical 21-gene RS results available. We cross-referenced this with an institutional registry of patients tested for hereditary cancer genes, either from the electronic health record or from a research study conducted using a custom QIAseq multiplex amplicon-based panel (Qiagen). The latter assay covered target regions of all coding regions for the following predisposition genes: ATM, BARD1, BLM, BRCA1, BRCA2, BRIP1, CDH1, CDKN2A, CHEK2, ERCC2, FANCC, FANCM, MLH1, MRE11A, MSH2, MSH6, MUTYH, NBN, NF1, PALB2, PMS2, PTEN, RAD50, RAD51C, RAD51D, RECQL, PRSS1, STK11, TP53, and XRCC2. Student’s t-test was used to compare 21-gene RS between carriers, non-carriers, and specific carrier genes (JMP v 14.0). Results: 282 patients were identified with early stage breast cancer and 21-gene RS available, including 85 known mutation carriers and 197 non-carriers (or unknown carrier status). Germline mutations were identified in CHEK2 (n=25), BRCA2 (n=18), ATM (n=10), BRCA1 (n=5), PALB2 (n=5), TP53 (n=3), FANCM (n=2), MUTYH (n=2), PMS2 (n=2), PPM1D (n=2), RECQL (n=2), BARD1 (n=1), BLM (n=1), CDH1 (n=1), ERCC2 (n=1), FANCC (n=1), MSH6 (n=1), NBN (n=1), PRSS1 (n=1), and PTEN (n=1). In the entire cohort, the mean 21-gene RS was 20.4 in carriers and 18.0 in non-carriers (p=0.11). We performed individual analyses on specific genes identified in ≥ 5 patients (table 1). Compared to the mean RS in non-carriers (18.0), higher mean RS were observed in patients harboring BRCA1 (38.8, p=0.002), PALB2 (26.6, p=0.0724), and BRCA2 (23.2, p=0.0453). The RS in CHEK2 and ATM carriers were comparable to those of non-carriers. Conclusions: We observed differences in RS score distribution based on specific mutated breast cancer predisposition genes. Further validation of the distribution and performance of the 21-gene RS in germline carriers is needed. 21-gene RS scorenMedian RSMean≤1516-25≥26n%n%n%CHEK2251616.61248728624BRCA2182123.2211.11161.1527.8ATM101716.9330550220BRCA153635.8120120360PALB253326.6120120360None1971518.09950.367343115.7 Citation Format: Karthik V Giridhar, Grace M Choong, Siddartha Yadav, Eric C Polley, Roberto A Leon-Ferre, Minetta C Liu, Matthew P Goetz, Fergus J Couch. Distribution of the 21-gene recurrence score (RS) assay (Oncotype DX) in carriers of germline mutations associated with breast cancer predisposition syndromes [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P5-03-07.

Abstract Background The risks of breast cancer in African American (AA) women associated with inherited mutations in breast cancer predisposition genes are not well defined. Thus, whether multigene germline hereditary cancer testing panels are applicable to this population is unknown. We assessed associations between mutations in panel-based genes and breast cancer risk in 5054 AA women with breast cancer and 4993 unaffected AA women drawn from 10 epidemiologic studies. Methods Germline DNA samples were sequenced for mutations in 23 cancer predisposition genes using a QIAseq multiplex amplicon panel. Prevalence of mutations and odds ratios (ORs) for associations with breast cancer risk were estimated with adjustment for study design, age, and family history of breast cancer. Results Pathogenic mutations were identified in 10.3% of women with estrogen receptor (ER)-negative breast cancer, 5.2% of women with ER-positive breast cancer, and 2.3% of unaffected women. Mutations in BRCA1, BRCA2, and PALB2 were associated with high risks of breast cancer (OR = 47.55, 95% confidence interval [CI] = 10.43 to >100; OR = 7.25, 95% CI = 4.07 to 14.12; OR = 8.54, 95% CI = 3.67 to 24.95, respectively). RAD51D mutations were associated with high risk of ER-negative disease (OR = 7.82, 95% CI = 1.61 to 57.42). Moderate risks were observed for CHEK2, ATM, ERCC3, and FANCC mutations with ER-positive cancer, and RECQL mutations with all breast cancer. Conclusions The study identifies genes that predispose to breast cancer in the AA population, demonstrates the validity of current breast cancer testing panels for use in AA women, and provides a basis for increased referral of AA patients for cancer genetic testing.

PURPOSE To determine the sensitivity and specificity of genetic testing criteria for the detection of germline pathogenic variants in women with breast cancer. MATERIALS AND METHODS Women with breast cancer enrolled in a breast cancer registry at a tertiary cancer center between 2000 and 2016 were evaluated for germline pathogenic variants in 9 breast cancer predisposition genes ( ATM , BRCA1, BRCA2, CDH1, CHEK2, NF1, PALB2, PTEN, and TP53). The performance of the National Comprehensive Cancer Network (NCCN) hereditary cancer testing criteria was evaluated relative to testing of all women as recommended by the American Society of Breast Surgeons. RESULTS Of 3,907 women, 1,872 (47.9%) meeting NCCN criteria were more likely to carry a pathogenic variant in 9 predisposition genes compared with women not meeting criteria (9.0% v 3.5%; P < .001). Of those not meeting criteria (n = 2,035), 14 (0.7%) had pathogenic variants in BRCA1 or BRCA2. The sensitivity of NCCN criteria was 70% for 9 predisposition genes and 87% for BRCA1 and BRCA2, with a specificity of 53%. Expansion of the NCCN criteria to include all women diagnosed with breast cancer at ≤ 65 years of age achieved > 90% sensitivity for the 9 predisposition genes and > 98% sensitivity for BRCA1 and BRCA2. CONCLUSION A substantial proportion of women with breast cancer carrying germline pathogenic variants in predisposition genes do not qualify for testing by NCCN criteria. Expansion of NCCN criteria to include all women diagnosed at ≤ 65 years of age improves the sensitivity of the selection criteria without requiring testing of all women with breast cancer.

Abstract Survival analysis is a widely used method to establish a connection between a time to event outcome and a set of potential covariates. Accurately predicting the time of an event of interest is of primary importance in survival analysis. Many different algorithms have been proposed for survival prediction. However, for a given prediction problem it is rarely, if ever, possible to know in advance which algorithm will perform the best. In this paper we propose two algorithms for constructing super learners in survival data prediction where the individual algorithms are based on proportional hazards. A super learner is a flexible approach to statistical learning that finds the best weighted ensemble of the individual algorithms. Finding the optimal combination of the individual algorithms through minimizing cross-validated risk controls for over-fitting of the final ensemble learner. Candidate algorithms may range from a basic Cox model to tree-based machine learning algorithms, assuming all candidate algorithms are based on the proportional hazards framework. The ensemble weights are estimated by minimizing the cross-validated negative log partial likelihood. We compare the performance of the proposed super learners with existing models through extensive simulation studies. In all simulation scenarios, the proposed super learners are either the best fit or near the best fit. The performances of the newly proposed algorithms are also demonstrated with clinical data examples.

Abstract Background The germline cancer predisposition genes associated with increased risk of each clinical subtype of breast cancer, defined by estrogen receptor (ER), progesterone receptor (PR), and HER2, are not well defined. Methods A total of 54 555 invasive breast cancer patients with 56 480 breast tumors were subjected to clinical hereditary cancer multigene panel testing. Heterogeneity for predisposition genes across clinical breast cancer subtypes was assessed by comparing mutation frequencies by gene among tumor subtypes and by association studies between each tumor subtype and reference controls. Results Mutations in 15 cancer predisposition genes were detected in 8.6% of patients with ER+/HER2-; 8.9% with ER+/HER2+; 7.7% with ER-/HER2+; and 14.4% of ER-/PR-/HER2- tumors. BRCA1, BRCA2, BARD1, and PALB2 mutations were enriched in ER- and HER2- tumors; RAD51C and RAD51D mutations were enriched in ER- tumors only; TP53 mutations were enriched in HER2+ tumors, and ATM and CHEK2 mutations were enriched in both ER+ and/or HER2+ tumors. All genes were associated with moderate (odds ratio &gt; 2.00) or strong (odds ratio &gt; 5.00) risks of at least one subtype of breast cancer in case-control analyses. Mutations in ATM, BARD1, BRCA1, BRCA2, CHEK2, PALB2, RAD51C, RAD51D, and TP53 had predicted lifetime absolute risks of at least 20.0% for breast cancer. Conclusions Germline mutations in hereditary cancer panel genes confer subtype-specific risks of breast cancer. Combined tumor subtype, age at breast cancer diagnosis, and family history of breast and/or ovarian cancer information provides refined categorical estimates of mutation prevalence for women considering genetic testing.

PURPOSE In studies of men of European ancestry, rare pathogenic variants in DNA repair pathway genes have been shown to be associated with risk of aggressive prostate cancer. The contribution of rare coding variation to prostate cancer risk in men of African ancestry has not been established. METHODS We sequenced a panel of 19 DNA repair and cancer predisposition genes in 2,453 African American and 1,151 Ugandan cases and controls with prostate cancer. Rare variants were classified as pathogenic or putatively functionally disruptive and examined in association with prostate cancer risk and disease aggressiveness in gene and pathway-level association analyses. RESULTS Pathogenic variants were found in 75 of 2,098 cases (3.6%) and 31 of 1,481 controls (2.1%; odds ratio [OR], 1.82; 95% CI, 1.19 to 2.79; P = .0044), with the association being stronger for more aggressive disease phenotypes (OR, 3.10; 95% CI, 1.54 to 6.23; P = .0022). The highest risks for aggressive disease were observed with pathogenic variants in the ATM, BRCA2, PALB2, and NBN genes, with ORs ranging from approximately 4 to 15 in the combined study sample of African American and Ugandan men. Rare, nonpathogenic, nonsynonymous variants did not have a major impact on risk of overall prostate cancer or disease aggressiveness. CONCLUSION Rare pathogenic variants in DNA repair genes have appreciable effects on risk of aggressive prostate cancer in men of African ancestry. These findings have potential implications for panel testing and risk stratification in this high-risk population.

PURPOSE To compare the prevalence of germline mutations in metastatic hormone-sensitive prostate cancer (mHSPC) and metastatic castrate-resistant prostate cancer (mCRPC) and assess the impact of mutations on progression to castration resistance and overall survival. METHODS Targeted sequencing of germline DNA from 704 men (221 at the time of mHSPC and 483 at the time of mCRPC) enrolled in two advanced prostate cancer registries at Mayo Clinic between 2003 and 2013 was performed for 21 predisposition genes. Frequencies of pathogenic mutations were compared in patients and reference controls to identify genes enriched in metastatic prostate cancer. Multivariable Cox proportional hazards regression was used to identify predictors of progression to mCRPC and overall survival. RESULTS Sixty-eight germline mutations in 12 genes were identified in 66 men (9.4%). Mutations in ATM, BRCA2, CHEK2, FANCM, and TP53 were significantly enriched (odds ratio greater than 2.0) in the metastatic cohorts compared with reference controls. The frequency of germline mutations was similar for patients with mHSPC and mCRPC (11.8% v 8.3%; P = .16). The median time to progression from mHSPC to mCRPC was 23.1 and 32.5 months for patients with and without mutations, respectively ( P = .96). Although older age at diagnosis, Gleason score greater than 7, elevated alkaline phosphatase level, and high volume of disease were associated with shorter duration of progression to mCRPC and poor overall survival, mutation status was not (progression to mCRPC hazard ratio, 0.81; 95% CI, 0.61 to 1.09; P = .17; overall survival hazard ratio, 1.00; 95% CI, 0.75 to 1.34; P = .98). CONCLUSION Similarly elevated rates of germline predisposition gene mutations in mHSPC and mCRPC suggest that germline genetic testing may help to guide medical management for all patients with advanced metastatic prostate cancer. Mutation status was not associated with shorter progression to mCRPC or poor overall survival.

Abstract Recent years have seen an explosion in the availability of paired molecular profiling and drug screen data, providing an unprecedented opportunity for the development of targeted therapies based on an individual’s genetic background. Despite a number of recent successes in diseases ranging from cystic fibrosis to cancer, significant hurdles remain in our ability to accurately predict treatments based on molecular profiling data. In particular, few such tools exist that allow the integration of heterogeneous data types (e.g. genomic, transcriptomic, and somatic mutations), along with high-throughput drug screen data to make predictions about treatment efficacy. Here, we describe a generalized open-source pipeline developed for the analysis of precision medicine data, Pharmacogenomics Prediction Pipeline, or “P3”. The modular design of P3 enables the inclusion of arbitrary input data types and the selection from multiple alternative machine learning algorithms, while automated statistical and visualization reporting steps incorporated throughout the pipeline assist in parameter tuning and early detection of problematic data components. Molecular profiling data is further enriched by the incorporation of external biological information in the form of pathway and gene set annotations such and Gene Ontology (GO) and The Molecular Signatures Database (MSigDB). To demonstrate the use of P3 for preclinical biomarker prediction, we applied P3 to an unpublished multiple myeloma dataset consisting of exome, RNA-Seq, CNV, and drug screen data for 1,912 compounds across 47 tumor cell lines. Specifically, P3 was used to predict molecular features associated with response to treatment for all drugs where a differential response to treatment was observed across patients. Furthermore, molecular profiling and drug screen data for 267 drugs and over a thousand cell lines spanning multiple cancer types from the Genomics of Drug Sensitivity in Cancer (GDSC) project were analyzed using P3, providing insights into shared mechanisms of drug sensitivity and resistance across different cancer and treatment types. Citation Format: V Keith Hughitt, Sayeh Gorjifard, Aleksandra M. Michalowski, John K. Simmons, Ryan Dale, Eric C. Polley, Jonathan J. Keats, Beverly A. Mock. A flexible pipeline for precision medicine biomarker detection and prediction of treatment response [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 5113.

Abstract Background: Understanding the joint effect on breast cancer risk of rare pathogenic variants in cancer predisposition genes and polygenic risk scores (PRS) from common variants can enable a precision approach to breast cancer management. Previous work found that PRS were associated with breast cancer risk in BRCA1 and BRCA2 mutation carriers recruited from cancer genetics clinics (Kuchenbaecker et al. JNCI 2017). The joint effects of pathogenic variants and PRS have not been studied in samples drawn from the general population. There is also no existing study evaluating the effect of PRS in mutation carriers in genes other than BRCA1/2. Here we evaluate the joint effects of PRS and pathogenic mutations in established cancer predisposition genes in a population-based case-control sample. Method: We analyzed 53,199 European-ancestry individuals (20,730 controls and 21,272 cases) drawn from 7 cohorts and 2 population-based case-control studies in the CAnceR RIsk Estimates Related to Susceptibility” (CARRIERS) consortium. Targeted sequencing was performed using dual bar-coded QIAseq. Mutation calling was conducted with Haplotype Caller and Vardict. We sequenced 18 breast cancer predisposition genes: ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CDKN2A, CHECK2, FANCC, MLH1, MSH2, MSH6, NF1, PALB2, PTEN, RAD51C, RAD51D and TP53. The PRS was calculated based on 128 common variants using effect estimates from the largest published breast cancer GWAS. The PRS was standardized to a mean of 0 and standard deviation of 1. Logistic regression was performed to assess the combined effect of identified mutation and common variants (including main and interaction effects for carrier status and PRS). Results: A total of 1,770 pathogenic mutations were observed. About 1% of the study sample had mutations in either BRCA1 (n=209) or BRCA2 gene (n=305), and 2.2% had mutations in other genes (n=1,176). The effect of PRS in BRCA1 carriers was OR= 0.97 (95%CI: 0.62, 1.52); however, this is not statistically significantly different from the PRS in non-carriers [OR=1.23 (95%CI: 1.20, 1.25)] or previous estimates in carriers [OR: 1.14, 95%CI: 1.11-1.17]. The effect of PRS in BRCA2 carriers was OR=1.87 (95%CI: 1.31, 2.78) which was statistically significantly different from that of the non-carriers. Among the mutation carriers in genes other than BRCA, the effect of PRS on breast cancer was OR=1.00 (95%CI: 0.88,1.15). Conclusion: Consistent with previous studies, we did not find evidence that the effect of the PRS among BRCA1 carriers was statistically significantly different than non-carriers. We found some evidence that the PRS effect may be larger among BRCA2 carriers than non-carriers. Our results also suggest that PRS may not have a strong effect on breast cancer risk in mutation carriers of other predisposition genes; further work is needed for verification. Citation Format: Chi Gao, Chunling Hu, Steven N. Hart, Rohan Gnanaolivu, Kun Y. Lee, Jenna Lilyquist, Nicholas J. Boddicker, Bruce Eckloff, Raed Samara, Josh Klebba, Paul Auer, Leslie Bernstein, Mia Gaudet, Hoda Anton-Culver, Amy Trentham-Dietz, Julie R. Palmer, Song Yao, Christopher Haiman, Janet E. Olson, Susan Domchek, Jeffrey Weitzel, David Goldgar, Katherine L. Nathanson, Eric C. Polley, Fergus J. Couch, Peter Kraft. The joint effects of polygenic risk scores and pathogenic variants in cancer predisposition genes on breast cancer risk in the general population: results from the CARRIERS study [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 4177.