Sara H. Olson’s research while affiliated with Memorial Sloan Kettering Cancer Center and other places

Glioma is a rare brain tumor with a poor prognosis. Familial glioma is a subset of glioma with a strong genetic predisposition that accounts for approximately 5% of glioma cases. We performed whole-genome sequencing on an exploratory cohort of 203 individuals from 189 families with a history of familial glioma and an additional validation cohort of 122 individuals from 115 families. We found significant enrichment of rare deleterious variants of seven genes in both cohorts, and the most significantly enriched gene was HERC2 (P = 0.0006). Furthermore, we identified rare noncoding variants in both cohorts that were predicted to affect transcription factor binding sites or cause cryptic splicing. Last, we selected a subset of discovered genes for validation by CRISPR knockdown screening and found that DMBT1, HP1BP3, and ZCH7B3 have profound impacts on proliferation. This study performs comprehensive surveillance of the genomic landscape of familial glioma.

Purpose The incidence of endometrial cancer (EC) has been increasing faster among Black women than among other racial/ethnic groups in the United States. Although the mortality rate is nearly twice as high among Black than White women, there is a paucity of literature on risk factors for EC among Black women, particularly regarding menopausal hormone use and severe obesity. Methods We pooled questionnaire data on 811 EC cases and 3,124 controls from eight studies with data on self-identified Black women (4 case–control and 4 cohort studies). We analyzed cohort studies as nested case–control studies with up to 4 controls selected per case. We used logistic regression to estimate multivariable-adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Results We observed a positive association between BMI and EC incidence (Ptrend < 0.0001) The OR comparing BMI ≥ 40 vs. < 25 kg/m² was 3.92 (95% CI 2.91, 5.27). Abdominal obesity among those with BMI < 30 kg/m² was not appreciably associated with EC risk (OR 1.21, 95% CI 0.74, 1.99). Associations of reproductive history with EC were similar to those observed in studies of White women. Long-term use of estrogen-only menopausal hormones was associated with an increased risk of EC (≥ 5 years vs. never use: OR 2.08, 95% CI: 1.06, 4.06). Conclusions Our results suggest that the associations of established risk factors with EC are similar between Black and White women. Other explanations, such as differences in the prevalence of known risk factors or previously unidentified risk factors likely underlie the recent increases in EC incidence among Black women.

Background Epidemiological studies suggest that coffee consumption may be inversely associated with risk of endometrial cancer (EC), the most common gynecological malignancy in developed countries. Furthermore, coffee consumption may lower circulating levels of estrogen and insulin, hormones implicated in endometrial carcinogenesis. Antioxidants and other chemopreventive compounds in coffee may have anticarcinogenic effects. Based on available meta-analyses, the World Cancer Research Fund concluded that consumption of coffee probably protects against EC. Objective Our main aim was to examine the association between coffee consumption and EC risk by combining individual-level data in a pooled analysis. We also sought to evaluate potential effect modification by other risk factors of EC. Patients and Methods We combined individual-level data from 19 epidemiologic studies (6 cohort, 13 case-control) of 12,159 endometrial cancer cases and 27,479 controls from the Epidemiology of Endometrial Cancer Consortium (E2C2). Logistic regression was used to calculate odds ratios (OR) and their corresponding 95% confidence intervals (CI). All models were adjusted for potential confounders including age, race, body mass index, smoking status, diabetes status, study design and study site. Results Coffee drinkers had a lower risk of EC compared to non-coffee drinkers (multi-adjusted OR = 0.87, 95% CI = 0.79,0.95). There was a dose-response relationship between higher coffee consumption and lower risk of EC: compared to non-coffee drinkers, the adjusted pooled ORs for those who drank 1, 2–3 and more than 4 cups/day were 0.90 (95% CI = 0.82,1.00), 0.86 (95% CI = 0.78,0.95), and 0.76 (95% CI = 0.66,0.87), respectively (p for trend < 0.001). The inverse association between coffee consumption and EC risk was stronger in participants with body mass index (BMI) over 25 kg/m2. Conclusion The results of the largest analysis to date pooling individual-level data further support the potentially beneficial health effects of coffee consumption in relation to EC, especially among females with higher BMI.

Polygenic risk scores (PRS) for epithelial ovarian cancer (EOC) have the potential to improve risk stratification. Joint estimation of Single Nucleotide Polymorphism (SNP) effects in models could improve predictive performance over standard approaches of PRS construction. Here, we implemented computationally efficient, penalized, logistic regression models (lasso, elastic net, stepwise) to individual level genotype data and a Bayesian framework with continuous shrinkage, "select and shrink for summary statistics" (S4), to summary level data for epithelial non-mucinous ovarian cancer risk prediction. We developed the models in a dataset consisting of 23,564 non-mucinous EOC cases and 40,138 controls participating in the Ovarian Cancer Association Consortium (OCAC) and validated the best models in three populations of different ancestries: prospective data from 198,101 women of European ancestries; 7,669 women of East Asian ancestries; 1,072 women of African ancestries, and in 18,915 BRCA1 and 12,337 BRCA2 pathogenic variant carriers of European ancestries. In the external validation data, the model with the strongest association for non-mucinous EOC risk derived from the OCAC model development data was the S4 model (27,240 SNPs) with odds ratios (OR) of 1.38 (95% CI: 1.28-1.48, AUC: 0.588) per unit standard deviation, in women of European ancestries; 1.14 (95% CI: 1.08-1.19, AUC: 0.538) in women of East Asian ancestries; 1.38 (95% CI: 1.21-1.58, AUC: 0.593) in women of African ancestries; hazard ratios of 1.36 (95% CI: 1.29-1.43, AUC: 0.592) in BRCA1 pathogenic variant carriers and 1.49 (95% CI: 1.35-1.64, AUC: 0.624) in BRCA2 pathogenic variant carriers. Incorporation of the S4 PRS in risk prediction models for ovarian cancer may have clinical utility in ovarian cancer prevention programs.

Jews are estimated to be at increased risk of pancreatic cancer compared to non-Jews, but their observed 50–80% excess risk is not explained by known non-genetic or genetic risk factors. We conducted a GWAS in a case–control sample of American Jews, largely Ashkenazi, including 406 pancreatic cancer patients and 2332 controls, identified in the dbGaP, PanScan I/II, PanC4 and GERA data sets. We then examined resulting SNPs with P < 10–7 in an expanded sample set, of 539 full- or part-Jewish pancreatic cancer patients and 4117 full- or part-Jewish controls from the same data sets. Jewish ancestries were genetically determined using seeded FastPCA. Among the full Jews, a novel genome-wide significant association was detected on chromosome 19p12 (rs66562280, per-allele OR = 1.55, 95% CI = 1.33–1.81, P = 10–7.6). A suggestive relatively independent association was detected on chromosome 19p13.3 (rs2656937, OR = 1.53, 95% CI = 1.31–1.78, P = 10–7.0). Similar associations were seen for these SNPs among the full and part Jews combined. This is the first GWAS conducted for pancreatic cancer in the increased-risk Jewish population. The SNPs rs66562280 and rs2656937 are located in introns of ZNF100-like and ARRDC5, respectively, and are known to alter regulatory motifs of genes that play integral roles in pancreatic carcinogenesis.

Polygenic risk scores (PRS) for epithelial ovarian cancer (EOC) have the potential to improve risk stratification. Joint estimation of Single Nucleotide Polymorphism (SNP) effects in models could improve predictive performance over standard approaches of PRS construction. Here, we implemented computationally-efficient, penalized, logistic regression models (lasso, elastic net, stepwise) to individual level genotype data and a Bayesian framework with continuous shrinkage, “select and shrink for summary statistics” (S4), to summary level data for epithelial non-mucinous ovarian cancer risk prediction. We developed the models in a dataset consisting of 23,564 non-mucinous EOC cases and 40,138 controls participating in the Ovarian Cancer Association Consortium (OCAC) and validated the best models in three populations of different ancestries: prospective data from 198,101 women of European ancestry; 7,669 women of East Asian ancestry; 1,072 women of African ancestry, and in 18,915 BRCA1 and 12,337 BRCA2 pathogenic variant carriers of European ancestry. In the external validation data, the model with the strongest association for non-mucinous EOC risk derived from the OCAC model development data was the S4 model (27,240 SNPs) with odds ratios (OR) of 1.38(95%CI:1.28–1.48,AUC:0.588) per unit standard deviation, in women of European ancestry; 1.14(95%CI:1.08–1.19,AUC:0.538) in women of East Asian ancestry; 1.38(95%CI:1.21-1.58,AUC:0.593) in women of African ancestry; hazard ratios of 1.37(95%CI:1.30–1.44,AUC:0.592) in BRCA1 pathogenic variant carriers and 1.51(95%CI:1.36-1.67,AUC:0.624) in BRCA2 pathogenic variant carriers. Incorporation of the S4 PRS in risk prediction models for ovarian cancer may have clinical utility in ovarian cancer prevention programs.

A full‐term pregnancy is associated with reduced endometrial cancer risk; however, whether the effect of additional pregnancies is independent of age at last pregnancy is unknown. The associations between other pregnancy‐related factors and endometrial cancer risk are less clear. We pooled individual participant data from 11 cohort and 19 case‐control studies participating in the Epidemiology of Endometrial Cancer Consortium (E2C2) including 16 986 women with endometrial cancer and 39 538 control women. We used one‐ and two‐stage meta‐analytic approaches to estimate pooled odds ratios (ORs) for the association between exposures and endometrial cancer risk. Ever having a full‐term pregnancy was associated with a 41% reduction in risk of endometrial cancer compared to never having a full‐term pregnancy (OR = 0.59, 95% confidence interval [CI] 0.56‐0.63). The risk reduction appeared the greatest for the first full‐term pregnancy (OR = 0.78, 95% CI 0.72‐0.84), with a further ~15% reduction per pregnancy up to eight pregnancies (OR = 0.20, 95% CI 0.14‐0.28) that was independent of age at last full‐term pregnancy. Incomplete pregnancy was also associated with decreased endometrial cancer risk (7%‐9% reduction per pregnancy). Twin births appeared to have the same effect as singleton pregnancies. Our pooled analysis shows that, while the magnitude of the risk reduction is greater for a full‐term pregnancy than an incomplete pregnancy, each additional pregnancy is associated with further reduction in endometrial cancer risk, independent of age at last full‐term pregnancy. These results suggest that the very high progesterone level in the last trimester of pregnancy is not the sole explanation for the protective effect of pregnancy.

PALB2 has been identified as a breast and pancreatic cancer susceptibility gene. Utilizing a targeted sequencing approach, we discovered two novel germline missense PALB2 variants c.191C>T and c.311C>T, encoding p.Ser64Leu and p.Pro104Leu, respectively, in individuals in a pancreatic cancer registry. No missense PALB2 variants from familial pancreatic cancer patients, and few PALB2 variants overall, have been functionally characterized. Given the known role of PALB2, we tested the impact of p.Ser64Leu and p.Pro104Leu variants on DNA damage responses. Neither p.Ser64Leu nor p.Pro104Leu had clear effects on interactions with BRCA1 and KEAP1, which are mediated by adjacent motifs in PALB2. However, both variants are associated with defective recruitment of PALB2, and the RAD51 recombinase downstream, to DNA damage foci. Further, p.Ser64Leu and p.Pro104Leu both largely compromise DNA double‐strand break‐initiated homologous recombination, and confer increased cellular sensitivity to ionizing radiation (IR) and the poly (ADP‐ribose) polymerase (PARP) inhibitor Olaparib. Taken together, our results represent the first demonstration of functionally deleterious PALB2 missense variants associated with familial pancreatic cancer and of deleterious variants in the N‐terminus outside of the coiled‐coil domain. Further, our results suggest the possibility of personalized treatments, using IR or PARP inhibitor, of pancreatic and other cancers that carry a deleterious PALB2 variant. This article is protected by copyright. All rights reserved.

Background: Germline copy number variants (CNVs) increase risk for many diseases, yet detection of CNVs and quantifying their contribution to disease risk in large-scale studies is challenging due to biological and technical sources of heterogeneity that vary across the genome within and between samples. Methods: We developed an approach called CNPBayes to identify latent batch effects in genome-wide association studies involving copy number, to provide probabilistic estimates of integer copy number across the estimated batches, and to fully integrate the copy number uncertainty in the association model for disease. Results: Applying a hidden Markov model (HMM) to identify CNVs in a large multi-site Pancreatic Cancer Case Control study (PanC4) of 7598 participants, we found CNV inference was highly sensitive to technical noise that varied appreciably among participants. Applying CNPBayes to this dataset, we found that the major sources of technical variation were linked to sample processing by the centralized laboratory and not the individual study sites. Modeling the latent batch effects at each CNV region hierarchically, we developed probabilistic estimates of copy number that were directly incorporated in a Bayesian regression model for pancreatic cancer risk. Candidate associations aided by this approach include deletions of 8q24 near regulatory elements of the tumor oncogene MYC and of Tumor Suppressor Candidate 3 (TUSC3). Conclusions: Laboratory effects may not account for the major sources of technical variation in genome-wide association studies. This study provides a robust Bayesian inferential framework for identifying latent batch effects, estimating copy number, and evaluating the role of copy number in heritable diseases.