Evaluating polygenic risk scores in assessing risk of nine solid and hematologic cancers in European descendants

Abstract

Genome-wide association studies (GWAS) have identified many genetic risk variants for cancers. The utility of these variants in assessing risk of esophageal, gastric and endometrial cancers, as well as melanoma, glioma, diffuse large B-cell lymphoma, follicular lymphoma, chronic lymphoid leukemia and multiple myeloma, has not been adequately investigated. We constructed a site-specific polygenic risk score (PRS) for each of these nine cancers using their GWAS-identified risk variants. Using data from 400 807 participants of European descent in the UK Biobank, a population-based cohort study, we estimated the hazard ratios of each cancer associated with its PRS using Cox proportional hazard models. During a median follow-up of 5.8 years, 3905 incident cases of these nine cancers were identified in the cohort. The area under the receiver operating characteristic curve ranged from 0.53 to 0.69 for these cancers. Except for esophageal cancer, significant dose-response associations were observed between PRS and cancer risk. Compared to individuals in the middle quintile (40%-60%) at an average risk, those among the highest 5% of the PRS had a twofold elevated risk of melanoma, glioma, follicular lymphoma or multiple myeloma, and a fourfold elevated risk of chronic lymphoid leukemia. Using PRS, 63.0% of the participants could be classified as having an over twofold elevated risk for at least one cancer. The PRS derived using risk variants identified to date by GWAS showed the potential in identifying individuals at a significantly elevated risk of cancer for prevention.

Associations of genetic susceptibility to 16 cancers with risk of breast cancer overall and by intrinsic subtypes

“…We selected genetic risk variants, including single-nucleotide polymorphisms (SNPs) or small insertions or deletions from the most recent studies with the largest sample sizes of individuals of European ancestry (sample size varied from 5,415 to 299,686; Table S2 ). 25 , 26 Using the conventional genome-wide significance threshold (p < 5 × 10 −8 ), variants showing an association with p values at or below this threshold were included in our study. We also included some risk variants with an established association at p < 5 × 10 −8 from previous studies with the cancer of interest even if they were not significant at p < 5 × 10 −8 in the latest studies due to small sample sizes.…”

“…Details on the derivation of the genetic risk score have been published recently. 17 , 25 , 26 Hazard ratios (HRs) and 95% confidence intervals (CIs) associated with each PRS were estimated by Cox proportional hazard models using age as the underlying timescale left-truncated at the age of baseline interview and adjusted for age at enrollment, genotype array type (UKBL or UKBB), the 10 PCs for ancestry, and stratified by birth cohorts. The assumptions of proportionality were examined using Schoenfeld residuals.…”

10 Years of GWAS discovery in endometrial cancer: Aetiology, function and translation

“…61 PRS generated from currently known endometrial cancer risk variants are limited by their low discriminatory power, with area under the curve (AUC) results for PRS predictability reported around 0.56-0.57. 62,63 Including risk factors, such as BMI and parity, into the PRS, has led to a slight improvement in distinguishing cancer cases from cancer-free individuals. 61 This suggests that risk prediction for endometrial cancer could be further improved by the integration of disease PRS with risk factors to generate a multi-trait PRS (multi-PRS), as has been shown for other complex traits, including ischemic stroke 58 and type 2 diabetes.…”

The Polygenic Risk Score Knowledge Base: A Centralized Online Repository for Calculating and Contextualizing Polygenic Risk Scores

“…Therefore, choosing an appropriate GWA study to calculate polygenic risk scores is paramount to the delity of the calculations because the accuracy and predictive power of a polygenic risk score is dependent on the power and scope of the corresponding GWA study data 29,30 . When used appropriately, polygenic risk scores can capture missing heritability and are a measure of genetic risk for a trait compared to the average risk in a speci c population [31][32][33][34] . Because polygenic risk scores can stratify populations based on distinct risk, they are useful in determining clinical and personal interventions 35,36 .…”