Abstract P3-19-29: Polygenic risk scores predict toxicities in racially diverse breast cancer patients following radiotherapy

Abstract

Abstract Purpose: Fatigue, fibrosis, and breast asymmetry are the most common side effects of radiotherapy, experienced in up to 40% of survivors long term. Due to the prevalence of these toxicities, a number of patient, tumor, and treatment-related characteristics have been studied in an attempt to identify at risk individuals. Additionally, several single nucleotide polymorphisms (SNPs) such as those that reside in gene TGFB1 have been associated with long-term breast fibrosis following radiotherapy by genome-wide association studies (GWAS). However, these findings are often not consistently replicated in various cohorts, particularly those with racially diverse patients, indicating that these significant SNPs (with p < 5e-8) or clinical factors alone may not provide enough information to determine individual risk for radiation-related toxicities. We hypothesized that the Polygenic Risk Scores (PRS) derived from the GWAS summary statistics of a larger number of potentially significant SNPs (e.g. with p-values < 0.05) have the potential to improve risk prediction, especially by using GWAS data of a racially diverse cohort. Methods: We conducted three prospective, longitudinal studies of breast cancer patients treated with whole breast radiotherapy with or without regional nodal irradiation. In all studies, fatigue, skin thickening, and breast asymmetry were assessed before and 12-18 months post radiotherapy using the following identical measures: percentage breast retraction assessment (pBRA) calculated from digital photographs, skin thickening objectively measured on ultrasound images (STRA), and fatigue assessed by the Multidimensional Fatigue Symptom Inventory (MFI) 20-item scale. We profiled genome-wide SNP data from peripheral blood samples of enrolled patients (n=176; 44% African Ancestry; 56% European Ancestry) by Affymetrix arrays and then imputed these genotyped SNPs to 1000 Genome Phase III reference panel. Genotyped or imputed SNPs with imputation R2>0.3 with minor allele frequency (MAF) > 5% (~10M) were analyzed in our GWAS with baseline, post radiotherapy, and changes in pBRA, STRA, and MFI. GWAS was first conducted separately in African and European ancestry samples, and then combined by meta-analysis for baseline and post radiotherapy measurements. Joint GWAS was conducted on all samples for the change phenotypes as we observed no heterogeneous phenotype distribution for samples of different ancestries. PRS was derived using the GWAS summary statistics for each phenotype. Results: Twelve independent significant SNPs were identified for baseline and post radiotherapy measurements in pBRA, STRA, and MFI, all of which were identified only in patients of African and not European ancestry due to low MAF < 5%. All post radiotherapy PRS were significantly correlated with the corresponding baseline (r>0.29, p < 0.05) and change PRS (r>0.19, p <0.05). Importantly, PRS accounted for 76%, 73%, and 79% of the variance of changes in pBRA, STRA, and MFI after radiation. When clinical (e.g. breast volume and age) and treatment-related factors (e.g. radiation dose, chemotherapy treatment) were added as additional covariates to the respective PRS, the prediction R2 increased to 92%, 96%, and 96% for changes in pBRA, STRA, and MFI, respectively. Conclusions: Our findings indicate that over 70% of the variation in fatigue, skin thickening and breast asymmetry changes experienced by breast cancer patients after radiotherapy may be explained by PRS, which can be derived from our GWAS summary statistics and SNP data profiled from blood samples prior to radiotherapy. Genome-wide significant risk SNPs of baseline and post radiotherapy measurements of pBRA, STRA, and MFI were found only in samples of patients of African ancestry. Future studies aimed at replicating our findings in larger cohorts are needed. Citation Format: Mylin Ann Torres, Tian Liu, Boran Zhou, Xiaofeng Yang, Gabrielle Brown, India Green, Jolinta Lin, Andrew H Miller, Jingjing Yang. Polygenic risk scores predict toxicities in racially diverse breast cancer patients following radiotherapy [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-19-29.