Development of a method for generating SNP interaction-aware polygenic risk scores for radiotherapy toxicity

Abstract

Aim: To identify the effect of single nucleotide polymorphism (SNP) interactions on the risk of toxicity following radiotherapy (RT) for prostate cancer (PCa) and propose a new method for polygenic risk score incorporating SNP-SNP interactions (PRSi). Materials and methods: Analysis included the REQUITE PCa cohort that received external beam RT and was followed for 2 years. Late toxicity endpoints were: rectal bleeding, urinary frequency, haematuria,

Polygenic Risk Scores Associated with Tumor Immune Infiltration in Common Cancers

“…Third, it should be noted that our analysis was based on the PRS constructed from common GWAS-identified risk variants. Other risk variants (i.e., rare) with epistatic effects, or expression of inflammatory genes (i.e., COX1 , COX2 , LOX5 and ALOX5AP and correlated oncogenes), together with advanced approaches (e.g., interaction-aware [ 54 ] or machine learning-based PRSs), could be used to improve PRS-based analysis in future studies. Therefore, further molecular and cellular analyses are required to incorporate them and accurately assess their functional consequences with PRS.…”

The Normal, the Radiosensitive, and the Ataxic in the Era of Precision Radiotherapy: A Narrative Review

“…NTCP models can also include dose-modifying factors explicitly describing individual patients’ radiosensitivity, such as polygenic risk scores or results from biomarker assays [ 43 , 44 , 45 , 46 ]. These biologically-extended NTCP models can drive personalized decision-making and personalized optimization of treatments by setting goals on dose distributions that are tuned to the patient’s own genetics/biology [ 45 , 47 ].…”

“…Other possible examples of such genetically extended NTPC models are those developed by Rancati et al for predicting five urinary and rectal symptoms after radiotherapy for prostate cancer [ 47 ]. These models include clinical/treatment-related features (diabetes, presence of mild symptoms before radiotherapy, a previous transurethral resection of the prostate, a previous prostatectomy) and a polygenic risk score calculated from a cluster of SNPs identified within a validation study in the REQUITE population [ 43 , 44 ]. Deneuve et al proposed the inclusion of results from the RADIODTECT© assay in NTCP models predicting acute, moderate, and severe oral mucositis and dysphagia after postoperative irradiation for head and neck cancers [ 46 ].…”

“…Even if most studies have produced statistically significant results, the models were often not applied to validation cohorts ( Table 4 ). The most significant progress in these areas has been made in identifying specific SNPs associated with late toxicities in breast and prostate cancer via the European REQUITE study [ 40 , 43 , 44 ]. The French study PROUST is in progress and centralizes data and blood samples to validate these predictive markers of individual RT-induced toxicity [ 112 ].…”

Transfer Learning with False Negative Control Improves Polygenic Risk Prediction

“…To accommodate the potential ancestral heterogeneities between base sample and target sample, in this work, we treat the GWAS summary information obtained in the base data as knowledge learned from a pre-trained model, and adopt a transfer learning framework to leverage the knowledge learned from the relatively ancestry-diversified base data to build the PRS of target individuals. Our proposed trans-learning framework consists of two main steps: (1) conducting false-negative control (FNC) screening to extract useful knowledge from the base data; (2) parsimonious PRS model that includes fewer SNPs whilst maintaining the same or higher predictive R 2 can facilitate PRS interpretation and enable downstream analyses with more complex modeling techniques, such as incorporating SNP-SNP interactions in polygenic prediction models as discussed in [15,16]. We also explore interactive models with the lipid prediction in the CoLaus/PsyCoLaus study.…”