Use of polygenic risk scores in the prediction of COVID-19 non-recovery

Résumé fourni par la source

Polygenic Risk Scores (PRS), which combine genetic risk from across the genome into a score for each individual, are being developed to improve screening, diagnosis, prognosis and treatment response across a range of diseases. We hypothesised that incorporating genetic risk for organ impairment might improve prediction of those at risk of post-COVID-19 conditions. We used data from genome-wide association studies for traits and diseases that share pathophysiology, clinical features, or were previously associated with COVID-19. We calculated PRS in the Post-HOSPitalisation COVID-19 (PHOSP-COVID) study to determine their association with self-reported non-recovery 12 months post COVID-19 (450 cases, 389 controls). We then tested whether using these PRS improved predictive ability when added to a clinical prediction model. PRS of several traits –BMI (p=0.0002), chest pain (p=0.0035), hypertension (p=0.0054), type 2 diabetes (p=0.0057), atrial fibrillation (p=0.006), chronic fatigue syndrome (p=0.0111), type 1 diabetes (0.0114), inflammatory bowel disease (p=0.0352), FVC (p=0.0374), anxiety/depression (p=0.0452) – were nominally associated with non-recovery at 12 months. A PRS of anxiety/depression improved the clinical prediction model (AUC: 0.714 [CI: 0.680 – 0.749] vs. 0.696 [0.661 – 0.731], p=0.018), increasing total prediction accuracy from 65% to 66.5%. Other traits conferred smaller increases in predictive performance. Whilst there was significant genetic overlap between COVID-19 non-recovery and a number of traits, inclusion of genetic risk in a clinical prediction model did not improve accuracy. The PRS will be further evaluated with organ-specific outcomes in post-COVID-19 conditions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.