Abstract| Volume 28, SUPPLEMENT 1, S24, April 2020

Polygenic risk score and its potential to improve diagnostic ability in knee and hip osteoarthritis

      Purpose: Osteoarthritis (OA) is the most common joint disorder worldwide, and due to pain and disability it markedly decreases the quality of life of affected patients. Estimates of heritability of hip osteoarthritis (HipOA) and knee osteoarthritis (KneeOA) have fluctuated around 40 to 60%. A polygenic risk score (PRS) is a practical tool used by precision medicine that stratifies an individual's genetic risk of disease using common single nucleotide polymorphisms (SNPs). Applications toward the clinic are currently explored in a diverse number of diseases and are reaching clinical application fast (ref). The value and applicability of a PRS is dependent on the genetic knowledge of the disease under study. The largest genome-wide association study (GWAS) meta-analysis in OA has recently been performed by the Genetics of Osteoarthritis (GO)-consortium, and discovered >100 loci associated with OA. The purpose of this study was to evaluate the ability of the 63 SNPs as a PRS in predicting hip and knee OA prevalence and incidence. This project is part of the ErasmusMC project “Genotyping On All Patients (GOALL)” , in which the aim is to genotype all patients coming into the clinic and return valuable information back to the patient and/or clinician.
      Methods: Data from a total of 7,983 individuals from the Rotterdam Study-I was used. Knee and hip OA was defined as a KL-score of 2 or higher. We studied prevalent knee (n= 1177 cases) as well as hip (n=561 cases). In addition, incident knee and hip OA was examined during a mean follow-up time of mean 10 years (n=463 for incident knee, n=404 for incident hip). We constructed a 43-SNPs PRS for HipOA and a 23-SNPs PRS for KneeOA. Eleven SNPs are shared in both scores. Standardized PRSs (zPRS) were constructed for each subject, weighted by the effect sizes reported by the GO Consortium. Age and sex adjusted logistic regression models were applied to estimate the odds ratios per standard deviation (SD) of zPRS. We also assessed the PRS performance in the highest ten percentiles. All primary analyses were done in the Rotterdam Study-I (RS-I) cohort. Subsequently, external validation of the PRS is being performed in the Cohort Hip and Cohort Knee (CHECK) study and RS cohorts II and III.
      Results: The mean zPRS was -0.004 (SD=1) in controls and 0.033 (SD=0.99) in prevalent HipOA patients and was-0.03 (SD=1) in unaffected participants and 0.13 (SD=1) prevalent KneeOA cases. The zPRS were averaged to 0 with a SD of 1. The zPRS for HipOA significantly predicted prevalent HipOA (p=6.7 × 10-4 ; OR=1.1) and incident HipOA (6.2 × 10-4; OR=1.4). The zPRS for KneeOA significantly predicted prevalent KneeOA (p=1.5 × 10-6; OR=1.19) but failed to reach significance for incident KneeOA (p=0.08; OR=1.11). Subsequently, we examined individuals in the top 10 percentile of the polygenetic risk score and compared them with the rest of the population for incident OA risk. We observed an almost 2 times higher risk for HipOA incidence (p=7.7 × 10-4; OR=1.95) and 50% higher risk for KneeOA incidence (p=3.7 × 10-2; OR=1.55).
      Conclusions: Our results show that the OA polygenetic risk score is associated with prevalent and incident Hip and Knee OA. Importantly, individuals with a PRS in the highest 10% of the population, had an almost 2 times higher risk for incident hip OA, and 50% increased risk for knee OA. To evaluate the power of PRS fully, future studies should focus on age of onset of disease. In addition, the predictive power of genetics can be different across (patient) populations and ongoing efforts will address this by validating the predictive power of the genetic risk scores in diverse populations.