DEVELOPMENT AND EVALUATION OF THE FIRST POLYGENIC RISK SCORE FOR PEDIATRIC NEPHROTIC SYNDROME IN A JAPANESE COHORT

Polygenic risk scores (PRS) integrate the effects of many genetic variants into a single quantitative measure of genetic susceptibility. They have become powerful tools for dissecting the genetic architecture of complex diseases and are increasingly considered for risk prediction and clinical translation. In pediatric nephrotic syndrome (NS), genome-wide association studies (GWAS) have identified multiple susceptibility loci, and PRS derived from these variants have shown significant associations with clinical outcomes in European cohorts. However, the performance and clinical utility of PRS in non-European populations remain largely unknown. Notably, no PRS has been developed or validated in Japanese children, despite potential population-specific genetic architecture. The incidence of pediatric NS varies by ancestry; populations in East Asia—including Japan—generally have higher incidence than those of European ancestry. We therefore sought to construct and evaluate the first PRS for pediatric NS in a Japanese cohort and to investigate its associations with key clinical characteristics.

PRS was constructed using PRS-CS with GWAS summary statistics of pediatric NS (902 cases and 6,558 controls). PRS-CS is a Bayesian polygenic risk scoring framework that estimates variant effect sizes from summary statistics under a continuous shrinkage prior while accounting for linkage disequilibrium via an external reference panel.　Twelve parameter settings were tested. For model selection, we randomly sampled 388 cases and 2,811 controls from the Japanese dataset, using 80% for training and tuning and 20% for independent testing. Predictive performance was evaluated using the area under the receiver operating characteristic curve (AUC). Associations between PRS and clinical features (age of onset, sex, relapse pattern) were assessed with regression models adjusting for the other clinical features and four genetic principal components.

The PRS significantly distinguished Japanese NS cases from population controls, achieving an AUC of 0.706 (95% CI: 0.688–0.724). At a cutoff value of 0.104, sensitivity was 60.1% and specificity 71.9%, indicating moderate discriminatory ability. Clinical association testing revealed that higher PRS was inversely correlated with age of onset: each +0.1 increase in PRS corresponded to an approximately 1.5-month earlier disease onset (P=0.045). No significant associations were observed between PRS and sex (OR=0.946, 95%CI: 0.882–1.014, P=0.119) or relapse pattern (OR=1.009, 95%CI: 0.930–1.094, P=0.836).

We report the first PRS for pediatric NS in a Japanese cohort. The PRS showed moderate case–control discrimination and a reproducible association with age at onset, consistent with European data. These findings indicate that PRS derived from Japanese GWAS data capture biologically meaningful risk in this population. Establishing feasibility in Japanese children lays groundwork for clinical translation, including early risk stratification and prognosis. Larger, multi-ethnic studies are needed to validate and refine predictive models and to determine clinical utility in pediatric NS.