A combined polygenic score of 21,293 rare and 22 common variants significantly improves diabetes diagnosis based on hemoglobin A1C levels

Abstract

Polygenic scores (PS), constructed from the combined effects of many genetic variants1, have been shown to predict risk or treatment strategies for certain common diseases2-6. As most PS to date are based on common variants7, the benefit of adding rare variation to PS remains largely unknown and methodically challenging. We developed and validated a method for constructing a rare variant PS and applied it to a previously identified clinical scenario, in which genetic variants modify the hemoglobin A1C (HbA1C) threshold recommended for type 2 diabetes (T2D) diagnosis6, 8-10. The resultant rare variant PS is highly polygenic (21,293 variants across 144 genes), depends on ultra-rare variants (72.7% of variants observed in <3 people), and identifies significantly more undiagnosed T2D cases than expected by chance (OR=2.71, p=1.51x10-6). A model combining the rare variant PS with a previously published common variant PS6 is expected to identify 4.9M misdiagnosed T2D cases in the USA, nearly 1.5-fold more than the common variant PS alone. These results provide a method for constructing complex trait PS from rare variants and suggest that rare variants will augment common variants in precision medicine approaches for common disease.