PgmNr 40: Quantifying the polygenic contribution to variable expressivity in eleven rare genetic disorders.Authors:
M.T. Oetjens 1; M.A. Kelly 1; A.C. Strum 1; R.G.C. Regeneron Genetics Center 2; C.L. Martin 1; D.H. Ledbetter 1
View Session Add to Schedule
1) Geisinger Health System, Danville, PA; 2) Regeneron Genetics Center, Tarrytown, NY
Rare genetic disorders (RGDs) often exhibit significant variation in clinical severity among affected individuals. However, the genetic and environmental factors that contribute to variable expressivity are not well understood. Recently, the aggregate effect of common variation, quantified as polygenic scores (PGSs), has emerged as an effective tool for predictions of disease risk and trait variation in the general population. To determine if common variants contribute to variable expressivity of RGDs, we measured the effect of PGSs in RGDs affecting three quantitative phenotypes: low-density lipoprotein (LDL-C) levels, body-mass index (BMI), and height. From 92,455 patients in the DiscovEHR cohort, an unselected health system-based population, we identified rare pathogenic variants underlying 11 RGDs that affect these three phenotypes, including familial hypercholesterolemia (FH; LDLR and APOB), familial hypobetalipoproteinemia (FHBL; PCSK9 and APOB) for LDL-C, 16p11.2 deletions and duplications, melanocortin 4 receptor deficiency (MC4R) for BMI, and four sex-chromosome aneuploidies (47,XXX; 47,XXY; 47,XYY; 45,X) for height. We first characterized the strength of these rare variants and showed each has a large effect size (0.5 - 2.6 SD) on the affected trait. Next, we developed PGSs for the three quantitative phenotypes (PGSLDL-C, PGSBMI, and PGSHEIGHT). PGSs were strong predictors of phenotypic variance in the general population (R2 = 8 - 21%), and in some cases we observed an equivalence between the effect size of an extreme PGS and the rare variant. The effect size of highly penetrant rare variants in LDLR (p =1.83 x 10-28; n=146) and APOB (p = 1.59 x 10-7; n=87) are significantly greater than a PGSLDL-C in the 100th percentile. However, the BMI of individuals with rare pathogenic MC4R variants is approximately equal to variant-negative individuals in the ~99th percentile of PGSBMI. Our examination of PGSs within RGDs revealed a polygenic component to variable expressivity in most RGD/trait pairs tested, including FH caused by pathogenic LDLR variants (p = 7.69 x 10-3), obesity caused by pathogenic MC4R variants (p = 3.71 x 10-3, n=58), and tall stature caused by 47,XXY (p = 3.87 x 10-3, n=44) and 47,XXX (p = 1.33 x 10-3, n=42). These results demonstrate that common, polygenic factors often contribute in an additive fashion to variable expressivity in RGDs and PGSs may a useful metric for stratifying affected individuals by clinical severity.