Abstract 11533: Genetic Prediction Tool Using LPA Haplotypes Improves Identification of Patients with Elevated Lipoprotein(a)

• Published in: Circulation (New York, N.Y.) Vol. 144; no. Suppl_1; p. A11533
• Main Authors: Atri, Deepak, Mukamel, Ronen E, Loh, Po-ru, Blankstein, Ron, Gupta, Rajat M
• Format: Journal Article
• Language: English
• Published: Lippincott Williams & Wilkins 16.11.2021
• DOI: 10.1161/circ.144.suppl_1.11533
• ISSN: 0009-7322

IntroductionMultiple lines of evidence support the causal role of serum lipoprotein (a) [Lp(a)] in the pathogenesis of cardiovascular disease, including CAD/MI and ischemic stroke. Mendelian randomization studies demonstrate that genetic variation is the principal determinant of serum Lp(a) level, and that variants associated with higher Lp(a) are associated with higher vascular disease risk. Though 80-90% of variation in serum Lp(a) levels is genetically-determined, single nucleotide polymorphisms (SNPs) contribute a small fraction of this risk. SNPs, however, may be used to identify LPA alleles based on a reference panel. These LPA alleles are the primary genetic determinants of serum Lp(a). New methods that identify these alleles using extended SNP haplotypes may improve serum Lp(a) prediction and identify at-risk patients. HypothesisThe use of reference panels of extended LPA haplotypes can identify patients with elevated Lp(a) in a genotyped, hospital-based cohort. Methods & ResultsWe identified all genotyped subjects in the Massachusetts General Brigham Biobank with a history of CAD or ischemic stroke for genetic prediction of serum Lp(a) (n = 3202). The majority of these patients (n = 2578, 80.5%) did not have a prior Lp(a) measurement as part of clinical care. We computed genetic predictions of Lp(a) in all patients by matching their SNP haplotypes to a reference panel of ~50,000 UK Biobank participants (and ~440,000 haplotypes) for whom LPA alleles had been matched to Lp(a) levels. The 90th percentile, 258 patients, had predicted Lp(a) of >75mg/dl (~159nM), equal to the 90th percentile of the general population. We observed strong correlation between predicted and clinically-measured Lp(a) (R2 = 0.75-0.92 across four clinical assays; p<0.0001 for all; n = 372). ConclusionsWe demonstrate the validity of a serum Lp(a) genetic prediction with very high fidelity in a disease cohort using a new method of imputed SNP haplotypes. Using LPA alleles, as determined by SNPs, we identified patients who are likely to have elevated serum Lp(a) without a prior measurement. This method may be applied to biobank populations to identify patients at risk for cardiovascular disease mediated by elevated serum Lp(a).