The results of a population-based study suggest that lipid levels and other electronic health record-derived clinical data points, combined with exome sequencing for familial hypercholesterolemia (FH) variants, can be used to risk-stratify individuals for adverse cardiovascular outcomes.
Prashant Patel, MD, presented the study’s findings Tuesday at Scientific Sessions.
“Our goal in this study was to employ a population-based approach using EHR-based algorithms to risk-stratify hyperlipidemia using available clinical variables,” said Patel, of Geisinger Clinic and Medical Center in Danville, Pennsylvania. “In this process, we envisaged identifying the hidden burden of familial hypercholesterolemia and the associated trends of major adverse cardiovascular events. We also sought to study the role of genomic testing in the prediction of clinical outcomes.”
A Geisinger Health System cohort of 41,649 patients was included in the study (57.61 percent female, median age 62 years). The researchers deployed five phenotyping methods to risk-stratify severe hypercholesterolemia in the cohort. These included an LDL-C cutoff, addition of clinical data (family and personal history of heart disease) from the EHR and presence of FH pathogenic variants. A priori associated confounders were used for multivariate analyses using binary logistic regression for outcome measures of myocardial infarction, coronary interventions and a composite ischemic heart disease burden.
Patel reported that baseline characteristics identified by all hypercholesterolemia phenotype definitions showed that each was associated with a high prevalence of the outcome measures. When compared to mutation-negative individuals with LDL-C ≤ 130mg/dl, there was a stepwise increase in the odds for outcomes, starting with a simple LDL-C based cutoff (≥190mg/dl) to addition of data elements from EHR and finally FH mutation positivity. Subjects with LDL ≥190 and FH mutation-positive status identified the highest risk with an OR of 28.9 (95 percent CI 15.0-55.8, p<0.0001), 26.8 (13.9-51.6, p<0.0001) and 44.3 (23.7-82.9, p<0.0001) for MI, coronary interventions and IHD burden, respectively.
“These findings suggest an important role for biochemical and genetic phenotype determination,” said Vishal C. Mehra, MD, PhD, a senior author of the study. “The ever-increasing use of EHR systems may broaden the appeal of well-designed, population-based screening to identify high-risk groups, which can then be preferentially targeted by cascade genetic screening or aggressive, novel therapies to improve outcomes.”