Measured Lp(a) levels and LPA GRS similarly associated with ASCVD risk in primary prevention setting
Clinical utility of LPA genetic characterization for primary prevention of atherosclerotic cardiovascular disease
Presented at the virtual EAS 2020 by Mark Trinder (Broad Institute, Boston, MA, USA)
Introduction and methods
Variants in the LPA gene, which encodes lipoprotein(a), predominately determine Lp(a) levels in plasma. Mendelian randomization analyses have suggested that elevated Lp(a) levels are a causal risk factor of ASCVD. Conclusive testing of this association between Lp(a) and CV risk is currently in progress. However, recommendations for testing of Lp(a) levels in regard to primary prevention for ASCVD remain contentious.
This study assessed whether measurement of Lp(a) levels and/or LPA genetic risk score (GRS) have clinical utility in risk prediction of ASCVD in the primary setting.
Firstly, 374,099 individuals from the UK biobank cohort that had an Lp(a) measurement, genotyping info and had less than third degree relatedness were selected. LPA genetic risk score was determined using 43 LPA variants. Subjects were grouped into different ethnic groups (South Asian (SAS), European/white (EUR), East Asian (EAS), African (AFR) descent, and admixed or unknown) and their LPA GRS was tested for association with measured Lp(a) levels.
Secondly, 283,540 individuals from EUR descent from this cohort were tested for association of the LPA GRS and Lp(a) with incident ASCVD. Individuals using cholesterol-lowering medication were excluded.
Thirdly, individuals within this cohort that had no prior history of diabetes, <190 mg/dL LDL-c concentration and were classified as borderline to intermediate 10-year risk (5-20%) by Pooled Cohort Equations (n=113,703) or QRISK3 (n=144,350) were selected. This analysis assessed whether the addition of continuous levels of measured Lp(a) or LPA GRS improved risk prediction.
- Measured Lp(a) levels in ethnic groups are right skewed and was slightly higher in people from AFR descent. Data of LPA GRS show similar results. Correlation between Lp(a) levels and LPA GRS performs best in individuals from EUR descent (Spearman correlation 0.72) and poor in individuals from AFR ethnicity (Spearman correlation 0.07).
- Associations for the measured Lp(a) and the LPA GRS with ASCVD risk were similar (per 120 nmol/L increase in Lp(a): HR 1.26 95% CI:1.23-1.28 vs. HR 1.29, 95% CI: 126-1.33, respectively). When the LPA GRS was adjusted for Lp(a) levels, the HR for ASCVD risk was greatly attenuated (HR 1.06, 95% CI:1.01-1.11). Alternatively, this was not observed when Lp(a) levels were adjusted for the LPA GRS (HR 1.22, 95% CI:1.18-1.25).
- Measured Lp(a), LPA genetic risk score, and Lp(a) levels plus LPA GRS give a slight, but significant improvement to estimate the risk in individuals that were classified as borderline to intermediate risk for an ASCVD event by QRISK3 (AUROC by QRISK3 calculator: 0.640; adding Lp(a) to QRISK3 resulted in AUROC of 0.640, P=0.005; adding LPA GRS resulted in AUROC of 0.642, P=0.01; and adding Lp(a) plus LPA GRS gave AUROC of 0.642, P=0.005).
An LPA genetic risk score is, relative to Lp(a) levels, a comparable prediction marker for incident ASCVD. LPA genetic risk score and measured Lp(a) or combined improve risk assessment in borderline to intermediate risk groups for primary prevention and point to Lp(a) as an risk enhancer.
-Our reporting is based upon the information provided at the EAS 2020 congress-