Background and Aims

Whereas the identification of causal mutations leads to a molecular diagnosis of familial hypercholesterolemia (FH), it is now established that the cumulative effects of common SNPs on LDL-cholesterol (LDL-c) can phenocopy the FH conditions. Polygenic Risk Scores (PRSs) have been developed using few to millions of SNPs to identify patients with polygenic-FH. We here set up and compared the performance of several PRSs for the diagnosis of polygenic-FH.

Methods

Four PRSs for LDL-c were set up using PLINK, PRSice2, Lassossum and LDpred2 softwares, including 165 to 1633 SNPs sequenced using the Dysliseq panel. These PRSs were tested in 758 hypercholesterolemic patients, and 1082 French healthy controls and then compared with previously described PRSs. The fraction of LDL-c variance explained by PRSs was defined in controls and the ability of PRS to discriminate mutation negative patients (FH/M-) vs. healthy individuals was determined using ROC curves. Patients with PRSs > 95^th percentile of controls were considered as polygenic-FH.

Results

The 165-SNP PRS developed with PLINK explains the highest part of LCL-c variance (adjusted R²: 0.187) and shows the best diagnosis ability (AUC: 0.759, significantly over performing previously-published PRSs: p-value < 1x10^-5). Within FH/M- patients, 22% of polygenic forms were detected using this 165-SNP PRS versus 14% with the previously-published 12-SNP PRS (p-value: 5.41x10^-5).

Conclusions

Notably, this new PRS improves the diagnosis ability and is compatible with routine genetic diagnosis. Indeed, the identification of polygenic-FH allows the etiologic diagnosis in FH/M- and the identification of an additional cause of hypercholesterolemia in FH/M+.

Background and Aims

Common cardiovascular risk equations are imprecise for heterozygous Familial Hypercholesterolemia (HeFH). Coronary Artery Calcium (CAC) score could help to better stratify the risk of major cardiovascular events (MACE). We investigated the additional contribution of CAC Score to SAFEHEART risk equation (SAFEHEART-RE) for MACE prediction in HeFH.

Methods

We analyzed data from primary prevention HeFH patients undergoing CAC quantification from two ongoing national registries , REFERCHOL and SAFEHEART. CAC score was expressed as log(CAC + 1).We used probability-weighted Cox proportional hazard models to estimate hazard ratios (HR). Area under the receiver operator characteristic curve (AUC) and net reclassification improvement (NRI) were used to compare incremental contribution of CAC score to SAFEHEART-RE for MACE prediction. MACE were defined as coronary heart disease, stroke or transient ischemic attack, peripheral artery disease, resuscitated sudden death and cardiovascular death.

Results

We included 1424 patients (age 48.9±12.8, men 45.9%).

After a 2.4-years follow-up, MACE occurred in 70 subjects. The addition of log(CAC+1) to SAFEHEART-RE was associated with an improved prediction of MACE in intermediate-risk (from HR 3.2 [95%CI 1.77-5.59] to HR 8.18 [95%CI 3.26-20.37]) and in high-risk subjects (from HR 3.5 [95%CI 1.93-6.30] to HR 20.21 [95%CI 8.58-47.6]) (log-rank p <0.0001). The c-statistics confirmed a significant improvement in MACE prediction by the addition of log(CAC+1) to SAFEHEART-RE (AUC 0.896 [0.889-0.903]) versus SAFEHEART-RE alone (AUC 0.859 [0.852-0.866]) (p= 0.004). The addition of CAC score was associated with an overall NRI of 46.8%.

Conclusions

Identification of very-high risk HeFH patients is possible by combining information from SAFEHEART-RE and CAC score.