Phenomapping of subgroups in high-Lp(a) patients: a data-driven cluster analysis in RED-CARPET study.

BACKGROUND

The association between high levels of lipoprotein (a) [Lp(a)] and cardiovascular disease (CVD) is influenced by clinical characteristics. We aimed to explore the heterogeneity in high Lp(a) population with different clinical phenotypes and their relationship with atherosclerosis cardiovascular disease (ASCVD) risk.

METHODS AND RESULTS

We included 11,629 participants with Lp(a) measurement in RED-CARPET Study (ChiCTR2000039901) from the First Affiliated Hospital of Sun Yat-Sen University. The primary outcome was the occurrence of ASCVD events. The k-means clustering method was performed for baseline variables in participants with high Lp(a) levels (Lp(a) ≥ 50 mg/dL). Multivariate logistic regression model was used to assess the association between high Lp(a) level and ASCVD across clusters, with the low-Lp(a) group (Lp(a) < 50 mg/dL) serving as reference. Propensity score matching (PSM) was used to validate thefindings. High-Lp(a) group was categorized into four clusters: cluster 1 (dyslipidemia); cluster 2 (aged females); cluster 3 (males with an unhealthy lifestyle) and cluster 4 (anemia, renal insufficiency and hypercoagulability). Patients in different clusters exhibited differences in ASCVD risk. Patients with high-Lp(a) had significantly highest risk for ASCVD in cluster 3 (OR 2.12, 95% CI 1.62-2.76, p < 0.001) after adjusting for traditional risk factors. However, no significant association was observed in cluster 4 (OR 0.82, 95% CI 0.58-1.16, p = 0.233). These findings remained consistent after PSM.

CONCLUSIONS

Using a data-driven approach, high-Lp(a) patients can be stratified into four phenotypically distinct subgroups with different ASCVD risk.