Spherical Coordinate System for Dyslipoproteinemia Phenotyping and Risk Prediction

Area of Science:

• Cardiovascular Medicine
• Biostatistics
• Lipidomics

Background:

• Residual atherosclerotic cardiovascular disease (ASCVD) risk factors remain incompletely understood.
• Specific dyslipoproteinemia types may enhance ASCVD risk assessment.
• Current risk models may not fully capture lipid-related risks.

Purpose of the Study:

• To develop a novel, automated phenotyping and risk assessment system for ASCVD.
• To classify dyslipidemia using standard lipid panel parameters.
• To refine ASCVD risk prediction by incorporating detailed lipoprotein phenotypes.

Main Methods:

• Utilized NHANES data (37,056 participants, 1999-2018) to develop a 3D dyslipidemia phenotype classification.
• Employed ARIC data (14,632 participants) for logistic regression model training and validation.
• Converted Cartesian coordinates to spherical coordinates for risk modeling, validated with UK Biobank data (354,344 participants).

Main Results:

• Defined nine distinct lipidemia phenotypes based on HDLC, non-HDLC, and TG levels.
• Phenotypes correlated with metabolic syndrome prevalence, Pooled Cohort Equation (PCE) scores, and ASCVD-free survival.
• A logistic regression model using spherical coordinates achieved predictive accuracy comparable to PCEs.

Conclusions:

• Demonstrated the utility of a multidimensional coordinate system for novel lipoprotein phenotyping and disease association studies.
• Developed a fully automated composite risk marker based on spherical coordinates for ASCVD risk.
• The novel risk marker showed performance nearly equivalent to PCEs, offering a lipid-centric, automated alternative.