Cardiometabolic risk factor clustering in persons with spinal cord injury: A principal component analysis approach

Abstract

Context/Objective To identify cardiometabolic (CM) measurements that cluster to confer increased cardiovascular disease (CVD) risk using principal component analysis (PCA) in a cohort of chronic spinal cord injury (SCI) and healthy non-SCI individuals. Approach A cross-sectional study was performed in ninety-eight non-ambulatory men with chronic SCI and fifty-one healthy non-SCI individuals (ambulatory comparison group). Fasting blood samples were obtained for the following CM biomarkers: lipid, lipoprotein particle, fasting glucose and insulin concentrations, leptin, adiponectin, and markers of inflammation. Total and central adiposity [total body fat (TBF) percent and visceral adipose tissue (VAT) percent, respectively] were obtained by dual x-ray absorptiometry (DXA). A PCA was used to identify the CM outcome measurements that cluster to confer CVD risk in SCI and non-SCI cohorts. Results Using PCA, six factor-components (FC) were extracted, explaining 77% and 82% of the total variance in the SCI and non-SCI cohorts, respectively. In both groups, FC-1 was primarily composed of lipoprotein particle concentration variables. TBF and VAT were included in FC-2 in the SCI group, but not the non-SCI group. In the SCI cohort, logistic regression analysis results revealed that for every unit increase in the FC-1 standardized score generated from the statistical software during the PCA, there is a 216% increased risk of MetS (P = 0.001), a 209% increased risk of a 10-yr. FRS ≥ 10% (P = 0.001), and a 92% increase in the risk of HOMA2-IR ≥ 2.05 (P = 0.01). Conclusion Application of PCA identified 6-FC models for the SCI and non-SCI groups. The clustering of variables into the respective models varied considerably between the cohorts, indicating that CM outcomes may play a differential role on their conferring CVD-risk in individuals with chronic SCI.