The progressive impact of obesity severity on systemic inflammation and atherogenic dyslipidemia: a multicenter cohort study
TL;DR
A composite score derived from universally available routine laboratory parameters — CORI_Ridge — derived from universally available routine laboratory parameters — integrates this signal into a single graded score with moderate out-of-sample discrimination for Grade III obesity; it does not, however, exceed the discriminative performance of a simple three-variable clinical model.
Abstract
Obesity-associated cardiometabolic risk involves a complex interplay of systemic inflammation and atherogenic dyslipidemia that escalates with increasing adiposity. This study investigated the relationship between obesity severity and established hematological inflammatory indices and lipid-based metabolic markers, and derived a composite score (CORI_Ridge) as a proof of concept for integrating these routine parameters. In this retrospective multicenter cohort study, 1,107 adults with obesity were stratified by BMI into Grade I ( n = 475), Grade II ( n = 396), and Grade III ( n = 236) categories. Inflammatory indices (NLR, PLR, MLR, SII, SIRI, AISI) and metabolic indices (Castelli Risk Index-1 and − 2, AIP) were calculated from routine complete blood count and fasting lipid parameters. CORI_Ridge was derived via Ridge regression with continuous BMI as the dependent variable. Because the score was derived in the same cohort, performance was evaluated not only apparently but with a strict out-of-sample framework: 50 repeated stratified 70:30 holdout splits in which z-score standardization, λ selection and weight derivation were repeated from scratch within each training partition. Progressive and significant increases were observed across obesity grades in neutrophil counts, platelet counts, SII, AISI, Castelli Risk Index-1, Castelli Risk Index-2, and AIP (all p < 0.001); metabolic indices showed a true dose-response gradient, whereas inflammatory indices escalated predominantly at Grade III. CORI_Ridge correlated most strongly with obesity severity (Spearman ρ = 0.420, p < 0.001) and achieved the highest apparent discrimination of Grade III obesity (AUC 0.796, 95% CI 0.763–0.829), significantly above every individual index (all DeLong p < 0.05). Out-of-sample, CORI_Ridge achieved a median AUC of 0.779 (IQR 0.763–0.793). A simple reference model of waist circumference, fasting glucose and triglycerides performed better (median AUC 0.827), and adding CORI_Ridge to it yielded a small increment (median AUC 0.845) that reached statistical significance in only 6 of 50 splits (median DeLong p = 0.38). Systemic inflammatory and atherogenic burden escalates progressively with obesity severity. CORI_Ridge — derived from universally available routine laboratory parameters — integrates this signal into a single graded score with moderate out-of-sample discrimination for Grade III obesity; it does not, however, exceed the discriminative performance of a simple three-variable clinical model and should be regarded strictly as a proof of concept. External validation against clinical outcomes is required before any clinical application can be considered.