Computed tomography-derived cardiac parameters versus biochemical inflammatory indices for discriminating disease...

It is unclear whether haematological inflammatory indices or CT-derived cardiac parameters add discriminative information beyond standard clinical variables in obstructive sleep apnoea syndrome (OSAS). We compared both modalities, alone and combined, against age, sex and body mass index (BMI) for identifying severe OSAS and associated cardiovascular comorbidity.

This was a retrospective cross-sectional study of 401 participants (240 OSAS, 161 controls). Four nested logistic regression models were fitted in the OSAS cohort: M0 (age, sex, BMI); M1 (M0 plus biochemical indices CAR, NLR, PLR, SII); M2 (M0 plus CT-derived cardiac parameters: pulmonary artery [PA] diameter, PA/aorta ratio, RV/LV ratio, ventricular wall thicknesses); and M3 (M0 plus CT plus biochemical). We compared discrimination (AUC, DeLong), reclassification (continuous NRI, IDI) and model fit (likelihood ratio tests), with internal validation by 500-iteration bootstrap optimism correction and repeated tenfold cross-validation. Missing predictor data were handled by multiple imputation (MICE, m = 20) as the primary approach. Cardiovascular outcomes were defined from echocardiography to avoid predictor-outcome overlap.

For severe OSAS, M2 improved on M0 (AUC 0.755 vs 0.716; LR p = 0.005; NRI + 0.60 [95% CI 0.35-0.83]; IDI + 0.055 [0.027-0.083]), whereas M1 did not (LR p = 0.62). M3 added nothing over M2 (ΔAUC = 0.007; LR p = 0.63). Optimism-corrected AUCs were 0.705, 0.695, 0.721 and 0.716 for M0-M3, so the incremental gain was modest. Nocturnal hypoxaemia indices discriminated severe OSAS far better than CT (ODI AUC 0.888 vs PA diameter 0.699), and once ODI and T90 were available CT added little (ΔAUC < 0.01). CT parameters gave modest reclassification for hypertension, coronary artery disease and the composite cardiovascular outcome; biochemical indices did not.

CT-derived cardiac parameters, particularly PA diameter, added discrimination over age, sex and BMI for severe OSAS; haematological indices did not. When thoracic CT is already available for another indication, opportunistic measurement of PA dimensions may complement clinical risk stratification. These findings are exploratory and require prospective external validation.