Can Quantitative CT Scan Analysis of Pulmonary Parenchymal and Vascular Characteristics Help to Differentiate Among Various Phenotypes of COPD Associated With Pulmonary Hypertension?

BACKGROUND

Quantitative CT imaging has been applied to pulmonary vascular disease, but its role in characterizing phenotypes of COPD associated with pulmonary hypertension (COPD-PH) has not been evaluated systematically.

RESEARCH QUESTION: Can quantitative CT imaging reliably identify and differentiate major COPD-PH phenotypes?

STUDY DESIGN AND

METHODS

We conducted a retrospective cohort study of patients with COPD-PH at Temple University Hospital from 2010 through 2022. Patients were categorized into 3 phenotypes: pulmonary vascular (pulmonary vascular resistence [PVR] > 5 Wood units [WU]; FEV1 > 50% predicted), pulmonary parenchymal (PVR < 5 WU; FEV1 < 50% predicted), and mixed (PVR > 5 WU; FEV1 < 50%). Quantitative CT imaging-derived vascular measures (proportion of blood volume in pulmonary vessels with cross-sectional area of < 5 mm2 [BV5] and proportion of blood volume in pulmonary vessels with cross-sectional area of < 10 mm2 [BV10]) and parenchymal measures (regional lung volume, total emphysema volume, emphysema percentage) were compared using analyses of variance.

RESULTS

Ninety patients met inclusion criteria. Mean PVR was 9.5 WU, 7.5 WU, and 2.8 WU in the vascular, mixed, and parenchymal phenotypes, respectively; corresponding mean FEV1 values were 67.4%, 35.8%, and 22.3%. Survival analysis demonstrated significantly higher mortality in the mixed phenotype (31%) vs the vascular (14.2%) and parenchymal (6.7%) groups (P = .04, log-rank test). Cox regression confirmed increased mortality risk in the mixed phenotype (OR, 2.12; 95% CI, 1.17-3.86; P = .001). Quantitative CT imaging-determined phenotypes showed similar patterns, with mortality of 75%, 25%, and 7.1% in the mixed, vascular, and parenchymal groups, respectively. The BV5 to total vessel volume ratio predicted elevated PVR of > 5 WU (area under the receiver operating characteristic curve [AUC], 0.75; P < .001) with an optimal cutoff of 0.48. Emphysema percentage predicted severe airflow obstruction (FEV1 < 50%; AUC, 0.83; P = .001). Using quantitative CT imaging-derived thresholds, phenotype classification demonstrated high specificity (vascular, 86.2%; parenchymal, 89.7%; mixed, 100%) with sensitivities of 46.7%, 73.3%, and 7.1%, respectively.

INTERPRETATION: Our results show that quantitative CT imaging can differentiate COPD-PH phenotypes using vascular and parenchymal markers and may serve as a noninvasive tool to characterize disease heterogeneity.