Phenotypic Clustering of Idiopathic Pulmonary Arterial Hypertension: Insights Into Pulmonary Vascular and Cardiometabolic Co-Morbidity Trajectories

Cihangir Kaymaz, Barkin Kultursay, Hacer Ceren Tokgoz, Seda Tanyeri, Cagdas Bulus, Berhan Keskin, Caglar Emre Cagliyan, Metehan Kibar, Can Erdem, Aziz Vezir, Ipek Akkus, Muhammet Bulut, Aykun Hakgor, Ibrahim Halil Tanboga, Nihal Ozdemir 

https://doi.org/10.1002/pul2.70362 

Abstract

Idiopathic pulmonary arterial hypertension (IPAH) exhibits significant clinical heterogeneity, necessitating a precision medicine approach. This study aimed to identify distinct IPAH phenotypes using machine learning-based clustering and to evaluate their longitudinal therapeutic responses and long-term survival. We analyzed 297 IPAH patients using hierarchical agglomerative clustering based on baseline demographics, hemodynamics, and comorbidities. The cohort was characterized by two distinctive yet overlapping phenotypic patterns: Cluster 1 (n = 198, “Pulmonary Vascular Phenotype”), comprising younger patients (median age: 40) with low comorbidity burden but severe hemodynamic impairment, and Cluster 2 (n = 99, “Cardiometabolic Phenotype”), consisting of older, predominantly female patients (median age: 65) with high prevalence of obesity (78%), diabetes (73%), and hypertension (57%).

Despite comparable baseline multiparametric risk scores, Cluster 1 exhibited more impaired echocardiographic right ventricular-arterial coupling, lower left ventricular filling pressure and higher pulmonary vascular resistance (p < 0.001). Triple combination therapy was significantly more frequent in Cluster 1 (41.6% vs. 25.5%, p = 0.048). Linear mixed-effects modeling over 12 months demonstrated that Cluster 1 achieved superior functional recovery (6-min walk distance increase: 75 m vs. 42 m, p < 0.001) and more profound neurohormonal stabilization (Final NT-proBNP: 55 vs. 145 pg/mL, p = 0.017). However, during a median follow-up of 763 days, 5-year survival was similar between groups (Log-rank p = 0.452), and cluster membership was not associated with mortality in adjusted Cox regression (HR: 0.71; 95% CI: 0.47–1.07; p = 0.099). These findings suggest that while vascular phenotypes demonstrate more robust responses to therapy, mortality is driven by distinct factors—hemodynamic severity in younger patients versus metabolic frailty in the comorbid population.

Read the full research article