Non-invasive PECS model for detection of combined post-capillary pulmonary hypertension

TL;DR

This study introduces a non-invasive model called PECS to help detect a severe type of pulmonary hypertension, reducing the need for invasive procedures.

Contribution

The novel contribution is the development and validation of the non-invasive Predictive Echocardiography Cpc-PH Score (PECS) model.

Findings

The PECS model achieved an AUC of 0.761 in predicting Cpc-PH.

It demonstrated a sensitivity of 66.7% and specificity of 72.0%.

5-fold cross-validation confirmed the model's performance with an AUC of 0.752 ± 0.070.

Abstract

Combined post-capillary pulmonary hypertension (Cpc-PH) is a severe form of pulmonary hypertension associated with high morbidity and mortality. Early identification and intervention are crucial but challenging due to the invasive right heart catheterization (RHC). This study aimed to develop and validate a non-invasive diagnostic model, the Predictive Echocardiography Cpc-PH Score (PECS), using echocardiographic parameters to facilitate detection of Cpc-PH.

A retrospective analysis encompassing 198 patients with suspected PH-LHD, admitted from July 2010 through December 2023, was executed. Patients were divided into Cpc-PH and Ipc-PH/No-PH groups based on RHC in accordance with the 7th World Symposium on Pulmonary Hypertension criteria for PECS model construction. Chi-square and L1-regularized backward elimination refined predictive indicators. Model efficacy and stability were appraised via receiver operating characteristic and 5-fold cross-validation.

The PECS model, incorporating a suite of indicators including valvular heart disease, left atrial systolic diameter, interventricular septal thickness, mitral valve E/Em ratio, left ventricular fractional shortening, and tricuspid regurgitation velocity, demonstrated good predictive performance, achieving an area under characteristic (AUC) of 0.761 (95% CI: 0.692–0.823, P < 0.001). It demonstrated a sensitivity of 66.7%, specificity of 72.0%, a positive predictive value of 72.9%, a negative predictive value of 65.7%, and an overall accuracy of 69.2%. A total of 5-fold cross-validation confirmed these findings, yielding an AUC of 0.752 ± 0.070.

The PECS model provides a non-invasive and precise approach to diagnosing Cpc-PH, potentially acting as a practical screening tool.

Figures