Residual risk identified in routine noninvasive follow-up assessments in pulmonary arterial hypertension

Abstract

Background: The 2022 ESC/ERS guidelines on pulmonary hypertension recommend noninvasive risk assessments based on three clinical variables during follow-up in patients with pulmonary arterial hypertension (PAH). We set out to test whether residual risk can be captured from routinely measured noninvasive clinical variables during follow-up in PAH.

Methods: We retrospectively studied 298 incident PAH patients from a German pulmonary hypertension centre who underwent routine noninvasive follow-up assessments including exercise testing, echocardiography, electrocardiography, pulmonary function testing and biochemistry. To select variables, we used least absolute shrinkage and selection operator (LASSO)-regularised Cox regression models. Outcome was defined as mortality or lung transplant after first follow-up assessment.

Results: 12 noninvasive variables that were associated with outcomes in a training sub-cohort (n=208) after correction for multiple testing entered LASSO modelling. A model combining seven variables discriminated 1-year (area under the curve (AUC) 0.83, 95% confidence interval (CI) 0.68-0.99, p=8.4×10^-6) and 3-year (AUC 0.81, 95% CI 0.70-0.92, p=2.9×10^-8) outcome status in a replication sub-cohort (n=90). The model's discriminatory ability was comparable to that of the guideline approach in the replication sub-cohort. From the individual model components, World Health Organization functional class, 6-min walking distance and the tricuspid annular plane systolic excursion to systolic pulmonary arterial pressure (TAPSE/sPAP) ratio were sensitive to treatment initiation. Addition of TAPSE/sPAP ratio to the guideline approach numerically increased its ability to discriminate outcome status.

Conclusion: Our real-world data suggest that residual risk can be captured by noninvasive clinical procedures during routine follow-up assessments in patients with PAH and highlights the potential use of echocardiographic imaging to refine risk assessment.

FIGURE 1

Study flowchart showing the selection of patients. 6MWD: 6-min walking distance; ALT: alanine aminotransferase; AST: aspartate aminotransferase; AUC: area under the curve; D_LCO: diffusing capacity of the lung for carbon monoxide; PH: pulmonary hypertension; PAH: pulmonary arterial hypertension; RAAi: right atrial area index; sPAP: systolic pulmonary arterial pressure; TAPSE: tricuspid annular plane systolic excursion; TLC: total lung capacity; WHO-FC: World Health Organization functional class.

FIGURE 2

Comparison of the areas under the curve (AUC) from receiver operating characteristic (ROC) with the corresponding 95% confidence intervals between the newly derived and COMPERA 2.0 model for the a) 1-year and b) 3-year outcomes after first follow-up assessment in the replication sub-cohort. ROC analyses were repeated using a four-strata approach to the newly derived model (based on quartiles) on c) 1-year and d) 3-year transplant-free survival. The grey dashed line marks the reference. TPR: true positive rate; FPR: false positive rate.

FIGURE 3

Transplant-free survival from first follow-up assessment according to the newly derived model (divided into four-strata based on quartiles). a) For the entire cohort of patients with pulmonary arterial hypertension (PAH), b) after exclusion of patients with congenital heart disease-associated PAH and c) after exclusion of all associated forms of PAH.

FIGURE 4

Sankey diagram illustrating the change of risk status according to the newly derived model (divided into four strata based on quartiles) from baseline to follow-up. The width of the bars is proportional to the number of patients. The grey section marks the patients without follow-up assessment, which were excluded from development of the model.

FIGURE 5

Kaplan–Meier survival plot of TAPSE to sPAP ratio from follow-up assessment. Variables were grouped by their optimal cut-offs (0.2343 mm·mmHg⁻¹) and whether the variable improved or worsened as compared to baseline (i.e., crossed the cut-off). Overall log-rank test is displayed. Comparisons of the curves for “improved” versus “worsened” were significant (p=0.003 for log-rank test). sPAP: systolic pulmonary arterial pressure; TAPSE: tricuspid annular plane systolic excursion.