Predicting survival across chronic interstitial lung disease: the ILD-GAP model

Area of Science:

• Pulmonology
• Clinical Epidemiology
• Medical Statistics

Background:

• Chronic interstitial lung disease (ILD) presents heterogeneous patient- and disease-specific variables, complicating accurate risk prediction.
• The GAP model, based on sex, age, and lung physiology, was previously validated for idiopathic pulmonary fibrosis mortality prediction.
• Evaluating the GAP model's utility in diverse chronic ILD populations is crucial for improving patient outcomes.

Purpose of the Study:

• To assess the accuracy of the GAP model in predicting mortality across various chronic interstitial lung disease subtypes.
• To compare the performance of the GAP model against novel prediction models within specific ILD cohorts and a combined cohort.
• To develop and validate a modified ILD-GAP Index for disease-specific survival estimates in a unified risk prediction model.

Main Methods:

• A cohort of 1,012 patients with various ILD subtypes (idiopathic pulmonary fibrosis, chronic hypersensitivity pneumonitis, connective tissue disease-associated ILD, etc.) was analyzed.
• The predictive performance of the established GAP model was compared with alternative models across ILD subtypes and the overall cohort.
• Longitudinal model validation was performed using patients with available follow-up pulmonary function data.

Main Results:

• The GAP model demonstrated good predictive performance (c-index, 74.6 in the combined cohort) across all evaluated ILD subtypes.
• Model performance remained consistent across different disease severity stages and during follow-up.
• A modified ILD-GAP Index was developed, incorporating disease subtype to enhance survival prediction accuracy for specific ILDs.

Conclusions:

• The GAP model accurately predicts mortality risk in patients with chronic interstitial lung disease.
• The developed ILD-GAP model provides accurate mortality predictions for major chronic ILD subtypes and across all disease stages.