Baseline and Change in Peripheral Blood Transcriptome Enhance Outcome Prediction in Fibrotic Hypersensitivity Pneumonitis: A Prospective Multicenter Cohort Study.

Rationale: Data on the prognostic value of peripheral blood mononuclear cell (PBMC) expression profiles, when used in fibrotic hypersensitivity pneumonitis (fHP) patients as an adjunct to traditional clinical assessment in predicting disease progression, remains limited. Objectives: To determine whether a baseline and time-course transcriptomic signature in PBMC from patients with fHP can enhance progression-free survival (PFS) prediction and complement clinical risk stratification. Methods: The prospective multicenter study cohort included 133 participants with fHP. Lasso regression was employed to create a baseline and time-course (baseline to 12-month change) gene signature. We developed multivariable models incorporating clinical variables (age, sex, smoking status, exposure history, FVC% and quantitative measurements of lung fibrosis-derived from data-driven textural analysis, DTA) both with and without the baseline and time-course gene expression, and evaluated these models using receiver operating characteristic curves. Kaplan-Meier curves displaying high and low gene expression risk scores are presented. Results: The addition of a baseline gene expression profile to the logistic regression model of 24-month PFS using baseline clinical parameters markedly improved the predictive accuracy, increasing the area under the curve (AUC) from 0.77 to 0.93. Similarly, a logistic regression model of 12-month PFS using age, sex, smoking status, exposure history, and changes in FVC% and DTA over 12 months had an AUC of 0.81, which improved to 0.95 with the inclusion of changes in gene expression over the same period. A difference in PFS was noted when the cohort was divided based on whether their gene principal component analysis (PCA) scores were above or below the median. At baseline, patients with higher gene PCA scores had lower median survival than those with lower scores (267 days vs. not available; P = 0.008). Similarly, when assessing the change in gene expression from baseline to 12 months, patients with higher gene PCA scores had lower median survival than those with lower scores (203 days vs. 449 days; P = <0.001). Conclusions: In fHP, a risk-indicative gene expression signature in peripheral blood at baseline, along with its changes over 12 months, predicts disease progression at 24 months and during the subsequent 12 months, respectively. Prognostic models based on gene expression and clinical factors strongly outperform models based solely on clinical factors.