Interstitial lung disease progression after genomic usual interstitial pneumonia testing.
Sachin ChaudharyS Sam WeigtManuel L Ribiero NetoBryan S BennJanelle Vu PugashettiRebecca KeithArista ChandScott OhFayez KheirVijaya RamalingamJoshua SolomonRichart HarperJoseph A LaskyJustin M OldhamPublished in: The European respiratory journal (2022)
A genomic classifier for usual interstitial pneumonia (gUIP) has been shown to predict histologic UIP with high specificity, increasing diagnostic confidence for idiopathic pulmonary fibrosis (IPF). Whether those with positive gUIP classification exhibit a progressive, IPF-like phenotype remains unknown. A pooled, retrospective analysis of patients who underwent clinically indicated diagnostic bronchoscopy with gUIP testing at seven academic medical centers was performed. We assessed the association between gUIP classification and eighteen-month progression-free survival (PFS) using Cox proportional hazards regression. PFS was defined as the time from gUIP testing to death from any cause, lung transplant, ≥10% relative decline in forced vital capacity (FVC) or censoring at the time of last available FVC measure. Longitudinal change in FVC was then compared between gUIP classification groups using a joint regression model. Of 238 consecutive patients who underwent gUIP testing, 192 had available follow-up data and were included in the analysis, including 104 with positive gUIP classification and 88 with negative classification. In multivariable analysis, positive gUIP classification was associated with reduced PFS (HR 1.58, 95% CI 0.86-2.92; p=0.14), but this did not reach statistical significance. Mean annual change in FVC was -101.8 mL (95% CI -142.7 mL, -60.9 mL; p<0.001) for those with positive gUIP classification and -73.2 mL (95% CI -115.2 mL, -31.1 mL; p<0.001) for those with negative classification (difference 28.7 mL; 95% CI -83.2 mL, 25.9 mL; p=0.30). Genomic UIP classification was not associated with differential rates of PFS or longitudinal FVC decline in a multi-center ILD cohort undergoing bronchoscopy as part of the diagnostic evaluation.
Keyphrases
- deep learning
- machine learning
- idiopathic pulmonary fibrosis
- interstitial lung disease
- systemic sclerosis
- artificial intelligence
- free survival
- end stage renal disease
- clinical trial
- gene expression
- intensive care unit
- ejection fraction
- chronic kidney disease
- randomized controlled trial
- data analysis
- dna methylation
- acute respiratory distress syndrome
- study protocol
- peritoneal dialysis
- electronic health record
- mechanical ventilation
- respiratory failure