Highlight
- A Bayesian joint mixed-effects model (Disease Progression Model, DPM) integrates longitudinal lung function data with mortality risk in fibrotic ILD.
- DPM addresses bias from mortality in estimating forced vital capacity (FVC) decline compared to traditional linear models.
- By capturing the nonlinear FVC trajectory, the model improves precision and reduces time to treatment effect detection.
- The approach could refine clinical trial analyses and population characterization in fibrotic interstitial lung diseases.
Study Background
Fibrotic interstitial lung diseases (ILDs) represent a heterogeneous group of chronic pulmonary disorders characterized by progressive fibrosis leading to declining lung function. Forced vital capacity (FVC) is widely used as a surrogate endpoint in clinical trials, as declining FVC correlates with disease severity and prognosis. However, these diseases also carry significant mortality risk, which complicates the longitudinal assessment of lung function because patients with lower FVC are at higher risk of death, creating potential biases in analyses that do not account for mortality. Accurately capturing the trajectory of lung function decline alongside mortality risk is vital for evaluating treatment efficacy, informing prognosis, and optimizing patient management.
Study Design
The authors developed a Bayesian joint mixed-effects Disease Progression Model (DPM) that simultaneously models the time-dependent FVC trajectory and hazard of ILD-related mortality. This modeling approach uses minimally informative prior distributions to avoid imposing strong assumptions while integrating survival data with longitudinal lung function measurements. The DPM was applied retrospectively to individual patient-level data pooled from prospective cohort studies of fibrotic ILD, encompassing a broad patient population with varied disease severity. This framework allows the disentanglement of FVC decline dynamics from the censoring effect imposed by mortality.
Key Findings
The joint DPM yielded a higher estimated annual rate of FVC decline at 6.0% compared with 4.7% per year using a conventional linear mixed-effects model analyzing FVC alone. This indicates that models ignoring mortality may underestimate disease progression due to informative censoring of rapidly declining patients who die earlier. The DPM also provided a more precise fit to the data and successfully replicated the nonlinear decline pattern observed clinically, capturing acceleration or deceleration phases often seen in fibrotic ILD progression.
In contrast to analysis strategies focused solely on change from baseline at fixed timepoints, modeling the entire longitudinal FVC trajectory increased the information extracted from individual patients. This approach reduces variability stemming from baseline measurement fluctuations and decreases the time needed to detect statistically significant treatment effects, potentially accelerating clinical trial timelines. By jointly modeling mortality, this method also prevents bias toward underestimating treatment benefit or disease progression caused by differential dropout due to death.
Expert Commentary
The integration of survival outcomes with repeated lung function measures addresses a fundamental methodological challenge in progressive pulmonary diseases. Previous approaches often treated death as a censoring event, but this can lead to misleading conclusions about lung function trends if mortality is linked to disease severity. The Bayesian joint modeling framework provides a statistically robust and clinically meaningful solution.
This model’s flexibility to reflect the nonlinear progression of fibrotic ILD aligns well with pathophysiological insights, where fibrosis may progress at uneven rates. The increased precision can improve decision-making about continuing or modifying therapy and enhances the power of clinical trials to detect true treatment effects. Limitations include the need for careful distributional assumptions, computational complexity, and the requirement for high-quality longitudinal data sets with mortality follow-up.
Conclusion
The presented Bayesian joint mixed-effects Disease Progression Model represents a significant advancement in modeling fibrotic ILD progression by jointly assessing lung function trajectory and mortality risk. Its ability to reduce bias and improve precision in estimating FVC decline and treatment effects supports optimized clinical trial design and potentially more accurate prognostication in clinical practice. Future research should focus on validating its utility in diverse ILD subtypes, integrating additional biomarkers, and exploring its prospective application in clinical trials.
Funding and ClinicalTrials.gov
The original study was conducted by the REMAP-ILD Consortium Investigators as cited; no specific funding details or trial registration information was provided in the summarized content.
References
Wendelberger B, Jensen TP, Quintana M, et al. A statistical model for lung function trajectory and mortality in patients with fibrotic interstitial lung disease. Am J Respir Crit Care Med. 2026 Jul 1;212(7):1533-1547. PMID: 42085272.
