Abstract
Background
Idiopathic pulmonary fibrosis (IPF) is one of the interstitial lung diseases (ILDs) with poor prognosis. Multiple regulated cell death (RCD) pathways are involved in regulating the progression of pulmonary fibrosis at different stages.
Methods
A total of 20 RCD pathways and crucial regulatory genes were collected from available literature. The study initially elucidated the profiling of 20 kinds of RCD pathways in normal and fibrotic lung tissues based on the scRNAseq dataset and bulk-RNAseq datasets. IPF-related prognostic genes were identified based on differential expression analysis, univariate Cox regression analysis, the "Scissor" program, and the "Findmarkers" program. A combination of 101 distinct machine-learning algorithms was utilized to develop a prognostic signature.
Results
The study developed a cell death index (CDI) signature consisting of 12 genes. As proven by independently validated datasets, IPF patients with high CDI scores were more likely to have a shorter overall survival (OS). By employing several machine-learning algorithms, the CDI signature was able to accurately predict the prognosis of IPF patients, with higher predictive performance than other feature factors including age, gender, and GAP score. Furthermore, drug sensitivity analysis suggested the guiding role of CDI signature in the treatment of IPF patients. Ultimately, to enhance its practical application in clinical settings, a nomogram was developed by integrating CDI with relevant clinical characteristics, resulting in a very accurate predicting tool.
Conclusion
In summary, we have developed a novel CDI model that effectively forecasts the clinical prognosis of IPF patients by integrating various cell death patterns.