Leveraging Genetic Instrumental Variables and Sequencing Analysis to Identify a Prognostic Signature Based on Epithelial Cell Markers in Lung Adenocarcinoma

TL;DR

This study identifies a new prognostic model for lung adenocarcinoma based on genes expressed during malignant cell dedifferentiation.

Contribution

A novel prognostic signature using dynamically expressed genes in malignant alveolar type II cells for LUAD prognosis.

Findings

Pseudo-time analysis identified 3526 dynamically expressed genes during malignant AT2 cell dedifferentiation.

A four-gene prognostic model achieved AUC values of 0.649–0.675 for predicting LUAD patient survival.

High-risk patients had significantly poorer overall survival compared to low-risk patients in training and validation sets.

Abstract

The treatment and prognosis of lung adenocarcinoma (LUAD) remain challenging. The study aimed to identify prognostic genes and construct a prognostic model for LUAD.

After identifying malignant alveolar type II (AT2) cells using InferCNV, we applied CytoTRACE, pseudo‐time analysis, Mendelian randomization (MR), and univariate Cox regression analysis to identify prognostic genes. A prognostic model was then developed using an optimized subset of these genes, selected through the least absolute shrinkage and selection operator (LASSO) algorithm. Further analyses included Gene Ontology enrichment analysis and the construction of a protein–protein interaction (PPI) network.

Pseudo‐time analysis identified 3526 dynamically expressed genes during malignant AT2 cell dedifferentiation. Subsequent multi‐omics integration refined the gene selection, yielding four prognostic genes for the final predictive model. The resulting model achieved area under the receiver operating characteristic (ROC) curve (AUC) values of 0.649, 0.675, and 0.654 for predicting 1, 2, and 3‐year overall survival (OS) in the training set, respectively, and was successfully validated in two external cohorts at the corresponding time points. Moreover, survival analysis demonstrated that patients in the high‐risk group had significantly poorer OS than those in the low‐risk group, both in the training set and the validation sets (p < 0.01).

The study developed a novel signature based on genes dynamically expressed during malignant AT2 cell dedifferentiation, capable of predicting the prognosis of LUAD patients, and offered four accurate prognostic biomarkers (ADM, MARK4, PARVA, and RPS6KA1).

It begins with the identification of malignant AT2 cells, progresses through pseudo‐time analysis and Mendelian randomization to identify prognostic genes, and leads into the construction and validation of the prognostic signature. The visual story also includes functional exploration through Gene Ontology enrichment analysis and protein–protein interaction network analysis.

Figures