Integrated radiomics and immune infiltration analysis to decipher immunotherapy efficacy in lung adenocarcinoma.

BACKGROUND

Research in recent years has witnessed unprecedented improvements in immunotherapy, especially immune checkpoint blockade (ICB) for the treatment of lung adenocarcinoma (LUAD) patients. Nevertheless, due to the heterogeneity of immunotherapy response, reliable biomarkers are urgently needed to guide precision cancer therapy. In this study, we aimed to identify immune subtypes in LUAD and develop a radiogenomic model to improve immunotherapy predictive accuracy.

METHODS

In this study, clinical data of LUAD patients were downloaded from The Cancer Genome Atlas (TCGA) databases, and immune subtypes were identified using the ConsensusClusterPlus package in R. Biological, genomic, and epigenomic distinctions were compared. The TCGA cohort and clinical cohort from the Third Xiangya Hospital were utilized to demonstrate no significant differences of survival probability between sexes. Feature extraction and definition were conducted from 103 computed tomography (CT) images from The Cancer Imaging Archive (TCIA) dataset via the "PyRadiomics" embedded in Python. A series of machine learning techniques were applied to build a radiogenomic model.

RESULTS

Two LUAD subtypes with different molecular and immune characteristics were identified. Significant differences in biological, genomic, and epigenomic distinctions among the two subtypes were observed (P<0.05). The immune subtype A participated in pathways related to immune activation and displayed a higher tumor microenvironment (TME) score (P<0.001) with a better prognosis of LUAD [overall survival (OS), P=0.037; disease-specific survival (DSS), P=0.034]. Besides, the model appears to show better fit for females (P=0.015) than for males (P=0.641). Our constructed radiogenomic model incorporating 12 radiomics features displayed satisfactory potential to facilitate the predictive accuracy of immunotherapy in LUAD [test area under the curve (AUC) =0.89; train AUC =0.95].

CONCLUSIONS

Our study presented a promising avenue to harness the rich radiomics data to identify the specific immune subtype and integrate it into the existing clinical decision-making system to facilitate the predictive accuracy of immunotherapy in LUAD.