Abstract
Background: Immunotherapy has revolutionized non-small cell lung cancer (NSCLC
) therapy, only a small proportion of patients demonstrate durable antitumor response due to the heterogeneity. Emerging evidence has linked inflammatory response to progression, anticancer therapeutic resistance, and poor prognosis of NSCLC. This study aimed to classify distinct inflammatory response-related subtypes and constructed a new risk model to predict the prognosis and guide immunotherapeutic treatments.
Methods:
The gene expression, somatic mutation and clinical parameters for patients with NSCLC were obtained from TCGA-LUAD, and were used as the training dataset. GSE41721 and GSE54467, from GEO database including gene expression and clinical parameters were used as external validation datasets. We distinguished the patients of TCGA-LUAD into three clusters by Consensus clustering analysis based on the “ConsensusClusterPlus” package. It was validated through clinical features, prognosis, tumor microenvironment, expression of immune checkpoints and somatic mutation profile in distinct inflammation-associated subtypes. The risk model was construct by the hub-gene, which were screened out from the inflammation-related genes by univariate Cox and LASSO Cox regression, and verified by time-dependent ROC and Kaplan–Meier analyses. In addition, IC50 was implemented with “oncoPredict” package and GDSC datasets to evaluate the difference of drug susceptibility, the immunotherapy response were investigated by the IMVigor210 datasets. Finally, the single cell RNA seq analysis was preformed to validate of inflammation genes expression pattern.
Results:
Our findings demonstrated that NSCLC can be devided into three subtypes by inflammatory response-related signature, namely, inflammation-low, inflammation-mid, and inflammation-high, each exhibiting distinct clinicopathological characteristics, prognostic implications, somatic mutation profile and tumor microenvironments. We have affirmed the reproducibility and predictability of this categorization. The inflammation-high subtype generally represents a poor prognosis characterized by high immune cell infiltration, high immune score, low tumor purity high expression of immune checkpoints and a high frequency of oncogene mutations. Conversely, the inflammation-low subtype exhibit favorable clinical outcomes, low immune cell infiltration, low immune score, high tumor purity low expression of immune checkpoints and a low frequency of oncogene mutations. Furthermore, we have developed an inflammatory response-related risk model that demonstrates robust efficiency in assessing prognosis, drug sensitivity and immunotherapy response.
Conclusions: In conclusion, we devided NSCLC into three subtypes and constructed a risk model based on the inflammatory response . This model was highly effective in predict the prognosis, as well as the immunotherapy response.