Conclusion
PECMS is an effective predictor of prognosis and drug sensitivity in pancreatic cancer patients. KLHL32 may play an important role in the construction of ECM, and the mechanism is worth further study.
Methods
GSVA score calculation and clustering were performed in TCGA-PAAD patients. Lasso regression was used to construct the PECMS model. The association between PECMS and patient survival was analyzed and validated in the CPTAC-3 dataset of TCGA and our single-center retrospective cohort. The relationships between PECMS and features of the matrix microenvironment were analyzed. Finally, PECMS feature genes were screened and verified in pancreatic cancer specimens to select key genes associated with the ECM microenvironment. Result: The survival of the PECMS-high group was significantly worse. The PECMS-high group showed higher oxidative stress levels, lower levels of antigen presentation- and MHC-I molecule-related pathways, and less immune effector cell infiltration. Data from IMvigor-210 cohort suggested that PECMS-low group patients were more sensitive to immune checkpoint blockers. The PECMS score was negatively correlated with chemotherapy drug sensitivity. The negative association of PECMS with survival and drug sensitivity was validated in our retrospective cohort. KLHL32 expression predicted lower oxidative stress level and more immune cells infiltrate in pancreatic cancer.
Purpose
Pancreatic cancer is one of the deadliest cancers worldwide. The extracellular matrix (ECM) microenvironment affects the drug sensitivity and prognosis of pancreatic cancer patients. This study constructed an 8-genes pancreatic ECM scoring (PECMS) model, to classify the ECM features of pancreatic cancer, analyze the impact of ECM features on survival and drug sensitivity, and mine key molecules that influence ECM features in pancreatic cancer.