Identifying early-stage cancer patients at risk for progression is a major goal of biomarker research.This report describes a novel 19-gene signature (19-GCS) that predicts stage Ⅰ lung adenocarcinoma (LAC) recurrence and response to therapy and performs comparably in pancreatic adenocarcinoma (PAC),which shares LAC molecular traits.Kaplan-Meier,Cox regression,and cross-validation analyses were used to build the signature from training,test,and validation sets comprising 831 stage Ⅰ LAC transcriptomes from multiple independent data sets.A statistical analysis was performed using the R language.Pathway and gene set enrichment were used to identify underlying mechanisms.19-GCS strongly predicts overall survival and recurrence-free survival in stage Ⅰ LAC (P=0.002 and P＜0.001,respectively) and in stage Ⅰ-Ⅱ PAC (P＜0.0001 and P＜0.0005,respectively).A multivariate cox regression analysis demonstrated the independence of 19-GCS from significant clinical factors.Pathway analyses revealed that 19-GCS high-risk LAC and PAC tumors are characterized by increased proliferation,enhanced steness,DNA repair deficiency,and compromised MHC class Ⅰ and Ⅱ antigen presentation along with decreased immune infiltration.Importantly,high-risk LAC patients do not appear to benefit from adjuvant cisplatin while PAC patients derive additional benefit from FOLFIRINOX compared with gemcitabine-based regimens.When validated prospectively,this proof-of-concept biomarker may contribute to tailoring treatment,recurrence reduction,and survival improvements in early-stage lung and pancreatic cancers.