Abstract Background: Microsatellite stability (MSS) is a biomarker for worse prognosis than microsatellite instability (MSI) in non-metastatic tumors. Tumors with MSS status are more probable to occur postoperative recurrence and metastasis. Some studies had revealed that endometrial and gastrointestinal tumors have distinct MSI-related features compared with other types of cancers, indicating there are potential similarities in these cancer types. The proportion of MSS population in non-metastatic endometrial and gastrointestinal tumors were about 90% and 75%. There are high heterogeneity in such wide range of patient population. The status of immune microenvironment based on immunohistochemistry or RNA-Seq profiling had been proved effective prognostic biomarker in some previous investigation. Therefore, we assume a subtype model of immune microenvironment based on transcriptomics features in endometrial and gastrointestinal tumors. Method: 128 patients with I-III stage MSS endometrial and gastrointestinal tumors from TCGA project were included as training cohort and a meta cohort of 603 patients of colorectal and stomach cancers from GEO datasets (GSE39582,GSE62254) were included as validation cohort in our study. The RNA-Seq profiling data and disease-free survival (DFS) of patients were collected. CIBERSORT tool was conducted to evaluate the immune-related cells' enrichment. A lasso-cox model of DFS was developed based on the CIBERSORT features. Patients in training cohort were scored the recurrence risk by the model. We further investigate the differentially expressed genes between high-risk and low-risk patients. Log-rank test was used to evaluate the difference of DFS. Result: 76 immune-related genes were identified to be differentially expressed significantly. The patients in training cohort were divided into two groups (group1 n=81, group2 n=41) based on non-negative matrix factorization (NMF) method. The DFS of group1 and group2 were significantly different (log-rank p=0.005). In validation cohort, patients were divided into group1 and group2 (group1 n=401, group2 n=202) using clustering method above. A significant difference between subgroups was also tested (log-rank p<0.0001). Conclusion: We identified 76 immune-related genes using CIBERSORT tool. Based on this gene set, early stage MSS endometrial and gastrointestinal tumors could be divided into two subgroups with distinct disease-free survival. Citation Format: Shi Zhang, Xinyi Liu, Lan Zhang, Mengli Huang. Subgroups of early-stage microsatellite stability endometrial, colorectal and stomach tumors based on transcriptomics have distinct disease-free survival [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2149.