Exploratory Analysis of Age-Associated Somatic Mutation Patterns and Clinicopathologic Features in Endometrial Cancer

Background: Age-related heterogeneity in endometrial cancer may reflect differences in histologic composition, molecular subtype distribution, and tumor mutational burden. Methods: We matched TCGA clinical age data with MAF somatic mutation data and analyzed 508 patients with non-missing age and mutation information; all displayed patient-level TCGA mutation spectra and age-related comparisons used this common analytic cohort. The primary age analysis modeled age continuously using multivariable logistic regression adjusted for histologic grade and histologic type. An observed-age cutoff scan was performed only as an exploratory descriptive analysis using patient-level two-sided Fisher exact tests, followed by Benjamini-Hochberg false-discovery rate (BH-FDR) correction across all gene-cutoff comparisons. A 49-gene candidate set, defined from somatic variants with definite or potential clinical significance in a 28-patient SYSU-3 targeted-panel cohort, was evaluated in TCGA using restricted cubic spline (RCS) logistic regression. Mutation-count and POLE-exclusion sensitivity analyses were performed. Results: In the primary continuous-age analysis, age-related mutation odds were estimated per one-year increase with adjustment for histologic grade and histologic type. A secondary exploratory scan identified 57 years as the cutoff that maximized the number of uncorrected significant genes and separated 140 younger from 368 older patients. At this exploratory cutoff, 3308 genes were BH-FDR significant, but 3307 showed higher mutation frequency in the younger group. The younger group had a higher nonsynonymous mutation-count proxy and a higher proportion of nonsynonymous POLE variants. After excluding POLE-variant patients, only one gene remained significant in the all-gene cutoff analysis. Among the 49 SYSU-3-derived candidates, 22 showed significant linear age effects and 21 showed significant RCS overall effects; 19 genes overlapped. In SYSU-3, PTEN, PIK3CA, and ARID1A were the most frequently affected genes among variants with definite or potential clinical significance. Conclusions: Age was statistically associated with somatic mutation patterns in TCGA, but the broad directional signal was strongly influenced by tumor molecular composition and POLE-related hypermutation. The 57-year value should be considered an exploratory, data-driven threshold rather than a validated biological or clinical breakpoint.