A sex-informed transcriptomic prognostic score for gynecologic cancers: Multiplatform validation and spatial...

This study develops and validates a sex-informed transcriptomic prognostic score derived from sex-stratified survival analyses, with a focus on gynecologic malignancies.

This retrospective, computational multi-cohort study analyzed transcriptomic and clinical data from The Cancer Genome Atlas (TCGA; n ≈ 5000) and CPTAC-3 (n = 2191). A 10-gene score was constructed using sex-stratified Cox models and LASSO regression across 20 TCGA tumor types. Prognostic performance was evaluated using hazard ratios and time-dependent AUCs at 1, 3, and 5 years. Analyses followed a four-phase design: discovery in TCGA, internal cross-cancer validation, external validation in independent cohorts (MSK-IMPACT, CPTAC-3, and LIHC-FR), and biological validation using single-cell and spatial transcriptomic data from formalin-fixed paraffin-embedded tissues. Patients were stratified into high- and low-risk groups based on the median gene-expression score, with optimal cutpoints determined using maximally selected rank statistics where indicated. While the discovery analyses were sex-stratified, the final 10-gene score is sex-agnostic and applicable to both sexes.

The score showed strong prognostic discrimination (hazard ratio = 2.15; 95% confidence interval 1.60-2.88; P < 0.001), with areas under the curve ranging from 0.69 to 0.72 across timepoints. It remained robust across datasets and analytic platforms. In gynecologic tumors, high-score regions colocalized with fibroblast-rich, immune-depleted areas, reflecting transcriptional programs of stromal remodeling and immune exclusion linked to immunotherapy resistance.

This sex-informed, spatially validated score provides a reproducible and biologically interpretable framework for transcriptomic risk stratification in gynecologic cancers. By capturing immune-evasive and aggressive tumor states, it might inform biomarker-guided clinical trials and support context-appropriate implementation of precision oncology strategies.