Title: AI-based quantification of tumor-infiltrating lymphocytes with integrative transcriptomics in ovarian clear cell carcinoma: JGOG3025-TR1/A1 study

Transcriptomic classification methods have been proposed for ovarian clear cell carcinoma. However, their clinical significance and association with pathologically evaluated tumor-infiltrating lymphocytes (TILs) remain unclear. We established two large transcriptomic datasets and analyzed RNA-sequencing data from 189 (JGOG3025-TR1 cohort) and 38 (Kyoto cohort) ovarian clear cell carcinomas. Representative histopathological slides were also digitized (102 and 38, respectively). Cell types were classified by two state-of-the-art artificial-intelligence models, and TILs were quantified. The transcriptomically defined immune subtype was associated with significantly poor prognosis (hazard ratio, 2.54; 95% CI, 1.42–4.54; p = 0.002 for OS). However, this group also contained significantly higher proportion of advanced-stage cases (p = 0.003), and multivariate analyses showed no independent prognostic effect (hazard ratio, 1.32; 95% CI, 0.68–2.58; p = 0.42 for OS). In contrast, the pathologically defined inflamed group demonstrated a trend toward improved survival, and the inflamed phenotype emerged as a statistically significant favorable prognostic factor for both OS and PFS in multivariate analyses (hazard ratio, 0.32; 95% CI, 0.13–0.78; p = 0.013 for OS. hazard ratio, 0.32; 95% CI, 0.15–0.67; p = 0.0026 for PFS). These findings indicate a discordance between transcriptome- and pathology-based immune classifications and suggest greater prognostic relevance of pathology-derived immune status.