Abstract
Background/Objectives: Most studies investigating prognostic biomarkers in cervical cancer (CC) analyze patients irrespective of FIGO stage, potentially masking molecular features that underlie the aggressiveness of some FIGO II tumors. To address this, we investigated differential gene expression in a FIGO II CC cohort to identify a gene signature predictive of progression-free survival (PFS) within five years of treatment initiation. Methods: Tumor samples from 15 CC patients were analyzed using RNA sequencing, bioinformatics, and machine learning to identify differentially expressed genes (DEGs) associated with prognosis. Findings were validated in an independent CC cohort (n = 174). Results: High expression of B3GALT1 (HR = 5.11), GTF3C2-AS1 (HR = 18.73), and ZKSCAN4 (HR = 5.18) was significantly associated with an increased risk of recurrence in our cohort. Elevated expression of these transcripts is also associated with shorter PFS in the external dataset. Notably, GTF3C2-AS1 expression alone was sufficient to classify all fifteen patients into their respective prognostic groups using a decision tree model, achieving 93.3% accuracy in leave-one-out cross-validation (LOOCV). Additional candidates, including RCAN2-DT, MYH9-DT, IGKC, IGHG1, and IGHG3, were associated with PFS in our cohort but could not be externally validated due to a lack of available data. Conclusions: Transcriptomic profiling revealed potential biomarkers that refine prognostic stratification in cervical cancer beyond FIGO staging. Among them, GTF3C2-AS1 consistently emerged as a potential predictor of recurrence risk. Additional candidates, including B3GALT1, ZKSCAN4, and immunoglobulin transcripts, provided complementary insights but require further validation. These preliminary results highlight intra-stage heterogeneity in FIGO II CC and underscore the promise of molecular markers to improve risk assessment.