Serum CA19-9 Adds Histotype Information to CA125 in Epithelial Ovarian Cancer: A Retrospective Cohort Study of 157 Chemotherapy-Naive Patients

Background and Objectives: Histological subtype in epithelial ovarian cancer determines chemosensitivity and prognosis but is usually confirmed only on the surgical specimen. We asked whether any preoperative serum tumour marker, or any ratio between two, is associated with non-serous histology and adds to International Federation of Gynecology and Obstetrics (FIGO) stage. Materials and Methods: Retrospective cohort of 157 women operated on for epithelial ovarian cancer in 2016–2024 with complete chemotherapy-naive CA125, carcinoembryonic antigen (CEA), CA19-9 and CA15-3 values (pre-cycle-one values in 64 neoadjuvant patients). Four markers and six pairwise ratios were assessed by Mann–Whitney tests with Benjamini–Hochberg correction, receiver-operating-characteristic analysis and Firth penalised regression; discrimination was internally validated by bootstrap resampling that repeated the selection of the leading index. Results: Twenty-four tumours (15.3%) were non-serous. The CA125/CA19-9 ratio ranked first (area under the curve 0.796, 95% confidence interval 0.696–0.883) but was not superior to CA125 alone (0.708; p = 0.072), and stage alone performed comparably (0.805). Adding the ratio to stage improved fit (p = 0.001; odds ratio 0.36 per ten-fold increase, 0.17–0.71; selection-adjusted permutation p = 0.021). Optimism-corrected areas were 0.766 for the ratio alone and 0.869 for stage plus ratio; CA15-3/CA19-9 and CA19-9 alone added to stage similarly. The threshold of 12.2 was unstable (bootstrap range 2.6–58.7) and discrimination was 0.575 in early-stage disease. Raised CA19-9 arose from clear cell (p = 0.002) and mucinous (p = 0.022) but not endometrioid tumours (p = 0.92). Conclusions: CA19-9 carries histotype information beyond CA125 and stage, arising from clear cell and mucinous tumours, whether expressed alone or as a ratio; neither is usable as a stand-alone preoperative test or to guide treatment. The finding is hypothesis-generating and requires external validation alongside human epididymis protein 4 (HE4) and the Risk of Ovarian Malignancy Algorithm (ROMA).