In the present retrospective study, 256 patients (age range, 22–88 years) with a confirmed diagnosis of ovarian cancer at Nara Medical University Hospital (Kashihara, Japan) were recruited between January 2008 and January 2022. The inclusion criteria were as follows: i) Confirmed pathological diagnosis of ovarian cancer and ii) received treatment, not only supportive care. The exclusion criteria were as follows: i) Combined with other malignancies; ii) pregnant women; and iii) patients with concomitant serious comorbidities. The ovarian cancer staging was determined using the International Federation of Gynecology and Obstetrics (FIGO) classification 2014 (16). Patients were diagnosed with ovarian cancer based on histopathology, pelvic magnetic resonance imaging and chest and abdominal computed tomography (Definition Flash, Siemens AG; Definition AS, Siemens AG; and Aquilion ONE, Canon, Inc.). Histopathology involved primary staining with hematoxylin-eosin (room temperature, hematoxylin for 5 min and eosin for 2 min), with additional p53 staining [primary antibody at 4°C for 12 h (rabbit polyclonal anti-p53; dilution, 1:200; cat no. NCL-L-p53-CM5p; Leica Biosystems, Ltd.); secondary antibody at room temperature for 1 h (mouse anti-rabbit IgG-HRP; dilution, 1:25; cat no. sc-2357; Santa Cruz Biotechnology, Inc.)] for high-grade serous carcinoma and HNF-1β staining [primary antibody at 4°C for 12 h (rabbit polyclonal anti-HNF1β; dilution, 1:200; cat no. 12533-1-AP, Proteintech Group, Inc.); secondary antibody at room temperature for 1 h (mouse anti-rabbit IgG-HRP; dilution, 1:25; cat no. sc-2357; Santa Cruz Biotechnology, Inc.)] for OCCC as an auxiliary diagnosis. CT images were evaluated using 5 mm thick axial images. Patient clinical data were collected, including age, body mass index (BMI), parity, menopausal status, histological type and FIGO stage. Proteins of interest were quantified using a fluorescence or chemiluminescence immunoassay, including TFPI2 [E-Test TOSOH II (TFPI2); Tosoh Corporation; cat. no. #0025245], CA125 (CL AIA-PACK^® OVCA; Tosoh Corporation, cat. no. #0029114; ARCHITECT CA125 II; Abbott Japan LLC, cat. no. #2K45-28), carbohydrate antigen (CA) 19-9 (CL AIA-PACK^® Sla; Tosoh Corporation, cat. no. #0029112) and carcinoembryonic antigen (CEA) (CL AIA-PACK^® CEA; Tosoh Corporation, #0029108) according to the manufacturer's instructions. TFPI2, CA125 (CL AIAPACK OVCA), CA19-9, and CEA concentrations were determined by Tosoh Corporation using serum obtained before the surgery. Tumor marker concentrations were measured by clinical laboratory technologists blinded to the study. The present study adhered to the guidelines of the Declaration of Helsinki. This was a single-center retrospective study based on medical records and all patient information was anonymized. The research project was announced on an opt-out basis.

Only variables that could be assessed preoperatively were included in uni- and multivariate analyses, including age, BMI, menopausal status and tumor marker levels. The analysis was performed in patients with OCCC and non-OCCC separately because diagnostic cut-off value differs between OCCC and ovarian cancer (11). The cut-off value for OS was applied to analyze PFS and OS. OS was defined as the period from treatment initiation until death or the last follow-up examination. PFS was defined as the period from treatment initiation until date of diagnosis as a progressive disease or the last follow-up examination.

Patients diagnosed with ovarian cancer underwent primary debulking surgery (PDS) if optimal surgery was possible. If the surgery was more extensive than bilateral oophorectomy and no second surgery was performed, it was considered PDS. If surgery was performed following chemotherapy, it was considered as an interval debulking surgery (IDS). If surgery was minor relative to bilateral oophorectomy, it was considered as no surgery.

The completion of surgery was considered optimal if the diameter of the remaining tumor was <1 cm. The surgery was considered suboptimal if the diameter of the remaining mass was ≥1 cm. Furthermore, information regarding lymphadenectomy was collected. Unless the patient had a poor performance status and was considered incapable of enduring a high invasive surgery, both paraaortic and pelvic lymphadenectomy were performed. Adjuvant chemotherapy, mostly comprised of taxane and carboplatin (TC) therapy, was generally performed upon obtaining consent from the patient.

The receiver operating characteristic (ROC) curve was used to determine optimal cut-off points of TFPI2, CA125, CA19-9, and CEA levels to predict OS for OCCC and non-OCCC. The optimal cut-off value was determined using the Youden index to predict OS. The outcome on the ROC curve was defined as survival or death. The Kaplan-Meier life table analysis and the log-rank tests were used to assess survival rates and differences based on prognostic factors. Multivariate analysis of prognostic factors for PFS and OS was performed using the Cox proportional hazard regression model where univariate analysis revealed significant differences. All statistical analyses were performed using SPSS software (version 29.0, IBM Corp.). P<0.05 was considered to indicate a statistically significant difference.

The present study included 256 patients with ovarian cancer with a median age of 60 years (range, 22–88 years). The median follow-up period was 52.4 months (range, 0.8–190.8 months).

A total of 121 cases (47.3%) were FIGO stage I or II while 135 (52.7%) were stage III or IV (Table I). There were 109 cases (42.6%) of serous carcinoma, 40 cases of mucinous carcinoma (15.6%), 15 cases of endometrial carcinoma (5.9%), 66 cases of OCCC (25.8%) and 26 cases of others (10.2%; Table SI). Other histological types included carcinosarcoma, malignant Brenner tumor, seromucinous carcinoma, mixed epithelial tumor, neuroendocrine carcinoma and undifferentiated carcinoma (data not shown). The median preoperative serum levels of TFPI2, CA125, CA19-9 and CEA were 219.0 (82.5–5,824.2) pg/ml, 278.6 (0.5–43,170.9) U/ml, 21.5 (0.0–217,474.9) U/ml and 2.1 (0.4–142.9) ng/ml, respectively (Table I).

A total of 163 (63.7%) patients underwent PDS, 75 (29.3%) underwent IDS and 18 (7.0%) underwent biopsy (Table II). Moreover, surgery was optimal and suboptimal in 184 (71.9%) and 72 (28.1%) patients, respectively.

Among the 256 patients, 106 (41.4%) underwent lymphadenectomy while 150 (58.6%) did not. The adjuvant first-line chemotherapy regimen mostly comprised TC therapy (n=198, 77.3%). Of the remaining patients, most cases underwent platinum-based chemotherapy, including docetaxel and carboplatin (n=5, 2.0%), TC and bevacizumab (n=5, 2.0%), dose-dense TC (n=5, 2.0%), weekly TC (n=2, 0.8%) and irinotecan and cisplatin therapy (n=2, 0.8%). Weekly paclitaxel therapy was performed in two patients (0.8%), while docetaxel and gemcitabine therapy was performed in one patient (0.4%). However, 36 patients (14.1%) did not receive adjuvant chemotherapy (Table SII).

For non-OCCC, the cut-off value of TFPI2 for predicting OS was 201 pg/ml based on the Youden index [area under the curve (AUC), 0.646; sensitivity, 73.1%; specificity, 59.8%; 95% confidential interval (CI), 0.568–0.724], while that for CA125 was 394 U/ml (AUC, 0.648; sensitivity, 69.2%; specificity, 60.7%; 95% CI, 0.569–0.727; Fig. 1). CA19-9 and CEA were excluded because they showed negative associations with OS based on ROC curve results. TFPI2 values <201 and ≥201 pg/ml were defined as negative and positive, respectively. Similarly, CA125 values <394 and ≥394 U/ml were defined as negative and positive, respectively.

In the univariate analysis, TFPI2 ≥201 pg/ml was significantly associated with PFS (Fig. 2A) and OS (Fig. 2B). Table III shows uni- and multivariate analyses of prognostic factors for PFS and OS. For PFS. Univariate analysis showed significant differences in age ≥60 years, post-menopausal status, TFPI2 ≥201 pg/ml and CA125 ≥394 U/ml. For OS, univariate analysis showed significant differences in age ≥60 years, post-menopausal status, TFPI2 ≥201 pg/ml and CA125 ≥394 U/ml. Cox multivariate analysis revealed that TFPI2 was a significant independent prognostic factor affecting OS.

Next, analysis was conducted for patients with OCCC. The Youden index was used to calculate a cut-off value of 255 pg/ml for TFPI2 for predicting OS (AUC, 0.653; sensitivity, 87.5%; specificity, 55.2%; 95% CI, 0.494–0.812) and 363 U/ml for CA125 (AUC; 0.655, sensitivity; 62.5%, specificity; 82.8%, 95% CI; 0.421–0.890; Fig. 3). TFPI2 levels <255 and ≥255 pg/ml were defined as negative and positive, respectively. Similarly, CA125 values <363 and ≥363 U/ml were defined as negative and positive, respectively. In univariate analysis, TFPI2 ≥255 pg/ml was significantly associated with PFS (Fig. 4A) and OS (Fig. 4B). Table IV shows univariate and multivariate analyses of prognostic factors for PFS and OS. For PFS, univariate analysis showed significant differences in TFPI2 ≥255 pg/ml and CA125 ≥363 U/ml. Contrastingly, for OS, univariate analysis showed significant differences in BMI ≥25, TFPI2 ≥255 pg/ml and CA125 ≥363 U/ml. When Cox multivariate analysis was applied, only TFPI2 was a significant independent prognostic factor affecting PFS in patients with OCCC.