Assessment of tissue factor pathway inhibitor 2 (TFPI2) as a novel serum marker for malignant tumors of the ovary before and after treatment: A case-control study.

Yuki Ogawara: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; writing – original draft. Naho Ruiz Yokota: Conceptualization; data curation; formal analysis; investigation; methodology; project administration. Yuki Yamada: Data curation. Noriaki Arakawa: Formal analysis; methodology; validation. Kentaro Sakamaki: Formal analysis. Hiroshi Kobayashi: Data curation; investigation; writing – review and editing. Kazumi Kubota: Formal analysis. Fuminori Kimura: Writing – review and editing. Taichi Mizushima: Conceptualization; methodology; supervision; writing – review and editing. Etsuko Yamazaki: Writing – review and editing. Etsuko Miyagi: Investigation; project administration; validation; writing – original draft; writing – review and editing.

This study included BOT and EOC patients who had participated in our previous study from June 2016 to March 2018. 14 We performed further analyses for each patient group as follows: CCC, non‐clear cell BOT + EOC, non‐clear cell BOT, and non‐clear cell EOC. Patients were treated and followed up at Yokohama City University Hospital and Nara Medical University Hospital. Written informed consent was obtained from all participating patients. This study was approved by the institutional ethics committee (June 20, 2022, B180400049) and followed the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan. All patients underwent at least surgical removal of adnexal masses as debulking surgery; clinical staging of patients and histological diagnoses by pathologists at each hospital following the FIGO 2014 classification have been reported. 14 We measured TFPI2 and CA125 in the blood samples obtained in the study, which were collected within 1 month before surgery as baseline levels. 14 Routine patient blood samples were collected approximately every 3–4 months, some of which were used to analyze TFPI2 and CA125 until March 2019 for this study. Blood samples were collected in Venoject II serum separator tubes (VP‐AS109K60, Terumo, Tokyo, Japan). The tubes were stored for 2–3 h at 4°C or for 30 min at room temperature and then centrifuged at 1000–1500 g for 10 min. Serum aliquots were stored at −40°C to −80°C. TFPI2 and CA125 concentrations were measured using reagents provided by Tosoh Diagnostics Product Divisions (Tosoh Corporation, Tokyo, Japan). The technicians involved in the measurements are aware that TFPI2 is being measured as part of a clinical trial, but are not aware of the details of the study other than the sampling procedure and the clinical background of each sample. The TFPI2 concentration was measured by the direct assay method on an automated immunoassay analyzer system (Tosoh Corporation) as described before. 14 CA125 was also measured by the automated immunoassay analyzer system and a diagnostic reagent, E test TOSOH II (Tosoh Corporation). The cutoff value was 35 U/mL CA125. The cutoff value of 191 pg/mL for TFPI2 was approved for insurance coverage by the Ministry of Health, Labour and Welfare in Japan in 2021 for comprehensive detection of ovarian malignant tumors. 15 We also evaluated the cutoff of 270 pg/mL TFPI2 used to detect CCC in accordance with the previous studies. 13 , 14 Demographic factors were summarized as follows: continuous variables are represented as means with ranges, whereas categorical variables are depicted as frequencies and percentages. Changes in serum TFPI2 and CA125 levels are summarized as means with ranges. Positivity rates were calculated as the percentage of positive numbers above each designated cutoff value in samples. Confidence intervals were calculated using the exact method. Statistical analysis was performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC) and Microsoft Excel version 16.90.2 (24102719).

Among the patients evaluated in our previous study for the performance of TFPI2 to discriminate CCC from other EOCs, 14 47 patients who had previously undergone surgical treatment for BOT or EOC at Yokohama City University or Nara Medical University agreed to participate in this follow‐up study. In this study, standard taxane and platinum‐based chemotherapy were administered to patients who required them according to the Guidelines for the Treatment of Ovarian, Fallopian Tube, and Primary Peritoneal Cancer 2020 Edition published by the Japanese Society of Gynecologic Oncology. One patient discontinued visits and was excluded from the study. The demographics of the 46 patients in the final group are shown in Table  1 . Eleven patients had a history of BOT and no recurrences occurred during the observation period. One patient with stage II disease in the BOT group received taxane and platinum‐based chemotherapy (TP therapy). The 35 patients with a history of EOC included 13 CCC (37.1%), 15 serous carcinoma (42.9%), four endometrioid carcinoma (11.4%), and three mucinous carcinoma (8.6%) cases. In the EOC group, 16 (45.7%) and 19 (54.3%) patients had stage I and stage II–IV disease, respectively. All patients in the EOC group received TP therapy after surgery. In the EOC group, 26 (74.2%) patients had complete remission and nine (25.7%) patients had recurrent disease during the observation period. Demographics of patients with ovarian borderline tumors and epithelial cancers. Abbreviation: FIGO, International Federation of Gynecology and Obstetrics. The changes in mean serum TFPI2 and CA125 levels in patients without recurrence after surgery are shown in Table  2 . In the overall group, the preoperative mean levels of TFPI2 (235.3 pg/mL, range: 78.3–607.7) and CA125 (1125.5, range: 6.2–6272.0) were higher than the cutoff values (TFPI2: 191 pg/mL; CA125: 35 U/mL). We performed further analyses for each patient group as follows: CCC, non‐clear cell BOT + EOC, non‐clear cell BOT, and non‐clear cell EOC. In the CCC group, the preoperative mean level of TFPI2 exceeded 270 pg/mL (the cutoff value for CCC). At 3 months or longer after primary debulking surgeries in the overall group, the mean levels of TFPI2 (150.2 pg/mL) and CA125 (12.7 U/mL) decreased to below the cutoff values. Changes in serum TFPI2 and CA125 levels in patients without recurrence after primary debulking surgery. Note : TFPI2 is measured in pg/mL; CA125 is measured in U/mL. Abbreviation: BOT borderline tumor, EOC epithelial ovarian cancer. Postoperative value shows the minimum one. The CA125 value was missing in one patient with BOT. The changes in positivity rates in patients without recurrence using the two cutoff values of TFPI2 (cutoff values of 270 pg/mL for CCC and 191 pg/mL for BOT and EOC) and the standard cutoff value of CA125 (35 U/mL) are shown in Table  3 . In the overall group, the preoperative positivity rates of TFPI2 (cutoff value: 191 pg/mL) and CA125 were 54.0% (95% CI: 36.9–70.5) and 70.2% (95% CI: 53.0–84.1), respectively. In the CCC group, the preoperative positivity rate of TFPI2 (cutoff value: 270 pg/mL) was higher (72.2%, 95% CI: 39.0–94.0) than that of CA125 (54.5%, 95% CI: 23.4–83.3). In the non‐clear cell EOC group, the positivity rates of TFPI2 (cutoff value: 191 pg/mL) and CA125 were 21.4% (95% CI: 4.7–50.8) and 92.9% (95% CI: 66.1–99.8), respectively, with a large and significant disparity. In the non‐clear cell BOT group, the difference in the positivity rate was relatively small for both TFPI2 (41.7%, 95% CI: 15.2–72.3) and CA125 (58.3%, 95% CI: 27.7–84.8). Pre‐ and postoperative positive rates of serum TFPI2 and CA125 in patients without recurrence after primary debulking surgery. Note : The cutoff of TFPI2; 270 pg/mL for clear cell carcinoma, 191 pg/mL for BOT and EOC. Abbreviations: BOT, borderline tumor; CI, confidence interval; EOC, epithelial ovarian cancer. Postoperative positive rate is at the minimum value. The CA125 value was missing in one patient with BOT. The postoperative TFPI2 level was higher than the cutoff (191 pg/mL) in 11 patients (29.7%, 95% CI: 13.8–44.1) among 37 patients without recurrence (Table  3 ). However, in eight out of 11 patients, TFPI2 values decreased from 9.5% to 49.9% (Figure  1 ) after treatment. The pre‐ and postoperative TFPI2 data of patients with postoperative TFPI2 higher than 191 pg/mL are shown in Figure  1 . No specific tendency was found in terms of tissue type. Comparison of postoperative TFPI2 in patients without recurrence and with TFPI2 over 191 pg/mL. Horizontal axis: Individual patients; vertical axis: TFPI2 level (pg/mL) pre‐ and postoperative minimum values. Changes in mean TFPI2 and CA125 levels before and after primary debulking surgery in patients who had recurrent disease after debulking surgery and chemotherapy are shown in Table  4 . The maximum mean serum level of TFPI2 in the overall group after recurrence (492.6 pg/mL, range 114.6–2703.3) was higher than the preoperative level (421.5 pg/mL, range 120.2–2002.4). In terms of CA125, the mean preoperative level was much higher (2903.8 U/mL, range 28.0–12898.0) than the maximum mean level after recurrence (727.4 U/mL, range 21.0–2759.0). In the non‐CCC EOC group, the mean TFPI2 level was relatively low before (222.3 pg/mL, range 120.2–325.2) and after (229.5 pg/mL, range 114.6–392.7) recurrence, although both mean levels were more than the cutoff value. Changes in serum TFPI2 and CA125 in patients with recurrence after primary debulking surgery. Note : TFPI2 is measured in pg/mL; CA125 is measured in U/mL. The cutoff level; 191 pg/mL for TFPI2, 35 U/mL for CA125. In patients with recurrence, the positive rate of TFPI2 using the cutoff value of 191 pg/mL after recurrence was 55.6% (95% CI: 21.2–86.3) and that of CA125 was 66.7% (95% CI: 29.9%–92.5%), which was comparable (Table  S1 ). The two cases of recurrent CCC had TFPI2 levels above and below the cut‐off at the time of recurrence (Table  S2 ). Because only seven patients in the non‐CCC group had relapsed, statistical evaluation was not conducted.

In our previous study, we identified TFPI2 with a cutoff value of 270 pg/mL as a novel ovarian CCC‐specific serum tumor marker. 14 The sensitivity and specificity of TFPI2 to detect CCC among all EOC and BOT patients were 44% and 80%, respectively, and the high specificity for CCC was notable. In addition, TFPI2 is often expressed in malignant ovarian tumors, including BOT, and can be used as a comprehensive ovarian tumor marker to distinguish malignant and benign tumors. 14 , 15 These data suggest that TFPI2 should be evaluated together with CA125, which has high non‐specific sensitivity to detect EOC and BOT, but relatively low sensitivity in CCC patients. Additionally, serum TFPI2 efficiently discriminates CCC from benign ovarian lesions, including benign endometriotic cysts, the precursor of ovarian CCC. 14 After the study was published, the TFPI2 test became covered by health insurance in Japan in 2021. 15 The 191 pg/mL cutoff value is used to detect CCC, BOC, and EOC at present. Using both TFPI2 and CA125, the positivity rate to detect BOT and EOC has been determined to be 82%. 15 In this study, we examined whether this cutoff value was appropriate as a preoperative and postoperative serum tumor marker to predict the existence of malignant ovarian tumors. From our postoperative data, we consider that TFPI2 can be useful as a tumor marker for all EOCs and BOTs. In the present study, 56% of patients with BOT or EOC had preoperative positive serum levels of TFPI2 above 191 pg/mL and confirmed that patients who participated in this study had equal TFPI2 positivity with 55% of our previous study. 15 Among the 11 CCC patients without recurrence, five (45.5%) patients had a high TFPI2 level above 270 pg/mL, which TFPI2 level decreased to below 191 pg/mL postoperatively. In the non‐clear cell EOC group, the positivity rate of CA125 was 92.9%, but only 21.4% of patients had a TFPI2 level exceeding 270 pg/mL. This large disparity also indicated that the TFPI2 level should be considered together with the CA125 level to estimate non‐CCC malignant tumors or CCC. Because the TFPI2 level is not influenced by menstrual cycles or the presence of endometriotic cysts 13 , 14 unlike CA125, 9 , 10 we speculate that using TFPI2 instead of CA125 for preoperative evaluation of ovarian tumors will be useful in Japan and Asian countries where CCC is a major EOC type. Predicting intractable ovarian CCC before an operation changes the strategy of the surgical procedure to extended removal of whole tumors even in advanced‐stage patients because of the chemo‐resistant nature of CCC. We are currently attempting to improve the accuracy of preoperative diagnosis of each histological type of ovarian tumor by adding machine learning that combines MRI images and TFPI2 and CA125 levels. The positivity rate of TFPI2 after complete remission was ~30% (Table  3 ) at a cutoff value of 191 pg/mL, which was based on the preoperative data of 351 ovarian tumor patients in our previous study using the Youden index. 13 , 14 Because the specificity to detect BOT and EOC at the cutoff was 86% in our previous study, 15 the false positive rate was estimated to be ~15% before this study. In contrast to such expectations, 11 out of 37 patients without recurrence had high TFPI2 levels over the cutoff after complete remission following individual treatments. However, TFPI2 decreased in eight patients after complete remission, as shown in Figure  1 . In these eight patients, the TFPI2 level was considered to be a true tumor marker that changed in accordance with disease existence. However, in three patients, TFPI2 did not reflect the disease status. In our previous study, 14 one patient with benign ovarian tumors had very high TFPI2 above 270 pg/mL, and she had renal failure. In the present study, we could not identify the reason for the relatively high TFPI2 values in patients without malignant tumors by reviewing the available data including fundamental blood tests and histological types (Figure  1 ). CA125 increases during the menstrual period, pregnancy, endometriosis, diseases with ascites, and pleural effusion, 16 and elevation of CEA is related to smoking. 17 Thus, we should explore conditions with a high basal level of TFPI2. Concerning the relationship between the malignant potential of cancer and TFPI2 expression, a pro‐invasive function of TFPI2 in hepatocellular carcinoma cells in vitro has been reported, 18 which was associated with TFPI2 binding to the tissue factor‐activated coagulation factor VII complex. We previously found that factor VII expression occurs frequently in EOC with tissue factor expression, particularly in CCC cells. 19 Additionally, the malignant potential of EOC is highly associated with the complication of deep vein thrombosis known as Trousseau's syndrome. A recent study indicated that high TFPI2 may predict asymptomatic venous thromboembolism in patients with EOC. 20 TFPI2 is a serine protease inhibitor related to the blood coagulation system. Therefore, whether the status of venous thromboembolism, the use of direct oral anticoagulants, and factors related to blood coagulation, such as D‐dimer, fibrinogen, prothrombin time, and activated partial thromboplastin, affect TFPI2 or the malignant potential of EOC should be explored. In addition, the effects of antiplatelet drugs on TFPI2 are also unknown. In the two CCC cases that recurred during follow‐up in the present study, the preoperative TFPI2 levels were both above the cut‐off value. After recurrence, the TFPI2 level increased in one case who was taking aspirin and was below the cut‐off in another case who was taking Limaprost Alfadex (Table  S2 ). It is also necessary to evaluate the effect of antiplatelet drugs on TFPI2 levels. This exploratory study had several limitations. Only a small number of patients were included, and the observation period was relatively short (up to 27 months). Additionally, each patient was followed up in accordance with the institutional instructions after the primary debulking surgery. Despite these limitations, the present study highlights the potential clinical use of TFPI2 and future steps, including the identification of factors that increase serum TFPI2 influence independent of ovarian malignancy and whether the combination of TFPI2 and CA125 is a prognostic factor. In conclusion, this follow‐up study showed that TFPI2 can be used as a serum tumor marker to estimate the status of ovarian CCC, BOT, and non‐CCC EOC. The physiological role of TFPI2 and the development of thrombosis should be investigated in further basic and translational research. Large‐scale clinical research should focus on TFPI2, which has a particularly high level in patients with advanced CCC, together with CA125 as a credible non‐specific tumor marker to detect malignant ovarian tumors and explore the potential utility of this combination for medical treatment.

Ovarian cancer was the third most common gynecological cancer in 2020 with over 300 000 new cases and over 200 000 deaths worldwide. 1 In Japan, 13 388 women were diagnosed with ovarian cancer in 2019, and 5182 women died of ovarian cancer in 2022. 2 The incidence and mortality rates of ovarian cancer have been gradually increasing over the past 40 years. The annual patient report from the Japan Society of Obstetrics and Gynecology for 2020 stated that clear cell carcinoma (CCC) was the second most common (27.6%) subtype of epithelial ovarian cancer (EOC) in Japan, 3 with a frequency two‐fold higher than that in Western countries (5%–10%). 4 , 5 CCC is an endometriosis‐associated EOC and exhibits chemoresistance in advanced or recurrent disease, 6 a low prevalence of germline BRCA mutation (0%–7%), 7 and low or normal levels of serum cancer antigen 125 (CA125). 8 Patients with benign ovarian endometriosis often have high levels of serum CA125. 9 , 10 Standard chemotherapy with paclitaxel and carboplatin or poly (ADP)‐ribose polymerase inhibitors is not generally effective for patients with advanced CCC. Therefore, radical and/or extended surgery should be scheduled if advanced CCC is preoperatively predicted. We previously established tissue factor pathway inhibitor 2 (TFPI2), a Kuniz‐type serine protease inhibitor, also known as placental protein 5, 11 as a novel serum biomarker specific to CCC by translational research using modified proteomic techniques. 12 , 13 TFPI2 has 80% specificity to discriminate CCC from other EOC subtypes at the appropriate cutoff value of 270 pg/mL in clinical settings. 14 An efficient automated enzyme‐linked immunosorbent assay system was applied for TFPI2 detection as a clinical test. Because TFPI2 is often expressed in malignant tumors that originate from the Müllerian duct, the sensitivity and specificity of TFPI2 to discriminate EOC and borderline ovarian tumors (BOT) from benign ovarian tumors were 54% and 86%, respectively, when the cutoff value was lowered to 191 pg/mL from 270 pg/mL to detect CCC. 15 The TFPI2 test was covered by insurance in Japan to detect EOC and BOT with a cutoff value of 191 pg/mL in April 2021. In this study, we conducted a prospective observational study to clarify whether TFPI2 reflects the postoperative condition of patients with ovarian BOT and EOC.

Dr. E. Miyagi and Dr. N. Arakawa had grants issued to Yokohama City University from Tosoh Corporation during the conduct of the study. Dr. E. Miyagi and Dr. N. Arakawa have patents for the use of TFPI2 in the diagnosis of ovarian cancer issued to Yokohama City University. Dr. E. Miyagi and Dr. N. Arakawa had honoraria and a travel grant from Tosoh Corporation. Dr. F. Kimura had Payment or honoraria for lectures from Mochida Pharmaceutical Co., Ltd. and ASKA Pharmaceutical Co., Ltd. Another author declares that they have no competing interests. Dr. Fuminori Kimura is an Editorial Board member of JOG Journal and a co‐author of this article. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication.

Table S1. Pre‐ and postoperative positive rates of serum TFPI2 and CA125 in patients with recurrence after surgery. Table S2. The trend in TFPI2 levels (pg/mL) in two cases of recurrent CCC.