Background

Endometrial cancer is the fourth most common cancer in women, and its incidence is increasing in the United States [1]. In Japan, there were 18,338 new cases of endometrial cancer and 3,630 deaths from the disease reported in 2022 [2]. Cancer antigen 125 (CA125) is the most commonly used serum tumor marker in patients with endometrial cancer, but it has a low sensitivity of 23.5–34.9% [3, 4]. Although some new serum biomarkers for the diagnosis of endometrial cancer have been proposed, further investigation is needed for clinical use [5, 6]. Zhou et al. reported the efficacy of exosomal microRNA as a diagnostic biomarker, with an area under the curve value of 0.899 when combined with CA125 and carcinoembryonic antigen (CEA) [7]. However, microRNA testing is currently not commonly used for cancer diagnosis in clinical settings. Therefore, the identification of new serum biomarkers is essential to facilitate the diagnosis of endometrial cancer and determine the appropriate surgical approach.

Tissue factor pathway inhibitor 2 (TFPI2) is a secreted Kunitz-type inhibitor of serine proteases involved in various coagulation and fibrinolysis systems, including the TF/VIIa complex, plasmin, plasma kallikrein, trypsin, and chymotrypsin [8]. TFPI2 is highly expressed in the mature placenta [9, 10] and is physiologically produced in vascular endothelial cells [11], fibroblasts, platelets [12], and macrophages [13]. TFPI2 is a tumor suppressor gene that is silenced in many types of cancer [14, 15] and is involved in tumor apoptosis and in the inhibition of invasion, growth, and metastasis [16, 17]. We previously discovered that TFPI2 is specifically expressed in ovarian clear cell carcinoma [9, 16, 18,19,20]. Since 2021, the TFPI2 test has been included in the public health insurance coverage for ovarian cancer in Japan. For endometrial cancer, Kawaguchi et al. first demonstrated the efficacy of serum TFPI2 for the diagnosis of endometrial cancer in a retrospective single-center study of 207 patients [21]. Based on the receiver operating characteristic curve for predicting the prognosis of endometrial cancer, they established a cutoff value of 177 pg/mL. This group was also the first to demonstrate that TFPI2 is highly expressed in clear cell carcinoma of the endometrium and can differentiate this type from other histological types with 100% sensitivity and 73.8% specificity through immunohistochemistry [22]. To date, no additional studies exist, and no previous report has analyzed the correlation between TFPI2 expression in serum and tissue samples from the same patients.

In this study, we enrolled patients with endometrial cancer who had undergone surgical removal as their initial treatment. We first assessed the effectiveness of TFPI2 as a serum diagnostic marker by comparing patients with endometrial cancer with healthy controls. We also analyzed the clinicopathological characteristics and prognoses of TFPI2-positive and TFPI2-negative patients following initial treatment for endometrial cancer. Furthermore, we investigated the correlation between TFPI2 expression in serum and tissue samples in high-risk histological types of endometrial cancer. This study aimed to provide a more comprehensive understanding of the condition of patients with endometrial cancer based on serum TFPI2 levels. Overall, our findings demonstrate the potential utility of TFPI2 in the preoperative diagnosis of patients with endometrial cancer.

Methods

Study design and participants

This was a retrospective single-center study. The protocol was reviewed and approved by the Institutional Review Board of Kanagawa Cancer Center (approval no. Ethics-2021-60). We preoperatively collected serum samples from 401 patients with endometrial cancer who underwent primary treatment at the Kanagawa Cancer Center between 2010 and 2020. Among them, we included patients who were treated with primary surgery for endometrial cancer of the following histological types: endometrioid carcinoma with any grade, serous carcinoma, clear cell carcinoma, carcinosarcoma, and mixed-type adenocarcinoma, according to the World Health Organization classification 5th edition [23]. We excluded patients with a history of other malignancies, those who underwent neoadjuvant therapy, those who did not undergo primary surgery, and those with neuroendocrine carcinoma or mucinous carcinoma of the endometrium. We defined the type I histological type as low-risk histology (endometrioid carcinoma grades 1 and 2) and type II histological type as high-risk histology (endometrioid carcinoma grade 3, serous carcinoma, clear cell carcinoma, carcinosarcoma, and mixed adenocarcinoma) based on the World Health Organization 2020 classification [23,24,25]. We used serum from 65 women as healthy controls, which were purchased from Bio IVT [26] for 31 cases and Trina Bioreactives AG [27] for 34 international healthy volunteers. The immunohistochemical cohort comprised 105 patients with high-risk histological features of endometrial cancer who underwent primary surgery between 2010 and 2015. We analyzed TFPI2 expression levels in serum and tissue samples from 33 patients for whom samples were available.

Serum data

We employed an automated immunoassay analyzer (AIA) system (TOSOH, Japan) as described previously [16, 20]. This system uses a sandwich-type one-step immunofluorometric assay involving two different anti-TFPI2 monoclonal antibodies produced by TOSOH corporation, and immunoreaction reagents are provided as the E-test “TOSOH” II (TFPI2) (product number: 0025245; TOSOH, Japan). For our experiments, we utilized recombinant TFPI2 protein, which was obtained from SP 2/0 cells transfected with a TFPI2 expression vector and added it to the sample dilution buffer.

Tissue microarray

We used tissue microarrays (TMA) of high-risk histological endometrial carcinomas that were surgically removed at the Kanagawa Cancer Center. Five 2.0 mm-diameter tissue cores of formalin-fixed paraffin-embedded (FFPE) tissues for each case were prepared for TMA from three sites showing representative primary tumor histology, a site from the non-neoplastic endometrium, and a site from the non-neoplastic myometrium as normal controls.

Evaluation of TFPI2 immunohistochemistry for TMA

Immunohistochemistry (IHC) was performed using the same concentration of mouse monoclonal anti-TFPI2 antibody (sc-48380, diluted 1:200, Santa Cruz Biotechnology, Inc., CA, USA) as the primary antibody to ensure consistency of our experimental approach [20]. Immunohistological evaluations were conducted by two individuals (MU and YO) under the supervision of a board-certified surgical pathologist (SS). We selected and evaluated the cores that exhibited the highest staining intensity among the three cores. The IHC score was calculated by multiplying the intensity (0, 1+, 2+, 3+) and labeled tumor cell frequency (%) at the core of the most predominant intensity. For example, when the most predominant intensity was 2+, and the proportion of 2 + accounted for 30%, the IHC score was calculated as 2 × 30. An IHC score of 0 was considered negative. We analyzed the agreement between serum TFPI2 levels and IHC scores in high-risk histological endometrial cancer.

Clinical data analysis

The cutoff value of serum TFPI2 levels was set at 191 pg/mL, which is clinically established for the diagnosis of ovarian cancer, and a value less than the cutoff was defined as negative, while a higher value was defined as positive [18, 28]. We calculated sensitivity and specificity for endometrial cancer and analyzed clinical characteristics of patients, including serum TFPI2 level, age, parity, serum CA125 level, treatment period, histology, International Federation of Gynecology and Obstetrics (FIGO) stage (2009) [29], TNM classification (8th edition) [30], and ascites fluid cytology. The positivity rates of tumor markers (TFPI2, CA125, CA19-9, and CEA) were calculated, and the correlations between TFPI2 and other tumor markers were analyzed. Furthermore, we calculated the positivity of the combination of TFPI2 and CA125. The 5-year overall survival was compared between TFPI2-positive and TFPI2-negative groups (n = 328). Furthermore, the sensitivity and specificity were evaluated using the existing cutoff value of 177 pg/mL, which has been reported to be useful for predicting the prognosis of patients with endometrial cancer [21].

Statistics

Continuous variables were assessed for normality using the Shapiro–Wilk test. The Mann–Whitney U test was used for comparisons between two groups; the Kruskal–Wallis test was employed for comparisons among three or more groups. Categorical variables were analyzed using Fisher’s exact test. Pearson’s correlations were used for correlation analysis, and Cohen’s kappa test was used for agreement analysis. The 5-year overall survival rates were estimated using the Kaplan–Meier method and compared using the log-rank test. The Cox proportional hazards regression model was used for multivariate analysis of the 5-year overall survival. We used SPSS Statistics version 19 (IBM, Armonk, NY, USA) and EZR version 1.61 (64-bit) [31]. p < 0.05 was considered to indicate a statistically significant difference.

Results

Effectiveness of TFPI2 as a serum diagnostic marker

The clinicopathological characteristics of patients with endometrial cancer are shown in Table 1. In total, 328 patients were included and compared with 65 healthy controls (Additional file 1).

Table 1 Baseline characteristics of 328 patients with endometrial cancer in the TFPI2-positive and TFPI2-negative groups
Full size table

Patients with endometrial cancer had significantly higher serum TFPI2 levels than controls (median 196.7 pg/mL vs. 83.3 pg/mL; p < 0.001) (Fig. 1). Sensitivity and specificity were 54.3% and 95.4% for a cutoff value of 191 pg/mL and 66.2% and 93.8% for a cutoff value of 177 pg/mL (Additional file 2 A, 2B, Additional file 3) [21]. In the present study, we set the cutoff value at 191 pg/mL, which is clinically established for the diagnosis of ovarian cancer [18, 28].

Fig. 1
figure 1

Comparison of serum TFPI2 levels between healthy controls and patients with endometrial cancer. The Mann–Whitney U test showed that the serum TFPI2 levels of patients with endometrial cancer were significantly higher than those of healthy controls (median 196.7 pg/mL vs. 83.3 pg/mL). *** indicates p < 0.001. The dotted line indicates 191 pg/mL. TFPI2, tissue factor pathway inhibitor 2

Full size image
Fig. 2
figure 2

Serum TFPI2 levels of patients with endometrial cancer by histological types and FIGO stages. (A) TFPI2 levels were significantly higher in the high-risk group than in the low-risk group (220.8 pg/mL vs. 187.7 pg/mL, p < 0.001). (B) TFPI2 levels were significantly higher in stage III and IV than in stage I (stage I, 189.6 pg/mL; stage III, 230.9 pg/mL; stage IV, 312.5 pg/mL). *** indicates p < 0.001 in the Mann–Whitney U test. The dotted line indicates 191 pg/mL. CS, carcinosarcoma; EMG, endometrioid carcinoma grade; FIGO, International Federation of Gynecology and Obstetrics; TFPI2, tissue factor pathway inhibitor 2

Full size image

Clinicopathological characteristics of the TFPI2-positive and TFPI2-negative groups

Patients in the TFPI2-positive group were older than those in the TFPI2-negative group (median 57.0 years vs. 59.5 years, p = 0.005) and had higher proportions of CA125-positive cases (10.0% vs. 28.7%, p < 0.001), high-risk histological cases (14.7% vs. 35.4%, p < 0.001), and cases with stage II or higher according to the FIGO classification (13.3% vs. 32.0%, p < 0.001) (Table 1). There were significant differences in surgical procedures and whether patients underwent adjuvant chemotherapy between TFPI2-negative group and TFPI2-positive group (Additional file 4).

Next, we evaluated serum TFPI2 levels according to the histological type and FIGO stage. TFPI2 levels were significantly higher in the high-risk histological group than in the low-risk histological group (220.8 pg/mL vs. 187.7 pg/mL, p < 0.001) (Fig. 2A). The positivity rate was higher in the high-risk histological group than in the low-risk histological group (74.1% vs. 47.3%, p < 0.001) (Additional file 5 A). Compared to stage I, stages III and IV had significantly higher TFPI2 levels (stage I, 189.6 pg/mL; stage III, 230.9 pg/mL; stage IV, 312.5 pg/mL; p < 0.001, Fig. 2B). The TFPI2 positivity rate was also higher in stages III and IV (stage I, 48.2%; stage II, 63.2%; stage III, 75.6%; stage IV, 84.6%; p < 0.001) (Additional file 5B, Additional file 6).

Correlation between TFPI2 and other tumor markers

The positivity rates for TFPI2, CA125, CA19-9, and CEA were 54.3%, 20.1%, 24.7%, and 20.1%, respectively (Additional file 7). Combining TFPI2 and CA125 increased the positivity to 58.8% (Additional file 7). The correlation between serum TFPI2 and other tumor markers was low or absent; for example, TFPI2 and CA125 only had a low correlation (r = 0.203; p < 0.001, kappa coefficient, 0.176; 95% confidence interval [CI], 0.074–0.278) (Fig. 3, Additional file 8, Additional file 9.

Fig. 3
figure 3

Pearson’s correlation analysis between serum TFPI2 and CA125 levels. There was a low correlation between serum TFPI2 and CA125 levels (r = 0.203, p < 0.001). CA, cancer antigen; TFPI2, tissue factor pathway inhibitor 2

Full size image

Efficacy of serum TFPI2 for prognostic prediction

We analyzed the association between preoperative serum tumor marker levels and the 5-year survival rates. The 5-year survival rate was significantly worse in the TFPI2-positive group than in the TFPI2-negative group (hazard ratio [HR], 8.22; 95% CI, 2.49–27.1; log-rank test, p < 0.001) (Fig. 4). Serum TFPI2, CA125, CA19-9, and CEA levels, parity, treatment period, and age were used as covariates in the Cox proportional hazards regression model. Multivariate analysis showed that TFPI2 positivity was associated with poor prognosis (HR, 6.00; 95% CI, 1.79–20.1); CA125 positivity (HR, 3.05; 95% CI, 1.34–6.95) and age over 65 years (HR, 1.05; 95% CI, 1.01–1.09) were also associated with poor prognosis (Table 2).

Fig. 4
figure 4

Kaplan–Meier curve of preoperative serum TFPI2 levels and 5-year survival. Serum TFPI2 positivity is associated with poor prognosis in patients with endometrial cancer. P-value was calculated using the log-rank test (HR, 8.22; 95% CI, 2.49–27.1; p < 0.001). CI, confidence interval; HR, hazard ratio; TFPI2, tissue factor pathway inhibitor 2

Full size image
Table 2 Multivariate analysis of prognostic factors for overall survival
Full size table

Immunohistological analysis of TFPI2

IHC for TFPI2 was performed on 105 patients with endometrial cancer of high-risk histological types, and immunolabeled TFPI2 was detected in the cytoplasm of the tumor cells (Fig. 5). An IHC score greater than 0 was considered positive, and the positivity rate was 37.1% (Additional file 10). There were no significant differences in age, parity, CA125 levels, treatment period, histological type, FIGO stage, or TNM stage between the positive and negative groups. None of the non-neoplastic control tissues from the endometrium or the myometrium exhibited TFPI2 expression (Additional file 11). There were no significant differences in the IHC scores according to histological type (p = 0.427) (Fig. 6). In addition, the agreement rate of TFPI2 positivity between the serum and tissue was indicated by a kappa coefficient of -0.039, demonstrating poor agreement (95% CI, -0.344–0.265); there were no significant differences between serum TFPI2 levels of the IHC-positive and IHC-negative groups (p = 0.427) (Additional file 12, Additional file 13).

Fig. 5
figure 5

Microscopic immunohistochemistry images of TFPI2 expression in endometrial cancer tissues. Microscopic images show immunohistochemistry for TFPI2 in (A) clear cell carcinoma with intensity 0, (B) endometrioid carcinoma grade 3 with intensity 1+, (C) clear cell carcinoma with intensity 2+, and (D) serous carcinoma with intensity 3+. There were both IHC-positive and IHC-negative cases for all histological types. HE, hematoxylin eosin; IHC, immunohistochemistry; TFPI2, tissue factor pathway inhibitor 2

Full size image
Fig. 6
figure 6

Boxplots of IHC scores by each histological type. The Mann–Whitney U test showed no significant differences in IHC scores among histological types (p = 0.427). CS, carcinosarcoma; EMG, endometrioid carcinoma grade; IHC, immunohistochemistry; TFPI2, tissue factor pathway inhibitor 2

Full size image

Discussion

This study demonstrates the potential of TFPI2 as a serum diagnostic marker for endometrial cancer, using the cutoff value of 191 pg/mL clinically applied for ovarian cancer, with elevated performance in combination with serum CA125. The evaluation of serum TFPI2 levels before surgery alone also raises the possibility of predicting patient prognosis. The serum level of TFPI2 was not simply correlated with its expression in tumor cells.

Kawaguchi et al. reported that TFPI2 with a cutoff value of 177 pg/mL may serve as an independent prognostic factor in such patients (69.4% sensitivity; 69.4% specificity) [21]. In the present study, we validated the results of this previous study with an increased number of patients from another institution. Using a cutoff value of 191 pg/mL, the established value for ovarian cancer diagnosis, decreased the sensitivity from 66.2 to 54.3% but slightly elevated the specificity from 93.8 to 95.4%, compared to a cutoff value of 177 pg/mL. This decrease in sensitivity partly improved to 58.8% when combined with serum CA125 testing. TFPI2 is independent of CA125; therefore, a combination of the two biomarkers may benefit more patients with endometrial cancer in terms of diagnosis and disease monitoring. However, we should use these markers in combination with other examinations, such as sonography and endometrial cytology, because the individual sensitivity values of these markers are not sufficiently high for accurate diagnosis.

Generally, TFPI2 acts as a suppressor of cancer progression by inhibiting the proteolytic degradation of proteins, leading to the suppression of tumor proliferation, invasion, and angiogenesis, as well as the promotion of apoptosis [11, 32]. Epigenetic silencing by the aberrant methylation of CpG islands in the TFPI-2 promoter has been reported in many types of cancer [33,34,35]. In breast and pancreatic cancers, low TFPI2 expression in tumor cells is associated with poor patient survival [36, 37]. One study reported the pro-tumorigenic effect of TFPI2 on the hepatocyte growth factor-induced invasion of hepatocellular carcinoma cells [32]. However, the effect of serum TFPI2 levels on cancer progression remains unclear. The present study revealed that serum TFPI2 levels were elevated in patients with high-risk histological types and advanced stages of endometrial cancer. Although the pro-tumorigenic aspects of TFPI2 remain to be clarified biologically in endometrioid cancer, this finding indicates that even with low-risk histology on pathologic examination or in early-stage disease by imaging, elevated serum TFPI2 levels should lead to consideration of the possibility of high-risk histology or advanced disease when planning preoperative treatment strategies.

We demonstrated that TFPI2 was detected via IHC in all high-risk histological types. In a previous study involving 55 patients with endometrial cancer, TFPI2 IHC was able to differentiate clear cell carcinoma from other histological types with a sensitivity of 100% and specificity of 73.8% [22]. However, in the present study, no differences in the IHC scores were observed across the various histological types of endometrial carcinomas. Nonetheless, several factors differed between the two studies, i.e., the sample sizes were different, and the target populations were notably different. The previous study included grade 1 and 2 endometrioid carcinoma (11 out of 55) but excluded carcinosarcoma and mixed-type endometrial carcinomas. Furthermore, a prior study reported a localized positive staining pattern for TFPI2, which might have been overlooked in our TMA examination. We used the same antibody as that used by Kawaguchi et al. [22] but adopted the same concentration as that used in our previous research on ovarian cancers [20], which differed from that used by Kawaguchi et al. [22]. The sensitivity and specificity of TFPI2 immunostaining for clear cell endometrial carcinoma require further characterization.

We conducted the first analysis comparing TFPI2 expression levels in serum and tissues derived from the same patients, demonstrating a low agreement rate. This may be partly consistent with our findings in ovarian cancer that TFPI2 was exclusively found in clear cell carcinoma [20], whereas elevated serum TFPI2 was observed in 29.4% of serous adenocarcinoma cases [16, 18]. It is conceivable that the secretory capacity of TFPI2 varies among cells; however, several other factors can be postulated for this apparent discrepancy. First, in some cases, TFPI2 expression may be localized and prone to be missed by immunostaining of representative sections or TMAs [22]. Second, TFPI2 secreted from tumor cells can be trapped on the cell surface or in the extracellular matrix (ECM), preventing its migration into the bloodstream. Both TFPI and TFPI2 carry strong positive charges at their C-terminal regions, allowing them to adhere to negatively charged ECM or cell surface heparan sulfate proteoglycans, such as glypican-3, via glycosylphosphatidylinositol anchors [38]. Third, TFPI2 is released or secreted into the serum by non-cancer cells physiologically producing TFPI2, such as vascular endothelial cells [11], platelets [12], or macrophages [13], under conditions that are currently undefined. In endometrial cancer, discordance between serum levels and tissue IHC staining for CA125 has been reported, and CA125 production from sources other than endometrial cancer tissues has been suggested [39]. The mechanism of elevated serum TFPI2 remains largely unknown; thus, understanding this mechanism may promote the clinical use of this biomarker.

Limitations

This study has several limitations. First, it was a retrospective single-center study, and the results may not be generalizable to other settings. Second, we compared the serum TFPI2 levels of patients with endometrial cancer with those from commercially available serum samples of international healthy volunteers used as controls. Consequently, we were unable to [1] accurately assess the incidence with precise sensitivity and specificity, [2] determine an accurate cutoff value for TFPI2 from the ROC curve, and [3] thoroughly evaluate the efficacy of combining of TFPI2 with CA125 versus using CA125 alone in terms of specificity, positive predictive value, and negative predictive value. To address these limitations, a future prospective study is necessary. Third, we did not use the FIGO 2023 classification but instead applied the FIGO 2009 classification for endometrial cancer owing to the lack of molecular and biological data [29, 40]. Fourth, when evaluating protein expression levels by IHC using FFPE specimens, the heterogeneity of formalin fixation intensity within and between specimens is known to affect stainability, which is another limitation of this study. Paired TFPI2 analyses of IHC using FFPE specimens and transcriptomics with frozen tissues in at least some cases are needed in future research. Fifth, we attempted to incorporate other factors such as family histories of endometriosis, endometrial cancer, and other types of cancer in multivariate analysis for prognosis of patients with endometrial cancer. However, our medical records did not provide complete data on these variables, thus we were unable to include these factors in our analysis. To address this limitation, we recommend conducting prospective, multicenter collaborative research to ensure the clinical application of TFPI2 as a serum biomarker for endometrial cancer.

Conclusions

Serum TFPI2 has been suggested as a useful diagnostic and prognostic predictor of endometrial cancer.