Combination of serum CST1 and HE4 for early diagnosis of endometrial cancer

Study participants

We collected 300 retrospective serum samples from January 2021 to December 2022 from the Fujian Provincial Maternal and Child Health Hospital, including 100 early EC patients, 100 EBL patients, and 100 HCs. All study participants provided informed consent, and the study design was approved by the appropriate ethics review board (Grant No. 2023KYLLRD01058). The 100 early EC patients at TNM I/II stage were diagnosed by medically histopathology or cytology, and received no anti-cancer treatment, such as surgery, radiotherapy, or molecular targeted therapy. A total of 100 EBL patients were admitted to the Fujian Provincial Maternal and Child Health Hospital for primary treatment, including endometrial thickening, uterine fibroids, and endometriosis, during the same period. The 100 HCs were age-matched women receiving physical examinations at Fujian Provincial Maternal and Child Health Hospital, and showed no evidence of acute or chronic disease. All serum samples were collected in strict accordance with the following requirements: collection of five mL of peripheral blood from each subject prior to surgery; serum was separated at 3,000 rpm for 5 min, and frozen at −80 °C before use. This study was approved by the Institutional Review Board of Fujian Provincial Maternal and Child Health Hospital, and all enrolled participants provided written informed consent.

Immunohistochemistry

CST1 primary antibody (1:100, Abcam, Cambridge, UK) was added dropwise to the repaired and stained tissue sections and incubated overnight at 4 °C (Fuzhou Maixin Biotechnology Limited Company, Fuzhou, China). The tissues were then incubated with biotinylated secondary antibody and streptomyces anti-biotin protein-peroxidase solution, dried, and sealed by DAB color development and hematoxylin re-staining. Each section was randomly observed at five high-magnification fields and analyzed by two experienced pathologists. The staining intensity was graded as 0 (negative), 1 (weakly positive), 2 (moderately positive), or 3 (strongly positive).

Detection of serum CST1

The serum test kit for CST1 was the Human Cysteine Protease Inhibitor 1 Assay Kit (Jiangsu Enzyme-Free Industrial Co, Jiangsu, China). Serum CST1 was measured using enzyme-linked immunosorbent assay (ELISA), and the threshold value for serum CST1 in patients with early EC was determined by the highest discriminatory capacity (maximum sum of sensitivity and specificity) through the receiver operating characteristic (ROC) curve.

In addition, the performance of ELISA for the detection of serum CST1 was evaluated in terms of linearity, limit of detection, accuracy, precision, and resistance to interference.

For the detection limit and linearity, the CST1 calibrator (0 µg/L) was subjected to 20 intra-batch replicate determinations and the mean ($\overline{X}$) and standard deviation (SD) of the 20 OD values measured were calculated, with $\overline{X}$ ± 2 SD equal to the detection limit. After serial multiplicative dilutions, each dilution concentration of the CST1 calibrator was tested twice, and the mean OD was calculated. A linear regression curve of CST1 was plotted using the dilution concentration (C) as the vertical coordinate Y and the OD value as the horizontal coordinate X. The linear regression equation “C = a∗OD + b” was obtained, where a and b are the ELISA parameters indicating the conversion relationship between the OD value and CST1 level of the sample. The results demonstrated that the minimum detection limit for CST1 was 3.294 g/L (Table S1) and the correlation coefficient for CST1 in the range 0–1,600 µg/L was 0.993 (Fig. S1, Table S2).

To assess accuracy, the CST1 calibrators at concentrations of 100, 800, and 1,600 µg/L were mixed with certain concentrations of serum samples in a volume ratio of 1:9. The mixed samples were tested twice, the diluted samples were tested three times, and the mean level was calculated. The recoveries were calculated using the following formula: ${\displaystyle R=\frac{\left(\mathrm{V}0+\mathrm{V}1\right)\times \mathrm{C}-\mathrm{V}1\mathrm{C}1}{\mathrm{V}0\mathrm{C}0}\times 100\text{\%}}$ where V0 and C0 are the volume and level of calibrator, respectively; V1 and C1 are the volumes and levels of the serum samples, respectively; C is the concentration of the mixed samples; and R is the recovery rate. An 85%–115% recovery rate was acceptable. The results showed a recovery rate of 102.82% for CST1 (Table S3).

Precision was assessed according to the CLSI EP15-A2 standard. Samples with low and high CST1 levels were tested 20 times separately, and intra-batch CV < 10% was considered acceptable. CV values of 4.92% and 8.35% were determined for low and high CST1 samples, respectively (Table S4). Low and high CST1 samples were tested separately for five days, and each level was repeated three times to obtain mean values; inter-batch CV < 15% was considered acceptable. The results showed CV values of 7.63% and 9.27% for the low and high CST1 levels, respectively (Table S5).

Interference assessment was carried out according to the WS-T 416-2013 interference test guidelines: serum samples with CST1 levels of 48.5 and 87.61 µg/L were added to 2 g/L of hemoglobin, 342 µmol/L of bilirubin, and 37 mmol/L of triglycerides in a ratio of 1:9. Each sample was measured three times, and a relative deviation (δ) of <10% was considered acceptable. The results excluded that hemoglobin, bilirubin and triglycerides did not significantly interfere with CST1 determination (Table S6).

Detection of serum CA125 and HE4

Serum CA125 and HE4 levels were measured by chemiluminescent immunoassay kits on the Cobas 602 analyzer (Roche, Mannheim, Germany) according to the manufacturer’s instructions. The cutoff values for serum CA125 and HE4 levels in patients with early EC were determined based on the highest discriminatory capacity (maximum sum of sensitivity and specificity) using receiver operating characteristic (ROC) curves.

Statistical analysis

The SPSS software (version 25.0) was used for statistical analysis. One-way analysis of variance (ANOVA) was used for comparisons among three or more groups if the chi-square test was satisfied, and the Student–Newman–Keuls (SNK) method was chosen for multiple comparisons between groups. The Kruskal–Wallis test was used for non-parametric comparisons, and Kruskal–Wallis one-way ANOVA (K-sample comparison) was used for two-way comparisons. GraphPad Prism 9 was used to produce ROC curves, and the AUC values were calculated. P < 0.05 was considered statistically significant.

Expression of CST1 protein in EC cancer tissues

Immunohistochemical results showed that the CST1 protein exhibited focal or diffuse distribution of tan granules in EC early-stage cancer tissues (Fig. 1A), whereas not expressed or weakly expressed in paired paracancerous tissues (Fig. 1B). Moreover, as shown in Table 1, CST1 was positive in 67.6% of CST1 protein in EC cancer tissues, which was significantly higher than that in paraneoplastic tissues (P < 0.01). The above studies indicated that CST1 was aberrantly highly expressed in the cancer tissues of patients with early EC.

Levels of serum biomarkers in the training set

As shown in Table 2 and Figs. 2A–2C, the levels of serum CST1 and HE4 were significantly higher in the early EC group than in the EBL/HC group (P < 0.05). However, there was no significant difference in serum CA125 levels between the early EC and EBL/HC groups (P > 0.05).

Diagnostic performance of serum biomarkers in the training set

As shown in Table 3 and Fig. 2D, serum CST1 had an AUC of 0.715, sensitivity of 31.3%, and specificity of 90.3% for patients with early EC; serum CA125 and HE4 had AUCs of 0.577 and 0.706, sensitivity of 20.9% and 23.9%, and specificity of 88.1% and 95.5%, respectively, for patients with early EC. Among all possible combinations, the combination of serum CST1 and HE4 had the best diagnostic performance for early EC, with a maximum AUC of 0.788 (95% CI [0.720–0.856]), sensitivity of 49.3%, specificity of 92.5%, maximum positive predictive value of 76.7%, and negative predictive value of 78.5%.

Serum biomarker levels in the validation set

Table 4 and Figs. 3A, 3B showed that the levels of serum CST1 and HE4 expression were significantly higher in the early EC group than in the EBL/HC groups (P < 0.05). However, the difference in serum CA125 levels was not significant.

Diagnostic performance of serum biomarkers in the validation set

The results of the validation set demonstrated that serum CST1 had an AUC of 0.719, with a sensitivity of 39.4% and specificity of 86.4%; serum HE4 had an AUC of 0.686, with a sensitivity of 24.3% and specificity of 96.9%; the combination of serum CST1 and HE4 had an AUC of 0.824, with a sensitivity of 48.5% and specificity of 92.4% (Table 5, Fig. 3C).