The ratio of serum Angiopoietin-1 to Angiopoietin-2 in patients with cervical cancer is a valuable diagnostic and prognostic biomarker

Study groups and inclusion criteria

We conducted a non-matched case-control study using serum samples obtained from patients who received treatment at Wuhan Union Hospital (Wuhan, Hubei, China) between February 2012 to March 2014, including 44 patients with CIN, 77 patients with cervical cancer (61 squamous cell carcinomas, CSCC, and 16 adenocarcinoma, CADC) and 43 patients with benign disorders of uterus such as myoma or adenomyosis as controls. The diagnoses were pathologically verified and patients with medical diseases such as other type of cancer or inflammatory, atherosclerotic and connective tissue disease were excluded. The age range of the 164 patients was 23–72 years with mean of 44.26 years, and there was no significant difference in age among the cervical cancer, CIN and control groups. All patients or their next of kin provided written informed consent for the collection of samples and subsequent research. The present study was approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology (IORG0003571).

Clinical definitions

The medical records of the patients with cervical cancer were reviewed to collect clinical and pathological characters, including age at diagnosis, FIGO stage, pathological type, tumor differentiation, pelvic lymph node metastasis, tumor size, and lymphovascular space invasion (LVSI). The patients were classified according to the revised FIGO staging system for cervical cancer of 2009. The tumor size was the largest diameter of tumor determined by the attending gynecologic oncologist during a pelvic examination preceding surgery. The patients of IA1 stage underwent hysterectomy, and patients between IB1 and IIB underwent radical hysterectomy and pelvic lymph node dissection. Patients with ≥IIB stage underwent radiotherapy or radiotherapy combined with chemotherapy. Stage-specific follow-up after treatment was performed in accordance with the FIGO guidelines. Progression-free survival (PFS) was defined as the time from treatment to the first appearance of tumor recurrence or to the date of last contact. Overall survival (OS) was defined as the time from treatment to death from any cause or to the date of last contact.

Sample collection, ELISA, and IHC

Serum samples were collected before primary treatment such as surgery, radiation, or chemotherapy. Blood was drawn into serum tubes (10 ml) and centrifuged at 2,200 G/min for 10 min. Serum was harvested, aliquoted and stored at −80 °C until usage.

The concentration of sAng-1 and sAng-2 in serum was assayed by a standardized sandwich enzyme-linked immunosorbent assay (ELISA) in triplicate according to the protocol recommended by the manufacturer (Uscn Life Science Inc., Wuhan, China).

For immunohistochemistry (IHC), paraffin-embedded sections (4 µm) were hydrated by alcohol and then subjected to antigen retrieval in 10 mM sodium citrate buffer (pH 6.0). After being treated with 3% H₂O₂ for 10 min, samples were incubated with primary antibodies overnight at 4 °C. Three primary antibodies were used: human antibodies for detection of Ang-1 (AF923, Polyclonal, Goat IgG), Ang-2 (AF623, Polyclonal, Goat IgG) and CD34 (AF7227, Polyclonal, Sheep IgG) in cervical tissue were all purchased from R&D (Minneapolis, MN, USA); the working concentrations of antibodies were 5 µg/ml. Standard horseradish peroxidase staining using an appropriate biotinylated secondary antibody (ZSGB-BIO, Peking, PRC), the elite ABC kit (Avidin: Biotinylated enzyme Complex; ZSGB-BIO, Peking, PRC) and diaminobenzidine (ZSGB-BIO, Peking, PRC) were performed according to the manufacturers’ instructions. Sections were counterstained with Gill’s haematoxylin (Sigma, Aldrich, St. Louis, MO USA). The expression of angiopoietins in epithelial cells and endothelial cells (EC) in cervical tissues were separately evaluated and scored as negative staining and positive staining. MVD was assessed using CD34 (Yang et al., 2009).

Statistical analysis

Statistical analysis was carried out with SPSS 13.0 statistical software (SPSS, Chicago, IL, USA). Continuous variables were expressed as median and range; the data from each category was presented as frequency and percentage. Non-parametric tests were used for comparisons between different groups. The differences in sAng-1, sAng-2, and sAng-1/sAng-2 ratio between groups were analyzed with the Kruskal–Wallis test combined with Dunn’s multiple comparison, the Jonckheere–Terpstra test or Mann–Whitney U test, as appropriate. ROC curve was used to analyze the diagnostic value of individual factors. Correlations were analyzed with the Spearman’s correlation coefficient. Survival analyses were performed and Kaplan–Meier survival curves were generated; the survival distributions were compared by log-rank test. All tests were two-tailed, and a P value of less than 0.05 was considered to be statistically significant.

sAng-2 concentration and sAng-1/sAng-2 ratio are altered in patients with cervical neoplasia

The sAng-1and sAng-2 in 43 patients of normal control, 44 CIN patients and 77 cervical cancer patients (61 CSCC patients and 16 CADC patients) were detected by ELISA. Given that CINs are the precancerous lesions of CSCCs, we first compared the sAng-1, sAng-2 and sAng-1/sAng-2 ratio in the normal, CIN and CSCC patients. We found that the sAng-1 concentration was not significantly different between these three groups, with a median concentration of 58.13 pg/ml (range, 6.79–194.1 pg/ml) in the control group, 28.53 pg/ml (range, 7.22–248.5 pg/ml) in the CIN group and 37.79 pg/ml (range, 1.57–361.2 pg/ml) in the CSCC group (Kruskal–Wallis test, P = 0.068, Fig. 1A). However, there was a significant difference in sAng-2 between the three groups (Kruskal–Wallis test, P = 0.000), and from normal control (28.87(1.86–235.7) pg/ml), to CIN (44.35(0.70–395.6) pg/ml), then to CSCC (109.90(1.09–431.70) pg/ml), in that with the increasing severity of cervical lesions, the concentration of sAng-2 gradually increased (Jonckheere–Terpstra test, P < 0.001; Fig. 1B). The sAng-1/sAng-2 ratio was also significantly different between the three groups (Kruskal–Wallis test, P < 0.001; Fig. 1C). The sAng-1/sAng-2 ratio decreased gradually from the control group (median, 2.07; range, 0.25–30.36), to the CIN group (median, 0.85; range, 0.08–17.54), then to the CSCC group (median,0.54; range, 0.02–42.91) (Jonckheere–Terpstra test, P < 0.001; Fig. 1C).

We then compared the concentration of sAng-1and sAng-2 and the sAng-1/sAng-2 ratio in the CADC group relative to the normal groups. The differences in sAng-1(37.79 vs 58.13 pg/ml) and sAng-2 concentrations (62.97 vs 28.87 pg/m) were not significant (Figs. 1A and 1B). However, the sAng-1/sAng-2 ratio was significantly decreased in CADC patients compared with the normal group (0.58 (0.06–33.33) vs 2.07 (0.25–30.36), P = 0.008) (Fig. 1C).

sAng-2 concentration and the sAng-1/sAng-2 ratio are valuable diagnostic biomarkers for cervical lesions

Having determined the differential sAng-2 concentrations and sAng-1/sAng-2 ratio in cervical cancer lesions compared with normal, we then used ROC curves to test whether the sAng-2 concentration and sAng-1/sAng-2 ratio can serve as biomarkers to distinguish CSCC from normal and CIN. The data of the 43 patients of normal control, the 44 CIN patients, and the 61 CSCC patients were included in the analysis. The area under curve (AUC) of the ROC curves of sAng-2 for discriminating CSCC from normal (Fig. 2A), CSCC from CIN (Fig. 2B), CSCC from normal and CIN (Fig. 2C), CIN and CSCC from normal (Fig. 2D), and invasive cervical cancer from normal (Fig. 2E) were 0.769, 0.662, 0.715, 0.706 and 0.744, respectively. The sAng-2 cutoff value of 48.24 pg/ml provides a high diagnostic accuracy to discriminate CSCC from normal, with a sensitivity of 76.67% and a specificity of 74.42% (Fig. 2F).

The AUC value of sAng-1/ sAng-2 as a diagnostic biomarker for normal to CIN, normal to CSCC, normal to CIN and CSCC and normal to cervical cancer was 0.657, 0.719, 0.693 and 0.705 (Figs. 2G–2J), respectively. The cutoff value of 1.09 had a sensitivity of 68.83% and specificity of 72.09% to distinguish CSCC from normal (Fig. 2K).

sAng-1 and sAng-2 correlate to clinicopathological characteristics in patients with cervical cancer

The correlations between the concentrations of serum angiopoietins and clinicopathological characteristics of cervical cancer were analyzed in a subset of 77 patients with cervical cancer, including 41 stage IA1 to IB1, 20 stage IB2 to IIA and 16 stage ≥IIB based on classification criteria (FIGO, 2009). Characteristics of the patients are summarized in Table 1. We found that sAng-1 concentration was significantly decreased in groups of stage ≥IB2, tumor size >2 cm and LVSI compared with stages IA1-IB1, tumor size ≤2 cm, or no LVSI, respectively. sAng-1 concentrations had no significant correlation with differentiation and pelvic lymphatic status. Although the sAng-2 concentration was significantly increased in the poorly differentiated group, no significant correlation with other clinicopathological characteristics were noted. sAng-1/sAng-2 was significantly decreased in stages ≥IB2, poor differentiation and LVSI, but had no significant correlation with pelvic lymphatic metastasis status and tumor size (Table 2).

sAng-2 concentration positively correlates with Ang-2 expression on epithelia and MVD in cervical tissues

In a subset of 25 patients with invasive cervical cancer (FIGO stage IA2–IIA) and 10 normal control patients, angiopoietin expression and MVD were evaluated in paraffin-embedded cervical tissue samples. Among the 25 cervical cancers, 21 cases were squamous cancer and four were adenocarcinoma; 15 patients had a tumor stage of FIGO IA1 to IB1 and 10 had a tumor stage of FIGO IB2 to IIA; seven tumors were well differentiated (G1), eight were moderately differentiated (G2) and 10 were poorly differentiated cancer (G3); nine patients had lymphovascular space invasion (LVSI); four patients had pelvic lymph node metastasis; the tumor sizes were greater than 2 cm in nine cases.

The IHC assays revealed that both Ang-1 and Ang-2 were highly expressed on tumor cells compared to normal cervical epithelia. The positive expression rate of Ang-1 in cervical cancer epithelia was 60% (15/25) and Ang-2 was 68% (17/25), while normal cervical epithelia had no positive staining. For the expression of Ang-1 and Ang-2 in endothelial cells, there was no significant difference between cervical cancer tissue specimens and normal controls (Figs. 3A and 3B).

MVD, as assessed by CD34 staining, was significantly increased in cervical cancer compared with normal (Fig. 3C). The patients with positive Ang-2 in cervical epithelia had significantly elevated sAng-2 concentration when compared to those with negative Ang-2 expression in cervical epithelia (80.89 (15.54–431.7) pg/ml vs 24.66 (1.86–299.3) pg/ml) (P = 0.029, Fig. 3D). In addition, the sAng-1 concentration did not correlate with MVD (r = 0.256, P = 0.138, Fig. 3E); the sAng-2 concentration positively correlated to MVD (r = 0.440, P = 0.008, Fig. 3F); the sAng-1/sAng-2 ratio was negatively correlated to MVD (r =  − 0.516, P = 0.002, Fig. 3G).

High sAng-1/sAng-2 ratio predicts poorer prognosis in cervical cancer patients

The prognostic value of sAng-1, -2 and sAng-1/sAng-2 was evaluated in 66 cervical cancer patients with follow-up data. The patients were followed up to October 2016. The median follow-up time was 33.62 months (range, 9–48 months). Of the 66 patients, 17 suffered from recurrence and 8 died. According to the ROC analyses, we choose 131.00 pg/ml, 156.40 pg/ml, and 0.26 as the cut-off values of sAng-1, sAng-2 and sAng-1/sAng-2, respectively, as these values were associated with a specificity of 95% for discriminating between invasive cervical cancer and normal. We found that there were no significant differences in the PFS and OS between the patients with high versus low sAng-1 or sAng-2 (Figs. 4A–4D). However, the sAng-1/sAng-2 ratio was significantly associated with PFS (Fig. 4E, P = 0.046) and OS (Fig. 4F, P = 0.040) in cervical cancer patients. In the low sAng-1/sAng-2 ratio group, 7 of 14 (50.0%) patients developed recurrent disease and 4 (28.6%) patients died. The estimated 3-year PFS and OS were 42.86% and 71.43%, respectively. In the high sAng-1/sAng-2 ratio group, 10 of 52 (19.23%) patients had recurrence and 4 (7.7%) patients died, resulting in an estimated 3-year PFS and OS of 78.96% and 91.55%, respectively.