Clinical value of detecting IQGAP3, B7-H4 and cyclooxygenase-2 in the diagnosis and prognostic evaluation of colorectal cancer

We retrospectively examined the medical records of patients with CRC at The Third Affiliated Hospital of Soochow University from May 2011 to May 2013. In order to obtain normal para-carcinoma tissues, same source of CRC patients from January 2018 to January 2019 were selected. The enrolled patients met the following criteria: (1) histological confirmed adenocarcinoma, (2) not received neoadjuvant chemoradiation, (3) underwent curative R0 resection, (4) with no evidence of distant metastasis. Patients were excluded if they had died or had incomplete clinicopathological data. Tumor stage was classified according to the 8th edition of TNM staging system. Follow-up was conducted by telephone calls, e-mails, and on-site visits. Besides, eighty-five healthy age-and sex-matched individuals were recruited from people who came for general health examinations in our hospital, with an exclusion of history of past diseases. Informed written consent was obtained from all patients and healthy individuals and this study was approved by the Ethics Committee of The Third Affiliated Hospital of Soochow University. The approval number was 201165.

Tumor tissue samples and normal para-carcinoma tissues were collected after surgery in CRC patients. Immunohistochemistry was performed to detect the IQGAP3, B7-H4 and COX-2 in tumor tissues from patients of May 2011 to May 2013 and normal para-carcinoma tissues from patients of January 2018 to January 2019. Specimens were cut into 5 mm thick slices and then subjected to routine deparaffinization and rehydration. The original anti-IQGAP3 antibody (ab88353, Abcam, Shanghai), anti-B7-H4 antibody (EPR20236, Abcam, Shanghai) and anti-COX-2 antibody (ab15191, Abcam, Shanghai) were used in a 1:200 dilution. After incubation at 4 °C, a microscopic examination was performed to determine the percentage of cells positively stained for IQGAP3, B7-H4 and COX-2.

All slices were analyzed by two clinical pathologists independently. When disagreement occurred, the third pathologist was asked to confirm the assessment. The inter-assessor agreement rate between the two pathologists was 95.6%. As for the classification of cases into low or high expression group, we referred to the study reported by Albasri AM [19]. The extent of staining was scored as follows: 0 (no staining), 1 (1–10% positive staining), 2 (10–50% positive staining), 3 (50–70% positive staining) and 4 (70–100% positive staining). The intensity of staining was scored as follows: 0 (negative), 1 (weakly positive), 2 (moderately positive) and 3 (strongly positive). The final expression score was determined as follows: − (scores 0), + (scores 1–3), ++ (scores 4–6) and +++ (scores > 6). For statistical analysis, the cases that scored as − and + were combined as the low expression group, which was compared to the high expression group consisted of the cases scored as ++ and +++.

Serum samples were collected prior to any therapy in CRC patients and as part of a routine examination in healthy individuals. All samples were collected in anti-coagulated tubes containing ethylene diamine tetraacetie acid (EDTA). Serum IQGAP3, B7-H4 and COX-2 levels were measured using commercially available ELISA kits (RENJIEBIO, Shanghai, China) at the tumor laboratory of The Third Affiliated Hospital of Soochow University. Coefficient of variation (CV) was used to validate the precision of the ELISA assays, which was defined as the ration of standard deviation and mean value. Serum CEA and CA19-9 levels were achieved directly for they were conventional examination items in our hospital.

Youden Index, which was calculated as sensitivity + specificity − 1, was used to determine the optimum cut-off values of the three tumor markers. The maximum Youden Index corresponds to the optimum cut-off value. We divided the patients into low expression group and high expression group based on the cut-off values of serum tumor markers.

One hundred eighteen CRC patients from May 2011 to May 2013 and sixty-two CRC patients from January 2018 to January 2019 who met the inclusion criteria shown above and eighty-five healthy individuals were analyzed in this study. Clinicopathological characteristics of 118 patients and their correlation with serum tumor marker levels were listed in Table 1. There were 49 female patients and 69 male patients, with age ≤ 60 and > 60 patients accounting for 44.9% and 55.1%, respectively. Owing to the late diagnosis, the majority of patients were in a later stage, with stage T3 and T4 patients accounting for 28.8% and 25.4%, respectively. The percentage of patients in stage I–II and III were 40.7% versus 59.3%, respectively. There appears to be an overrepresentation of the later stage CRC cases which would inflate the sensitivity and specificity of the discussed biomarkers.

IQGAP3, B7-H4 and COX-2 immunohistochemistry staining was quantitatively assessed and divided into low and high groups (Table 2). Positive staining showed a brown color. Immunohistochemical expression of IQGAP3 in tumor tissues was observed in the cytoplasm with different intensities, with 61 cases showing low expression (20% IQGAP3, Fig. 1a) and 57 cases showing high expression (70% IQGAP3, Fig. 1d). However, no IQGAP3 expression was found in normal para-carcinoma tissues (Fig. 1g). Similar condition was observed in B7-H4 immunohistochemistry staining in the cytoplasm, 52 cases showed low B7-H4 expression (10% B7-H4, Fig. 1b) and 66 cases showed high B7-H4 expression (80% B7-H4, Fig. 1e). Normal para-carcinoma tissues showed no B7-H4 expression (Fig. 1h). COX-2 immunoreactivity was also observed in the cytoplasm in all CRC cases, with 55 cases low expression (10% COX-2, Fig. 1c) and 63 cases high expression (80% COX-2, Fig. 1f). In comparison, COX-2 staining of normal para-carcinoma tissues was negative (Fig. 1i).

As shown in Table 3, 118 CRC patients were set as CRC group including T1–T4 groups and 85 healthy individuals were set as control group. Serum levels of IQGAP3, B7-H4, COX-2, CEA and CA19-9 in CRC group were significantly higher than those in control group (IQGAP3: 312.68 ± 101.91 vs. 191.97 ± 103.96, P < 0.001; B7-H4: 97.85 ± 15.56 vs. 64.78 ± 16.73, P = 0.003; COX-2: 48.55 ± 13.43 vs. 25.78 ± 16.09, P < 0.001; CEA: 6.62 ± 4.49 vs. 3.41 ± 1.27, P < 0.001; CA19-9: 29.92 ± 8.97 vs. 19.24 ± 10.08, P < 0.001). However, we can see that IQGAP3 in T1 and T2 groups were not higher than that in control group (T1 180.33 ± 71.10: Control 191.97 ± 103.96, P = 0.284; T2 185.57 ± 84.10: Control 191.97 ± 103.96, P = 0.412). Besides, CVs of each group were also shown.

The results of Spearman correlation analysis were shown in Fig. 2. IQGAP3, B7-H4 and COX-2 these three tumor markers were compared in three groups. Figure 2a indicated that there was a correlation between IQGAP3 and B7-H4 levels (r = 0.710, P < 0.001). Points show IQGAP3 or B7-H4 levels of each patient, with X axis indicating B7-H4 levels (ng/ml) and Y axis indicating IQGAP3 levels (pg/ml). Figure 2b showed that there was a correlation between IQGAP3 and COX-2 levels (r = 0.860, P < 0.001). Points show IQGAP3 or COX-2 levels of each patient, with X axis indicating COX-2 levels (ng/ml) and Y axis indicating IQGAP3 levels (pg/ml). Figure 2c also showed that there was a correlation between B7-H4 and COX-2 levels (r = 0.724, P < 0.001). Points show B7-H4 or COX-2 levels of each patient, with X axis indicating COX-2 levels (ng/ml) and Y axis indicating B7-H4 levels (ng/ml).

The correlation of serum IQGAP3, B7-H4 and COX-2 expression levels with clinicopathological features were shown in Table 1. Serum IQGAP3, B7-H4 and COX-2 levels were correlated with T stage, N stage, and TNM stage (P < 0.05). The correlation of serum IQGAP3, CEA and CA19-9 expression levels with clinicopathological features were shown in Additional file 1. Serum CEA and CA19-9 levels were correlated with T stage, N stage, differentiation degree and TNM stage (P < 0.05). The correlation of tissue IQGAP3, B7-H4 and COX-2 expression levels with clinicopathological features were shown in Table 2. Tissue IQGAP3, B7-H4 and COX-2 expression levels were also correlated with T stage, N stage and TNM stage (P < 0.05).

We analyzed the ROC curves of serum IQGAP3, B7-H4, COX-2, CEA and CA19-9 in CRC patients (Table 4). The IQGAP3 AUC was 0.799, with 95% confidence interval (CI) 0.736–0.861 (P < 0.001) (Fig. 3a). The B7-H4 AUC was 0.795 (95% CI 0.731–0.858, P < 0.001) (Fig. 3b) and the COX-2 AUC was 0.796 (95% CI 0.737–0.856, P < 0.001) (Fig. 3c). The CEA AUC was 0.786 (95% CI 0.725–0.847, P < 0.001) (Fig. 3h). The CA19-9 AUC was 0.777 (95% CI 0.714–0.840, P < 0.001) (Fig. 3i).

When the cut-off value for IQGAP3 was chosen as 186.535 pg/ml, as determined by the maximum value of Youden index, the sensitivity and specificity for IQGAP3 were 89.8 and 58.8%, respectively. When the cut-off value for B7-H4 was selected at 81.055 ng/ml, the sensitivity and specificity for B7-H4 were 88.1 and 62.4%, respectively. When the cut-off value for COX-2 was selected at 38.025 ng/ml, the sensitivity and specificity were 79.2 and 69.4%, respectively. When the cut-off value for CEA was chosen as 4.195 ng/ml, the sensitivity and specificity for CEA were 60.3 and 71.8%, respectively. When the cut-off value for CA19-9 was selected at 28.795 U/ml, the sensitivity and specificity for CA19-9 were 50.2 and 81.2%, respectively. IQGAP3 seemed to be inferior to B7-H4, COX-2, CEA and CA19-9 in specificity when detecting CRC, however, its sensitivity and AUC were all the highest among the three markers.

We analyzed the ROC curves of serum IQGAP3 + B7-H4, IQGAP3 + COX-2, B7-H4 + COX-2 in CRC patients (Table 4). As shown in Fig. 3d, the IQGAP3 + B7-H4 AUC was 0.876 (95% CI 0.825–0.927, P < 0.001). The IQGAP3 + COX-2 AUC was 0.889 (95% CI 0.842–0.936, P < 0.001) (Fig. 3e) and the B7-H4 + COX-2 AUC was 0.875 (95% CI 0.827–0.924, P < 0.001) (Fig. 3f). When determined by the maximum value of Youden index, the sensitivity and specificity for IQGAP3 + B7-H4 were 92.9 and 62.5%, respectively; the sensitivity and specificity for IQGAP3 + COX-2 were 91.8 and 56.8%, respectively; the sensitivity and specificity for B7-H4 + COX-2 were 90.6 and 63.5%, respectively. Thus, IQGAP3 + COX-2 showed superiority in detecting CRC.

We analyzed the ROC curves of IQGAP3 + B7-H4 + COX-2 in CRC patients (Table 4). As shown in Fig. 3g, the AUC of IQGAP3 + B7-H4 + COX-2 was 0.926 (95% CI 0.887–0.966, P < 0.001). When determined by the maximum value of Youden index, the sensitivity and specificity for IQGAP3 + B7-H4 + COX-2 were 94.1 and 74.5%. IQGAP3 + B7-H4 + COX-2 was superior to other markers in detecting CRC.

All 118 patients from May 2011 to May 2013 in this study were followed up for more than 5 years. The 5-year overall survival (OS) rate was 52.5%. Univariate analysis showed that tumor size, T stage, N stage, differentiation degree, TNM stage, both serum and tissue IQGAP3, B7-H4 and COX-2 levels and serum CEA and CA19-9 levels were significant prognostic factors for CRC. However, based on multivariate analysis, tumor size was not statistically significant as a prognostic risk factor (Table 5). Our study showed that the 5-year OS rate in both s-IQGAP3 and t-IQGAP3 low groups were significantly higher than those in s-IQGAP3 and t-IQGAP3 high groups, respectively (s-IQGAP3 low 85.7% vs. s-IQGAP3 high 48.1%, P = 0.007, Fig. 4a; t-IQGAP3 low 65.6% vs. t-IQGAP3 high 38.6%, P < 0.001, Fig. 4d). Similarly, the 5-year survival rates between B7-H4 or COX-2 levels low and high groups also showed statistical significance, either in serum levels or tissue levels (s-B7-H4 low 78.6% vs. s-B7-H4 high 49.0%, P = 0.015, Fig. 4b; t-B7-H4 low 73.1% vs. t-B7-H4 high 36.4%, P < 0.001, Fig. 4e; s-COX-2 low 73.8% vs. s-COX-2 high 40.8%, P < 0.001, Fig. 4c; t-COX-2 low 67.3% vs. t-COX-2 high 39.7%, P < 0.001, Fig. 4f). We also compared the survival curves according to TNM stage in these patients. The 5-year survival rates in the four T stages were significantly different (T1 77.3% vs. T2 59.4% vs. T3 47.1% vs. T4 33.3%, P < 0.001, Fig. 4g). The 5-year survival rates in the three N stages were significantly different (N0 68.8% vs. N1 48.7% vs. N2 32.3%, P < 0.001, Fig. 4h). The 5-year survival rates in the three TNM stages were significantly different (stage I 80.0% vs. stage II 60.7% vs. stage III 41.4%, P < 0.001, Fig. 4i).