A total of 157 subjects from Yonsei University Health System were enrolled in this study. All subjects underwent upper gastrointestinal endoscopy (Types XQ-260, Olympus, Tokyo, Japan) with biopsy. The final diagnosis was made based on histological findings from biopsy or surgical specimens. All patients were diagnosed for the first time during the enrollment period, and blood samples were collected before they received any treatments. Blood samples were stored as serum fractions at -80 °C until analysis. The Institutional Review Board of Yonsei University Health System approved the current study, and written informed consent was obtained from all participants in accordance with the Declaration of Helsinki.

Subjects who suffered from chronic diseases such as liver cirrhosis, chronic renal disease, and diabetes mellitus were excluded from this study. Subjects with other cancers and other gastric neoplasms such as gastrointestinal stromal tumors, mucosa-associated lymphoid tissue lymphomas, and neuroendocrine tumors were also excluded. Patients who previously received any treatment for GC or its premalignant lesions were also excluded.

Subjects were classified into the following four groups according to the “gastritis-dysplasia-carcinoma” sequence of gastric carcinogenesis[21]: control group, which included normal mucosa or acute and chronic gastritis; high-risk group, which included intestinal metaplasia (IM) and dysplasia; EGC group; and AGC group. Both age and sex were matched in all groups. All patients in the cancer groups underwent imagining studies including chest X-ray, abdominal-pelvic helical computed tomography, and whole-body positron emission tomography to determine TNM stage. TNM stage for GC was evaluated according to the 7^th International Union Against Cancer-TNM stage guidelines for GC[22] based on radiological studies or surgical findings. Helicobacter pylori (H. pylori) infection was determined by staining of gastric tissue with Giemsa solution (Sigma, MO, United States). Glandular atrophy and IM were diagnosed according to the updated Sydney classification[23], and pathological determination of differentiation status (well, moderate, poor, and signet-ring cell) was performed according to the Lauren classification.

Serum CEA levels were measured by the Beckman Access CEA assay (Beckman Coulter Inc., MN, United States). Serum sHB-EGF levels were measured by a commercially available human HB-EGF ELISA Kit (DY259, RD, MN, United States) according to the manufacturer’s instructions. Briefly, 96-well microplates were coated with capture antibody (80 μg/well, goat anti-human HB-EGF) at 4 °C for 16 h. After washing, the plates were blocked with Reagent Diluent (provided in kit) and then incubated for 1 h at room temperature (RT). After washing, 100 μL of diluted sample, standard, and control were added to each well. The microplates were then incubated for 2 h at RT. Subsequently, microplates were washed and then detection antibody was added (10 ng/well, biotinylated goat anti-human HB-EGF). The plates were then incubated for 2 h at RT. After washing, streptavidin-HRP was added and incubated for 20 min at RT in the dark place. Plates were then washed again, and 100 μL of chromogen (H₂O₂ and tetra-methylbenzidine) was added to each well. The enzyme reaction proceeded for 20 min at RT in the dark place. The chromogenic substrate reaction was stopped by the addition of stop solution (2 mol/L H₂SO₄) and the absorbance was read at 450/570 nm. The final values were calculated based on a calibration curve prepared from standards. The ELISA for sHB-EGF levels was tested in triplicate.

To calculate the appropriate sample size for each group, Russ Lenth’s interactive power/sample size online calculator was used. Under assuming that there were 4 comparison groups, the estimated standard deviation (SD) was 1, and the confidence level was 0.05, sample size of ≥ 30 in each group achieved a statistical power > 80% using one-way analysis of variance (ANOVA).

For statistical analysis for current data, SPSS version 20.0 (IBM Corp., NY, United States) was used. P values < 0.05 were considered statistically significant. Values (sHB-EGF, CEA) were expressed as the mean with the 25%-75% SD. Means of each group was compared by ANOVA test with multiple comparisons by using the post-hoc Bonferroni method. An independent sample t-test was used to compare the mean between the cancer groups vs non-cancer groups. Spearman’s correlation (coefficient, γ_s) was used to assess the relationship between continuous variables and non-continuous variables, and Pearson’s correlation (coefficient, γ_p) was used to assess the relationship between continuous variables. Nominal data were compared by χ² test. The receiver operating characteristic (ROC) curves was conducted and area under the curve (AUC) was calculated to compare the diagnostic accuracy between serum sHB-EGF and serum CEA. Logistic regression analysis was performed to obtain the best sensitivity/specificity to predict the presence of GC as a single-marker or as a part of multiple-markers panel. Each marker was included as a linear term.

The 157 subjects are composed of 60 individuals/patients with normal mucosa or gastritis (control group), 30 patients with IM/dysplasia (high-risk group), 37 patients with EGC (EGC group), and 30 patients with AGC (AGC group). The control group was further subdivided into two subgroups; patients with normal mucosa/chronic superficial gastritis (CSG, n = 30) and patients with chronic atrophic gastritis (CAG, n = 30) because the risk of GC development was different between CSG and CAG. The normal mucosa/CSG group was also further subdivided into normal mucosa (n = 15) or CSG (n = 15) because gastric inflammation status may affect sHB-EGF levels comparing to normal mucosa. The clinical and histopathological features of subjects in each group are described in Table 1. There were no significant differences in distribution of age and sex, and the status of H. pylori infection among the disease groups (χ², all P > 0.05). In the cancer groups, the location of primary tumor did not differ (P > 0.05), while histological differentiation, primary tumor size, and TNM stage were significantly different between the EGC and AGC groups (all P < 0.05).

Serum sHB-EGF levels increased along the GC carcinogenic sequence, and the differences among the groups were statistically significant (ANOVA, P < 0.001; Table 2). Serum sHB-EGF levels were significantly higher in the AGC group (314.4 ± 127.5 pg/mL) compared with those of EGC (165.3 ± 123.2 pg/mL), high-risk (98.7 ± 67.3 pg/mL), and control (94.7 ± 83.6 pg/mL) groups (post-hoc Bonferroni, all P < 0.001), respectively. Serum sHB-EGF levels were also significantly higher in the EGC groups than the high-risk (P = 0.049) and control (P = 0.006) groups, respectively. However, there was not a significant difference between the high-risk and control groups (P > 0.05). Serum sHB-EGF levels were also not significantly different between the CAG and CSG groups (P > 0.05), or between the CSG and normal mucosa groups (P > 0.05), respectively (Table 3). On the other hand, serum CEA levels were not significantly different among the control, high-risk, and EGC groups. The AGC group was the only population with significantly elevated serum CEA levels (P = 0.001), especially in the case with distant metastasis. When serum sHB-EGF levels were compared between cancer and non-cancer groups, they were significantly higher in the cancer groups (232.1 ± 144.9 pg/mL) than in the non-cancer groups (96.0 ± 78.2 pg/mL; t-test, P < 0.001; Table 4).

Table 5 shows that serum sHB-EGF levels were not affected by sex (γ_s = 0.138, P = 0.076) or the status of H. pylori infection (γ_s = -0.054, P = 0.486), whereas these levels were negatively correlated with age (γ_p = -0.265, P = 0.001). However, serum sHB-EGF levels were not closely correlated with age when analysis was performed in just non-cancer groups (γ_p = 0.108, P = 0.313; Table 6). In contrast, when analysis was performed within the cancer groups, the serum sHB-EGF levels were negatively correlated with age (γ_p = -0.314, P = 0.010). This result implies that serum sHB-EGF levels are affected by the age of patients with GC; relatively younger patients with GC had more highly elevated sHB-EGF levels compared with relatively older patients with GC. On the other hand, serum CEA levels were not affected by sex, age and the status of H. pylori infection (all P > 0.05; Table 5).

Histopathologically, there were no significant relationships between serum sHB-EGF levels and the histological differentiation of GC (Lauren classification system), although sHB-EGF levels tend to be higher in diffuse-type than in intestinal-type (γ_s = 0.214, P = 0.078; Table 5). Serum sHB-EGF levels were not also affected by primary tumor location (γ_s = −0.054, P = 0.652; Table 5).

On the other hand, serum sHB-EGF levels were closely correlated with depth of invasion (T-stage, γ_s = 0.669, P < 0.001), lymph node metastasis (N-stage, γ_s = 0.407, P= 0.001), distant metastasis (M-stage, γ_s = 0.261, P = 0.030), and overall stage (γ_s = 0.570, P < 0.001) respectively (Table 5). To analyze the relationship between serum sHB-EGF levels and primary GC size, patients were divided into 3 groups based on the tumor size: < 3 cm, 3-5 cm, and > 5 cm. Table 5 shows that primary GC size was positively correlated with serum sHB-EGF levels (γ_s = 0.237, P = 0.048). On the other hand, serum CEA levels were only correlated with tumor size (γ_s = 0.382, P = 0.006) and distant metastasis (γ_s = 0.362, P = 0.002). Collectively, histopathological results suggest that serum sHB-EGF levels were closely correlated with advanced stage and poor prognosis of GC.

ROC curve was generated and AUCs were calculated to compare the diagnostic accuracy of serum sHB-EGF with serum CEA for prediction of GC (Figure 1). The AUC of serum sHB-EGF was 0.85 (95%CI: 0.79-0.91), and those of serum CEA was 0.64 (95%CI: 0.55-0.73). This analysis indicates that serum sHB-EGF has a higher diagnostic accuracy to predict the presence of GC compared with CEA.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Receiver operating characteristic curve for serum soluble heparin-binding epidermal growth factor-like growth factor compared with the curve for serum carcinoembryonic antigen to predict gastric cancer. ROC: Receiver operating curve; sHB-EGF: Soluble heparin-binding epidermal growth factor-like growth factor; CEA: Carcinoembryonic antigen; AUC: Area under the ROC curve.

Logistic regression analysis further confirmed the remarkable diagnostic accuracy of serum sHB-EGF for GC; the sensitivity and specificity of serum sHB-EGF levels for diagnosis of GC were 76.1% and 76.5% (cut-off point, 0.38; Table 7). These values are superior to those of serum CEA (sensitivity, 62.1%; specificity, 51.8%; cut-off point, 0.38). When serum sHB-EGF was combined with serum CEA, the sensitivity was slightly increased; the sensitivity was 77.3% and specificity was 76.5% (cut-off point, 0.38), respectively. When serum sHB-EGF was combined with serum CEA, the sensitivity was slightly increased; the sensitivity and specificity were 77.3% and 76.5% (cut-off point, 0.38). Collectively, serum sHB-EGF exhibited a remarkable diagnostic accuracy to predict GC both as a single biomarker and as a part of multiple-markers panel in GC (Table 7).

Early detection of gastric cancer (GC) is the most important clinical issue. Although endoscopic examination is an ideal, highly reliable technique for early detection of GC, it has some limitations as a routine screening method because of the risk associated with invasive procedure. Therefore, identification of inexpensive, reliable and less-invasive serum biomarkers is a great clinical challenge. However, research is still underway to identify effective serum biomarkers for GC.

Heparin-binding epidermal growth factor-like growth factor (HB-EGF) has been thought to be associated with GC development and progression, and demonstrated to be overexpressed in human GC tissues. Because HB-EGF can be released into circulation as a mature soluble form of HB-EGF (sHB-EGF), it can be measured in serum and can be used as a serum biomarker for GC. The authors determined how serum levels of sHB-EGF related to the ‘gastritis-dysplasia-carcinoma’ sequence of gastric carcinogenesis and validated its usefulness as a biomarker for GC compared with serum CEA, a classic biomarker for gastrointestinal tumors.

Recent reports showed that increased epidermal growth factor receptor levels are associated with poor prognosis in patients with GC and HB-EGF expression is linked to tumor formation although HB-EGF is a critical molecular component of many normal physiological processes. Thus, HB-EGF may become a promising biomarker or treatment target for GC. Several studies have shown that HB-EGF is overexpressed in GC tissues, and overexpressed HB-EGF is correlated with far-advanced stage of GC. However, there is little quantitative data demonstrating the clinical significance of serum sHB-EGF in relation to GC tumorigenesis and progression such as TNM stage.

This is the first study to evaluate serum sHB-EGF levels quantitatively according to the gastric carcinogenic sequence, to analyze the correlations between serum sHB-EGF and clinicopathological features of GC, such as TNM stage, and to validate serum sHB-EGF as a desirable serological biomarker for GC.

The study results suggest that sHB-EGF are closely correlated with advanced TNM stage and higher in early gastric cancer (EGC) group than high-risk group, and higher in advanced gastric cancer group than EGC group. Additionally, this study demonstrated a remarkable diagnostic accuracy of serum sHB-EGF for GC.

This is an interesting manuscript with innovative endpoint and potential clinical impact. In this manuscript, authors studied the clinical usefulness of soluble heparin-binding EGF-like growth factor as a biomarker for GC. HB-EGF is overexpressed in several cancer cell lines and cancer tissues including human GC tissues. HB-EGF is also reported to be involved in malignant phenotype and chemo-resistance of cancer cells. Therefore this study would be useful to develop a useful biomarker for GC diagnosis.