Fresh specimens were collected from 32 patients with gastric adenocarcinoma who underwent radical resection at National Cheng Kung University Hospital (Tainan, Taiwan) between August 2003 and August 2008. The mean age was 60±11 years old (range, 35–82 years old; 20 males, 12 females). A total of 32 pairs of cancerous and matched adjacent normal gastric mucosa tissues were collected and analyzed as previously described (30). The Tumor, Node, Metastasis system by the American Joint Committee on Cancer was used for the classification, grading and staging of gastric cancer (31). The specimens were preserved in the Human Biobank within the Research Center of Clinical Medicine of the National Cheng Kung University Hospital (Tainan, Taiwan). All the patients provided written informed consent, and the study was approved by the Institutional Review Board of National Cheng Kung University Hospital (approval no., ER-97-148).

Total cell lysates were prepared and analyzed by 10% SDS-PAGE as previously described (30,32,33). Membranes were blocked with 5% (w/v) skimmed milk (Merck KGaA, Darmstadt, Germany) for 1 h at room temperature and incubated with the following primary antibodies overnight at 4°C: Anti-CCND1 (cat. no., 2922; dilution, 1:2,000; Cell Signaling Technology, Inc., Danvers, MA, USA) and anti-β-actin (cat. no., GTX26276; dilution, 1:5,000; GeneTex, Inc., Irvine, CA, USA). Membranes were then incubated for 1 h at room temperature with peroxidase-conjugated goat anti-rabbit IgG (cat. no. 7074S; 1:3,000; Cell Signaling Technology, Inc.) or peroxidase-conjugated sheep anti-mouse IgG antibody (ECL anti-mouse IgG; cat. no. NA931V; 1:3,000) (Amersham Pharmacia Biosciences, Buckinghamshire, U.K.). Immunodetection was performed using the horseradish peroxidase-based SuperSignal Chemiluminescent Substrate (Pierce; Thermo Fisher Scientific, Inc., Waltham, MA, USA). For quantification, the bands were measured with the AlphaImager 2200 Imaging System (Alpha Innotech; Bio-Techne, Minneapolis, MN, USA), and the densities of the CCND1 bands were normalized to those of the β-actin bands. CCND1 expression was quantified and described as a ratio to β-actin expression (CCND1/β-actin ratio).

A search of the Oncomine database (http://www.oncomine.com) (34) was initially conducted to systematically assess the expression level of the CCND1, CCND2 and CCND3 genes in gastric cancer. For this, normal vs. cancer tissues were compared in the differential analysis. The results were analyzed for their P-values, fold change and cancer subtype. The prognostic value of the CCND1, CCND2 and CCND3 genes in gastric cancer was also analyzed using the Kaplan-Meier Plotter (http://kmplot.com/analysis/), as described previously (35). The following settings were used for the analysis: ‘Overall survival’; ‘progression-free survival’; ‘autoselect best cutoff’; ‘censore at threshold all’ (patients surviving over the selected threshold are censored instead of excluded); ‘tumor stage all’; ‘tumor stage T all’; ‘tumor stage N all’; ‘tumor stage M all’; ‘grade all’; ‘Lauren classification all’ (2); ‘differentiation all’; and ‘moderate and poor differentiation’. Tumors were classified according to WHO histopathological type (36). Three cyclin D genes probe sets were available: 208712_at at CCND1, 200953_s_at at CCND2 and 201700_at at CCND3, and patients were split according to median expression or to expression at best cut-off for each probe. A total of 1,065 patients with gastric cancer were assessed using a Kaplan-Meier plot (36) and HER2 status was identified using the gene chip probe set 216836_s_at, as previously described (37). The hazard ratio (HR) 95% confidence intervals and logrank P-values were calculated and described. P<0.05 was considered to indicate a statistically significant difference. The data were extracted from the Oncomine database and Kaplan-Meier Plotter between March 2015 and August 2015. Finally, the association between CCND1 protein expression, assessed according to previously published protocols (30,38) (CCND1/β-actin ratio), and differentiation type (moderate and poor differentiation) in fresh specimens derived from patients with gastric adenocarcinoma was assessed using the Student's t-test. The statistical differences between two groups were assessed, and P<0.05 was considered to indicate a statistically significant difference. Statistical analysis was performed using GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA).

Data for CCND1, CCND2 and CCND3 transcript expression were extracted from the Oncomine database for gastric cancer, focusing on cancer vs. normal patient datasets. The statistical significance, fold change, patient number, and type of tissues that displayed upregulation or downregulation of CCND1, CCND2 and CCND3 gene expression were also analyzed in normal vs. cancer tissues from the Oncomine database. The Derrico and Cho datasets obtained from Oncomine is embedded in the NCBI GEO database (https://www.ncbi.nlm.nih.gov/geo/) at accession numbers GSE13911 and GSE13861, respectively (Fig. 1). Overexpression and downregulation of the CCND1, CCND2 and CCND3 genes were identified in gastric cancer (Fig. 1). To determine the clinical relevance of CCND1, CCND2 and CCND3 expression in human gastric cancer, their expression profiles in the oncomine cancer microarray database were analyzed. Information on the expression of CCND1, CCND2 and CCND3 in normal and cancerous gastric tissues was compiled from all of the microarray studies in the database (39,40). The histological type of gastric adenocarcinoma was divided into intestinal, diffuse and mixed types (2). As demonstrated in Fig. 1A, CCND1 expression was significantly increased in gastric intestinal-type adenocarcinoma of gastric cancer (40). CCND2 expression was significantly increased in several types of gastric cancer, including diffuse gastric adenocarcinoma, gastric intestinal-type adenocarcinoma and gastric mixed adenocarcinoma (Fig. 1B-E) (39,40). By contrast, CCND3 expression was significantly decreased in gastric mixed adenocarcinoma (Fig. 1F) (39). Oncomine analysis of neoplastic vs. normal tissue revealed that CCND1 and CCND2 were overexpressed in gastric cancer from the GSE13911 and GSE13861 datasets, respectively.

To analyze the association of CCND1, CCND2 and CCND3 expression with gastric cancer patient survival, Kaplan-Meier survival curves were constructed (Fig. 2). A significant association was identified between CCND1, CCND2 and CCND3 mRNA and survival (P<0.05, log-rank test). Overexpression of CCND1 (Fig. 2A and B) was correlated with lower OS and PFS. By contrast, CCND2 overexpression was correlated with increased survival (Fig. 2C and D). Overexpression of CCND3 was correlated with lower OS and PFS (Fig. 2E and F).

When the analysis was restricted by differentiation type, significant differences in OS and PFS were observed for the expression of the CCND1, CCND2 and CCND3 genes in patients with moderately and poorly differentiated tumors. Specifically, high CCND1 expression was not associated with PFS compared with that of patients with low CCND1 expression in patients with moderately differentiated gastric cancer (Fig. 3A), but in low CCND1 expression was significantly (P=0.0079) longer associated with PFS compared with that of patients with high CCND1 expression in patients with poorly differentiated tumors (Fig. 3B). By contrast, high CCND2 expression was associated with a significantly poorer (P=0.032) PFS in patients with moderately differentiated tumors (Fig. 3C), but not in those with poorly differentiated gastric cancer (Fig. 3D). Additionally, low CCND3 expression was associated with significantly (P=0.039) longer PFS in patients with moderately differentiated tumors (Fig. 3E), but not in patients with poorly differentiated gastric cancer (Fig. 3F). The results demonstrated that the high expression of CCND1 was significantly correlated with poor differentiation and poor survival, while high expression of CCND2 and CCND3 was significantly correlated with moderate differentiation and poor survival.

HER2 overexpression has been correlated with poor outcomes and a more aggressive disease (41); however, the association between HER2 status and the prognosis of patients with gastric cancer remains controversial (42). To analyze the association of CCND1, CCND2 and CCND3 expression and HER2 status with survival, Kaplan-Meier PFS curves of PFS stratified by CCND1, CCND2 and CCND3 mRNA expression in HER2-negative and HER2-positive tumors were constructed. Overexpression of CCND1 (Fig. 4A and B) was associated with reduced PFS in HER2-negative and HER2-positive tumors. By contrast, reduced CCND2 expression was associated with lower PFS in patients with either HER2-negative or HER2-positive tumors (Fig. 4C and D). Overexpression of CCND3 (Fig. 4E and F) was associated with reduced PFS in HER2-negative and HER2-positive tumors. These results indicated that CCND1, CCND2 and CCND3 expression, and HER2 status were associated with survival.

When the analysis was limited to those patients with gastric cancer with poorly differentiated tumors and was restricted by HER2 status, significant differences in PFS were observed between high and low expression levels of CCND1, CCND2 and CCND3 (Fig. 5). Low CCND1 expression was associated with significantly longer PFS in patients with poorly differentiated, HER2-negative tumors (Fig. 5A), but not in those with poorly differentiated, HER2-positive tumors (Fig. 5B). By contrast, low CCND2 expression was associated with significantly poorer PFS in patients with poorly differentiated, HER2-negative tumors (Fig. 5C), but not in those with poorly differentiated, HER2-positive tumors (Fig. 5D). PFS was not affected by CCND3 gene expression level in poorly differentiated, HER2-negative or HER2-positive tumors (Fig. 5E and F, respectively). However, moderately differentiated tumors were not analyzed in the present study due to the small number of patients. Overexpression of CCND1 was significantly correlated with HER2-negative tumors, poor differentiation and poor survival. Additionally, downregulation of CCND2 was significantly correlated with poor differentiation, HER2-negative tumor status and poor survival. Taken together, these results suggest that overexpression of CCND1 is predictive of a poor prognosis and serves an important role in poorly differentiated, HER2-negative gastric tumors.

CCND1 expression was also compared in moderately and poorly differentiated gastric cancer, and CCND1 protein expression was examined by western blot analysis in tumor and adjacent normal gastric tissues of 32 patients. All 32 cases of gastric cancer were adenocarcinomas, including 13 moderately differentiated and 19 poorly differentiated tumors (Table I). Analysis of the relative expression of CCND1 in moderately (Fig. 6A) and poorly (Fig. 6B) differentiated tissues indicated that the CCND1/β-actin ratio in poorly differentiated samples was significantly greater compared with that in moderately differentiated samples (P<0.0018) (Fig. 6C). In summary, the overexpression of CCND1 is associated with poorly differentiated gastric cancer.