Lower Bmi-1 Expression May Predict Longer Survival of Colon Cancer Patients

Colon cancer (CC) is the fourth common malignancy worldwide [1]. In China, both the incidence and mobility rate are increasing recently [2]. Bmi-1 is a member of the polycomb group which functions as a transcriptional repressor [3,4]. Bmi-1 gene is associated with tumor invasion and metastasis [5]. Several lines of evidence suggest that Bmi-1 blocks cell senescence and proliferation [6,7]. Aberrant expression of Bmi-1 has been detected in colorectal carcinoma, breast carcinoma, prostate cancer, liver carcinoma, medulloblastoma, lymphoma, acute myeloid leukemia, head and neck squamous cell carcinoma, non-small cell lung cancer and glioblastoma [5,6,8,9,10,11,12,13]. Bmi-1 has been identified as a predictor of the response to therapy and poor survival of patients with various tumors [14,15,16].

Bmi-1 has been detected in colorectal cancer tissue [17,18,19,20] but with smaller sample sizes. In this study, we investigated the expression of Bmi-1 in CC using a larger sample size of tumor tissue and analyzed the correlation between Bmi-1 expression and clinical significance and outcome. We aimed to confirm the potential role of Bmi-1 as a biomarker for the prognosis of CC.

This study was approved by the Ethics Committee of The Third Affiliated Hospital of Soochow University (also named The First People's Hospital of Changzhou). Human CC high density tissue microarray obtained from Shanghai Outdo Biotech Co. Ltd. contained 267 primary tumor specimens and the corresponding tissue with follow-up information included. The patients' baseline clinicopathological characteristics are listed in Table 1. Some of the patient's information was not available and thus were secluded from the statistical analysis.

IHC analysis of Bmi-1 was performed using a standard streptavidin-biotin-peroxidase complex method described in our previous article [21]. Two pathologists with no knowledge of the patients' information examined the stained sections independently. The Bmi-1 immunostaining densities were assessed according to the H-score method described [22]. H-score = (% of tumor cells unstained × 0) + (% of tumor cells stained weak × 1) + (% of tumor cells stained moderate × 2) + (% of tumor cells stained strong × 3). The H-scores ranged from 0 (100% negative tumor cells) to 300 (100% strong staining tumor cells). Results from the two pathologists were averaged and used for statistical analysis.

All data were analyzed with SPSS statistical software (SPSS Standard version 13.0, SPSS Inc.). The Bmi-1 expression between tumor and the corresponding tissues were compared with the Wilcoxon test. The association between Bmi-1 expression and the clinicalpathological characteristics was assessed with the chi-square test. For univariate survival analysis, all CC patients were analyzed by Kaplan-Meier analysis. The log-rank test was used for comparing different survival curves. A P value less than 0.05 was considered statistically significant. Data were expressed as mean ± standard error.

Table 1 summarizes the patients' available clinicalpathological characteristics. There were 127 male and 140 female cases. There were 17 cases with stage I, 132 with stage II, 107 with stage III and 9 with stage IV.

Immunoreactivity for Bmi-1 protein was observed as brown, granular staining on the cell nuclei of tumor cells (Fig. 1). The cut-off value of Bmi-1 expression was 135. The median H-scores of Bmi-1 in CC tissues and the corresponding tissues were 80.0 (0-270) and 5.0 (0-90), with no statistically significant difference (Z = -13.7, P < 0.001).

Univariate analysis revealed that there was no statistically significant difference between Bmi-1 expression level and any clinicopathological characteristics (including age, gender, pathological differentiation degree, tumor size, lymph node involvement status, distant metastasis status and TNM stage, all P > 0.05). See Table 1.

Based on the Kaplan-Meier survival analysis, cumulative survival curves were calculated, and differences in survival time were assessed with the log-rank test. The median OS times for CC patients with high or low Bmi-1 expression were 53.7 months and 44.9 months, respectively (P = 0.123). See Figure 2.

Table 2 summaries the 1-year to 5-year of CC patients. For patients with high Bim-1 expression (n = 74), the 1-year to 5-year survival rates were 75.7%, 67.6%, 59.5%, 50.3% and 46.3%, respectively. For patients with low Bim-1 expression (n = 193), the 1-year to 5-year survival rates were 81.3%, 72.5%, 66.8%, 61.1% and 56.5%, respectively. We found that the survival rates of patients with low Bmi-1 expression were higher that of patients with high Bmi-1 expression but none of the differences of various years were statistically significant (P > 0.05). See Table 2.

In the present study, we found that the OS of patients with low Bmi-1 expression tended to be longer, however not significant

Other investigators also find that low Bmi-1 expression had a better survival outcome [20,23]. And Bmi-1 immunoreactivity emerged as an independent prognostic factor [19]. In pancreatic cancer [4], patients with low Bmi-1 expression had a better survival outcome. We have meta-analyzed the association between Bmi-1 expression and the clinicopathological characteristics and prognosis of patients with various malignancies [24]. The results suggest that high Bmi-1 expression predict poorer clinical outcome.

However, Espersen et al. observed Bmi-1 expression in primary tumors of stage II colon cancer patients could not predict relapse or OS of the patients, thus having a limited prognostic value in stage II colon cancer patients [25]. Furthermore, some other studies suggest opposite conclusions to ours. In breast cancer, patients with high levels of Bmi-1 have a more favorable outcome than patients with low expression of Bmi-1 [26]. And we recently found that high levels of Bmi-1 in gastric cancer patients are significantly associated with better OS (P = 0.02) [21].

The differences among the results of various studies might be explained by the complex function of polycomb protein Bmi-1. Bmi-1 functions as a double edged sword. It has been reported that leukemic stem cells lacking Bmi-1 do not proliferate [27,28]. Bmi-1 protein might be involved in human colorectal carcinogenesis [17]. Bmi-1 is required for colon cancer proliferation in vitro and in vivo [29]. Downregulation of Bmi-1 inhibits cell proliferation and cycle progression and promotes apoptosis of tumor cells of colorectal cancer [30].

Chemoresistance can be facilitated via the upregulation of Bmi-1 [31]. Reducing Bmi-1 expression can inhibit the proliferation of tumor cells and inverse chemoresistance [32]. Meanwhile, tumor cells with high Bmi-1 expression proliferate faster and hence are more sensitive to chemotherapy, leading to longer survival.

Furthermore, our sample size is larger than other studies, although the size alone does not explain the difference of results. Majority of our patient population has stage II and III disease, the omission of stage I and IV patients may alter the results. Besides, some information is unavailable so it is not certain how treatment modalities influence survival. Treatment effect on outcome and clinical follow up for recurrences will further strengthen the findings. This may also be a reason why there is no significant difference in the OS times between the two groups of patients.

Conclusively, while there may be a trend towards improved survival, this is not statistically significant.

This work was supported by grants from National Natural Science Foundation of China (No. 31570877 and 31570908), Jiangsu Engineering Research Center for Tumor Immunotherapy (BM2014404), and Society development plans, Department of Science and Technology Changzhou (CJ20159018).

The authors declare that they have no conflict of interest.

X. Li and X. Zheng contributed equally to this work.