[Chinese Journal of Cancer 27:12, 568-573; December Expression 2008]; 2008 and significance Sun Yat-sen of University Bmi-1 and Cancer Ki67 in Center colorectal carcinoma tissues Clinical Research Paper Expression and significance of Bmi-1 and Ki67 in colorectal carcinoma tissues Miao-Xia Lin, 1 Zhuo-Fu Wen, 1, * Zhi-Ying Feng 2 and Dan He 2 1 Department of Gastroenterology; and 2 Department of Pathology; The Third Affiliated Hospital of Sun Yat-sen University; Guangzhou, Guangdong P.R. China Key words: Bmi-1, Ki67, colorectal carcinoma, immunohistochemistry *Correspondence to: Zhuo-Fu Wen; Department of Gastroenterology; and Department of Pathology; The Third Affiliated Hospital of Sun Yat-sen University; Guangzhou, Guangdong P.R. China; Tel.: 86.20.85252360; Email: WENZFU@163.com Submitted: 04/08/08; Revised: 06/16/08; Accepted: 06/20/08 This paper was translated into English from its original publication in Chinese. Translated by: Beijing Xinglin Meditrans Center (http://www.59medtrans.com) and Hua He on 10/20/08. The original Chinese version of this paper is published in: Ai Zheng (Chinese Journal of Cancer), 27(12); http://www.cjcsysu.cn/cn/article.asp?id=14824 Previously published online as a Chinese Journal of Cancer E-publication: http://www.landesbioscience.com/journals/cjc/article/8027 Background and Objective: Bmi-1, a putative oncogene, is a member of the polycomb group genes, and is expressed in many human tumors. Ki67 is an important nuclear antigen associated with cell proliferation. This study investigates the expression, significance and correlation of Bmi-1 and Ki67 in colorectal carcinoma (CRC) tissues. Methods: Protein expressions of Bmi-1 and Ki67 in 60 CRC, 30 adenomas and 20 normal colorectal mucosal tissues were detected using immunohistochemistry. Correlations of the expression of Bmi-1 and Ki67 to clinicopathological features and patient survival were analyzed. Results: The expression rates of Bmi-1 and Ki67 were 25.0%, 6.7%, 0% and 18.3%, 3.3% and 0% in CRC, adenomas and normal colorectal mucosal tissues, respectively. The expression of Bim-1 and Ki67 were significantly higher in CRC than in adenomas and normal colorectal mucosal tissues (p < 0.05). A chi-square test revealed that the overexpression of Bmi-1 protein in CRC was correlated to distant metastasis (p < 0.01) and TNM stage (p < 0.05), while the overexpression of Ki67 protein was associated with age, distant metastasis and TNM stage (p < 0.05). Further logistic regression analysis showed that only distant metastasis was correlated with Bmi-1 overexpression (p < 0.01, OR > 1), and only age was correlated with Ki67 overexpression (p < 0.05, OR < 1). Kaplan-Meier survival analysis showed that the survival was significantly shorter in CRC patients with overexpression than low expression of Bmi-1 and Ki67 (p < 0.05). Conclusions: Overexpression of Bmi-1 and Ki67 protein are significantly correlated with tumorigenesis, metastasis and prognosis of CRC. Bmi-1 and Ki67 might be used to predict the prognosis of CRC. Bmi-1 might regulate the expression of Ki67 indirectly in CRC patients. Colorectal carcinoma (CRC) is a common digestive tract cancer, with the incident rate on the rise. In economically developed countries such as Western Europe, the incidence of CRC ranks second, while in China ranks third in incidence and fourth in mortality. 1,2 Bmi-1 gene is a member of Polycomb group genes (PcG), which is first identified in passaging lymphoma cells from transgenic mice in 1991. It functions as a proto-oncogene and causes cell transformation and tumor formation with the assistance of another oncogene, c-myc. 3 Currently, the association between Bmi-1 and malignant tumor attracts more attention and cells with high expression of Bmi-1 are considered to be the cancer stem cells of tumor cells. Various literature reports that the origination and development of a variety of tumors are related to the enhanced expression of Bmi-1 gene, which also correlates with tumor prognosis. There are few overseas reports and no domestic reports on the correlation between the Bmi-1 gene and colorectal cancer. Ki67 is a non-histone protein located at nucleus, with a short half-life and it is known to be associated with cell proliferation. 4 Many studies confirm that, the expression of Ki67 closely correlates with cell proliferation and its expression level reflects the degree of cell proliferation. The objective of this study is to detect the protein expression level of Bmi-1 and Ki67 in colorectal cancer tissues by immunohistochemical staining method as well as its association with clinicopathological features and the survival rate of patients, and to investigate the significance of those two genes in colorectal cancer. Data and Methods Clinical data. Sixty paraffin specimens from colorectal cancer patients preserved at the Pathology Department in the Third Affiliated Hospital of Sun Yat-sen University from January 2000 to August 2004 were collected and all samples were reviewed again for affirmative pathological diagnosis. The sixty colorectal cancer patients were all hospitalized and treated at the Gastrointestinal Surgery Department in our hospital, with complete clinical information and at least three years of follow-up data, no pre-operative chemotherapy and radiotherapy. Among all cases, there were 40 men 568 Chinese Journal of Cancer 2008; Vol. 27 Issue 12
and 20 women, aged 19 to 84 years old, with a median age of 64 years. Tumor clinical staging referred to the TNM classification of cancer stage criteria released by International Union Against Cancer (UICC) in 2002. Case-by-case follow-up was done for the above colorectal cancer patients mainly by phone interview. All patients had follow-up data, with the shortest follow-up period at 36 months and the longest at 81 months. At the end of follow-up, 35 cases survived (six cases with post-operative metastasis) and 25 cases died; at year three, 44 cases had survived and 16 patients died. Reagents and methods. Tissue paraffin blocks were sliced continuously at 4 μm thickness, dewaxed with xylene, dehydrated by gradient ethanol treatments with 3% hydrogen peroxide solution used to block endogenous peroxidase, then washed by PBS buffer (ph = 7.2~7.4), with EDTA (ph = 8.0) antigen repair solution applied to recover antigens under high-pressure; the tissue was then cooled, with mouse anti-human Bmi-1 monoclonal antibody (1:150 dilution, Upstate, USA) added and mouse antihuman Ki67 monoclonal antibody working solution (Shanghai CHANGDAO company) added, respectively, then incubated at 37 C in a water bath for 1 h, washed by PBS buffer, EnVision TM secondary antibody added, then incubated at 37 C in a water bath for another 30 min, washed again by PBS buffer and developed by freshly made DAB solution. Developing time was controlled under microscope and stopped by tap water. Cell nucleus was slightly double-stained by Mayer dye, and then slices were dehydrated by gradient ethanol, sealed by Balsam neutral and observed under a microscope. Previously positive-stained slices were used as positive controls and PBS-replacing secondary antibody blocked slices were included as negative controls. Results assessment criteria. Bmi-1 positive protein expression mainly showed as tan or brown particles presenting in the nucleus, while a small portion had brown staining inside cytosol. Under high magnification, four different fields were randomly selected for counting the total cell number and nuclear-positive cell number. Slices were scored according to the percentage of positive cells: positive cells rating 10% as 1, positive cells rating >10% but 50% as 2, positive cells rating >50% but 75% as 3, while positive cells rating >75% as 4; Meanwhile, slices were scored according to staining intensity: negative staining as 1, weak staining as 2, moderate as 3 and strong staining as 4. Final results were determined by the product of the above two scores: 4 as (-), >4 but 8 as (+), >8 but 12 as (++) and >12 as (+++). Only (++) and (+++) groups were considered high expression group, while (-) and (+) groups were classified as low expression group. Positive Ki67 expression in nucleus was identified by presence of tan or brown particles in nucleus. Contiguous 10 high magnification fields were observed and 100 tumor cells inside each field were counted, for a total of 1,000 cells. Samples were scored by percentage of nuclear positive cells and scoring method for positive cells was: <10% as (+); 10% 50% as (++); >50% as (+++). Only (+++) was classified as high expression group, while (+) and (++) were classified as low expression groups. Results were confirmed by independent examinations by two pathologists. Statistical analysis. Differences between quantitative data was detected by chi-square test, multi-factor interactions were detected using logistic regression, Kaplan-Meier survival curves used the log-rank test and correlation between two parameters used the Spearman correlation analysis. Our group set p < 0.05 as the criteria for statistical significance and all statistical analysis were processed by the SPSS13.0 software package. Results Expression of Bmi-1 and Ki67 proteins in colorectal carcinoma. Immunohistochemical analysis showed that: Bmi-1 positive protein expression mainly showed as tan or brown particles presented in nucleus, while a small portion had brown staining inside cytosol (see Fig. 1). In this study, the positive expression rate of 60 cases colorectal cancers was 51.7% (31/60), 25% of which had a high expression rate (15/60). Among 30 cases of colorectal adenomas, 6.67% (2/30) had high expression while 0% of the 20 normal controls had high expression in their colorectal mucosa. Therefore, the expression of Bmi-1 protein in colorectal cancer was significantly higher than the adenomas group and the normal control group (p < 0.05), while its expression in the adenoma group was not significantly different from that of the normal group (p > 0.05). Ki67 expression in nucleus and its positive expression were marked by the presence of tan or brown particles inside the nucleus (see Fig. 1). In this study, 11 cases colorectal cancers out of 60 cases had Ki67 high expression, so the high expression rate was 18.3% (11/60); among 30 cases of colorectal adenomas, there was one case of high expression (severe atypical hyperplasia), so the high expression rate was 3.33% (1/30); All 20 cases of normal colorectal mucosa had low expression. So, the expression of Ki67 protein in colorectal cancer was significantly higher than that of the adenoma group and normal control group (p < 0.05), while the expression between the adenoma group and the normal group had no significant difference (p > 0.05). The correlation between Bmi-1, Ki67 protein expression and the clinicopathological features of colorectal cancer. Bmi-1 protein expression correlated with the occurrence of colorectal cancer metastasis and TNM staging i.e., the Bmi-1 protein expression in patients with distant metastasis was significantly higher than those without (p < 0.01). Under TNM staging, the Bmi-1 protein expression in stage III/IV patients was significantly higher than that of stage I/II patients (p < 0.05), but did not correlate with the gender of patients, their age, tumor size, tumor location, degree of differentiation, histological classification or occurrence of lymphatic metastasis (p > 0.05) (see Table 1). Furthermore, logistic regression analysis showed that only distant metastasis correlated with Bmi-1 protein expression i.e., distant metastasis is the highest risk factor for Bmi-1 protein high expression (p < 0.01, B = 1.705, OR = 5.50) (see Table 2). Ki67 protein high expression correlated with the age of the patient, distant metastasis and TNM staging (p < 0.05) i.e., the Ki67 protein expression in younger patients, patients with distant metastasis and a TNM staging of III/IV was significantly higher than that of older patients, those with no distant metastasis and those with TNM staging of I/II, and it was not related to the gender of patients, tumor size, location, degree of differentiation, histological classification or occurrence of lymph node metastasis (p > 0.05). Further logistic regression analysis showed that only www.landesbioscience.com Chinese Journal of Cancer 569
Figure 1. Expression of Bmi-l and Ki67 proteins in colorectal neoplasm tissues (SPX400). (A) Negative expression of Bmi-l in normal colorectal tissues; (B) Weak expression (yellow) of Bmi-l in the nuclei of colorectal adenoma cells; (C) Intensive expression (brown) of Bmi-l in the nuclei of colorectal cancer cells; (D) Positive nuclear staining (brown-yellow) of Ki67 in normal colorectal tissues; (E) Positive nuclear staining (brown-yellow) of Ki67 in CRC; (F) Strong nuclear staining (brown) of Ki67 in CRC. patients age correlated with high expression of Ki67 protein, i.e., the younger the age was, the higher the Ki67 expression was (p < 0.05, B = -1.921, OR = 0.146) (Table 3). Correlation between Bmi-1, Ki67 protein expression and the prognosis of colorectal cancer. We studied the correlation between Bmi-1 protein expression in colorectal cancer and patient prognoses by analyzing the association between Bmi-1 protein expression in 60 cases of colorectal cancer patients and their survival rate using the Kaplan-Meier survival curve. Furthermore, a logrank test was applied to detect any statistical significance in the survival rate between these two groups. The results showed that the survival rate of patients with Bmi-1 protein high expression was significantly lower than patients with low expression (χ 2 = 5.929, p = 0.015), while the survival rate of patients with Ki67 protein high expression was significantly lower that patients with low expression (χ 2 = 14.374, p = 0.000) (Fig. 2). The significance of Bmi-1 and Ki67 protein expression in colorectal cancer. Spearman correlation analysis showed that Bmi-1 and Ki67 protein expression were positive correlated to a certain degree, but the correlation was relatively weak and not statistically significant (r = 0.224, p = 0.086) (Table 4). Discussion The mechanisms for the origination, development and metastasis of colorectal cancer have not yet become clear; however it is believed to be related to the transformation of adenoma to adenocarcinoma and the dysregulation of multiple oncogenes and tumor suppressor genes, forming a multi-factor, multi-step, graduated and complicated process. 5 The human Bmi-1 gene is located at the thirteenth region of the No. 10 chromosome short arm (10p13) and contains 10 exons and 10 introns. The Bmi-1 gene encodes a 326 amino acid nuclear protein with a relative molecular weight of 44,000 46,000. The structure of Bmi-1 protein is similar to the structure of drosophila proteins Psc and Su(z)2. Its N-terminus has a ring finger domain (RF) while its middle part displays the Helix- Turn-Helix-Turn-Helix-Turn (H-T-H-T-H-T) structure which is a DNA-binding domain. These two regions are associated with cellular transcription and tumor formation. 3 Currently, research has confirmed that Bmi-1 shows high expression in a variety of tumors. Kim et al. 6 tested the Bmi-1 gene expression in 51 cases of colorectal cancer tissues and paracancerous normal mucosa tissues and found that the expression of Bmi-1 mrna level is 2- to 3-fold 570 Chinese Journal of Cancer 2008; Vol. 27 Issue 12
Table 1 Correlation between the clinical pathologic features and the expression of Bmi-1 and Ki67 protein in colorectal cancer Table 2 Logistic regression analysis of the factors to the expression of Bmi-1 Table 3 Logistic regression analysis of the factors to the expression of Ki67 higher in colorectal cancer tissues than in normal intestinal tissues. Immunohistochemical staining shows that 65% (30/46) of cases have moderate to strong staining for Bmi-1 protein, in which the expression levels of p16 INK4a and p14 ARF are significantly declined and correlate with the origination and development of colorectal cancer. However, the relationship to prognosis was not indicated by our data. This finding parallels similar data achieved by other recent studies which confirm the direct involvement of Bmi-1 in the origination, development, invasion and metastasis of tumors. One study by Yan Feng et al. 7 found that the overexpression of Bmi-1 protein is closely correlated with lymph node metastasis and the clinical staging of breast cancer, indicating that a tumor may have a high degree of metastatic potential and that Bmi-1 could be a new molecular marker to predict the metastasis of breast cancer. A study by Song et al. 8 on nasopharyngeal carcinoma proved that the five-year survival rate of Bmi-1-positive patients is significantly lower than those of Bmi-1 negative. In a multi-factor analysis, Bmi-1 has been identified as a factor affecting the survival rate, and its upregulation is a marker for poor prognosis of nasopharyngeal cancer patients. KaiHong Huang et al. 9 tested 42 specimens of gastric cancer surgical resection using the RT-PCR method; their study found that the mrna expression of Bmi-1 gene is closely correlated with the size of gastric cancer, lymph node metastasis and the degree of invasion (p < 0.05). In addition, patients with positive Bmi-1mRNA expression have a much lower survival rate than those with negative Bmi-1mRNA expression, thus, detection www.landesbioscience.com Chinese Journal of Cancer 571
Table 4 Correlation of the expression of Bmi-1 to Ki67 protein in colorectal cancer tissues Figure 2. Survival curves of colorectal cancer patients with low or high expression of Bmi-1 or Ki67. of Bmi-1 mrna will help determine the development and prognosis of tumors. This study used immunohistochemistry to detect the Bmi-1 protein expression levels in colorectal cancer, colorectal adenomas and normal colorectal tissue. Results showed that the positive expression rate of Bmi-1 in colorectal cancer tissue was 51.7%, slightly lower than the result by Kim et al. The high expression rate of Bmi-1 in cancer tissue in our study was 25.0%, while the Bmi-1 protein expression rate in colorectal adenomas and normal colorectal tissue were 6.67% (2/30) and 0% (0/20) respectively. Among the two cases of colorectal adenoma with Bmi-1 protein high expression, one case had severe atypical dysplasia and the other case had cancer transformation. Upon statistical analysis, significant differences were identified for Bmi-1 protein high expression between colorectal cancer, colorectal adenomas, and normal colorectal tissue (p < 0.05). Our results showed that Bmi-1 protein expression in normal colorectal mucosa, adenoma and colorectal cancer was gradually increased. The existence of highly expressed Bmi-1 protein in colorectal cancer suggests that Bmi-1 protein correlates with the origination and development of colorectal cancer. This study used a chi-square test and showed that Bmi-1 protein high expression did not associate with the gender of patients, age, tumor size, tumor location, degree of differentiation, histological classification or the presence of lymph node metastasis (p > 0.05), but did correlate with the occurrence of distant metastasis and TNM staging. Specifically, the expression of Bmi-1 protein in colorectal cancer patients with distant metastasis was significantly higher than those without distant metastasis (p < 0.01), while the Bmi-1 protein expression in TNM staging III/IV patients was significantly higher than those of I/II patients (p < 0.05). Furthermore, logistic regression analysis showed that only the distant metastasis factor fit into this model, indicating that distant metastasis was the highest risk factor for Bmi-1 protein high expression (p < 0.01, B = 1.705, OR = 5.50). Our results suggest that Bmi-1 protein high expression in colorectal cancer tissue is a reference factor for determining the invasion and metastasis of colorectal cancer. Meanwhile, the Kaplan-Meier survival curve analysis showed that the survival rate of patients with Bmi-1 protein high expression was significantly lower than those with Bmi-1 low expression (χ 2 = 5.929, p = 0.015), consistent with the findings by Song et al. showing that the five-year survival rate of nasopharyngeal cancer patients with positive Bmi-1 expression was significantly lower than those with negative Bmi-1 expression. This result suggests that high expression of Bmi-1 protein is a reference parameter to determine the prognosis of the colorectal cancer patient. Many studies show that Ki67 is an important reference index for cell proliferation activity and an important marker to determine the degree of tumor malignance and invasion ability. WenXiu Gao et al. 10 applied SP immunohistochemistry to detect the Ki67 expression in 120 cases of colorectal cancer patients and their results show that Ki67 expression correlates with tumor location, lymph node metastasis, Dukes staging and survival time. Specifically, expression of Ki67 increases in groups with lymph node metastasis more than groups with no lymph node metastases, and it significantly increases with advance of Dukes staging. The postoperative survival rate of patients with Ki67 positive expression was significantly lower than those with negative expression, indicating a poorer prognosis in patients with Ki67 positive expression than those with negative Ki67 expression. Therefore, Ki67 has important reference value in predicting the prognosis of colorectal cancer and can be used as an independent prognostic indicator. Our results revealed that the rate of Ki67 protein high expression in 60 cases of colorectal cancer was 18.3% (11/60); in contrast, the high expression rate of 30 cases of colorectal adenomas was only 3.33% (1/30), in which the sole case with Ki67 572 Chinese Journal of Cancer 2008; Vol. 27 Issue 12
high expression had severe atypical hyperplasia; 20 cases of normal colorectal tissues all had low Ki67 expression. The high expression of Ki67 in colorectal cancer tissues was significantly different than that of colorectal adenomas and normal colorectal mucosa (p < 0.05) indicating that the expression of Ki67 protein in normal colorectal mucosa, adenoma and colorectal cancer gradually increased and the proliferation activity of the cancer cells gradually increased as well. This suggests that Ki67 expression correlates with the origination and development of colorectal cancer. Further study regarding the association between Ki67 protein expression and clinicopathological features found that, Ki67 high expression was correlated with age of patients, occurrence of distant metastasis and TNM staging (p < 0.05), while not related with the patients gender, tumor size, location, degree of differentiation, histological classification, lymph node metastasis and other clinicopathological features (p > 0.05), i.e., the expression of Ki67 protein in younger patients is significantly higher than in older patients, as well as higher in patients with distant metastasis as opposed to those without distant metastasis and its expression level increases along with the advance of TNM staging. Further logistic regression analysis showed that only the age of patients fit this model, indicating that the younger the patients were, the higher the Ki67 expression would be (p < 0.05). This suggests that the younger the clinical colorectal cancer patients are, the more vigorously tumor cells proliferate, the lower the degree of differentiation will be, the more malignant the cells will be, the easier the invasion and metastasis, the worse the prognosis. Meanwhile, analysis of the survival curve showed that the survival rate of patients with high Ki67 protein expression was significantly lower than those with low protein expression (χ 2 = 14.374, p = 0.000), suggesting that Ki67 high expression is an effective reference to determine the prognosis of colorectal cancer. This is consistent with the results by WenXiu Gao et al. The origination and development of cancer is a multi-gene and multi-factor process, and is the result of a dynamic imbalance between tumor cell proliferation and apoptosis. Through p16 INK4a /cyclin D/Rb and p19 ARF /MDM2/p53 pathways, Bmi-1 protein can negatively regulate the protein expression of p16 INK4a and p19 ARF, and therefore modulate the proliferation and apoptosis of cells, 11 while Ki67 is a good molecular marker reflecting the proliferation index of tumor cells. 12 The expression of Bmi-1 protein may correlate with the expression of Ki67 to a certain degree, but there are still no reports on the issue, neither domestically nor abroad. Our results showed that, the higher the Bmi-1 protein expression, the higher the Ki67 protein expression level, and the more vigorously tumor cells proliferate. Spearman correlation analysis showed that Bmi-1 protein expression in colorectal cancer was positively correlated with protein expression of Ki67 to a certain degree, but not statistically significant (r = 0.224, p = 0.086). The reason may be the relatively small sample number in this study or the involvement of multiple factors in regulating Ki67 protein expression by Bmi-1 protein, or the indirect regulation of Bmi-1 protein on Ki67 protein. Further study is needed to resolve those problems. Acknowledgements Grant: Sci-Tech Subject Foundation from Science Committee of Guangdong Province (No. 2006B35502009). References [1] Tsukuma H, Ajiki W. Descriptive epidemiology of colorectal cancer international comparison. Nippon Rinsho, 2003, 61(Suppl 7):25-30. [2] Dong Z W, Qiao Y L, Li L D, et al. Report of Chinese cancer control strategy. Zhong Guo Zhong Liu, 2002,11(5):250-260. [in Chinese] [3] Haupt Y, Alexander W S, Barri G, et al. Novel zinc finger gene implicated as myc collaborator by retrovirally accelerated lymphomagenesis in Emu-myc transgenic mice. Cell, 1991, 65(5):753-763. [4] Schluter C, Duchrow M, Wohlerberg C, et al. The cell proliferation associated nuclear antigen of antibody ki-67: A very large, ubiquiyous nuclear protein with numerous repeated elements, representing a new kind of cell cycle maintaining proteins. J Cell Biol, 1993,123(3):513-522. [5] Lv Y, Liu B, We J F. Colorectal cancer-related biological markers of progress in pathology. He Bei Bei Fang Xue Yuan Xue Bao (Medical Sciences). 2007, 24(1):80-82. [in Chinese] [6] Kim J H, Yoon S Y, Kim C N, et al. The Bmi-1 oncoprotein is overexpressed in human colorectal cancer and correlates with the reduced p16 INK4a /p14 ARF proteins. Cancer Lett, 2004, 203(2):217-224. [7] Feng Y, Song L B, Guo B H, et al. Expression and Significance of Bmi-1 in Breast Cancer. Ai Zheng, 2007,26(2):154-157. [in Chinese] [8] Song L B, Zeng M S, Liao W T, et al. Bmi-1 is a novel molecular marker of nasopharyngeal carcinoma progression and immortalizes primary human nasopharyngeal epithelial cells. Cancer Res, 2006, 66(12):6225-6232. [9] Huang K H, Liu J H, Li X X, et al. Association of Bmi-1 mrna expression with differentiation, metastasis and prognosis of gastric carcinoma. Nan Fang Yi Ke Da Xue Bao, 2007,27(7):973-979.[in Chinese] [10] Gao W X, Feng J G, Yang Y F, et al. PCNA in colorectal carcinoma and the prognosis. Xian Dai Yu Fang Yi Xue, 2007,34(2):225-229. [in Chinese] [11] Park I K, Morrison S J, Clarke M F. Bmi-1, stem cells, and senescence regulation. J Clin Invest, 2004,113(2):175-179. [12] Brown D C, Gatter K C. Ki67 protein: the immaculate deception?. Histopathology, 2002, 40(1): 2-11. www.landesbioscience.com Chinese Journal of Cancer 573