Two Stem Cell Markers, ATP-Binding Cassette, G2 Subfamily (ABCG2) and BMI-1, Predict the Transformation of Oral Leukoplakia to Cancer

Transcription

1 Two Stem Cell Markers, ATP-Binding Cassette, G2 Subfamily (ABCG2) and BMI-1, Predict the Transformation of Oral Leukoplakia to Cancer A Long-Term Follow-Up Study Wei Liu, MD 1 ; Jin-Qiu Feng, DDS 1,2 ; Xue-Min Shen, DDS 1 ; Hai-Yan Wang, DDS 1 ; Yang Liu, DDS 1 ; and Zeng-Tong Zhou, DDS 1 BACKGROUND: Although oral leukoplakia (OL) is the best-known potentially malignant disorder, the risk of OL malignant transformation is difficult to assess. ATP-binding cassette, G2 subfamily (ABCG2) and BMI-1 are stem cell markers that have been found to be associated with head and neck tumorigenesis. The objective of the current study was to evaluate the usefulness of ABCG2 and BMI-1 in predicting OL transformation. METHODS: In a retrospective cohort of 135 patients with OL from the study institution who had a mean follow-up of 5.5 years, 32 developed cancer between 1985 and The expression of ABCG2 and BMI-1 was determined using immunohistochemistry in samples from these patients, and included untransformed OL (n ¼ 103) and malignant-transformed OL (n ¼ 32). The association between protein expression and clinicopathological parameters and transformation was analyzed. RESULTS: Expression of ABCG2 and BMI-1 was observed in 58 (43.0%) and 44 (32.6%) of 135 patients, respectively. The correlation between ABCG2 and BMI-1 expression was significant (P ¼.024). Kaplan-Meier analysis revealed that 37.9% of patients with ABCG2 positivity developed cancer compared with 13.0% of patients with ABCG2 negativity (P ¼.014, log-rank test). Approximately 40.9% of patients with BMI-1 positivity developed cancer compared with 15.4% of patients with BMI-1 negativity (P ¼.029, log-rank test). Multivariate analysis revealed that ABCG2 and BMI-1 expression was associated with a 3.24-fold (95% confidence interval [95% CI], ; P ¼.011) and 4.03-fold (95% CI, ; P ¼.003) increased the risk of transformation, respectively. CONCLUSIONS: ABCG2 and BMI-1 expression was found to be associated with the development of oral cancer in a large cohort of patients with OL for whom long-term follow-up was available, which suggests that ABCG2 and BMI-1 may be used as predictors of OL transformation. Cancer 2012;118: VC 2011 American Cancer Society. KEYWORDS: ATP-binding cassette, G2 subfamily (ABCG2), BMI-1, stem cell marker, oral cancer, leukoplakia, malignant transformation. INTRODUCTION Oral cancer is the sixth most frequent leading cause of cancer death worldwide; the early diagnosis of high-risk, potentially malignant lesions is a high priority for reducing deaths due to oral cancer. 1,2 Oral leukoplakia (OL) is the bestknown potentially malignant disorder, with a rate of malignant transformation of between 17% and 35%. 3 Histologic assessment of epithelial dysplasia is currently the gold standard for evaluating the risk of OL malignant transformation. However, the diagnosis and grading of epithelial dysplasia is subjective, with interexaminer and intraexaminer variability. 4-7 Moreover, some dysplastic lesions may be remain static or even regress and nondysplastic lesions also may become malignant. 7,8 Therefore, objective biomarkers are needed to evaluate and clarify the risk of OL malignant transformation. Corresponding author: Zeng-Tong Zhou, DDS, Department of Oral Mucosal Diseases, Ninth People s Hospital, Shanghai Jiao Tong University School of Medicine, 639 Zhizaoju Road, Shanghai , China; Fax: (011) ; liuweb@hotmail.com 1 Department of Oral Mucosal Diseases, Ninth People s Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai Key Laboratory of Stomatology, Shanghai, People s Republic of China; 2 Department of Preventive Dentistry, Shanghai Municipal Hospital for Oral Health, Shanghai, People s Republic of China The first 3 authors contributed equally to this work. We thank Drs. Jiang Li and Li-Zhen Wang (Department of Oral Pathology, Ninth People s Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, People s Republic of China) for technical support. DOI: /cncr.26483, Received: April 24, 2011; Revised: July 3, 2011; Accepted: July 12, 2011, Published online August 25, 2011 in Wiley Online Library (wileyonlinelibrary.com) Cancer March 15,

2 ATP-binding cassette, G2 subfamily (ABCG2) isa member of the ATP-binding cassette transporter protein superfamily that produces multiple drug-resistant cancers. It is also known as a molecular determinant for maintaining the side population phenotype in stem cells. 9 This phenotype has been isolated from several primary tumor types and cancer cell lines, including those of the lung, esophagus, nasopharynx, and oral cavity Furthermore, ABCG2 expression has been found in solid tumors of hepatocellular, lung, esophageal, and head and neck squamous carcinomas Increasing evidence suggests that ABCG2 is recognized as an universal marker of stem cells and plays a central role in tumorigenesis. BMI-1 is a polycomb group transcription repressor that mediates gene silencing by regulating chromatin structure. It is considered to be a stem cell-related gene that has been demonstrated to play a role in the self-renewal of cancer stem cells. 17 BMI-1 is increasingly believed to be one of the stem cell markers that has been reported to be essential for maintaining the selfrenewal properties of and tumorigenicity in head and neck cancer. 11,18-20 Furthermore, it has been demonstrated to be a prognostic marker in lung cancer, gastric cancer, and head and neck cancer BMI-1 was also found to be overexpressed in human gastrointestinal and oral dysplastic lesions, indicating that its abnormal expression occurs early in gastrointestinal and oral tumorigenesis. 26,27 Recently, Kawaguchi et al demonstrated that podoplanin was a promising biomarker for predicting the risk of oral cancer in patients with OL. 28 Shi et al reported that the expression of podoplanin and ABCG2 in oral lichen planus was significantly associated with the risk of malignant transformation. 29 Vormittag et al found that coexpression of podoplanin and BMI-1 significantly predicted overall survival in patients with head and neck squamous cell carcinoma. 30 Tokar et al reported that ABCG2 and BMI-1 in combination with other stem cell makers maintained the self-renewal properties of the prostate cancer stem-like phenotype. 31 Therefore, we hypothesized that the expression patterns of ABCG2 and BMI-1, 2 stem cell markers, would also be dysregulated in patients with OL. In this retrospective cohort study, we examined the protein expression of ABCG2 and BMI-1 in samples from 135 patients with OL for whom a mean follow-up of 5.5 years was available, and evaluated their usefulness as biomarkers for predicting the transformation of OL to oral cancer. MATERIALS AND METHODS Patients and Tissue Specimens All 135 patients with a clinical and pathological diagnosis of OL in the current study were retrieved from the archived files at the Department of Oral Mucosal Diseases at the Ninth People s Hospital, Shanghai Jiao Tong University School of Medicine, between 1985 and The inclusion criteria and detailed clinicopathological information collection methods were previously described. 32 Biopsy specimens were obtained and subsequently fixed in formalin and embedded in paraffin. In this retrospective cohort study, malignant transformation versus nontransformation of OL was considered to be the surrogate for the clinical outcome of the patients. Of 135 patients with OL with a mean follow-up period of 5.5 years, 32 (23.7%) developed oral squamous cell carcinoma that was confirmed by histopathology. This study was approved by the Institutional Review Board. Of 135 lesions, 73 (54.1%) were located at the tongue, followed by the cheek (28.1%), gingiva (8.1%), palate (5.9%), and floor of the mouth (3.7%). All lesions were classified as tongue and nontongue sites according to our previous report. 32 In addition, all lesions were reclassified into low-grade dysplasia and high-grade dysplasia in the current study. The architecture (a total of 7) and cytology (a total of 9) criteria for epithelial dysplasia were described previously. 32 In the current study, a low-grade lesion was based on observing < 4 architectural changes or < 5 cytological changes. A high-grade lesion was based on observing at least 4 architectural changes and 5 cytological changes. For all the subjects, the treatments were grouped into 3 categories (Table 1). Tissue Processing and Immunohistochemistry Serial tissue sections (measuring 5 lm in thickness) from formalin-fixed, paraffin-embedded tissue blocks of OL were mounted on positively charged glass slides. Immunohistochemical staining was performed using the streptavidin-peroxidase method as described previously. 29 The anti-abcg2 monoclonal antibody (1:100 dilution, sc ; Santa Cruz Biotechnology, Inc., CA) was used to detect ABCG2 expression. The anti-bmi-1 monoclonal antibody (1:150 dilution, ab14389; Abcam, Cambridge, UK) was used to detect BMI-1 expression. Cell membrane and/or cytoplasmic immunoreactivity in the epithelium was considered to indicate positive expression of ABCG2. Nuclear immunoreactivity in the epithelium was considered to indicate positive expression of BMI Cancer March 15, 2012

3 ABCG2 and BMI-1 in Leukoplakia/Liu et al Sample evaluation of immunoreactivity was performed by 2 oral pathologists (J.L. and L.Z.W.) according to the staining intensity criteria described previously by Kang et al. 27 Statistical Analysis Statistical analysis was performed with the chi-square test and Fisher exact test among qualitative variables, the Student t test among quantitative variables, and the Mann- Whitney U test among categorical variables. Pearson correlation analysis was used to determine the relation between ABCG2 and BMI-1 expression for all cases. For time-to-event analysis, Kaplan-Meier curves were plotted Table 1. Patient Characteristics Oral Leukoplakia Characteristic Total UT MT P All patients, no. (%) Age, y.561 Mean (SD) 52.9 (10.7) 54.2 (12.5) Range Gender.107 Female (48.5) 21 (65.6) Male (51.5) 11 (34.4) Lesion site.158 Nontongue (49.5) 11 (34.4) Tongue (50.5) 21 (65.6) Dietary habits.403 Bland (81.4) 25 (89.3) Spicy (18.6) 3 (10.7) Missing 10 Smoking history.148 Never (73.2) 26 (86.7) Past and present (26.8) 4 (13.3) Missing 8 Alcohol intake 1.00 Never (91.7) 27 (90.0) Past and present 11 8 (8.3) 3 (10.0) Missing 9 Epithelial dysplasia.016 Low grade (81.6) 19 (59.4) High grade (18.4) 13 (40.6) Treatment.858 Vit A (52.4) 18 (56.3) Vit A 1 Chinese herb (35.0) 9 (28.1) Vit A 1 surgery (12.6) 5 (15.6) Follow-up, y.569 Mean (SD) 5.6 (3.7) 5.1 (3.6) Range Abbreviations: MT, MT, malignant-transformed oral leukoplakia; SD, standard deviation; UT, untransformed oral leukoplakia; Vit A, vitamin A. and the log-rank test was used to provide analysis of oral cancer-free survival (OCFS) analysis. The logistic regression model was applied to evaluate hazard ratios for OL malignant transformation. Odd ratios (ORs) with 95% confidence intervals (95% CIs) and P values were reported. All tests were 2-sided, and P values <.05 were considered statistically significant. RESULTS Patient Characteristics In this retrospective cohort study, participants were grouped as untransformed OL (n ¼ 103) and malignanttransformed OL (n ¼ 32). Table 1 summarizes the characteristics of these patients. A significant difference with regard to epithelial dysplasia (P ¼.016, Fisher exact test) was found, whereas differences with regard to age, gender, lesion site, dietary habits, smoking history, alcohol intake, treatment, and follow-up time were not observed between the 2 groups. ABCG2 and BMI-1 Expression in OL Among the 135 patients with OL lesions, 58 (43.0%) demonstrated positive expression of ABCG2 (Figs. 1A and 1B). We observed ABCG2 expression in 36 of 103 (35.0%) patients with untransformed OL and in 22 of 32 (68.8%) patients with malignant-transformed OL (P ¼.001, Fisher exact test) (Fig. 1E). Conversely, 44 of 135 patients with OL lesions (32.6%) demonstrated positive expression of BMI-1 (Figs. 1C and 1D). We noted BMI-1 expression in 26 of 103 (25.2%) patients with untransformed OL and in 18 of 32 (56.2%) patients with malignant-transformed OL (P ¼.002, Fisher exact test) (Fig. 1F). The correlation between ABCG2 and BMI-1 expression was found to be statistically significant (P ¼.024; Pearson correlation coefficient, 0.195). Elevated the predictors of OL transformation To investigate the time to OL malignant transformation, we performed Kaplan-Meier analysis for OCFS using ABCG2 and BMI-1 expression and clinicopathological parameters. In this cohort, ABCG2 and BMI-1 expression and epithelial dysplasia were found to be the significant predictors using the log-rank test. Among the 135 OL lesions, 10 of 77 (13.0%) ABCG2-negative lesions and 22 of 58 (37.9%) ABCG2-positive lesions, respectively, developed into cancer. The 5-year OCFS rate for ABCG2-negative lesions was 88.1% (95% CI, ) compared with 68.3% (95% CI, Cancer March 15,

4 Figure 1. Expression of ATP-binding cassette, G2 subfamily (ABCG2) is shown in (A) basal layer cells and (B) suprabasal layer cells of oral leukoplakia (OL). BMI-1 expression is shown in (C) basal layer cells and (D) suprabasal layer cells of OL (original magnification 400). (E) The frequency of ABCG2 expression is 35.0% in untransformed OL and 68.8% in malignant-transformed OL. * indicates P ¼.001 using the Fisher exact test. (F) The frequency of BMI-1 expression is 25.2% in untransformed OL and 56.2% in malignant-transformed OL. ** indicates P ¼.002 using the Fisher exact test ) for ABCG2-positive lesions (P ¼.014, log-rank test) (Fig. 2A). Conversely, 14 of 91 (15.4%) BMI-1 negative lesions and 18 of 44 (40.9%) BMI-1 positive lesions developed into cancer, respectively. The 5-year OCFS rate for BMI-1 negative lesions was 81.7% (95% CI, ) compared with 73.9% (95% CI, ) for BMI-1 positive lesions (P ¼.029, log-rank test) (Fig. 2B). It is interesting to note that we observed that coexpression of both ABCG2 and BMI-1 in patients with OL was not predictive of malignant transformation (P ¼.082, log-rank test) (Fig. 2C). However, expression of either ABCG2 or BMI-1 in OL lesions was found to be strongly predictive of malignant transformation. The 5-year OCFS rate for lesions with negative expression of the 2 proteins was 93.2% (95% CI, ) compared with 70.6% (95% CI, ) for OL lesions with positive expression of at least 1 of the proteins (P ¼.002, log-rank test) (Fig. 2D). In addition, patients with OL with high-grade dysplasia (5-year OCFS rate, 57.2%) had a much higher incidence of malignant 1696 Cancer March 15, 2012

5 ABCG2 and BMI-1 in Leukoplakia/Liu et al Figure 2. Oral cancer-free survival rates are shown by (A) ATP-binding cassette, G2 subfamily (ABCG2) negativity and ABCG2 positivity, (B) BMI-1 negativity and BMI-1 positivity, (C) both ABCG2 positivity and BMI-1 positivity, (D) either ABCG2 positivity or BMI-1 positivity, and (E) low-grade and high-grade dysplasia. transformation compared with those with OL with lowgrade dysplasia (5-year OCFS rate, 85.4%) (P <.001, log-rank test) (Fig. 2E). Elevated the risk of OL transformation To evaluate the risk of OL malignant transformation, clinicopathological parameters and ABCG2 and BMI-1 expression were analyzed by logistic regression (Table 2). In the univariate analysis, expression of ABCG2 and BMI-1 was found to be associated with a 4.09-fold (95% CI, ; P ¼.001) and 3.81-fold (95% CI, ; P ¼.002), respectively, increased risk of malignant transformation. In addition, the risk of malignant transformation of OL with high-grade dysplasia was higher than that of OL with low-grade dysplasia (OR, 3.03; 95% CI, [P ¼.012]). To further assess the influence of each factor, we performed multivariate analysis. All 3 factors retained statistical significance. The adjusted OR for malignant transformation was 3.24 for ABCG2 expression (95% CI, ; P ¼.011) Cancer March 15,

6 Table 2. Logistic Regression Analysis of Oral Cancer Risk Variate OR (95% CI) P Univariate analysis Age 1.01 ( ).558 Gender 0.49 ( ).094 Lesion site 1.87 ( ).136 Dietary habits 0.53 ( ).335 Smoking history 0.42 ( ).138 Alcohol intake 1.22 ( ).778 Epithelial dysplasia 3.03 ( ).012 ABCG2 expression 4.09 ( ).001 BMI-1 expression 3.81 ( ).002 Multivariate analysis Epithelial dysplasia 3.86 ( ).008 ABCG2 expression 3.24 ( ).011 BMI-1 expression 4.03 ( ).003 Abbreviations: 95% CI, 95% confidence interval; ABCG2, ATP-binding cassette, G2 subfamily; OR, odds ratio. and 4.03 for BMI-1 expression (95% CI, ; P ¼.003), respectively. DISCUSSION To the best of our knowledge, the current study is the first to evaluate ABCG2 and BMI-1, 2 stem cell markers, as predictors of the risk of oral cancer in a large cohort of 135 patients with OL for whom long-term follow-up information (mean, 5.5 years) was available. Kaplan- Meier analysis for the 5-year OCFS rate revealed that 31.7% of patients with OL demonstrating ABCG2 positivity developed oral cancer, compared with 11.9% of those with OL demonstrating ABCG2 negativity; 26.1% of patients with OL demonstrating BM1-1 positivity developed oral cancer, compared with 18.3% of patients with OL that was negative for BMI-1. Multivariate analysis revealed that expression of ABCG2 and BMI-1 was associated with a 3.24-fold and 4.03-fold, respectively, increase in the risk of malignant transformation. Kawaguchi et al 28 have suggested that podoplanin may be used as a biomarker for oral cancer risk assessment. Nonetheless, podoplanin expression alone may not be sufficient to clarify carcinogenesis. Podoplanin-positive expression extending to the suprabasal layer may represent upward clonal expansion of stem cells during carcinogenesis, and OL lesions with such clonal expansion may indicate a significantly higher risk of malignant transformation. 28 In the current study, we attempted to elucidate this clonal expression pattern of stem cells in the epithelia of patients with OL using 2 stem cell markers. ABCG2 is a molecular determinant for maintaining the side population phenotype in stem cells through its ability to exclude Hoechst dye. 10 Side population cells have been reported to possess cancer stem cell-like properties A recent study suggested that side population cells may play an important role in oral carcinogenesis. 14 Furthermore, ABCG2 as a stem cell marker has been demonstrated to be upregulated in a subset of cancer stem-like cells from patients with head and neck/oral squamous cell carcinoma. 33,34 In the current study, ABCG2 expression was observed in 43% of patients with OL. In a previous study, Shi et al reported 21% of lesions demonstrating ABCG2 expression in the epithelia of oral lichen planus, and 67% of these occurred in tumor cells of oral squamous cell carcinoma. 29 These patterns of ABCG2 expression may reflect a stepwise transformation of potentially malignant oral disorders to cancer, although the biological function of ABCG2 in these lesions remains unclear. BMI-1 has been studied as an oncoprotein associated with poor prognosis in several cancers Recent studies have reported that BMI-1 as a stem cell marker was essential for maintaining the self-renewal properties of and tumorigenicity in head and neck/oral squamous cell carcinoma. 11,19-21 In potentially malignant lesions, Tateishi et al 26 found that BMI-1 was expressed in 13% of low-grade gastrointestinal dysplasia, and Kang et al 27 observed it to be constantly overexpressed in a small series of patients with oral dysplasia. In the current study, BMI-1 expression was observed in 33% of patients with OL. The correlation between ABCG2 and BMI-1 expression in patients with OL was significant (P ¼.024), most likely because of the similar nature of the stem cell marker. Taken together, these data not only support the potential significance of ABCG2 and BMI-1 in early oral tumorigenesis but also suggest that they may be used as predictors of oral cancer risk in patients with potentially malignant disorders. In our clinic, periodic follow-up examinations at intervals of every 6 months or fewer were recommended for patients with OL. Based on the clinical implications of the current study, the shorter (eg, 3 months) and longer (eg, 9 months) follow-up intervals could be recommended for those patients with positive or negative expression of ABCG2 or BMI-1. Individualized treatment may reduce the costs, labor, and expertise involved. In addition, previous follow-up studies reported loss of heterozygosity patterns; p16 INK4A and SIBLING family proteins also could be used as markers to predict the risk of malignant transformation These biomarkers still 1698 Cancer March 15, 2012

7 ABCG2 and BMI-1 in Leukoplakia/Liu et al need to be validated in further prospective, longitudinal, large cohort studies. In the retrospective cohort in the current study, 23.7% of patients with OL developed oral cancer, which was in agreement with data documented in the literature. 3 Herein, differences in treatments administered to the patients with OL were not observed, and to the best of our knowledge only a few previous prevention studies have demonstrated effectiveness in preventing the transformation of OL to cancer. 38 Age, gender, lesion site, dietary habits, smoking history, and alcohol intake were not found to be significant risk factors for OL malignant transformation in the current study cohort. It should be noted that epithelial dysplasia was found to be a significant indicator associated with OL malignant transformation, as previously reported by us. 32 Actually, many clinicians currently rely on the degree of dysplasia present in patients with OL as an important indicator of oral cancer risk in routine practice. However, it is well known that the histological classification of potentially malignant lesions is insufficient. 4-8 In addition, the interpretation of protein expression of significant biomarkers is relatively simple in clinical practice. Therefore, immunohistochemical staining of biomarkers is more promising for the evaluation of oral cancer risk. To the best of our knowledge, the current study is the first to report that the expression of ABCG2 and BMI-1, 2 stem cell markers, was associated with the development of oral cancer in a large cohort of patients with OL for whom long-term follow-up data were available. The current study data suggest that ABCG2 and BMI-1 may be used as powerful biomarkers for oral cancer risk assessment in patients with OL, but these need to be validated in further prospective longitudinal studies. FUNDING SUPPORT No specific funding was disclosed. CONFLICT OF INTEREST DISCLOSURES Supported by the Shanghai Jiao Tong University School of Medicine (BXJ to W.Z.), the State Administration of Traditional Chinese Medicine of China (09J1X1L420B227 to Z.T.Z.), the Science and Technology Commission of Shanghai (08DZ and 10DZ to Z.T.Z.), and the Shanghai Leading Academic Discipline Project (S30206 to X.M.S.). REFERENCES 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, CA Cancer J Clin. 2008;58: Neville BW, Day TA. Oral cancer and precancerous lesions. CA Cancer J Clin. 2002;52: van der Waal I, Schepman KP, van der Meij EH, Smeele LE. Oral leukoplakia: a clinicopathological review. Oral Oncol. 1997;33: Karabulut A, Reibel J, Therkildsen MH, Praetorius F, Nielsen HW, Dabelsteen E. Observer variability in the histologic assessment of oral premalignant lesions. J Oral Pathol Med. 1995;24: Abbey LM, Kaugars GE, Gunsolley JC, et al. Intraexaminer and interexaminer reliability in the diagnosis of oral epithelial dysplasia. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1995;80: Kujan O, Khattab A, Oliver RJ, Roberts SA, Thakker N, Sloan P. Why oral histopathology suffers inter-observer variability on grading oral epithelial dysplasia: an attempt to understand the sources of variation. Oral Oncol. 2007; 43: Ho PS, Chen PL, Warnakulasuriya S, Shieh TY, Chen YK, Huang IY. Malignant transformation of oral potentially malignant disorders in males: a retrospective cohort study. BMC Cancer. 2009;9: World Health Organization. World Health Organization international histological classification of tumors. In: Pindborg JJ, Reichart PA, Smith CJ, Van der Waal I, eds. Histological typing of Cancer and Precancer of the Oral Mucosa. Berlin: Springer; 1997: Zhou S, Schuetz JD, Bunting KD, et al. The ABC transporter Bcrp1/ABCG2 is expressed in a wide variety of stem cells and is a molecular determinant of the side-population phenotype. Nat Med. 2001;7: Ho MM, Ng AV, Lam S, Hung JY. Side population in human lung cancer cell lines and tumors is enriched with stem-like cancer cells. Cancer Res. 2007;67: Huang D, Gao Q, Guo L, et al. Isolation and identification of cancer stem-like cells in esophageal carcinoma cell lines. Stem Cells Dev. 2009;18: Wang J, Guo LP, Chen LZ, Zeng YX, Lu SH. Identification of cancer stem cell-like side population cells in human nasopharyngeal carcinoma cell line. Cancer Res. 2007;67: Zhang P, Zhang Y, Mao L, Zhang Z, Chen W. Side population in oral squamous cell carcinoma possesses tumor stem cell phenotypes. Cancer Lett. 2009;277: Zen Y, Fujii T, Yoshikawa S, et al. Histological and culture studies with respect to ABCG2 expression support the existence of a cancer cell hierarchy in human hepatocellular carcinoma. Am J Pathol. 2007;170: Tsunoda S, Okumura T, Ito T, et al. ABCG2 expression is an independent unfavorable prognostic factor in esophageal squamous cell carcinoma. Oncology. 2006;71: Diestra JE, Scheffer GL, Catala L, et al. Frequent expression of the multi-drug resistance-associated protein BCRP/MXR/ ABCP/ABCG2 in human tumours detected by the BXP-21 monoclonal antibody in paraffin-embedded material. J Pathol. 2002;198: Sparmann A, van Lohuizen M. Polycomb silencers control cell fate, development and cancer. Nat Rev Cancer. 2006;6: Prince ME, Sivanandan R, Kaczorowski A, et al. Identification of a subpopulation of cells with cancer stem cell properties in head and neck squamous cell carcinoma. Proc Natl Acad Sci U S A. 2007;104: Cancer March 15,

8 19. Chen YC, Chang CJ, Hsu HS, et al. Inhibition of tumorigenicity and enhancement of radiochemosensitivity in head and neck squamous cell cancer-derived ALDH1-positive cells by knockdown of Bmi-1. Oral Oncol. 2010;46: Ganguli-Indra G, Wasylyk C, Liang X, et al. CTIP2 expression in human head and neck squamous cell carcinoma is linked to poorly differentiated tumor status. PLoS One. 2009;4:e Vrzalikova K, Skarda J, Ehrmann J, et al. Prognostic value of Bmi-1 oncoprotein expression in NSCLC patients: a tissue microarray study. J Cancer Res Clin Oncol. 2008; 134: Liu JH, Song LB, Zhang X, et al. Bmi-1 expression predicts prognosis for patients with gastric carcinoma. J Surg Oncol. 2008;97: Huber GF, Albinger-Hegyi A, Soltermann A, et al. Expression patterns of Bmi-1 and p16 significantly correlate with overall, disease-specific, and recurrence-free survival in oropharyngeal squamous cell carcinoma [published online ahead of print March 29, 2011]. Cancer. doi: /cncr Song LB, Zeng MS, Liao WT, et al. Bmi-1 is a novel molecular marker of nasopharyngeal carcinoma progression and immortalizes primary human nasopharyngeal epithelial cells. Cancer Res. 2006;66: Hayry V, Makinen LK, Atula T, et al. Bmi-1 expression predicts prognosis in squamous cell carcinoma of the tongue. Br J Cancer. 2010;102: Tateishi K, Ohta M, Kanai F, et al. Dysregulated expression of stem cell factor Bmi-1 in precancerous lesions of the gastrointestinal tract. Clin Cancer Res. 2006;23: Kang MK, Kim RH, Kim SJ, et al. Elevated Bmi-1 expression is associated with dysplastic cell transformation during oral carcinogenesis and is required for cancer cell replication and survival. Br J Cancer. 2007;96: Kawaguchi H, El-Naggar AK, Papadimitrakopoulou V, et al. Podoplanin: a novel marker for oral cancer risk in patients with oral premalignancy. JClinOncol. 2008;26: Shi P, Liu W, Zhou ZT, He QB, Jiang WW. Podoplanin and ABCG2: malignant transformation risk markers for oral lichen planus. Cancer Epidemiol Biomarkers Prev. 2010; 19: Vormittag L, Thurnher D, Geleff S, et al. Co-expression of Bmi-1 and podoplanin predicts overall survival in patients with squamous cell carcinoma of the head and neck treated with radio(chemo)therapy. Int J Radiat Oncol Biol Phys. 2009;73: Tokar EJ, Diwan BA, Waalkes MP. Arsenic exposure transforms human epithelial stem/progenitor cells into a cancer stem-like phenotype. Environ Health Perspect. 2010;118: Liu W, Wang YF, Zhou HW, Shi P, Zhou ZT, Tang GY. Malignant transformation of oral leukoplakia: a retrospective cohort study of 218 Chinese patients. BMC Cancer. 2010;10: Chiou SH, Yu CC, Huang CY, et al. Positive correlations of Oct-4 and Nanog in oral cancer stem-like cells and highgrade oral squamous cell carcinoma. Clin Cancer Res. 2008;14: Okamoto A, Chikamatsu K, Sakakura K, Hatsushika K, Takahashi G, Masuyama K. Expansion and characterization of cancer stem-like cells in squamous cell carcinoma of the head and neck. Oral Oncol. 2009;45: Rosin MP, Cheng X, Poh C, et al. Use of allelic loss to predict malignant risk for low-grade oral epithelial dysplasia. Clin Cancer Res. 2000;6: Montebugnoli L, Cervellati F, Cocchi R, et al. Immunohistochemical expression of p16(ink4a) protein as a helpful marker of a subset of potentially malignant oral epithelial lesions: study on a series with long-term follow-up. Histopathology. 2010;57: Ogbureke KU, Abdelsayed RA, Kushner H, Li L, Fisher LW. Two members of the SIBLING family of proteins, DSPP and BSP, may predict the transition of oral epithelial dysplasia to oral squamous cell carcinoma. Cancer. 2010;116: Lodi GSA, Sardella A, Bez C, Demarosi F, Carrassi A. Interventions for treating oral leukoplakia. Cochrane Database Syst Rev. 2006;(4):CD Cancer March 15, 2012