Cancer Stem Cell and its Niche in Malignant Progression of Oral Potentially Malignant Disorders

Abstract

Objective

The purpose of this study was to determine association between cancer stem cells (CSCs) and their niche with progression of oral potentially malignant lesions.

Materials and Methods

Patients with histologically confirmed oral potentially malignant disorders, stratified into high/low risk lesions based on the degree of dysplasia and oral cancer were included in this study. Immunohistochemical profiling of markers of CSCs (CD44), endothelial cells (CD31) and CSC-vascular niche cross-talk (CXCR4 and SDF1) were carried out. Statistical analysis was performed to correlate the relationship of markers with histopathology grade (ANOVA, and χ² test, unpaired t test) using GraphPad InStat v3.06.

Results

The study included 550 samples (349 patients) and analysis showed progressive increase in expression levels of CSC and its niche markers with increase in grade of dysplasia as compared to the normal cohort (p<0.05). Co-expression analysis revealed that, in comparison to the normal cohort, a larger percentage of patients showed increased expression of CD31 and CD44 (CD31^high/CD44^high; p<0.05) and of CXCR4 and SDF1 (CXCR4^high/SDF1^high; p=0.04), suggesting an association of the CSCs and the vascular niche. Further, distribution of patients with CD44^high/CXCR4^high (p<0.05) and CD31^high/SDF1^high (p=0.01) was significantly increased in the high-risk group (18%), suggesting a correlation between CD44⁺/CXCR4⁺ cells, the vascular niche and progression of oral dysplastic lesions.

Conclusion

The increased expression of CSCs, the vascular niche and their cross talk markers are associated with increase in severity of dysplasia suggesting their role in the progression of oral potentially malignant lesions and may hence be used in identifying high-risk OPMD.

INTRODUCTION

Oral cancer, which is the sixth most common cancer, accounts for 300,000 cases worldwide [1]. A large proportion of oral cancer is preceded by the development of oral leukoplakia [2, 3], an oral potentially malignant disorder (OPMD). The histologic progression of OPMD from hyperplasia, different grades of dysplasia to carcinoma-in-situ and invasive carcinoma and the associated genomic changes are well studied [4]. Cancer stem cells (CSCs) have been increasingly implicated in oral carcinogenesis and field cancerization [5], and are known to be regulated by stroma and endothelial cells constituting the CSC-niche [6, 7]. We hypothesized that increase in expression of markers specific to the CSCs, its vascular niche, and their cross talk correlates with progression of OPMD. These molecules, if validated in OPMD, could serve as potential markers for malignant transformation and targets for chemoprevention.

Among the multiple pathways implicated in the epithelial cell-niche cross talk (Notch1, TGFβ, and SDF-1/CXCR4) [8–11], the SDF-1/CXCR4 axis has been implicated in CSC-niche cross talk in many cancers including head and neck (HNSCC) [12]. In vitro studies investigating their role in HNSCC have shown that CXCR4 promotes migration and invasion specifically in CD44+ cells [13], while SDF-1 released from the Cancer Associated Fibroblast (CAFs) and endothelial cells induces increased migratory and invasive properties in the cancer cells [14]. Additionally, SDF-1/CXCR4 has been associated with poor prognosis in patients diagnosed with head and neck cancer [15, 16]. A definitive correlation between the CSCs, their vascular niche and the molecular signals involved in this cross-talk, should provide evidence towards their role in oral carcinogenesis.

MATERIALS AND METHODS

Patient Details

This is a retrospective study of patients with histologically confirmed OPMD and oral cancer presented to the Department of Head and Neck Surgery, Roswell Park Cancer Institute, New York, USA (2006 to 2013) and Department of Head and Neck Oncology at Mazumdar Shaw Medical Centre, Bangalore, India (2011 to 2013). The clinical and demographic details of these patients were obtained from the hospital medical records. Patients who have been previously treated for malignancy, diagnosed with infectious/inflammatory lesions and non-squamous malignancies were excluded from the study. Patients with clinically and histologically normal mucosa were used as controls for the study. The study was approved by the Institutional Ethical Committee of the respective hospitals (RPCI IEC No: I66805 and MSMC IRB approval dated 10.08.2010).

The oral mucosa of the patients enrolled in the study were categorized into normal, oral potentially malignant disorders (OPMD) and oral squamous cell carcinoma based on the histopathology reports. The OPMD patients were categorized as high and low risk based on established binary classification criteria [17]. Hyperplasia, parakeratosis and mild dysplasia were categorized as low-risk lesions, while moderate and severe dysplasia and carcinoma in situ as high-risk lesions. The carcinoma patients included micro-invasive carcinoma, and cancers of all stages (I-IV) and differentiation (well, moderate and poor).

Immunohistochemistry

The Formalin Fixed Paraffin Embedded (FFPE) sections were de-paraffinized, re-hydrated and antigen retrieval was carried out according to standard protocols. The sections were then incubated with the primary antibodies overnight at 4°C in a humidified chamber. The antibodies used were for CD44 (AM310-5M; Biogenex Life Sciences Pvt. Ltd., Hyderabad, India; Ready-to-use and MA4400; Thermo Fischer Scientific, Rockford, IL, USA; dilution: 1:50), CD31 (endothelial cell clone JC70A (M0823, DAKO, California, USA; dilution: 1:50), CXCR4 (ab2074; Abcam, Cambridge, MA, USA; 1:50) and SDF-1 (97958; Cell signaling technology, Danvers, MA, USA; 1:200). The two CD44 antibodies were evaluated for their comparative staining patterns (Supplementary Figure 1). Sections were then washed twice with 1X-TBST buffer, stained with a secondary detection kit (Real TM EnVision™ Detection system DAKO, Denmark) and counterstained using Mayers-Hematoxylin and mounted using DPX mountant. All the slides were scanned (Aperio ScanScope XT 1509, AT2, Leica biosystems, IL, USA) at 20× magnification, reviewed independently by two observers and images analyzed using the Aperio Imagescope v.11.2.0.780 (Leica biosystems, IL, USA). The reviewers were blinded to the histopathological diagnosis during the entirety of scoring.

Immunohistochemical scoring and analysis

The expression of CD31 was scored to denote the presence or absence of microvessels. Slides were initially viewed under minimal magnification to identify areas of maximal staining, or “hot spots”, a minimum of three CD31 positively stained hotspots were identified. The micro vessel density (MVD) was evaluated by counting the number of CD31 positive vessel lumen under a high power field (hpf) using a magnification of 400× field [18–20]. Four non-overlapping fields were counted per hot spot, the total number of vessel lumens/hpf recorded, and an average number of lumens per hot spot calculated.

Analysis of CD44 and SDF-1 was carried out by evaluating the staining in terms of the distribution (percentage of positive epithelial cells) and level of expression (intensity of staining) [21, 22]. Sections were considered positive when >10% of the tumor cells were stained with strong intensity. The percentage of positive cells (0-100%) was multiplied with the intensity of staining (weak/1+, moderate/2+, strong/3+) to obtain the H-score of CD44 and SDF1 expression with a maximum score being 300 (100% × 3+). Cytoplasmic/membrane and nuclear low and high expression of CXCR4 evaluated by counting percentage of positive cells (1 = <10%, 2 = 10–50%, 3 = >50%) for cytoplasmic and membrane staining (1–3). Positive nuclear staining was defined as a nuclear score of 6 or more [i.e. any slide with >50% of the cells expressing nuclear expression (3) with intermediate or strong]. Cells were counted in at least three fields (at ×400) within the lesion [23]. The scoring for all the antibodies was done by a panel of three experienced observers.

Statistical Analysis

All statistical analyses were performed using the R 3.3.1 statistical computing language. Statistical testing included Pearson correlation, ANOVA, and χ2 test of independence. Unpaired t test was carried out using GraphPad InStat version 3.06 (Graph Pad Software, San Diego California USA). Statistical significance was set at p<0.05 for all tests. The data is presented as Mean ± SEM for all relevant analyses. Univariate and multivariate analysis of frequency distributions of the patient cohorts (normal, low risk, high risk and carcinoma) within the marker expression patterns (marker^high and/or marker^low) was carried out and the statistical significance evaluated by Chi square or Fisher’s test.

RESULTS

Clinical Characteristics of the patients

Five hundred and fifty (n=550) samples from 349 patients were categorized into normal, different grades of dysplasia and squamous cell carcinoma based on their histopathological diagnosis. Among the study cohort (n=349), majority of the patients were males (n=211, 60.6%) with a median age of 57 years (range 21 to 91). While the most common histologic diagnosis was non-dysplastic lesions (n=227, 41.27%), the remaining patients were distributed into mild (n=148, 26.9%), moderate (n=47, 8.45%) and severe dysplasia (n=34, 6.2%) and carcinoma (n=94, 17.09%) (Table 1). The clinically and histologically normal samples (n=6) from healthy subjects, undergoing dental procedures, served as the control.

Table 1.

Patient details

	Total No.	550
Age	Median (Range)	57 (21-91)
Gender	Male	211 (60.6)
	Female	138 (39.6)
Habits	Yes	257 (73.9)
	No	63 (18.1)
Grades of Dysplasia	Normal	6
	Non-Dysplasia	227 (41.27)
	Mild Dysplasia	148 (26.9)
	Moderate Dysplasia	47 (8.45)
	Severe Dysplasia	34 (6.2)
	Carcinoma	94 (17.09)

Correlation of Microvascular Density (MVD) with grades of dysplasia

The expression of CD31 (Figure 1A), as an indicator of MVD, was scored and comparison of the average scores indicated a trend towards an increase in expression in dysplastic epithelia (low-risk: 12.78±0.65, high-risk: 13.63±0.68) as compared to normal (9.03±1.54), though the difference was not statistically significant. There was no increase in MVD in cancer patients in comparison to normal or OPMD (9.68±0.66) (Figure 1B). Categorization of the patients into CD31^high or CD31^low based on the average score in the normal cohort (9.03±1.54) revealed statistically significant distribution of patient cohorts between the CD31^high and CD31^low categories (p=0.01) (Table 2).

Figure 1. Immunohistochemical staining and average expression of CD31 and CD44 markers.

The expression profiling of CD31 and CD44 was carried out by IHC in the different patient cohort, low-risk (non-dysplasia, Mild dysplasia), high-risk (moderate and severe dysplasia) and carcinoma. (A) Representative images of the expression of these markers in the cohorts are provided. The average scores for all the markers were calculated and correlated for significance in the low risk, high risk and carcinoma. (B) Expression levels CD31 within the risk groups showed a trend of increase, although the difference was not statistically significant. (C) Statistically significant difference was observed between the average expression of CD44 in high risk (p=0.0193) and low risk (p=0.0001) when compared to the normal. All the immunohistochemical figures were taken at a magnification of 400×.

Table 2.

Individual correlation

Markers	CD31		CD44		CXCR4c*		CXCR4n*		SDF1
Risk Level	High	Low	High	Low	High	Low	High	Low	High	Low
N	107	102	145	63	152	288	126	312	204	219
Normal	8%	8%	4%	16%	4%	4%	2%	5%	3%	5%
Low risk	50%	43%	41%	59%	61%	75%	64%	73%	67%	75%
High risk	35%	25%	39%	12%	22%	9%	18%	11%	20%	5%
Cancer	7%	24%	16%	13%	13%	12%	16%	11%	10%	15%
p-value	0.01#		0.0001#		0.001#		0.04#		0.0001#

The distribution of the patient risk groups between the expression level groups was tested for significance by the Chi-square test.

^#p<0.05 was considered significant.

^*CXCR4c: CXCR4-Cytoplasmic; CXCR4n: CXCR4-Nuclear

Correlation of CD44 with grades of dysplasia

Evaluation of CD44 expression (Figure 1A) indicated a significant increase in levels in the dysplastic cohort (97.24±4.04; p=0.0015) compared to the non-dysplastic patients (73.15±5.32). The difference was also significant in all the patient groups compared to normal (67±12.03, p<0.05) samples. A comparison between the low and high risk groups, showed significant increase in the expression levels within high risk (114.84±5.44) as compared to the normal (67.02±12.03; p=0.019) and low risk groups (73.5±3.45; p=0.0001) (Figure 1C). The expression in carcinoma was also high (158.29±13.98) as compared to normal (p=0.034), low risk (p<0.0001) and high risk (p=0.001) OPMD groups (Figure 1C). A univariate analysis of the risk groups and carcinoma with the CD44 scores categorized as low or high (cut-off value of normal cohort: 100) indicated a significant association (p=0.0001).

Correlation of CXCR4 with grades of dysplasia

CXCR4 expression levels (nuclear staining, CXCR4n) (Figure 2A) showed statistically significant difference in average scores in high risk (1.39±0.095) and low risk (0.60±0.01) OPMD groups compared to normal (0, p=0.0001) (Figure 2B & C). Cytoplasmic CXCR4 (CXCR4c) also showed a significant increase in expression in high risk (4.45+0.07; p=0.0001) and low risk (2.6+0.0008, p=0.0001) when compared to normal (2.5±1.29). There was no significant difference observed in cancer patients in either cytoplasmic or nuclear staining of CXCR4. Univariate analysis of the risk groups and carcinoma with the cytoplasmic and nuclear CXCR4 scores categorized as low or high (cut-off value of normal cohort: CXCR4c: 2.5±1.29; CXCR4n: 0) indicated a significant association between the groups (p<0.05) (Table 2).

Figure 2. Immunohistochemical staining and average expression of CXCR4 and SDF1 markers.

The expression profiling of CXCR4 and SDF1 was carried out by IHC in the different patient cohort, low-risk (non-dysplasia, Mild dysplasia), high-risk (moderate and severe dysplasia) and carcinoma. The images for CXCR4 represent both cytoplasmic and nuclear staining together. (A) Representative images of the expression of these markers in the cohorts are provided. (B & C) CXCR4 expression levels were assessed for their cytoplasmic and nuclear expression wherein a significant difference was observed in high risk (p=0.0001) and low risk (p=0.0001) in comparison to normal samples in cytoplasmic expression levels. Nuclear expression of CXCR4 also showed significant difference in risk groups (p<0.05) when compared to normal. (D) SDF1 expression was also observed to show a significant difference in high risk and low risk (p=0.0001) groups. All the immunohistochemical figures were taken at a magnification of 400×.

Correlation of SDF1 with grades of dysplasia

Correlation of the expression in the patient cohorts (Figure 2A) showed a significant difference in high-risk (170.6+6.91) as compared to the low-risk (109.7+3.55, p=0.0001) groups. No significant difference was observed in cancer (65.8±10.4) patients when compared to the normal cohort (98.3±18.14) (Figure 2D). However, univariate analysis with SDF1 categorized as low or high (cut-off value of normal cohort: 98.3±18.14) indicated a significant association between the groups (p=0.0001) (Table 2).

Correlation of MVD and CD44 with grades of dysplasia

In an effort to correlate the CSCs with their microvascular niche, the expression of CD31 and CD44 were analyzed in combination. All the samples were grouped into four categories CD31^high/CD44^high, CD31^high/CD44^low, CD31^low/CD44^high and CD31^low/CD44^low as per the cutoff value detailed before. The distribution of the patients in the four groups indicated >25% of the CD31^high/CD44^high group being the low/high risk group (normal: 4%; low risk: 44%; high risk: 44%; carcinoma: 8%). Further, among the groups, the lowest percentage of high risk (4%) patients was observed in the CD31^low/CD44^low cohort. This indicated that the MVD in the vicinity of the CD44+ cancer stem cell correlated with advances in grades of dysplasia (Figure 3A, Supplementary Table 1).

Figure 3. Co-expression of the CSC-Niche and cross-talk pathway markers.

The distribution of patients in the different marker categories CD44/CD31 (A), CXCR4c/SDF1 (B), CXCR4n/SDF1 (C). The statistical significance of the distribution in also represented (h: high and l: low).

Correlation of CXCR4 and SDF1 with grades of dysplasia

In an effort to correlate the cross-talk pathway, the expression of CXCR4 and SDF1 were analyzed in combination. All the samples were grouped into four categories CXCR4^high/SDF1^high, CXCR4^high/SDF1^low, CXCR4^low/SDF1^high and CXCR4^low/SDF1^low as per the cutoff details mentioned previously. The correlation of expression levels of CXCR4 and SDF1 was evaluated by chi-square and Fisher exact testing with p<0.05 being considered statistically significant. The distribution of the patients in the four groups showed a statistically significant correlation between the co-expression of CXCR4, nuclear (CXCR4n) and cytoplasmic (CXCR4c), with SDF1. The CXCR4c^high/SDF1^high category included a larger percentage of high risk patients (22%) as compared to the CXCR4c^low/SDF1^low cohort (7%), with the overall distribution being statistically significant (p=0.02) (Supplementary Table 1) (Figure 3B). A similar correlation was observed with CXCR4n expression, wherein a higher percentage of high risk patients were in the CXCR4n^high/SDF1^high (20%) category, as compared to the CXCR4n^low/SDF1^low (4%), with the overall distribution being statistically significant (p=0.04) (Supplementary Table 1) (Figure 3C).

Correlation of CD31, CD44, CXCR4 and SDF1 with grades of dysplasia

To correlate MVD and cross-talk with CSCs, the expression of CD31, CD44, CXCR4 and SDF1 were analyzed in combination. All the samples were grouped into eight categories based on the expression of CD44 and CD31 as high/low in combination with SDF-1/CXCR4 as high/low as per the cutoff scores in the normal. Initially, epithelial markers, CD44 and CXCR4 were correlated with the stromal markers, CD31 and SDF1 in combination.

Co-expression of CD44 and CXCR4c/CXCR4n showed a statistically significant distribution between the four groups (CD44^high/CXCR4^high, CD44^high/CXCR4^low, CD44^low/CXCR4^high and CD44^low/CXCR4^low). In CD44^high/CXCRc^high category, a higher percentage of high risk patients were included (46%) as compared to CD44^low/CXCRc^low (11%) group, with the overall distribution being highly statistically significant (p= 0.0000092) (Supplementary Table 1) (Figure 4A). A similar correlation was observed in CXCR4n expression with a higher percentage of high risk (37%) and carcinoma (22%) patients in CD44^high/CXCRn^high as compared to CD44^low/CXCRn^low (high risk: 7% and carcinoma: 10%). None of the patients in the normal cohort (0%) were in the high group, while 19% of the CD44^low/CXCRn^low were normal (p=0.02) (Table 3) (Figure 4B).

Figure 4. Co-expression of the CSC markers and stromal component.

Correlation of MVD and cross talk with CSC markers was carried wherein the epithelial markers (CD44 and CXCR4) were correlated with stromal markers (CD31 and SDF1) both individually and in combination. (A) CD44/CXCR4c (B) CD44/CXCR4c (C) CD31/SDF1 (D) CD44/CXCR4c with CD31/SDF1 (E) CD44/CXCR4n with CD31/SDF1. The distribution of the patients in the different expression categories (h: high and l: low) of the marker pairs are statistically significant (p<0.05, Table 4)

Table 3.

Co-expression of CD44^* CXCR4c/CD31^* SDF1 and CD44^* CXCR4n/CD31^* SDF1 marker pairs

Markers	N	Normal	Low risk	High risk	Cancer	p-value
CD44/CXCR4c High, CD31/SDF1 High*	34	3%	70%	18%	9%	0.01
CD44/CXCR4c High, CD31/SDF1 Low*	54	7%	50%	17%	26%
CD44/CXCR4c Low, CD31/SDF1 High*	25	12%	64%	12%	12%
CD44/CXCR4c Low, CD31/SDF1 Low*	27	26%	67%	0%	7%
CD44/CXCR4n High, CD31/SDF1 High*	36	3%	72%	17%	8%	0.03
CD44/CXCR4n High, CD31/SDF1 Low*	55	9%	49%	16%	26%
CD44/CXCR4n Low, CD31/SDF1 High*	23	13%	61%	13%	13%
CD44/CXCR4n Low, CD31/SDF1 Low*	26	23%	69%	0%	8%

^*CXCR4c: CXCR4-Cytoplasmic; CXCR4n: CXCR4-Nuclear

Similar analysis with CD31 and SDF1, also showed a statistically significant correlation of marker expression with the different groups (CD31^high/SDF1^high, CD31^high/SDF1^low, CD31^low/SDF1^high and CD31^low/SDF1^low). CD31^high/SDF1^high category had the highest percentage of low risk (59%) and high risk (27%) patients when compared to CD31^low/SDF1^low (low risk: 39% and high risk: 11%). Accordingly, in the CD31^high/SDF1^high group, normal cohort showed the lowest percentage (7%) as compared to CD31^low/SDF1^low (14%), with the overall distribution being statistically significant across the different expression categories (p=0.01) (Supplementary Table 1) (Figure 4C).

Multivariate analysis with co-expression of CD44/CXCR4 and CD31/SDF1 with the risk groups and carcinoma showed a statistical significance in the distribution between the groups. In CD44/CXCR4c^high, CD31/SDF1^high category, a higher percentage of high risk patients (18%) were observed as compared to CD44/CXCR4c^low, CD31/SDF1^low (0%), with the overall distribution being statistically significant among the different expression categories (p=0.01) (Table 3) (Figure 4D). Similar analysis was carried with CXCR4n, wherein high risk patients were of a larger percentage in CD44/CXCR4n^high, CD31/SDF1^high (17%) groups, as compared to CD44/CXCR4n^low, CD31/SDF1^low (0%), with the overall distribution across the different expression categories, being statistically significant (p=0.03) (Table 3) (Figure 4E).

DISCUSSION

Understanding how dysplastic lesions progress to oral cancer is important for both diagnosis and management of these lesions. Additionally, elucidation of the relevant biologic pathways at both the cellular and molecular level is critical to understanding the biologic behavior of dysplastic lesions. Defining and understanding these biologic pathways are a necessary step in the development of predictive biomarker panels as well as novel chemopreventive strategies.

Cancer stem cells (CSCs) have the capacity to divide asymmetrically and expand into heterogeneous cells, which is believed to be a critical and necessary step in cancer progression [24]. However, these cells do not act in isolation and are biologically effective due to their surrounding microenvironment. In vivo, the vascular and stromal niche plays an important role in CSC maintenance [25]. Of the three recognized biologic niche components (CSC, fibroblast, endothelium), the vascular component is likely the most significant. There is evidence demonstrating the independent correlation of micro vascular density (MVD) in cancer progression and the central role played by their inter cellular communication (cross-talk) with CSC in malignant transformation [26]. Which of the molecular markers, when expressed in the niche could predict progression of the dysplastic lesion and the applicability of these markers towards identifying high risk lesions, are important questions for both diagnosis and treatment.

Disease progression in oral cavity cancer appears to be associated with the increased presence of CSCs. [27]. CD44 is one of the well-studied markers of CSCs of oral cancer [28, 29]. In our study, the expression of CD44 in patients strongly correlated with the progression of lesions to the more aggressive histologic type (mild to severe dysplasia and carcinoma), suggesting that the CSCs may play a role in the progression of dysplasia to invasive squamous cell carcinoma of the oral cavity. The role of tumor associated endothelial cells in tumor progression is well established. In breast carcinoma, intra-tumoral endothelial cells are reported to have a higher proliferative rate as compared to endothelial cells in benign stroma [5–7]. The prognostic value of microvessel density (MVD) has also been documented in a variety of cancers [7], with increased intra-tumoral MVD being positively associated with the rate of tumor progression [18, 30, 31], and metastatic potential of the primary tumor [30, 32–38]. In this study, lesions divided into low (parakeratosis, hyperplasia, mild dysplasia) and high risk categories (moderate and severe dysplasia) showed a statistically significant correlation with MVD expression. A correlation between the expression patterns of CSCs and MVD indicated that a majority of the patients with increasing levels of CD44 (CSC)/CD31 (endothelium) were associated with high-risk OPMD and carcinoma, indicating their potential, synergistic and biologically interdependent roles in tumor progression. These findings suggest that progression of epithelial dysplasia from benign to malignant phenotypes histopathologically correlates with CSC enrichment, which is in turn associated with the vascular component of its niche.

SDF1 (CXCL12) is a chemokine that is expressed by most components of the CSC niche and is known to increase vasculogenesis by recruiting CXCR4 (receptor on CSC) and endothelial progenitor cells [39]. In this study, the expression of SDF-1 and CXCR4 showed significant correlation with increasing dysplasia, indicating that the expression of these markers can determine high-risk lesions and hence may be a poor prognostic factor. Although evidence of CSC markers in the progression of premalignant lesions is scant, this is in accordance with evidence in head and neck cancers, wherein these markers are shown to be indicative of poor prognosis [40]. It is important to note that though this study has indicated association of CSC markers and degree of dysplasia; to determine whether CSC markers can prognosticate malignant transformation, it is essential to establish this association in a follow up study with malignant transformation as the end-point.

The analysis in this study indicated a significance of co-expression of CD44/CXCR4 as well as CD31/SDF1, wherein increasing the percentage of the high-risk groups had co-expression of these marker pairs. Prior studies have reported the effect of SDF-1 on CXCR4+/CD44+ cells in terms of their tumorigenic properties [41]. Accordingly, combinatorial analysis of the co-expression of all the marker sets also indicated a strong correlation in the high-risk group (p<0.05). Interestingly, except for CD44, the other markers did not show increasing trend of expression between dysplastic lesions and carcinoma. Similar observation was reported with other markers such as cytokeratins, wherein their expression is high during the initial stages of carcinogenesis and then down-regulated in carcinomas [42], suggesting the involvement of different processes in the later stages of cancer development as compared to tumor initiation. Nevertheless, the marker expression patterns demonstrated synergy of expression of these markers in high-risk dysplasia. These evidences suggest a molecular signature for the low (CD44^low/CD31^low/CXCR4^low/SDF-1^low) and high risk (CD44^high/CD31^high/CXCR4^high/SDF-1 ^high) lesions, apart from the clinical and histological characteristics. In combination with evidences that attribute the properties of drug resistance to CSCs, this signature is also possibly, indicative of their response to chemopreventive treatment (Figure 5); the low risk lesions being responsive to treatment, while the high risk ones show an initial response and then develop relapse on withdrawal of treatment. A major limitation towards establishing these markers as predictors of malignant progression is the lack of follow up data in this study. Follow up data is a challenge in such studies due to the long time to progression (5-10 years). Nevertheless, longitudinal follow up studies are mandatory to confirm the significance of this model in malignant transformation as well as to assess novel strategies in chemoprevention.

Figure 5. CSC-Niche in malignant transformation.

The model indicates the molecular signature that specifies low and high risk lesions of oral cancer and the possible correlation with chemoprevention outcome. The signature also may indicate to chemopreventive agents with the low risk lesions (CD44^low/CD31^low/CXCR4 ^low/SDF-1^low) responding better as compared to the high risk lesions (CD44^high/CD31^high/CXCR4^high/SDF-1^high), which after an initial response, show a relapse. Additional, novel strategies might be essential to obtain an improved outcome in high risk lesions.

Despite considerable evidences suggesting the role of CSCs and its niche in carcinogenesis, their role in oral cancer progression and/or their utility as biomarkers individually or in combination, has not been previously studied. Oral carcinogenesis, as is the case with other cancers, is known to follow a sequence of molecular alterations that leads to histological and clinical manifestations; whether these changes occur preferentially in the CSCs and aided by the niche are questions for further investigations. Findings from this study suggesting a role for CSCs, their vascular niche and the intercellular cross-talks during the progression of OPMD, lays the foundation for further investigation towards the development of these markers for possible clinical applications. Further studies are required to decipher the mechanistic role of the SDF1/CXCR4 axis in promoting CSC-driven oral carcinogenesis.

Supplementary Material

Supplementary Figure 1

Supplementary Figure 1: Comparative staining of CD44 antibodies. A comparison between the staining pattern between the two antibodies (AM310-5M and MA4400) indicated a similar staining for both the antibodies in the positive controls (n=2). Representative images for both antibodies AM310-5M (A) and MA4400 (B) are provided (400× magnification). Scale bars are represented.

Highlights.

• CD31, CD44, CXCR4 and SDF1 markers showed progressive increase in expression levels with an increase in grade of dysplasia as compared to the normal cohort (p<0.05)

• Co-expression of markers specifying CSCs (CD44), endothelial niche (CD31) and their cross talk (SDF-1, CXCR4) correlated with severity of dysplasia

• CSCs in combination with its niche and their cross-talk may be used in identifying high-risk OPMD