Abstract
Background/Aim: Spindle-shaped stromal cells of tumors are derived from various cellular origins, including mesenchymal stem cells (MSCs). MSCs express CD73, CD90 and CD105 antigens. Herein, the aim was to investigate the association between the expression of specific MSC markers in gastric cancer stromal cells and the clinicopathological features of the disease. Materials and Methods: The expression of CD73, CD90 and CD105 in spindle-shaped cancer stromal cells was studied by immunohistochemistry in tissue arrays containing 546 gastric cancer cases. Univariate and multivariate analyses were performed to evaluate the association of MSC marker expression with clinicopathological variables. Results: Spindle-shaped cancer stromal cells expressing the MSC markers CD73, CD90 or CD105 were associated with larger tumor size, advanced cancer, venous infiltration, lymphatic infiltration, and lymph node metastasis. Statistical analysis demonstrated that the presence of CD105-positive spindle cells was an independent prognostic factor of advanced cancer, lymph node metastasis and EBV infection in multivariate analysis. Conclusion: Spindle-shaped gastric cancer stromal cells expressing CD73, CD90 or CD105 are involved in disease progression, and among them, CD105-positive cells are strongly associated with poor prognosis.
The prognosis of advanced gastric cancer is still poor (1, 2) and identifying the mechanism of cancer progression remains challenging (2). It is well known that interactions between carcinoma cells and the tumor stroma are crucial for tumor growth and progression (3, 4). Among cancer stromal cells, research has largely focused on cancer-associated fibroblasts (CAFs); however, their study is hampered due to the lack of specific markers to identify CAFs (3). CAFs comprise a heterogeneous population of cells with various origins, including resident fibroblasts, fibrocytes, pericytes, endothelial cells, smooth muscle cells, adipocytes, epithelial cells, bone marrow-derived cell and mesenchymal stem cells (MSCs) (3-5). The latter MSCs seem to play a critical role in cancer progression (6).
Multipotent MSCs are spindle-shaped and are found at perivascular sites in various human tissues and organs (7, 8). MSCs are introduced to a tumor microenvironment by soluble factors released from inflammatory cells or tumor cells (9). MSCs, as defined by the International Society for Cellular Therapy (ISCT), express CD73 (Ecto-5’-nucleotidase), CD90 (Thy-1) and CD105 (Endoglin) (10). Previous cancer studies used MSCs mainly isolated from bone marrow or umbilical cord blood (11-14). MSC-like cells isolated from human gastric cancer stroma (GC-MSCs) were found to possess similar characteristics to bone marrow-derived MSCs (BM-MSCs), including spindle-shaped morphology and MSC immunophenotype (15). However, the two cell populations differed in the number of cytoplasmic organelles and the higher growth ability and long-term proliferation of GC-MSCs compared to BM-MSCs (16). In gastric cancer, MSCs that come in contact with premalignant or malignant epithelial cells enhance growth and promote epithelial to mesenchymal transition, thus promoting cancer development and progression (6); whereas gastric cancer cells can also trigger MSCs to differentiate into CAFs (17).
To our knowledge, this is the first study to investigate the association of the expression of MSC markers with clinicopathological factors in a large number of gastric cancer cases. The expression of CD73, CD90 or CD105 in spindle-shaped gastric cancer stromal cells was assessed, and their association with cancer progression was examined.
Patients and Methods
Tissue arrays. A total of 546 primary gastric cancers from 524 patients surgically resected from 2000 to 2011 were obtained from the archives of the Department of Pathology, Teikyo University Hospital, Tokyo, Japan. Some of the gastric cancer cases were included in our previous report (18). This study received approval from the Ethics Committee of the Teikyo University School of Medicine (No. 13-021, 8 May 2013). Formalin-fixed, paraffin-embedded (FFPE) tissue blocks were used for staining. Histopathological data were evaluated according to the Japanese classification of gastric cancer and the Lauren classification. The clinicopathological data of the patients and tumors are summarized in Table I. Cylindrical tissue cores of 2 mm from the primary FFPE blocks were arrayed by using the manual device KIN-1 (Azumaya, Tokyo, Japan). Two representative cores were taken from each tissue block. The resulting tissue array blocks were then sectioned at 4 μm in thickness and placed on slides. Cancers occupying more than half of the core area were evaluated by immunohistochemistry.
Immunohistochemical staining. Paraffin sections were dewaxed in xylene and rehydrated through graded ethanol to distilled water. Tissue arrays were stained with rabbit polyclonal antibodies against human CD73 (HPA017357, 1:5000, Atlas Antibodies, Stockholm, Sweden), human CD90 (HPA003733, 1:100, Atlas Antibodies, Stockholm, Sweden) and human CD105 (HPA011862, 1:500, Sigma-Aldrich, St. Louis, USA) using the EnVision FLEX system (Agilent, Santa Clara, USA). Antigen retrieval was performed via water bath heating for 40 min in retrieval buffers (pH9 for CD73 and CD105, pH6 for CD90). Positive staining was scored only in spindle cells. Since endothelial cells are also spindle-shaped, spindle cells forming a lumen structure were excluded from scoring (19). Scoring scale was as follows (Figure 1): for CD73, according to the positivity rate in a high power field; 0 [0-10%], 1 [10-50%], 2 [50-100%]. For CD90 and CD105, 0 [absent], 1 [positive in small aggregates], 2 [positive in diffuse fashion], 3 [highly positive area in more than half of core field]. Scoring was performed by two independent pathologists. In case of discordance, a consensus was reached through discussion. After scoring, gastric cancer cases were divided into negative- (score 0) and positive- (score ≥1) expression groups for CD73, CD90 and CD105. Epstein-Barr virus (EBV) infection was evaluated by EBER in situ hybridization as previously described (18).
Statistical analyses. Chi-square test was used to determine whether there was significant association between MSC marker expression and clinicopathological features. The t-test was used for comparisons between two samples from the same population after stratification. Survival curves were estimated by the Kaplan–Meier method and compared by the log-rank test. Multivariate analyses were performed using Cox proportional hazards model. For the latter, scoring data of miR-21 expression and EBER in situ hybridization from our previous report were used (18). All p-values of less than 0.05 were regarded as significant.
Results
Expression of MSC markers in spindle-shaped gastric cancer stroma cells. All markers showed membranous staining, while CD73 and CD90 were also found in the cytoplasm. The expression of CD73, CD90 and CD105 was detected in mononuclear cells and, in some cases, in vascular endothelial cells, and spindle-shaped cells; CD73 was also expressed in cancer cells. In gastric cancer stroma, 190 (34.8%) cases were positive for CD73, 333 (61.0%) cases for CD90 and 61 (11.2%) cases for CD105 (Table II). In addition, there was a positive correlation among the expression of CD73, CD90 and CD105 (p<0.0001 for CD73 and CD90, p=0.0002 for CD73 and CD105, and p<0.0001 for CD90 and CD105) (Table III).
Patients and gastric cancer cases characteristics.
Distribution of immunostaining scores in spindle cells.
Correlation between the expressions of the examined MSC markers.
MSC marker expression was related to disease progression. The association of each MSC marker with clinicopathological features was examined (Table IV). The expression of all markers was positively correlated with larger tumor size (p=0.0001 for CD73, p<0.0001 for CD90, p=0.0009 for CD105), advanced cancer (p<0.0001 for CD73, CD90 and CD105), venous infiltration (p<0.0001 for CD73 and CD90, p=0.001 for CD105), lymphatic infiltration (p<0.0001 for CD73, CD90 and CD105) and lymph node metastasis (p<0.0001 for CD73 and CD90, p=0.0002 for CD105). Concerning CD73 and CD90 markers, positivity was also associated with diffuse type in Lauren's classification (p=0.009 for CD73, p=0.0002 for CD90). These results suggest that the expression of the examined MSC markers in gastric cancer stroma is associated with disease progression.
Immunohistochemical assessment of the expression of each marker in spindle cells of gastric cancer stroma. Representative cases are shown; CD73 (A: score 0-2), CD90 (B: score 0-3), CD105 (C: score 0-2). Scale bar=100 μm.
Clinicopathological features and MSC markers.
MSC markers and prognosis. The expression of CD73, CD90 and CD105 was associated with poorer prognosis (p<0.0001 for CD73 and CD105, p=0.0001 for CD90) (Figure 2). A multivariate analysis was performed that included CD73-positive spindle cells, CD90-positive spindle cells, CD105-positive spindle cells, stage, venous infiltration, lymphatic infiltration, lymph node metastasis, and miR-21 expression in cancer stroma (previous data (18)) and EBV status (previous data (18)). CD105-positive spindle cells were strongly associated with poor prognosis (Adjusted risk ratio=2.407, 95% CI=1.463-3.961, p=0.0005) (Table V).
Discussion
Previous studies indicated that the interaction between cancer cells and MSCs is involved in the epithelial to mesenchymal transition and in the differentiation of MSCs to CAFs, as well (6, 17). Hence, this interaction is a critical event in gastric cancer progression. To investigate the possible association of MSC features with gastric cancer progression, the correlation between the expression of three MSC markers, CD73, CD90 and CD105, in spindle-shaped stromal cells, and clinicopathological features was examined in a large number of cases. We found that the expression of these markers was associated with larger tumor size, advanced stage, venous infiltration, lymphatic infiltration and lymph node metastasis. These data suggested the expression of MSC markers was associated with disease progression.
There are several reports regarding the relationship between the tested MSC markers and cancer. Contrary to our results, CD73 expression in cancer stroma was associated with better prognosis in rectal cancer (20) and prostate cancer (21). This could be attributed to the possibility that the stromal component included round immune cells in these reports. However, this discrepancy needs to be further explored. CD90 can be expressed in the cancer stroma of prostate (22), lung (23) and gastric cancer (24). In lung cancer, it was that suggested CD90 is a CAF marker (23). CD90-positive CAFs affect the prostate cancer stem cell niche (22) and promote lung cancer cell proliferation and dissemination (25). Moreover, in vivo studies in mice and in vitro experiments in human prostate-derived CAFs have demonstrated that CD105 expression was also involved in prostate cancer progression (26). Authors suggested that CD105-positive fibroblasts were derived from myofibroblasts rather than from resident fibroblasts (26).
A recent report, based on human clinical samples, showed that high percentage of triple-positive (CD73+/CD90+/CD105+) stromal cells from high-grade glioma tissue (Glioma-associated MSCs) were correlated with poor prognosis (27). Herein, co-expression of CD73, CD90, and CD105 was not examined, but a positive correlation among them was revealed. According to multivariate analysis, among the three MSC markers, CD105 was the most important prognostic factor of overall survival of gastric cancer. This predominance of CD105 over other MSC markers was also suggested in multipotency, one of the essential traits of MSCs, where CD73−/CD105+ dermal fibroblasts showed differentiation into both adipocytes and osteocytes, whereas CD73−/CD105− dermal fibroblasts did not (28). This difference among markers could be partly explained from the close relationship between CD105 and TGF-beta signaling. CD105 is a co-receptor of TGF-beta, which directly modulates signaling (29). This process is necessary for angiogenesis and vasculogenesis in endothelial cells or immature mesenchymal cells, as well as for fibrosis in fibroblasts in benign and malignant states (30). In the human stomach, TGF-beta 1 was expressed in epithelial cells and cancer cells closely associated with CD105-positive neo-vessels (31). High density of CD105-positive micro-vessels (neo-vessels) was associated with poorer prognosis in several types of cancer (29).
Overall survival and MSC marker expression. In the examined gastric cancer cases, cells expressing the marker CD73 (A), CD90 (B), or CD105 (C) (Positive group) were significantly associated with poorer overall survival (p<0.0001, p=0.0001, and p<0.0001, respectively), compared to the cells that did not express these markers (Negative group).
Prognostic factors for patient survival in univariate and multivariate analyses.
Since endothelial cells are spindle-shaped (3), we defined “spindle cells” as cells with a spindle-shaped morphology without lumen formation (19). However, there is a possibility that “spindle cells” included the endothelium of collapsed or immature vessels with obscured lumens. On the other hand, it is also possible that aggregates of CD105-positive spindle cells were immature mesenchymal cells associated with vasculogenesis.
Our study showed that CD105-positive spindle stromal cells are strongly associated with disease progression and poor prognosis of gastric cancer, suggesting that these cells may represent a promising therapeutic target for gastric cancer.
Acknowledgements
This work was supported by our host institutions. The Authors declare no conflicts of interest.
- Received October 18, 2018.
- Revision received November 14, 2018.
- Accepted November 20, 2018.
- Copyright© 2019, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
