Galectin 3 expression in primary oral squamous cell carcinomas

A consecutively treated oscc patient collective was selected for this retrospective study. The patient selection was described in previous reports [3, 24]. Biopsies and tumor resection specimens from a total of 34 patients histologically diagnosed with primary oscc were analyzed in this study. Tumor resection specimens from 34 patients and biopsies from 26 patients were available. The biopsies from nine cases were not available for analysis or too small for analysis. Because we counted macrophage infiltration in at least 1.1 mm² of carcinoma tissue, specimens with smaller carcinoma fractions were excluded. All patients were treated in 2011 at the Department of Oral and Maxillofacial Surgery of the University Hospital Erlangen. The study protocol was approved by the ethical committee of the University of Erlangen-Nuremberg (Ref.-No. 45_12 Bc). The specimens used in this study were obtained from tissue samples collected for routine histopathologic diagnosis. Each included specimen was judged to be a representative squamous cell carcinoma. In addition to the diagnosis of oscc, the following inclusion criteria were defined: pT1 or pT2 tumors, no restrictions in the grading of the tumor, no adjuvant preoperative radio- or chemotherapy and no distant metastasis at the time of diagnosis. Patients with former radio- or chemotherapy and pT3 and pT4 tumors were excluded. No study-related changes in the patients’ treatments took place.

The immunohistochemical staining procedure was performed as previously described [3, 24]. The following primary antibodies were used: anti-Galectin 3 (sc-20,157, clone H-160, Santa Cruz, Dallas, Texas, USA) and anti-CD68 (11,081,401, clone KP1, Dako, Hamburg, Germany). An appropriate positive control was included in each series.

The tumor and biopsy sections were completely scanned and digitized using the method of “whole slide imaging”. The scanning procedure was performed in cooperation with the Institute of Pathology of the University of Erlangen-Nürnberg using a Panoramic 250 Flash III Scanner (3D Histech, Budapest, Hungary) in 40× magnification. All samples were digitally analyzed (Case viewer, 3D Histech, Budapest, Hungary). Quality controls were performed using a bright-field microscope (Zeiss Axioskop and Axiocam 5, at 100–400 × magnification).

For each sample, three visual fields showing the highest infiltration rate of positive cells were selected (hot spot analysis). The complete area of all three visual fields of one specimen was between 1.1 and 1.5 mm² (Case viewer, 3D Histech, Budapest, Hungary).

Micrographs of the selected areas were imported into the Biomas analysis software (modular systems of applied biology, Erlangen, Germany) for cell counting. Two regions of interest were defined in the visual fields using the Biomas software: the epithelial tumor compartment and the stromal compartment.

A quantitative analysis was performed to determine the numbers of Galectin 3- and CD68-positive cells in the epithelial tumor compartment and the surrounding stroma. A Gal3 background staining, as it was visible in the epithelial tumor compartment of all cases, was not counted. Cells with strong Gal3 expression significantly surpassing the background expression of epithelial oscc tumor cells were counted as positive. Assessment of the cell density per mm² was performed as previously described [16, 24].

To analyze the immunohistochemical staining, the cell count per mm² was determined as the number of positive cells per mm² of the specimen. Multiple measurements were pooled for each sample group prior to analysis. The results are expressed as the median and standard deviation (SD) and range. Box plot diagrams represent the median, the interquartile range, minimum (Min) and maximum (Max).

Galectin 3 (Gal3) was expressed in all investigated specimens. An expression of Gal3 could be observed in all cellular compartments (Fig. 1a). Oscc tumor cells in all patients showed a low level of Gal3 expression. For statistical analysis, the cells with a Gal3 expression significantly surpassing the background Gal3 staining of epithelial ossc tumor cells were counted. These cells could be found in the epithelial tumor compartment as well as in the tumor stroma. The distribution pattern of these Gal3 highly positive cells was similar to the distribution of CD68 positive macrophages (Fig. 1b). The number of CD68 positive macrophages exceeded the number of Gal3 highly positive cells (except in G3 oscc cases).

Assessing the epithelial compartment of biopsy specimens, the Galectin 3 (Gal3) cell count in T2 oscc cases was significantly higher than in T1 cases (median 256 cells/mm² and 103 cells/mm², respectively, p = 0.032) (Table 1, Fig. 2a). Comparably, Gal3 expression in the whole analyzed specimen area (epithelial + stroma) of biopsies was significantly higher in T2 cases compared to T1 cases (median 293 cells/mm² and 101 cells/mm², respectively) (p = 0.024) (Table 1, Fig. 2b).

The ratio between Gal3-expressing cells and CD68-positive macrophages (Gal3/CD68-ratio) in the epithelial fraction of T2 oscc biopsy samples was significantly higher than in biopsies of T1 cases (median values 0.73 and 0.43, respectively, p = 0.045) (Table 1, Fig. 2c). The whole analyzed specimen area (epithelial + stroma) in T2 oscc biopsy specimens revealed a significantly higher Gal3/CD68-ratio than in T1 cases (median value 0.71 and 0.19, respectively, p = 0.018) (Table 1, Fig. 2d).

Oscc biopsy specimens in cases with lymph vessel infiltration (L1) showed a significantly higher Gal3 expression compared to L0 cases assessing the whole analyzed specimen area (epithelial + stroma) (median 369 cells/mm² and 110 cells/mm², respectively, p = 0.013) (Table 1, Fig. 2e).

The Gal3/CD68-ratio in the whole analyzed specimen area (epithelial + stroma) of biopsies from L1 oscc cases was significantly higher than in L0 cases (median value 0.95 and 0.31, respectively, p = 0.008) (Table 1, Fig. 2f). In the epithelial compartment, the Gal3/CD68-ratio in L1 biopsy specimens was also significantly higher than in L0 cases (median value 0.88 and 0.55, respectively, p = 0.020) (Table 1).

There was no significant association apparent between Gal3 expression and N-status, Pn-status and tumor grading in oscc biopsy specimens.

Analyzing the epithelial fraction of tumor resection specimens, the Galectin 3 (Gal3) cell count in T2 cases was significantly higher than in T1 cases (median 241 cells/mm² and 97 cells/mm², respectively, p < 0.001) (Table 2, Fig. 3a). In the whole analyzed specimen area (epithelial + stroma) of tumor resection samples the Gal3 expression in T2 cases was also significantly higher than in T1 cases (median 302 cells/mm² and 116 cells/mm², respectively, p = 0.002) (Table 2, Fig. 3b).

The Gal3/CD68 expression ratio in the epithelial fraction of T2 oscc tumor resection specimens was significantly higher than in T1 cases (median value 0.73 and 0.30, respectively, p = 0.002) (Table 2, Fig. 3c). The whole analyzed specimen area (epithelial + stroma) in T2 oscc tumor resection specimens also showed a significantly higher Gal3/CD68-ratio than in T1 cases (median value 0.58 and 0.24, respectively, p = 0.010) (Table 2, Fig. 3d).

The epithelial compartment of oscc tumor resection specimens in cases with lymph node metastases (N+) revealed a significantly higher Gal3 cell count than N0 cases (median 234 cells/mm² and 126 cells/mm², respectively, p = 0.030) (Table 2, Fig. 3e).

The Gal3/CD68-ratio in the epithelial compartment of G3 oscc tumor resection specimens was significantly higher compared to G2 cases (median value 1.31 and 0.52, respectively, p = 0.008) (Table 2, Fig. 3f).

There was no significant association apparent between Gal3 expression and L-status or Pn-status in oscc tumor resection specimens.

The current study revealed an association between Galectin 3 (Gal3) expression in oral squamous cell carcinoma (oscc) tissue and histomorphologic parameters of tumor progression (T-, N-, L-, Pn-stage, grading). In biopsy and tumor resection specimens, the number of Gal3 positive cells as well as the ratio between Gal3 expressing cells and CD68 positive macrophages was higher in T2 oscc cases compared to T1 cases (Figs. 2 and 3). Considering the tumor promoting and immunosuppressive function of Gal3 [23, 25], this finding could implicate that local immunosuppression promotes increasing tumor growth and consequently size. The correlation between tumor size (T-status) and Gal3 expression resp. Gal3/CD68 ratio was seen in the epithelial tumor compartment as well as the whole analyzed specimen area including stroma and epithelium (Figs. 2a-d and 3a-d). In a previous study in the patient cohort analyzed here, no association was detected between T-status and the density of CD68 positive macrophages [3].

An association between Gal3 expression and T-status was seen in biopsies and in tumor resection specimens. As previously reported, there are immunologic changes in tumor tissue occurring during the time interval between the diagnostic biopsy and the definitive tumor resection [24]. The association between T-status and Gal3 expression in biopsies and tumor resection samples observed in the current report (Fig. 2a-d and 3a-d) indicates that Gal3 might be a possible robust biologic parameter that is not relevantly influenced by biopsy-derived tissue trauma.

Additionally, the current analysis showed an association of higher Gal3 expression in tumor resection specimens with the occurrence of lymph node metastases (N+) (Fig. 3e). A high Gal3/CD68 ratio correlated with higher grading (G3) in oscc tumor resection specimens (Fig. 3f). In biopsies, the Gal3 expression and the Gal3/CD68 ratio was associated with the presence of lymph vessel infiltration (L1) (Fig. 2e-f). As the N-status, the tumor grading and the L-status are important parameters of malignant behavior, the association of Gal3 expression with these parameters underlines a possible tumor-promoting role of Gal3 in oscc.

Gal3 can interact with macrophages and modulate macrophage polarization [7, 26]. M2 polarized macrophages show a significantly increased Gal3 expression and secretion [7, 26, 27]. In contrast, induction of M1 polarization in macrophages leads to an inhibition of Gal3 expression compared to unstimulated monocytes [7].

Additionally, Gal3 itself can shift macrophage polarization towards M2 polarized cells. In murine cells, Gal3 deficiency prevented the M2 polarization of macrophages while there was no influence observed on the induction of M1 polarization [7]. Furthermore, M2 polarization can be prevented by the use of small inhibitory RNA (siRNA) for Gal3 or for the Gal3 receptor CD98. CD98 shows strong expression on macrophages and leads to the activation of the second messenger phosphatidylinositol 3-kinase (PI3K) [7]. PI3K activation is a relevant pathway for the induction of M2 macrophage polarization. Hence, Gal3 can induce PI3K activation via CD98 and thereby induce M2 polarization of macrophages [7]. Consequently, there might be a Gal3 triggered positive feedback loop for M2 polarization of macrophages as Gal3 contributes to the induction of M2 polarization and M2 macrophages show an increased Gal3 production [7].

This loop could be targeted by pharmacologic inhibition of Gal3 e.g. with citrus pectin [25] or by blocking CD98 [7]. As we detected an association of Gal3 with parameters of malignancy in oscc, targeting the Gal3 – CD98 – PI3K pathway could be a potential immune modulating treatment option for oral cancer patients.

These findings suggest the possibility of an increased Gal3 production in the context of a malign disease leading to an increased degree of M2 polarization of macrophages. On the other hand, it would be conceivable that macrophage-derived Gal3 itself leads to a promotion of malign transformation and metastatic spread of oscc tumor cells. Besides its influence on macrophage polarization, Gal3 can inhibit T-cell activation by destabilizing the immunologic synapse [28]. Additionally, Gal3 can inhibit CD8 positive cytotoxic T-cells in an LAG-3 dependent manner and thereby contribute to local immune tolerance and tumor promotion [29]. Consequently, Gal3 could contribute to an immunosuppressive local tumor environment at the level of macrophages and T-cells which could be targeted by Gal3 inhibitors like citrus pectin [25].

The current study reveals an association between tumor size (T-status) and parameters of malignancy (L-status, grading) with the Gal3/CD68 ratio. In most cases the detected Gal3/CD68 ratio was smaller than 1, meaning that less Gal3-positive cells than CD68-positive macrophages were present in the specimens (Tables 1 and 2). An association between the absolute cell count of CD68-positive macrophages in the epithelial tumor compartment and the occurrence of lymph node metastases was already published in a previous report of our group [3].

Cells with high Gal3 expression were detected in the epithelial tumor compartment and in the tumor stroma (Fig. 1). This indicates that a relevant proportion of the highly Gal3 expressing cells in oscc specimens are no tumor cells but stroma cells or tumor infiltrating immune cells.

One part of the detected Gal3 expressing cells might be M2 polarized macrophages. An association between M2 polarization of macrophages and high Gal3 expression is already shown [7, 26, 27]. A previous report analyzing cervical squamous cell carcinomas revealed that many intratumoral Gal3 expressing cells are CD163 positive M2 polarized macrophages [27]. However, the Gal 3 positive cell population identified in the current study might also include cancer stem cells [30, 31].