Signal transducer and activator of transcription 3 (STAT3) is involved in the JAK-STAT signaling pathway that mediates a wide variety of cellular responses depending on the tissue type such as differentiation, proliferation, apoptosis and survival [1] . STAT3 is constitutively activated in various human epithelial malignancies, including prostate, breast, lung, head and neck, brain, pancreas and colorectal cancer [2] -[4] as well as haematopoietic malignancies [2] suggesting its importance in oncogenesis. Cellular transformation by activated STAT3 occurs through the transcriptional regulation of specific target genes including those encoding the anti-apoptotic proteins Bcl-xl, Mcl-1, and Bcl-2, the proliferation-associated proteins Cyclin D1 (CCND1) and Myc, and the pro- angiogenic factor VEGF [2] [5] [6] . Activation of STAT3 also contributes to invasive growth of cancer cells and confers resistance to chemotherapy [3] [7] . Elevated levels of STAT3 were significantly associated with poor prognosis [8] [9] and metastasis [3] in colorectal carcinomas (CRCs).

The matrix metalloproteinases (MMPs) family consists of proteolytic enzymes that are involved in the degradation of extracellular matrix components [10] . MMPs are often overexpressed during tumor progression and have been associated with tumor invasion and metastasis [11] . MMP-7, a member of the MMP gene family, has proteolytic activity against a wide range of substrates such as collagens, proteoglycans, elastin, laminin, fibronectin and casein [12] . It has been reported that MMP-7 was overexpressed both in benign and malignant colorectal tumors [13] [14] but the level of expression in colorectal cancer was known to be 50-fold greater than normal mucosa [15] . It has also been reported that MMP-7 was associated in the development and progression of CRC, and correlates with distant and lymph node metastasis [16] [17] .

Vascular endothelial growth factor (VEGF) plays a critical role in the process of angiogenesis by stimulating the proliferation and migration of vascular endothelial cells, and modelling of the tumor neovasculature. Neo- angiogenesis, mediated by angiogenic cytokines, contributes to the growth, progression and survival of tumors [18] . VEGF is a potent growth factor that stimulates a strong angiogenic response by increasing microvascular permeability [19] . The formation of new blood capillaries provides nutrients for rapid growth, and increases the opportunity for tumor cells to metastasize [18] . VEGF expression in CRC was found to be associated with poor prognosis and development of metastatic disease [20] [21] , and could be an independent risk factor for tumor recurrence and might assist in selecting patients who could benefit from adjuvant therapy [22] .

Currently colorectal malignancy is an important disease in Malaysia. The incidence of CRC is rising rapidly and ranked third highest in both males and females among the top ten most common cancers in Malaysia. It accounts for 14.2% of male cancers and 10.1% female cancers [23] . Although a number of etiological factors have been identified, the genetic information on the disease among the Malaysians is unavailable. Though STAT3, pSTAT3, MMP-7 and VEGF have been widely studied, in the present study we select these biomarkers to determine the expression pattern in colorectal adenocarcinomas using immunohistochemistry in a hospital-based Malaysian population. We also evaluate association with the clinico-pathological parameters such as patients’ age, gender, ethnicity, histological grade, Dukes’ stage and nodal status, and co-expression of these biomarkers.

The immunostaining of STAT3 (Figure 1(a)) and pSTAT3 (Figure 1(b)) exhibited heterogeneous staining pattern. Expression of STAT3 was present either in the cytoplasm or nuclear region in colorectal tumor cells; occasional staining was present in both areas. However, pSTAT3 immunoreactivity appeared to be predominant in the nuclear region, whereas the cytoplasmic area showed very weak immunostaining. However, intratumor observation showed some heterogeneous immunoreactivity. MMP-7 (Figure 1(c)) and VEGF (Figure 1(d)) immunopositivity was detected in the cytoplasm of the tumor cells. Nuclear immunoreactivity was not noted in all the stained tumor cells. None of the stromal cells and negative control exhibited immunopositivity.

Overexpression of STAT3 was detected in 26.7% (20/75), pSTAT3 in 13.4% (10/75), MMP-7 in 38.8% (36/93) and VEGF in 59.2% (55/93) of CRCs. Tumors weakly positive for STAT3, pSTAT3, MMP-7 and VEGF were demonstrated in 1.3% (1/75), 1.3% (1/75), 1.1% (1/93) and 5.4% (5/93) CRCs, respectively. Tumors were considered as negative for STAT3 in 72.0% (54/75), pSTAT3 in 85.3% (64/75), MMP-7 in 60.2% (56/93) and VEGF in 35.5% (33/93) of CRCs.

Statistical analysis was performed to evaluate the relationship between STAT3, pSTAT3, MMP-7 or VEGF immunoreactivity and clinico-pathological parameters such as patients’ age, gender, ethnic group, histological grade, Dukes’ stage and nodal status (Table 2 and Table 3). STAT3, MMP-7 and VEGF immunopositivity were significantly correlated with poorly-differentiated tumors (P = 0.004; P = 0.03; P = 0.002, respectively) but not with other parameters. However, pSTAT3 immunostaining was not significantly associated with the clinico-pathological characteristics. Elevated expression of STAT3, MMP-7 and VEGF was demonstrated in 44.8%, 41.7% and 72.2% of the poorly-differentiated tumors, and 30.0%, 56.0% and 72.0% of the moderately-differ- entiated tumors compared to 3.8%, 21.9% and 34.4% of the well-differentiated tumors, correspondingly.

Co-expression between STAT3, pSTAT3, MMP-7 and VEGF genes was also analysed (Table 4). Significant relationship was noted between overexpression of pSTAT3 and STAT3 (P < 0.001), pSTAT3 and VEGF (P = 0.044), pSTAT3 and MMP-7 (P = 0.003), and STAT3 and VEGF (P = 0.037) but marginal association was detected between STAT3 and MMP-7 (P = 0.057), and MMP-7 and VEGF (P = 0.052).

In the present study we have examined the expression of STAT3, pSTAT3, MMP-7 and VEGF protein by immunohistochemistry in a cohort of colorectal adenocarcinomas. There was overexpression of STAT3 in 26.7% of CRCs but smaller proportion (13.4%) of the tumors demonstrated the activated form of STAT3 (pSTAT3). Similarly Hbibi et al. noted that 62% and 27% of 126 CRCs were positive for STAT3 and pSTAT3, respectively [25] . Other investigators reported that the frequencies of overexpressed tumors range from 44% - 78% [4] [8] [9] [26] [27] . Interestingly, Xiong et al. reported 100% immunopositivity for both STAT3 and pSTAT3 protein in 38 cases [28] . Significant correlation was noted between increased expression of STAT3 and the phosphorylated form of STAT3 which is in agreement with findings by Lassman et al. [29] . Interestingly our data showed lower frequency of pSTAT3 overexpression among the CRCs compared to other findings which could be attributed to scoring methods, random selection of patients and types of antibodies used. Our findings suggest that a proportion of the tumors that overexpress STAT3 might harbour higher levels of pSTAT3 which could have important implications in colorectal cancer biology.

Increased expression of STAT3 was significantly associated with high grade tumors but this was not seen for the phosphorylated form of STAT3. Although the association was not supported statistically, higher frequency of poorly-differentiated tumors (24.1%) was positive for pSTAT3 compared to moderately (10.0%) and well (3.8%) differentiated tumors. Overexpression of STAT3 and pSTAT3 was not associated with other clinico-pa- thological parameters such as age, gender, ethnicity, Dukes’ classification and nodal metastases. Contrary to our results, some studies showed significant correlation between pSTAT3 and presence of nodal metastases [4] [9]

^*Represents significance at P < 0.05; Note: S, significant at P < 0.05. NS, not significant.

[25] [27] and advanced Dukes’ stage [8] [27] but not with age, gender and/or tumor grade [4] [9] [27] . In a recent study, Xiong et al. showed that all the well (n = 6), moderately (n = 27) and poorly-differentiated (n = 3) tumors were positive for both the activated and unactivated form of STAT3 protein [28] . Though constrasting results were noted probably due to smaller sample size, our data suggests that upregulation of STAT3 and pSTAT3 in CRC could have an important role in the tumor progression as defined by its association with high grade tumors.

MMP-7 is widely expressed in the epithelial tumor cells of malignant colorectal adenocarcinomas and believed to be associated with metastasis [13] [14] [30] [31] . In the current study we found that MMP-7 was overexpressed in 38.8% of CRCs and this positive rate is comparable with previous reports of 30% to 46% [15] [17] [30] [32] . In a recent study with a larger sample size of 620 cases, Fang et al. reported MMP-7 immunopositivity in 88.8% of the cancer [33] . Roca and co-investigators demonstrated 81.7% of 60 CRC tumors were positive for MMP-7 [34] . In the present study, clinico-pathological analysis demonstrated that the frequency of MMP-7 immunopositivity was significantly higher in moderately- and poorly-differentiated tumors compared to well- differentiated tumors. However, earlier studies did not observe such association [17] [34] . Some of the studies have evaluated expression of MMP-7 in association with clinico-pathological parameters but did not correlate with histological grade [15] [30] [32] [33] . There was no significant association between MMP-7 expression and

^*Represents significance at P < 0.05; Note: S, significant at P < 0.05. NS, not significant.

Note: S, significant at P < 0.05. NS, not significant.

the age, gender, ethnicity, Dukes’ staging and lymph node status of the patients. Roca and coauthors found no significant correlation between MMP-7 and lymph node metastases which is consistent with our findings [34] . In contrast, Adachi and co-investigators reported that elevated levels of MMP-7 was significantly associated with lymph node metastases and advanced Dukes’ stage suggesting its role in tumor metastases [15] [30] . Our findings suggest that overexpression of MMP-7 could be a biological indicator for aggressive characteristics of colorectal cancer.

VEGF has been extensively investigated in relation to the development of novel drugs against CRCs [20] . In the current study, elevated levels of VEGF were demonstrated in 59.2% of CRC. Among the biomarkers evaluated in this study, VEGF immunopositivity was the most frequent in CRC. In a series of studies, overexpression of VEGF was detected in 45% to 70% of CRCs [21] [35] -[42] which is comparable with the present study. However, Khorana et al. reported lower incidence of VEGF immunopositivity (34%) [43] whereas Uner et al. showed higher frequency of 86% in CRCs [44] . In the present study, a significant association was observed between VEGF expression and high histological grade tumors. Although consistent with other published reports [40] [42] , majority of the studies did not confirm the association [21] [35] -[39] [41] [44] [45] . Statistical correlation was not found with Dukes’ stage and lymph node involvement which is in agreement with other reports [37] [44] [46] . However most of the studies have demonstrated significant correlation between VEGF immunoreactivity and nodal metastases [21] [36] [39] [40] [42] [45] or advanced Dukes’ stage [36] . The patients’ characteristics (age at diagnosis, gender and ethnicity) were not significantly associated with VEGF expression and similar observation was also noted by other investigators [21] [35] [36] [38] [39] [41] [42] [45] [46] . Our observation suggests that overexpression of VEGF could be an important marker for aggressive phenotype of colorectal tumors.

Apart from determining the frequencies of the genes overexpressed as well as correlation with clinico-patho- logical parameters we also determined co-overexpression of STAT3, pSTAT3, MMP-7 and VEGF in CRCs. Overexpression of STAT3 was significantly associated with VEGF but showed marginal correlation with MMP-7. However, the activated form of STAT3 was found to be significantly associated with MMP-7 and VEGF. Our data also showed that elevated levels of MMP-7 was marginally associated VEGF. Earlier studies have reported that pSTAT3 promotes tumor angiogenesis by acting as a direct transcriptional activator of VEGF and mediates expression of genes related to tumor growth and metastasis [6] . Recently, Cascio and co-investi- gators supported this evidence in colon cancer cells by demonstrating the binding of STAT3 to the promoter of VEGF [47] . Elevated levels of MMP-7 mRNA was detected in colorectal tumor tissues with activated form of STAT3 [3] . In addition to MMP-7, aberrant activity of STAT3 could be a major factor of local tumor progression and metastasis of colorectal cancer through upregulation of other MMPs including MMP-1, MMP-3 and MMP-9 [3] . This could also explain the reason that we noted pSTAT3 was significantly associated with MMP-7 but STAT3 showed marginal association with MMP-7.

Among the MMPs (MMP-1, MMP-2, MMP-3, MMP-7, MMP-9 and MMP-13), MMP-7 is primarily expressed in cancer cells during tumor progression [48] and implicated in promoting angiogenesis by remodeling the ECM [49] . Ito et al. reported that despite increased expression of VEGF, upregulation of MMP-7 in addition to MMP-1 and MMP-3 may regulate the switch from vascular quiescence to angiogenesis in the human colorectal tissues [50] . Furthermore, high levels of MMP-7, MMP-1 and MMP-3 along with VEGF is associated with increased expression of angiogenesis markers in the CRC suggesting that MMP-7 could be one of the regulators of VEGF activity in human colorectal cancer cells. These observations support our data that demonstrated marginal association between MMP-7 and VEGF.

The possible reasons for the differences between the studies could be attributed to the sample size of the cases used in the investigations, scoring methods including variations in the threshold definitions for overexpression between the laboratories, selection bias with regard to recruitment of patients such as patients’ characteristics, ethnicity, different histopathological types of the tumors and antibodies used. These differences could alter the distribution of the immunopositivity that might influence the statistical power to determine the significance of association.

In our study, we showed an interesting observation that STAT3, MMP-7 and VEGF were consistently associated with high grade tumors. Histological grade or tumor differentiation has been regarded as one of the predictive factors associated with poor prognosis for colorectal carcinomas. Furthermore we have also demonstrated significant correlation between pSTAT3 and VEGF or MMP-7, with marginal association between VEGF and MMP-7.

In conclusion, our data suggest that expression of these genes may have an important role in tumor dedifferentiation and may be useful indicators of biologic aggressiveness. Co-expression of the biomarkers by cancer cells might have important implications in colorectal cancer biology and could be useful biological markers of the malignant phenotype.

This study was financially supported by National Biotechnology Directorate, Medical Biotechnology Cooperative Centre, Ministry of Science, Technology and Innovation, Malaysia.

The authors declare that they have no conflict of interest related to the publication of this manuscript.