Prognostic role of Gli1 expression in breast cancer: a meta-analysis

Glioma-associated oncogene 1 (Gli1) is a critical transcriptional factor of Sonic hedgehog pathway which has been proved to participate in the initiation and progression of tumor in mammalians. However, its clinical value in breast cancer remains unknown. Thus, a meta-analysis was performed to clarify the association of Gli1 over-expression, clinic-pathological characteristics, molecular subtypes and prognosis in breast cancer. According to included criteria, 13 eligible studies containing 2816 patients all around the world were selected in this study. Our results indicated no significant association of Gli1 expression and histological grade (RR = 1.20, 95% CI: [0.98, 1.47]), T stage (RR = 1.05, 95% CI: [0.87, 1.27]), clinical stage (RR = 1.04, 95% CI: [0.93, 1.18]) and lymph node metastasis (RR = 1.12, 95% CI: [0.92, 1.37]). In addition, pooled RR showed no correlation of Gli1 expression and progesterone receptor (PR) (RR = 0.92, 95% CI: [0.70, 1.21]), estrogen receptor (ER) (RR = 1.03, 95% CI: [0.74, 1.42]), human epidermal growth factor receptor 2 (HER-2) (RR = 1.12, 95% CI: [0.90, 1.39]). Nonetheless, up-regulated Gli1 expression predicts shorter disease-free survival (DFS) (HR = 1.38, 95% CI: [1.05, 1.81]), 3-year survival (HR = 1.74, 95% CI: [1.28, 2.36]), 5-year survival (HR = 2.04, 95% CI: [1.62, 2.57]) and overall survival (OS) (HR = 2.05, 95% CI: [1.60, 2.64]). In conclusion, over-expression of Gli1 tends to progressive stages and is related to unfavorable prognosis of breast cancer, which may become a potential prognosis indicator and therapy target in breast cancer.


INTRODUCTION
Worldwide, morbidity of breast cancer has been increased to 1.7 million in 2012. For females, breast cancer has the highest mortality rate among all cancer types, accounting for 15% of all cancer deaths [1]. In China, its age-standardized incidence rate has been increasing significantly in the last decade [2]. The general therapeutic strategies are based on locoregional tumor load, molecular subtype and patients' preferences in early breast cancer [3]. In metastatic breast cancer, principles of systemic therapy mainly include chemotherapy, targeted therapy and endocrine therapy [4]. Breast cancer biology plays an important role in the selection of therapeutic plan [5]. Moreover, progesterone receptor (PR), estrogen receptor (ER) and human epidermal growth factor receptor 2 (HER-2) are the only clinically relevant biomarkers and verified therapeutic targets in metastatic breast cancer. However, potential heterogeneity between primary tumor and metastasis, even between metastases, find a potential therapeutic target becomes particularly important, especially in triple-negative metastatic breast

Meta-Analysis
Oncotarget 81089 www.impactjournals.com/oncotarget cancer [6,7] . Therefore, more effective therapeutic strategies of breast cancer underlie a better understanding of novel molecular targets and signaling pathways that are closely related to the clinic-pathological prognostic factors of breast cancer.
Sonic hedgehog (Shh) signaling pathway, one of the components of the hedgehog (hh) pathway, was originally known to play an important role in embryonic development, cell maturation including differentiation, proliferation and maintenance of tissue polarity [7][8][9]. Recent studies have demonstrated that the SHH pathway was also involved in the invasion and metastasis process of solid tumor via its interaction with cancer stem cells (CSC) [10,11] This signaling pathway is initiated with the secretion of Shh glycoprotein, activating the transmembrane protein Patched 1 (PTCH1) by binding with it. The activation of the PTCH1 relieves the inhibition of the Smoothened (Smo), thereby leading to the activation of Glioma-associated oncogene 1 transcription factors [7]. As the final effective factor of the Shh pathway, Gli1, which is considered as a valuable maker of Shh pathway activation, has been reported to be involved in cell selfrenewal, cell proliferation, survival, invasion, angiogenesis and epithelial-mesenchymal transition through regulating expressions of certain genes [12,13]. Thus, abnormal cell development and even oncogenesis tend to happen when Gli1 is dysregulated. Several types of carcinoma have been reported to have aberrant activation of Gli1, including hepatocellular carcinoma, gastric cancer, lung cancer, breast cancer and basal cell carcinoma, indicating the dysregulation of Gli1 may contribute to malignant biological behavior [14][15][16]. Clinical studies have also demonstrated that the expression of Gli1 can be utilized as a potential maker and imply poor prognosis in lung squamous cell carcinoma, basal cell carcinoma, head and neck squamous carcinoma which indicting the potential prognostic value of Gli1 [17,18]. However, the functional and prognostic significance of Gli1 in breast cancer still remains unclear. Some studies have indicated that overexpression of Gli1 predicted poor outcome of breast cancer with higher tumor stage and increased number of tumor-positive axillar lymph nodes [19,20]. While other study showed no significant correlation between expression level of Gli1 and cancer-specific survival in ERα-positive breast cancer [20].
Hence, to clarify the association of Gli1 overexpression and clinicopathological features, molecular subtypes, and clinical outcomes in breast cancer, a metaanalysis was performed via acquired available published data.

Literature selection and characteristics
We recruited 706 eligible studies from Pubmed/ MEDLINE (n = 387) and EMBASE (n = 319) according to the literature retrieval method mentioned above. Two independent investigators (Bilan Wang and Ting Yu) went through title and abstract of these articles and excluded 535 unrelated citations. Meanwhile, 134 studies were removed from scope because of duplicate data. The remaining 37 candidate studies were reviewed carefully by full text. Among them 12 were conference abstracts and 10 were not exploitable for survival data or clinicpathological data associated with Gli1. In addition, 1 non-English (Chinese) and 1 review article were excluded as well. Ultimately, our study included 13 articles for further data extraction and analysis ( Figure 1). The 13 included studies [8,19,20,[23][24][25][26][27][28][29][30][31][32] were published between 2009 and 2016. A total number of 2816 patients from Australia, America, Sweden, China, Korea and Germany were investigated and sample size varied from 83 to 334. All studies performed immunohistochemistry (IHC) to evaluate the expression of Gli1. In data analysis, 6 studies [8,19,25,29,31,32] provided information for overall survival (OS), 4 studies [8,25,31,32] provided disease free survival (DFS), 1 study [20] provided cause specific survival (CSS), 1 study [23] provided distal metastasis free survival (DMFS), 1 study [27] for regression free survival (RFS), 1 study [28] for event free survival (EFS). Detailed Characteristics were demonstrated in Table1, and main outcomes presented in this meta-analysis were summarized in Table 2.

Gli1 expression and immunohistochemical parameters
Considering breast cancer is a typical hormonerelated cancer, we performed pooled analysis on correlations between Gli1 expression and three specific immunohistochemical parameters (PR, ER, HER-2) in breast cancer patients (Figure 3). Our results reported there was no significant association between PR [19,26,27

Gli1 expression and breast cancer survival outcome
Survival outcomes of breast cancer with high and low Gli1 expression from 6 studies were extracted and analyzed ( Figure 4). Result showed Gli1 over-expression was correlated with shorter DFS in breast cancer patients

DISCUSSION
Gli1 overexpression is revealed to have a close connection with breast cancer as a significant maker of aberrant activation of SHH pathway [33][34][35]. It involves in the formation and progression of breast cancer in many important ways such as activating tumor associated target genes [36,37], promoting mammary epithelial cell mesenchymal transition (EMT) [38], regulating mammary CSC self-renewal [10] and facilitating angiogenesis [39]. In    Oncotarget 81094 www.impactjournals.com/oncotarget addition, researchers have confirmed that inhibiting the Gli1 expression could effectively attenuate tumor growth and migration and indicated its potential role as a therapeutic target in breast cancer management [40,41]. However, its clinical value as a prognostic marker remains unclear due to differences in detective methods of Gli1 and selected patients, etc among researches. Therefore we carried out a meta-analysis based on available evidences in order to investigate Gli1 expression with survival and several clinicpathological charactereristics in breast cancer patients.
Our study was the first one to summarize researches of prognostic role of Gli1 in breast cancer. The result indicated no significant association between Gli1 expression and several clinic-pathological characteristics such as histological grade, T stage, clinical stage and lymph node metastasis. Moreover, we reported PR, ER and HER-2 were not correlated with Gli1 expression in breast cancer. But the relationship between Gli1 over-expression and survival in breast cancer patients might be influenced by the clinicpathological characteristics. Nevertheless, it was remarkable that we confirmed that breast cancer patients with overexpression of Gli1 tended to obtain a worse survival outcome referring to DFS, 3-year survival, 5-year survival and OS.
Diao's study [23] clarified that Gli1 would be a potential therapeutic target, moreover, could also act as a prognostic marker in breast cancer. The research observed that high Gli1 expression was associated with poor distant metastasis free survival (DMFS) in 126 patients (HR = 4.87, 95% CI: [1.34, 17.67]). So the clinical staff should pay more attention to patients with breast cancer with high Gli1 expression in systemic screening, such as increasing the frequency of bone scan. Li's study [27] showed that Gli1 expression is significantly correlated with aggressive features and unfavorable recurrence free survival (RFS). The breast cancer with nuclear Gli1 overexpression signified early relapse after radical operation, therefore mammary gland color ultrasound is particularly necessary in patient reexamination. Moreover, summarized DFS in our study also supported that the regular rechecks are more necessary in postoperative patients with Gli1 over expression. What's more, our results reported no significant association between Gli1 expression and hormone receptors expression, which is controversial with Makoto Kubo and Sun reported that Gli1 expression is positively correlated with ER expression [42,43]. Further researches are required to clarify relationship between Gli1 and hormone receptors expression.
There still exist some limitations in our study, which deserve attention. Firstly, definitions of Gli1 overexpression, detection of subcellular Gli1 localization and antibody source varied in different studies. Secondly, only part of recruited patients has undergone surgery which may contribute to potential bias. Thirdly, for the reason that the number of eligible studies was limited, further detailed analysis of correlation between Gli1 overexpression and clinicopathological features hadn't been carried out. In addition, the heterogeneities across articles cannot be ignored.
Our meta-analysis is the first one to explain Gli1 expression as an aggressive biological behavior in breast cancer patients. It integrated convincing evidence to elucidate relationship between Gli1 expression and prognosis of breast cancer. The over expression of Gli1 could predict poor survival in patients with breast cancer. Moreover, detection of Gli1 can provid more convincing evidences for guiding the diagnosis and treatment in breast cancer patients.

Literature retrieval
We retrieved articles from PubMed/Medline and EMBASE electronic databases by variablely combining the following terms "Glioma associated oncogene 1", "Gli1", "Hh", "breast cancer", "breast carcinoma". The search ended on 10th March, 2017. Meanwhile, the references for articles and reviews were also screened to identify additional relevant articles. This meta-analysis included publications meeting the following criteria: (1) patients had a diagnosis of breast cancer; (2) expression level of Gli1 were determined with method such as IHC or reverse transcription-polymerase chain reaction (RT-PCR); (3) the studies provided direct hazard ratio (HR) and 95% confidence interval (CI) for survival or Kaplan-Meier survival curves at different Gli1 expression level; (4) the studies reported Gli1 expression according to different clinic-pathological parameters of patients. When there is possible duplication of data in several publications, we prefer the most recent or the most integrated study. Studies of letters, case reports, conference abstracts, editorials and reviews without original data; non-English papers; animal or laboratory studies were not in our scope.

Data extraction
Two investigators reviewed eligible studies to obtain the following information: first author's name, year of publication, source of patients, sample size, mean age of patients, assay method, cut-off definition, TNM stage, histological grade, immunohistochemical features (ER, PR, HER-2), Gli1 expression rate and survival data according to Gli1 expression. Disagreements were discussed by the two investigators and settled by consulting a third author to reach consensus. Moreover, quality assessments of all studies included were conducted by Newcastle-Ottawa-Scale (NOS) criteria. High quality studies refer to those scored 5 or above 5.

Statistical analysis
RevMan 5.0 software (Copenhagen, Denmark) was applied to analyze extracted data and calculate RR.
To investigate the impact of Gli1 expression on survival