Meta-analysis of the likelihood of FOXC1 expression in early- and late-stage tumors

Background Aberrations in the expression of the transcription factor forkhead box C1 (FOXC1) have been linked to a number of malignancies. Here, we characterized the relationship between FOXC1 and cancer progression by conducting a meta-analysis of studies that reported the frequency of FOXC1 expression in tumors of different stages (T1, T2, T3, T4). Materials and Method Relevant articles were retrieved from the Medline database by searching for the terms “FOXC1” and “cancer”; then, the retrieved articles were reviewed individually, and studies that were of multivariate cohort design, evaluated FOXC1 expression via immunohistochemical staining, and assessed the relationship between FOXC1 expression and cancer T-stage were included in our meta-analysis. Results Our search terms identified 128 studies, 5 of which met all inclusion criteria. A total of 850 tumor samples were evaluated in the 5 studies; 452 samples were from early-stage (T1-T2) tumors, and 398 were from late-stage (T3-T4) tumors. FOXC1 was expressed in 60.7% (516/850) of all samples, in 54.6% (247/452) of early-stage tumor samples, and in 67.5% (269/398) of late-stage tumor samples. When calculated relative to early-stage samples, the pooled risk for FOXC1 expression in late-stage samples was 1.238 (95% CI = 1.061–1.444, p = 0.007). Conclusions The results from our meta-analysis of 5 studies indicate that FOXC1 is 23.8% more likely to be expressed in late-stage tumors than in early-stage tumors.


INTRODUCTION
Cancer can arise via the accumulation of single or multiple genetic mutations, which cause the cancer cells to proliferate without restriction: however, the molecular and genetic cascades involved in tumor formation and cancer progression are largely unknown. The forkhead box (FOX) family of transcription factors includes 17 subfamilies, from FOXA to FOXR, that control a wide range of biological processes such as cell growth, proliferation, differentiation, and longevity [1].
The goal of this study was to characterize the relationship between FOXC1 expression and cancer progression by conducting a meta-analysis of studies that reported the frequency of FOXC1 expression in tumors of Meta-Analysis www.oncotarget.com different stages (T1, T2, T3, T4) [13], and then calculating the pooled relative risk of FOXC1 expression in stage T1-T2 (early) and in stage T3-T4 (late) tumors. We identified 5 reports that met all inclusion criteria and evaluated a total of 850 samples from a wide range of cancer types. Our results suggest that the frequency of FOXC1 expression is significantly higher in late-stage than in early-stage tumors.

MATERIALS AND METHODS
Our meta-analysis was conducted and reported according to PRISMA for Network Meta-Analyses (PRISMA-NMA) checklist [14]. We used a single database (Medline), which is consistent with the PRISMA-NMA requirements. Relevant studies were retrieved by using the PubMed interface to search for the terms "FOXC1" and "cancer," and the studies included in our metaanalysis were of multivariate cohort design, evaluated FOXC1 expression via immunohistochemical staining, and assessed the relationship between FOXC1 expression and cancer T-stage (T1, T2, T3, T4). The meta-analysis was performed with an open-source program as described previously [15][16][17][18].

Study selection
The initial Medline search was conducted on September 1st, 2018 and identified 128 articles (Supplementary Table 1); 106 of the studies were excluded because they were not a multivariant cohort studies. Seventeen of the 22 multivariant studies were excluded either because they did not investigate the relationship between FOXC1 expression and cancer, did not categorize their results by tumor stage, or did not evaluate FOXC1 expression immunohistochemically ( Figure 1). The remaining 5 studies (Table 1) were multivariate analyses and met all inclusion criteria.

Risk of bias in individual studies
Each individual study had a potential selection bias as the samples for each study were not randomly selected. Furthermore, the immunohistochemical methods used to detect FOXC1 expression varied across the studies, and this variation could also induce bias. The authors evaluated 121 tumor samples from patients with salivary adenoid cystic carcinoma; 48 samples were categorized as stage T1-T2 and 62 samples were categorized as stage T3-T4. The proportion of FOXC1-positive tumors in each group was 50% (24/48) and 53% (33/62), respectively (p-value = 0.737) ( Figure 2). For our meta-analysis, "high" expression levels were considered positive for FOXC1 and "low" expression levels for FOXC1 were considered negative.

Exploration for inconsistency and risk of bias across studies
Differences in the immunohistochemical protocols and techniques used for FOXC1 staining could produce some inconsistencies in FOXC1 detection. Biases that may affect the cumulative evidence include sample selection, because samples were not chosen randomly in all studies, and our inclusion of multivariate cohort studies which, because they evaluate multiple parameters simultaneously, could increase the heterogeneity of our results. We attempted to minimize heterogeneity by limiting study eligibility. We only selected the multivariate cohort design which evaluated FOXC1 expression via immunohistochemical staining. The I 2 of the meta-analysis study has shown a mild heterogeneity with 24.47% with a p-value = 0.258 (Table 3).

DISCUSSION
Accumulating evidence indicates that FOXC1 is involved in tumor development and metastasis. In particular, FOXC1 is a prognostic biomarker for BLBC [24,26,27], which is a form of triple-negative breast cancer for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). Elevated FOXC1 mRNA expression is associated with a worse overall survival of breast cancer patients [27], as well as with brain and lung metastasis of breast cancer [26,27]. Therefore, the function of FOXC1 in breast cancer, specifically BLBC, has been extensively investigated. FOXC1 plays a critical role in proliferation, migration, invasion, and epithelial-to-mesenchymal transition (EMT) of breast cancer cells through regulation of EGFR, NF-kB, and MMP7 [28][29][30], as well as in control of breast cancer stem cell properties by directly interacting with the Gli2 transcription factor [31].
Most of the studies chosen in our meta-analysis suggested that FOXC1 expression and cancer were associated, but none of them reported the relative risk between tumor stage and FOXC1 levels. Our metaanalysis compared the expression of FOXC1 in stage T1-T2 and stage T3-T4 tumors, and we reduced  selection bias by restricting our analysis to multivariate cohort studies that detected FOXC1 expression via immunohistochemical staining and based the T-stage definition on tumor size, which does not predict morbidity but is a definitive indicator of tumor growth. Our results indicate that FOXC1 expression is significantly more common in late-stage (T3-T4) tumors than in early stage (T1-T2) tumors and, consequently, that FOXC1 may be a marker for the T-stage of cancer.

ACKNOWLEDGMENTS AND FUNDING
This work was supported by the National Institutes of Health (NIH) (HL126920, EY028304, and HL144129 to T. K.). The authors thank W. Kevin Meisner, PhD, ELS, for editorial assistance.

CONFLICTS OF INTEREST
We certify that there is no conflict of interest with any financial organization regarding the material discussed in the manuscript.