Abstract
Background: The Forkhead box transcription factor (FOX) family plays an essential role in embryogenesis, especially during brain development. Our hypothesis is that de-regulation of FOX genes may contribute to aggressive tumor biology and therapy resistance in patients with glioblastoma multiforme (GBM). Materials and Methods: Univariate and multivariate analyses were performed to evaluate prognostic significance of transcript levels of 31 FOX genes in a test set of GBM patients (n=191) and validated them in The Cancer Genome Atlas (TCGA) cohort comprising of 508 adult cases of GBM. The predictive significance of key FOX genes was investigated in patients who received chemotherapy or radiotherapy. Results: Low FOXA2 mRNA, low FOXN2 mRNA, low FOXN3 mRNA and high FOXG1 mRNA were associated with poor survival in the test and TCGA validation cohorts. In multivariate analysis, low FOXA2 mRNA, low FOXN2 mRNA, low FOXN3 mRNA and high FOXG1 mRNA remained independently associated with poor survival in the test and TCGA validation cohorts. In patients who received chemotherapy or radiotherapy, low FOXA2 mRNA, low FOXN2 mRNA and high FOXG1 mRNA correlated with adverse outcomes in the TCGA validation cohort. Conclusion: To our knowledge, our data provide the first comprehensive clinical evidence that FOXA2, FOXN2, FOXN3 and FOXG1 are promising biomarkers of GBM and warrant further investigation.
Despite advances in surgery, chemotherapy and concurrent chemoradiotherapy strategies, the overall prognosis of patients with glioblastoma multiforme (GBM) remains poor, with a 3-year survival of less than 10% (1). Forkhead box transcription factors (FOX) are responsible for regulating the transcription of several proteins involved in embryogenesis, cell proliferation, differentiation, DNA repair and cell survival (2-4). There exist at least 50 known FOX genes in the human genome, categorised into 19 sub-groups (from A to S) (2-4). FOX genes may have a role in cancer pathogenesis. FOXOA1 has been linked to prostate cancer (5). Overexpression of FOXM1 has been identified in cancer of the liver, brain, and pancreas (6). FOXP1 may act as a tumor suppressor in breast cancer and paradoxically as an oncogene in certain types of lymphoma (7). In addition, FOXA1 and FOXG1 may also be involved in gliomagenesis (8, 9). We hypothesized that FOX genes may have a role in GBM pathogenesis.
Materials and Methods
We investigated FOX gene expression in GBMs in two datasets, a test set and a validation set.
Test set. The dataset E-GEOD-13041 (publically available from http://www.ebi.ac.uk/arrayexpress/) was used as a test set. This dataset contained microarray gene profiling data for 267 patients using three different Affymetrix platforms. A total of 191 patients with GBM were included in the subsequent data analysis for the test dataset, all of whom were profiled using the Affymetrix U133A array. The median age of the patients was 54 years, with a range of 18-86 years. One hundred and eighteen of the patients (61.8%) were male. The patients were followed-up for a median of 385 days (range=7-3353 days). At the end of the follow-up, 92.1% of patients had died (176/191). Limited treatment data was available in this dataset.
Validation set. The dataset obtained from The Cancer Genome Atlas (http://cancergenome.nih.gov/) was used as a validation set. The dataset consisted of 548 patients out of which 508 were selected for analysis after duplicates and patients with missing survival data were excluded. The median age in this set was 59 years, with a range of 10-89 years. 60.6% of the patients were male (308/508). Follow-up was undertaken for a median of 353 days (range=2-3880 days). At the end of follow-up, 81.9% of patients had died (416/508). Within this set, 69.1% received chemotherapy (351/508) and 73.4% received radiotherapy (373/508). Of these 65.7% received both chemotherapy and radiotherapy (334/508), 3.1% received just chemotherapy and no radiotherapy (16/508), 7.5% received only radiotherapy and no chemotherapy (38/508) and 23% (116/508) received neither treatment option.
FOX gene probes in test and validation sets.
Statistical analysis. Out of the 50 known human FOX genes, 31 genes, represented by 42 probes, were present in both datasets and included in subsequent analyses (Table I). Along with this expression data, demographic data including: age, gender and survival data were also included. X-Tile (version 3.6.1; Yale University, New Haven, Connecticut, USA) was used to classify gene expression data into high and low expression for the 42 probes. SPSS (IBM SPSS Statistics for Windows, Version 22.0; IBM Corp., Armonk, NY, USA) was used to generate Kaplan-Meier survival curves for each probe in both datasets. The Benjamini-Hochberg false-discovery rate (BH FDR) (10) was then applied to the values to allow for multiple comparisons. Cox multivariate regression models were constructed for each dataset including probes which were significant (with BH FDR correction) in both datasets. Four probes were found to be significant in both datasets. Subgroup analysis based on treatments received (chemotherapy, no chemotherapy, radiotherapy, and no radiotherapy) were also performed in the TCGA dataset.
FOX genes and association with survival in test and validation datasets.
Results
FOX gene expression and survival in adult patients with GBM (Table II). In the test set, 17 probes were found to be significantly associated with survival (p<0.05). In the TCGA validation cohort, 25 probes were significantly associated with survival (p<0.05). Following BH FDR assessments, the number of significant probes (p<0.05) was 11 in the test set and 21 in the TCGA validation cohort. As shown in Table II, FOXA2 mRNA (probe 210103_s_at), FOXC1 mRNA, FOXD3 mRNA, FOXF1 mRNA, FOXG1 mRNA (probe 206018_at), FOXN2 mRNA and FOXN3 mRNA (probe 205022_s_at) were significantly expressed in both datasets. The probes for FOXC1 mRNA, FOXD3 mRNA and FOXF1 mRNA were eliminated due to discordant significance, as they were represented by high expression in one dataset and low expression in the other. At the end of univariate analysis, FOXA2 mRNA (probe 210103_s_at), FOXG1 mRNA (probe 206018_at), FOXN2 mRNA and FOXN3 mRNA (probe 205022_s_at) were consistently significantly associated with poor survival in the test set as well as in the TCGA validation set. The Kaplan-Meier survival curves according to expression of these genes are shown in Figure 1. Low expression of FOXA2 mRNA, FOXN2 mRNA and FOXN3 mRNA were associated with poor survival; conversely, high expression of FOXG1 mRNA was associated with poor survival. We then proceeded to multivariate analysis.
FOXA2, FOXG1, FOXN2 and FOXN3 are independently associated with survival in adult patients with GBM (Table III). Multivariate analysis in the test set demonstrated that FOXA2 mRNA (p=0.006), FOXG1 mRNA (p=0.044), FOXN2 mRNA (p=0.004), FOXN3 mRNA (p=0.001) were significant independent predictors of survival. Similarly, FOXA2 mRNA (p=0.019), FOXG1 mRNA (p=0.016), FOXN2 mRNA (p=0.000101), FOXN3 mRNA (p=0.013) were also significant independent predictors of survival in the TCGA cohort (Table III). Predictive significance of FOXA2, FOXG1 and FOXN2 gene expression in adult patients with GBM. The data presented above provide evidence that FOXA2 mRNA, FOXG1 mRNA, FOXN2 mRNA, and FOXN3 mRNA have prognostic significance. To investigate if they also have predictive significance, we conducted analysis in various groups in the TCGA cohort that received chemotherapy with/without radiotherapy. As shown in Figure 2, low FOXA2 mRNA, low FOXN2 mRNA and high FOXG1 mRNA expression were significantly associated with poor survival in those patients who had received chemotherapy. There was no significance of the expression of these genes in patients who received no chemotherapy. Similarly, low FOXA2 mRNA, low FOXN2 mRNA and high FOXG1 mRNA expression were significantly associated with poor survival in patients who had received radiotherapy (Figure 3). There was no significance in patients who received no radiotherapy. Interestingly, low FOXN3 mRNA expression was significantly associated with poor survival only in the group not treated with chemotherapy or radiotherapy. Taken together, the data suggest that FOXA2 mRNA, FOXN2 mRNA and FOXG1 mRNA have predictive significance in GBM.
Kaplan-Meier survival curves in Test and TCGA validation cohorts for genes shown to be significant after BH correction.
Kaplan-Meier survival curves in TCGA validation cohort patients that either received Chemotherapy or no chemotherapy for genes shown to be significant after BH correction.
Discussion
As far as we are aware, this is the first study to comprehensively evaluate the FOX gene family in GBM. FOXG1 is involved in the early development of the brain and has been linked to CNS tumors, including GBM (9). During normal development in mice, FOXG1 has been shown to be crucial in many aspects of the development of the forebrain, more specifically the telecephalon. It acts as a transcriptional repressor, not only during early development, but also into adulthood. In adulthood, it is thought to influence neuronal survival (11). A variant of Rett syndrome, known as FOXG1 syndrome, is characterised by mutation in the FOXG1 gene and manifests as severe mental retardation, severe post-natal microcephaly, lack of language development, epilepsy and autism-like features (12). FOXG1 has also been implicated in medulloblastoma, hepatoblastomas (13, 14) and ovarian cancer (15). A direct role for FOXG1 in GBM was recently described (9). A link between the FOXG1 protein and the member of the Groucho family human transducin-like enhancer of split (TLE) proteins has been shown. Verginelli et al. observed that FOXG1 and TLE form a complex in brain tumor-initiating cells that have stem cell-like properties (9). Interestingly, inhibiting the function of this complex reduced tumor growth. Our data would concur with pre-clinical observations.
We also found that low mRNA expression of FOXA2, FOXN2, and FOXN3 was associated with worse clinical outcomes. FOXA2 is involved in the embryonic development of the liver (2) and pancreas (16). In lung cancer models, FOXA2 acts to suppress metastasis by preventing epithelial- to mesenchymal transition (17). FOXN2 (murine) is known to be involved in the embryogenesis of the central nervous system (18). In humans, FOXN2 is also known as human T-cell leukaemia virus enhancing factor (19) and may be involved in the pathogenesis of adult T-cell leukaemia (19). FOXN3, if mutated in mice, causes craniofacial defects which are very similar to those seen in humans that have FOXN3 mutations (20). In humans, reduced FOXN3 expression has been described in carcinomas of the mouth, larynx, liver (16) and Hodgkins lymphoma (21).
Kaplan-Meier survival curves in TCGA validation cohort patients that either received radiotherapy or no radiotherapy for genes shown to be significant after BH correction.
Taken together, the above data provide clinical evidence for a potential role for FOX genes in gliomagenesis.
Multivariate analysis in the test and the validation sets.
- Received March 12, 2015.
- Revision received March 23, 2015.
- Accepted March 26, 2015.
- Copyright© 2015, International Institute of Anticancer Research (Dr. John G. Delinasios), All rights reserved
References
In this issue
