Improved grading and survival prediction of human astrocytic brain tumors by artificial neural network analysis of gene expression microarray data

Abstract

Histopathologic grading of astrocytic tumors based on current WHO criteria offers a valuable but simplified representation of oncologic reality and is often insufficient to predict clinical outcome. In this study, we report a new astrocytic tumor microarray gene expression data set (n = 65). We have used a simple artificial neural network algorithm to address grading of human astrocytic tumors, derive specific transcriptional signatures from histopathologic subtypes of astrocytic tumors, and asses whether these molecular signatures define survival prognostic subclasses. Fifty-nine classifier genes were identified and found to fall within three distinct functional classes, that is, angiogenesis, cell differentiation, and lower-grade astrocytic tumor discrimination. These gene classes were found to characterize three molecular tumor subtypes denoted ANGIO, INTER, and LOWER. Grading of samples using these subtypes agreed with prior histopathologic grading for both our data set (96.15%) and an independent data set. Six tumors were particularly challenging to diagnose histopathologically. We present an artificial neural network grading for these samples and offer an evidence-based interpretation of grading results using clinical metadata to substantiate findings. The prognostic value of the three identified tumor subtypes was found to outperform histopathologic grading as well as tumor subtypes reported in other studies, indicating a high survival prognostic potential for the 59 gene classifiers. Finally, 11 gene classifiers that differentiate between primary and secondary glioblastomas were also identified.

Dual-specificity phosphatase DUSP6 has tumor-promoting properties in human glioblastomas

“…Comparing the neoplastic cells with the normal counterpart (normal human astrocytes), we observed an increased expression of DUSP6 (three-to sixfold) at baseline that increased upon mitogenic stimulation ( Figure 1a, lower panel). Consistently, gene expression studies revealed DUSP6 upregulation (mean fold changes: 3.14) in human astrocytic brain tumors compared with normal brain tissue (Petalidis et al, 2008). All cells were serumstarved overnight and stimulated with 100 ng/ml of EGF for 3 h. Despite some fluctuation monitored, northern hybridization of non-neoplatic cells (unstimulated cultures of murine pheochromocytoma PC12 cells, mouse fibroblasts NIH-3T3 and normal human astrocytes) showed basal negligible mRNA levels and powerful induction of DUSP6 mRNA upon mitogenic stimulation ( Figure 1a, upper panel).…”

“…Interestingly, DUSP6 is a selected gene for astrocytic classification in angiogenic functional class and, recently, a potential transcriptional effect of hypoxia on the DUSP6 gene has been reported (Petalidis et al, 2008;Bermudez et al, 2011). Moreover, inducible expression of DUSP6 specifically blunts in vivo fibroblasts growth but, in our opinion, this effect probably relies on Ha-Ras oncogene transformation (Marchetti et al, 2004).…”

Integrin α5β1 and p53 convergent pathways in the control of anti-apoptotic proteins PEA-15 and survivin in high-grade glioma

“…A tumorsuppressor role of PEA-15 was proposed in brain tumors based on its decreased expression at the mRNA and protein level in GBM compared with low-grade astrocytoma with 50% of GBM nevertheless having high PEA-15 expression. 41,42 Interestingly, expression of pser116-PEA-15 was confined in specific intratumoral area with the strongest immunoreactivity in perinecrotic area. 43 Our preliminary results indicated that α5 integrin could also be localized in these particular area in GBM specimens (data not shown).…”

Identification and validation of a gene expression signature that predicts outcome in malignant glioma patients

“…To quantify the accuracy of prediction, we utilized independent test data sets from four studies (25,26,37,38). Firstly, ensemble mortalities for each individual in the test data were computed using the random survival forests model fitted to our data set with selected genes.…”

“…Recently, microarray technology has permitted development of multi-organ cancer classification including gliomas (4-6), identification of glioma subclasses (7)(8)(9)(10)(11)(12)(13)(14)(15), discovery of molecular markers (16)(17)(18)(19)(20)(21)(22)(23) and prediction of disease outcomes (24)(25)(26)(27). Unlike clinicopathological staging, molecular staging can predict long-term outcomes of any individual based on the gene expression profile of the tumor at diagnosis, helping clinicians make an optimal clinical decisions.…”