Benign
neurofibromas, the main phenotypic manifestations of the rare
neurological disorder neurofibromatosis type 1, degenerate to malignant
tumors associated to poor prognosis in about 10% of patients. Despite efforts in the field of (epi)genomics, the lack of prognostic
biomarkers with which to predict disease evolution frustrates the adoption of appropriate early therapeutic measures. To identify potential
biomarkers of malignant
neurofibroma transformation, we integrated four human experimental studies and one for mouse, using a gene score-based meta-analysis method, from which we obtained a score-ranked signature of 579 genes. Genes with the highest absolute scores were classified as promising disease
biomarkers. By grouping genes with similar
neurofibromatosis-related profiles, we derived panels of potential
biomarkers. The addition of promoter methylation data to gene profiles indicated a panel of genes probably silenced by hypermethylation. To identify possible therapeutic treatments, we used the gene signature to query
drug expression databases.
Trichostatin A and other
histone deacetylase inhibitors, as well as
cantharidin and
tamoxifen, were retrieved as putative therapeutic means to reverse the aberrant regulation that drives to malignant cell proliferation and
metastasis. This in silico prediction corroborated reported experimental results that suggested the inclusion of these compounds in clinical trials. This experimental validation supported the suitability of the meta-analysis method used to integrate several sources of public genomic information, and the reliability of the gene signature associated to the malignant evolution of
neurofibromas to generate working hypotheses for prognostic and
drug-responsive
biomarkers or therapeutic measures, thus showing the potential of this in silico approach for
biomarker discovery.