Radiogenomic features of CNS tumours and MiRNAs correlation phenotypes analysis

Radiogenomic features of CNS tumours and MiRNAs correlation phenotypes analysis Gliomas are the most common, invasive and aggressive tumours of the adult central nervous system (CNS), representing 81% of malignant brain tumours, with 5-year survival rate of ~ 5%. Gliomas arise from astrocytic cells, oligodendrocytic cells, or a mixture of both cell types, and can spring up from all the districts in the CNS. The classifying principle for gliomas is based on their morphological features, cytoarchitecture and immunohistological asset. Despite notable advances in surgery and therapy research efforts, the outcome for patients with gliomas remains poor and patients will eventually die due to complications of gliomas. MicroRNAs (miRNAs) are short sequences RNAs, 18-22 nucleotides long, which do not encode proteins, however these molecules are strong post-transcriptional modulators through binding messenger RNAs that reduce the translation and stability. miRNAs are key regulators of physiological neural development, they play a strategic role in glioma’s initiation and progression, they reversibly regulate protein-coding genes expression by targeting the regions of genes regulation. Texture analysis, a relatively recent method of investigation, is a non-invasive radiomic technique. It allows analysis of the texture parameters of digital radiological images in order to extract characteristic of the heterogenous macroscopic tissue and comparing them to heterogenous features of the microscopic tissue, overcoming the limits of human visual perception. Material and methods 13 patients were selected who underwent surgery due to suspicion of a primary tumor of the central nervous system. The pathological analysis confirmed the diagnosis of primary tumor of the Central Nervous system (III e IV WHO). All tissues were immediately stored in RNA-later and 48 hours later were subjected to the procedure of RNA. All extracted RNA samples were quantitatively and qualitatively evaluated. the samples were sent for deep sequencing through Illumina HiSeq3000. The NGS performed by CRS4 show us many miRNAs. We select the miRNAs present in the two samples presenting a normal distribution and then we performed a paired Student’s T test in order to verify if the expression of miRNA were statistically different between glioma and surrounding tissue samples. All the miRNA considered resulted statistically different between the two groups; in particular 10 miRNA (miR-25; miR-339; miR-362; miR-92b-5p_mature; miR-4677; miR-106b-5p; miR-501; miR-505; miR-542; miR-92b-5p_star) resulted more expressed in glioma samples, and the other 15 (miR-25; miR-339-; miR-362; miR-92b; miR-4677; miR-106b; miR-501; miR-505; miR-542-5p_star; miR-92b) in surrounding tissue samples At the same moment, for every patient we performed texture analysis in five types of MR sequences. Results We identified the miRNA that resulted statistically different in terms of number of copies between the two groups, we verified if their expression in glioma samples was correlated with the chosen parameters of the texture analysis above mentioned. We then performed a Pearson’s correlation test for every single miRNA, and we selected only the miRNA that showed a Pearson’s correlation coefficient (xy) > 0.6 (strong positive correlation) or < -0.6 (strong negative correlation). Our research obtained impressive results: first, through the use of Next Generation Sequencing, differentially expressed between healthy tissue and pathological tissue miRNA’s were confirmed; secondly, there are significant differences between texture analysis MR data of healthy and glioma tissue; lastly correlation between differentially expressed miRNA’s and MR data creates a “molecular footprint” of gliomas, thus opening a new path of search in order to reasonably improve diagnostic accuracy in the studyin of brain tumors