11066 Background: Oncogenic driving alterations define types of lung adenocarcinoma with different prognosis and sensitivity to targeted agents. MicroRNAs (miRNAs) are a new class of non-coding RNAs involved in gene expression regulation. How miRNAs are dysregulated in lung cancer with ALK translocation, EGFR or KRAS mutation is unknown. In this study we aimed to identify miRNA signatures according to the presence of specific driver and to correlate miRNAs deregulation with patient outcome. Methods: The study was conducted in a cohort of 70 lung cancer patients (pts) including 18 ALK+ tumors, 11 ALK-/EGFR mutation+, 15 ALK-/KRAS mutation+, 26 ALK-/EGFR and KRAS wild-type and defined as triple negative. Matched normal lung tissue from 18 cases representative of the entire cohort were also included onto the analysis. RNA was isolated from formalin-fixed paraffin-embedded tissue (FFPE), using the Recover ALL kit (Ambion). NanoString nCounter system platform was used to generate the miRNA profile. We used Limma to test for differential expression analysis of data. The miR-515 family expression between tissues was validated by RT-qPCRs, analyzed using the parametric t-test (unpaired, 2-tailed for validation). Results: miRNA expression profile clusters distinctly ALK+ pts from ALK- and normal lung tissue. Within the ALK- group we found specific miRNAs subsets able to sub-stratify KRAS versus EGFR careers clustering sharply triple negative versus EGFR mutation+ and triple negative versus KRAS mutation+. miRNAs belonging to the miR-515 family seems to be the most deregulated in the ALK+ versus ALK-. Although their expression is stably high in normal tissues and ALK+ class, they are highly downregulated in KRAS mutated versus EGFR mutated and versus triple negative (p-value <0.001 for all comparisons). Conclusions: miRNAs profile significantly differs in pts with ALK translocation, EGFR mutations and KRAS mutations. Analysis of miR-515 family members is ongoing in order to correlate their expression levels with pts’ outcome. In vitro modulation of miR-515 family expression levels, together with drugs treatment are ongoing in order to find possible chemo-resistance/chemo-sensitivity miRNA dependent, in ALK+ and ALK- model.
MO14-4 Machine learning-based clustering of scRNAseq data from normal lung tissue and application to lung cancer
