Abstract 4305: NMI as a biomarker of response to CDK4/6 inhibition in a preclinical model of neuroblastoma

Background Hyper-activation of the Cyclin D-CDK4/6 signaling pathway is a common feature of neuroblastoma. Previously, it has been shown that while a majority of preclinical models of neuroblastoma are highly sensitive to pharmacologic inhibition of CDK4/6 by LEE011 (Novartis Oncology), a subset are remarkably resistant. In an effort to aid in the selection of a patient population that will best benefit from this therapy, we sought to identify genomic biomarkers of response to CDK4/6 inhibition. Methods LEE011 IC50 values were determined using the RT-CES cell impedance assay, and gene expression was established at baseline using HuGene 1.0ST microarrays (Affymetrix). Expression data was then coupled with IC50 values to identify a set of genes significantly associated with LEE011 response via Pearson correlation. Using this gene set, a Lasso linear regression model was built in order to predict sensitivity to LEE011 based upon the expression of a subset of genes chosen to be most predictive of response. The model was tested in a separate cohort of neuroblastoma cell lines by comparing predicted responses to LEE011 with experimentally determined LEE011 IC50 values. Functional RNAi and over-expression studies are ongoing in order to validate genes identified to be correlated with or predictive of response. Results The Lasso model was able to predict the response of neuroblastoma cell lines to LEE011 based upon the expression pattern of 11 genes. A comparison of Lasso predictions to the LEE011 IC50 values of a separate panel of cell lines yielded a correlation coefficient of 0.63. The correlation of predicted to actual response across our entire cell line panel was 0.96, suggesting that response to CDK4/6 inhibition by LEE011 can be accurately predicted based on the expression of these genes. Of the 11 genes, NMI (N-myc and stat interactor) was prioritized for further validation and functional studies as it was identified by Pearson correlation to be the third most correlated gene with response (r = 0.86; p = 3.1×10-7) and because it interacts with MYCN, a known prognostic indicator in neuroblastoma. We next showed that NMI was expressed higher at both the mRNA and protein levels in neuroblastoma cell lines resistant (N = 6) to LEE011 compared to those demonstrating sensitivity to the compound (N = 16). Studies are ongoing to confirm the role of NMI as a predictor of resistance as well as to assess its contribution, if any, in desensitizing neuroblastoma to CD