Combination of Gene Expression Data from Myeloid Cell Lines with In Vitro Response to Molecularly Targeted Therapies.

AML is currently classified using a combination of morphology and major cytogenetic abnormalities. There is considerable heterogeneity with respect to secondary cytogenetics, molecular mutations, response to therapy, remission rate, and overall survival. New molecular therapies are being developed constantly and are tested in vitro using patient samples in order to balance the drug-patient relationship. Use of the same drug produces patient specific responses, whereas different molecular drugs give drug specific responses.

In order to rationalise and enhance the drug-patient relationship we have, using MTS assays, tested a range of molecular inhibitors against FLT3, PKC, HSP90, aminopeptidases and NFκB as well as modified nucleoside analogues on a panel of myeloid cell lines. These cell lines have also been profiled for gene expression. The combination of drug response and gene expression levels, were analysed by statistical algorithms including SAM, ANOVA and T-test to identify genes that differentiated between molecular responses or the overall therapeutic index (TI). Forty one genes distinguished between the molecular responses of FLT3, PKC, HSP90, and NFkB inhibitors. Around 100 genes were needed for the Ara-C response. The gene expression profiles of 187 diagnostic AML samples were also available with full clinical and molecular analysis. Some of these samples have been tested for in vitro responses to the molecular therapies, such as those targeted to FLT3 or HSP90. The TI gene list was used to analyse AML sub-groups including elderly patients (>60 yrs old); younger patients (<60 yrs old); cytogenetic risk groups, and those patients with FLT3 mutations (either ITD or TKD mutations) in an attempt to identify patients who may respond to specific molecularly targeted therapy. The Ara-C genes were used to predict whether a CR response was achieved. In patients less than 60 yrs old this correlation was not significant due to the great majority of patients achieving a complete remission. However, this was significant (p<0.031) in patients >60 yrs. Around 12% of patients were predicted to be responsive to HSP90 inhibitors whilst 18% might respond to FLT3 inhibitors. Several patients were in both groups, so 17% of patients were sensitive to a least one of the drugs. Whilst the potential added value of identifying patients sensitive to one of these drugs was only an additional 1–3%, this study shows the potential of using expression data to identify patients for specific molecular therapies and increasing the possibility of achieving a remission.