Identification of Molecular Predictors of Response to ZARNESTRA™ (Tipifarnib, R115777) in Relapsed and Refractory Acute Myeloid Leukemia.

The orally available farnesyltransferase inhibitor (FTI) tipifarnib has demonstrated clinical efficacy in hematological disease. In an effort to identify patients with a higher likelihood of response, we have employed microarray technology to identify gene expression markers that predict response to tipifarnib. In an open label, single arm, Phase II study of patients with relapsed or refractory acute myeloid leukemia, tipifarnib was administered at a dose of 600 mg twice daily for the first 21 days of each 28-day cycle. Gene expression profiles from 80 bone marrow samples, collected before drug treatment, were analyzed on the Affymetrix U133A gene chip that contains approximately 22,000 genes. Statistical analyses were performed to identify genes that predict patient outcome following treatment with tipifarnib. Nineteen genes were selected because they showed significant differences in gene expression between the responder group and the non-responder group of patients. Several of these genes are associated with oncogenesis, cell proliferation, and FTI biology. One of the candidate genes, lymphoid blast crisis oncogene (oncoLBC or AKAP13), was over-expressed in patients who were resistant to tipifarnib. A leave-one-out cross validation (LOOCV) showed that the predictive value of this gene provided a sensitivity of 93% in identifying responders and a specificity of 61%, providing an overall diagnostic accuracy of 69%. In addition, supervised analysis identified a combination of 3 gene expression markers that together can potentially be used to predict patient response to the drug with the best overall accuracy of 74%. This 3-gene expression signature provided a sensitivity of 86% and a specificity of 70% in the LOOCV. In both instances, classification of patients using the gene expression signatures demonstrated significantly enhanced overall survival in those predicted to be responders. These data indicate that diagnostic gene expression signatures can potentially double the response rate to tipifarnib from 24% to 48%. The identification of these gene expression markers is an important step towards defining diagnostic signatures that could be used to identify AML patients who are likely to respond to tipifarnib and for understanding pathways that are affected by farnesyltransferase inhibition in AML.