A 2-gene classifier for predicting response to the farnesyltransferase inhibitor tipifarnib in acute myeloid leukemia

At present, there is no method available to predict response to farnesyltransferase inhibitors (FTIs). We analyzed gene expression profiles from the bone marrow of patients from a phase 2 study of the FTI tipifarnib in older adults with previously untreated acute myeloid leukemia (AML). The RASGRP1/APTX gene expression ratio was found to predict response to tipifarnib with the greatest accuracy using a "leave one out" cross validation (LOOCV; 96%). RASGRP1 is a guanine nucleotide exchange factor that activates RAS, while APTX (aprataxin) is involved in DNA excision repair. The utility of this classifier for predicting response to tipifarnib was validated in an independent set of 58 samples from relapsed or refractory AML, with a negative predictive value (NPV) and positive predictive value (PPV) of 92% and 28%, respectively (odds ratio of 4.4). The classifier also predicted for improved overall survival (154 vs 56 days; P < .001), which was independent of other covariates, including a previously described prognostic gene expression classifier. Therefore, these data indicate that a 2-gene expression assay may have utility in categorizing a population of patients with AML who are more likely to respond to tipifarnib.