Dataset Information

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

ABSTRACT: Currently there is no method available to predict response to farnesyltransferase inhibitors (FTI). 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. This two-gene ratio was validated by quantitative PCR (QPCR) in the newly diagnosed AML cohort. We further demonstrated that this classifier could predict response to tipifarnib in an independent set of 54 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 = 0.0001), which was shown to be independent of other prognostic factors including a previously described gene expression classifier predictive of overall survival. Therefore, these data indicate that a two-gene expression assay may have utility in categorizing a population of AML patients who are more likely to respond to tipifarnib. Keywords: classification

ORGANISM(S): Homo sapiens

PROVIDER: GSE8970 | GEO | 2007/09/12

SECONDARY ACCESSION(S): PRJNA102421

REPOSITORIES: GEO

ACCESS DATA

Similar Datasets

Project description:Background: Validated markers to predict outcome in rectal cancer patients treated with multimodal therapy remain elusive. Identifying molecular profiles for disease prognosis would be required for the design of clinical trials aimed at optimizing risk-adapted therapies. We have therefore used whole genome expression profiling of tumors of a large cohort of patients enrolled in the multicenter trials of the German Rectal Cancer Study Group (GRCSG) to identify molecular profiles for individualized therapy. Methods: We prospectively collected pretherapeutic biopsies from patients (n=300) treated according to the GRCSG trial guidelines from seven different German surgical departments using rigid quality controls. These samples were profiled by global gene expression analysis and a classifier developed to predict postoperative lymph node status and Disease Free Survival (DFS). The performance of the classifier was validated with an independent, prospectively collected set of samples. Findings: The final training and test set included 198 patients. Analyzes for postoperative nodal status and DFS revealed 69 and 674 differentially regulated genes, respectively. Depending on the classifier the accuracy to predict lymph node status ranged from 64% to 69% with negative predictive values between 72% and 74%. Stratification according to DFS resulted in a good (n=99), bad (n=96) and a small very bad prognosis group (n=3). Based on linear discriminant analysis the classifier for positive lymph node status was validated in 47 independent patients and revealed an accuracy of 72% and a positive predictive value of 100%. Thereby, prediction based on molecular profiling is superior to prediction based on conventional clinical markers. Interpretation: Whole genome expression analysis of pretherapeutical biopsies resulted in a molecular classifier of disease prognosis. This classifier was successfully validated. The high positive predictive value for post therapeutic lymph node status allows identification of patients requiring alternative or intensified treatment protocols. These data are currently validated for their clinical applicability and should be taken as basis for future molecular driven clinical trials to develop risk-adapted treatments.

Dataset Information

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure