Gene Expression Profiling of Acute Myeloid Leukemia Shows Therapeutically Meaningful Patterns of Association with Ara-CTP Pharmacokinetics and Pharmacodynamics

Abstract

Cytarabine (ara-C) is one of the most effective agents for the treatment of AML; however development of drug resistance remains a major limitation. To identify genes that impact ara-C therapy, we explored the association of leukemic blast gene expression (measured using the Affymetrix U133A microarray) with intracellular ara-CTP pharmacokinetic and pharmacodynamic variables (DNA synthesis relative to baseline measured in leukemia cells, peripheral blast clearance) measured in leukemia cells from 42 patients treated under St. Jude AML97 protocol. In this study, patients received either a daily short infusion of ara-C or a continuous infusion of ara-C in combination with cladribine (CdA). Probe sets with mRNA expression levels (at diagnosis) that positively correlate with leukemic blast ara-CTP levels at day 1 (after ara-C alone) and day 2 (after ara-C and CdA), negatively correlate with extent of inhibition in days 1 and day 2 DNA synthesis relative to baseline, and negatively correlate with absolute blast count at 48 hours relative to baseline were considered to exhibit a therapeutically beneficial pattern of association. Probe sets with correlations opposite to those mentioned above were considered to exhibit a therapeutically detrimental pattern of association. An association pattern test found statistically significant evidence (p ≤ 0.005, FDR ≈ 0.30) that the expression of 304 probe sets associates with a beneficial pattern of association and of 61 probe sets associates with a detrimental pattern of association. These results include 34 cell cycle genes with beneficial association patterns and 2 cell cycle genes with detrimental association patterns (PPP1R13B and CUL4B). Also, 2 tumor necrosis factor genes (TNFRSF25 and FASLG) showed significant association with beneficial association patterns. The significant probe sets included RAS oncogene pathway genes, RAP2A and RAB9A with a detrimental pattern of association and the RAS-like gene RIT2 with a beneficial pattern of association. It has been recently shown that AML patients with mutations in RAS genebenefit from high ara-C doses1. The erythroblastic leukemia viral oncogene homologs EGFR and THRB demonstrated a statistically significant beneficial pattern of association. The T-cell leukemia homeobox gene also showed a statistically significant beneficial pattern of association. We further analyzed these 365 genes using Ingenuity Pathway Analysis tools and identified 16 networks with scores greater than 11 (p value < 10−11). In addition to the apoptotic genes, genes in the metabolic pathway of ara-C as well as few previously identified genes were also found to be associated with the pharmacokinetic and/or pharmacodynamic measures of ara-C in circulating blasts. In conclusion, our study has identified molecular signatures (in leukemic blasts at diagnosis) associated with the ara-CTP pharmacokinetics and pharmacodynamics. Validation of these signatures will help in better understanding of ara-C response as well as the mechanism underlying drug resistance. We are concentrating on these candidate genes and pathways for functional validation studies.