Comparison of Two Rapid Predictive Methods for Therapeutic Targeting in Acute Myeloid Leukaemia Cells

Background. AML blasts from different patients vary in the agents to which they are most responsive. With a plethora of novel agents to evaluate, there is a lack of predictive biomarkers to precisely assign targeted therapies to individual patients. Primary AML cells often survive poorly in vitro, thus confounding conventional cytotoxicity assays. Dynamic profiling, i.e. functional biomarker assays of responsiveness to chemotherapeutic agents, is an alternative approach to help further the move towards personalised therapy. The purpose of this work was to assess the potential of two same-day dynamic profiling assays in AML cell lines to predict response to chemotherapy. (i) Ribosomal protein S6 (rpS6) is a downstream substrate of PI3K/akt/mTOR/ p70S6 kinase and MAPK/p90S6 kinase pathways and is also dephosphorylated following DNA double strand breaks. It thus has the potential to function as a biomarker of responsiveness to several therapeutic agents. (ii) Cellular propensity for apoptosis can be interrogated via a functional assay termed BH3 profiling. Dynamic BH3 profiling can be used to predict cellular responses to therapy based on priming mitochondria with BH3 domain peptides and thus allowing early detection of pro-apoptotic drug effects on mitochondrial outer membrane permeabilisation.

Methods.We measured effects of a short (four hour) incubation with drugs of interest on rpS6 phosphorylation and BH3 peptide-accelerated cytochrome C release by flow cytometry. Baseline rpS6 phosphorylation was determined by culture with the mTOR inhibitor Rapamycin and the MEK inhibitor U0126. BH3 profiling included permeabilisation with digitonin followed by mitochondrial exposure to PUMA BH3-derived peptide. To establish specificity for sensitive cells we also measured dose responses to the drugs in 12 diverse AML cell lines after 48 hours' culture.

Results.

RpS6 dephosphorylation at four hours closely predicted the 48 hour IC50 for FLT3 inhibitors, an hsp90 inhibitor and topoisomerase II inhibitors. ROC (predictive test) analysis of small molecule inhibitors showed that the assay was highly sensitive and specific with area under the curve (AUC) values of 1.0 (sorafenib), 1.0 (AC220) and 1.0 (17-AAG). RpS6 phosphorylation also predicted response to the double strand break inducing drugs, with AUC values of 1.0 (mylotarg), 0.83 (etoposide) and 0.82 (vosaroxin). In contrast, responses to cytarabine and ABT-199, likely independent of mTOR, MAPK or double strand break response pathways, are not predicted by rpS6 dephosphorylation. PUMA-BH3 peptide-induced cytochrome C release also closely predicted the 48 hour IC₅₀ with AUC values for FLT3 inhibitors of 1.0 (sorafenib) and 1.0 (AC220), double strand break inducing drugs 1.0 (mylotarg), 1.0 (etoposide) and 1.0 (vosaroxin) and a Bcl-2 targeting agent 0.875 (ABT-199). Response to 17-AAG and cytarabine were not predicted by this assay. Preliminary analysis shows that differential responses within primary AML sample subsets can be interrogated with these methodologies.

Conclusions. In conclusion, we have established that rpS6 dephosphorylation and/or PUMA-BH3 peptide-induced cytochrome C release predict chemoresponsiveness to tyrosine kinase inhibitors, topoisomerase II inhibitors, an hsp90 inhibitor and a Bcl-2 targeting agent in AML cell lines after short term culture. Both assays are sensitive, specific and amenable to leukaemic sub-population analysis and are thus suitable assays for further development towards use in a clinical setting. pRS6 can be performed on smaller samples and is less technically challenging than priming with PUMA-BH3 peptide, but does not take into account apoptosis resistance that may occur independently of pathways effecting pRS6 inhibition.