Development and Use of Clinical Pharmacodynamic Assays to Demonstrate Target Engagement for AZD4573, a Selective and Potent CDK9 Inhibitor Currently in Phase I Clinical Trials

CDK9 (cyclin-dependent kinase 9) regulates RNA transcription through its phosphorylation of RNA polymerase II and subsequent release from a "paused" elongation state, resulting in transcriptional activity. A transient inhibition of CDK9 activity will significantly impact transcripts with a short half-life, which includes the transcript for MCL1, an anti-apoptotic protein. AZD4573 is a highly potent and selective CDK9 inhibitor. We have previously reported that AZD4573 treatment results in dose-dependent inhibition of pSer2 of RNA polymerase II, reductions in MCL1 mRNA and protein, and an increase in cleaved caspase 3, resulting in cell death of hematological tumor cell lines and regressions in AML/DLBCL in vivo models [Cidado et. al., AACR Annual Meeting, Abstract 310 (2018)].

AZD4573 is currently being evaluated in a Phase I all-comers relapsed/refractory hematological malignancy trial (NCT03263637). To generate clinical evidence that the compound is inhibiting its target and having the expected downstream effects, we have developed pharmacodynamic biomarker (PD) assays for clinical use to measure the changes in pSer2/RNApol II, MCL1 protein, cleaved Caspase 3 and cleaved PARP. We have chosen a capillary electrophoresis platform which allows quantification and multiplexing while requiring lower sample quantities (<20ng/capillary) than Western blotting. Data from AZD4573-treated MV4,11 AML cells using the clinical assays were similar to what had been shown previously with Western blotting, i.e. similar dose-dependence across 0.1nM to 1uM and >80% inhibition of both pSer2 and MCL1 at 100nM (in 10% FCS). Next, we validated the use of these assays with human peripheral blood mononuclear cells (PBMCs) since we expected that some patients from the trial are likely to have significant numbers of circulating tumor cells (i.e. AML/CLL). After incubation of human whole blood with 1uM AZD4573 for different times, PBMCs were isolated and analyzed using the capillary electrophoresis assays. These results showed at least ~80% inhibition of pSer2 at 2h, consistent with preclinical efficacy modeling and predicted clinically achievable levels. In addition, we optimized the PBMC isolation protocol for the clinical sites and showed that compound washout during the procedure did not diminish AZD4573-mediated effects. In another study, we mimicked potential clinical conditions (e.g. 30 min, 1h, and 2h between blood collection and PBMC isolation, ambient vs 37°C storage conditions) and demonstrated that significant ex vivo apoptosis (i.e. activation of cleaved caspase 3) was unlikely to occur if PBMCs were isolated within 1-2h of blood collection. Finally, by showing that AZD4573 target engagement and pathway inhibition can be measured in normal PBMCs, these assays will be informative for all patients, including those without circulating tumor cells (e.g. many lymphomas).

Here we describe the development and characterization of a suite of pharmacodynamic assays for clinical use with blood samples for our CDK9 program. Preliminary data from our Phase I study demonstrate successful implementation of these assays and support that AZD4573 inhibits CDK9 with the expected downstream effects, in the clinical setting. Updated results will be presented at the meeting.