Targeting Mitochondrial Apoptotic Priming State to Personalize Therapy for Relapsed Acute Myeloid Leukemia

Despite of remarkable changes in therapeutic landscape of AML in past 5 years, relapsed AML patients still succumb to disease. In majority of patients, the common cause of death ultimately is the emergence multi-drug resistance at relapse. Using patient-derived xenograft (PDX) as our models, we investigated the basis of multi-drug resistance and methods for identifying drug sensitivity in relapsed AML. We hypothesized that there is differential activation of drug-induced mitochondrial apoptotic signaling between treatment-naïve and relapsed myeloblasts. To study mechanisms of drug resistance in AML, we developed PDX models (n=18 models) of in vivo acquired resistance to single agents (BCL-2/MCL-1 inhibitor, FLT3 inhibitor, SMAC mimetics and BRD-4 inhibitor) of distinct mechanisms, which caused board chemoresistance across all models. We tested whether multidrug resistance is associated with reduction in apoptotic signaling, using BH3 profiling, a method for measurement of mitochondrial sensitivity to apoptosis via exposure to synthetic BH3 domain peptides, on matched pre- and post-resistant PDXs. Using BH3 profiling, we found that a reduction in mitochondrial apoptotic priming consistently accompanied the acquisition of multidrug resistance irrespective of the mechanism of drug action and genetic status of PDX models. Next we created pharmacologic sensitivity landscape for 22 AML PDX models by exposing splenic myeloblasts isolated after confirmed engraftment (>70% circulating hCD45+ blasts) to drug a drug panel (n=40) for 14 hours. We first demonstrate that ex vivo drug-induced mitochondrial sensitivity singling measured via dynamic BH3 profiling (DBP) segregates PDXs according to prior treatment. By using unsupervised hierarchical clustering we found that AML PDXs developed from treatment naïve patients clustered together and showed greater apoptotic priming responses to a larger number of targeted agents than PDXs from R/R patients, which also formed discrete clusters. We replicated drug-response patterns from PDX models in primary and relapsed AML patients by analyzing ex vivo sensitivity responses to >500 drugs across relapse and diagnosis patients (Malani et al, 2021 and Tyner et al, 2018). We show that human AML patients at relapse also broadly lose sensitivity to agents in a similar pan-resistant manner as PDX models.

Using unbiased whole transcriptomic analysis of naïve and resistant PDXs (n=78 samples), we found common pathways of resistance where positive enrichment was observed for cytokine-cytokine receptor interactions, xenobiotics biodegradation, metabolic pathways, Hippo signaling, and Ras signaling in resistant compared to treatment-naïve PDXs. Conversely, mechanisms relating to DNA repair and replication, such as base excision repair, mismatch repair, and homologous recombination, were negatively enriched. Moreover, pathways associated with specific drug resistance were also observed. One of the known multi-drug resistance mechanisms to chemotherapy drugs in AML is increased expression of efflux pumps from the superfamily of ATP binding cassette. We found enrichment in ABC transporter family pathway signatures in all 3 PDX models, DFAM-61786, DFAM-15354, and DFAM-61345, resistant to quizartinib, birinapant, and JQ-1 and validated higher drug-efflux activity in quizartinib and birinapant resistant model using calcein-AM efflux assay.

We finally tested whether DBP could predict in vivo response of patient myeloblasts in PDX models. We tested in vivo sensitivity to 5 drugs of disparate mechanisms of action: birinapant and LCL-161 (SMAC mimetics), JQ-1 (BRD-4 inhibitor), venetoclax (BCL-2 antagonist), and quizartinib (FLT-3 inhibitor) in 4-9 different PDX models each. Using ex vivo DBP assay, we identified drugs that prime not only treatment naïve PDXs but also R/R PDXs. We identified persistent in vivo activity in a drug-specific manner, with BH3 mimetics and HDAC inhibitors showing overlapping activity across different drug-resistant models. Overall, we demonstrate that acquired resistance to targeted therapy in AML is accompanied by common mechanism of reduction in mitochondrial priming along with drug-specific resistance mechanisms. Further, we find that, even in the context of a multiply-resistant PDX model, DBP can still identify therapeutic vulnerabilities that can be efficaciously exploited in vivo.