A New CDK9 Inhibitor on the Block to Treat Hematologic Malignancies

In this issue of Clinical Cancer Research, Cidado and colleagues describe a new selective CDK9 inhibitor effective against hematologic malignancies (1). Uncontrolled proliferation and gene dysregulation are two hallmarks of cancer, which are in part regulated by cyclin-dependent kinases (CDK). These serine-threonine kinases are often overexpressed in many malignancies, making them attractive therapeutic targets. In most human cancer types, the cyclin D-CDK4/6-p16-retinoblastoma signaling pathway is commonly altered, leading to abnormal cell proliferation. In fact, CDK4/6 inhibitors like palbociclib, ribociclib, and abemaciclib are already approved by the FDA to treat advanced hormone receptor–positive breast cancer, and have shown clinical activity in lymphoma, leukemia, glioblastoma, melanoma, among others (2). Within the CDKs implicated in gene transcription, the most relevant are CDK7 and CDK9 that phosphorylate RNA polymerase II at specific serine residues; the first is implicated in the start of transcription while the latter regulates elongation. Therefore, CDK9 is an important target because, in contrast to other CDK proteins that are mostly involved in cell-cycle progression, it controls gene expression, especially of those regulated by super enhancers (3).

Many CDK inhibitors have been developed and clinically tested, but until now they were quite unspecific, in part because the kinase domains of these proteins are very similar. While the first generation of CDK inhibitors, the so-called pan-CDK inhibitors, target different CDKs and show high toxicity, the second generation is more specific and present a tolerable adverse event profile (2). Pan-CDK inhibitors targeting CDK9 like alvocidib (flavopiridol), atuveciclib, seliciclib, or dinaciclib were evaluated in multiple clinical trials focused on liquid tumors but, in most cases, they were discontinued because of significant adverse events. However, precise CDK9 inhibition would impair expression of cancer-promoting genes with short half-life mRNAs such as MYC or MCL-1, which are particularly important for most hematologic malignancies. In fact, blood cancers such as acute myeloid leukemia (AML), acute lymphoblastic leukemia, or chronic lymphocytic leukemia, often exhibit CDK9 dysregulation that promotes cancer progression (3).

AZD4573 is a highly selective CDK9 inhibitor, with an IC₅₀ in the low nanomolar range, and is at least 10-fold more specific compared with other CDK inhibitors. The authors observed that cell death was rapidly and extensively achieved in different hematologic cancer cell lines and that 6–8 hours of CDK9 inhibition was sufficient to engage maximal cell death. When comparing hematologic and solid tumor cell lines, the latter were clearly less sensitive, with geometric mean EC₅₀ and GI₅₀ >30-fold greater. These observations indicate that CDK9 inhibition could be effective to treat different forms of leukemia and lymphoma. Gene expression and proteomic analyses revealed that the antiapoptotic protein MCL-1, with a short half-life, was significantly diminished with AZD4573, while other members of the BCL-2 family remained unchanged. To further investigate that CDK9 inhibition induces apoptosis through MCL-1 downregulation, they compared AZD4573 with the specific MCL-1 inhibitor AZD5991 (4), showing a strong correlation between both agents in vitro. Indeed, a dose-dependent reduction of phosphorylated Ser2-RNAP2, MCL-1 depletion, and caspase-3 cleavage (Fig. 1) was observed after only 6 hours of exposure to this CDK9 inhibitor. Although they elegantly demonstrate this main mechanism of action, some cells were sensitive to AZD4573 but not AZD5991, pointing to another possible factor for cancer cell survival also targeted by CDK9 inhibition. To confirm that AZD4573-mediated caspase activation is engaged by the intrinsic pathway of apoptosis, they analyzed and detected mitochondrial outer membrane permeabilization, caspase 3/7 activation, and phosphatidylserine exposure upon treatment (1).

Further in vivo studies demonstrated similar effects for both AZD4573 and AZD5991 using intermittent drug dosing in MV-4-11 subcutaneous xenograft models (but not in the insensitive OCI-AML3). Moreover, an AZD4573 pharmacokinetic and pharmacodynamic evaluation on MV-4-11 xenografts demonstrated that the in vivo results were consistent with the in vitro data. The authors optimized the drug administration to maximize the therapeutic window and established that 2 days on/5 days off dosing schedules in mice sustain a progressive reduction in tumor volume. Besides MV-4-11, 10 hematologic models were evaluated with this regime, and eight of them, that were previously reported sensitive in vitro to AZD4573, achieved either regression or tumor growth inhibition. In addition to subcutaneous xenografts, they explored patient-derived xenograft (PDX) models. Five of nine AML PDXs responded to the CDK9 inhibitor showing a >50% reduction of leukemic blasts in the bone marrow, and a T-cell lymphoma PDX displayed increased apoptosis in human CD45⁺ tumor cells and significant survival benefit compared with vehicle (1). These in vivo results suggest that AZD4573 is well tolerated and could be potentially used to treat hematologic malignancies.

While AZD4573 was effective in multiple cancer models explored by the authors, similarly to what has been previously observed with BH3 mimetics, some tumors displayed resistance. Notably, OCI-AML3 did not respond, and SU-DHL-4 showed no tumor regression when exposed to anti-CDK9 monotherapy in vivo. However, AZD5991, that like AZD4573 induces apoptosis by undermining MCL-1, has been reported to produce more durable responses when combined with the BCL-2 inhibitor venetoclax (4). Indeed, they found that both in vitro and in vivo venetoclax treatment leads to MCL-1 stabilization and dependency that can be pharmacologically exploited with AZD4573. Consequently, they combined AZD4573 with venetoclax in both resistant models, detecting a significant increase in caspase activation in vitro and highly durable regressions in vivo, while either single agent was not effective as monotherapy (1). Moreover, this synergistic regimen was well-tolerated by mice with no observed body weight loss.

Overall, this interesting work by Cidado and colleagues suggests that the specific CDK9 inhibitor AZD4573, currently in a phase I clinical trial (NCT03263637), may be effective for the treatment of different hematologic malignancies, with particular promise in AML. Its mechanism of action is mostly mediated by phosphorylated Ser2-RNAP2 reduction causing MCL-1 depletion and leading to a similar effect as the BH3 mimetic AZD5991 (4). While other CDK9 inhibitors are currently being explored (5, 6), AZD4573 was well tolerated across preclinical hematologic cancer models using intermittent dosing that could maximize its therapeutic index in the clinic. Moreover, if synergistically combined with the FDA-approved BCL-2 inhibitor venetoclax, it may overcome resistance to therapy. This exciting new drug represents a promising clinical candidate to explore for the treatment of patients with different hematologic malignancies and should be further examined not only as a monotherapy but also in combination with BH3 mimetics.

J. Montero is a paid consultant for Oncoheroes Biosciences and Vivid Biosciences, and is an unpaid board member for The Society for Functional Precision Medicine. No potential conflicts of interest were disclosed by the other authors.

The authors would like to thank Drs. Davids and Pérez-Galán for their insightful comments. J. Montero is supported by Ramon y Cajal Programme, Ministerio de Economia y Competitividad (RYC-2015-18357). C. Alcon, A. Manzano-Muñoz, and J. Montero are supported by the CELLEX foundation.