BCL2 and MCL1 inhibitors for hematologic malignancies

The regulatory approvals for venetoclax herald the arrival of a new class of small-molecule cancer therapeutics that target prosurvival proteins that protect against apoptosis. Venetoclax, a potent and selective inhibitor of the intracellular protein BCL2, has received US Food and Drug Administration (FDA) approval for the treatment of chronic lymphocytic leukemia (CLL) and acute myeloid leukemia (AML). Fast on its heels in development are similarly acting drugs that target either BCL2 or proteins with comparable prosurvival function, particularly MCL1. This review focuses on key areas relevant to clinical use and development of these BH3 mimetic drugs. Underpinning preclinical data will be only selectively touched on, given previous comprehensive reviews. Neither have we attempted to summarize all early-phase trial data in hematologic diseases given the currently rapidly changing trial landscape. Rather, we distil principles that have been established and identify major questions that the field needs to address to enable BCL2 and MCL1 inhibitors to achieve maximum clinical impact.

The development this new class of antineoplastics required a deep understanding of the intrinsic (mitochondrial) pathway to apoptosis.^1-5 The pathway is orchestrated by the BCL2 network, a family of partially homologous cytoplasmic proteins, best considered as 3 functional subfamilies. Pivotal in this are the effectors of mitochondrial apoptosis: BAX and BAK.

Within a cell, BAX/BAK are kept inert by the direct and indirect actions of prosurvival family members. These proteins (BCL2, MCL1, BCLxL, BCLw, and BCL2A1) act to prevent BAX/BAK-driven apoptosis, thereby maintaining cellular viability. Physiologic stresses (eg, lack of growth signals, oncogenic stresses, DNA damage) trigger increased expression of the BH3-only proteins, the third subfamily in the network. These BH3-only proteins (eg, BIM, BAD, NOXA) act to initiate apoptosis by binding tightly to prosurvival BCL2 proteins. If present in sufficient amounts, the BH3-only proteins overwhelm the capacity of prosurvival proteins to hold BAX/BAK in check, thus allowing these effectors to oligomerize and form activated pores that damage the mitochondrial outer membrane. Certain BH3-only proteins have also been reported to activate BAX/BAK directly; however, BAX/BAK-driven apoptosis can proceed even in the complete absence of any BH3-only protein.⁶ 

BAX/BAK-driven mitochondria outer membrane permeabilization (MOMP) commits a cell toward apoptosis undermining normal mitochondrial function, including energy production, and allowing the leakage of mitochondrial contents, including cytochrome c, which trigger proteolytic enzymes (caspases) causing cellular demolition.

Most normal cells rely on multiple prosurvival BCL2 proteins to maintain their viability although in some cell types particular prosurvival proteins predominate. For example, mature B lymphocytes are highly reliant on BCL2, whereas MCL1 plays a central role in plasma cells.⁷ This is exaggerated by overexpression of specific prosurvival proteins in certain neoplasms, such as BCL2 in CLL. Regardless, once the action of the prosurvival proteins is blocked by the BH3-only proteins, BAX and BAK are freed to drive apoptosis. Conversely, mitochondrial apoptosis cannot proceed in the absence of BAX and BAK.

A cardinal feature of this cellular life-death switch is that it is underpinned by protein–protein interactions, distinct from enzymatic reactions such as phosphorylation by kinases (eg, BCR-ABL, FLT3, BTK). To initiate apoptosis, the killer BH3 domain of the BH3-only proteins engages a large, poorly defined groove on the prosurvival proteins. Unlike enzymes such as protein kinases that bind their substrates via small binding interfaces (akin to a “lock-and-key”), the BH3:prosurvival groove interactions use an “induced-fit” mechanism mediated largely by hydrophobic interactions.⁸ Moreover, multiple molecular interactions are required for tight binding across the large protein-protein interfaces.

Small molecules have been developed to mimic the action of the BH3-only proteins, hence BH3 mimetics. Such small-molecule entities have been generated to preferentially bind ≥1 prosurvival proteins: cells reliant on BCL2 will be highly sensitive to a BCL2-selective inhibitor. By mimicking the action of the BH3-only proteins, a selective chemical inhibitor will induce BAX/BAK-mediated mitochondrial apoptosis unless other prosurvival proteins are present in sufficient amounts to restrain endogenous apoptosis effectors (Figure 1).

BH3 mimetics are therefore defined by their ability to kill cells by apoptosis in a strictly BAX/BAK-dependent manner.³ BH3 mimetics targeting BCL2 (±BCLxL) or MCL1 that have progressed into clinical trials are listed in Table 1. Other than venetoclax and navitoclax, it can be appreciated that the clinical development of other BH3 mimetics is only in the earliest stages. Navitoclax⁹ was the first potent BCL2 inhibitor to enter clinical trials and demonstrated moderate single-agent activity in patients with relapsed CLL and indolent B-cell lymphomas.^10-13 However, acute reduction in platelet count is a direct legacy of its cotargeting of BCLxL, as survival of platelets is highly dependent on BCLxL.¹⁴ Consequently, thrombocytopenia proved to be the dose-limiting toxicity, precluding full exploration of BCL2 inhibition. Navitoclax remains in clinical development only for hematologic diseases where targeting of BCLxL or both BCL2 and BCLxL, is anticipated to be beneficial, such as myelofibrosis and acute lymphoblastic leukemia (ALL), respectively.^15,16

To avoid the on-target thrombocytopenia associated with navitoclax, venetoclax was designed to be a BCL2-specific inhibitor. No naturally occurring BH3-only proteins inhibit BCL2 exclusively. Venetoclax binds and inhibits BCL2 with picomolar affinity and displays greater than 100-fold selectivity for BCL2 over BCLxL and >1000-fold for MCL1.¹⁷ The demonstration that a single prosurvival protein can be inhibited and have important clinical activity turbo-charged drug discovery and development of this class. Similarly, MCL1 can now be selectively targeted by potent BH3 mimetics.^18-22 

Dependence on BCL2 is high and relatively uniformly so in CLL. BCL2 is required but not essential for normal B-cell development, and mature B cells normally express high levels of BCL2.²³ CLL cells in all patients express high levels of BCL2 and typically much lower levels of MCL1 or BCLxL.^23-27 CLL bearing del(13q) have loss of mir15/16, negative regulators of BCL2 expression, with enhanced levels of BCL2 expression.^28,29 In vitro, CLL cells are highly sensitive to BCL2-selective inhibition (Figures 1 and 2).^17,30,31

Venetoclax was first approved by the FDA as monotherapy for relapsed or refractory del(17p) CLL in April 2016. It now has multiple listings internationally for relapsed/refractory disease in combination with rituximab and for unfit patients with newly diagnosed CLL in combination with obinutuzumab. Monotherapy studies in relapsed/refractory disease indicated an immediate reduction in CLL burden in almost all patients. Objective responses occurred in 79% of patients, complete remissions (CRs) in 20% and undetectable measurable residual disease (uMRD) in up to 30%^32-34 with acceptable toxicity.³⁵ Indeed, the dose-limiting toxicity (DLT) was biochemical tumor lysis, with several patients experiencing severe or fatal clinical tumor lysis syndrome (TLS).³² Patients with high burden disease or reduced renal function were at highest risk. This necessitated introduction of a standard 4-week ramp-up phase commencing at 20 mg/day and rising incrementally each week to the target dose of 400 mg/day. In patients, apoptosis of CLL cells was readily detectable within 6 to 8 hours of dosing,³¹ consistent with the rapid induction observed in vitro and the timing of peak plasma venetoclax concentrations (6 hours).³² Importantly, venetoclax is highly active after prior therapy with chemo-immunotherapy, BTK inhibitors, or phosphatidylinositol 3-kinase inhibitors.^32,34,36,37

Apparently higher CR (51%) and uMRD (57%) rates were observed in early-phase trials combining rituximab with venetoclax,³⁸ prompting this combination to be compared in a randomized trial against bendamustine-rituximab in relapsed/refractory CLL. Venetoclax-rituximab demonstrated superior efficacy and safety; respective hazard ratios (HRs) for progression-free survival (PFS) and overall survival (OS) were 0.17 and 0.48, with a number-needed-to treat (NNT) of 2 for PFS at 2 years.^39,40 Despite its now uniform adoption as the standard venetoclax regimen for relapsed/refractory CLL, in the absence of a randomized comparison vs venetoclax monotherapy, the evidence for additive benefit from rituximab is modest and confined to improvement in CR rate in a multivariable analysis of pooled trial data.⁴¹ Uncontrolled real-world data analyses of PFS do not suggest additive benefit,⁴² and it seems most probable that venetoclax is delivering the majority of the observed clinical efficacy.

In first-line treatment of patients unfit for fludarabine-based regimens, venetoclax-obinutuzumab proved superior to chlorambucil-obinutuzumab with respect to PFS (HR, 0.31; NNT of 3 at 3 years), with similar toxicity and OS.⁴³ As previously observed for fludarabine-cyclophosphamide-rituximab chemo-immunotherapy, the durability of benefit with venetoclax-based regimens correlated with achievement of deeper responses (measured either by CR or uMRD rate).^38,40,41,44 In the salvage setting, the probability of achieving uMRD with continuing venetoclax plateaus around 2 years^34,40,41,44 and earlier in the first-line setting.⁴⁴ This has provided a rational basis for the introduction of regimens with fixed-duration venetoclax (2 years for relapsed/refractory, 1 year for first-line) in registration studies rather than continuous use which may not provide additional benefit.⁴⁵ 

However, long-term follow-up data indicate an ongoing risk of relapse with these venetoclax-based regimens, and it seems improbable that these treatments are curative. Accumulated trial data now enable analyses of which factors either modify the response to venetoclax and/or predict for particular benefit from the use of venetoclax-based regimens over chemo-immunotherapy and/or are prognostic with venetoclax-based regimens. In relapsed CLL, the validated genetic biomarkers most relevant to clinical practice are del(17p), TP53 mutation, and immunoglobulin heavy chain variable region (IGHV) mutational status.⁴⁶ Table 2 summarizes the key findings from randomized trials. Other factors have been associated with relatively reduced response rates or inferior duration of response compared with their absence (bulky lymphadenopathy >10 cm; refractoriness to ibrutinib; NOTCH1 mutations) in multivariable analyses of pooled data from single arm trials,⁴¹ but these have not yet been confirmed in prospective trials.

Abnormalities of TP53 function are known negative response-effect modifiers for DNA-damaging therapy in CLL (and other hematologic malignancies), characterized by consistent associations with inferior CR, uMRD, PFS, and OS rates in trials and mechanistic studies in model systems.⁴⁶ With venetoclax ± anti-CD20 antibodies, no negative effect is observed for overall response, CR, or MRD clearance rates or with CLL killing in vitro.^{31,39,43,47,48} However, the presence of a TP53 abnormality before therapy is associated with an increased rate of relapse, compared with CLL with normal TP53, and remains prognostic.^41,47,48 The magnitude of benefit for venetoclax-based regimens over chemo-immunotherapy is greater for patients with TP53-aberrant CLL than observed for those with TP53 wild-type CLL.^47,48 Similarly, patients with unmutated IGHV CLL have inferior outcomes with chemo-immunotherapy compared with those whose CLL has mutated IGHV. Although unmutated IGHV is a negative treatment effect modifier for chemo-immunotherapy, it is not for venetoclax-based regimens, and its presence predicts for major benefit from the use of venetoclax.^41,47 Both mutated and unmutated IGHV CLL have similar rates and depths of response. However, unmutated IGHV may remain a negative prognostic factor (compared with mutated IGHV; Table 2), in long-term follow-up of approved venetoclax-based regimens.⁴⁷ 

Robust data are also emerging about the molecular basis for treatment failure in CLL (Table 3). Early progression (within the first year) is uncommon and closely associated with complex karyotype and Richter transformation in heavily pretreated patients.^49,50 Secondary resistance in patients receiving continuous venetoclax can be caused by several independently occurring mechanisms (Table 3; Figure 3). These include mutations in BCL2 (eg, Gly101Val), overexpression of MCL1, and overexpression of BCLxL.^51-54 All have been validated functionally, with each reducing the intrinsic sensitivity of CLL cells to venetoclax by 1.5 to 3 logs in both patient cells and model systems. BCL2 mutations reduce binding of the drug but retain pro-survival function of the protein.^51,55 Heightened expression of MCL1 or BCLxL provides cells with sufficient alternative prosurvival function to avoid apoptosis in the presence of venetoclax. It is common for relapse to be polyclonal, with independent mechanisms evident in different subclones. For example, as well as the 2 more common BCL2 mutations (Gly101Val and Asp104Tyr), 6 other BCL2 mutations have been described, with many patients demonstrating several independent subclones, each characterized by a different mutation.⁵⁶ How the BCL2 mutations are generated remains unknown. The great majority have never been observed outside the context of secondary venetoclax resistance. MCL1 overexpression can be the consequence of amplification of the MCL1 locus on chromosome 1q21 but can also occur in the absence of gene amplification.⁵³ The presumed epigenetic mechanisms leading to MCL1 overexpression in those circumstances and BCLxL overexpression in other patients’ cells remain to be elucidated. With most patients treated with venetoclax now completing time-limited therapy in objective response and early data suggesting that re-exposure to the drug induces secondary responses in 12 of 17 cases reported in 2 prospective studies to date,^45,48 further studies are required to understand whether the landscape of resistance will be different with repeat exposure.

BTK inhibitors (BTKi) are the other class of targeted novel drugs that have revolutionized care in CLL. Like venetoclax, they are highly effective as monotherapy and against TP53-aberrant disease.^57,58 Unlike venetoclax, they are not rapidly cytotoxic; rather CLL cells die of neglect because of blockade of B-cell receptor and chemokine receptor signaling.⁵⁸ Indefinite continuous rather than fixed-duration therapy appears required. Preclinical data indicate synergy between BTKi and BCL2 inhibition, and circulating CLL cells from patients taking ibrutinib are sensitized to in vitro killing by venetoclax.^59-61 Together with the largely nonoverlapping toxicity profiles, these findings made exploration of the combination of BTKi and venetoclax logical in CLL (and mantle cell lymphoma [MCL]). Phase 1/2 trials indicate 90% to 100% rates of response and very high levels of CRs (50% to 88%) and uMRD (60% to 70%) in both the first-line and relapsed settings.^62,63 Phase 3 trials comparing various combinations of venetoclax with a BTKi ± anti-CD20 antibody are ongoing, including randomized studies vs chemoimmunotherapy in fit patients.

AML is a more heterogeneous disease than CLL, representing neoplastic transformation of either hematopoietic stem cells (HSCs) or myeloid progenitor cells. BCL2 expression varies across its spectrum and can be heterogeneous even within a patient’s leukemic cell population. Genetic evidence indicates that BCL2 is not essential for HSC or progenitor cell function, whereas MCL1 is required for HSC function⁶⁴ and for expansion of AML in model systems.⁶⁵ Nevertheless, ABT-737 had in vitro activity against primary patient samples.⁶⁶ Subsequent studies revealed venetoclax and ABT-737 to be equipotent in killing AML cells, supporting the dominant prosurvival role played by BCL2.⁶⁷ Venetoclax was also shown to induce apoptosis more effectively in leukemic compared with normal CD34⁺ HSCs and progenitor cells, suggesting a therapeutic window.⁶⁸ In the first study of venetoclax monotherapy in relapsed/refractory AML, CRs were observed in 19%.⁶⁹ These responses, however, were not durable, in contrast to those seen in CLL.

Combination therapy is therefore necessary, and the first regimens trialed were those incorporating conventional treatments for patients unfit for intensive induction therapy. Azacitidine can enhance venetoclax activity against AML through TP53-independent NOXA induction after activation of the integrated stress response pathway,⁷⁰ whereas genotoxic drugs do so through TP53-mediated upregulation of NOXA and PUMA.^67,71

In November 2018, venetoclax was granted accelerated approval for use in adults with newly diagnosed AML ≥ 75 years of age or with comorbidities precluding intensive chemotherapy. Phase 1b/2 studies had shown that venetoclax in combination with either low-dose cytarabine or hypomethylating agents produced high rates of CR (54%-67%) in patients with a median age of 74 years.^72,73 These remissions were achieved rapidly and with low early (30-day) mortality (3% to 6%), consistent with the profile expected for a low-intensity therapy. In October 2020, venetoclax was granted full approval after 2 confirmatory phase 3 trials, conducted in the same unfit patient population, demonstrated a significant survival benefit over placebo for the addition of venetoclax to azacitidine (VIALE-A; HR 0.66, NNT of ∼5 at 2 years) or to low-dose cytarabine (VIALE-C).^74,75 In VIALE-C, improvement in survival was only significant after a post hoc longer-term follow-up analysis. Therefore, venetoclax-azacitidine has been widely adopted as the preferred first line therapy for older and unfit AML patients. In these patients, the major toxicities relate to myelosuppression which is increased and can be prolonged.

With the graduation of venetoclax into the routine management of AML, a current priority is the identification of clinically relevant correlates of response and resistance to guide patient selection and inform design of new drug combinations aimed at subverting drug resistance or salvaging patients with disease progression.

However, biomarkers of primary response to venetoclax in AML are not straightforward to delineate, as AML has diverse genetics, and the drug has not been widely used as monotherapy. Two clinical studies have examined bone marrow blast responses to venetoclax monotherapy. In relapsed/refractory AML, blast reductions were greatest in AML bearing IDH, SRSF2, or ZRSR2 mutations.⁷⁶ In previously untreated AML, a 7-day prephase of venetoclax monotherapy suppressed marrow blasts most effectively in NPM1, IDH, and SRSF2 mutant AML.⁷⁷ Preclinical studies have suggested the presence of a HOX gene expression signature,^78,MLL-X fusion, RAD21, GATA2, or WT1 mutations to correspond with venetoclax sensitivity.^78,79 It has yet to be determined, however, how these entities perturb AML biology in favor of venetoclax responsiveness, and they have not been studied prospectively in trials. Importantly, the presence of either IDH1 or IDH2 canonical mutations is a bone fide treatment effect modifier in the context of venetoclax-azacitidine treatment. Consistent with preclinical discoveries and early signals from the phase 1 study, IDH-mutant AML is relatively highly susceptible to BCL2 inhibition compared with other AMLs and compared with azacitidine therapy alone (Table 2).^69,74,80

In terms of secondary venetoclax resistance, longitudinal studies have identified some factors associated with adaptive resistance (Table 3). Selection of clones enriched for FLT3-ITD or mutant TP53 has been found at the time of AML relapse.⁸¹ The addition of exogenous growth factors reduces sensitivity to venetoclax in vitro and activating mutations in growth factor signaling pathways (eg, FLT3-ITD) appear to drive expression of prosurvival proteins, including MCL1, which can mediate resistance to venetoclax and other cytotoxic drugs.⁸¹ Unlike in CLL, BCL2 mutations impairing venetoclax affinity have not been reported in patients with relapsed AML. Loss-of-function CRISPR (clustered regularly interspaced short palindromic repeats) screens have identified TP53 and BAX as candidate loss-of-function routes to venetoclax resistance, but exactly how TP53 dysfunction achieves this remains uncertain.⁸² In part, it is explained by recent data demonstrating that TP53 loss impairs BAX/BAK activation, conferring a competitive advantage for subclones with aberrant TP53 during ongoing BCL2 (or MCL1) inhibition.⁸³ This appears particularly important when BH3 mimetic concentrations are suboptimal and resolves the apparent disconnect between similarly high response rates in TP53-aberrant and TP53-sufficient AML and CLL but inferior long-term benefit in TP53-aberrant leukemias. Emergent BAX mutations have been identified recently in patients with AML relapsing after venetoclax and also in the myeloid compartment of patients with clonal hematopoiesis receiving venetoclax for treatment of CLL.^84,85 Higher constitutive expression of MCL1 in monocytes may also result in lineage-associated adaptive resistance during venetoclax therapy (Table 3).⁸⁶ Other reported associations with venetoclax resistance in AML in vitro include increased expression of alternative prosurvival proteins, such as BCLxL, MCL1, BCL2A1, downregulation of proapoptotic BH3-only members PUMA or NOXA, or activating mutations affecting other kinases in the MAPK pathway, such as RAS or PTPN11.^78,82,87,88 However, whether these lesions do cause resistance in patients in vivo is currently unknown.

Previous studies have identified MCL1 to be amplified or overexpressed in AML and models conditionally targeting MCL1 have confirmed its prosurvival role in certain subsets of AML, including MLL-X–driven disease.⁶⁵ Xenograft model systems demonstrated synergistic antileukemic activity in vivo when venetoclax was combined with an inducible vector targeting MCL1.⁶⁷ Although a subset of primary AML samples also appear sensitive to MCL1 inhibition with BH3 mimetics, monotherapy activity in vitro against primary AML is generally modest.^18,89 However, combined targeting of BCL2 and MCL1 (using S63845,¹⁸ AMG176,¹⁹ AZD5991,²² or VU661013⁹⁰) has been reported by 4 independent groups to be synergistic against AML, including TP53-aberrant AML,⁸³ confirming the importance of these 2 prosurvival targets in sustaining survival of a diverse spectrum of leukemic blasts (Figures 1 and 2). Although toxicity from dual inhibition may be a barrier, to date, lethal damage to normal tissues in these models has not been observed, perhaps suggesting that clinical development of such combinations may be feasible.^19,89

Other opportunities to improve venetoclax efficacy in AML relate to the apparent reliance of leukemic stem and progenitor cells (LSPCs) on glycolytic metabolism as an energy source.⁹¹ Following BAX/BAK-induced MOMP, the oxidative phosphorylation pathway is disrupted, including in cells incompletely committed to apoptosis (Figure 1). Adding drugs that compromise amino acid metabolism⁹² or transportation of fatty acids into the cell can deprive AML of alternative energy sources to compensate for the effect of venetoclax-azacitidine on oxidative phosphorylation in LSPCs.⁹³ In fit patients with poor prognosis AML, venetoclax is now being combined with more intensive cytotoxic regimens. Preliminary data indicate higher initial response rates but also more pronounced hematopoietic toxicity, particularly with repetitive cycles of therapy.⁷⁷ 

Neoplastic plasma cells not only typically express high levels of BCL2 but commonly express BCLxL and MCL1.⁹⁴ Indeed, MCL1 is required for normal plasma cell development.⁷ In vitro testing of primary cells indicate broad activity of BCL2 inhibitors against myeloma, with marked sensitivity observed for myeloma bearing t(11;14).^94-96 BCL2 inhibition as treatment of patients with myeloma remains experimental. Although durable remissions in the t(11;14) subset have been observed with venetoclax monotherapy,⁹⁷ combination therapy appears necessary in most patients.⁹⁸ The most advanced combination, venetoclax with bortezomib-dexamethasone, has been investigated in a placebo-controlled randomized trial in relapsed/refractory disease.⁹⁹ Venetoclax substantially improved PFS but not OS, at a cost of increased morbidity and some excess mortality in patients with myeloma progression (Table 4). Increased risk of infections was an important contributor to this excess toxicity and current trials in multiple myeloma include specific recommendations on how this risk can be mitigated. t(11;14) is clearly a positive response-effect modifier (Table 2). Ongoing trials are required to identify how this drug can be incorporated into well tolerated regimens with enhanced efficacy.

B-cell lymphomas by virtue of their lineage commonly express BCL2.¹⁰⁰ Overexpression is driven in some through translocation (eg, t[14;18]),¹⁰¹ in follicular lymphoma and some double-hit lymphomas, gene amplification (eg, some diffuse large B-cell lymphomas [DLBCL]),¹⁰² or epigenetic dysregulation (eg, MCL).¹⁰³ High-level MCL1 expression is common in MYC-driven lymphomas such as Burkitt lymphoma.¹⁰⁴ 

BCL2 inhibition as treatment of patients with lymphoma remains experimental. Early-phase trials established modest single agent activity in MCL, follicular lymphoma, and Waldenström macroglobulinemia, but minimal activity in DLBCL.¹⁰⁵ Various combination regimens, rational for each disease, are being tested in ongoing trials. Published results for completed phase 2 trials are summarized in Table 4. The results of randomized trials of R-CHOP (rituximab-cyclophosphamide-doxorubicin hydrochloride-vincristine sulphate-prednisone), with or without venetoclax for first-line high risk DLBCL and ibrutinib, with or without venetoclax, in relapsed MCL are awaited with interest.

Single-agent activity has also been observed in small series of patients with diseases with high levels of BCL2 expression: blastic plasmacytoid dendritic cell neoplasm,¹⁰⁶ prolymphocytic leukemia,¹⁰⁷ and t(11;14) light chain amyloidosis.¹⁰⁸ In relapsed acute lymphoblastic leukemia, venetoclax in combination with navitoclax (to inhibit BCLxL) and nonmyelotoxic chemotherapy is being explored, with a phase 1b study reporting a high response rate.¹⁰⁹ 

As listed in Table 1, there are multiple specific MCL1i in early-stage clinical development. Based on preclinical data and lessons learned from selective BCL2 inhibition with venetoclax, several predictions can be made for selective MCL1i undergoing clinical trials. First, apoptosis of susceptible cells is likely to be rapid, with biological activity evident within hours after exposure. TLS can be anticipated to occur occasionally, especially in patients with bulky or proliferative forms of myeloma, lymphoma, or acute leukemia. Second, "on target" hematopoietic toxicities, including neutropenia, monocytopenia, and depletion of plasma cells (with consequent hypogammaglobulinemia), are likely to occur in patients receiving antitumoral doses.^19,110 Gene ablation models have also highlighted the biologically important role played by MCL1 in cardiomyocyte,^111,112 neuronal,¹¹³ hepatocyte,^114,115 and intestinal cell survival.¹¹⁶ Consequently, collateral toxicities that should be anticipated in clinical trials include biochemical derangement of liver and cardiac enzymes and gastrointestinal symptoms. Patients should be screened for pretreatment hepatic and/or cardiac abnormalities and monitored closely for evidence of biochemical damage after commencing therapy. As MCL1i are likely to be less well tolerated than BCL2i, intermittent schedules (eg, weekly or bi weekly) are being explored in initial first-in-human trials. Third, objective responses with monotherapy should be observed in myeloma,^18,19,117 AML,^{18,19,22,65,90} and MYC-driven lymphomas,^18,104 but these may not be durable. Although some exceptional responders are likely to be seen, at this point, no putative biomarkers have been validated. Candidate biomarkers of response include amplification of MCL1 on chromosome 1q21¹¹⁸ and monocytic differentiation in AML.^86,119

Analogous to venetoclax in various blood cancers (CLL, AML), aberrant TP53 is unlikely to diminish response to MCL1i but durability of benefit is likely to be reduced.⁸³ Consequently, combination therapy will be required.^18,19,22,90 As with venetoclax, potential partners should be tailored to specific diseases and can be anticipated to include monoclonal antibodies , tyrosine kinase inhibitors (TKIs), hypomethylating agents, and conventional cytotoxics. Furthermore, once tolerable monotherapy regimens demonstrating biological activity without DLTs are established, it seems appropriate to commence assessment of novel combinations. Based on a narrower therapeutic window than venetoclax, it is probable that MCL1i will most likely succeed if efficacy only requires brief bursts of exposure to maximal tolerated doses.

To date, the only clinical trial data reported for MCL1 inhibition are preliminary results from the first 26 patients with relapsed refractory myeloma enrolled to the first-in-human trial of intravenous AMG-176. Emergent side effects were mostly hematologic (neutropenia and anemia) and gastrointestinal (nausea and diarrhea). An antitumor effect was reported in 3 patients (1 CR and 2 partial responses).¹²⁰ Enrollment to another trial investigating the orally administered MCL1 inhibitor AMG-397 in various blood malignancies was placed on clinical hold after a cardiac toxicity signal was seen. Clinical trial data from all MCL1i are awaited to determine how appropriate patient selection and dosing schedules can result in identification of a safe and efficacious clinical pathway for these new BH3 mimetic drugs.

For highly susceptible diseases (eg, CLL with BCL2 inhibition), the outstanding questions relate to (1) the optimum duration of treatment: is this for a fixed duration or adapted to the tempo and depth of MRD clearance, and (2) whether triplet regimens combining venetoclax with BTKi and monoclonal antibodies add substantially to durability of disease clearance? The latter question also applies for mantle cell lymphoma. Long term, it is conceivable that combinations of BCL2i and BTKi may be curative for some patients with CLL. For less susceptible diseases (eg, BCL2-expressing DLBCL), the critical question is whether routine addition of venetoclax adds to the cure fraction currently achievable with chemo-immunotherapy alone. This question is also of great interest in patients with AML suitable for standard intensive chemotherapy. BH3 mimetics have long promised to augment efficacy when combined with DNA-damaging cytotoxics in preclinical models, and it seems these 2 clinical settings will inform us as to whether the inevitable increase in hematologic toxicity is outweighed by substantially higher proportions of patients with durable responses. In AML, another important question is whether treatment should continue until disease progression or whether limited-duration therapy should be considered, similar to CLL. This is especially pertinent in subgroups such as IDH-mutated leukemia and where MRD is absent.

As outlined above, the immediate research imperative for MCL1 inhibitors is definition of safe and biologically active regimens with preliminary evidence of efficacy in relapsed and refractory disease settings. The field is expecting that some clinically important dose-limiting toxicities will emerge necessitating use of submaximal doses in combination regimens in order to limit exposure at peak concentrations while maximizing clinical efficacy. Priorities for phase 2 exploration should be guided by where the preliminary clinical data suggest the greatest impact may be, with likely candidate areas including relapsed AML, first-line AML bearing highly unfavorable genomic features, multiply relapsed MYC-driven lymphomas, and relapsed myeloma bearing amplifications of chromosome 1q21 (containing the MCL1 locus).

The potential of BH3 mimetics as anticancer drugs is only beginning to emerge. These rapidly acting, non-DNA, damaging cytotoxics may fill key gaps in our current armamentaria against hematologic malignancies. Lessons learned thus far form sound principles for future therapeutic endeavors. The small, potential risk of tumor lysis syndrome, as well as bone marrow and other organ toxicity, mandates that new combinations are carefully tested in clinical trials. As long as therapeutic safety margins can be maintained, BH3 mimetic agents have the potential to further improve clinical outcomes in multiple hematologic malignancies.

The authors thank their many long-term collaborators for advice and important contributions to the generation of knowledge included in this review.

Academic research in the A.W.R., A.H.W., and D.C.S.H. laboratories has been supported over the last 20 years by grants from the National Health and Medical Research Council of Australia , the Leukemia and Lymphoma Society, Cancer Council of Victoria, Victorian Cancer Agency, Australian Cancer Research Foundation, Leukaemia Foundation of Australia, the Snowdome Foundation, and the Medical Research Future Fund.

Contribution: A.W.R., A.H.W., and D.C.S.H. conceived, reviewed the literature, and wrote the manuscript.

Conflict-of-interest disclosure: A.W.R. and D.C.S.H. are employees and A.H.W. is a former employee of the Walter and Eliza Hall Institute, which receives milestone and royalty payments related to venetoclax; each receives a share of these royalties from the Institute. A.W.R. has received research funding to his institutions from Abbvie, Janssen, and Servier for investigator-initiated clinical trials or laboratory research. A.H.W. has served on advisory boards for Novartis, Janssen, Amgen, Roche, Pfizer, Abbvie, Servier, Celgene-BMS, Macrogenics, Agios, and Gilead; receives research funding to the institution from Novartis, Abbvie, Servier, Celgene-BMS, Astra Zeneca, and Amgen; and serves on speaker bureaus for Abbvie, Novartis, and Celgene. D.C.S.H. has received research funding from Genentech and Servier.

Correspondence: Andrew Roberts, The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, VIC 3052, Australia; e-mail: roberts@wehi.edu.au.