Harnessing the Therapeutic Value of Venetoclax: A Breakthrough Therapy in Acute Myeloid Leukemia

INTRODUCTION

Between 2017 and 2020, 10 agents for the treatment of different subsets of acute myeloid leukemia (AML) have been approved, most with randomized trial data showing increased overall survival (OS) (Fig 1). The most broadly practice-changing agent, in our opinion, is the BCL2 inhibitor venetoclax (VEN). Based on phase III trials, VEN combined with either azacitidine (AZA) or low-dose cytarabine (LDAC) improves OS for patients with AML either ≥ 75 years or unfit for intensive chemotherapy.^1,2 This commentary reflects lessons learned from the design of the phase III studies, controversies regarding which cytotoxic backbone to use, the role of targeted therapies, and potential for abbreviated VEN dosing in AML.

FIG 1.

Randomized survival outcomes for approved therapies in acute myeloid leukemia including midostaurin,²⁶ CPX-351,²⁷ glasdegib,²⁸ ENA,¹⁵ GO,²⁹ gilteritinib,³⁰ oral AZA (CC-486),³¹ and VEN.² Comparison of the survival curves from the VIALE-C study showing the primary analysis: (A) 7 + 3 + midostaurin, (B) CPX-351, (C) 7 + 3 + GO, (D) gilteritinib, (E) LDAC + glasdegib, (F) AZA + ENA, (G) CC-486, and (H) AZA + VEN. AZA, azacitidine; ENA, enasidenib; GO, gemtuzumab ozogamicin; HR, hazard ratio; LDAC, low-dose cytarabine; VEN, venetoclax. (A) From The New England Journal of Medicine, Stone RM, Mandrekar SJ, Sanford BL, et al, Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation, 377:454-464, Copyright © 2017, Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.²⁶ (B) From Lancet JE, Uy GL, Cortes JE, et al, CPX-351 (cytarabine and daunorubicin) liposome for injection versus conventional cytarabine plus daunorubicin in older patients with newly diagnosed secondary acute myeloid leukemia. J Clin Oncol, 36:2684-2692, 2018.²⁷ (C) Reprinted from The Lancet, 379, Castaigne S, Pautas C, Terre' C, et al, Effect of gemtuzumab ozogamicin on survival of adult patients with de-novo acute myeloid leukaemia (ALFA-0701): A randomised, open-label, phase 3 study, 1508-1516, 2012, with permission from Elsevier.²⁹ (D) From The New England Journal of Medicine, Perl AE, Martinelli G, Cortes JE, et al, Gilteritinib or chemotherapy for relapsed or refractory FLT3-mutated AML, Copyright © 2019, Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.³⁰ (E) Republished with permission of Elsevier Science & Technology Journals, from Venetoclax plus LDAC for patients with untreated AML ineligible for intensive chemotherapy: Phase 3 randomized placebo-controlled trial, Wei AH, Montesinos P, Ivanov V, et al, 135, 2137-2145, 2020; permission conveyed through Copyright Clearance Center, Inc.¹ (G) From The New England Journal of Medicine, Wei AH, Döhner H, Pocock C, et al, Oral azacitidine maintenance therapy for acute myeloid leukemia in first remission, 383:2526-2537, Copyright © 2020, Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.³¹ (H) From The New England Journal of Medicine, DiNardo C, Jonas B, Pullarkat V, et al, Azacitidine and venetoclax in previously untreated acute myeloid leukemia, 383:617-629, Copyright © 2020, Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.²

LESSONS LEARNED FROM THE STATISTICAL DESIGN OF THE VIALE-A AND VIALE-C TRIALS

Two recently published phase III studies compared VEN versus placebo (PBO) combined with either AZA (VIALE-A trial) or LDAC (VIALE-C trial) in patients with newly diagnosed AML unfit for intensive chemotherapy.^2,3 Both studies significantly improved overall response rates (ORRs) and OS for the VEN-containing arms, without excess early mortality. In VIALE-A, for the VEN arm, there was concordance between the dose finding and pivotal study stages in terms of ORR (complete remission [CR] plus CR with incomplete hematologic recovery) (67% v 66%), median OS (17.5 v 14.7 months), and 30-day mortality (3% v 7%). In VIALE-C, outcomes for VEN + LDAC were also similar between the dose-finding and phase III stages in terms of ORR (54% v 48%), median OS (10.1 v 8.4 months), and early death (6% v 13%).^1,4 In VIALE-A, OS was significantly improved by the addition of VEN to AZA (Fig 1). In the primary analysis of VIALE-C, however, VEN + LDAC did not meet its primary OS end point, with the hazard ratio (HR) for survival being 0.75 (P = .11). In a post hoc analysis with 6 months additional follow-up, the HR was 0.70 (P = .04) (Fig 2).¹ For most approved therapies in AML, the HR achieved ranged between 0.64 and 0.78 (Fig 1).

FIG 2.

(A) OS from the primary analysis with 12 months follow-up; (B) OS with 18 months median follow-up (additional 6 months in a post hoc analysis). In the primary analysis, a cluster of patients was censored between 3 and 6 months before the median OS timepoint was reached, compared to most censored events between 9 and 12 months in the updated analysis. In both analyses, there are few recorded observations at 24 months. LDAC, low-dose cytarabine; OS, overall survival; PBO, placebo; VEN, venetoclax.

A major difference between VIALE-A and VIALE-C was the inclusion of patients with prior hypomethylating agent (HMA) exposure in VIALE-C, which accounted for 20% of the study population. In VIALE-C, the HR for OS in the HMA-exposed group was 0.82 (0.39-1.7) versus 0.73 (0.49-1.1) in the HMA-naive population, suggesting that outcomes were more favorable among patients without prior HMA exposure. With the benefit of hindsight, it would have been advantageous if the VIALE-C study was stratified and powered to independently analyze survival in the HMA nonexposed group. In VIALE-C, the number of recruited patients was substantially smaller than the number enrolled to VIALE-A (211 v 431 patients), with VIALE-C pursuing a more aggressive HR for OS than VIALE-A (0.5 v 0.7). For LDAC + VEN, the phase Ib data available in August 2016 (median follow-up of 7.9 months) estimated the 12-month OS to be 64%.⁵ Recruitment to VIALE-C commenced in 2017, targeting an HR of 0.5, based on an expected improvement in the median OS from 6 to 12 months. A more mature analysis of the phase Ib LDAC + VEN study showed a median OS of 10.1 months (median follow-up approximately 20 months).⁴ Therefore, in the planning of phase III AML studies based on early-phase data with short follow-up, a more conservative HR should be considered to account for the likely overestimation of efficacy associated with analysis of survival data with limited follow-up time.

In the VIALE-C study, an unusual survival bump in the LDAC control curve was observed (Fig 2), potentially related to poststudy salvage with intensive chemotherapy, which was more frequent in the PBO + LDAC arm (22% v 8%).⁶ The impact of poststudy therapy was likely exacerbated by the relatively small number of patients in the control arm (n = 68), which in turn was amplified by the 2:1 allocation ratio in favor of the VEN arm. For studies with unequal allocation ratios, the perceived advantages of greater enrollment to investigational therapy need to be balanced by larger overall sample sizes to achieve the same level of statistical power.⁷

Recruitment to VIALE-C spanned the period between May 2017 and November 2018. The preplanned primary analysis was undertaken on February 15, 2019. At this analysis, the median follow-up time was only 12.0 months. Consequently, a large proportion of patients in the VEN-LDAC arm were censored with a follow-up duration shorter than the median OS for the study arm (Fig 2). With 6 months additional follow-up, allowing more events to be observed, the HR was 0.70 (P = .04). Historically, AML survival beyond 2 years is rare among patients treated with LDAC alone.⁸ In the phase Ib experience with VEN + LDAC, 22% of patients were alive at 2 years (median follow-up 3.5 years).⁹ Similarly, after VEN + AZA, approximately 40% of patients were alive at 2 years (median follow-up 2.5 years).¹⁰ Therefore, future trials in older AML populations should ensure that follow-up is sufficient to document extended survival outcomes. In the VIALE-C study, observations at 24 months in the LDAC + VEN arm were negligible (Fig 2). Therefore, short follow-up may limit the capacity to analyze long-term OS, an important benefit for patients receiving VEN-based therapy. Most cancer studies use proportional hazards for comparing survival curves, which assume an equal proportion of events per unit time. For therapies associated with a prolonged survival tail, the survival curve may take the shape of an ice-hockey stick, as exemplified by inotuzumab in B-cell acute lymphoblastic leukemia.¹¹ In such instances, alternative statistical methods may better appreciate these longer-term benefits, such as a late-emphasis Wilcoxon test or a restricted mean survival time analysis, which compares the difference between the areas under the curve for both treatment arms to quantify overall years of life gained for those receiving investigational therapy.¹²

With the benefit of hindsight, we suggest that future phase III studies in AML consider preplanned stratification of patient subgroups represented by very poor outcomes. Study assumptions based on extrapolation of outcomes linked to phase Ib or II studies should be conservative or associated with adequate patient follow-up. Finally, if a survival tail is expected with investigational therapy, the statistical plan should include a minimum median follow-up time, as opposed to a solely event-driven primary analysis timepoint. In addition, consideration should be given to coadoption of a late-emphasis statistical method to assist in the analysis of survival curves displaying nonproportionality.

WHICH THERAPY FOR NPM1-, IDH-, AND FLT3-MUTANT AML?

With the availability of VEN-based therapy, an important question is whether clinicians treating older and/or unfit patients with AML should wait for molecular subtyping results to make treatment decisions. Since the approval of IDH1/2- and FLT3-targeted agents, there has been a push to counsel clinicians to defer induction until molecular data become available, to potentially enable more effective treatment choices. For patients with severe leukocytosis and/or leukemia-related complications, emergency intervention is clearly indicated. Patients enrolled to clinical trials represent a selected group characterized by limited comorbidities and commonly treated at specialized centers. Clinical studies evaluating potential risks of delaying induction likely exclude patients at greatest risk of early death. In a comparison of early death among non-M3 AML, mortality (< 1 month) was lower among cases enrolled to SWOG trials than among 26,272 patients recorded in the SEER18 registry.¹³ The early rate among patients < 56 years in SWOG trials compared with SEER was 3% and 12%, whereas among patients > 75 years, the early mortality was 32% and 42%, respectively. Now that VEN-based regimens have demonstrated activity across broad molecular and cytogenetic subgroups, the potential advantages of delaying treatment in older populations must be revisited.

Among molecular subgroups without a specific targetable molecular abnormality, ORR generally favored VEN-AZA over VEN-LDAC, apart from patients with NPM1 mutation, where the ORR for VEN-LDAC was higher (79% v 67%) (Fig 3). In the phase Ib VEN-LDAC experience with long-term follow-up, patients with NPM1-mutated AML had an ORR of 89% and a 2-year OS of 64%.⁴ Similar outcomes for NPM1-mutated AML were observed for VEN + HMA.¹⁴ In phase III studies, current follow-up remains limited and the impact of NPM1 mutation on treatment outcomes remains uncertain: VEN + LDAC (HR 0.46; CI, 0.14 to 1.5) and VEN-AZA (0.73; 0.36 to 1.51). In addition, the impact of NPM1 comutations, such as FLT3–internal tandem duplication (ITD), is likely to be important, emphasizing the need for pooled patient analyses on larger data sets to provide clinical guidance regarding efficacy in molecular subgroups. Therefore, either VEN + LDAC or AZA represent reasonable options until the survival impact of VEN in NPM1-mutant cases with longer-term follow-up becomes available.

FIG 3.

Comparison of responses in the VIALE-A and VIALE-C studies. Numbers above bars represent % CR/CRi. AZA, azacitidine; CR, complete remission; CRi, CR with incomplete hematologic recovery; HMA, hypomethylating agent; LDAC, low-dose cytarabine; PBO, placebo; VEN, venetoclax.

Among patients with IDH mutations, VEN-AZA was associated with an ORR of 75% (Fig 3). Notably, the HR for survival was low (0.34), suggesting strong survival benefit for VEN-AZA.² A key question is whether there is a role for IDH-targeted therapy in the first-line setting. In an open-label study of AZA with or without enasidenib for IDH2-mutant AML, significant improvements in ORR (71% v 42%) and progression-free survival were observed in favor of enasidenib. OS, however, was not significantly different between the two treatment arms (Fig 1).¹⁵ In the AZA control arm of the enasidenib study, 21% were salvaged with enasidenib, likely confounding interpretation of OS. Furthermore, OS for patients in the AZA control arms was strikingly different between the two trials, with 2-year OS of 45% in the enasidenib study and 12% in VIALE-A.^15,16 For the investigational arms, the 2-year OS for enasidenib-AZA and VEN-AZA was 50% and 52%, respectively, suggesting that the major difference between the two studies was the behavior of the AZA control arm. A prospective study comparing enasidenib-AZA with VEN-AZA will be needed to determine the benefit of one approach over the other. With respect to IDH1 mutations, a randomized study combining ivosidenib in frontline combination with AZA is ongoing (ClinicalTrials.gov identifier: NCT03173248). For now, it is difficult to argue that waiting for IDH subtyping (outside of clinical trials) is beneficial for up-front treatment decisions in older or unfit patients. An exception may be very frail patients unable to tolerate myelosuppression related to VEN-AZA, but able to tolerate single-agent induction with ivosidenib or enasidenib. Despite improved outcomes from VEN-AZA, most will not be cured. Therefore, an important future consideration is whether improved outcomes can be achieved by combining VEN-AZA with an IDH1 or IDH2 inhibitor in so-called triplet combinations or whether the sequential or tandem use of HMAs with VEN and targeted IDH inhibitors will be more efficacious and/or better tolerated.

Among patients with FLT3-mutated AML, the ORR for VEN-AZA in VIALE-A was 70%, compared with 36% for PBO-AZA (Fig 3).¹⁷ Despite a higher ORR for VEN-AZA among patients with either FLT3–tyrosine kinase domain or ITD, OS was more convincingly prolonged among patients with FLT3–tyrosine kinase domain (median OS 19.2 v 10.0 months) than those with FLT3-ITD (median OS 11.5 v 8.5 months). Although high response rates have been reported among FLT3-mutated older patients treated with AZA combined with FLT3 inhibitors,^18,19 a phase III trial of AZA with or without gilteritinib in newly diagnosed FLT3-mutated AML ineligible for intensive chemotherapy (the LACEWING trial) was terminated for futility. It remains to be determined whether off-protocol salvage therapy with FLT3 inhibitors in the control arm nullified the benefit of gilteritinib. An alternative strategy is combination of VEN, AZA, and gilteritinib in an up-front triplet, with dose optimization to mitigate the risk of myelosuppression. Currently, outside of clinical trials, there is no benefit in delaying treatment while waiting for the FLT3 results in older or unfit patients otherwise planned for treatment with AZA-VEN. Despite the activity of AZA-VEN in patients with FLT3-mutated AML, OS is still limited, and clinical trials are strongly encouraged for this high-risk population.

OPTIMIZING VENETOCLAX DOSING DURATION

Although the US Food and Drug Administration recommends VEN for 28 days each cycle, controversy exists regarding biologic rationale for dosing VEN beyond its combination with cytotoxic therapy. Among patients responding to VEN-AZA, grade 4 cytopenia occurred in 87%, resulting in 70% of patients switching to a 21-day VEN schedule for subsequent cycles. Analysis of OS showed no difference in outcome for those switching immediately to 21-day VEN, compared with those persisting with a 28-day schedule.²⁰ To guide decision making, we recommend a marrow assessment on day 21 and ceasing VEN if blast clearance is demonstrated. A VEN-free interval enables neutrophil and platelet recovery, for example, to CR with partial hematologic recovery, before commencing the next cycle. Further guidance on management of VEN-based therapy has been published.^21,22

Truncated VEN schedules have been investigated in several combination regimens. In elderly patients with newly diagnosed AML, infusional cytarabine and idarubicin (5 + 2 schedule) combined with 7 days of VEN after a 7-day monotherapy prephase (CAVEAT study) produced a response rate of 72% in patients mainly > 65 years considered fit for modified intensive chemotherapy.²³ FLAG-Ida combined with VEN on days 1-14 achieved ORRs of 89% and 74% in patients with newly diagnosed and relapsed or refractory AML, respectively.²⁴ In treatment-naïve AML, cladribine/LDAC combined with VEN D1-21 and D1-14 in cycles 1 and 2, respectively, followed by VEN-AZA with 14-day VEN dosing produced high rates of response (93%) and flow minimal residual disease clearance (89%), supporting a growing body of evidence that abbreviated VEN schedules are effective.²⁵ Comparative studies are warranted to determine whether shorter VEN exposures (14-days or less) can maintain the positive efficacy observed in the VIALE-A and C trials, with less toxicity.

QUALITY OF LIFE

In addition to 8-week RBC- and platelet transfusion–free intervals, which were increased in the VEN arms of VIALE-A and VIALE-C, patient-reported outcomes were also studied, using the EORTC QLQ-C30, EQ-5D-5L, and PROMIS Cancer Fatigue Short Form 7a scales. Although patient-reported outcomes are frequently measured in trials and improved in the VEN-based arms, such end points are not easy to translate to patients. An alternative measure of patient impact is hospital burden associated with continuing VEN-based therapy, including days spent in hospital for treatment, managing complications, attending clinic or visiting day centers for routine monitoring, AZA injection, or transfusion support. For patients with prolonged remission, a common question is whether treatment can be ceased. Future strategies could investigate the potential of fixed interval treatment (eg, 12 months) for those in CR and who are minimal residual disease–negative. The benefit of a treatment-free interval lasting months or years would be considerable, as long as it can be shown that survival outcome is not compromised.

SUMMARY

New therapies promise a brighter future for patients with AML. The pace of change has been rapid, with clinical challenges and dilemmas often arising before the availability of published answers. We believe that it is important for clinicians to share and debate these controversies to highlight current perspectives and opinions regarding new drugs not always highlighted in scientific publications. Candidly, we hope that this brief discourse will help doctors in making informed and balanced decisions on behalf of, and in collaboration with, their patients.

AUTHOR CONTRIBUTIONS

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Harnessing the Therapeutic Value of Venetoclax: A Breakthrough Therapy in Acute Myeloid Leukemia

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).