Ivosidenib or venetoclax combined with hypomethylating agents in IDH1-mutated acute myeloid leukemia: a real-world study Open Access

Acute myeloid leukemia (AML) is a molecularly heterogeneous disease commonly affecting older adults, with a median age at diagnosis of 69 years.^1,2 Within this older patient population, use of standard intensive induction chemotherapy using anthracycline and cytarabine is associated with increased toxicity,³ even within genetic subgroups typically considered favorable risk. For newly diagnosed (ND) patients with certain molecular mutations, both “targeted” and “mutation-agnostic” frontline combination regimens are available in the United States and certain other countries.

Approximately 6% to 16% of patients with ND-AML have mutations in the isocitrate dehydrogenase 1 (mIDH1) gene, resulting in the transcription of a mutant IDH1 protein that generates 2-hydroxyglutarate, an oncometabolite that promotes epigenetic dysregulation and subsequent myeloid maturation arrest.^4,5 Ivosidenib (IVO) is an orally available, IDH1-targeted drug approved by the US Food and Drug Administration (FDA) for use as a single agent or in combination with hypomethylating agents (HMAs) in frontline AML treatment, as well as monotherapy in relapsed/refractory disease.

Venetoclax (VEN) combined with the HMAs azacitidine (AZA) or decitabine (DEC) has emerged as a standard-of-care treatment for patients with ND-AML.⁶ However, post hoc analyses demonstrated relatively poor complete remission (CR) rates (27.3%) and overall survival (OS; median, 10.2-15.2 months) in patients with ND-AML and mIDH1 treated with VEN+HMA.^7,8 In contrast, in an international, multicenter, randomized, placebo-controlled phase 3 trial, IVO combined with AZA demonstrated superior efficacy compared with AZA monotherapy in patients with mIDH1, with 47% of patients achieving CR and a resultant median OS of 29.3 months in long-term follow-up.⁹ These data suggest that combinations incorporating targeted therapy in molecularly actionable subgroups may enhance response and survival, consistent with the recent European LeukemiaNet (ELN) recommendation of IVO+HMA as the standard of care for patients with mIDH1.¹⁰ 

Direct prospective and retrospective comparisons between these regimens in patients with mIDH1 have not been performed to date. This comparative retrospective analysis is, to our knowledge, the first to evaluate a large, real-world cohort of patients with mIDH1 ND-AML to understand treatment patterns, efficacy, and safety characteristics of these 2 frontline treatment regimens.

This retrospective cohort study included adult patients across the United States with ND-AML and mIDH1 deemed ineligible for intensive chemotherapy by the treating physician and treated with IVO+HMA or VEN+HMA in community and academic settings between November 2021 and November 2022. Hematologists and medical oncologists were invited to complete a standardized web-based form for up to 5 of their patients meeting the study’s inclusion criteria, beginning with the most recent eligible case. The study was conducted in a double-blind manner: the physician was unaware of the study sponsor and the third-party firm that collected these data; similarly, these entities were unaware of the physicians' identities.

Baseline patient and practice characteristics, mutational testing timing and results, treatment patterns and rationale, and efficacy and safety outcomes were collected. Categorical variables were tested for differences using χ² test, and continuous variables were tested using Mann-Whitney U test. Due to prior reports of varied treatment schedules for VEN, the duration of VEN treatment per cycle was also captured, and safety and efficacy analyses were performed between subgroups receiving differing VEN durations per cycle.

End points required a minimum of 6 months of follow-up and included CR, composite CR/CR with incomplete hematologic recovery (CRi) rate, proportion of patients proceeding to allogeneic hemopoietic cell transplantation (HCT), selected anticipated safety and tolerability measures, and acute care episodes. Event-free survival (EFS) was defined as not having an event of death, relapse, or treatment failure (no CR by 24 weeks) and was compared using Kaplan-Meier survival analysis for the proportion of events and logistic regression to adjust for covariates. The covariates used for adjusted analyses included patient age, performance status (Eastern Cooperative Oncology Group performance score 0-1 vs 2-4), ELN (2017 version) cytogenetic risk (favorable vs others), and antecedent hematologic disorders (myelodysplastic neoplasms, myeloproliferative neoplasms, secondary AML, and myelodysplasia-related mutations). All statistics were conducted using SAS/STAT software, version 9.4 of the SAS System for Windows.

Two-hundred and eighty patients were included in the final analysis (IVO+HMA, n = 181; VEN+HMA, n = 99). Baseline characteristics are displayed in Table 1. Between patients treated with IVO+HMA and VEN+HMA, no significant difference was observed with respect to baseline characteristics including age (median, 61 [range, 18-89] vs 65 years [range, 18-86]; P = .122), non-White race (34.8% vs 42.4%; P = .208), median bone marrow blasts at diagnosis (35% vs 37%; P = .185), HMA backbone (AZA, 78.5% vs 71.7%; P = .207), and performance status (Eastern Cooperative Oncology Group performance score 0-1, 79% vs 72%; P = .170). Fewer patients treated with IVO+HMA were classified as AML with myelodysplasia-related changes (13% vs 24%; P = .014) and dependent on red blood cell transfusions (55.2% vs 68.7%; P = .028). In response to direct questions on the basis of treatment choice, the most common reason was the efficacy profile (86.7% for clinicians selecting IVO+HMA vs 85.9% for those selecting VEN+HMA; P = .837); clinicians reported the biggest differential driver for selecting VEN+HMA was hospital protocol (31.3% vs 16.6%; P = .004).

Overall, 77 patients receiving IVO+HMA and 26 receiving VEN+HMA achieved CR (42.5% vs 26.3%; P = .007; Figure 1).^7,9 IVO+HMA and VEN+HMA resulted in composite CR/CRi in 63.0% (114/181) and 48.5% (48/99) of patients, respectively. After adjusting for differences in baseline characteristics, IVO+HMA retained significance for improved CR rates compared with VEN+HMA (P = .018) and demonstrated a trend toward improved CR/CRi rates (P = .075).

The median time to first bone marrow biopsy on treatment was similar between patients receiving IVO+HMA and VEN+HMA at 52 and 59 days, respectively. The median time to best response was shorter for IVO+HMA than for VEN+HMA (3.3 vs 4.1 months; P = .006). Although AZA was the predominant HMA backbone used, no significant differences in outcomes were observed between those receiving IVO+AZA vs IVO+DEC or those receiving VEN+AZA vs VEN+DEC. Improvement in performance status permitted transition to HCT in 22 patients (8%). More patients treated with IVO+HMA successfully transitioned to HCT than with VEN+HMA (11% [n = 19] vs 3% [n = 3]; P = .029; Figure 1).

With a median follow-up time of 7.4 months overall (IVO+HMA, 7.1 months; VEN+HMA, 8.8 months; P = .096), the 6-month EFS was higher for IVO+HMA than VEN+HMA (55.8% vs 38.4%; P = .006; Figure 1). After adjusting for differences in baseline characteristics between treatment cohorts, a statistically significant advantage favoring IVO+HMA remained (P = .015).

Rates of prespecified selected anticipated toxicities of grade $≥$3 were similar between cohorts, except for higher rates of febrile neutropenia within 30 days of treatment initiation for VEN+HMA than IVO+HMA (8.1% vs 1.7%; P = .008; Table 2 and 3). Unscheduled acute care use (defined as emergency department use or inpatient hospitalization) occurred for 38.6% of patients receiving IVO+HMA in the first 12 weeks vs 70.7% for VEN+HMA (relative risk, 0.55; 95% confidence interval, 0.44-0.69; adjusted P < .001).

Among the 280 included patients, only 5 in the IVO group and 6 in the VEN group had changes from the initial dose or schedule (apart from any planned VEN initial ramp-up). Treatment discontinuation was 37% for each of the 2 regimens.

The length of VEN administration was also assessed, demonstrating a minority of patients (n = 22 [22%]) received the full FDA-approved 28 days of VEN during the 28-day cycles (Figure 2); 4 (4%) received 22 to 27 days, 31 (31%) received 8 to 21 days, and 42 (42.4%) received ≤7 days of VEN per cycle. Patients with short VEN schedules had similar or healthier baseline status and similar proportion of patients aged <65 years compared with those with full schedules; they were more often treated in academic settings. CR rates were lower in patients receiving ≤7 days vs 22 to 28 days of VEN per cycle (9.5% vs 46.2%; P = .001). However, composite CR/CRi rates did not differ significantly based on the duration of VEN per cycle. Time to best response was longer in patients receiving VEN ≤7 days per cycle than in patients receiving 22 to 28 days of VEN per cycle (4.8 vs 2.7 months; adjusted P = .039), and short-duration VEN was associated with a trend toward decreased 6-month EFS rates (26.2% for ≤7 days vs 50.0% for 22-28 days; adjusted P = .087; Table 4).

No significant differences in grade $≥$3 adverse event rates were observed between patients receiving ≤7 vs 22 to 28 days of VEN per cycle (Table 3), although patients receiving ≤7 days of VEN per cycle had higher rates of unscheduled acute care within 12 weeks (90.5% vs 50.0%; P < .001) of treatment initiation. Use of a VEN duration of ≤7 days vs the FDA-approved schedule appeared to be a preemptive modification at treatment initiation; clinicians more often cited safety and underlying patient characteristics as the reason for this modification vs a VEN duration of 22 to 28 days (78.6% vs 46.2%; P = .006).

Median turnaround time (TAT) for mIDH1 results was 7 days in each cohort (Figure 3); two-thirds of patients had results by day 11 (IVO+HMA) and day 9 (VEN+HMA). Once mIDH1 testing was available, the median time from test results to treatment was 1 day (IVO+HMA) and 4 days (VEN+HMA), resulting in a median time from diagnosis to treatment of 14 days for IVO+HMA vs 21 days for VEN+HMA (P = .021). Median TATs were similar for academic vs community sites, but the time from diagnosis to treatment was longer in academic settings (19 vs 13 days). Seventy-two patients (72%) treated with VEN+HMA initiated treatment on or after the receipt of IDH1 test results; the remaining patients initiated treatment after a median of 6 days before mutational test results became available.

Two frontline therapeutic regimens are now available for patients with ND-AML and mIDH1. In this comparative, real-world analysis, we demonstrate, to our knowledge, for the first time that IVO+HMA compared with VEN+HMA results in superior CR rates and prolonged EFS. IVO+HMA also resulted in a faster time to response, greater likelihood of proceeding to HCT, and fewer infectious complications and acute care visits when used as frontline therapy in this patient population.

VEN+HMA therapy represents a mutationally agnostic standard-of-care regimen for patients with AML; however, outcomes vary across molecularly defined subgroups.^6,IDH1 and IDH2 mutations have been shown to confer an increased dependency on BCL2, translating into enhanced clinical efficacy in patients with mIDH1/2 AML when analyzed in combination.¹¹ However, variations in efficacy appear to exist across IDH isoforms. In a post hoc analysis of the initial phase 1b and phase 3 studies of VEN+HMA, the superior efficacy in patients with ND-AML and mIDH1/2 was driven predominantly by the mIDH2 subgroup, with CR and partial hematologic recovery (CRh) rates of 56% and 80%, respectively. Response rates were lower in patients with mIDH1 (CR, 27.3%; CR/CRh, 60.4%).⁷ Additionally, in patients receiving VEN+HMA, the median OS in those with mIDH1 was 15.2 months, whereas it was not reached in those with mIDH2. This analysis supports these findings of lower CR rates (26.3%) in patients with mIDH1 treated with VEN+HMA.

Both CR and EFS are potential surrogate end points corresponding with improved survival in patients with AML.¹² In patients receiving lower-intensity targeted therapies or VEN+HMA, a benefit continues to be observed for attainment of a true CR compared to lesser responses such as CRh or CRi.¹³ In this analysis, IVO+HMA therapy improved CR rates vs VEN+HMA (43% vs 26%; adjusted P = .018). Similarly, 6-month EFS rates were improved with IVO+HMA vs VEN+HMA (56% vs 38%; adjusted P = .015). These efficacy results for IVO+HMA in this real-world study are consistent with those reported in the AGILE registrational trial.⁹ Results for VEN+HMA are consistent with those reported by Abaza et al¹⁴ for 204 patients treated in a real-world setting at 8 US academic institutions; when those results were corrected to report on an intention-to-treat basis, composite CR/CRi was 52.0%, well below the levels reported in VIALE-A and similar to findings here.

Despite IVO’s mechanism of action working predominantly through differentiation, the median time to best response was shorter for IVO+HMA than VEN+HMA (3.3 vs 4.1 months; P = .006), which was slightly shorter than that reported in the phase 3 AGILE study (median time to CR, 4.3 months, range 1.7-9.2). The comparative time to response of IVO+HMA relative to VEN+HMA may be due to the shortened number of VEN treatment days per cycle received by many patients in the VEN+HMA arm.

Short-schedule VEN+HMA, reported here and in other US real-world evidence (RWE) studies as making up 60% to >75% of VEN use, has been used with the aim of mitigating toxicity and improving tolerability of the full-schedule regimen. This RWE study shows impaired effectiveness with short schedules of VEN, compromising CR and EFS proportionate to schedule length, with results significantly below the efficacy observed with IVO+HMA; however, it is recognized that OS (unstudied in this analysis) is the ultimate measure of efficacy. Similar studies comparing VEN schedules only reported on the composite end point CR/CRi.¹⁵ Thus, it remains unclear whether reduced CR rates and shortened EFS were observed in those studies. This shortened schedule may be in part responsible for the lower overall toxicity level observed with VEN+HMA (vs that reported in the phase 3 VIALE-A trial) and highlights clinician efforts to optimize efficacy while mitigating myelosuppression from VEN. However, no significant safety improvements were observed with short-schedule VEN, a finding consistent with other RWE evaluations.¹⁶ Despite the modified schedule of VEN, patients receiving VEN+HMA continued to experience higher early rates of febrile neutropenia and a greater need for unscheduled acute care visits than those receiving IVO+HMA.

Bazinet et al previously demonstrated that receipt of longer VEN durations corresponded to improved response and survival vs shorter VEN durations per cycle, despite OS not differing overall between ≤7 days vs 21 to 28 days of VEN.¹⁶ Whether these data apply to mIDH1 or these findings are driven predominantly by the IDH2 subgroup is unknown. However, the 29.3-month median OS in the AGILE study suggests IVO+AZA as a preferred option.^9,16 

The time required for receipt of mutational testing is often cited as an obstacle to initiation of molecularly targeted therapy in AML, with concern surrounding increased mortality while awaiting test results. Recent analyses demonstrate no difference in survival when delaying treatment to obtain diagnostic test results in both fit and more frail patients.^17,18 In this analysis, the median time from diagnostic testing to results was 7 days in both cohorts. Although longer than the TAT proposed by ELN 2022 guidelines (3-5 days), these results suggest that for many clinicians and sites of care, the TAT to aid in therapy selection does not pose an increased risk of morbidity or mortality. Both IVO+HMA and VEN+HMA cohorts experienced a considerably longer time from AML diagnosis to treatment initiation (14 vs 21 days, respectively), suggesting the current TAT is sufficient to guide therapeutic decision-making without delaying treatment; institutions with TAT >10 days may want to examine their processes to move more in line with national averages.

Our study has several limitations, including its retrospective design. Because the study was retrospective, there exists the possibility of underreporting of adverse events. We attempted to mitigate this by providing a prespecified list of targeted adverse events. It is also important to recognize that treatment responses in real-world data are not formally and independently adjudicated because they are in registrational trials such as AGILE and VIALE-A. Finally, although we did not see any significant differences in terms of baseline characteristics between the treatment groups, there is a possibility that patients receiving VEN+AZA may have certain high-risk characteristics, which could contribute to worse outcomes. We attempted to mitigate this through multivariate regression by controlling for characteristics that were numerically different and clinically relevant.

In a large, balanced cohort of 280 patients with mIDH1 ND-AML, those treated with IVO+HMA had higher rates of CR, attained responses sooner, had greater rates of HCT, and had improved EFS compared to those treated with VEN+HMA, with less acute care utilization. Many patients receiving VEN had preemptively shortened durations of VEN per cycle starting cycle 1 day 1, and short-duration VEN was associated with poor rates of CR without clearly improving significant toxicities. The marked efficacy observed for IVO+HMA in this real-world assessment was consistent with the efficacy demonstrated in the AGILE clinical trial of IVO+AZA and supports the recently published 2024 ELN guidelines that stratify patients with mIDH1 as favorable risk when treated with IVO+AZA.¹⁰ This enhanced efficacy of IVO+HMA, combined with its favorable toxicity profile, suggests that IVO+HMA should be considered as the preferred standard-of-care treatment for patients with ND-AML and mIDH1, and calls for more robust and rapid mIDH testing in ND-AML at those institutions unable to meet the national average TAT.

This study is funded by Servier Pharmaceuticals LLC. Medical writing support was provided by Envision Pharma Group (Fairfield, CT) and was funded by Servier Pharmaceuticals LLC. T.W.L. is a Scholar in Clinical Research of the Leukemia & Lymphoma Society.

Contribution: All authors reviewed the manuscript for important intellectual content, provided final approval of the version to be submitted, and agree to be accountable for all aspects of the manuscript.

Conflict-of-interest disclosure: C.A.L. received consultancy fees from COTA Healthcare; acted on the advisory board for Rigel Pharmaceuticals, Servier, Astellas, Syndax, Bristol Myers Squibb (BMS), and AbbVie; and receives research funding from AbbVie. B.D.S. received consultancy fees from Azurity, Enliven, Servier, and Takeda. G.B. was an employee of Servier at the time the study was conducted. A.A. is an employee of Servier. R.P. and E.P. are employees of Putnam Associates. T.W.L. received consultancy fees and honoraria from Servier and Agilix; received research funding from Jazz Pharmaceuticals, Duke University, Deverra Therapeutics, and American Cancer Society; received patents and royalties from UpToDate; is a current equity holder in private companies, Dosentrx and Thyme Care; received honoraria and speaker bureau fees from Incyte; received consultancy fees and honoraria from Pfizer, Novartis, Meter Health, Lilly, Genentech, Flatiron, Carevive, Blue Note, and BeiGene; received consultancy fees, honoraria, research funding, and speaker bureau fees from BMS/Celgene; received consultancy, honoraria, and research funding from AstraZeneca; received consultancy fees, honoraria, and speaker bureau fees from Astellas and Agios/Servier; received consultancy fees, honoraria, research funding, and speaker bureau fees from AbbVie; and received research funding from Leukemia and Lymphoma Society, National Institute of Nursing Research/National Institutes of Health, and Seattle Genetics. A.J.A. received an honorarium from Astellas.

Correspondence: Curtis A. Lachowiez, Knight Cancer Institute, Hematology and Medical Oncology, Oregon Health & Science University, 3181 SW Sam Jackson Park Rd, Portland, OR 97239; email: lachowie@ohsu.edu.