Oral azacitidine maintenance therapy for patients with acute myeloid leukaemia and myelodysplasia‐related changes: Post hoc analysis of the QUAZAR AML ‐001 trial

Approximately 25%–34% of patients with acute myeloid leukaemia (AML) have AML with myelodysplasia-related changes (AML-MRC), which generally includes AML arising from myelodysplastic syndrome (MDS) or MDS/myeloproliferative neoplasm (MDS/MPN), de novo AML in patients with specific MDS-related cytogenetic abnormalities or MDS-related gene mutations.1-4 Secondary AML (sAML) can arise from previous myeloid disease (including MDS/MPN) or may arise following exposure to cytotoxic therapy.1, 2, 4 AML-MRC and sAML are associated with poor outcomes, lower response to standard therapy and a high risk of relapse.3, 5 The phase 3 QUAZAR AML-001 trial (NCT01757535) established the role of oral azacitidine (Oral-AZA) maintenance therapy in AML.6-8 Oral-AZA significantly prolonged overall survival (OS) and relapse-free survival (RFS) compared with placebo in patients with AML in first remission following intensive chemotherapy (IC) who were ineligible for haematopoietic stem cell transplantation (HSCT).7, 8 The OS benefit was maintained up to 5 years (median OS: Oral-AZA, 24.7 months vs. placebo, 14.8 months; p < 0.001), and conversion from detectable to undetectable measurable residual disease (MRD) was significantly higher with Oral-AZA (37%) versus placebo (19%).8, 9 This post hoc analysis evaluated outcomes of patients with AML-MRC versus those with non-AML-MRC who received Oral-AZA or placebo in the QUAZAR trial.7 The QUAZAR trial details have been previously reported.7 In brief, eligible patients were ≥55 years of age with AML and intermediate- or poor-risk cytogenetics, Eastern Cooperative Oncology Group (ECOG) performance status of 0–3, who were in first remission after chemotherapy and ineligible for HSCT. Enrolled patients were randomised after achieving complete remission (CR; <5% bone marrow [BM] blasts) following induction chemotherapy; this made genomic assessments challenging due to the low number of BM blasts. Patients received Oral-AZA 300 mg daily or placebo for 14 days in 28-day cycles. MRD status was assessed by multiparameter flow cytometry (MRD-positive threshold of ≥0.1%)9 in patient BM aspirates. This analysis included patients from QUAZAR with AML-MRC (n = 91) or sAML (n = 8 secondary to MDS or chronic myelomonocytic leukaemia; n = 2 therapy-related), hereinafter collectively referred to as AML-MRC (n = 101). In QUAZAR, AML-MRC was classified using the World Health Organization (WHO) 2008 criteria.10 This analysis uses a prior AML-MRC classification3, 11, 12 since comprehensive molecular profiling as per the WHO 2022 criteria4 was not feasible due to the historic nature of the cohort (study period between 2012 and 2019) and the unavailability of diagnostic samples for retrospective next-generation sequencing (NGS). Between-group differences in characteristics at diagnosis and at randomisation were assessed using Fisher's exact test. OS was defined as the time from randomisation to death from any cause after censoring for HSCT, and RFS from randomisation until relapse or death. OS and RFS were estimated using Kaplan–Meier methods and were compared using unstratified log-rank tests. The hazard ratios (HRs) between treatments with 95% confidence intervals (CIs) were estimated using Cox proportional hazards models. Univariate analyses of prognostic factors for OS (censoring for HSCT) and RFS among patients with AML-MRC were also performed. The duration of MRD negativity was calculated from the first assessment for patients who were MRD-negative at baseline or achieved MRD negativity on the study and was analysed using Cox proportional hazards models. p-values were estimated using an unstratified log-rank test. This analysis evaluated 101 patients with AML-MRC (AML-MRC, n = 91; sAML, n = 10) and 371 patients with non-AML-MRC at baseline (Table 1). Fewer patients with AML-MRC had NPM1 mutations (15.8% vs. 32.8%; p < 0.001) or FLT3-ITD mutations (7.9% vs. 15.7%; p = 0.052; Table 1). Significantly more patients with AML-MRC had the WHO 2022 del(5q) (11.5% vs. 2.7%; p = 0.002) and monosomy 7/del(7q) (10.3% vs. 4.2%; p = 0.035), and more were categorised as poor risk (19.8% vs. 12.4%; p = 0.074) based on National Comprehensive Cancer Network (NCCN®) Clinical Practice Guidelines in Oncology (NCCN Guidelines®) 2011.13 Median platelet count was significantly lower for patients with AML-MRC than for those with non-AML-MRC (137 vs. 172 × 109/L; Table 1). Thirteen (AML-MRC) and 35 (non-AML-MRC) patients underwent HSCT after the study drug was discontinued. Approximately half of patients in both the AML-MRC group (n = 56/101, 55.4%) and the non-AML-MRC group (n = 182/371, 49.1%) were randomised to Oral-AZA. Characteristics were generally similar between treatment groups and AML-MRC status, with some notable differences. In the Oral-AZA arm, patients with AML-MRC had lower median platelet counts than those with non-AML-MRC (135 vs. 164 × 109/L; p = 0.0062) and higher rates of CR with incomplete haematological recovery (CRi) (33.9% vs. 17.6%; Table S1). In the placebo arm, more patients with AML-MRC than with non-AML-MRC were MRD-positive at randomisation (64.4% vs. 46.0%; p = 0.05). Among patients with AML-MRC, five in the Oral-AZA arm and eight in the placebo arm underwent HSCT after a median of 6.0 (range, 1–33) and 13.5 (range, 2–24) maintenance cycles, respectively. In patients with AML-MRC, Oral-AZA significantly prolonged OS versus placebo (median, 19.9 vs. 14.8 months; HR, 0.59 [95% CI, 0.36–0.94]; p = 0.0261; Figure 1A) at a median follow-up of 17.4 (range, 0.6–81.2) and 12.2 (range, 1.1–55.1) months, respectively. Median RFS was also significantly longer with Oral-AZA versus placebo (7.5 vs. 3.7 months; HR, 0.57 [95% CI, 0.36–0.88]; p = 0.0111; Figure 1A). No significant differences in OS in patients with AML-MRC versus non-AML-MRC were observed in either the Oral-AZA or placebo arms (Figure 1B). However, RFS was significantly shorter for patients with AML-MRC versus non-AML-MRC receiving either Oral-AZA (7.5 vs. 10.5 months; HR, 1.45 [95% CI, 1.01–2.08]; p = 0.0430) or placebo (3.7 vs. 4.9 months; HR, 1.56 [95% CI, 1.11–2.21]; p = 0.0109; Figure 1B). Among patients with AML-MRC, univariate analysis showed that NPM1 mutations, intermediate-risk cytogenetics (vs. poor risk) and prior consolidation therapy (vs. none) were all associated with significantly prolonged OS with Oral-AZA versus placebo (Table S2). The same variables were predictive of RFS, in addition to CR status (vs. CRi) at randomisation (Table S2). Among patients who achieved MRD negativity on study, those receiving Oral-AZA maintained MRD negativity for a significantly longer duration versus placebo in both the AML-MRC (8.1 vs. 0 months; HR, 0.47 [95% CI, 0.29–0.76]; p = 0.0032; Figure S1A) and non-AML-MRC groups (12.0 vs. 5.2 months; HR, 0.64 [95% CI, 0.48–0.84]; p = 0.0012; Figure S1B). This post hoc analysis of QUAZAR showed that, in patients with AML-MRC, Oral-AZA significantly prolonged OS, RFS and duration of MRD negativity versus placebo. Patients with AML-MRC have an unfavourable prognosis, and accordingly, this analysis showed a significantly shorter median RFS for patients with AML-MRC than for those with non-AML-MRC. In patients with AML-MRC, MRD negativity was 55.4% (Oral-AZA) and 31.1% (placebo) at baseline in this analysis, similar to a real-world study by Rautenberg et al. reporting MRD-negative status in 64% of patients with AML-MRC after induction therapy.14 Consistent with previous reports,2 unfavourable karyotypes were prevalent in patients with AML-MRC, while NPM1 mutations were rare. While several adverse-risk genetic mutations associated with AML-MRC are included in the 2022 WHO and International Consensus Classification systems,4 comprehensive molecular data were not available for this analysis. BM multilineage dysplasia may predict the presence of myelodysplasia-related mutations in >75% of normal-karyotype AML,15 supporting the diagnostic value of BM morphological assessment. This may be especially important where NGS is not available and AML classification relies on morphological and cytogenetic data. While secondary-type mutations may result in inferior patient outcomes, there is no consensus on the impact of NPM1 mutations on survival outcomes in patients with AML-MRC.SR1 In this analysis, normal karyotypes and NPM1 mutations appeared to maintain prognostic relevance for improvements in OS and RFS in the context of Oral-AZA treatment for AML-MRC. Similarly, achievement of CR after induction and having received ≥1 consolidation cycle significantly reduced the probability of relapse after Oral-AZA treatment. The prognostic value of flow cytometry for MRD assessment has been recently demonstrated in secondary AML following IC.14 In this line, Oral-AZA maintained MRD negativity for a significantly longer time in both AML-MRC and non-AML-MRC groups.SR2 We observed delayed relapse in the Oral-AZA group versus placebo among patients who were MRD-negative and became MRD-positive, suggesting a role for Oral-AZA in controlling the expansion of the AML clone. A limitation of this analysis is the use of a prior AML-MRC classification that primarily uses cytogenetic and clinical criteria, rather than the more recent WHO 2022 criteria,4 which includes comprehensive molecular profiling, meaning the 2022 European LeukemiaNet (ELN) risk stratification could not be applied. At the time, the WHO 2022 panel genes were not tested routinely and mutation data beyond NPM1 and FLT3 were limited in the site-reported mutational results. While this limitation reflects the standard of care at the time of the study period, we recognise that future studies incorporating expanded molecular data will be essential for accurate disease classification, risk assessment and therapeutic guidance in AML-MRC. Despite AML-MRC representing a more challenging population to treat with a particularly unfavourable prognosis, this analysis of the QUAZAR trial shows the potential of Oral-AZA to improve outcomes in these patients. Although limitations include lack of availability of secondary genetic mutation data and an imbalance of baseline MRD positivity rate in the AML-MRC group, the improved survival benefit supports the use of Oral-AZA as maintenance therapy in patients with AML-MRC. M.T.V. and G.J.R. participated in writing the original draft, study conception and design, data acquisition and data interpretation. S.d.B., M.P. and A.F.A. participated in data acquisition and data interpretation. K.W., W.L.S., M.U.G. and D.L.d.M. participated in data analysis and data interpretation. E.P. and T.P. participated in study conception and design, data acquisition, data analysis and data interpretation. All authors participated in reviewing the draft and approving the submitted and final versions and are fully responsible for its content. We would like to acknowledge the patients and families who made this study possible, and the clinical study teams who participated in the study. The study was supported by Celgene, a Bristol Myers Squibb Company. Writing and editorial assistance were provided by James Matthews, PhD; Emma Rathbone, PhD; and Jeff Frimpter, MPH, of Excerpta Medica, funded by Bristol Myers Squibb. This study was sponsored by Bristol Myers Squibb. M.T.V. reports speaker's bureau participation for AbbVie, Astellas Pharma, Bristol Myers Squibb, Jazz Pharmaceuticals and Servier; research funding from Bristol Myers Squibb and Novartis; consultancy for Bristol Myers Squibb and Jazz Pharmaceuticals. S.d.B. reports honoraria from AbbVie, Astellas Pharma, Bristol Myers Squibb, Jazz Pharmaceuticals, Loxo and Servier; consultancy for Bristol Myers Squibb, GlaxoSmithKline, Remix, Servier and Syndax; speakers' bureau participation for AbbVie, Astellas Pharma, Bristol Myers Squibb, Jazz Pharmaceuticals and Servier; research funding from Auron and Forma; travel expenses from AbbVie and Servier. M.P. reports consultancy for AbbVie, AOP Orphan Pharmaceuticals and Bristol Myers Squibb; speaker's bureau participation for Bristol Myers Squibb and Sandoz; travel expenses from Alexion Pharmaceuticals and Sobi; honoraria and research funding from Bristol Myers Squibb. A.F.A. reports honoraria from Bristol Myers Squibb. K.W. reports employment and stock ownership from Bristol Myers Squibb. W.L.S. reports employment, travel expenses and patents, royalties and other intellectual property from Bristol Myers Squibb. M.U.G. reports employment, stock ownership, honoraria and patents, royalties and other intellectual property from Bristol Myers Squibb. D.L.d.M. reports employment, stock ownership and patents, royalties and other intellectual property from Bristol Myers Squibb. E.P reports employment and travel expenses from Bristol Myers Squibb. T.P. reports employment and stock ownership from Bristol Myers Squibb. G.J.R. reports consultancy for AbbVie, Amgen, AstraZeneca, Astellas Pharma, Bristol Myers Squibb, Caribou Biosciences, Celgene, Daiichi Sankyo, Ellipses Pharma, Genoptix, Geron, Gilead Sciences, GlaxoSmithKline, GlycoMimetics, Janssen, Jasper Pharmaceuticals, Jazz Pharmaceuticals, Molecular Partners, MorphoSys, NeoGenomics, Novartis, OncoPrecision, OncoVerity, Pfizer, Rigel Pharmaceuticals, Roche, Syndax Pharmaceuticals and Telix Pharmaceuticals; research funding from Janssen. This study was conducted in accordance with the principles of the Declaration of Helsinki. The protocol was approved by an institutional review board or ethics committee at each participating site. All patients participating in the QUAZAR AML-001 trial provided written informed consent. NCCN makes no warranties of any kind whatsoever regarding its content, use or application and disclaims any responsibility for its application or use in any way. The Bristol Myers Squibb policy on data sharing may be found at https://www.bms.com/researchers-and-partners/clinical-trialsand-research/disclosure-commitment.html. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.