Gemtuzumab ozogamicin in (KMT2A)-rearranged adult acute myeloid leukaemia (AML) in the UK Medical Research Council AML15 and AML16 trials.

Abstract

A recent report from the USA Children’s Oncology Group (COG) acute myeloid leukaemia (AML)0531 trial1 described a surprisingly marked improvement in survival outcomes for children with (KMT2A)-rearranged AML who received gemtuzumab ozogamicin (GO) with induction chemotherapy compared with those who did not. As these rearrangements occur in approximately 5–10% of adults with AML,2 we examined the outcomes of patients with KMT2A-rearrangements in the UK Medical Research Council (MRC) AML15 and AML16 trials, which predominantly enrolled adults (up to and over the age of 60 years respectively) and included a randomisation between administration of GO or not with induction chemotherapy. The major outcomes of these trials have been previously reported3, 4 and meta-analysis (n = 2228 patients) showed benefit of GO in patients with favourable- and intermediate- but not adverse-risk cytogenetics.4 These findings were confirmed in a subsequent meta-analysis of five randomised trials of GO in patients with newly diagnosed AML.5 Patients with KMT2A-rearrangements fall in the intermediate- or adverse-risk cytogenetic groups depending on the fusion partner. The GO dosing schedules were comparable between the AML0531 and AML15/16 studies, using a single dose of 3 mg/m2 in cycle one with some patients in AML15 and AML0531 receiving a second dose in cycle three. In AML15, 908 patients with available cytogenetics were randomised to receive or not one dose of GO with induction cycle one, and 38 of these (4·2%) had a KMT2A-rearrangement. In AML16, there were a further eight of 866 (1%) for a total of 46 patients with a KMT2A-rearrangements participating in a GO randomisation. The majority of these (41/46, 89%) were in the intermediate-risk cytogenetic group according to the MRC cytogenetic classification6 (this differs from the European Leukaemia Network classification, which places more KMT2A-rearranged patients in the adverse-risk group).7 The median (range) age of these patients was 39.5 (6–720) years (one patient was aged <16 years). The 5-year relapse-free survival (RFS) for patients with KMT2A-rearrangements who did and did not receive GO was 45% and 36% respectively [hazard ratio (HR) 0·91, 95% confidence interval (CI) 0·41–2·02, P = 0·8, Fig 1]. This is in contrast to the results of the paediatric cohort reported by Pollard et al.,1 who identified a very large effect of GO in children with KMT2A-rearrangements: 5-year disease-free survival for patients achieving complete remission (CR) after course one of therapy in this study was 57% and 33% respectively, for a HR of 0·50 (95% CI 0·32–0·78, P = 0·002). While the effect on survival was marginally non-significant (HR 0·66, 9%% CI 0·44–1·01, P = 0·056), the CIs are wide and the result here is inconclusive, with either a minimal or a large effect of GO on survival both within the range of the CI. The effect of GO on outcome according to cytogenetic group in AML15 and AML16 is shown in Fig 2. Importantly, while there was a significant interaction between KMT2A-rearranged and other intermediate-risk cytogenetics for achievement of CR (P = 0·03 for interaction, i.e. patients with KMT2A-rearrangements receiving GO had a lower rate of CR), there was no heterogeneity in the effect of GO on RFS or overall survival (OS) between KMT2A-rearranged and other intermediate-risk groups. Comparing patients with KMT2A-rearrangements and those with favourable-risk cytogenetics (who had the greatest benefit from GO in meta-analysis)4, 5 showed a significant interaction for CR (P = 0·008) and OS (P = 0·04), indicating that patients with favourable-risk cytogenetics benefit more from GO than patients with KMT2A-rearrangements. However, there was no evidence of heterogeneity for RFS (P = 0·6). In the cohort there were 22 transplants (GO n = 10; no GO n = 12), of which 13 were in first remission 7vs6; analyses here showed results consistent with the whole cohort (HR for RFS censored at transplant 0·73, 95% CI 0·27–1·96, P = 0·5; OS censored at transplant HR 1·80, 95% CI 0·55–5·87, P = 0·3). Overall, these data show that in a predominantly adult cohort, the effect of GO in patients with a KMT2A-rearrangement is clearly smaller than that seen in a paediatric cohort, and also smaller than that seen in adults with favourable-risk cytogenetics. On the other hand, there is no evidence that the effect of GO in patients with KMT2A-rearrangements is different to that in patients with other intermediate-risk cytogenetics, who benefit from the addition of GO (HR 0·85, 95% CI 0·74–0·96). Supported by the UK Medical Research Council, UK National Institute for Health Research, Cancer Research UK and Blood Cancer UK. Alan K. Burnett and Nigel H. Russell performed the research. Robert K. Hills performed the statistical analyses. Richard Dillon wrote the first draft of the manuscript. All authors contributed to the final manuscript.