Molecular monitoring versus standard clinical care in younger adults with acute myeloid leukaemia: results from the UK NCRI AML17 and AML19 randomised, controlled, phase 3 trials.

BACKGROUND In patients with acute myeloid leukaemia treated with curative intent, the detection of measurable residual disease (MRD) generally confers a poor prognosis. This study aimed to identify whether altering treatment based on MRD results can improve survival. METHODS In the UK NCRI AML17 and AML19 randomised, controlled, phase 3 trials, performed in the UK, Denmark, and New Zealand, we screened patients aged 16-60 years with newly diagnosed acute myeloid leukaemia for molecular markers suitable for disease monitoring, including NPM1 mutations and fusion genes. Patients with a marker were randomly assigned (2:1) to either sequential molecular MRD monitoring during treatment and for 3 years after, or standard clinical care only with no molecular monitoring. In the monitoring group, treating physicians decided whether and how to incorporate the MRD results into the patient's therapy, including in cases of MRD relapse. The primary endpoint was overall survival. Prespecified subgroup analysis of the primary outcome included analysis by molecular group (NPM1mut with FLT3-ITD, NPM1mut without FLT3-ITD, and fusion gene transcripts). Both trials were registered with ISRCTN, ISRCTN55675535 and ISRCTN78449203, and are completed. FINDINGS In the AML17 trial, 1836 patients were enrolled between June 1, 2012 and Dec 31, 2014. In the AML19 trial, 965 patients were enrolled between Nov 9, 2015, and Jan 23, 2018. 637 patients were randomly assigned across both trials (289 to MRD monitoring and 144 to no monitoring in AML17 and 136 to MRD monitoring and 68 to no monitoring in AML19). With a median follow-up time of 4·9 years (IQR 3·6-5·9), overall survival at 3 years was 70% (95% CI 66-75) in patients in the monitoring group and 73% (68-80) in patients in the no-monitoring group. Meta analysis of the two studies showed no difference in overall survival (hazard ratio [HR] 1·11, 95% CI 0·83-1·49; p=0·25). In the pre-specified subgroup analysis of the primary endpoint, overall survival at 3 years in patients with both NPM1 and FLT3 internal tandem duplication (ITD) mutations was 69% (95% CI 60-79) in the monitoring group and 58% (45-74) in the no-monitoring group (HR 0·53, 95% CI 0·31-0·91; p=0·021). However there was no difference in survival by randomisation in patients with NPM1 mutations without FLT3-ITD (overall survial 69% [95% CI 62-77] in the monitoring group and 78% [70-87] in the no monitoring group; HR 1·56, 95% CI 0·96-2·52) or those with fusion gene transcripts (overall survial 72% [95% CI 65-79] in the monitoring group and 77% [68-87] in the no monitoring group; HR 1·28, 95% CI 0·80-2·18). INTERPRETATION Sequential molecular MRD monitoring, coupled with MRD-guided treatment, did not improve overall survival in the entire study population; however, in the subgroup of patients with baseline NPM1 and FLT3 ITD mutations, we observed a survival benefit for MRD monitoring. FUNDING National Institute for Health Research, Blood Cancer UK, and Cancer Research UK.