Background
Phase 3, randomized controlled trial nested within the UK NCRI AML17 and AML19 studies (UK, Denmark, New Zealand). Patients aged 16–60 with newly diagnosed AML and a suitable molecular marker (NPM1 mutation or fusion genes RUNX1-RUNX1T1, CBFb-MYH11, DEK-NUP214) were randomized 2:1 to sequential molecular MRD monitoring vs standard clinical care. In the monitoring arm physicians could act on MRD results at their discretion. AML17 enrolled 1836 (Jun 2012–Dec 2014) and AML19 enrolled 965 (Nov 2015–Jan 2018); 637 were randomized across both. Published Lancet Haematology 2025.
Interventions and follow up
Arm A (monitoring): Sequential molecular MRD monitoring during treatment and for 3 years post-treatment; treating physicians decide whether/how to act on MRD results (N=425)
Arm B (control): Standard clinical care only; MRD results not provided to physicians (N=212)
Primary endpoint: Overall survival
Median follow-up: 4.9 years (IQR 3.6–5.9)
Arm B (control): Standard clinical care only; MRD results not provided to physicians (N=212)
Primary endpoint: Overall survival
Median follow-up: 4.9 years (IQR 3.6–5.9)
Results
OS overall (meta-analysis AML17+AML19, primary): HR 1.11 (95% CI 0.83–1.49), P=.25 — primary endpoint NOT met
3-year OS (overall): 70% (monitoring) vs 73% (no monitoring)
3-year OS (NPM1 + FLT3-ITD subgroup): 69% vs 58%; HR 0.53 (95% CI 0.31–0.91), P=.021 — prespecified subgroup benefit
OS (NPM1 without FLT3-ITD): 69% vs 78%; HR 1.56 (95% CI 0.96–2.52) — no benefit; possible harm signal
OS (fusion-gene transcripts): 72% vs 77%; HR 1.28 (95% CI 0.80–2.18) — no benefit
3-year OS (overall): 70% (monitoring) vs 73% (no monitoring)
3-year OS (NPM1 + FLT3-ITD subgroup): 69% vs 58%; HR 0.53 (95% CI 0.31–0.91), P=.021 — prespecified subgroup benefit
OS (NPM1 without FLT3-ITD): 69% vs 78%; HR 1.56 (95% CI 0.96–2.52) — no benefit; possible harm signal
OS (fusion-gene transcripts): 72% vs 77%; HR 1.28 (95% CI 0.80–2.18) — no benefit
Adverse events
Trial design: Monitoring-strategy trial — no investigational drug toxicity; harms tracked as clinical events, not drug AEs
MRD-directed interventions: Monitoring arm received intensification, alloSCT, or treatment changes at physician discretion; possible excess mortality observed in the NPM1-without-FLT3-ITD subgroup (HR 1.56) raises concern for harm from unnecessary escalation
MRD-directed interventions: Monitoring arm received intensification, alloSCT, or treatment changes at physician discretion; possible excess mortality observed in the NPM1-without-FLT3-ITD subgroup (HR 1.56) raises concern for harm from unnecessary escalation
Conclusions
Sequential molecular MRD monitoring did not improve overall survival in the overall population of younger adults with AML (HR 1.11, P=.25) — a negative result for the primary endpoint. A prespecified subgroup analysis showed an OS benefit in NPM1 + FLT3-ITD patients (HR 0.53, P=.021), suggesting MRD monitoring may have value in this specific high-risk subgroup.
Key Limitations
Negative primary endpoint; the NPM1+FLT3-ITD subgroup finding, although prespecified, is hypothesis-generating. The intervention in the monitoring arm was uncontrolled — physicians acted on MRD as they saw fit, so the specific MRD-guided action driving any benefit cannot be identified. Possible excess harm in NPM1-without-FLT3-ITD patients (HR 1.56) is concerning. Conducted largely before venetoclax combinations and standardized contemporary MRD assays, limiting current applicability; MRD thresholds and assays varied over time.
Clinical Context
AML17/AML19 is the largest randomized trial of MRD monitoring as a therapeutic strategy in AML and delivers a negative verdict for the overall population. The NPM1+FLT3-ITD benefit is biologically plausible and aligns with ELN 2022 guidance recommending molecular MRD monitoring as prognostic in NPM1-mutated AML. The signal of possible harm in NPM1-without-FLT3-ITD patients underscores that MRD-guided treatment intensification must be validated before universal adoption; MRD remains established as prognostic rather than as a proven trigger for therapy change.