From prognostication to precision in acute myeloid leukemia

In this issue of Blood, Levis and colleagues¹ use results from highly sensitive measurable residual disease (MRD) testing of adults in first remission from acute myeloid leukemia (AML) to confirm a highly quantitative relationship between peritransplant FLT3-ITD MRD levels and clinical outcomes after allogeneic hematopoietic cell transplantation (alloHCT) and to show how such testing can identify those most likely to benefit from additional posttransplant treatment with gilteritinib, an FLT3 inhibitor. This important post hoc data analysis moves MRD-guided therapy for patients with AML from an appealing theoretical concept to a practical and immediately applicable, although imperfect, reality.

BMT CTN 1506 (“MORPHO,” NCT02997202) was a phase 3 randomized controlled trial in adults with FLT3-ITD AML in continuous first remission after intensive therapy with either gilteritinib (120 mg once daily) or placebo for 24 months after alloHCT.² No statistically significant difference was observed between these arms for the primary study end point of relapse-free survival (RFS). Given its known prognostic importance, MRD status was tested as part of the protocol, using a polymerase chain reaction–based DNA-sequencing assay for FLT3-ITD on remission marrow, and included as 1 of 3 key stratification factors (along with conditioning intensity and time from transplant to randomization). However, MRD status was not part of the primary end point analysis and was mentioned in the protocol secondary end points only generically in terms of prognostication (“to examine the effect of pre- and posttransplant MRD on RFS and OS”). Nevertheless, in a subgroup analysis, the initial clinical report noted that the benefit of gilteritinib appeared restricted to those who tested positive for MRD.² This new study extends this finding by further exploration enabled by the rich quantitative data resulting from the protocol-mandated use of this highly sensitive MRD assay.

Patients with FLT3-ITD AML in remission are commonly offered alloHCT as a final consolidative therapy. Posttransplant pharmacological treatment to reduce relapse risk is not universally given to such patients despite published evidence of efficacy.³ Each individual drug treatment decision must balance benefit against risk; for patients in remission from FLT3-ITD AML, only 25% to 30% will ordinarily experience relapse in the posttransplant setting,⁴ yet “maintenance” approaches ask all to tolerate additional treatment and place themselves at risk of suffering adverse effects. Molecularly targeted therapy is not without cost, beyond financial toxicity, in this case off-tumor but likely on-target toxicity from the sequelae of myelosuppression was common. Although gilteritinib does have a side-effect profile that appears to compare favorably with another FLT3 inhibitor previously tested,³ in the BMT CTN 1506 study patients randomized to gilteritinib were more likely to experience adverse events and nonrelapse mortality than those randomized to placebo.² 

Detection of persistent FLT3-ITD by DNA-sequencing MRD methods in patients with FLT3-ITD AML in remission before alloHCT is known to identify a subset of patients at significantly higher risk of posttransplant relapse and death compared with those testing negative,^4-6 opening up the potential for MRD-guided treatment after transplant limited to only those with the greatest unmet clinical need. An earlier BMT CTN study suggested conditioning intensification in younger patients could partially mitigate the negative impact of AML MRD in those undergoing alloHCT,⁷ but that intervention is not available or appropriate for all patients. In contrast, gilteritinib is relatively well-tolerated across age groups and is specifically targeted to the persistent disease biology identified by the MRD test used in this study. Although “MRD eraser” aspirations are currently fanciful, the collection of high-quality FLT3-ITD MRD data during BMT CTN 1506 allowed the authors to evaluate both the impact of MRD burden on posttransplant outcomes and whether MRD test status could be used as a predictive biomarker.

This new analysis confirmed both that FLT3-ITD MRD is associated with increased relapse and decreased RFS,^4-6 and that higher burdens of FLT3-ITD MRD are associated with higher risks of relapse after alloHCT,⁸ with the greatest risk being observed in those with an FLT3-ITD variant allele fraction of ≥0.01% (“MRD4”). A prior retrospective cohort study using the same MRD method showed that approximately one-third of patients tested positive in pretransplant blood,⁸ whereas pretransplant testing of marrow here identified ∼46% as MRD positive. The optimal MRD sampling and testing strategy will continue to undergo refinement, but it is already clear that no level of detectable FLT3-ITD MRD in remission before alloHCT can be considered “safe.” A reduction in relapse rates with gilteritinib compared with placebo was seen across all MRD levels. Differences in nonrelapse mortality between the treatment arms, however, resulted in the summative clinical RFS benefit of gilteritinib only being observed in those testing MRD-positive (ie, those with the most potential to benefit given the high rates of relapse on placebo alone).

Gilteritinib is already US Food and Drug Administration approved and is available in the United States for patients with relapsed and refractory AML. Given that the FLT3-ITD MRD test is also commercially available⁸ and modifications of an open-source protocol method⁹ are possible, the stage is now set. Although the prospect of only a subset of patients being considered eligible for a therapy may not immediately cause hearts to race in the pharmaceutical industry, the drug development and regulatory implications are evident. MRD testing integrated into the protocol allowed identification of a patient population likely to benefit within this “negative" clinical trial. Although a confirmatory trial would likely be required for regulatory approval (which, given the biomarker-enriched population, would be smaller, faster, and cheaper than the original study), clinical practice guidelines are updated frequently based on the latest evidence. MRD testing may also help early in drug development by accurately quantifying any anti-leukemic efficacy to “pick winners” sooner. It would certainly take a brave clinical investigator to forgo MRD testing in their upcoming AML therapeutic protocols, and increasing appreciation of the value of MRD testing in trial design will now facilitate answering the next generation of randomized controlled trial questions: What about treatment for other types of FLT3 MRD¹⁰? Which targeted agent(s) should be used for NPM1 and FLT3 comutated AML MRD? Do patients who test negative for FLT3-ITD MRD after intensive therapy and an FLT3 inhibitor still require alloHCT in first remission? What evidence on serial MRD monitoring should trigger escalation or discontinuation of posttransplant therapy? The future will not immediately be perfect, but the improvements will be measurable.

Conflict-of-interest disclosure: C.S.H. has received research funding from the National Cancer Institute, National Institutes of Health (R44CA233381) and the Foundation of the National Institutes of Health Biomarkers Consortium in AML MRD, and has been involved in a research collaboration with Illumina.