Analysis of the clinical impact of NPM1 mutant allele burden in a large cohort of younger adult patients with acute myeloid leukaemia

Abstract

Summary Although an NPM1 mutation is generally considered to be a good prognostic marker in acute myeloid leukaemia, it has recently been suggested that a higher level of NPM1 mutant (NPM1MUT) alleles relative to wild‐type alleles is associated with poor clinical outcome. We therefore sought to confirm this finding in a larger study of 876 NPM1MUT cases entered into UK national trials. In univariate analysis, the higher NPM1MUT allele burden was associated with a lower complete remission (CR) rate, higher relapse rate and reduced overall survival, but this was largely attributable to the association of the higher NPM1MUT allele burden with other known poor risk factors, particularly the presence of a concomitant FLT3 internal tandem duplication. In multivariate analysis, there was no significant impact of the NPM1MUT allele burden on CR rates, and the impact on relapse and overall survival, whilst still significant, was greatly reduced. This impact was similar in patients who did or did not receive an allogeneic transplant in first CR. We conclude that the binary presence or absence of an NPM1 mutation, combined with minimal residual disease levels following induction therapy, should continue to be used in therapeutic management rather than stratification according to the NPM1MUT level.

The clinical impact of mutant DNMT3A R882 variant allele frequency in acute myeloid leukaemia

“…However, in two other patient cohorts there was either no impact of NPM1 MUT VAF on outcome (Abbas et al , ) or the impact on outcome was seen only in univariate analysis and not MVA (Rothenberg‐Thurley et al , ). In our own large cohort study of 876 NPM1 MUT cases, we confirmed the Patel et al () study showing that the NPM1 MUT VAF had a major impact on event‐free survival and OS in univariate analysis ( P < 0·0001 for both), but we found that this was largely, although not entirely, attributable to an association with other known prognostic factors, particularly the white blood cell (WBC) count and the presence of a FLT3 ITD (Linch et al , ). For instance, the impact of the NPM1 MUT VAF on OS had a hazard ratio (HR) of 1·81 [95% confidence interval (CI) 1·41–2·32; P < 0·0001] in univariate analysis and 1·29 (95% CI 1·02–1·63; P = 0·03) in MVA.…”

Mutant allelic burden in acute myeloid leukaemia: Why bother?

“…While Patel et al (2018) proposed that a high NPM1 mutant variant allele frequency (VAF) at diagnosis predicts an unfavourable outcome of treatment, which they more recently extended by demonstrating that this correlates with minimal residual disease status in first remission , other comparable studies showed that NPM1 mutant VAFs correlated with leukaemia burden but not with survival (Abbas et al, 2019). In line with this latter publication, the data presented by Linch et al (2020) also show that different levels of the NPM1 mutational burden do not have predictive value for complete remission (CR) and overall survival rates in a multivariate analysis. This also applied to patients receiving an allograft in first CR, a subset of patients in which high NPM1 VAF had a particularly adverse impact in the studies reported by Patel et al (2018Patel et al ( , 2019.…”

“…In this issue of the journal, David Linch and colleagues present a comprehensive analysis on the NPM1 mutant allele frequency in a large cohort ( n = 876) of younger acute myeloid leukaemia (AML) patients (Linch et al , ). The significance of the NPM1 mutant allelic burden for prediction of clinical outcome in AML had previously been addressed in smaller cohorts of patients, leading to conflicting conclusions.…”

“…Conceivably, discrepancies between the studies may be partly explained by differences in sample purities. Irrespective of the underlying explanation, the recommendation of Linch et al (), that the binary presence or absence of an NPM1 mutation rather than variations in VAFs is the most robust and should therefore preferably be used in therapeutic management makes sense, also because inter‐laboratory variations in sample preparations will make VAF cut‐offs ambiguous.…”

“…Whether the current and future trials will provide meaningful answers to discriminate between these possibilities will depend on how patients will be stratified based on VAFs, essentially along the lines presented by Linch et al (). However, here it will be essential to use pure (>90%) AML blasts for the analysis to avoid normal cell admixture as a major confounder.…”