Molecular Detection of Minimal Residual Disease Based on NPM1-Real-Time PCR At the Time of Complete Remission Is a Strong Predictor of Disease Free and Overall Survival in AML

Several recently identified molecular markers, i.e. FLT3-ITD, NPM1 and CEBPA and a number of others, have completely redefined our understanding of disease development in patients with acute myeloid leukemia and have already profound effects on risk assessment and treatment in AML. In addition, some of these novel markers, especially NPM1 -mutations, can be very efficiently used to detect minimal residual disease (MRD). In this study we investigated the suitability of NPM1 as MRD-marker, evaluated optimal cut-off points to define molecular relapse and analyzed the impact of MRD and additional risk factors on survival in a large cohort of patients with AML. Methods: The study population consisted of 164 NPM1 -mutant AML patients (pts) (median age 51 yrs. (range, 20–79 yrs.)), having one of the three most common NPM1 -mutations, A (N=139), B (N=16) and D (N=9). The patients were treated in protocols of the Study Alliance Leukemia (SAL) and were prospectively monitored. Consolidation treatment included autologous (N=18) and allogeneic transplantation from HLA-matched related (N=20) and unrelated (N=21) donors. An optimized assay for the sensitive cDNA based detection of the three most common NPM1-mutations using a Real-Time-Q-PCR based on locked-nucleic acid (LNA) containing primers was used. MRD-levels were expressed as percent of ABL1 -copies, which was assessed in all samples to control for sample quality. Results: Of the primary group, 154 patients were available for further analysis, 10 patients did not achieve CR and were excluded. A total of 1671 samples (972 BM; 699 PBL) were analyzed, the median number of samples per patient was 8 (range, 3–59), PBL and BM were analyzed in parallel at 138 time points. Although in general the median NPM1 levels in PBL samples were alsmost one log10 lower, a good qualitative concordance was found between both sources, with 48.6% of the samples being positive and 34.1% negative in both materials, whereas BM was pos. and PBL neg. in 14.5% and PBL pos and BM neg. in 2.9% of the samples, respectively. Using receiver operator characteristic (ROC) curves and regression models, the optimal cut-off value for detection of subsequent clinical relapse was found to be 1% NPM1/ABL1. In a competing risk analysis using Gray's algorithm, the cumulative incidence of relapse in patients >1% and <=1% at twelve months was 34.8% [95% CI: 18.4–51.2%] vs. 7.7% [95% CI: 0–16.1%] (p=.0007). We also analyzed, whether the level of MRD at the time of achievement of CR had an impact on outcome. Both, NPM1 -MRD negativity and MRD-values below 1% were associated with significantly improved disease free (DFS) and overall (OS) survival, e.g. median OS in patients with NPM1 >1%: 20.2 months vs. NPM1 <1% not reached (p=.003), median DFS 13.6 months [95% CI: 9.7-35.3 mo] vs. 61.1 months [95% CI: 27.3 mo -not reached] (p=.0003). Not surprisingly, when we further differentiated the NPM1- positive patients into FLT3-ITD positive and negative, the subgroup of patients with FLT3-ITD andNPM1 >1% was associated with the poorest outcome. Conclusions: Our data clearly support the reported strong predictive value of NPM1 -positivity for the detection of relapse. In addition, our data show that the quality of CR as assessed by molecular methods profoundly affects long term outcome. Taken together, in NPM1 positive patients, MRD should be prospectively monitored to detect impeding relapse and to potentially start preemptive therapy, e.g. treatment with 5-Azacytidine or allo-SCT in eligible patients.

Thiede:AgenDix GmbH: Employment, Equity Ownership. Platzbecker:Celgene: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding.