Acute myeloid leukemia (AML) is a heterogeneous disease where selected subgroups of patients, linked by the presence of biological and clinical high-risk features, are candidates to receive allogenic hematopoietic stem cell transplantation HSCT) as post-remission consolidation treatment. The achievement of morphological complete remission (CR) before HSCT is an important pre-requisite to optimize the chance of successful post-transplant outcome. Minimal residual disease (MRD) assessment by quantitative polymerase chain reaction (q-PCR) has been shown to increase the ability to monitor therapy response in AML, improving prognostic accuracy and allowing to refine transplant strategies. Although MRD assessment was shown to have potential benefit when measured after induction and consolidation therapy courses, its role before HSCT remains to be fully elucidated. In order to contribute to better clarify this issue, we conducted a q-PCR I-BFM-AML collaborative study to measure MRD in bone marrow samples collected within 5 weeks prior to HSCT of 108 pediatric AML patients harboring one of the main recurrent AML aberrancies t(8;21)(q22;q22); RUNX1-RUNX1T1
, inv(16)(p13.1q22)/t(16;16)(p13.1;q22); CBFB-MYH11
, t(9;11)(p22;q23); KMT2A
ITD. Sixty patients underwent HSCT in first complete remission (CR1) with an overall survival (OS) of 84% versus
54% for the 48 transplanted in CR2 achieved after an initial relapse. Sixty patients showed q-MRD negativity (defined as a value lower than 2.1x10-4
calculated by ROC curve analysis with respect to diagnosis or relapse), whereas in 48 patients we detected q-MRD levels >
. Five-year OS after HSCT was 83% for patients with q-MRD negativity, while that of patients with q-MRD above the cutoff was 57% (p=0.012). As regards, cumulative incidence of relapse (CIR), q-MRD above the cutoff was associated with a high risk of recurrence (26% versus
10% for patients with q-MRD <2.1x10-4
, p=0.036), q-MRD positivity representing an independent prognostic factor. When we interrogated the 3 genetic subgroups (namely CBF
r and FLT3
-ITD), despite the limited sample size, we found that OS was significantly influenced by q-MRD pre-HSCT in FLT3-
ITD (63% versus
100% for q-MRD negative patients, p=0.019) and in t(8;21)RUNX1-RUNX1T1
rearranged patients (50 % versus
84% for q-MRD negative patients, p=0.048). We further investigated the impact of higher levels of q-MRD: we found that the 17 patients showing a pre-transplant q-MRD reduction lower than 1x10-2
(2-log), with respect to either diagnosis or relapse value, had a worse outcome (OS=39%) when compared to the 91 patients who reduced q-MRD values more than 2-log (OS=78%, p=0.0019). These 17 patients, transplanted in CR1 (n=8) or CR2 (n=9), were heterogeneous in terms of genetic lesions (t(8;21) n=7, inv(16) n=2, t(9;11) n=5 and FLT3-
ITD n=3). Applying this 2-log cutoff by genetic subgroups, we found that cases with RUNX1-RUNX1T1
with q-MRD reduction above 2-log had the worst prognosis (OS 29% for q-MRD>2-log versus
73% for q-MRD<2-log, p=0.016). Overall, cases with FLT3
more often achieved a q-MRD reduction greater than 2-log. In line with these results we combined the two measurement approaches and proposed a model where the two cutoffs generate 3 risk groups stratification, namely low (q-MRD<2.1x10-4
, LR), intermediate (q-MRD>
and <2-log, IR) or high risk (q-MRD>2-log, HR). This combined stratification by q-MRD resulted into a better subdivision of the OS probability, which was 83%, 69% and 39% for LR, IR and HR respectively (p=0.004).
Finally, a multivariate Cox regression model revealed that, together with CR status at time of the allograft (CR2, hazard ratio 4.4, p=0.001), q-MRD was an independent factor (hazard ratio 0.5, p=0.001) predicting HSCT outcome.
In conclusion, this study supports the role of q-MRD pre-HSCT as a useful prognostic tool in childhood AML, able to provide information to tailor transplant strategies involving conditioning regimen intensity and graft-versus-host disease prophylaxis.