Outcome of (Novel) Subgroups in 1257 Pediatric Patients with KMT2A-Rearranged Acute Myeloid Leukemia (AML) and the Significance of Minimal Residual Disease (MRD) Status: A Retrospective Study By the I-BFM-SG

Introduction

Outcome of KMT2A-rearranged (KMT2A-r) pediatric AML (pAML) is in general poor with a 5-year probability of event-free survival (5y-pEFS) and overall survival (5y-pOS) of 44% and 56%, respectively (Balgobind et al., 2009). However, over the past decades, the heterogeneity of KMT2A-r pAML has emerged, showing differences in outcome between subgroups based on translocation partners. The predictive value of MRD in KMT2A-r pAML is undefined. This retrospective study aimed to confirm the outcome of pediatric KMT2A subgroups (Balgobind et al., 2009) in a more recent era and to study the significance of MRD status during and after induction.

Methods

Outcome and MRD data of 1257 KMT2A-r de novo pAML patients from 15 AML groups affiliated with the I-BFM-AML study group, diagnosed between 2005 and 2016 were retrospectively collected. Patients were assigned to KMT2A subgroups, or to the KMT2A-other group in case of unknown translocation partner. Flow cytometry MRD levels <0.1% were considered negative, and levels ≥0.1% positive. Kaplan-Meier methods were used to estimate probabilities of disease-free survival (pDFS), pEFS and pOS. Cox regression analyses were performed to study the independent impact of KMT2A subgroups and potentially prognostic factors: white blood cell count (WBC), age and MRD status.

Results

The 1257 patients were assigned to 13 KMT2A subgroups, or the KMT2A-other group. Two novel subgroups were identified: t(X;11)(q24;q23) (n=21, 2%) and t(1;11)(p32;q23) (n=12, 1%). The median age was 2.5 years (range, 0-18.9). The median WBC was 21.4 x 10⁹/L (range, 0.2-727). Overall complete remission rate was 91%. The 5y-pEFS was 46% [SE, 2%] and the 5y-pOS was 62% [SE, 2%]. Differences across subgroups in 5y-pEFS (Figure 1) ranged from 24% [SE, 5%] to 76% [SE, 9%], and in 5y-pOS from 25% [SE, 13%] to 92% [SE, 8%] (both p<0.0001). The median follow-up time of patients at risk was 5 years.

The subgroups t(10;11)(p12;q23) (HR 1.7, p<0.0001), t(6;11)(q27;q23) (HR 1.9, p<0.0001), t(4;11)(q21;q23) (HR 2.9, p=.003) and t(10;11)(p11.2;q23) (HR 2.7, p<0.0001), WBC of >100 x 10^9/L (HR 1.3, p=.006), and age >10y (HR 1.3, p=.005) were revealed as independent predictors of poor EFS. These factors also predicted OS.

MRD data after induction course one were available for n=635 (MRD-positivity (range, 0.1-94) n=126, 20%) and after course two for n=527 (MRD-positivity (range, 0.1-88) n=51, 10%). In the four KMT2A poor-risk subgroups, MRD-positivity was not significantly more common after induction course one (p=.0232) or two (p=.066), compared with the other subgroups.

MRD-positivity was associated with inferior 5y-pDFS after both induction course one (36% [SE, 4%] vs 48% [SE, 2%]; p=.002) and course two (28% [SE, 6%] vs 49% [SE, 2%]; p<0.0001) (Figure 2). Within the t(9;11)(p22;q23) subgroup, MRD-positivity after induction course one, and within the t(10;11)(p12;q23) subgroup after course two, was associated with inferior 5y-pDFS (36% [SE, 8%] vs 56% [SE, 4%]; p=.004, and 0% [SE, 0%] vs 35% [SE, 5%]; p<0.0001, respectively). After induction course one, the subgroups t(10;11)(p12;q23) (HR 1.7, p<0.0001) and t(10;11)(p11.2;q23) (HR 4.0, p<0.0001), and MRD-positivity (HR 1.5, p=.003) were revealed as independent predictors of poor DFS. After induction course two, the subgroups t(10;11)(p12;q23) (HR 1.8, p<0.0001), t(4;11)(q21;q23) (HR 4.9, p=.008) and t(10;11)(p11.2;q23) (HR 3.2, p<0.0001), MRD-positivity (HR 2.0, p<0.0001), and age >10y (HR 1.5, p=.002) were revealed as independent predictors of poor DFS.

Within the group of patients with MRD-negativity after induction course two, the subgroups t(10;11)(p12;q23) and t(10;11)(p11.2;q23) were independent predictors of poor EFS (5y-pEFS 35%, HR 1.7, p=.003 and 5y-pEFS 18%, HR 2.7, p=.004, respectively).

Conclusion

Outcome for KMT2A-r pAML patients has improved slightly, but similar subgroups were identified as poor risk (Balgobind et al., 2009), including t(10;11)(p12;q23), t(10;11)(p11.2;q23) and t(6;11)(q27;q23). In our study, t(4;11)(q21;q23) was poor risk as well. These subgroups should be considered for high-risk pAML therapy protocols. The favorable risk of t(1;11)(q21;q23) could not be confirmed in our cohort. MRD status is highly predictive of outcome within KMT2A subgroups. In MRD-negative patients after induction course two, both t(10;11) KMT2A subgroups were associated with poor outcome.