The effect of gemtuzumab ozogamicin on MRD and relapse in patients with NPM1-mutated AML

Advances in the molecular assessment of measurable residual disease (MRD) in patients with acute myeloid leukemia (AML) have led to the inclusion of MRD-negative molecular complete remission (CR) as a new response criterion in the 2017 European LeukemiaNet AML recommendations.¹ One of the most common genomic alterations in AML is a mutation in the gene encoding nucleophosmin (NPM1^mut). NPM1^mut represents an ideal marker for monitoring MRD because it is stable and homogeneous, and studies have shown that the transcript levels are prognostic.

Early results from the phase III AMLSG 09-09 trial (NCT00893399; previously reported on the AML Hub) showed that addition of gemtuzumab ozogamicin (GO) to intensive chemotherapy failed to provide significant improvement in event-free survival (EFS) in patients with NPM1^mut AML.² There was, however, evidence of anti-leukemic activity. In an explorative study published in Blood, Silke Kapp-Schwoerer et al. evaluated NPM1^mut transcript levels and determined the impact of NPM1^mut MRD and GO treatment on the prognosis of patients participating in the AMLSG 09-09 trial.³

Study design³

The study comprised 3,733 bone marrow (BM) and 3,793 peripheral blood (PB) samples from 469 patients with newly diagnosed NPM1^mut AML within the AMLSG 09-09 trial. Inclusion criteria included availability of a diagnostic and at least one subsequent BM or PB sample, and achievement of a CR or CR with incomplete blood count recovery (CRi) after induction therapy.

Patients were treated with intensive chemotherapy plus all-trans retinoic acid (ATRA) with or without GO.

• Standard arm: Two cycles of induction therapy with ATRA, idarubicin, cytarabine, and etoposide (ATRA-ICE) followed by ≤ 3 consolidation cycles of high-dose cytarabine with ATRA.
• GO arm: ATRA-ICE plus 3 mg/m² GO on Day 1 during the two induction cycles and first consolidation cycle.

NPM1^mut transcript levels (TL) were determined by quantitative reverse transcription polymerase chain reaction (RQ-PCR) after two treatment cycles: at the end of treatment (EOT) and during follow-up. Because the primary endpoint of the trial (improved EFS) was not met, cumulative incidence of relapse (CIR) was the endpoint used to assess the prognostic impact of NPM1^mut MRD.

Gene–gene interactions between NPM1^mut and other frequent mutations, including the gene encoding DNA methyltransferase 3A (DNMT3A^mut) and internal tandem duplications of FLT3 (FLT3-ITD), have been observed in NPM1^mut AML but the prognostic impact is unclear. In this study, the authors also evaluated the impact of NPM1^mut/DNMT3A^mut interactions on NPM1^mut TL, MRD negativity, and CIR.

Results³

Patient characteristics

The baseline characteristics of study patients are shown in Table 1. Clinical and disease features were well balanced between study arms; the only significant difference was in DNMT3A mutational status (DNMT3A^mut), which was higher in the GO arm than in the standard arm. FLT3-ITD were underrepresented overall since patients with this mutation were assigned to an alternative trial, (AMLSG 16-10, previously reported on the AML Hub). Post remission, 84% of patients received ≤ 3 cycles of high-dose cytarabine, 7% had an allogeneic hematopoietic cell transplant in the first CR or CRi, and 9% did not receive consolidation therapy.

Table 1. Baseline patient characteristics³

Prognostic impact of NPM1^mut

• A NPM1^mut TL log₁₀ reduction of ≥ 3 from diagnosis to the end of the second treatment cycle in both BM and PB was associated with a significantly lower 4-year CIR (29% vs 60%; p < 0.0001 and 34% vs 63%; p = 0.01, respectively).
• Patients with NPM1^mut TL levels which exceeded a cut-off of ≥ 200 in ≥ 1 sample during follow up, irrespective of sample type, had a 4-year CIR rate of around 70% and a median time to relapse of 3 months.
• MRD negativity was associated with lower 4-year CIR irrespective of sample type, both after two treatment cycles (BM: 25% vs 38%, p = 0.03; PB: 18% vs 53%, p < 0.0001) and at EOT (BM: 26% vs 45%, p < 0.0001; PB: 28% vs 70%, p < 0.0001).
• As shown in Table 2, MRD positivity after two treatment cycles was an independent prognostic factor for higher risk of relapse in PB and BM. The persistence of NPM1^mut TL remained significant at EOT.
• Although greater MRD sensitivity was achieved using BM samples, the prognostic value of MRD was significant in both BM and PB, supporting current recommendations for monitoring using both sample types.
• Other independent prognostic factors were FLT3-ITD with high allelic ratio (FLT3-ITD^high), older age, DNMT3A^mut, and GO treatment.

Table 2. Statistically significant prognostic factors identified in multivariate analysis³

Effect of GO treatment on outcome

• After the first induction cycle and throughout all subsequent treatment cycles, median NPM1^mut TL in BM and PB was significantly lower in patients treated with GO compared with patients in the standard arm.
• This translated into a lower CIR rate in the GO arm compared with the standard arm (32% vs 44%; p = 0.015) and a superior 4-year relapse-free survival (61% vs 49%; p = 0.028).
• At EOT, more patients in the GO arm achieved MRD negativity in BM (56% vs 41%; p = 0.01) and PB (85% vs 72%; p = 0.02) compared with the standard arm, and patients receiving GO treatment were more likely to have sustained MRD negativity during follow-up (BM: 41% vs 30%, p = 0.06; PB: 51% vs 39%, p = 0.05).
• GO treatment was identified as an independent prognostic factor for lower risk of relapse in BM after two treatment cycles and at EOT in both BM and PB (Table 2).

Impact of concurrent DNMT3A^mut

• From the second treatment cycle, patients with NPM1^mut alone had lower NPM1^mut TL than those with NPM1^mut/DNMT3A^mut, leading to a significantly higher proportion achieving MRD negativity after two cycles (19% vs 8%; p = 0.02) and at EOT (52% vs 36%; p = 0.04).
• This reduction in NPM1^mut TL translated into significantly reduced CIR in patients with NPM1^mut alone after two treatment cycles (26% vs 48%; p = 0.0003), and a trend to reduced CIR at EOT (30% vs 40%; p = 0.07).
• The impact of GO on increased MRD negativity at both timepoints was only observed in patients with NPM1^mut alone; there was no impact of GO on patients with NPM1^mut/DNMT3A^mut.
• The data suggest that concurrent DNMT3A^mut may confer resistance to treatment in NPM1^mut AML, but this finding requires confirmation in larger studies.
• Of note, the NPM1^mut/DNMT3A^mut/FLT3-ITD triple genotype could not be evaluated due to the small number of patients with FLT3-ITD.

Conclusion

This exploratory analysis supports a role for early reduction of NPM1^mut TL in lowering relapse rates and demonstrates the prognostic relevance of MRD monitoring in patients with NPM1^mut AML. The addition of GO to standard intensive chemotherapy led to reduced NPM1^mut TL and higher rates of MRD negativity, which translated into lower CIR and superior relapse-free survival. The authors conclude that monitoring MRD with NPM1^mut TL is a valuable tool for assessing risk of relapse and efficacy of novel treatments in NPM1^mut AML.