Clinical Relevance of Minimal Residual Disease Monitoring in NPM1 Mutated AML: A Study of the AML Study Group (AMLSG)

Background: Nucleophosmin (NPM1^mut) mutations represent one of the most common gene mutations in acute myeloid leukaemia (AML) and can be used for monitoring minimal residual disease (MRD). In a former study, we could define clinical relevant check-points and a cut-off value to identify patients (pts) at high risk of relapse.

Aims: To confirm our previous results on the clinical relevance of NPM1^mut transcript levels (TL) in an extended cohort of younger AML pts (18 to 60 years) harbouring NPM1^mut type A, B, C, D, JT, 4, QM, NM or KM, and to assess the impact of concurrent FLT3 internal tandem duplications (ITD) and DNMT3A (DNMT3A^mut) mutations on NPM1^mut TL kinetics.

Methods: All pts were enrolled in one of four AMLSG [AMLHD98A (n=46; NCT00146120); AMLSG 07-04 (n=199; NCT00151242); AMLSG 09-09 (n=179; NCT00893399); AMLSG 16-10 (n=75; NCT01477606)] treatment trials. Treatment comprised double induction therapy (DI) with ICE (idarubicin, cytarabine, etoposide) with or without ATRA or gemtuzumab ozogamicin, or 1 cycle of daunorubicin and cytarabine followed by 1 to 4 cycles of high-dose cytarabine (n=292), autologous (n=19) or allogeneic stem cell transplantation (n=141). NPM1^mut TL (ratio of NPM1^mut/ABL1 transcripts x 10⁴) were determined by RQ-PCR using TaqMan technology; the sensitivity of the assays was 10^-5 to 10^-6. DNMT3A and FLT3-ITD (FLT3-ITD^mut) mutation status was assessed by standard PCR-based methods.

Results: A total of 2835 samples from 499 NPM1^mut pts were analysed at diagnosis (n=439), after each treatment cycle (n=1394) and during follow-up (FU) (n=1002). Peripheral blood (PB) samples were only included in the advanced FU period (defined as at least 12 months after completion of therapy). NPM1^mut TL at diagnosis varied between 7.03 x10³ and 2.38 x 10⁷ (median 5.37 x 10⁵). Pretreatment NPM1^mut TL were not associated with clinical characteristics (e.g., age, WBC, BM blasts, FLT3-ITD^mut, DNMT3A^mut) with the exception of LDH level (p=0.006) and did not impact event-free survival (EFS), relapse-free (RFS) and overall survival (OS). NPM1^mut TL as log 10 transformed continuous variable at different time points during therapy were significantly associated with shorter remission duration (RD) and shorter OS. After DI therapy, the cumulative incidence of relapse (CIR) at 4 years was 10% for RQ-PCR-negative pts (n=41) versus 45% for RQ-PCR-positive pts (n=226) (p<0.0001); the lower CIR translated into a significant better OS (92% versus 60%, respectively; p=0.001). After completion of therapy, CIR at 4 years was 13% for RQ-PCR-negative pts (n=126) and thus significantly lower compared with 56% in RQ-PCR-positive pts (n=139; p<0.00001). Again, the lower CIR translated into a significantly better OS (81% versus 55%, respectively; p<0.00001). Multivariable analysis performed at both time points showed that NPM1^mut TL were significantly associated with a shorter RD (HR, 1.86; 2.30, respectively) and shorter OS (HR, 1.58; 1.72, respectively). During FU, 1002 bone marrow (BM) and PB samples from 280 pts were analysed. The relapse rate at 2 years for pts exceeding the previously defined cut-off value of >200 NPM1^mut copies was 90% with a median time to relapse of 1.38 months. In contrast, only 6/104 pts with sustaining RQ-PCR negativity relapsed. Finally, we evaluated the impact of concurrent FLT3-ITD^mut and DNMT3A^mut on kinetics of NPM1^mut TL. Following the first induction cycle, the median NPM1^mut TL was significantly lower in pts with the NPM1^mut/FLT3-ITD^wildtype/DNMT3A^wildtype genotype compared to pts with the genotype NPM1^mut/FLT3-ITD^mut/DNMT3A^mut. This effect could be observed throughout subsequent treatment cycles.

Conclusions: The results of our analysis on an extended cohort of younger AML pts with NPM1^mut highly confirmed the two clinically relevant MRD check-points, after DI and after completion of therapy; during the FU period, exceeding a cut-off value of >200 TL was highly predictive for relapse. Finally, we found a significant impact of concurrent FLT3-ITD^mut/DNMT3A^mut on the kinetics of NPM1^mut TL.