Assessment of Treatment Effects By Measurable Residual Disease Monitoring in NPM1-Mutated AML Patients Randomized for Gemtuzumab-Ozogamicin (GO) within the AMLSG 09-09 Trial of the German-Austrian AML Study Group (AMLSG)

Background: Measurable residual disease (MRD), as determined by quantitation of Nucleophosmin 1-mutated (NPM1^mut) transcript levels (TL), provides significant prognostic information independent of other risk factors in patients (pts) with acute myeloid leukemia (AML). This is also addressed by the 2017 European LeukemiaNet (ELN) risk stratification system, which recommends taking into account results from MRD monitoring when selecting the appropriate post-remission therapy. Furthermore, MRD monitoring provides a powerful tool to evaluate treatment effects within clinical trials investigating novel therapies.

Aims: To determine the impact of the anti-CD33 immunotoxin Gemtuzumab-Ozogamicin (GO) on kinetics of NPM1^mut TL in pts with newly diagnosed NPM1^mut AML [18 to 82 years (yrs), median age 58 yrs] enrolled in our randomized Phase III AMLSG 09-09 study (NCT00893399). In this study GO was randomized (1:1) to standard chemotherapy plus ATRA.

Patients and Methods: In total, 588 evaluable pts were enrolled in the clinical AMLSG 09-09 study. Standard treatment comprised two cycles of induction therapy with A-ICE (ATRA, idarubicin, cytarabine, etoposide; arm A) followed by three consolidation cycles of high-dose cytarabine (n=371, 63%) or allogeneic hematopoietic cell transplantation (n=42, 8%). In the investigational arm (arm B), GO (3 mg/m²) was given at d1 of each induction and in the first consolidation cycle. 296 pts were randomized to arm A and 292 pts to arm B. For this correlative study, outcome analysis was restricted to the clinical endpoint cumulative incidence of relapse (CIR) due to study protocol requirements allowing overall survival analysis to be performed only two years after the last pt had been enrolled. MRD monitoring was performed in a total 503 pts for whom at least one bone marrow (BM) sample was available using RQ-PCR technique; the median follow-up (FU) of the 503 pts was 2.8 yrs. NPM1^mut TL (ratio of NPM1^mut/ABL1 transcripts x 10⁴) were determined by RQ-PCR (sensitivity 10^-5 to 10^-6).

Results: In total, 3711 BM samples were analyzed (at diagnosis, n=415; during treatment, n=1765; during FU, n=1531). Both study arms were well balanced with regard to pts characteristics and pretreatment NPM1^mut TL. First, we evaluated the impact of GO on kinetics of NPM1^mut TL during treatment. After the first induction cycle, median NPM1^mut TL were significantly lower in the investigational arm (p=.001) and this was true for all subsequent treatment cycles [after induction II (p=.008), consolidation I (p<.001), consolidation II (p=.006), consolidation III (p=.009)]. Next, we evaluated treatment effects on NPM1^mut TL after two cycles of induction therapy in pts in complete remission (CR, n=378). At this time point, there was no difference in the proportion of pts achieving RQ-PCR negativity (RQ-PCR^neg) [arm A 15% (28/192), vs arm B 17% (32/186); p=.57] between the two treatment arms. However, treatment according to investigational arm B with GO was associated with a significantly lower CIR rate (CIR at 4 yrs: arm B 29% vs arm A 45%, p=.02). In multivariate analysis randomization to arm B revealed as an independent prognostic factor for remission duration (HR 0.63, p=.018). At the end of treatment (EOT, n=288 pts in CR) the proportion of pts achieving RQ-PCR^neg was significantly higher (55% vs 41%; p=.02) in the investigational arm; pts treated in arm B had a significantly lower CIR rate compared to arm A (CIR at 4 yrs: arm B 29% vs arm A 45%, p=.04).

Conclusion: In our randomized Phase III AMLSG 09-09 study, the addition of GO to intensive chemotherapy plus ATRA was associated with a significantly better reduction of NPM1^mut TL after each treatment cycle. This better clearance translated into a significantly lower CIR in the investigational arm with GO.