Phase 1b Dose Escalation Study of BI 836858 and Azacitidine in Previously Untreated AML: Results from Beat AML S2

Background:

The prognosis for acute myeloid leukemia (AML) patients age >60 years is poor. Rapid pre-treatment identification of molecular disease subsets may allow specific targeting with novel agents, presenting an opportunity to improve outcomes. The Leukemia and Lymphoma Society (LLS)-sponsored Beat AML master trial was initiated for previously untreated AML patients ≥60 years, with treatment assignment to a sub-study (S1, S2, etc.) based upon cytogenetics and dominant clone by next generation sequencing; patients with multiple mutations are assigned a sub-study according to variant allele frequency and a predetermined prioritization schema.

Methods:

Patients within two molecular subsets were selected for a phase 1b/2 dose escalation study (S2) of the Fc engineered (for increased binding to FcyRIIIa) CD33 antibody BI 836858 plus azacitidine (AZA). Group A was comprised of patients with mutations that drive hypermethylation, with potential for enhanced sensitivity to AZA: TET2/WT1 or IDH1/2 (for IDH1/2, if no specific targeting agent available). Group B was comprised of "marker-negative" AML (those not eligible for any available targeted sub-study). Both groups were pooled in the phase 1b portion reported herein. Patient eligibility included ECOG performance status < 3, no prior chemotherapy for AML or myelodysplastic syndrome, and preserved organ function. Patients received AZA (75 mg/m²) IV or SQ days 1-7 of 28 day cycles. BI 836858 was given weekly beginning day 9 (provided WBC < 10 x 10⁹/L) and monthly once complete remission with or without sufficient blood count recovery (CR/CRi) was obtained. A standard 3 + 3 design for toxicity was used to guide dosing decisions, with consideration of dose level expansion based on toxicity and CD33 saturation. Dose-limiting toxicity (DLT) was defined by select grade 3/4 toxicities during cycle 1, or failure to recover counts by day 56 in the setting of no residual AML.

Results:

At data cut off (April 30, 2018) 39 patients were enrolled; 31 were treated with the combination. Eight were not treated with BI 836858 due to failure to start any therapy (3), withdrawal (1), high white blood cell (WBC) count (2), adverse event (1), and early death (1). Median age of the 31 patients was 71 years (range 62-85); median WBC at study enrollment was 5.3 x 10⁹/L (range 0.5-46.7) and platelets 52 x 10⁹/L (range 10-681). Seventeen (55%) were in Group A (9 TET2, 1 with both TET2 and WT1, 3 IDH1, 4 IDH2). Among 27 with known cytogenetics, 16 (59%) were abnormal including 5 (19%) with complex karyotype. Of 31 patients, 8 were treated at 20 mg, 11 at 40 mg, 9 at 80 mg, and 3 at 160 mg. Six patients (3 at 80 mg and 3 at 160 mg) are currently in cycle 1 and have not fully been assessed for DLT or response. One patient had grade 4 decreased neutrophils in the absence of detectable AML at 20 mg and one had grade 3 portal hypertension at 40 mg, which were considered DLTs. Infusion reactions occurred in 5 BI 836858-treated patients, including 3 grade 3/4. Common grade 3/4 toxicities observed included: anemia (45%), febrile neutropenia (36%), hypophosphatemia (32%), hyponatremia (26%), and decreased platelet (42%), neutrophil (39%), and WBC count (36%). Pharmacokinetics of BI 836858 at dose levels 1-3 (20-80 mg) showed proportional increases by dose without trough accumulation at 4 weeks. Blast CD33 saturation by BI 836858 in the blood and bone marrow was not observed at dose levels 1-3. Pharmacodynamic studies of NK cells pre/post dose of BI 836858 demonstrated evidence of activation in a subset of patients. Among 28 eligible patients who received at least one dose of AZA, had adequate follow-up, and did not withdraw consent prior to response evaluation, there have been 5 CR/CRi (18%) and 4 morphologic leukemia free state (14%) responses, to date. Fifteen patients remain on treatment at data cut off. Reasons for discontinuing combination therapy include proceeding to allogeneic stem cell transplant (1), failure to respond (5), progression after response (4), adverse events (5), and patient withdrawal (1).

Conclusions:

This phase 1b sub-study of Beat AML demonstrates acceptable tolerability and activity of the combination of AZA with a novel CD33 glycoengineered antibody. Dose escalation as part of this sub-study continues, which will allow further assessment of efficacy and toxicity of this regimen in molecularly defined AML subsets.