Phase Ib/II trial of anti-CD33 monoclonal antibody BI 836858 and azacitidine in previously untreated older acute myeloid leukemia patients: Beat AML S2 sub-study results

The BCL-2 inhibitor venetoclax (VEN) plus azacitidine (AZA) improved outcomes compared to AZA monotherapy in the pivotal phase III VIALE-A trial.¹ Though the mechanism of action of VEN is not mutation-specific, clinical outcomes are not agnostic to mutation. Some patient subsets, such as those with TP53 mutations, do not benefit from the addition of VEN,² and often experience prolonged myelosuppression from drug or disease-related myelotoxicity. The Beat AML precision medicine program sponsored by Blood Cancer United (formerly the Leukemia & Lymphoma Society), genetically characterizes untreated, older AML patients in the “Master” study (M1), then assigns therapy to a “substudy” (e.g., S1, S2). The second treatment substudy (S2), designed in 2016 (prior to the approval of VEN and prior to the group’s opening of other molecularly targeted substudies), consisted of a novel immunotherapeutic CD33-targeting agent (BI 836858) given in combination with AZA as initial therapy for any patient without a targeted option (clinicaltrials gov. Identifier: NCT03013998). Beat AML S2 was conducted at the same time as an already ongoing similar industry-sponsored trial of decitabine plus BI 836858, conducted largely in Europe.³ Importantly, S2 served as an anchor for the Beat AML program at its beginning, before molecularly-targeted drug trials came online, and provided the proof of concept for a mutation-agnostic substudy within this precision medicine program.

Development of monoclonal antibody (mAb)-based therapies targeting the myeloid differentiation antigen CD33 in AML has been attractive due to its expression on leukemic blasts in almost all patients. Several anti-CD33 mAb-based therapies have been developed for the treatment of AML, most of which have been unsuccessful, with the notable exception of gemtuzumab ozogamicin.^7-11 BI 836858 is a fully humanized IgG1 unconjugated anti-CD33 mAb, fragment crystallizable (Fc)-engineered, with enhanced binding to the Fc receptor CD16 (FcyRIIIa) on natural killer (NK) cells, resulting in increased antibody-dependent cellular cytotoxicity (ADCC).^12,13 BI 836858 was shown to induce significantly higher ADCC in primary AML samples and cell lines in vitro, compared to lintuzumab, an anti-CD33 humanized mAb previously studied in AML. The addition of a hypomethylating agent enhanced BI 836858-mediated NK cell ADCC in vitro, providing the rationale for combining a hypomethylating agent with BI 836858.¹²

Patients ≥60 years with previously untreated AML according to the World Health Organization classification were enrolled per the Beat AML algorithm.¹⁴ Based on a first-in-human phase I monotherapy BI 836858 trial in patients with R/R AML (clinicaltrails gov. Identifier: NCT01690624),¹⁵ BI 836858 20 mg intravenously (IV)/weekly was the starting dose, with escalation to 40, 80, and 160 mg in combination with standard-of-care AZA in 28-day cycles (Online Supplementary Figure S1). Dose-limiting toxicity (DLT) was protocol-defined as drug-related non-hematologic adverse event (AE) of grade ≥3 toxicity, or failure to recover blood counts by day 56 of cycle 1 in the setting of no residual AML. Dose escalation decisions were based on typical toxicity/DLT criteria, as well as receptor occupancy (data not shown). At the recommended phase II dose (RP2D), the trial was powered to determine complete remission/complete remission with incomplete hematologic recovery (CR/ CRi) rate using Simon’s two-stage design. Response was determined by intention-to-treat analysis. The study was conducted with approval by the local institutional review boards and all participants gave written informed consent. BI 836858 was administered IV over up to 8 hours weekly starting on day 9 of cycle 1 to allow for cytoreduction to white blood cell count (WBC) <10x10⁹/L, if necessary. To mitigate against tumor lysis syndrome (TLS), a stepped-up dose-escalation strategy was used to administer BI 836858 in cycle I as follows: 20 mg on day 9, then remainder of the assigned dose on day 10, and from day 16 the full dose was administered on the same day/weekly. Premedications included acetaminophen, antihistamine, and glucocorticoid, which were tapered to tolerance in subsequent doses. Clinical response was assessed per 2017 European LeukemiaNet (ELN) guidelines.¹⁶

Eighty patients received combination BI 836858/AZA. In dose escalation, patients received BI 836858 20 mg/ AZA (N=10), 40 mg/AZA (N=11), 80 mg/AZA (N=10), and 160 mg/ AZA (N=7). Six additional patients received only AZA during phase I due to progression before day 9 and were replaced. All demographic/baseline disease characteristics are presented in Table 1. The composition by mutation subgroup had selection bias. For example, few FLT3-mutated patients enrolled, and enrollment was enriched with patients with mutations of TET2, IDH1, and WT1. Differences in mutation profiles between phase I and II studies were due to subsequent opening of competing studies with targeted agents. Three patients in dose escalation experienced a DLT: one grade 4 neutropenia (20 mg/AZA), one grade 3 portal hypertension (40 mg/AZA), and one grade 3 edema (160 mg/ AZA). No patient had a DLT at 80 mg/AZA. Atypical toxicities observed during dose escalation were one drug-induced liver injury (DILI), one hepatic ischemia, one grade 3 portal hypertension, and three grade ≥3 infusion-related reactions (IRR). No sinusoidal obstruction syndrome/venoocclusive disease (SOS/VOD) was observed, including later in the three patients who proceeded to allogeneic hematopoietic cell transplantation. Due to observed cumulative toxicity at 160 mg as well as pharmacokinetic and receptor occupancy studies, BI 836858 at 80 mg/AZA was selected as the RP2D for this patient population.

Table 1.Demographics and baseline disease characteristics.

The most frequent AE were gastrointestinal (GI; predominantly grades 1/2) and hematologic AE, consistent with historical data for single-agent AZA in AML (Online Supplementary Table S1). The most common non-hematologic AE in phase II were nausea and constipation (58.3% each). The most common grade ≥3 AE were hematologic. Two patients (5.6%) had grade ≥3 bleeding: one grade 4 cerebrovascular accident and one grade 3 GI hemorrhage. Among patients in dose escalation, infections of any grade occurred in 24 patients (66.7%). AE of special interest (AESI) in phase II included infusion-related reactions (IRR) in two patients (5.6%); no TLS or DILI were observed. Overall, 22 patients (61.1%) had serious AE (SAE), including four patients (11.1%) who had seven SAE considered treatment-related; most common treatment-related SAE were IRR (N=2). Twenty-two deaths (61.1%) occurred; four patients (11.1%) had AE leading to death, all of which were infections. Fifteen patients (41.7%) had AE leading to dose interruptions, most commonly IRR (N=5) and chills (N=4). Three patients (8.3%) had AE leading to dose reductions: neutropenia (N=2) and QT prolongation (N=1). Overall, BI 836858 (20-160 mg)/AZA combination demonstrated an acceptable tolerability and safety profile largely consistent with the known side effect profile of AZA, and AE were similar to those reported in a previous phase I/II trial of BI 836858/decitabine combination in older untreated AML patients and R/R AML patients (aged ≥18 years).³

In phase I (N=44), the CR/CRi rate was 25.0% with a median time to best response of 3.0 months (Table 2). Six patients (13.6%) died within the first 60 days of treatment, five of whom received single-agent AZA. In phase II (N=36), the CR/CRi rate was 50.0%, with CR 41.7% and CRi 8.3%; ORR was 52.8% (Table 2). The 30- and 60-day mortality rates were 1 (2.8%) and 4 (11.1%), respectively, with median overall survival of 15.1 months and median time to best response 3.6 months. Evaluating data among all patients (N=80) stratified by ELN 2017 risk, rates of CR/CRi were highest in the favorable risk group (58.3%) and lowest in the adverse risk group (28.6%). Median survival was 18.2 months in the favorable risk group versus 15.1 months in the adverse risk group (Online Supplementary Table S2). Three patients remained on treatment for an extended period (range, 17-46 months), one of whom remained on the investigational agent via single-patient IND after study closure until the drug was no longer made. Three patients proceeded to allogeneic bone marrow transplant on study: one died 4 months post-transplant without known relapse, one was alive at last follow-up 18 months post-transplant, and one relapsed post-transplant.

Current standard-of-care VEN/AZA therapy was previously reported to have CR/CRi rate of 66.4% with median overall survival of 14.7 months in the international phase III randomized study VIALE-A.¹ In contrast, the control group in VIALE-A (AZA monotherapy) only achieved a CR/CRi rate of 28.3% with median overall survival of 9.6 months.¹ Important differences exist between the BI 836858/AZA and VIALE-A trial designs and study populations, which temper strong conclusions about any perceived similarities between outcomes. Though the phase II study was prematurely ended after the industry partner halted development of BI 836858, the demonstration of efficacy and tolerable toxicities observed in the S2 trial suggests that additional agents targeting CD33 may yet have a place in the armamentarium for AML. Certainly, triplets combining an anti-CD33 agent with AZA/VEN are appealing at face value, but such combinations must account for potential excessive myelosuppression. Multiple ongoing studies are exploring additional approaches with chimeric antigen receptor (CAR) T cells or bispecific T-cell engagers. Whether more innovative approaches using CAR T-cell therapies directed at CD33 following inactivation of CD33 in normal hematopoietic stem cells¹⁷ will be more effective remains to be seen. Importantly, in the context of the Beat AML precision medicine program, S2 afforded patients without targetable genetic lesions an opportunity for novel therapy and was a critical building block in the Beat AML program’s success.

Table 2.Observed responses and survival in the efficacy evaluable population in phase Ib and phase II.

Footnotes

• Received January 26, 2026
• Accepted April 17, 2026

Correspondence

M.J. HOCHMAN - michael.jamie.hochman@emory.edu

Disclosures

MJH reports honoraria from Rigel Pharmaceuticals, and Blueprint Medicines. MB reports employment and stock program participation from OncoVerity. TL reports consulting from Daiichi Sankyo, Syndax, Johnson & Johnson, Servier, and Jazz Pharmaceuticals; research payments to institution from Biopath Holdings, Aptevo, Kura Oncology, and Lin Biosciences. TJG reports consulting role with Leukemia & Lymphoma Society (now Blood Cancer United) and stock ownership Bausch Health. All other authors have no conflicts of interest to disclose.

Contributions

Funding

The Beat AML Master Trial and substudies are sponsored by Beat AML, LLC, a division of Blood Cancer United (formerly the Leukemia & Lymphoma Society). Funding was also supported by Boehringer Ingelheim, who was given the opportunity to review the manuscript for medical and scientific accuracy as well as intellectual property considerations.

Acknowledgments

The authors acknowledge the enormous contribution of the patients who participated and likewise those who support research in Blood Cancer United.