The anti‐CD47 antibody magrolimab with obinutuzumab and venetoclax in relapsed or refractory indolent B‐cell lymphomas

Abstract

Follicular lymphoma (FL), marginal zone lymphoma (MZL), chronic lymphocytic leukaemia (CLL) and mantle cell lymphoma (MCL) are characterized by a continuous incidence of relapse and increasing resistance to therapy. Novel immunotherapy approaches are needed. Magrolimab, a CD47‐blocking antibody, disrupts CD47:SIRPα‐mediated antiphagocytic signalling. When combined with a prophagocytic signal from an anti‐CD20 antibody rituximab, it has shown activity in relapsed or refractory FL and MZL. In this phase 1 study, adding the BCL2‐inhibitor venetoclax to magrolimab and the anti‐CD20 antibody obinutuzumab resulted in complete responses in 6 of 10 (60%) evaluable patients with FL, MZL or CLL. Notably, we did not observe increased risk of infections previously reported from studies of magrolimab in acute myeloid leukaemia and higher risk myelodysplastic syndromes.

To the Editor,

Follicular lymphoma (FL), marginal zone lymphoma (MZL), chronic lymphocytic leukaemia (CLL) and mantle cell lymphoma (MCL) are indolent B‐cell lymphomas that often respond to initial therapy but demonstrate a continuous incidence of relapse and become increasingly refractory to therapy. ¹ Targeted therapies avoid the long‐term risks of chemotherapy and are highly active in indolent B‐cell lymphomas. ² , ³ However, they are typically delivered for extended durations and are not curative. Immunotherapies, including chimeric antigen receptor (CAR) T‐cell therapy and bispecific T‐cell engagers (BiTE), are active in FL and MCL but have unique toxicities that require close monitoring and intensive supportive care. ⁴ , ⁵

CD47 is a surface protein widely expressed on normal and malignant cells. ⁶ The interaction between CD47 and SIRPα on phagocytic cells recruits the src homology‐2 domain containing protein tyrosine phosphatases SHP‐1 and SHP‐2, which deliver an antiphagocytic or ‘don't eat me’ signal (Figure 1A). ⁶ Blocking CD47 alone is insufficient to induce phagocytosis and additional prophagocytic signals are required. Most normal cells lack these prophagocytic signals, whereas cancer cells frequently express them, rendering them susceptible to CD47 blockade. Magrolimab is a monoclonal antibody that blocks the CD47‐mediated antiphagocytic signalling, and rituximab provides a potent prophagocytic signal by stimulating antibody‐dependent cellular phagocytosis. ⁷ Together, they promote efficient phagocytosis of lymphoma cells.

FIGURE 1.

Pathways targeted and adverse events. (A) Malignant B‐cell survival pathways targeted by combination of magrolimab, obinutuzumab and venetoclax. Magrolimab blocks the CD47‐mediated antiphagocytic signalling and obinutuzumab provides a potent prophagocytic signal by stimulating antibody‐dependent cellular phagocytosis. Venetoclax induces apoptosis through inhibition of BCL2 and enhances anti‐CD20‐mediated antibody‐dependent phagocytosis (B) All treatment‐emergent adverse events (AEs) in >1 patient and all grade ≥3 AEs plotted as a percentage of affected patients. The numbers within the stacked bars reflect the number of affected patients.

In a phase I/II trial, indefinite therapy with magrolimab and rituximab induced responses in 52% of patients with relapsed/refractory FL and MZL, including a complete response (CR) rate of 30%. ⁸ , ⁹ The maximum tolerated dose was not reached and only 4% patients discontinued therapy due to toxicity. Approximately 30% of responding patients had not relapsed after 2 years and no late‐onset or cumulative toxicity was observed. ⁸

Venetoclax is a highly selective BCL‐2 inhibitor with activity in indolent B‐cell lymphomas. ¹⁰ , ¹¹ It restores intrinsic apoptosis via activation of caspases. ¹² Conversely, anti‐CD47 antibody and rituximab induce phagocytosis and kill non‐phagocytosed cells via a caspase‐independent mechanism. ¹² , ¹³ Venetoclax also enhances anti‐CD20‐mediated antibody‐dependent phagocytosis. ¹⁴ In xenograft models of CLL, the combination of anti‐CD47 antibody, rituximab and venetoclax led to durable tumour regressions. ¹² Obinutuzumab is a novel anti‐CD20 antibody which shares its mechanism of action with rituximab but can induce greater direct cell death by caspase‐independent mechanisms. ¹⁵ We hypothesized that venetoclax would synergize with magrolimab and obinutuzumab to induce deep and durable responses in patients with indolent B‐cell lymphomas without the need for indefinite duration of treatment.

In a phase 1 study, we enrolled adult patients with FL, MCL, MZL or CLL relapsed after or refractory to ≥2 prior lines of therapy (Data S1). Key exclusion criteria were prior anti‐CD47 treatment and active haemolytic anaemia or autoimmune thrombocytopenia. Prior venetoclax therapy was allowed. We were primarily concerned about neutropenia as an overlapping toxicity when combining venetoclax with magrolimab and obinutuzumab. Since obinutuzumab and magrolimab toxicity is not strictly dose‐dependent, we focused on venetoclax during dose finding. Two cohorts of six patients each were planned with different target doses of venetoclax. Cohort 1 (FL) started venetoclax on day 2 of Cycle 1 at the target dose of 800 mg. Cohort 2 (MCL, MZL, CLL) started venetoclax on day 2 of Cycle 1 at 20 mg, with a 5‐week escalation to the target dose of 400 mg. If ≥2 of six patients in a cohort had dose‐limiting toxicity (DLT), an additional six patients would be enrolled at a reduced target dose level (−1) of venetoclax: 600 mg for Cohort 1 and 200 mg for Cohort 2.

For both cohorts, magrolimab started at a 1 mg/kg priming dose on day 2, followed by 30 mg/kg on days 8, 15 and 22 of cycle 1, then on day 1 of each 4‐week cycle. Obinutuzumab dose was 100 mg on day 1, 900 mg on day 2, followed by 1000 mg on days 8 and 15 of the first cycle, then on the first day of each subsequent cycle. DLT was assessed during the first 4 weeks for Cohort 1 and 5 weeks for Cohort 2. Dose expansion cohorts targeting up to 18 FL, 6 MZL, 12 MCL and 6 CLL patients (total 42) were planned at the venetoclax target dose determined from the dose‐finding portion. Responses were determined by computed tomography (CT) every three cycles and fluorodeoxyglucose positron emission tomography (FDG‐PET) every six cycles and classified using the Lugano criteria for FL, MZL and MCL, and International Workshop criteria for CLL. Measurable residual disease (MRD) in CLL patients was assessed by peripheral blood flow cytometry at the end of therapy.

After six cycles, patients who achieved a CR stopped therapy while patients who achieved a partial response (PR) received an additional six cycles. The primary end‐point was safety assessed by the overall toxicity profile and DLTs. Secondary end‐points were overall response rate, duration of response, progression‐free survival (PFS) and overall survival. The study was registered at ClinicalTrials.gov (NCT04599634) and was conducted in accordance with the principles of the Declaration of Helsinki. The local institutional review board approved it, and all patients signed written informed consent.

Eleven patients were enrolled with a median age of 68 (range: 38–81) years (Table 1). Five patients had FL and two patients each had MCL, MZL and CLL. The median number of prior therapies was 2 (range: 2–4), and 6 (55%) patients received prior venetoclax. No patient had prior CAR‐T or BiTE therapy. Three (27%) patients were refractory to the last therapy, and 2 (18%) had relapsed within 6 months.

TABLE 1.

Patient characteristics.

Characteristics	N = 11
Median age (range)	68 (38–81)
Male sex, n (%)	5 (45)
Race, n (%)
White	6 (55)
Black	1 (9)
Asian	3 (27)
Other	1 (9)
Histology, n (%)
Follicular lymphoma	5 (45)
Mantle cell lymphoma	2 (18)
Marginal zone lymphoma	2 (18)
Chronic lymphocytic leukaemia (CLL)	2 (18)
Number of prior lines of therapy, median (range)	2 (2–4)
Prior therapy, n (%)
Bruton's tyrosine kinase inhibitor	8 (73)
Venetoclax	6 (55)
Anthracycline	5 (45)
Radiation	4 (36)
Elevated lactate dehydrogenase, n (%)	6 (55)
Advanced stage, n (%)	11 (100)
Bone marrow involvement, n (%)	6 (55)

Haematological adverse events were most common with grade ≥3 thrombocytopenia and neutropenia in 4 (36%) and 1 (9%) patient respectively (Figure 1B, Table S1). Grade 3 anaemia or any grade haemolytic anaemia were not observed. One patient had grade 3 pneumonia 3 weeks after treatment discontinuation, considered unrelated to study therapy. Another patient had grade 3 laboratory‐only tumour lysis syndrome, which resolved within 12 h after intravenous fluids, diuretics and potassium binders. Two DLTs occurred, both grade 4 thrombocytopenia in one patient each with MCL and CLL. Both had lymphocytosis >50 000/mm³, and the DLT occurred within 2 days of the first obinutuzumab infusion: in one patient, before the first magrolimab dose. Treatment was discontinued, and the platelet counts recovered within 1 week.

Patient #8 with relapsed MCL experienced a DLT on day 1 of Cycle 1 and began the next line of treatment without undergoing restaging scans. Of the remaining 10 response‐evaluable patients, 8 (80%) responded with 6 (60%) achieving CRs in patients with FL, MZL and CLL (Figure 2A). The 2 CLL patients had received prior Bruton's tyrosine kinase inhibitors but not a BCL2 inhibitor. Both patients with CLL achieved undetectable MRD. After a median follow‐up of 17.4 months, all six patients who achieved a CR remained alive in remission for a range of 3–21 months (Figure 2B). Two patients had PRs that lasted 3 months and 14 months. The 1‐year duration of CR was 100% and the 1‐year duration of PR was 50% (Figure 2C). The 1‐year PFS was 68.6% (95% CI, 30.5%–88.7%) and all patients remain alive (Figure 2D). Accrual was halted prematurely before expansion cohorts due to manufacturer discontinuation of magrolimab. ¹⁶

FIGURE 2.

Objective responses and survival outcomes for all patients. (A) Best objective tumour responses in all patients divided by histological subtype compared to baseline. (B) Swimmer's plot demonstrating timing and duration of responses in all patients divided by histological subtype, along with details of prior therapy. (C) Duration of complete response (DOCR) and duration of partial response (DOPR). (D) Progression‐free survival and overall survival for all patients. CI, confidence interval; CLL, chronic lymphocytic leukaemia; CR, complete response; FL, follicular lymphoma; MCL, mantle cell lymphoma; MZL, marginal zone lymphoma; PR, partial response; SPD, sum of product of diameters on computed tomography scan.

In a prospective trial, we show for the first time that the anti‐CD47 antibody magrolimab can be safely combined with both an anti‐CD20 antibody and a BCL2 inhibitor. Although limited numbers, this combination induced CRs in patients with relapsed and refractory FL, MZL and CLL, without extended treatment. All patients were previously treated with an anti‐CD20 agent, and 6 (55%) received venetoclax. Given the small sample size, larger studies are needed to confirm the rate and durability of the CRs. Nevertheless, our preliminary results suggest that this novel combination immunotherapy approach targeting CD47 may overcome resistance to standard therapy. We also report the first safety data for anti‐CD47 antibody‐based therapy in CLL.

The toxicity profile of this regimen was primarily haematological and both DLTs were infusion‐related thrombocytopenia, and platelet counts recovered without intervention or transfusion. The rapid onset of thrombocytopenia in patients with lymphocytosis suggests that it was primarily due to obinutuzumab, consistent with previous reports. ¹⁷ , ¹⁸ , ¹⁹ However, both patients also received venetoclax, and one experienced worsening thrombocytopenia after day 2 magrolimab infusion, indicating all three drugs may have contributed. Strategies to mitigate this complication could include allowing more time between the first obinutuzumab infusion and starting magrolimab and venetoclax, splitting the first obinutuzumab dose, or switching from obinutuzumab to rituximab in patients with lymphocytosis. In a trial with magrolimab and rituximab for relapsed or refractory indolent non‐Hodgkin lymphoma, rituximab started 8 days after magrolimab. ⁸ Only 1 (2%) of 46 patients stopped treatment due to Grade 4 thrombocytopenia, but the acuity of onset was not reported. We observed one serious infection, a pneumonia 3 weeks after treatment discontinuation, and after initiation of subsequent treatment. Only one case of lab‐only tumour lysis was noted during cycle 1 which rapidly resolved with medical management. No haemolysis or opportunistic infections were noted.

A major limitation of this study was premature closure due to the discontinuation of magrolimab manufacturing despite very promising early results. This decision was made due to increased infections observed in other trials testing combination therapy in patients with acute myeloid leukaemia (AML) and higher risk myelodysplastic syndromes (MDS) and unrelated to studies in lymphoma. ¹⁶ The early termination of this study precluded us from expanding upon these preliminary findings in expansion cohorts to better estimate the CR rate and durability. Patients with AML and MDS are prone to severe infections due to neutropenia caused by the disease itself. In contrast, most patients with indolent lymphoma do not suffer from severe disease‐related cytopenias, suggesting that similar toxicity concerns may not apply CD47‐targeting agents in this population. Taken together, our data support further development of other agents targeting the CD47:SIRPα pathway and provide foundational evidence that combinations can induce durable responses after short durations of therapy. Clinical trials with other CD47:SIRPα pathway targeting agents are ongoing (NCT05025800, NCT05507541, NCT05892718, NCT05896163, NCT05737628).

AUTHOR CONTRIBUTIONS

Drs Lakhotia and Roschewski had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Roschewski, Lakhotia, Wilson. Acquisition, analysis or interpretation of data: All authors. Drafting of the manuscript: Lakhotia, Roschewski. Critical review of the manuscript: All authors. Statistical design and data analysis: Lakhotia, Roschewski. Administrative and technical support: Evans. Supervision: Wilson, Staudt.

CONFLICT OF INTEREST STATEMENT

The authors report no conflict of interest.

ETHICS STATEMENT

This study was performed in accordance with the Declaration of Helsinki. Local ethics approval was obtained.

INFORMED CONSENT

All patients gave written informed consent before entering the study.

CLINICAL TRIAL REGISTRATION

The clinical trial is registered at www.clinicaltrials.gov (NCT04599634).

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

The data generated by this research will be made openly available within 1 year of the publication of this article on www.clinicaltrials.gov.