Safety and Efficacy of Bispecific Antibodies in Adults with Large B-Cell Lymphomas: A Systematic Review of Clinical Trial Data

Bispecific antibodies (bsAbs) are an emerging therapy in the treatment of large B-cell lymphomas (LBCLs). There is a gap in the research on the safety and efficacy of bsAbs in adults with LBCL, with current research focusing on the wider non-Hodgkin’s lymphoma population. To address this research gap, we conducted a systematic review aiming to evaluate the safety and efficacy outcomes of bsAbs in adults with LBCL. A systematized search was conducted in PubMed, EMBASE, and CENTRAL on 10 April 2024. Interventional clinical trials were eligible for inclusion. Observational studies, reviews, and meta-analyses were excluded. According to the Revised Risk of Bias Assessment Tool for Nonrandomized Studies, the included studies were largely of a high quality for safety outcome reporting, but of mixed quality for efficacy outcome reporting. Due to the heterogeneity of the included studies, the results were discussed as a narrative synthesis. Nineteen early phase studies were evaluated in the final analysis, with a pooled sample size of 1332 patients. Nine bsAbs were investigated across the studies as monotherapy (nine studies) or in combination regimes (10 studies). The rates of cytokine release syndrome were variable, with any grade events ranging from 0 to 72.2%. Infection rates were consistently high across the reporting studies (38–60%). Cytopenias were found to be common, in particular, anemia (4.4–62%), thrombocytopenia (3.3–69%), and neutropenia (4.4–70%). Immune effector cell-associated neurotoxicity syndrome (ICANS) and grade ≥3 adverse events were not commonly reported. Promising efficacy outcomes were reported, with median overall response rates of 95–100% in the front-line and 36–91% in terms of relapsed/refractory disease. The results of this systematic review demonstrate that bsAbs are generally well-tolerated and effective in adults with LBCL. BsAbs appear to have superior tolerability, but inferior efficacy to CAR T-cell therapies in adults with LBCL. Future research on safety and efficacy should focus on evaluating adverse event timing and management, the impact on the patient’s quality of life, the burden on the healthcare system, and overall survival outcomes.


Introduction
Large B-cell lymphomas (LBCLs) are a heterogeneous group of high-grade tumors within the non-Hodgkin's lymphoma (NHL) classification [1].These tumors arise from cells of mature B-cell lineage and are typically characterized by large lymphoid cells forming sheets or clusters [2].Diffuse large B-cell lymphomas (DLBCLs) are the most common form of LBCL in the Western world and represents 30-40% of cases of NHL globally [1].The highest proportion of cases are diagnosed in 65-74 year olds, with a slight male predominance [3].The first-line treatment for LBCL is chemoimmunotherapy consisting of an anthracycline backbone and an antigen-directed monoclonal antibody, with the most widely used regimen being rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) [4].For the 35-40% of patients with relapsed/refractory disease, current treatments include salvage chemotherapy with autologous stem cell transplant [4], antibody-drug conjugates [5], monoclonal antibodies [6], and chimeric antigen receptor (CAR) T-cell therapy [7].
Bispecific antibodies (bsAbs) are an emerging class of therapy in the treatment of adults with LBCL, with at least 16 bsAbs currently undergoing clinical trials in LBCL [8].BsAbs are an off-the-shelf immunotherapy constructed to target two antigens through two separate arms [8].By binding with immune effector cells with one antigen-binding region and the tumor-associated antigen with the other antigen-binding region, bsAbs promote immune-mediated immune cell activation and tumor cell death [8].The commonly targeted tumor-associated antigens (TAA) in lymphoma are CD19 and CD20, both highly expressed on the surface of B-cells.Upon contemporaneous binding of TAA on the tumor cell and an activating CD3 moiety on the T-cell surface, T-cell activation occurs resulting in perforin/granzyme release and subsequent tumor cell lysis [9].An alternative immune cell target is CD47, an antigen upregulated in B-cell malignancies and other cancers, that inhibits phagocytosis by macrophages and the so-called "Don't eat me" signals of associated cell types.Co-engaging CD47 with CD19 permits specific phagocytosis of B-cells and also appears to prevent CD19 cluster formation, thereby inhibiting B-cell receptor proliferation [10].
Overt immune activation is a class effect of bsAbs, resulting in characteristic toxicities, such as cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS) and, rarely, immune effector cell-associated hematologic toxicity (IC-AHT) [8].On-target off-tumor toxicities can result in B-cell depletion and subsequent hypogammaglobulinaemia with ensuant risk of infection, further compounded by immunosuppression due to CRS and ICANS [11].In 2023, the U.S. Food and Drug Authority approved the bsAbs, epcoritamab and glofitamab, for relapsed/refractory LBCL, paving the way for subsequent antibodies and trials [12,13].The growing number of new drugs and trials mandates a systematic analysis of the efficacy and adverse effects of bsAbs in adults with LBCL.
A preliminary search of the literature shows that current research is focused on the safety and efficacy of bsAbs in the wider NHL population [14][15][16][17].While patients with LBCL are included in these publications, combining outcomes for indolent and aggressive lymphomas could result in misestimation of the toxicities in LBCL.Patients with LBCL may be at differential risk of adverse events due to rapidly progressive lymphoma, comorbidities, and toxicities from prior therapies [18][19][20].Additionally, recent publications on lymphoma and other B-cell malignancies have highlighted infections as a significant toxicity risk of bsAbs [21,22].As infections have been established as an independent predictor for mortality in DLBCL [20], it is important to review the rates of this toxicity in adults with LBCL exposed to bsAbs in comparison to standard-of-care therapies.Further research is needed to address these gaps in the literature on the toxicities of bsAbs in adults with LBCL.
The aim of this systematic review is to evaluate the safety and efficacy of bsAbs in adults ≥18 years with LBCL.The primary objective is to identify the frequently reported treatment-emergent adverse events (TEAEs) in this population.The secondary objective is to evaluate the efficacy outcomes of bsAbs in this setting.By collating data on adverse events and efficacy outcomes associated with bsAbs in adults with LBCL, any prominent trends can be identified and evaluated.This may also highlight the potential impact of adverse events and any important differences between bsAbs and standard-of-care therapies.

Materials and Methods
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines and checklist [23].

Search Strategy
The search strategy was designed in consultation with Patrick Condron, Librarian at the University of Melbourne.The search was performed on 10 April 2024, using key databases reporting clinical trials results.These included PubMed, EMBASE, CENTRAL, and pre-print servers bioRxiv and medRxiv.The MeSH terms and keywords used included "toxicity", "safety", "tolerability", "lymphoma", "bispecific antibody", and "T-cell engager".The search was limited to English language results published between 1 January 2008 and 10 April 2024.A search of the grey literature was also performed through selected conference abstract books to capture data not yet published or indexed in the databases.A targeted search of the conference abstracts was repeated on 18 July 2024 to capture any new study results reported after 10 April 2024.The results of the search strategy were imported into EndNote 21.3 [24].For the full search strategy, refer to Supplementary Materials Table S1a,b.

Eligibility Criteria
Screening was performed using the online systematic review software Covidence [25], according to predetermined eligibility criteria.Duplicates were automatically removed by Covidence.Title and abstract screening were performed by a single author (EBM), while full-text screening was performed by two authors independently (EBM and SF).Disagreements were discussed, to reach a consensus.
The inclusion criteria included: (1) interventional clinical trials of bsAbs; (2) adults ≥ 18 years with newly diagnosed or relapsed/refractory LBCL; (3) published in English between 1 January 2008 and 10 April 2024; and (4) comprehensive adverse event data reported as TEAEs.The exclusion criteria included: (1) observational studies; (2) metaanalyses; and (3) adverse event data reported only as treatment-related adverse events (TRAEs).TEAEs were chosen for inclusion over TRAEs to reduce the risk of reporting bias by excluding subjective judgements of adverse event causality made in individual studies [26].To give priority to peer-reviewed publications with comprehensive outcomes reported, conference abstracts were only included where a full-text report had yet to be published for the same clinical trial cohort.To capture the most up-to-date data, only the most recent analysis from the same clinical trial cohort was included.

Data Extraction
The study characteristics and outcome data were extracted into predetermined data collection forms.The primary outcome data were extracted independently by two authors (EBM and MT).Additional information was sought from ClinicalTrials.gov(accessed on 26 May 2024) [27], publications related to the included studies, and by contacting the source authors.There were no assumptions made about missing or unclear information, with missing results documented as not reported (NR).
The extracted data for the key characteristics and outcomes of the included studies are presented in tabular form (Tables 1, S2 and S3).The study characteristics included trial design, key eligibility criteria, demographics of enrolled subjects, treatment schedule, median treatment exposure, adverse event mitigation strategies, and adverse event grading criteria used.The adverse event outcomes included CRS, ICANS, infection, fever/pyrexia, fatigue, cytopenias, and other frequently occurring adverse events.These were defined as any grade adverse events reported in >20% of participants to an individual study and grade ≥3 events reported in >5% of participants to an individual study.The efficacy outcomes included the median follow-up (MFU), overall response rate (ORR), complete response rate (CRR), partial response rate (PRR), median duration of the response (DoR), median progression-free survival (PFS), and response criteria used.Data on the adverse event outcomes were sought from the patients included in the safety analysis, while data on the efficacy outcomes were sought from patients included in the efficacy analysis from all the included studies.

Quality and Risk of Bias Assessment
The Risk of Bias Assessment Tool for Nonrandomized Studies (RoBANS 2) [28] was used to assess the risk of bias in the intervention and outcomes from the included studies.

Data Synthesis
Due to the heterogeneity of the included studies, the analysis was performed using a narrative synthesis.The safety synthesis was performed by grouping together related outcomes and reviewing the results from all the included studies reporting on each outcome.Within each outcome group, heterogeneity was explored by reviewing the differences between bsAbs in monotherapy or combination therapy, treatment-naïve or relapsed/refractory patients, and CD19 or CD20-targeting antibodies.The efficacy analysis was performed by grouping together front-line therapies and salvage therapies.The effect measures were explored through tabular presentation to display the spread of effects.The median effect was calculated for each outcome per bsAb from all the studies reporting results.Where the results for key outcomes were missing, selective non-reporting was assessed to determine the risk of reporting bias.The overall certainty of the evidence for each outcome was assessed based on the sample size and the quality of the contributing studies.

Study Selection
There were 1617 studies identified through the database searches and a further 106 studies identified through grey literature searching.After removing 415 duplicates, title and abstract screening were performed on 1308 studies.By sorting each individual study by the clinical trial identification number, only those studies that reported on the largest, most comprehensive, and recent results from the same dataset were included.Double-blind, full-text screening was performed independently on the remaining 127 studies.There were 19 studies selected for inclusion, according to Figure 1.
For the safety outcomes, 10 of the studies were assessed to have low risk of bias across all eight risk-of-bias domains [29][30][31][32]36,41,42,[45][46][47], as presented in Appendix A. The remaining nine studies were assessed to have unclear risk of bias, as they did not provide sufficient information on the grading criteria used for adverse event measurement [33][34][35][37][38][39][40]43,44].The open-label design potentially increases the risk of bias for safety outcomes in domain 5 (blinding of assessors) for all 19 included studies; however, the selection of TEAE over TRAE was chosen to limit this risk.
All of the studies were limited by their non-randomized, open-label trial design without an active comparator.This meant that any comparisons to standard-of-care therapies were of a lower certainty of evidence due to variability in the study design, population, and method of outcome measurement.
We demonstrate that the types of TEAEs from bsAbs in adults with LBCL generally aligns with their use in other hematologic malignancies; however, there exist differences in the incidence of TEAEs.Although significant infection rates were observed, the overall rates were lower than previously observed with bsAbs in multiple myeloma [22].Similar to bsAbs in other settings [11,16,53], the incidence of CRS was high, but the rate of severe CRS was low.The trends were less clear for ICANS, due to heterogeneity in the study design and grading criteria used.One study reported higher rates of ICANS than expected (27%); however, this was a first-in-human study of AZD0486, where the recommended step-up dosing schedule was still being determined [30].The study by Viardot et al., investigating blinatumomab, reported individual neurological events instead of ICANS [29], although many of these events were consistent with symptoms of ICANS [49].High rates of neurological events have also been highly reported in studies of blinatumomab in acute lymphoblastic leukemia (ALL), with any grade rates ranging from 20 to 53% [54].In contrast, Hawkes et al., who reported no cases of ICANS, were investigating a CD19/CD47 bsAb (TG-1801) [31], which does not engage the CD3 immune cells associated with ICANS [49].The rates of fever, fatigue, and cytopenias were generally consistent with previous experience of bsAbs in other hematological malignancies [11,16,53], with no notable new trends observed.The high rates of fatigue are important given the impact of this toxicity on the patient's quality of life [55].However, evaluating the impact of this toxicity is limited by the lack of published patient-reported outcome data from bsAbs in lymphoma.
In comparing the adverse events described with bsAbs in this review to current standard-of-care therapies, there were some similarities, although the overall safety profile of bsAbs appeared unique.The high rates of infection and cytopenias observed are also commonly seen in many other therapies, including R-CHOP [56,57], salvage chemotherapy [58][59][60], antibody-drug conjugates [5], and monoclonal antibodies [6].Some of the other adverse events frequently observed in bsAbs, including fatigue, fever, and peripheral neuropathy, are also associated with cytotoxic therapies [56][57][58][59] and antibody-drug conjugates [5].The rates of fatigue and fever reported here are comparable to polatuzumab vedotin (PV), while the rates of peripheral neuropathy were reported as lower with bsAbs in this review [5].The peripheral neuropathy observed with PV is due to the anti-microtubule agent conjugated to the antibody [60], while neuropathy in bispecific antibodies may be part of the neurotoxicity process triggered by systemic inflammation and cytokine release.Other adverse events commonly observed in standard-of-care therapies, such as cardiotoxicity, mucositis [56][57][58][59], or rashes [6], were not reported in the studies reviewed herein.This was an expected finding due to the unique mechanism of action of bsAbs, where toxicities are closely related to on-target off-tumor effects [8].
The most notable difference between the bsAbs reviewed here and standard-of-care therapies were the rates of CRS and ICANS.These toxicities are associated with treatments that target immune effector cells, so are primarily observed in patients exposed to bsAbs and CAR T-cell therapies [49].The rates of CRS in this review ranged from 0% to 72.2%, in comparison to the pivotal CAR T-cell trials, where the overall rates of CRS were reported to range from 42% to 92% [61][62][63][64].The incidence of ICANS was also lower in this review (ranging from 0 to 27%) compared to CAR T-cell therapies, where the rates ranged from 21% to 67% [61][62][63][64].The lower rates of CRS and ICANS found may be partly due to the dosing schedule for bsAbs.While CAR T-cell therapies are administered as a single dose, most of the included bsAbs were administered using step-up dosing schedules in order to modulate the T-cell activation and cytokine release, thus mitigating severe toxicities.The difference in toxicity rates may also be due to the more highly activated nature of the manufactured T-cells in CAR T-cell therapies, resulting in increased inflammatory cytokine release and associated toxicities [65].Along with lower rates of grade ≥3 cytopenias and infections, the bsAbs reviewed here appear less acutely toxic than CAR T-cell therapies in adults with LBCL.Long-term toxicities of CAR T-cell therapies in B-cell lymphomas include prolonged cytopenias and hypogammaglobulinemia [66].Data on the long-term adverse effects of bsAbs are not yet available, but such effects may be similar or possibly more severe due to sustained B-cell depletion as a result of repeated dosing [67].
The efficacy outcomes of bsAbs reported in this review show promising responses in both the front-line and relapsed/refractory setting for patients with LBCL.The ORR reported for the bsAbs used in combination, in newly diagnosed patients, show superior outcomes to the reported ORR with traditional R-CHOP therapy, of 70% [68].Responses to bsAbs in relapsed/refractory patients with LBCL also appear superior.The median ORR of bsAbs was reported as 40% to 91%, compared to approximately 26% with traditional salvage chemotherapy [69].In contrast, comparisons to CAR T-cell therapy show inferior efficacy outcomes reported here, than in the pivotal CAR T-cell clinical trials.In relapsed/refractory patients, the ORR for lisocabtagene maraleucel, tisagenlecleucel, and axicabtagene ciloleucel have been reported as 52% to 82% [61,63,64,70].
Of the studies excluded from this review due to insufficient safety outcome data, the efficacy outcomes recently reported by Abramson et al. in June 2024 are noteworthy [71].This phase III randomized trial (STARGLO) investigated glofitamab plus gemcitabine with oxaliplatin (Glofit-GemOx) versus rituximab-GemOx, in 274 patients with relapsed/refractory DLBCL.At an MFU of 20.7 months, the responses were superior in the Glofit-GemOx group, with a median overall survival of 25.5 versus 12.9 months, a median PFS of 13.8 versus 3.6 months, and a median CRR of 58.5% versus 25.3% [71].We await the final published report for both the efficacy and TEAE and subsequent inclusion within our specified criteria.Nonetheless, the data from STARGLO support the notion of the efficacy of bsAbs in both the front-line and relapsed/refractory setting when given as monotherapy or in combination with chemotherapy.
Two important questions arise for future empirical bedside and bench studies.The first is the utility, or lack thereof, of combining bispecific antibodies with chemotherapy.On one hand, the addition of chemotherapy may augment tumor clearance and promote a more immune-permissive microenvironment, thereby increasing the ORRs of bsAbs.On the other hand, chemotherapy may deplete important T-cell subsets that are co-opted by bsAbs.Further, bench-based correlative and comparative clinical trials will be needed to elucidate this important question.An additional, important question in the coming years will be the optimal sequencing of bsAbs and CAR T-cell therapies.Several of the included studies reviewed here found that bsAbs remained efficacious in patients previously exposed to CAR T-cell therapy [32,35,36,41,43,47].A recent retrospective study of 47 patients with LBCL also found no impact on the efficacy of CAR T-cell therapies in patients previously exposed to bsAbs [72].Without long-term efficacy data, it is not yet possible to confirm whether these observations will translate into positive overall survival outcomes.
Overall, the results of this systematic review provide evidence that bsAbs are generally safe and effective in adults with LBCL.While the observed adverse events were consistent with expectations, the frequently occurring adverse events observed require careful consideration due to their potential impact on the morbidity and mortality of patients.The high infection rates found are important to emphasize, as prior research has shown infection is an independent risk factor for mortality in DLBCL patients [20].Specialist care and inpatient management are likely to be required for the management of CRS, ICANS, and cytopenias [53,73].Adverse event mitigation strategies, such as those used in the studies in this review, along with regular monitoring and early intervention, will be important in managing the safety of bsAbs in adults with LBCL.

Limitations
There were several limitations in the study design and method of the outcome measurements used in the 19 studies included in this review.The heterogeneity in the study design and interventions precluded pooling of the studies, thus limiting the certainty of the observed effects.Although we undertook a broad search strategy for published and unpublished data, the reported data that may be within clinical trial registries were not included.Nonetheless, the targeted inclusion of trials assessing LBCL patients allows population-specific observations to be drawn.
All the studies were non-randomized trials without active comparators, thereby limiting the certainty of any comparisons to standard-of-care therapies [74].The open-label design may have also contributed to adverse event detection bias due to participants' and investigators' awareness of the treatment assignment [75].Individual trial results may have been influenced by selection bias, with small sample sizes and targeted recruitment of participants with high functional status [76].This also limits the generalizability of the results.Despite this, other demographic factors, such as age and gender, appear to be representative of the wider LBCL population [3].
Comparisons between the safety and efficacy in terms of bsAbs and standard-of-care therapies are further limited by the variability in outcome measurement methods used in the included studies.This was particularly true for ICANS, where only three of the studies reported using the currently recommended ASTCT consensus grading [49], meaning the overall ICANS rates may be underreported.Furthermore, the method of grading adverse events was unclear in nine of the studies.Also underreported, and thus not synthesized herein, was the timing, impact, and/or management of adverse events, an important aspect for the clinical implementation of these therapies.The overall impact of these limitations on the validity and accuracy of the results of this systematized review is unclear until results from randomized controlled trials (RCTs) using consensus grades are included.

Future Directions
There are several areas that would benefit from additional research to further describe the safety and efficacy of bsAbs in adults with LBCL.With RCTs currently underway for several of the bsAbs in adults with LBCL [77][78][79][80][81], this review of the literature and metaanalysis of the data should be repeated once these results become available.With larger sample sizes it would also be possible to perform subgroup analyses to investigate the differences in outcomes among different populations, such as age or gender groups.The impact of adverse events on the patient's quality of life and the burden on the healthcare system should also be explored in more detail.Future research could incorporate patientreported outcome data and invite consumer input to guide decisions on which adverse events are considered important.As access to drugs is often driven by health technology assessments, future research into the financial impact and burden on the healthcare system of managing the observed adverse events would be valuable.Longer term follow-up of the current clinical trials should be conducted to evaluate the impact of efficacy outcomes of bsAbs on overall survival trends.Overall, data on the safety and efficacy of bsAbs in adults with LBCL are still evolving and further research is required to gain a comprehensive understanding of this topic.

Figure 1 .
Figure 1.Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) flowchart for study selection.Adapted from the PRISMA statement, 2020 [23].n = number of studies.

Table 1 .
Key study design characteristics of the included studies.

Table 2 .
Median reported rates (%) of selected adverse events per bispecific antibody.

Table 4 .
Other frequently reported adverse events per bispecific antibody.