Toxicity, efficacy, and clinical management of bispecific antibodies in routine clinical practice: A real-world geltamo study in B-cell lymphoma

Abstract Background: CD20/CD3 bispecific antibodies (BsAbs) have reshaped the treatment of relapsed/refractory large B-cell lymphoma (LBCL) and follicular lymphoma (FL). Their use in routine care brings new challenges, especially in managing T cell–mediated toxicities such as CRS and ICANS, underscoring the need for multidisciplinary strategies and real-world data to refine clinical pathways and resource use. Methods: We conducted a retrospective, multicenter study of consecutive adults with relapsed/refractory B-cell lymphoma treated with commercial CD20/CD3 BsAbs across the GELTAMO network. Since glofitamab (glofit) is not reimbursed in Spain, that group was included through an early access program, as was epcoritamab (epco) in FL. Patient data included baseline demographics, prior treatments, toxicity, clinical management approaches, hospital resource utilization, response rates per Lugano 2014 and survival outcomes. Results: Among 168 patients, 86 (51%) received epco, 59 (35%) mosunetuzumab (mosun), and 23 (14%) glofit. The median age was 64 years (range 36-86), and 59% were male. Regarding histology, 58% had LBCL —including 12% with high-grade B-cell lymphoma (HGBCL)— and 42% had FL. Overall, 38% of patients would not have met the general eligibility criteria commonly applied in clinical trials, primarily due to organ dysfunction or cytopenias. In the LBCL cohort, 64% of patients were primary refractory, 51% were refractory to last previous treatment, 33% had bulky disease, and 78% presented extranodal involvement. The median number of prior treatment lines was 3 (range 2–4), with 45% having received two lines and 55% three or more. Additionally, 54% had previously received CAR-T therapy, and 9% had undergone a hematopoietic stem cell transplant. In the FL cohort, the median number of prior lines was also 3 (range 2–5); 44% had received two lines and 56% more than two. Progression within 24 months of first-line therapy (POD24) occurred in 62% of patients. Prior bendamustine exposure was reported in 44%, including 16% who received it within 9 months before BsAb initiation. Twenty-nine percent of patients had undergone a prior transplant, and 10% had received CAR-T therapy. Patients with aggressive lymphomas predominantly received epco (75%) and glofit (25%), while FL-patients received mosun (80%) and epco (20%). Treatment was delivered in an inpatient setting in 86% of cases and, among the 14% without a planned admission, most (90%) received mosun. The median cumulative hospital stay during the entire Step-Up dosing schedule was 12 days (range 1–92). Focusing on toxicity, CRS was observed in 42% of patients (45% epco, 45% glofit, 48% mosun): Severe CRS (Grade ≥3) appeared in 2% (0% epco, 4% glofit, 1.8% mosun). Most (71%) CRS events occurred during cycle 1. Tocilizumab was administered in 31% and corticosteroids in 37%. The median duration of steroid use was 3 days (range 1–12). ICANS occurred in 4.3% of patients, predominantly grade 1 (98%), with only one case each of grade 2 and 3. Any grade cytopenias occurred in 40% (grade ³3 in 30%). Infections were reported in 27% (9.5% were grade ³3) predominantly respiratory (77%) and mainly during the first four cycles. , Among the 8% of patients requiring ICU admission, 9 cases were due to CRS (in 2, ICU stay was solely for treatment administration), 3 were related to severe infections, and 1 patient was admitted for complete AV block requiring pacemaker placement. There were 34 deaths, 24 (71%) related with lymphoma progression (7 FL, 17 LBCL). Non-relapse mortality was 5.9%, mainly due to infections (80%). Regarding efficacy, overall response rate (ORR) and complete response (CR) rates were 43% (23%) in LBCL and 75% (49%) in FL. In patients previously treated with CAR-T therapy, ORR(CR) was 44.6% (21.7%) in LBCL and 66.7% (50%) in FL. With a median follow-up of 8.8 months, median progression-free survival and overall survival were 4.0 months (95% CI: 2.8–9.0) and 11 months (95% CI: 6.8–NR) for LBCL, and 17 months (95% CI: 7.1–NR) and not reached for FL. Conclusions: In this large, real-world BsAb cohort, efficacy was lower than pivotal trials, likely reflecting the broader and frailer patient population treated in routine practice. Overall toxicity was comparable to trials, but hospitalization duration remained substantial, emphasizing the need for optimized outpatient management. Infections were frequent and contributed significantly to non-relapse mortality.