Citations

Abstract Introduction: CAR-T therapy has transformed cancer therapy. with its first Food and Drug Administration (FDA) approved product in August 2017. Given the expanding indications and growing experience, particularly with management of immediate toxicities requiring specialized expertise such as Cytokine Release Syndrome (CRS) and Immune Effector Cell-Associated Neurotoxicity syndrome (ICANS), FDA recently removed the Risk Evaluation and Mitigation Strategies (REMS) for CD19 and BCMA targeting CAR-T on June 26th2025. In particular, given that CRS and ICANS typically occur in the first 2 weeks post CAR-T infusion, the minimum required stay near CAR-T treatment center is reduced from one month to 2 weeks. While this could reduce treatment burden for patients (pts) who need to travel away from home to receive CAR-T, there is a need to update the transition of care from CAR-T center to pts' local oncology practice. Pts carry in their rate of recovery and have varying supportive care needs in the second half of the month post CAR-T infusion. To assist with care transition and planning, we examine the use of AI multimodal large language model (LLM) to predict patients' supportive care needs 2 weeks after CAR-T infusion based on the clinical data at the time of CAR-T evaluation, prior to leukapheresis. This timepoint was selected to also maximize travel, lodging and caregiver planning for pts and family as well. Methodology: Pts who were treated with FDA-approved CAR-T at Mayo Clinic Rochester between December 2022 and June 2025 were examined. Based on experiences with referral practices in the Midwest region, we defined patients with high monitoring needs as those who between day 14-30 post CAR-T had either hospital admission due to recurrent CRS or ICANS or required transfusion of more than 1 blood product per week (red blood cell or platelets). Others were classified as low monitoring needs. Clinical notes, labs, vital signs and electrocardiogram at the time of CAR-T evaluation were abstracted and summarized by Gemini LLM for association with patients with high versus low monitoring needs post CAR-T. Model was developed and tested for predicted needs and concordance with actual outcome of these pts. Results: Among the 362 pts who received FDA approved CAR-T in the timeframe, 84 pts had high monitoring needs. An additional 84 pts with low monitoring needs were randomly selected from 278 pts. These were divided into a training cohort and test cohort of 84 pts each, with 42 pts with high and 42 pts with low monitoring needs. Demographics of the two cohorts displayed no statistical difference. LLM analysis of the clinical data at the time of CAR-T eval identified 5 categories associated with risk for high monitoring needs: 1) Disease status (clinical notes about bulky, progressive disease or CNS involvement); 2) inflammatory and tumor burden markers (CRP, LDH); 3) hematologic status (platelet, plt; neutrophil count ANC); 4) renal function; 5) performance status. Using common clinical lab parameter convention (platelets <100, ANC< 1, CrCl< 30ml/min), and prioritizing for not missing pts with high monitoring needs, LLM model was defined to predict patients for high monitoring needs if high risk feature was identified in category 1 or any category combinations; whereas patients with no high risk feature or only 1 category that is not disease status was predicted as low monitoring needs. In this model, the sensitivity was 85.7%, specificity 23.8%, F1 score 65.5%. Applying this model to test cohort 2, the sensitivity was 83.3%, specificity 28.6%, and F1 score 65.4%. Discussion: We demonstrate proof-of-concept that LLM can be used to analyze clinical data at the time of evaluation for CAR-T eligibility to predict for supportive care needs post treatment. Not surprisingly, features identified by LLM to be associated with high monitoring needs are those previously reported by us and others at the time of CAR-T evaluation and or pre-lymphodepletion to be associated with risk for cytopenia, infection, hospitalization and non-relapse mortality. Based on the summary output generated by LLM, additional opportunities for classification methods were identified to improve specificity and balanced accuracy of the LLM model. This is a promising tool for further development to assist clinicians and patients to plan their treatment journey.

Abstract Background Chimeric antigen receptor T cell (CART) therapy has revolutionized treatment for hematologic malignancies. However, durable remission following anti-CD19 CART therapy (CART19) remains limited to 40–50% of patients, often due to tumor relapse and CART dysfunction (PMID:31501612). We previously demonstrated that tumor-derived extracellular vesicles (EVs) from patients with chronic lymphocytic leukemia (CLL) can directly impair CART function (PMID:33388419). In this study, we aimed to identify tumor-derived EV microRNA (miRNA) cargo associated with clinical CART failure in B cell malignancies and define how they mechanistically drive CART cell dysfunction. Methods To achieve these aims, we profiled tumor-derived EV miRNA cargo using small RNA sequencing (RNAseq) in two independent patient cohorts. Cohort 1 compared plasma-derived EVs from untreated CLL patients to those from healthy donor EVs to identify miRNAs enriched in CLL (n=23). Cohort 2 included patients with diffuse large B cell lymphoma (DLBCL) treated with axicabtagene ciloleucel (CART19, n=12). Here, plasma EV miRNAs were analyzed at baseline (pre-CART treatment) and 1-month post-CART treatment and compared between CART responders (complete remission at 6 months) and nonresponders (no response at 1 month). Enriched miRNAs from both cohorts were identified following miRNA count normalization and differential expression analysis using adjusted p values of ≤0.05. Functional studies were conducted by transfecting CART19-28ζ cells with synthetic miRNA mimics or scrambled controls. CART cells were also generated with stable overexpression of miRNA via lentiviral shRNA delivery. CART function was assessed via in vitro killing, proliferation, apoptosis assays, RNAseq, and in vivo tumor xenograft models. Results In Cohort 1, we identified several miRNAs including miR-125a to be significantly enriched in CLL-derived EVs relative to healthy controls. Similarly, in Cohort 2, miRNA-125a was also identified among the miRNAs enriched in nonresponders vs. responders to CART treatment; Notably, miR-125a was significantly enriched in nonresponders at baseline (pre-CART treatment) and 1-month post-CART treatment (p=0.0177, p=0.0196). To define the functional effects of miR-125a, we delivered synthetic mimics or stable shRNA constructs encoding miR-125a into CART cells. Delivery of miR-125a was confirmed by RT-qPCR (p<0.0001 for mimic vs. scramble, p=0.0069 for shRNA vs. scramble). Here, miR-125a delivery and overexpression impaired CART cytotoxicity/killing of target tumor cells (~25% reduction, p= 0.0039), decreased antigen-specific proliferation (~47% reduction, p=0.0002), and increased apoptosis following antigen stimulation (1.8-fold increase, p=0.0002) as compared to scramble controls in vitro. We next studied how miR-125a impacts CART functions in vivo. Here, tumor-bearing mice treated with miR-125a–overexpressing CART cells (CART sh125a) exhibited accelerated tumor progression (4.1-fold increase, p=0.0019) and significantly reduced survival as compared to mice treated with scramble control CART treatment (CART shScr) (20% vs 80% survival at day 35, p=0.0064), underscoring the dramatic detrimental impact of miR-125a on therapeutic efficacy. To uncover changes in CART gene expression induced by miR-125a delivery, RNAseq was also performed across stimulated CART sh125a vs. CART shScr cells. CART sh125a cells demonstrated broad transcriptional changes, with gene set enrichment of E2F targets, G2/M checkpoint, and apoptosis pathways (FDR p<0.0001, p<0.0001, and p=0.006) further validating the role of miR-125a in disrupting CART cell cycle control, proliferation, and survival. Conclusion In summary, our study identifies miR-125a as a tumor-derived EV cargo that drives CART dysfunction in two separate B cell malignancies (CLL and DLBCL). miR-125a was enriched in both CLL EVs and nonresponders to CART19 therapy at baseline prior and post CART treatment. Functional studies demonstrate that miR-125a delivery and overexpression directly impair CART cytotoxicity and proliferation, increase CART apoptosis, reduce therapeutic efficacy in tumor xenograft models, and disrupt key pathways related to CART cell cycle regulation and survival. These findings position miR-125a as both a key mechanistic driver of CART failure and a potential predictive biomarker for treatment response, while highlighting it as a possible therapeutic target to improve CART outcomes in the clinic.

Abstract Introduction: Elranatamab (Elra) is a bispecific antibody (BsAb) targeting BCMA that is approved for patients with triple class exposed relapsed refractory myeloma (RRMM) on the basis of the MagnetisMM-3 phase 2 trial. For BCMA naïve patients, an overall response rate (ORR) of 61.0% and median progression free survival (PFS) of 17.2m was observed with a median overall survival (OS) of 24.6m. However, efficacy and safety information in the real world is lacking. Methods: This was an international retrospective study of patients with RRMM treated with Elra monotherapy outside of clinical trials. Data was collected through the International Myeloma Foundation Immunotherapy Database from 7 countries (UK, USA, Canada, Greece, Spain, Czech Republic and Singapore). High-risk cytogenetic abnormalities (HRCA) were defined as 1q+, t(4;14), t(14;16), t(14;20), and/or del(17p). Response was assessed using IMWG consensus criteria. Patient characteristics were summarized by frequency (percentage) or median (range). PFS, OS were evaluated using the Kaplan-Meier method and reported with 95% confidence intervals. Results: 79 patients were included. Median age was 67 years (range: 41-95y) with 20% ≥75 years. Racial distribution was: 75% White, 16% Black/African American, 8% Asian/Pacific. 55% were ECOG performance status (PS) 1-2. Patients received a median of 5 prior lines (range 1-11), and 35% were BCMA exposed (9% CAR-T, 5% ADC, 7% BsAb, 8% BsAb plus CAR-T, 4% ADC plus BsAb). Of those with available data, 45% had HRCA,13% had extramedullary disease (EMD) and 32% had penta-drug refractory RRMM. Overall, 48% would not have been eligible for the MagnetisMM-3 trial. Median follow-up was 6.3m (range 0.5-35.0). ORR was 53% (PR 11%, VGPR 33%, CR 8%, sCR 1%) and 67% for BCMA naïve patients. Responses were significantly lower for BCMA exposed patients (ORR 31%) but did not differ by age (51% <75yrs vs 60% ≥75yrs). On univariate analysis, prior BCMA exposure, platelets <50, presence of HRCA and penta-refractoriness were significantly associated with a lower response rate. PFS for the whole cohort at 6 and 12m was 64% (95%CI, 54-77%) and 52% (40-67%); for BCMA naïve patients, it was 77% (65-90%) and 67% (54-84%), respectively. Within the whole cohort, PFS was shorter for: BCMA exposed (median 4.0m vs not reached (NR)), platelets <50 (2.0 vs 21.7m), HRCA (7.2m vs NR), penta-refractory (4.3m vs NR). PFS did not differ significantly by EMD, PS, age, or creatinine clearance. OS for the whole cohort at 6 and12m was 74% (64-85%) and 58% (45-74%); for BCMA-naïve patients, it was 80% (43-85%) and 70% (57-87%), respectively. Within the whole cohort, OS was shorter for: BCMA exposed (median 10.3m vs NR), platelets <50 (4.3m vs NR), HRCA (10.7m vs NR) and penta-refractory (10.3m vs NR). Univariate regression analysis demonstrated that prior BCMA exposure, platelets <50 and penta-refractoriness were significantly associated with a worse PFS and OS, and HRCA predicted poorer survival CRS was observed in 41% of patients (grade(G)1: 35%, G2: 5%, ≥G3: 0%) with most occurring with the first step up dose (SUD): SUD1: 29%, SUD2: 15%, full dose: 4%. Tocilizumab was used in 20% and steroids in 3%. ICANS was observed in 4 cases (5%): G1: 2, G2: 1, G4: 1. Treatment was with steroids (n=3) and tocilizumab (n=2). G3-4 thrombocytopenia was observed in 19% and 14% required platelet transfusions. G3-4 neutropenia occurred in 25% with overall 31% requiring GCSF. Infections were reported in 33 (42%) patients with median time to infection of 41 days. A total of 94 infection episodes were observed over 581 personal months. There were 2 infection-related deaths reported (1 had IVIG, 1 did not). 64% had hypogammaglobulinemia (IgG < 400mg/dl) and IVIG was given in 65% of patients. Conclusions: In this first international real world analysis, Elranatamab achieved a response rate of 67% and a 12m PFS of 67% in BCMA naïve patients, similar to the MagnetisMM-3 trial despite almost half not meeting trial eligibility. However, prior BCMA exposure, platelets <50, high risk cytogenetics and penta refractory status were predictors of poorer outcomes. Acknowledgements: The International Myeloma Foundation, International Myeloma Working Group (IMWG)

Abstract Introduction: Isocitrate dehydrogenase (IDH) mutations are observed in ~20% of patients with acute myeloid leukemia (AML). In the U.S., ivosidenib (ivo) is approved with or without azacitidine (aza) in the frontline (FL) and as monotherapy in relapsed/refractory (R/R) IDH1-mutated (mut) AML. Olutasidenib (oluta) is approved as monotherapy for R/R IDH1-mut AML. Enasidenib (ena) is approved as monotherapy for R/R IDH2-mut AML. This study describes real-world IDH inhibitor (IDHi) administration patterns and evaluates their outcomes in on-label settings. Methods: A retrospective review of patients with IDH1/2-mut AML who received ivosidenib, olutasidenib, or enasidenib was conducted at 12 academic institutions in the U.S. Best response at any time during treatment, including complete response (CR) and complete response with partial hematologic recovery (CRh), were graded using European Leukemia Net (ELN) 2022 criteria. Results: A total of 387 patients with IDH1- and IDH2-mut AML were treated with an IDHi in the FL or R/R setting. Median follow-up was 9.6 months (range, 0.03-90.7). Demographics include 79% White, 14% Black, 3% Asian, and 4% other/unknown race; 5% were Hispanic ethnicity; 46% were female; the median age was 70 years (range, 28-95). 70% had an Eastern Cooperative Oncology Group (ECOG) performance status of 0-1, 26% had secondary AML, and 10% had therapy-related AML. Per the ELN 2024 less-Intensive risk classification, 58% were favorable risk, 29% intermediate, 7% adverse, and 6% unknown. Overall, 209 of 387 patients (54%) received IDHi regimens on-label. In FL IDH1-mut AML (n=45), 11 received ivo monotherapy, 25 aza/ivo, and 9 used off-label combinations: 3 hypomethylating agent (HMA) + venetoclax (ven) + ivo, 5 decitabine/ivo, 1 intensive chemotherapy (IC)/ivo. Among R/R IDH1-mut AML (n=109), 47 received ivo monotherapy, 7 oluta monotherapy, and 55 off-label (31 HMA/ivo, 16 HMA/ven/ivo, 3 IC/ivo, and 5 other). 44 received off-label ena for FL treatment of IDH2-mut AML (25 ena monotherapy, 9 HMA/ena, 8 HMA/ven/ena, and 2 IC/ena). In R/R IDH2-mut AML (n=189), 119 patients received ena monotherapy and 70 received off-label combinations (25 HMA/ena, 30 HMA/ven/ena, 7 IC/ena, 6 ena/gilteritinib, and 2 other). In on-label treatment of FL IDH1-mut AML, one of four patients with response assessment of FL ivo monotherapy had a CR/CRh. Of 19 patients with response assessment on FL aza/ivo, 13 patients (68.4%) had a CR as best response (95% CI, 46.0-84.6%). Median OS was 14.9 mo (95% CI, 4.17-NR) for ivo monotherapy (n=11) and 31.9 mo (95% CI, 9.1-NR) for aza/ivo (n=25), with 0 and 2 patients receiving allogeneic stem cell transplant (allo-SCT), respectively. In R/R IDH1-mut AML, the rate of CR/CRh was 35.5% (95% CI, 21.1-53.1) for on-label ivo monotherapy. All 4 patients with response assessment for on-label oluta monotherapy had a CR. Median OS was 14.0 mo (95% CI, 10.1-22.4) for ivo monotherapy and NR for oluta monotherapy (95% CI, 10.1-NR), with 10 and 0 receiving allo-SCT, respectively. The CR/CRh rate for ivo was 38.1% (8 of 21 assessed; 95% CI, 20.8-59.1) in age < 70, and 30.0% (3 of 10 assessed; 95% CI, 10.8-60.3) in age ≥ 70. Median OS was 15.2 mo (95% CI 9.3-NR) in age < 70 (n=27) and 10.4 mo (95% CI 8.0-18) in age ≥ 70 (n=20). In R/R IDH2-mut AML, the rate of CR/CRh for on-label ena monotherapy was 38.2% (95% CI, 27.6-50.1). Median OS was 12.1 mo (95% CI, 8.2-16.6), with 5 having received allo-SCT. In demographic subgroups for ena monotherapy, 52 non-Hispanic White (NHW) patients with response assessment had a CR/CRh rate of 42.3% (95% CI, 0.30-0.56). Only 1 of 13 non-Hispanic Black (NHB) patients with response assessment had a CRh (7.7%; 95% CI, 1.4-33.0) as best response, with none attaining CR. Median OS was 13.7 mo (95% CI 10.3-26.3) in NHW (n=82), 11.2 mo (95% CI, 6.1-NR) in NHB (n=18), and 69.1 mo (95% CI, 52.6-NR) in Hispanic patients (n=8). The CR/CRh rate was 48.6% (17 of 35 assessed; 95% CI, 33.0-64.4) for age < 70, and 27.3% (9 of 33 assessed; 95% CI, 15.1-44.2) for age ≥ 70. Median OS was 12.0 mo (95% CI, 9.9-22.1) in age < 70 (n=48) and 5.8 mo (95% CI, 4.1-8.7) in age ≥ 70 (n=69). Conclusion: Practice patterns for IDH inhibitors in AML vary considerably, with high amounts of off-label use. Response rates for ivosidenib and enasidenib in the real-world setting are comparable to those of clinical trials. Descriptive analyses of demographic subgroups highlight areas in need of further research.

Abstract Background: The phase II and III GRIFFIN and PERSEUS studies evaluated DVRd vs. VRd in NDMM pts that were transplant eligible (TE). These studies included consolidation (cons) and dual agent maintenance (maint) with dara and len (DR). Little is known about the efficacy of DVRd in RW practice including frequency of bor use, use of cons, use of DR maint, and subsequent treatment after progression (PD) on DVRd. We aim to evaluate the RW efficacy of NDMM pts treated with DVRd across the tri-site Mayo Clinic Comprehensive Cancer (MCCC). Methods: We retrospectively analyzed the medical records of pts treated with DVRd between June 2018 and June 2024 at the MCCC. Outcomes were estimated using the Kaplan-Meier method. Results: 464 pts were included; 278 pts were male, and the median (m) age of MM diagnosis was 64 years (yr). 114 pts had ISS III, 205 pts had R-ISS III, 266 pts had high risk (HR) FISH, 80 pts had extramedullary disease, 165 pts had ultra HR disease (dz) (as per 2025 IMS/IMWG consensus), and 47 pts had renal insufficiency (RI) with creatinine >2 mg/dL. Starting len dose was 25mg in 356 pts, 15 mg in 40 pts and ≤10mg in 65 pts. 399 pts received bor 1x per week and 63 2x per week, 91 pts received 1-2 cycles of a different regimen prior to DVRd due to hospitalization (n=23), RI (n=38), poor performance status (n=6), other (n=34). 315 (67.9%) pts underwent autologous transplant (ASCT). The m number of induction cycles was 5; 4 for pts who underwent ASCT and 6 for pts who did not. The best ORR to induction for pts who underwent ASCT was 98% with 247 (79%) achieving ≥VGPR, and 82 (26%) achieving minimal residual disease negativity (MRD-) to at least 10-5. The best ORR for pts who did not undergo ASCT was 87%. For those who underwent ASCT, (96% with melphalan 200mg/m2), the ORR at day 100 was 97% with 56% achieving MRD-. At 1-yr post ASCT, 30% (93/312) were MRD-. 42 pts (13%) received post ASCT cons with DVRd for a m of 2 cycles. 281 pts received post ASCT maint including 107 (38%) len, 53 (19%) bor-len, 91 (32%) dara-len, 14 (5%) dara, 16 (6%) other. 61(13%) pts had change in treatment, due to change to second line therapy (35; 55%), discontinuation (d/c) of len (11; 17%), d/c of dara (3; 5%), d/c of bor (15; 23%), d/c of dex (0; 0%). The m follow-up time for all pts was 26 months (mo) (17-39). The m PFS of pts who underwent ASCT was not reached (NR), and was 57 mo (52-NA) for pts who did not undergo ASCT. For pts who underwent ASCT, the PFS at 24mo for pts who did not have HR FISH was 93% (88-99%) compared to 80% (73-88%) for pts with HR FISH (p=0.12) and 77% (68-87%) for pts with ultra-HR dz (p=0.013).The PFS at 24mo of pts with RI who underwent ASCT was 81% (66-98%) compared to 86% (82-92%) for those without RI, p=0.57. The PFS at 24mo of pts with RI who did not undergo ASCT was 53% (28-100%) compared to 64% (55-75%) for those without RI, p=0.42. The ORR and MRD- rate for pts who received once-weekly bor was 94% and 24%, compared to 94% and 27% for pts who received 2x weekly bor (p=0.85, p=0.93). For pts who underwent ASCT, the PFS at 24mo for pts who received len maint was 91% vs. 83% for pts who receive dual agent maint (DR or VR) (p=0.25), although pts who got dual agent maint were more likely to have HR FISH than pts who received len maint (p<0.001). The PFS at 24mo for pts that received post-ASCT cons was 91% compared to 85% for those that did not (p=0.53); pts who got post-ASCT cons were more likely to have HR FISH (p=0.07). For pts that underwent ASCT, the 3 yr PFS rate is 85% and the 3 yrs OS rate is 96%. For pts that did not undergo ASCT the 3 yrs PFS rate is 60% and the 3 yr OS rate is 85%. 55 pts had functional HR (FHR) MM (relapse within 18 mo) and m PFS for FHR pts who underwent ASCT was 15 mo (11-NA), while m PFS was NR for pts without FHR MM (p<0.001). 79 pts relapsed after 1st line DVRd. The most common 2nd line therapies (tx) were dara-carfilzomib (K)-D, dara-pomalidomide(P)-D, KPD, and other. The m follow-up time from 2nd line tx was 10 mo, the best ORR to 2nd line tx was 76%, and the m PFS with 2nd line tx was 18 mo (7-NA). Conclusion: In this RW analysis, the efficacy and survival outcomes of NDMM pts treated with DVRd was comparable to that reported in the PERSEUS and GRIFFIN trials, although MRD- rates appear lower. DR maint and post-ASCT cons use was not common and was used more frequently in pts with HR FISH. Pts can be effectively salvaged following PD on front-line DVRd with available tx.

Abstract Introduction: Reporting patient outcomes by CAR-T treatment centers is part of the accreditation requirement under the FACT (Foundation for Accreditation of Cell Therapy) Immune Effector Cell (IEC) Program standard. Currently, this is a time-consuming and manual process, making it increasingly difficult to sustain with the growing CAR-T applications and patient volume. The objective of this study is to evaluate how a Large Language Model (LLM) can be used to automate the abstraction and summarization of medical record for CAR-T toxicities. Methods: Clinical data from patients treated with CAR-T between January 2018 and June 2025, including hematology clinical notes, lab results, flowsheets, vital signs, and medication administration record, were used as the source of patient information. The American Society of Transplant and Cell Therapy (ASTCT) CAR-T toxicity grading guideline for cytokine release syndrome (CRS) and immune effector cell associated neurologic syndrome (ICANS) were used to define these toxicities. With the large context window of the LLM (Gemini 2.5 pro), we created a prompt by embedding both the patient data and the ASTCT guidelines for CRS and ICANS, along with an outline of specific information we aimed to abstract from the data. Additionally, the LLM was instructed to provide contextual justification for the data it extracted from patient information. The LLM-generated output included event onset date, resolution date, maximum grade, date of maximum grade, and medications given for management. The CRS and ICANS outcome reports generated by the LLM were then compared to the IEC compliance program data. Any discrepancies identified were reviewed manually. Results: Among patients (pt) treated with CAR-T in the specified timeframe, 47 were selected as training cohort. This cohort included pt with toxicities events, no events, and a mix of events of varying grades. For the test cohort, 66 pts who received CAR-T from March to June 2025 were included. Demographics for the training and test cohorts are reflective of evolving clinical practice (training, test cohorts: median age 69, 65; male 55%, 74%; lymphoma/leukemia 66%, 48%; multiple myeloma 34%, 52%; CD19 CAR-T 46%, 50%). Compared to data from the IEC compliance program, the LLM-generated summaries for CRS events demonstrated an accuracy of 90%, precision 95%, sensitivity 100%, specificity 80%, F1 score 97%, and Matthew's correlation coefficient 87%. For ICANS event, the LLM achieved an accuracy of 91%, precision 92%, sensitivity 92%, specificity 90%, F1 score 92%, Matthew's correlation coefficient 83%. Maximum grade, for both CRS and ICANS, were identified with >90% accuracy. However, dates for CRS and ICANS were less accurate, with the lowest accuracy noted in identifying the event resolution date, 45% for CRS and 67% for ICANS. Medical records were reviewed for all discrepancies between the LLM and the IEC Compliance Program data. The most common discrepancy was a mismatch in event resolution date, determined to be due to differing criteria used by the compliance team and the LLM prompt. Of note, a few discrepancies were attributed to missed information by the IEC Compliance Team, which was correctly identified and categorised by the LLM. An updated LLM prompt (LLMv2) was developed and applied to the test cohort. In version 2, the prompt was revised to include clearer instructions on how to summarize CRS and ICANS events, as well as new instructions on additional data elements to be summarized. The performance of LLMv2 on this cohort for CRS event identification was 100% across all categories (accuracy, precision, sensitivity, specificity, F1 score, and Matthew's correlation coefficient). For ICANS the LLMv2 achieved accuracy of 96%, precision 69%, sensitivity 100%, specificity 93%, F1 score 82%, Matthew's correlation coefficient 80%. Accuracy in identifying event resolution date improved significantly from 64% to 99%. Conclusions: Our study demonstrates that LLMs can significantly reduce the time required to abstract and summarize CAR-T toxicities, while maintaining high accuracy. One limitation of our study is that it is based on data from a single institution. However, plans are already underway to evaluate the use of LLMs in a multi-institutional setting.

Abstract Introduction: Ciltacabtagene autoleucel (cilta-cel) is a highly effective BCMA-redirected CAR T-cell therapy for RRMM. While the pivotal CARTITUDE-1 trial established a target dose of 0.75 x 10^6 CAR+ viable T-cells/kg, the approved cell dose is 0.5-1.0 x 10^6 and the optimal dosing strategy to balance efficacy and toxicity remains unclear. This question is of interest given that a dose-response association has been demonstrated for idecabtagene vicleucel (a related BCMA-directed CAR T product). It has also been hypothesized that lower doses of cilta-cel may be associated with reduced toxicity. This study aims to evaluate the association between cilta-cel dose and key efficacy and toxicity outcomes using a large, real-world dataset from the US Multiple Myeloma Immunotherapy Consortium. Methods: We conducted a multicenter, retrospective analysis of 751 RRMM patients treated with commercial cilta-cel across 15 US academic centers. The administered cilta-cel dose (million cells/kg) was categorized into four groups: low (<0.5), standard-low (0.5 to <0.6), standard (0.6 to <0.8), and standard-high (0.8 to 1.0). We compared incidence rates of cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), delayed neurotoxicity (DNT), and immune effector cell-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) across these dose categories using logistic regression. Best response rates and progression-free survival (PFS), estimated using the Kaplan-Meier method, were also compared, with differences assessed using the log-rank test for the latter. Results presented are derived from multivariable models adjusting for extramedullary disease, ECOG status, penta-refractory disease and receipt of bridging therapy. Results: Among 751 infused patients, the dose distribution was: <0.5 (n=33), 0.5 to <0.6 (n=131), 0.6 to <0.8 (n=550), and 0.8 to 1.0 (n=37). The four dose groups were well-balanced in terms of baseline patient and disease characteristics, including age, high-risk cytogenetics, and extramedullary disease, with the exception of t(14;16), which was more common in the standard-low and standard dose groups (p=0.006). There were no statistically significant differences in the rates of any grade CRS (p=0.11), any grade ICANS (p=0.84), grade ≥3 ICANS (p=0.36), duration of CRS (p=0.18) or ICANS (p=0.64), or IEC-HS (p=0.16) across the four dose categories. Time to onset of CRS was shorter for the standard-high dose compared to the low dose (6 days vs. 8 days; p=0.08). Importantly, there was no significant difference in the rate of DNT (3.0%, 9.2%, 8.8%, 5.4%; p=0.24). Similarly, there was no significant difference in overall response rates (97%, 96%, 96%, 88%; p=0.50) or complete response rates (75%, 65%, 75%, 66%; p=0.10) by dose. The median follow-up was 12.3 months (IQR: 6.9, 20.5). For patients who received a standard-low dose (0.5, 0.6), the median PFS was 17.5 (IQR: 13.5, NR) months. For the standard dose range of (0.6, 0.8), the median PFS was 28.7 (95% CI: 19.8, NR) months, while the median PFS for the standard-high dose (0.8, 1.0) was not reached. For the low dose range of <0.5, the median follow up was only 8.7 months, which did not allow reliable estimation of the median PFS. While the overall difference in PFS among the groups did not reach statistical significance by the log rank test, there was a clear trend toward improved PFS with higher cell doses in the multivariable model, with outcomes clearly ordered in the same direction as the cell dose categories. A similar trend was also evident in the rates of 12-month PFS which showed an improvement with increasing dose (62% ,67%, 73%, 80%). Conclusions: In this large, real-world cohort of RRMM patients treated with cilta-cel, a lower cell dose was not associated with reduced incidence or severity of CRS, ICANS, DNT, or IEC-HS. Conversely, a dose-response correlation with PFS was observed, with higher doses trending toward longer median and 12-month PFS. It remains unknown if a lower cell dose in the commercial setting reflects poor T cell quality, yet these findings suggest that dose reduction of cilta-cel may not be an effective strategy to mitigate toxicities and might be associated with a trend toward lower efficacy. Therefore, the standard target dose should be utilized to maximize patient outcomes and further prospective dose exploration is warranted.

Abstract Introduction: The GPRC5D-directed bispecific antibody (BsAb) Talquetamab (Talq) has shown promising results for patients (pts) with relapsed/refractory multiple myeloma (RRMM) in the MonumenTAL-1 clinical trial (Chari et al. NEJM 2022). However, there are limited data on the use of Talq in the real-world setting. Methods: All pts with RRMM who started Talq between 8/2023-3/2025 across 15 US academic centers were included. High-risk cytogenetic abnormalities (HRCA) were defined as del(17p), t(4;14), t(14;16) and/or gain/amplification 1q21. Time-to-event analyses were estimated using the Kaplan-Meier method. Toxicities including CRS and ICANS were graded using CTCAE v5.0 or ASTCT criteria. Results: A total of 484 pts were included, with a median age of 65 years (range 35-89), and 83 pts (17%) were aged ≥75 years. 20% of pts were black. 21% had baseline CrCl<40, 65% had HRCA and 29% had >1 HRCA (i.e Ultra high-risk MM). The median number of prior lines of therapy was 6 (IQR 4, 8). 50% were penta-refractory, 69% received at least one prior BCMA-directed therapy (46% CART; 32% BsAb). Prior to initiation of Talq, 11 pts (2%) had CNS MM, 29 (6%) had PCL and 126 (26%) had true extramedullary disease (EMD). 70% would have been ineligible for the MonumenTAL-1 trial. For most pts (n=405; 84%) the dosing strategy from Cycle 2 was 0.8 mg/kg every 2 weeks. and the 0.4 mg/Kg weekly dosing was used in 28 patients (6%). In 6% of pts, an additional drug was added to Talq, mostly pomalidomide (n=20) or daratumumab (n=10). 101 pts (21%) received Talq as bridging for CART. Median duration of Talq treatment was 102 days. 62% developed CRS (Grade ≥3, 2%; including one Grade 5 event). 15% developed ICANS (Grades 3-4, 4%). Steroids were used in 213 patients (44%), tocilizumab in 145 (30%) and anakinra in 9 (2%). 22 patients (5%) required ICU stay during the first cycle, for a median of 6 days (range 1-31). A total of 200 pts (41%) had infections (median 1 infection each, range 1-6), including 99 (20%) with bacterial infections, 113 (23%) viral infections and 21 (4%) fungal infections. IVIG was used in 261 pts (54%). 79% of patients (n=384) developed Talq-related oral toxicity, 55% (n=269) skin toxicity and 199 (41%) nail toxicity. 157 patients died after a median follow up of 11.5 months (mos), and the leading causes of death were MM progression (n=120) and infection (n=10). 8 pts (1.7%) developed invasive second primary malignancies. The ORR was 73%, including 46% ≥VGPR. The median PFS (mPFS) for the entire cohort was 7.8 mos (95% CI 6.9-9.4) and the 12 months PFS rate was 39%. The median OS had not been reached and the 12 months OS rate was 63%. For pts who would have been eligible for MonumenTAL-1, the mPFS was 12.2 mos (95% CI 9.7 – not reached (NR); for trial-ineligible pts, the mPFS was 6.9 mos (5.8 – 8.0; p=0.003). For BCMA-naive pts, the mPFS was 15.7 mos (11.8-NR); for BCMA-exposed pts, the mPFS was 6.8 mos (6.0-7.8; p=0.0002). For patients who received Talq as bridging prior to CART, the median PFS was NR (14.1 – NR); for those who did not receive Talq as bridging the mPFS was 6.5 mos (5.3-7.6; p<0.0001). In multivariable analyses, worse OS and PFS were associated with trial ineligibility [(HR 1.38, 95% CI 1.03-1.86; p=0.03) and (2.33, 1.31-4.14; p=0.004)], BMI<30 [(1.42,1.06-1.89; p=0.02) and (1.50, 1.01-2.22; p=0.045)] and true EMD at baseline [(1.95, 1.49-2.56; p<0.001) and (1.65, 1.16-2.36; p=0.005). On the other hand, use of Talq as bridging to CART was associated with better PFS (0.27, 0.17-0.43; p<0.001) and OS (0.13, 0.05-2.56; p<0.001). Ultra high-risk MM (1.40, 1.06-1.84; p=0.02) and baseline albumin <3 (1.68, 1.18-2.37; p=0.004) were associated with worse PFS; ECOG ≥2 (1.68, 1.18-2.40; p=0.004), and baseline Plt<50K (2.14, 1.48-3.10; p<0.0001) was associated with worse OS. Conclusions: In this large real-world analysis of 484 Talq recipients, the ORR was comparable to MonumenTAL-1 but the median PFS was shorter. This likely reflects differences in patient selection, with 70% of patients in our cohort being trial-ineligible. Patients who received Talq as pre-CAR-T bridging had superior survival outcomes. Rates of toxicities were similar to MonumenTAL-1, including >70% of patients developing dysgeusia.

Abstract Introduction: Elranatamab, a BCMA-CD3 bispecific antibody, has shown efficacy in relapsed multiple myeloma (MM), but real-world data on its safety and effectiveness are limited. Methods: We conducted a retrospective, multi-center study of patients (pts) treated with commercial elranatamab from 8/2023–3/2025 at 9 U.S. academic centers. Logistic and Cox regression were used for binary and time-to-event outcomes, respectively. Multivariable analysis (MVA) adjusted for prior lines of therapy (LOTs), baseline Hgb (per g/dL), and LDH (per 100 U/L) as continuous variables; and for BCMA exposure, extramedullary disease (EMD; extraosseous soft tissue lesions), and ECOG ≥2 as binary variables. Results: Of 130 pts, median age was 71 yrs (range 39–95); 57% were female, 82% white, 13% black, and 35% had ECOG ≥2 (11% ≥3). Median LOTs was 6; 91% were triple-class refractory, 49% penta-refractory, 49% BCMA-exposed (median 15.5 mo [IQR 11.8–20.5] since last exposure), and 49% had prior anti-BCMA CAR-T. Most recent BCMA exposures included: CAR-T (84%), TCE (9.4%), and ADC (6.3%). Only 22% met MagnetisMM-3 cohort A eligibility (which excluded BCMA-exposed pts). High-risk cytogenetics (del(17p), t(4;14), t(14;16), or amp(1q)) were present in 44% (25%, 13%, 6.3%, and 7.9%, respectively). EMD was seen in 22%. Median treatment duration was 4.0 mo (95% CI 2.93–6.3) overall and 7.9 mo (95% CI 5.83–NA) with ≥VGPR. Discontinuation due to toxicity or non-relapse mortality occurred in 12% but was significantly lower among those receiving IVIg (5% vs. 17%, p=0.031); IVIg was given in 47%. The overall response rate (ORR) was 65%, including ≥VGPR in 46% and ≥CR in 36%, consistent with MagnetisMM-3 (ORR 61%, ≥CR 35%; Lesokhin et al. Nat Med. 2023). After median follow-up of 7.5 mo, median PFS was 4.3 mo (95% CI 3.4–11.3), OS was 14.6 mo (95% CI 8.6–NA), and DOR was 12.2 mo (95% CI 6.1–NA). In contrast, MagnetisMM-3 (Prince et al., ASH 2024) reported median PFS and OS of 17.2 and 24.6 mo, respectively, with DOR not reached despite 28.4 mo of follow-up. In univariate analysis, ECOG ≥2 was associated with lower ORR (45% vs. 71%; OR 0.34, 95% CI 0.16–0.73, p=0.006), shorter OS (4.3 mo vs. NA; HR 2.41, 95% CI 1.33–4.35, p=0.004), and shorter PFS (1.7 vs. 6.5 mo; HR 1.77, 95% CI 1.09–2.87, p=0.020). Prior BCMA-directed therapy was linked to reduced odds of ≥VGPR (OR 0.38, 95% CI 0.19–0.77, p=0.008) and ≥CR (OR 0.43, 95% CI 0.20–0.89, p=0.025). Higher baseline LDH was linked to shorter OS (HR 1.37, 95% CI 1.23–1.53, p<0.001) and PFS (HR 1.31, 95% CI 1.18–1.44, p<0.001). Higher baseline Hgb was associated with higher ORR (OR 1.46, 95% CI 1.20–1.81, p<0.001) and ≥CR (OR 1.54, 95% CI 1.26–1.92, p<0.001). Among BCMA-exposed pts, recent exposure (<1 [n=15] vs. ≥1 yr [n=32]) was associated with shorter OS (4.1 mo vs. NA; HR 3.66, 95% CI 1.26–10.6, p=0.017). In MVA, higher baseline LDH remained independently associated with shorter OS (aHR 1.36, 95% CI 1.19–1.54, p<0.001) and PFS (aHR 1.27, 95% CI 1.13–1.43, p<0.001); higher Hgb with greater odds of ≥CR (aOR 1.44, 95% CI 1.14–1.85, p=0.003) and longer PFS (aHR 0.84, 95% CI 0.72–0.99, p=0.034); and BCMA exposure with lower odds of ≥CR (aOR 0.32, 95% CI 0.10–0.91, p=0.037). CRS occurred in 40% (grade ≥2 in 12.3%, grade 3 in 2.3%), and ICANS in 17% (grade ≥2 in 7.7%). These rates contrast with MagnetisMM-3 (no grade ≥3 CRS, 3.4% ICANS). ICANS risk was lower in BCMA-exposed pts (9.4% vs. 24%; OR 0.32, 95% CI 0.11–0.85, p=0.029), and decreased with more LOTs (OR 0.73, 95% CI 0.55–0.92, p=0.016) and higher Hgb (OR 0.71, 95% CI 0.52–0.94, p=0.023). Tocilizumab and steroids were used in 36% (treatment 28%, prophylaxis 9.2%) and 23%, respectively. Steroid use was less common in BCMA-exposed pts (11% vs. 34%; OR 0.23, 95% CI 0.08–0.60, p=0.004) and declined with more LOTs (OR 0.68, 95% CI 0.52–0.85, p=0.002). Infections occurred in 40% of the cohort, more common in pts with CrCl <40 mL/min (59% vs. 36%; HR 2.58, 95% CI 1.02–6.80, p=0.049). IVIg was linked to improved infection-free survival (HR 0.59, 95% CI 0.37–0.93, p=0.023). Conclusions: Elranatamab demonstrated meaningful activity in real-world relapsed MM, though PFS and OS were shorter than in MagnetisMM-3, likely reflecting an older, frailer population with higher disease burden. ICANS was more frequent than in MagnetisMM-3. Baseline anemia and elevated LDH, markers of poor marrow reserve and higher disease burden, were strongly linked to worse outcomes.

Abstract Introduction: Patients (pts) with soft tissue plasmacytomas noncontiguous with bone (true extramedullary disease [EMD]) have poor outcomes with standard therapies shown by low overall response rates (ORRs), which are not durable. Talquetamab (Tal; anti-GPRC5D×CD3) and teclistamab (Tec; anti-BCMA×CD3) are first-in-class bispecific antibodies (BsAbs) approved as monotherapies for triple-class exposed (TCE) relapsed/refractory multiple myeloma (RRMM). Primary analysis from the RedirecTT-1 (NCT04586426) dedicated phase 2 EMD cohort (March 2025 data cut; median follow-up [mFU] 12.6 mo) showed Tal + Tec elicited an ORR of 78.9% (assessed by independent review committee) and a 12-mo progression-free survival (PFS) of 61.0%. Here, we report updated efficacy, using investigator assessment, and safety results from the phase 2 RedirecTT-1 EMD cohort with Tal + Tec. Importantly, we report EMD location and a novel analysis of tumor burden as a prognostic indicator of ORR. Methods: Pts had TCE RRMM and true EMD defined as ≥1 nonradiated soft tissue plasmacytoma noncontiguous with bone ≥2 cm in 1 dimension (with or without paramedullary plasmacytomas). Nonsecretory/oligosecretory disease was permitted. Prior CAR-T and non-BCMA/-GPRC5D BsAb therapies were permitted. Pts received Tal 0.8 mg/kg + Tec 3.0 mg/kg Q2W, preceded by step-up doses, with a permitted switch to monthly dosing at investigator's discretion after cycle 6 or after cycle 4 with confirmed ≥VGPR. Response was assessed per IMWG; EMD response was assessed by PET-CT or MRI whole-body scans. Tumor burden was assessed by total EMD tumor volume. Results: As of July 2025, 90 pts received Tal + Tec, with a median follow-up of 16.3 mo (range 0.5–23.7). Baseline characteristics were as previously reported; 39% had nonsecretory/oligosecretory disease, 20% had prior anti-BCMA CAR-T therapy, and 9% had prior BsAb therapy (all anti-FcRH5). The median number of plasmacytomas was 2 (range 1–7); 17 (18.9%), 29 (32.2%), and 74 (82.2%) pts, respectively, had ≥1 nodal, organ, or soft tissue EMD, while 11 (12.2%) also had ≥1 additional paramedullary plasmacytoma. Of 268 EMDs across all pts, 95 (35.4%) were nodal EMD, 77 (28.7%) were organ EMD (most commonly liver), and 64 (23.9%) were soft tissue EMD; 32 (11.9%) were paramedullary. Baseline EMD tumor volume was <25 cm2in 43 (47.8%) pts, 25–50 cm2 in 21 (23.3%) pts, and >50 cm2 in 26 (28.9%) pts. Per investigator assessment, ORR (95% CI) was 77.8% (67.8–85.9); 50.0% had a ≥CR. Median duration of response (DOR) was not estimable (NE; 12-mo DOR rate, 60.1%), 12-mo PFS was 55.6%, and median overall survival (OS) was NE (12-mo OS rate 73.8%). ORR was 90.7% (77.9–97.4; ≥CR 60.5%) in pts with EMD tumor volume <25 cm2, 66.7% (43.0–85.4; ≥CR 52.4%) for 25–50 cm2, and 65.4% (44.3–82.8; ≥CR 30.8%) for >50 cm2. Common adverse events (AEs) included CRS (77.8%; no grade [gr] 3/4) and neutropenia (72.2%; gr 3/4 62.2%). Taste changes (78.9%), non-rash skin AEs (68.9%), and nail AEs (55.6%) were all gr 1/2, and rash AEs (30.0%) were mostly gr 1/2. ICANS occurred in 11 (12.2%) pts (gr 3, 1.1%; gr 4, 1.1%). Infections occurred in 72 (80.0%) pts (gr 3/4 40.0%), most commonly upper respiratory tract infection (26.7%; gr 3/4 5.6%); IVIG was highly recommended to prevent and manage hypogammaglobulinemia and infection. Ten (11.1%) pts discontinued Tal + Tec due to treatment-emergent AEs (4 pts due to infections and 6 due to gr 5 AEs). Two pts discontinued Tal only due to AEs. In total, 11 (12.2%) pts had gr 5 AEs (6 due to infections), 6 of which were deemed to be drug related by investigators. Data will be updated with an additional 4 mo of follow-up for the presentation. Conclusions: With longer follow-up in pts with TCE RRMM with true EMD regardless of baseline tumor characteristics, Tal + Tec efficacy exceeded all approved therapies, including T-cell redirecting and cellular therapies, noting limitations of cross-study comparisons. Lower total EMD tumor volume was associated with a higher ORR but low pt numbers in each group and lack of statistical testing limits robust interpretation. The safety profile of Tal + Tec was generally consistent with each monotherapy; AEs were not exacerbated with the combination. These data continue to highlight the clinical benefit of the novel combination of Tal + Tec in pts with true EMD, a population with high disease burden and significant unmet need.