Citations

Abstract Background: Patients (pts) with myeloma (MM) receiving CAR-T therapy have usually been exposed to many prior therapies. Melphalan (HDM) and other chemotherapies can induce cellular senescence. It has recently been shown that MM cells can also exhibit features of therapy-induced senescence (TIS). TIS tumor cells are associated with resistance to therapy and relapse. To investigate the presence of TIS MM cells prior to CAR-T and association with response, we examined the transcriptome and phenotype of MM prior to CAR-T. Method: We used scRNA-seq data from bone marrow (BM) MM cells: Boiarsky, et al. (GSE193531, dbGaP: phs001323.v3.p1) newly diagnosed MM (NDMM) pts; Dhodapkar, et al. (GSE210079), pre- and post CAR-T; pts who received CAR-T at Mayo Clinic, pre- (N=10) and post- (N=2). To evaluate TIS, single-cell gene set enrichment analysis (scGSEA) for several gene sets (PMID: 40164720): SenUp (Senescence Upregulated), SenGA (Senescence Growth Arrest), and SCAPs (Senescent Cell Anti-apoptosis Pathways); and a Plasma Cell Senescence (PCSen) gene set (GSE5900, GSE47552). To assess TIS and myeloid phenotype, BM MM cells (CD138+) were stained for CD14, CD16, KLRG1 and GPNMB and analyzed by CytoFlex and Kaluza. Results: In the integrated dataset, scGSEA showed significantly higher enrichment of PCSen, SenUp, SenGA, and SCAPs in late-stage MM cells (both pre- and post-CAR-T) compared to NDMM and healthy donors (p<0.05, Kruskal-Wallis with Dunn’s posttest). Conversely, the Normal gene set was significantly enriched in healthy plasma cells relative to MM (p<0.05). Expression of CDKN2A, which encodes the senescence marker p16, was also increased in late-stage MM relative to NDMM or healthy plasma cells. Notably, in our newly generated scRNA-seq samples collected within 6 months before CAR-T, a significant inverse correlation between senescence marker expression and progression free survival (PFS) post-CAR-T was seen, suggesting that MM cells exhibiting TIS may be more resistant to T cell mediated cytotoxicity. We found that 62.5% (25/40) of pts had >1% MM cells that were CD14+. Among them, the median CD14+ MM cells were 5.5% (1-71%); and median CD14+CD16+ MM cells were 2.14% (0.4-67%). Among 7 pts with available samples for KLRG1 and GPNMB assessment, 100% (7/7) had KLRG+ MM cells (median: 59% (24-99%)) and 57% (4/7) had GPNMB+ MM cells (median: 15% (3.9-30%)), and KLRG1+GPNMB+ MM cells (median: 9.6% (3.8-17%)). Conclusion: Our study shows that late-line MM cells exhibit transcriptional and phenotype profiles consistent with TIS. This senescent phenotype may contribute to resistance to immunotherapies, including CAR-T. Additional phenotype data is collected and will be presented. A phase II clinical trial is ongoing (NCT06940297) to assess for the combined treatment with senolytic drugs dasatinib and quercetin peri-CART to deepen clinical response. Citation Format: Sofia Arbelaez, Andre De Menezes Silva Corraes, Gabriel Alvares Borges, Chen Wu, Christoph Schaefers, Malvika Gupta, Zuoyi Shao, Kevin Regan, Rayaan Kamal, Angelo Jose Guilatco, Neal I. Sannuli, Ying Li, Taxiarchis Kourelis, Wilson Gonsalves, Moritz Binder, Melinda Tan, Nadine Abdallah, Megan Weivoda, Yi Lin. Myeloma cells with therapy-induced senescence-like phenotype are associated with resistance to CAR-T therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6467.

Abstract Background: Therapy-induced senescence (TIS), a tumor cell state associated with distinct gene expression patterns and implicated in therapy resistance, may be activated in Multiple Myeloma (MM) patients (pt) by genotoxic chemotherapies, such as high-dose melphalan (HDM). We investigated whether TIS is activated in MM cells following chemotherapy and linked to resistance to T-cell directed therapies. Method: We integrated scRNA-seq data from healthy donors (CNTRL) and MM pts (newly-diagnosed (NDMM), pre/post-CAR-T) bone marrow plasma cells (PCs) using two public datasets (Boiarsky, et al.; Dhodapkar, et al.) and a newly generated pt dataset. Senescence was evaluated using scGSEA with published gene sets (SenUp, SenGA, and SCAPs). For in vitro MM cytotoxicity assay, human H929 MM cells were treated with vehicle (Veh) or HDM (3µM, IC90) for 6h and examined for TIS features. H929VEH or H929HDM were co-cultured with CNTRL T cells with or without teclistamab (Tec, 10 nM) or with day 14 PBMCs from MM pts post-CAR-T (E:T 5:1). CAR+ T cells were high among the day 14 PBMC (median±SD: 93.4±7.2% of CD3+ cells, n=6). Results: scGSEA revealed significantly higher enrichment of SenUp, SenGA, and SCAPs in late-stage MM cells compared to NDMM and CNTRL (p<0.05). Expression of CDKN2A, encoding the senescence marker p16, was also increased in late-stage MM relative to NDMM/CNTRL PCs. Of interest, cells enriched for senescence also exhibited enrichment for myeloid markers (TET2, CEBP family). Notably, in pre-CAR-T, increased baseline senescence or myeloid marker expression showed an inverse association with CAR-T progression-free survival (PFS), suggesting TIS confers resistance to T cell mediated cytotoxicity. This may be due to myeloid plasticity. TIS MM cells generated in vitro, H929HDM, exhibited a distinct morphology characterized by increased cell size and enhanced cytoplasmic granularity. Increased CD14 expression was seen in 4 of the 6 experiments (median, range % increase: 11.2%, 2.5%-21%), consistent with myeloid plasticity; Increased KLRG1 expression was seen in 5 of the 7 experiments (median MFI ratio, range: 3.7, 1.3-471). No significant change was seen in GPNMB expression. Compared to H929VEH, H929HDM had a reduction in killing by CAR-T cells (median±SD: 22.8±32.94% n=6, p=0.016, Wilcoxon paired test). Similarly, H929HDM had a 10.69% decrease in Tec-mediated T-cell killing compared to H929VEH (n=8, p=0.02, Wilcoxon paired test). Conclusions: Late-stage MM cells exhibit transcriptional signatures consistent with TIS, which can be associated with decreased PFS to CAR-T. We report an in vitro model of TIS MM cells for testing of resistance to T cell killing. Based on these findings, a phase II trial (NCT06940297) is underway to test the hypothesis that peri-CAR-T senolytic therapy with dasatinib and quercetin will deepen responses by targeting this resistant TIS MM population. Citation Format: Christoph Schaefers, Angelo Jose Guilatco, Sofia Arbelaez, Gabriel Alvares Borges, Chen Wu, Andre de Menezes Silva Corraes, Malvika Gupta, Zuoyi Shao, Kevin Regan, Rayaan Kamal, Neal I. Sannuli, Ying Li, Taxiarchis Kourelis, Wilson Gonsalves, Melinda Tan, Nadine Abdallah, Yi Lin, Megan Weivoda. Myeloma cells with therapy-induced senescence-like phenotype have increased resistance to killing by T cell directed therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6475.

Abstract The CoMMpass study (NCT01454297) is a prospective, longitudinal observational study involving 1,141 newly diagnosed multiple myeloma (NDMM) patients. Bone marrow aspirates from the study participants were collected for comprehensive molecular characterization of tumor (using bulk RNA sequencing and whole genome sequencing) and tumor immune microenvironment using 3’ single-cell RNA sequencing (scRNA-seq). To support exploration of this rich multi-omic dataset, we developed CoMMpass Explorer (CE), an intuitive interactive platform that enables real-time analysis of clinical and genomic data from CoMMpass. A central feature of CE is cohort building, where users can filter and stratify patients by clinical, genomic, and survival data elements to create custom cohorts for analysis. CE provides five main views: Overall Summary for clinical feature distribution, Kaplan-Meier curves, and multivariate Cox proportional hazards models; Mutational Profile for visualizing and comparing somatic mutations; Tumor Profile for bulk differential expression and gene set enrichment; Immune Microenvironment for comparing cell type abundance and cell cycle dynamics between cohorts using scRNA-seq data; and Pseudo-Bulk, which aggregates single-cell expression by cell type to enable cohort-level comparisons of transcriptional programs using DEseq2. CE reproduces prior CoMMpass studies, including WEE1 expression being associated with poorer progression-free survival (Simhal, et al.) and different mutation frequencies between African American and European American patients (Manojlovic, et al.). We further demonstrate CE’s utility by interrogating the 2025 Consensus Genomic Staging (CGS) risk classification system (Avet-Loiseau, et al). We compared CGS high-risk (HR, n=249) vs. CGS standard-risk (SR, n=573) patients. HR patients have significantly worse progression-free survival, overall survival, and time to second-line therapy compared with SR patients. Univariate models recapitulated expected differences in the clinical features that define CGS risk. Bulk tumor RNA-seq and ssGSEA highlighted a more proliferative and genomically unstable transcriptional program in HR group, with upregulation of angiogenesis and cell-cycle-related pathways. In the immune microenvironment, monocytes were more abundant in the SR group. Pseudo-bulk RNA-seq of plasma cells showed overexpression of ANXA1, NSD2, IGF1R, MAF and related genes in HR. CE democratizes access to the CoMMpass multi-omic resource, allowing researchers to interrogate clinical and molecular heterogeneity. By enabling hypothesis generation and replication of published results, CE serves as a valuable tool for identifying prognostic biomarkers and meaningful patterns in gene expression, mutation profiles, and the immune microenvironment. Citation Format: Wanxing Zhang, Chaitanya R. Acharya, Steven M. Foltz, David E. Avigan, Samir Parekh, Ravi Vij, Shaji Kunnathu Kumar, Taxiarchis Kourelis, Sagar Lonial, Hearn Cho, Ioannis S. Vlachos, Sacha Gnjatic, Li Ding, Manoj Bhasin, George Mulligan. CoMMpass Explorer: An interactive platform to explore clinical and genomic data from newly diagnosed multiple myeloma patients from the landmark CoMMpass observational study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5504.

Abstract Frailty predicts adverse outcomes in multiple myeloma (MM), yet an objective frailty biomarker is lacking. Circulating senescence-associated secretory phenotype (SASP) factors correlate with frailty and adverse outcomes in chronic diseases, but their relevance in MM is unclear. In this study, we investigated the relationship between 36 plasma SASP factors, frailty, and clinical outcomes in: (1) a historical cohort using stored specimens from 59 patients with newly diagnosed MM, and (2) a prospective cohort of 36 patients initiating new MM treatment. Several SASP factors were significantly associated with frailty after adjusting for age, sex, and body mass index; GDF-15, TNF-RI, IL-6, and IL-8 showed positive correlations with both cumulative-deficit based and physical frailty indices. Additionally, significant age-adjusted associations were observed between SASP factors and treatment outcomes, including overall survival (OS), high-grade toxicity, and healthcare utilization. We developed a 5-factor SASP model that demonstrated superior predictive performance for OS compared to chronological age, R-ISS stage, and clinical frailty, and TNF-RI alone (5-year OS AUC: 0.90). These findings highlight the association between senescence biomarkers and frailty and OS, and support further investigation of SASP factors as objective frailty biomarkers in MM.

Abstract Measurable residual disease (MRD) assessment has become central to risk-adapted treatment strategies in multiple myeloma, yet the prognostic value of MRD negativity following autologous stem cell transplantation (ASCT) across individual high-risk cytogenetic abnormalities (HRCAs) remains unclear. We analyzed 351 newly diagnosed patients who achieved MRD negativity (sensitivity 2.4×10 − 6 ) at Day 100 following ASCT and had baseline cytogenetic data available. Median follow-up was 50.4 months. Standard-risk cytogenetics were present in 46% (n = 162), while 34% (n = 118) had one HRCA and 20% (n = 71) had ≥ 2 HRCAs. Despite uniform MRD negativity, several HRCAs continued to confer inferior progression-free survival (PFS). Compared with the standard-risk group (4-year PFS 84%), patients with del(17p), 1q21+, t(4;14), or MAF translocations [t(14;16)/t(14;20)] had significantly lower 4-year PFS rates: 56% (HR 2.89; p = 0.0001), 60% (HR 2.46; p = 0.0001), 43% (HR 3.50; p = 0.0001), and 58% (HR 2.51; p = 0.01), respectively. Isolated del(17p) and 1q21 + also conferred inferior PFS. The 4-year PFS was 65% for patients with one HRCA (HR 2.31; p = 0.0004) and 55% for those with ≥ 2 HRCAs (HR 2.75; p = 0.0001). In this large MRD-negative post-ASCT cohort, several HRCAs, including isolated HRCAs, remained clinically meaningful drivers of inferior PFS, highlighting that early MRD negativity does not uniformly mitigate the impact of high-risk biology.

Abstract Castleman disease (CD) is a heterogeneous group of lymphoproliferative disorders anatomically classified by distribution (unicentric CD [UCD], oligocentric CD [oligoCD], or multicentric CD [MCD]). Human herpes virus 8–negative MCD is called idiopathic MCD (iMCD), which includes clinical subtypes with varying phenotypes and responses: TAFRO (thrombocytopenia, anasarca, fever, renal dysfunction and/or reticulin fibrosis, and organomegaly), IPL (idiopathic plasmacytic lymphadenopathy), and not otherwise specified. OligoCD has recently emerged as an intermediate form between UCD and MCD, with unclear clinical behavior, and IPL has not been validated in a Western cohort. We retrospectively analyzed 217 patients with CD evaluated at our institution between January 2004 and August 2024. Survival probabilities were compared using log-rank tests. Overall, 57% had UCD, 20% had oligoCD, and 23% had iMCD. Patients with oligoCD and iMCD more frequently exhibited systemic symptoms than those with UCD. Patients with oligoCD and iMCD had significantly shorter event-free survival (EFS) of 8.9 and 2.3 years, respectively, than those with UCD (not reached; P< .001 and P< .02). Among iMCD subtypes, iMCD-IPL demonstrated a longer EFS than iMCD-TAFRO (P = .02), with no deaths during the follow-up periods. The results validate oligoCD and iMCD-IPL as new subtypes in this Western cohort. The survival outcome in oligoCD was intermediate between UCD and iMCD, with only a subset of patients with oligoCD requiring systemic therapy. iMCD-IPL had a favorable survival outcome in this US cohort, similar to what has been reported in non-Western countries. Tailoring treatment strategies to disease subtypes and vigilant monitoring of oligoCD for progression may improve survival outcomes in CD.

Abstract Multiple myeloma is a malignancy of clonally expanded plasma cells shaped by complex interactions with the immune microenvironment (IME). To investigate immune correlates of treatment response and disease progression, we conducted multi-omics profiling including CD138neg single-cell RNA sequencing of 243 bone marrow samples from 102 patients (631,226 cells) and CD138pos bulk RNA and whole-genome sequencing from 209 samples. In longitudinal analyses, interferon-γ signaling associated with markers of impaired T-cell memory after autologous stem cell transplant, whereas naïve B-cell abundance and immunoglobulin diversity correlated with improved progression-free survival (HR = 0.48; P = 2.3e−4). At disease progression, multiple myeloma cells upregulated cancer–testis antigens (CTAg) and immune effector genes, with concurrent B-cell depletion, enrichment of myeloid-derived suppressor cell expression, and phenotypic T-cell exhaustion. These findings highlight dynamic immune–tumor interactions, identifying naïve B-cell reconstitution as a biomarker of durable response and CTAgs as potential targets for high-risk disease at progression. Significance: Longitudinal profiling of multiple myeloma and the IME revealed dynamic immune–tumor interactions across the disease course. Altered expression in CD8+ T cells limited memory phenotype after transplant, whereas naïve B-cell recovery associated with sustained treatment response. At progression, CTAg expression associated with immunosuppression, revealing novel mechanisms of immune dysregulation.

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: 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.