Rosenthal, Allison

e13622 Background: Chimeric antigen receptor T-cell (CAR-T) therapy has revolutionized treatment of hematologic malignancies; however, severe toxicities like cytokine release syndrome (CRS) remain a limitation to broader utilization. Using a large dataset of patients treated with CAR-T, we developed a machine learning (ML) model leveraging multi-model patient characteristics to predict the risk of CRS onset with clinically meaningful lead times. Methods: A prospectively maintained dataset of CAR-T patients treated at Mayo Clinic was used. Baseline disease/patients characteristics, labs, ECG, echocardiogram, and vital signs (including data from remote patient monitoring program) were collected. A 24-hour rolling window was used to generate daily averages, maximum, minimum, and deltas for each parameter. ECG and echocardiogram data were forward-imputed; vital signs and labs were not imputed. Each day-post-infusion served as a potential event (CRS onset vs. no CRS). A gradient boosting classifier was trained and tested on a subset of patients (n = 351) using 1,000 bootstrap iterations, evaluating 24-, 48-, and 72-hour lead times. Shapley-based feature importance guided simplification to an 11-feature model, externally validated on an independent cohort (n = 51). Models’ performance assessed using the area under the receiver operating characteristic curve (AUC). Results: A total of 402 patients treated with CAR-T were examined. Median age was 64 years and 36% were female. Primary diagnosis was B-cell lymphoma (n = 240) or multiple myeloma (n = 162). A total of 226,332 vital sign measurements and 372,871 lab values, as well as ECG and echocardiogram data, were collected for 30 days post-infusion. Overall, 79.8% of patients experienced CRS within 30 days. The full ML model demonstrated AUCs of 0.90 (95% CI 0.86–0.93), 0.83 (95% CI 0.77–0.88), and 0.82 (95% CI 0.77–0.87) for 24-, 48-, and 72-hour predictions of CRS, respectively. Including “day of CRS onset post CAR-T” as a feature improved the 72-hour AUC to 0.87 (95% CI 0.83–0.90). A model restricted to vital signs alone showed AUCs of 0.88, 0.75, and 0.71 for 24-, 48-, and 72-hour horizons. An 11-feature model—focusing on temperature, blood pressure, fibrinogen, CRP, serum sodium level, QRS interval, and lymphocyte counts—maintained high accuracy at 24 hours (AUC 0.90, 95% CI 0.86–0.93) and was validated externally using an independent dataset with an AUC of 0.92. Conclusions: Our ML model provides a reliable approach for predicting if/when CRS will occur, with up to three days of lead time before clinical manifestation, offering an opportunity for proactive management. The robust performance of a parsimonious 11-feature model supports practical adoption across clinical settings and real-time risk stratification to enhance patient safety. Prospective validation and integration into clinical workflows will be critical to translating these findings into practice.

e13575 Background: Cellular therapies continue to expand in diversity, complexity, and financial impact leading clinicians and administrators to seek innovative care options. The paradigm of late has been to move to an outpatient treatment model. However, centers may not find this model to be the safest option for their patients. This pilot project aims to use established outpatient resources to reduce hospital length of stay (LOS) for patients receiving ciltacabtagene autoleucel in an inpatient setting. Methods: Chart review was conducted for 42 patients treated for incidence of CRS, days to onset, LOS, readmissions, and hospital expenses and charges. The cohort showed an 83% incidence of CRS with a mean onset of 7 days. 71% of CRS was limited to Grade 1. The mean LOS was 10.8 days and the readmission rate was 21%. Financial modeling estimated overall cost reduction to range from $13,000 to $20,000 per patient for a LOS of 3 days. A pilot project to reduce LOS was developed with inclusion/exclusion criteria prioritizing patient safety and likelihood of remaining in the outpatient setting. A workflow for administering one dose of tocilizumab for Grade 1 CRS in the outpatient setting was also established to counterbalance readmissions. The goal of the pilot is to reduce the overall LOS from 10 days to 3 days while maintaining readmission rates and comparable charge to reimbursement ratios on net operating income. Results: In the first five months, 25 patients were screened for eligibility. 60% were deemed eligible and 66% of those were discharged after three days. No patients discharged after a minimum admission were readmitted within three days. Current data suggests the readmission rate within 30 days is higher than baseline at 40%. The overall LOS of patients enrolled in the pilot upon infusion was six days. Patients successfully remaining on the pilot have a 60% reduction in LOS. There have been no major safety events. Conclusions: We identified that select patients could be managed in the outpatient setting after a minimum three-midnight hospital admission. A pilot project was developed using our comprehensive care model coupled with a minimum LOS of 3 days. After discharge, patients are followed daily in our outpatient clinic using in-person visits and remote patient monitoring. To counterbalance readmissions, Grade 1 CRS has been treated in the outpatient clinic including administration of tocilizumab. Early data indicates that the workflow is reducing LOS by 60% under patient selection criteria capturing approximately 60% of patients. While readmissions in the first three days post discharge have been parried, there is opportunity to further reduce readmissions in the first 30 days following CAR-T infusion. We anticipate the pilot will reduce costs without negatively impacting reimbursement or readmissions.

Introduction: Outcomes from multidisciplinary lymphoma tumor boards in the molecular era are limited, and the effects of transitioning to virtual formats during the COVID-19 pandemic are not well-documented. This study evaluates how the transition of the Mayo Clinic Lymphoma Tumor Board to a virtual format has impacted its effectiveness and operations. The board, part of the International Mayo Clinic Care Network (MCCN), involves case presentations, subspecialist reviews, and collaborative recommendations. Methods: The Mayo Clinic Lymphoma Tumor Board, an integral part of the Mayo Clinic Care Network (MCCN), conducted a comprehensive review of lymphoma cases from 2014 to 2024. The board's review process involves a structured approach including the presentation of clinical case details, radiology and hematopathology findings, proposed treatment options, relevant literature review, and discussions. From 2014 to 2024, the Mayo Clinic Lymphoma Tumor Board reviewed a total of 722 patients. The review process involves a multidisciplinary team, including physicians, pharmacists, clinical research associates, and trainees. Recommendations and changes related to radiology interpretation, pathologic diagnosis, and treatment approaches were documented and tracked prospectively. In March 2020, the board transitioned to a virtual format due to the COVID-19 pandemic. This shift allowed for remote participation and continued collaborative review while maintaining the board's structured approach and documentation practices. Results: In a total of 722 patients, changes in care were identified in 423 patients, including 287 with newly diagnosed lymphoma and 136 with relapse or refractory disease. Among those who experienced changes, 94 had diffuse large B-cell lymphoma, 49 had follicular lymphoma, 39 had marginal zone lymphoma, 36 had classical Hodgkin lymphoma, and 29 had mantle cell lymphoma. Prior to the COVID-19 pandemic, 431 patients were analyzed, with 217 (50.3%) experiencing changes. A total of 268 changes were documented among these patients. Changes in radiologic interpretation occurred in 9 patients (2.1% of 431 total patients). Pathologic diagnosis changes were noted in 42 patients (9.7%). Additional testing was recommended for 73 patients (16.9%), and clinical management was altered for 144 patients (33.4%). After transitioning to a virtual format, 291 patients were reviewed post-COVID, with 206 (70.8%) having recommended changes. The total number of change recommendations was 322, reflecting increases across all components compared to previous data, with adjustments for varying numbers of patients reviewed. Specifically, radiologic interpretation changes occurred in 21 patients (7.2%), pathologic diagnosis changes in 57 patients (19.6%), additional testing recommendations for 85 patients (29.2%), and clinical management alterations in 159 patients (54.7%). Conclusion: Transitioning to a virtual format in the Mayo Clinic Lymphoma Tumor Board's continued to demonstrate the effectiveness, with increased changes in diagnostic interpretations and clinical management. These results illustrate the potential of virtual tumor boards to improve collaborative decision-making and adapt to evolving clinical needs.

Introduction: Patients with non-Hodgkin Lymphoma (NHL), especially adolescents and young adults, can face late effects from treatment, including decreased fertility. Due to time constraints prior to treatment initiation and sometimes financial limitations, implementing fertility counseling and referral-based guideline care remains challenging. We aim to understand the occurrence and predictors of fertility counseling and preservation in patients with NHL in the Lymphoma Epidemiology of Outcomes (LEO) cohort study. Methods: The LEO cohort prospectively enrolled newly diagnosed patients with NHL from 2015-2020 at 8 academic medical centers in the US. Clinical. Treatment data were abstracted from medical records, pathology was reviewed and classified by a LEO pathologist, and participants completed a self-administered questionnaire. Participants were then prospectively followed to identify outcomes and collect additional patient reported data, including a survivorship questionnaire at follow-up 3 years (FU3) after diagnosis that included items on fertility counseling and preservation. The questionnaire contained two fertility questions analyzed: whether healthcare providers discussed the impact of lymphoma treatment on fertility, and whether providers discussed the process of fertility preservation. This analysis was based on participants aged 18-50 years at enrollment and who completed the fertility questions at FU3. Women who had undergone natural or procedural menopause and men who had undergone vasectomy pre-treatment were excluded. We estimated the association of demographic and clinical factors with fertility preservation using odds ratios (ORs) and 95% confidence intervals (CI) from univariate and multivariable logistic regression models. Results: Seventy-seven patients were included in the analysis: 36 females (46%) and 41 males (53%). The median age of diagnosis was 40 years (IQR 30-45), and 36 (47%) participants were in the age 18-39-year category. Participant self-identified race/ethnicity was 69% non-Hispanic White, 22% Hispanic, 4% Black and 3% Asian. Sixty (78%) patients were treated with alkylators, 12 (15.6%) with non-alkylators, and 5 (6.5%) with radiation. Fifty-five (71%) of the patients were classified as having aggressive lymphoma, with follicular, extranodal marginal zone, primary cutaneous follicle center, and nodal marginal zone B-cell lymphoma defined as indolent and remaining subtypes aggressive. Around 6% of female patients underwent fertility preservation after lymphoma diagnosis, compared to 12.2% of male patients. All the patients who had undergone fertility preservation after lymphoma diagnosis were in the 18-39-year age group. On univariate analysis, females were more likely to have undergone fertility toxicity and preservation discussions compared to males, with 72% of females and 59% of males recalling a discussion regarding fertility (OR 1.8, 95% CI 0.7-4.9, p=0.21). Those age 18-39 were significantly more likely to have undergone fertility discussion compared to those aged 40-50 (OR 10.2, 95% CI 3.3-39.2, p<0.001). Patients treated with alkylating therapy were significantly more likely to undergo fertility discussion when compared to those on non-alkylating therapy (OR 9.0, 95% CI 2.3-45.5, p<0.01). In addition, 75% of those with aggressive lymphoma compared to 41% of patients with indolent lymphoma had fertility discussion (OR 4.2, 95% CI 1.5-12.4, p=0.01). White patients were significantly less likely than Hispanic, Black and Asian patients to have fertility discussion (OR 0.2, 95% CI 0.04-0.6, p<0.01). In the multivariable analysis, alkylating therapy (OR 8.9, 95% CI 1.6-66.7) and age 18-39 years (OR 10.9, 95% CI 2.9-55.3) remained significant predictors of fertility discussion; White race was also trending towards significance (OR 0.2, 95% CI 0.04-1.0, p=0.08). Aggressive lymphoma subtype was not a significant predictor of fertility discussion. Conclusions: In this cohort of lymphoma survivors 3 years post-diagnosis, we found that Hispanic patients, those diagnosed at a younger age and exposed to alkylating chemotherapy agents were significantly more likely to undergo discussions regarding the effects of lymphoma treatment on fertility. Our study highlights the importance of all cancer patients of childbearing age having the opportunity to address fertility concerns during clinic visits.

Background: Chimeric antigen receptor T-cell (CAR-T) therapy has shown substantial promise in treatment of hematological malignancies, including non-Hodgkin Lymphoma (NHL). A subset of patients develop prolonged cytopenia, comprising quality of life, increasing healthcare utilization, and imparting considerable morbidity and mortality. Therapy-related myeloid neoplasms (t-MN), typically seen in patients who receive DNA-damaging agents such as radiotherapy and alkylators, have also been observed in CAR-T recipients. CAR-Hematotox score predicts the development of hematotoxicity in CAR-T recipients using laboratory parameters obtained at lymphodepleting (LD) chemotherapy, a timepoint when the patient has already gone through leukapheresis thus committed to the CAR-T process. Therefore, it is critical to identify patients at the highest risk of prolonged cytopenia and/or t-MN prior to apheresis, making it possible for a better triage of those who will proceed with CAR-T and those who might benefit from other strategies. Methods: A retrospective analysis was conducted in NHL patients who received commercial CAR-T products between January 2018 and March 2024 at Mayo Clinic Rochester, Arizona and Florida. Clinical and laboratory data including age, hemoglobin (Hb), absolute neutrophil count (ANC), platelet (PLT), ferritin, and LDH, obtained at the initial CAR-T evaluation, typically within 1 week of leukapheresis, were used for analysis. Prior lines of therapies including autologous stem cell transplant (ASCT) were collected. Prolonged cytopenias were defined as Hb < 8 g/dL, ANC < 0.5 x 109/L, and/or PLT < 50 x 109/L at 3 months after CAR-T. t-MN was diagnosed per WHO 2016. For prolonged cytopenia, the optimal cutoff points for continuous variables were determined by maximizing the sum of sensitivity and specificity. Logistic regression was then used to determine the factors associated with cytopenia at 3 months post-CAR-T. For t-MN, maximally selected log-rank statistics was used to determine the optimal cutoffs. We used competing risk analysis to determine the factors associated with development of t-MN. Death without t-MN was considered a competing risk. For both analyses, the factors with P < 0.1 in univariate analysis (UVA) were included in multivariate analysis (MVA). Results: Among the 387 NHL patients, with a median follow-up of 11 months (95% CI 8.7 - 12.2 months), 56 (14.4%) developed (41, 10.5% with 1 cytopenia; 14, 3.6% with 2 and 1, 0.2% with cytopenia in all 3 lineages), 20 (5.2%) developed and 315 (81.2%) did not have either. Factors associated with prolonged cytopenia: were age >, PLT ≤ 150, ferritin ≥ 600, and LDH ≥ 250. Univariate analysis identified baseline Hb, ANC, PLT and ferritin to be significantly associated with prolonged cytopenia. MVA identified Hb ≤ 11 (HR = 1.98; 95% CI 1.03-3.85, P = 0.042) and PLT ≤ (HR = 2.56; 95% CI 1.39-4.78; P = 0.003) to be associated with the development of prolonged cytopenia. Factors associated with t-MN:were age > 65 years, Hb ≤ 9 g/dl, ANC < 3.8, PLT ≤ 80, Univariate analysis identified age, Hb, ANC, PLT, ferritin, and ≥ 3 prior lines of treatment to be significantly associated with t-MN. On MVA, age ≥ 65 (HR = 3.91; 95% CI 1.52-10.03; P = 0.004), Hb ≤ 9 (HR = 3.49; 95% CI 1.38-8.85; P = 0.008), ferritin ≥ 300 (HR = 3.06 ; 95% CI 1.08-8.67; P = 0.035) and ≥ 3 prior lines of therapy (HR = 3.25; 95% CI 1.33-7.94 ; P = 0.010) were associated with t-MN development. Association of prolonged cytopenia with t-MN development: Prolonged cytopenia was noted in 45% of patients who developed t-MN and 15% of those who did not (P = 0.002). Cumulative incidence of t-MN at 1 year was 3.3% and the presence of prolonged cytopenia was associated with a higher 1of t-MN (HR 8.93, 95% CI 2.15 - 37.1, P = 0.003). Conclusions: Readily accessible clinical and lab data such as blood count and inflammatory markers at the time of CAR-T evaluation were associated with the development of prolonged cytopenias. Similarly, advanced age, anemia, high ferritin levels and history of ≥3 lines of therapy were associated with the development of t-MN. These findings, when validated in larger dataset, can be used for patient counseling and therapy selection by identifying patients at high-risk of developing hematotoxicity and/or t-MN development.