Citations

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: Primary central nervous system T-cell lymphoma (PCNSTL) is a rare form of PCNSL, comprising 2–8% of cases. Prognostic factors remain undefined, and no standard treatment protocol currently exists. Prior studies have been limited by small sample sizes, incomplete molecular profiling, and poor representation of novel treatment modalities. In this study, we aimed to identify clinicopathologic factors associated with patient (pt) outcomes and delineate the spectrum of treatment regimens used to inform future clinical practice. Methods A multicenter retrospective study was conducted across 15 North American academic institutions. Pts aged ≥18 years with biopsy-confirmed PCNSTL between 2009 and 2025 with no evidence of systemic disease were included. Cases with CNS relapse of systemic T-cell lymphoma were excluded. Baseline characteristics were analyzed using independent t-tests, Fisher's exact tests, or Pearson's chi-square tests. Primary endpoints were overall survival (OS) and progression-free survival (PFS), evaluated by Kaplan–Meier, log-rank tests, and Cox regression. Results: Forty-two pts met inclusion criteria. Median age was 57 (range 19–77), 69% were male, and 79% were white. Most (69%) had ECOG 0–1. No patients had HIV, hepatitis B or C, or a history of solid organ transplant. LDH was elevated at diagnosis in 38%, and 24% had B-symptoms. Mean absolute lymphocyte count at diagnosis was 1.82/µL. The most common histologic subtypes were PTCL-NOS (76%), followed by ALK-negative ALCL (12%) and γδ T-cell lymphoma (5%). All patients had parenchymal brain lesions with 57% having multifocal disease. Additionally, CSF involvement was seen in 21%, vitreoretinal involvement in 3%, and deep brain involvement in 34%. Next-generation sequencing was performed in seven pts (17%), with disease-associated variants identified in three cases. Among these, two pts had mutations in TP53, and one had a mutation in DNMT3A. Cytogenetics were obtained in 7 pts with only one exhibiting a complex karyotype. Only one pt was EBV positive by EBER-ISH. Overall, 79% received HD-MTX-based therapy. First-line treatments were mainly HD-MTX alone (38%), HD-MTX + temozolomide (17%), or vincristine + procarbazine (12%). No patients were treated on clinical trials. Median treatment duration was 78 days and median number of treatment cycles was 4 (range 1–9 cycles). One patient received radiation alone; 5% received radiation with systemic therapy. A steroid pre-phase was used in 60%, with 44% exhibiting a clinical response. Overall response rate to first-line treatment was 63% (13 CR, 12 PR). Imaging alone was used in determining response in all patients. Repeated biopsy was not performed. Twelve patients (29%) underwent consolidative autologous transplant, most commonly with thiotepa/carmustine (TT/BCNU, 69%) or BEAM conditioning (25%). Median duration of follow-up was 10.4 months. Sixty-two percent of pts progressed during the study period, with 7% experiencing extra-CNS relapse. Median PFS was 8.2 months. Improved PFS was associated with >3 cycles of therapy (p=0.03) and transplant with TT/BCNU (p=0.0004). At last follow-up, 43% were in remission and 52% were deceased, mostly due to disease progression (77%). Median OS was 17.6 months. One- and two-year OS rates were 58% and 42%, respectively. OS was associated with Nottingham/Barcelona score <2 (p=0.02), >3 treatment cycles (p=0.0008), response to first-line therapy (p=0.0007), and TT/BCNU-based transplant (p=0.0001). IELSG, Taipei, and MSKCC risk scores were not significantly associated with survival, though IELSG approached significance (p=0.09). Discussion: This study is among the largest cohorts of PCNSTL pts compiled to date. PCNSTL remains rare and heterogenous. Treatment across centers largely mirrors management of B-cell PCNSL with high utilization of HD-MTX regimens and TT/BCNU conditioning autologous transplant. Survival outcomes within our cohort are consistent with those previously reported within the literature. Receipt of >3 cycles of therapy, TT/BCNU conditioning and Nottingham/Barcelona scores were associated with longer survival within our cohort. Limitations of our study include its retrospective design and heterogeneity within pt disease characteristics, treatments utilized and disease assessment intervals. Further exploration into pathophysiology and optimal management of this unique entity is needed.

Abstract Background Chimeric-antigen-receptor (CAR) T-cell therapy has transformed outcomes for relapsed or refractory hematologic malignancies, yet early identification of patients at highest risk for death remains elusive. A prognostic model that leverages only the information routinely available at the time of CAR-T evaluationcould guide pre-infusion counseling, intensity of follow-up, and resource allocation. Methods We analyzed a prospectively maintained cohort of CAR-T recipients treated at Mayo Clinic to develop and test our model. For each patient, data (vital signs, lab results, cardiac assessments [ECG and echocardiography], and PET/CT images) from initial CAR-T evaluation (up to 60 days prior to CAR-T infusion) was considered. Vital signs and labs were left un-imputed, while ECG, echocardiographic parameters, and PET/CT features were forward-filled from the preceding 60 days. Each patient-day was labeled as “outcome present” or “no outcome,” enabling supervised training of gradient-boosting models to predict overall survival (OS) at 6 months post CAR-T infusion. Model performance was evaluated using the area under the ROC curve (AUC) with 10,000-fold bootstrapping with data shuffling between each iteration to derive confidence intervals for stability. To enhance interpretability, we also developed a simplified model using the top 19 features (determined by Shapley value-based importance), aiming to preserve predictive accuracy with a smaller feature set. Results A total of 570 CAR-T cell treated patients were included in this study (476 patients in the training cohort and 94 patients in the validation cohort). The median age was 64 years and 61% males. The primary diagnoses were B-cell lymphoma (324 patients) and multiple myeloma (246 patients). A total of 4,851 individual vital-sign readings, 11,781 laboratory measurements were included. PET/CT images were available in 318 patients. The ML model demonstrated high discriminative ability for predicting OS at 6 months post CAR-T infusion. Vital signs alone produced a mean AUC of 0.63; radiomics features alone produced a mean AUC of 0.71, while laboratory tests alone produced a mean AUC of 0.84. The features most contributory to each single-modality model were then combined into a simplified multimodal model with a total of 19 features, which produced a mean AUC of 0.87, with a 95% CI of 0.73 - 0.97. The 19 features included in the simplified model were: LDH, ferritin, uric acid, eosinophils, fibrinogen, D-dimer, heart rate, mean arterial blood pressure, QTc, QT interval, and the radiomics features of sphericity, first-order Kurtosis, first-order mean, first-order energy, first-order total energy, shape maximum2D diameter (sagittal plane), SUV CoV, SUVp90, shape flatness. This multimodal predictive model was then applied to the validation cohort (uninvolved with model training or testing), which produced an AUC of 0.93, with a 95% CI of 0.84 - 0.98. Conclusions Our ML-driven approach successfully predicts overall survival 6 months post CAR-T infusion, highlighting those individuals at most risk of poor outcomes at the time of CAR-T evaluation. The strong performance of the multimodal version of the model suggests that each data domain contributes complementary resolving power, and by including novel data types, such as PET CT radiomics features, the most accurate predictions could be realized. By providing accurate, near-real-time risk stratification for poor outcomes within 6 months of CAR-T infusion, the model could facilitate proactive patient management. Prospective studies of the integration of this ML model into CAR-T treatment workflows will be crucial to confirm its clinical benefits and to facilitate its adoption in practice.

Abstract Background Neurolymphomatosis (NL) is a rare and frequently underrecognized manifestation of lymphoma, characterized by infiltration of peripheral or cranial nerves by malignant lymphocytes. Limited data exist to guide optimal management. We present the largest single-enterprise (included the 3 Mayo Clinic sites) cohort to date, evaluating disease characteristics, treatment patterns, and outcomes in patients (pts) with NL. Methods We conducted a retrospective analysis of pts ≥18 years (yrs) diagnosed with clinical, radiographic, or biopsy-confirmed NL at Mayo Clinic between January 2002 and December 2024. Baseline characteristics were compared using independent t-tests for continuous variables and Fisher's exact or Pearson's chi-square tests for categorical variables. Primary endpoints were overall survival (OS) and progression-free survival (PFS). Objective response rate (ORR) was a secondary endpoint. Kaplan–Meier methods, log-rank tests, and Cox proportional hazards regression were used to assess survival outcomes and associated factors. Results We included 149 pts. Median age at diagnosis was 64 yrs (range 22–83), and 23% were female. Most presented with stage IV disease (89%), and NL was part of the initial lymphoma presentation in 42% (primary NL). Diagnostic confirmation was based on histopathology and imaging in 55%, imaging alone in 26%, histopathology alone in 16%, and clinical features in 4%. Diffuse large B-cell lymphoma was the most common histology (61%), followed by marginal zone lymphoma (7%), and T-cell lymphoma (7%). Clinical presentation included painful neuropathy (65%), concurrent CNS involvement (37%), systemic disease (43%), elevated LDH (46.9%), and ECOG ≥2 (38%). Pattern of involvement included cranial nerves (29%), peripheral nerves (28%), and spinal nerve roots (27%). Median follow-up was 92 months (95% CI, 63–119). Among 137 treatment-evaluable pts, first-line regimens included methotrexate-based (MTX) (62%), bendamustine-based (B) (11%), cytarabine-based (Ara-C) (4%), and rituximab-based (R) (excluding MTX, B, or Ara-C) (15%) therapies. Radiation was used in 18%, and 21% underwent autologous stem cell transplant (ASCT). Among 103 response-evaluable pts, the ORR was 67% (52% complete response [CR], 16% partial response [PR]). ORRs by regimen were 78% (MTX), 92% (B), and 86% (R). On univariate analysis, primary NL (OR 4.3) and frontline ASCT (OR 8.47) were associated with higher odds of response (p<0.05), while ECOG ≥2 (OR 0.24) and localized disease (OR 0.33) were associated with lower odds. On multivariable analysis (MVA), primary NL (OR 5.0 [1.5–12.1], p=0.02) and ECOG >1 (OR 0.2 [0.06–0.79], p=0.018) remained significant predictors of response. Median PFS was 13 months (95% CI, 9–36), with a 2-yr PFS of 43% (95% CI, 35–53). On MVA, concurrent CNS involvement (HR 2.4, p=0.05), systemic disease (HR 2.7, p=0.02), and ECOG >1 (HR 2.4, p=0.02) were associated with inferior PFS, while primary NL was associated with improved PFS (HR 0.3, p=0.002). Patients undergoing frontline ASCT had a median PFS of 108 months (95% CI, 58–not reached) with a 7-yr PFS of 62% (95% CI, 46–84). Among those receiving second line and beyond (2+) therapy, 24 received MTX-based regimens, 10 received CAR-T, and 8 received tyrosine kinase inhibitors. Post-ASCT median PFS was 30 months (95% CI, 15–NR). Among CAR-T recipients, evaluable responses (n=8) included 4 CR, 1 PR, and 3 PD. Median OS was 56 months (95% CI, 29–133), with a 5-yr OS of 50% (95% CI, 41–60). On MVA, ECOG ≥2 (HR 2.5, p=0.02) predicted worse OS, while receipt of ASCT (HR 0.4, p=0.04) and primary NL (HR 0.5, p=0.04) were associated with improved OS. Conclusion This study represents the largest single-enterprise cohort of pts with NL, providing comprehensive insight into its clinical presentation, treatment patterns, and outcomes. Primary NL and preserved PS were independently associated with improved response rates and survival, while CNS and systemic involvement at diagnosis portended poorer PFS. MTX and bendamustine-based regimens achieved high ORR, and frontline ASCT was associated with significantly prolonged remission duration and OS. These findings highlight the prognostic importance of baseline disease burden and functional status, support the early incorporation of high-intensity regimens in eligible patients, and underscore the urgent need for prospective studies to guide evidence-based management in this rare and understudied population.

Abstract Introduction There is currently no standard-of-care for patients (pts) who cannot tolerate full dose anthracycline regimens in frontline large B-cell lymphoma (LBCL). R-miniCHOP is a commonly used regimen that provides curative intent treatment with manageable toxicity. Despite this, lymphoma and treatment-related mortality are the leading causes of death in these pts. Additionally, many pts may not be eligible to receive R-miniCHOP due to comorbidity or frailty. There are limited clinical trial and real-world data focused on this population. Methods Pts with LBCL age 18 years (yrs) and older were identified from the Lymphoma Epidemiology of Outcomes (LEO) cohort, a prospective study of newly diagnosed pts with non-Hodgkin lymphoma enrolled from 2015-2020 at 8 academic medical centers in the US. Pts with LBCL were included in this analysis if they had received R-miniCHOP or a regimen that did not include an anthracycline (nonAC). NonAC regimens included R-CVP, BR, rituximab, radiation and lenalidomide and rituximab. Pts completed the Vulnerable Elders Survey (VES-13) and were considered vulnerable if they had a score >3. Event-free survival (EFS) was defined as the time from start of treatment until progression/relapse, retreatment, or death due to any cause; EFS at 24 months (EFS24) was a secondary endpoint. Overall survival (OS) was defined as the time from start of treatment until death due to any cause. EFS and OS for R-miniCHOP and non-IC groups, as well as vulnerable and not vulnerable by VES-13 were estimated using Cox models. Results A total of 190 pts received R-miniCHOP (N=139) or nonAC (N=51). R-miniCHOP-treated pts were older (median 82 vs 72 yrs) and more commonly female (55% vs 35%). The majority in both groups (86% vs 78%) received their treatment at a LEO center and did not receive therapy on a clinical trial (89% vs 100%). Most pts receiving R-miniCHOP had high-intermediate (32%) or high (30%) international prognostic index (IPI) score, while both high-intermediate (22%) and high (20%) IPI were lower in the nonAC-treated pts. Cell of origin was determined by immunohistochemistry as available clinically, with GCB 42% vs 31%, non-GCB 32% vs 29%, and not done/unknown 26% vs 39% in R-miniCHOP vs nonAC pts. Double-hit status by FISH was known in 115 pts and was positive in 13% of R-miniCHOP pts and 10% of nonAC pts. The diagnosis-to-treatment interval (DTI) was ≤14 days in 30% of R-miniCHOP pts and 24% of nonAC pts. Of pts who had a VES-13 score (N=119), vulnerable status trended higher in the R-miniCHOP (66%) vs the non-IC group (50%; p=0.15). With a median follow-up of 50.5 months, the median EFS was 37.1 months and median OS was 50.0 months. R-miniCHOP, compared with non-IC, was associated with improved EFS (unadjusted HR=0.54, 95% CI 0.36-0.81; p=0.003) and OS (unadjusted HR=0.58, 95% CI 0.38-0.88; p=0.011). When adjusted for age, IPI, LBCL subtype (diffuse large B-cell lymphoma, not otherwise specified vs other), DTI and VES-13, R-miniCHOP was strongly associated with improved EFS (HR=0.26, 95% CI 0.14-0.50; p<0.001) and OS (adjusted HR=0.29, 95% CI 0.15-0.57; p<0.001). EFS24 rate was 56% in all pts and was superior in pts treated with R-miniCHOP vs non-IC (62.7% vs 34.1%; p=0.001). In the R-miniCHOP group, vulnerable status (adjusted for IPI) was associated with inferior OS (HR=2.22, 95% CI 1.09-4.50; p=0.028) and trended towards inferior EFS (HR=1.76, 95% CI 0.93-3.32; p=0.083). In a competing-risk cumulative-incidence analysis, lymphoma was the leading cause of death for pts receiving R-miniCHOP at 2 (0.16, 95% CI 0.11-0.24) and 5 (0.20, 95% CI 0.14-0.28) yrs, as well as for those receiving non-IC at 2 (0.35, 95% CI 0.24-0.53) and 5 (0.38, 95% CI 0.26-0.56) yrs. Treatment-related mortality was similar in both groups. In the R-miniCHOP group, the 2-yr (0.20 vs 0.03) and 5-yr (0.20 vs 0.11) lymphoma-related mortality was higher in pts who were vulnerable vs not by VES-13, while there were no differences for mortality due to therapy or other causes by vulnerable status. Conclusions These data support R-miniCHOP as a curative regimen in older and frail pts with the majority achieving EFS24. R-miniCHOP compared to non-IC use is strongly associated with improved EFS and OS when adjusted for disease and patient related variables. Within the R-miniCHOP group, vulnerable status is associated with inferior OS. Additional research efforts are needed to optimize outcomes for this group of pts.

Abstract Background: De novo secondary central nervous system (CNS) lymphoma (SCNSL), characterized by concurrent systemic and CNS aggressive B-cell lymphoma at diagnosis, is a rare entity with a dismal prognosis. The prognostic significance of the initial anatomical compartment of CNS involvement (parenchymal vs. leptomeningeal) is not well-defined. We sought to analyze outcomes based on CNS localization in a single-institution cohort. Methods We retrospectively reviewed 34 patients with de novo SCNSL (including diffuse large B-cell lymphoma [DLBCL], high-grade B-cell lymphoma, not otherwise specified (HGBCL NOS), and double-/triple-hit lymphoma) treated at the Mayo Clinic between 2002 and 2025. Clinicopathological features, radiologic findings, treatment details, and survival outcomes were abstracted from the medical records. Patients were categorized by the primary site of CNS involvement. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan-Meier method, and groups were compared with the log-rank test. Results The study cohort had a median age of 61.5 years (range 19-79) and a median International Prognostic Index score of 3 (range 2-4). Pathological subtypes included double-/triple-hit lymphoma (n=14, 41.2%), non-germinal center B-cell like (non-GCB) DLBCL (n=11, 32.4%), germinal center B-cell like (GCB) DLBCL (n=8, 23.5%), and HGBCL NOS (n=1, 2.9%). Among patients with available data, 14 (43.8%) were positive for MYC rearrangement, 13 (54.2%) for BCL2 rearrangement, and 10 (45.5%) for BCL6 rearrangement. First-line regimens received by patients included MR-CHOP (n=18, 52.9%), DA-R-EPOCH (n=5, 14.7%), combined CODOX-M/IVAC (n=4, 11.7%), R-CHOP with intrathecal methotrexate (n=3, 8.8%), miscellaneous therapies (n=3, 8.8%), and MRT (n=1, 2.9%). The complete response rate was 58.8% (n=20). Intrathecal chemotherapy was administered to 61.8% of patients, and consolidation with autologous stem cell transplant was performed in 35.3%. Based on the CNS compartment involved, de novo SCNSL (DN-SCNSL) were classified into two subtypes: the leptomeningeal subtype, defined as leptomeningeal involvement with or without parenchymal CNS disease (LM-DN-SCNSL; n=22, 64.7%), and the parenchymal subtype, defined as parenchymal-only CNS involvement (P-DN-SCNSL; n=12, 35.3%). Parenchymal disease was multifocal in 75% of cases, supratentorial in 58.3%, and associated with a midline shift in 30%. Although no tested factors (including pathological subtype, age, sex, and bone marrow involvement) were significantly associated with the anatomical compartment, a non-significant trend was noted for BCL6 rearrangement to be more common in the leptomeningeal group (8/13, 61.5%) than the parenchymal group (2/8, 25.0%) (P = 0.18). Survival outcomes were analyzed for both anatomical and pathological classifications. For the parenchymal versus leptomeningeal cohorts, 2- and 5-year Overall Survival (OS) rates were 75.0% vs. 45.1% and 41.0% vs. 38.7%, respectively (P = 0.34). For Progression-Free Survival (PFS), the 2- and 5-year rates were 48.6% vs. 32.5% and 29.2% vs. 26.0%, respectively (P = 0.53). The 2-year OS rates by pathological subtype were 81.8% for non-GCB DLBCL, 45.8% for DHL/THL, and 37.5% for GCB DLBCL (P = 0.27). Survival outcomes (OS/PFS) were not significantly associated with patient age, sex, pathological subtype, or anatomical compartment. Conclusion DN-SCNSL can be anatomically classified into two subtypes: LM-DN-SCNSL and P-DN-SCNSL. While a prognostic trend favoring the parenchymal subtype was observed, the difference did not achieve statistical significance, a finding likely attributable to the small cohort size. This classification may hold clinical utility for guiding risk stratification, and larger studies are needed to validate these findings and identify the most optimal treatments for each subtype of DN-SCNSL.

Abstract Background Secondary central nervous system lymphoma (SCNSL) is a rare but significant complication, affecting approximately 2–10% of patients with diffuse large B-cell lymphoma (DLBCL), often leading to poor outcomes. The CNS-International Prognostic Index (CNS-IPI) remains a widely adopted tool for estimating CNS relapse risk; however, recent advances in molecular profiling and evolving treatment paradigms have challenged the adequacy of CNS-IPI alone in guiding risk assessment. This study evaluates the prognostic relevance of CNS-IPI in conjunction with molecular and clinical factors, specifically cell-of-origin (COO), gene expression, and frontline treatment regimens. The aim is to refine risk stratification and identify patients at highest risk for SCNSL. Method Building on our previous study, we analyzed data from patients with newly diagnosed DLBCL between 2002 and 2023, enrolled in the Mayo Clinic/University of Iowa Lymphoma Molecular Epidemiology Resource. Patients with primary CNS lymphoma, CNS involvement at diagnosis, or who received CNS-directed therapy were excluded. COO was determined using the Hans algorithm and/or Nanostring profiling. Multivariate cause-specific Cox regression models were used to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for predictors of CNS relapse that were measured at the time of initial DLBCL diagnosis. Cumulative incidence with competing risks and cause-specific Cox models were used to analyze relapse risk. Result We included 2137 patients in our analysis, median age at diagnosis was 63 (IQR 53-72), with 41.6% (n=890) females. The albumin level was below normal in 32.5% (n=350), 64.0% (n=1367) had CNS-IPI of intermediate/high, 37.5% (n=586) had non-germinal center B-cell (GCB) COO, and 14.5% (n=309) were double expressor positive. At a median follow-up of 83 months (IQR 42 -143), 82 (3.8%) SCNSL relapses were observed; the 3 and 5 yrs cumulative incidence for SCNSL from time of DLBCL diagnosis were 3.7% (95% CI 3.0-4.7) and 3.8% (95% CI 3.1-4.8) respectively. In models accounting for death as a competing risk, the 5-year cumulative SCNSL incidence estimate for patients with a CNS-IPI low was 1.5% (95% CI 0.8-2.7), Intermediate was 4.9% (95% CI 3.7-6.4), and high was 6.1% (95% CI 3.8-9.6). The 5-year cumulative SCNSL incidence estimate for patients with non-GCB COO was 5.2% (95% CI 3.6-7.4) compared to 2.4% (95% CI 1.6-3.6; p=0.001) for GCB patients, and the 5-year cumulative SCNSL incidence for double expressor positive patients was 8.6% (95% 5.8-12.6) compared to 1.5% (95% CI 0.8-2.7; p<0.001) for double expressor negative patients. In a multivariable model (including CNS-IPI, COO, double expressor, treatment group, extranodal involvement, and albumin group), CNS-IPI intermediate (HR 2.85 [1.58-5.14]; p<0.001) and high (HR 4.31 [2.15-8.65]; p<0.001), non-GCB COO (HR 2.36 [1.37-4.06]; p=0.002], double expressor (HR 5.56 [2.88-10.73]; p<0.001), and below normal albumin (HR 1.89 [1.02-3.50]; p=0.042) were independently associated with increased risk of SCNSL. The majority of the SCNSL relapse occurred within the first 2 years after DLBCL diagnosis; 81.7% (n=67), with only 6.1% (n=5) occurring after 5 years. For patients who developed SCNSL, the median OS from the time of SCNSL diagnosis was 24 months (95% CI 17-34), with 5 years OS of 31% (95% CI 22-43). These findings are being validated using an external, international cohort, and final analysis will be presented at the time of presentation at ASH. Conclusion In this large, prospectively followed cohort with long-term follow-up data, SCNSL remains a rare but devastating complication of DLBCL, with a 5-year cumulative incidence of 3.8% and it is associated with very poor long-term survival. While CNS-IPI remains a strong predictor of SCNSL risk, our study identifies non-GCB COO and double expressor phenotype as independent, additive molecular risk factors. These findings suggest that current risk stratification tools may underestimate the risk of CNS relapse in molecularly defined high-risk subgroups. Considering the poor outcomes linked to SCNSL and the absence of a connection with frontline treatment types, future research should aim to enhance risk prediction models through molecular profiling and explore prophylactic strategies to reduce the likelihood of SCNSL.

Abstract Background Primary central nervous system lymphoma (PCNSL) is a rare but aggressive non-Hodgkin lymphoma with limited prospective data to guide optimal therapy. This study evaluates real-world treatment patterns and survival outcomes using a large national cohort. Methods We analyzed data from the National Cancer Database (NCDB) for patients diagnosed with PCNSL between 2004 and 2020. Baseline demographics and treatment characteristics were analyzed using descriptive statistics. Survival outcomes were estimated using the Kaplan-Meier method and compared using log-rank tests. Multivariable Cox proportional hazards models were constructed to identify independent predictors of overall survival (OS). Results A total of 14,450 patients with PCNSL were included; 14,439 had first-line treatment data available, and 13,390 had survival data. Median age at diagnosis was 67 years (range 18-90), 49.2% were female, and 6.6% were African American. Chemotherapy alone was the modality of treatment for 81.6% (n=11,787), 12.2% (n=1,769) received chemoradiation, 3.8% (n=545) received chemotherapy and underwent autologous stem cell transplant (ASCT) but no radiation, and 2.4% (n=348) received radiation only. The median follow-up time was 73.5 months (95% CI 71.4-75.6), and the median OS for the group was 17.4 months (95% CI 16.2-18.6). Survival differed significantly by age group (median OS: <65 yrs, 46.1 months [95% CI 41.7 -50.6]; 65-79 yrs, 12.5 months [95% CI 11.2-13.7]; ≥80 yrs, 3.2 months [95% CI 2.9-3.5]; p<0.001), by year of diagnosis (median OS: 2004-2009, 11.0 months [95% CI 9.9-12.1]; 2010-2015, 20.0 months [95% CI 17.4-22.6], 2016-2020, 29.7 months [95% CI 26.0-33.4]; p<0.001), and by frontline treatment strategy (p<0.001). Patients treated with chemotherapy followed by ASCT had the longest median OS (not reached), while those receiving radiation alone had the poorest outcomes with a median OS of 3.9 months (95% CI 3.2-4.7); chemoradiation patients had a median OS of 31.8 months (95% CI 27.5-36.1), and chemotherapy only patients had a median OS of 13.9 months (95% CI 12.7-15.1)(p<0.001). African American patients had an inferior outcome with a median OS of 10.2 months [95% CI 7.2-13.1], compared to “other” races with a median OS of 36.9 months [95% CI 27.6-46.1] (p<0.001), although this is more similar to the median outcome for white patients (17.0 months [95% CI 15.8-18.3]). Treatment at an academic center (25.7 vs 11.0 months) and a high case-volume center (defined as >median number of cases per facility) (18.6 vs 10.5 months), and lower Charlson-Deyo comorbidity scores (CDS) (defined as CDS 0-1) (21.7 vs 6.8 months) were also associated with superior OS (p<0.001). In multivariable analysis, increasing age ( 65-79 y.o HR 1.60 [1.52-1.1.68], ≥80 y.o HR 2.95 [2.75-3.17]), CDS≥2 (HR 1.36 [1.28-1.44]), and African American race (HR 1.12 [1.02-1.24]) were adverse predictors of OS (all p<0.001), while treatment strategy (ASCT [HR 0.17], chemoradiation [HR 0.53], chemotherapy only [HR 0.68] compared to radiation alone), more recent year of diagnosis (2010-2015 [HR 0.81], 2016-2020 [0.70]), and treatment at an academic center [HR 0.81] remained predictors of superior OS (all p<0.001). Conclusion In this large national cohort of patients with PCNSL, treatment with chemotherapy followed by ASCT was associated with the most favorable survival outcomes. Patients treated with chemoradiation or multi-agent chemotherapy also demonstrated improved survival compared to those receiving radiation monotherapy or single-agent chemotherapy alone, respectively. Age, comorbidities, race, and treatment at academic centers were also significant independent predictors of OS. These findings highlight substantial disparities in outcomes and underscore the importance of optimizing treatment strategies, improving access to high-intensity therapies, and supporting timely referral to academic or high case-volume centers, particularly for younger and medically fit patients.

Abstract Introduction: Brentuximab vedotin (BV) was approved by the US FDA for relapsed/refractory (R/R) classical Hodgkin lymphoma (cHL) in 2011 and for previously untreated advanced stage cHL in 2018. The immune checkpoint inhibitors (ICIs) nivolumab and pembrolizumab were approved for R/R cHL in 2016 and 2017, respectively. These novel agents have significantly changed the landscape of cHL management in recent years. In this study, we analyzed data from the National Cancer Database (NCDB) to investigate whether survival outcomes have improved in the novel agent era (2011 and beyond) for patients with cHL in the US. Methods: Patients diagnosed with cHL between 2004 and 2020 were identified in the NCDB. Those with missing survival data were excluded. Three treatment eras were defined based on the timing of FDA approval of novel agents: Chemo era (2004–2010), BV era (2011–2015), and ICI era (2016–2020). Patient and disease characteristics were compared using the Chi-square test. Survival analyses were conducted using the Kaplan-Meier method and Cox proportional hazards model. Results: A total of 85,488 patients with newly diagnosed cHL were identified: 36,150 in the Chemo era, 27,058 in the BV era, and 22,280 in the ICI era. Compared to the chemo era, the BV and/or ICI eras included slightly more patients aged ≥66 (17.3% vs 17.8% vs 19.5%), male (53.6% vs 54.6% vs 53.8%, p=0.042), Black (11.3% vs 12.9% vs 12.5%), Hispanic (9.0% vs 9.9% vs 10.9%), with Charlson-Deyo score ≥1 (14.2% vs 16.1% vs 17.2%), B symptoms (44.1% vs 46.9% vs 46.9%), treated at academic centers (32.4% vs 35.9% vs 36.4%), on Medicaid (9.5% vs 12.2% vs 13.7%) or Medicare (19.3% vs 20.3% vs 21.7%), and with higher income (≥4th quartile by 2008-2012 standard; 34.5% vs 34.3% vs 36.3%) (all p<0.001 except for sex, noted above). Ann Arbor stage data were not available in the ICI era. There were slightly more patients with advanced-stage disease in the BV era compared to the Chemo era (45.4% vs 40.3%, p<0.001). IPS scores were available for only 2,344 patients, with a similar distribution (0–2 vs 3–7) across the three eras (p=0.957). The median follow-up was 6.9 years (95% CI 6.9-7.0). Overall survival (OS) significantly improved in the novel agent eras: the 5-year OS rate was 79.8% (95% CI, 79.4%–81.2%) for patients diagnosed in the Chemo era, 81.9% (95% CI, 81.9%–82.4%) in the BV era, and 84.1% (95% CI, 83.5%–84.8%) in the ICI era (p<0.001). Statistically significant OS improvements were observed across all subgroups, including age (≤40: 92.8% vs 94.9% vs 96.5%, 41-65: 80.5% vs 81.8% vs 84.4%, ≥66: 41.4% vs 46.6% vs 52.7%), sex (male: 78.0% vs 80.0% vs 81.9%, female: 81.9% vs 84.1% vs 86.7%), race (White: 79.9% vs 81.6% vs 83.7%, Black: 77.2% vs 81.2% vs 84.6%, others: 84.5% vs 87.0% vs 89.3%), ethnicity (Hispanic 79.7% vs 82.3% vs 85.1%, Non-Hispanic 80.0% vs 81.9% vs 84.1%), Charlson-Deyo score (0: 83.5% vs 85.4% vs 87.9%, 1: 62.8% vs 68.2% vs 73.2%, 2: 42.5% vs 52.6% vs 58.3%, 3+: 49.0% vs 54.4% vs 51.7%), B symptoms (yes: 74.7% vs 78.9% vs 81.5%, no: 84.3% vs 85.1% vs 87.5%), treatment facility type (academic: 69.8% vs 74.0% vs 77.1%, non-academic: 66.4% vs 67.6% vs 70.2%, unknown: 92.9% vs 95.0% vs 96.5%), insurance type (private: 89.9% vs 91.2% vs 93.4%, Medicare: 46.5% vs 52.3% vs 57.2%, Medicaid: 79.1% vs 84.5% vs 87.7%) and income level (Q1: 74.1% vs 77.1% vs 81.1%, Q2 77.7% vs 80.1% vs 81.3%, Q3 79.8% vs 81.6% vs 85.2%, Q4: 82.8% vs 84.9% vs 87.0%) (all p<0.001). In multivariate Cox regression analysis adjusting for age, sex, race, ethnicity, Charlson-Deyo score, B symptoms, treatment facility type, insurance type, and income level, diagnosis in the BV era (HR = 0.856; 95% CI, 0.827–0.885; p<0.001) and ICI era (HR = 0.666; 95% CI, 0.636–0.698; p<0.001) was independently associated with improved OS compared to the Chemo era. Conclusions: In this large NCDB-based analysis, survival outcomes for patients with cHL significantly improved in the BV and ICI eras compared to the Chemo era. OS improvement was consistent across all subgroups and remained independent of age, sex, race, ethnicity, Charlson-Deyo score, B symptoms, treatment facility type, insurance type, and income level in multivariate analysis. These findings may reflect the positive impact of BV and ICIs on survival in patients with cHL.