Several recent studies have shown the promising efficacy of chimeric antigenic receptor (CAR) T cells in treating CNS lymphomas. However, data on neurotoxicity in this setting are limited. The objective of this study was to describe neurotoxicity in patients with CNS lymphoma treated with anti-CD19 CAR T cells and to identify risk factors. We retrospectively selected adult patients with isolated CNS relapse of B-cell lymphomas treated with CAR T cells at Pitié-Salpêtrière Hospital between January 2020 and January 2024 from the French Oculo-Cerebral Lymphoma network database. We collected clinical, biological, and imaging data before and after CAR T-cell infusion to investigate neurotoxicity. We considered only neurologic deterioration for which causes other than CAR T-cell toxicity were reasonably ruled out. According to the selection criteria, 48 patients (44% female, 28 with primary and 20 with secondary CNS lymphomas) were analyzed. The median age was 62 years (range: 30-82) at the time of CAR T-cell infusion, and the median Montreal Cognitive Assessment (MoCA) score was 23. Twenty-five patients received tisa-cel, 21 received axi-cel, and 2 received brexu-cel. Thirty-one patients (65%) experienced neurotoxicity, including 11 patients with grade 3-4 neurotoxicity (23%). The symptoms started at a median of 5 days (range: 1-10) after CAR T-cell infusion. The symptoms were cognitive disorders (N = 30), balance disorders (N = 18), consciousness disorders (N = 6), tremors (N = 6), seizures (N = 4), and motor deficits (N = 4). Brain MRI revealed pseudoprogression in 7 of 26 patients (27%), and there was a transient increase in CSF IL-10 levels in 7 of 29 patients (24%). Age 65 years or older (p = 0.04, OR: 4.4 [95% CI 1.1-19.3]) and a MoCA score <26 at the time of CAR T-cell infusion (p = 0.04, OR: 12 [95% CI 4-29]) were significantly associated with a greater risk of grade 3-4 neurotoxicity (exploratory analysis). Twenty patients (42%) received steroids. The median duration of neurologic impairment was 100 days (range: 4 days-18 months) in patients with grade 3-4 neurotoxicity. Although the rate of neurotoxicity seems acceptable in CNS lymphomas, the risk of unusual prolonged neurologic deterioration is high in patients with grade 3-4 neurotoxicity. Special attention should be given to older patients with cognitive impairment who seem at greater risk of severe forms of neurotoxicity. Larger series are warranted to confirm these results.

Purpose To develop and validate a deep learning model for automatic segmentation of primary central nervous system lymphoma (PCNSL) at postcontrast T1-weighted MRI. Materials and Methods Data were retrospectively collected from patients with pathologically proven immunocompetent PCNSL between September 2010 and February 2022. Postcontrast T1-weighted MRI scans were used to train and validate a deep learning model based on the nnU-Net framework. Manual segmentation by neuroradiologists served as the reference standard. The model was trained using an internal dataset from a single center and tested on both internal and external test sets from seven additional centers. Performance was assessed using Dice score, mean average surface distance, and F1 score. Statistical comparisons were performed using Mann-Whitney U test and bootstrap resampling for CIs. Results The study included 135 patients (68 female, 66 male, and one of unspecified sex; internal dataset: mean age ±SD, 67.0 years ± 12.0; external dataset: mean age, 75.5 years ± 13.6). The model achieved a mean Dice score of 0.84 (95% CI: 0.79, 0.88) on the internal test set (n = 44) and 0.88 (95% CI: 0.84, 0.91) on the external test set (n = 48), with no evidence of a difference between test sets (P = .59). Performance varied by lesion type; accuracy was highest in homogeneous discrete lesions, and performance was slightly decreased when numerous poorly defined infracentimetric lesions occurred. Strong volumetric correlation was observed between automatic and manual segmentations (internal: r = 0.99, P < .001; external: r = 0.98, P < .001). Conclusion A deep learning model achieved accurate and robust automatic segmentation of PCNSL across multiple clinical centers with different MRI acquisition parameters. Keywords: Brain Lymphoma, Brain Tumor, Automatic Segmentation, Artificial Intelligence, Deep Learning Supplemental material is available for this article. © RSNA, 2025.

The objective of this study was a multicentric evaluation of professional practices, analyzing the irradiation technique itself and its impact on survival and recurrence sites, in primary central nervous system lymphomas (PCNSLs). We retrospectively analyzed the technical and clinical records of 79 PCNSL patients included in the database of the national expert network for oculocerebral lymphoma ('LOC') who were treated with brain radiotherapy as first-line treatment for newly diagnosed primary central nervous system lymphoma between 2011 and 2018. The number of patients treated with brain radiotherapy gradually decreased over time. The heterogeneity of radiotherapy prescriptions was significant, and 55% of them did not comply with published recommendations in terms of irradiation dose and/or volume. The proportion of complete responders to induction chemotherapy treated with reduced-dose radiotherapy increased over time. Partial brain radiotherapy was associated with significantly lower overall survival in univariate analysis. In partial responders to induction chemotherapy, increasing the total dose to the brain >30 Gy and adding a boost to the WBRT induced a trend toward improved progression-free and overall survival. Five recurrences (13%) occurred exclusively in the eyes, all in patients whose eyes had been excluded from the irradiation target volume and including 2 patients without ocular involvement at diagnosis. The visibility of recommendations for prescribing brain radiotherapy for the treatment of newly diagnosed primary central nervous system lymphoma needs to be improved to harmonize practices and improve their quality. We propose an update of the recommendations.

The optimal consolidation strategy for primary central nervous system lymphoma (PCNSL) remains controversial. Preventing radio-induced neurotoxicity of consolidation treatment through reduced-dose whole-brain radiotherapy (rdWBRT) at a dose of 23.4 Gy is an interesting alternative to conventional WBRT in patients aged <60 years. From the LOC Network (Network for Oculo-cerebral Lymphomas) database, we retrospectively selected patients with PCNSL aged <60 years who showed complete (CR) or unconfirmed CR after high-dose methotrexate-based chemotherapy and had received consolidation rdWBRT as the first-line treatment. If available, prospective neuropsychological follow-ups were reported. Twenty-nine patients diagnosed between 2013 and 2018 met the study selection criteria. Nine (31%) patients experienced relapse during the follow-up, with a median time from radiotherapy to recurrence of 8.7 months (interquartile range, 4-11.5). Five of those patients received salvage treatment and consolidation with intensive chemotherapy and autologous stem cell transplantation. Progression-free survival rates were 89% (95% confidence interval [CI] 79%-100%), 72% (95% CI, 56%-88%), and 69% (95% CI, 52%-85%) at 1, 2, and 5 years, respectively. Overall survival rates were 100%, 89% (95% CI, 79%-100%), and 86% (95% CI, 74%-99%) at 1, 2, and 5 years, respectively, and were consistent with those observed for standard-dose WBRT (sdWBRT). No prognostic factor was identified. The results of the 36-month neuropsychological follow-up for a subset of patients appeared reassuring, with most patients exhibiting maintenance of or improvements in their baseline conditions. Our results, combined with phase 2 study results, support the use of rdWBRT instead of sdWBRT as a consolidation treatment in <60-year-old patients showing CR after induction treatment.