Search for a command to run...
To the Editor, Chimeric antigen receptor (CAR) T-cell therapies have revolutionized the treatment of plasma cell dyscrasias.1 While associated with excellent long-term outcomes, CAR T-cell therapy is also associated with significant side effects. With increasing use, novel adverse events, such as immune colitis, are now being reported.2 Non-Immune Effector cell-associated neurotoxicity syndrome (ICANS) neurotoxicity (NINT) has been increasingly described post CAR T-cell therapy. It has a wide range of clinical presentations, ranging from cranial nerve palsies to Parkinson-like movement and neurocognitive toxicity (MNT).3 The data regarding the incidence of MNTs and NINTs after B-cell maturation antigen (BCMA)= CAR T are evolving, with growing evidence from real-world studies on the true incidence and potential insights into the pathogenesis and appropriate therapy for these syndromes.4 Given the growing utilization of BCMA CAR T-cell therapy, and its use in patients in earlier lines of therapy, more information is needed regarding NINTs to counsel patients better and to have a better sense of the risk–benefit ratio. We conducted a systematic review to study the incidence and outcomes of patients with NINTs post-CAR T-cell therapy in patients with plasma cell dyscrasias. We conducted a comprehensive literature search in PubMed/MEDLINE, EMBASE, Web of Science and CENTRAL using tailored strategies for each database. The study was registered on PROSPERO (CRD420251074357) and followed the PRISMA reporting guidelines. Two independent authors performed a systematic search. The detailed search strategy is presented in Table S1. Studies published between January 2015 and May 2025 were screened, and the search was limited to prospective clinical trials, randomized controlled trials and prospective and retrospective observational studies with 10 or more patients. Case reports, retrospective case series with fewer than 10 patients, conference abstracts and publications in a language other than English were excluded from the analysis. This was done as the primary outcome of interest was the pooled incidence of NINTs, which could not be calculated or may have been overestimated by including case reports or small case series. Search results were imported into EndNote for deduplication and then screened in Covidence by two independent reviewers. Disagreements were resolved by consensus or adjudication by a third independent reviewer. Data extraction was performed in duplicate using a standardized form. The risk of bias was assessed using the Cochrane RoB 2 tool for randomized studies and the Newcastle–Ottawa Scale for observational studies. Data were synthesized narratively and, where appropriate, quantitatively using a random-effects meta-analysis with heterogeneity assessed via the I2 statistic. R software (version 4.4.0) was used for analysis, and the pooled incidence was reported as a percentage with 95% confidence intervals (CIs) using a random-effects model. Publication bias was assessed using doi plot. The primary outcomes were the incidence, category and outcomes of NINT, while secondary outcomes included timing, reversibility, risk factors, treatment interventions and neurological sequelae. Our initial search yielded 8719 studies, which were screened for eligibility and inclusion in the study (Figure S1). After removing studies that reported on the same dataset, 10 studies (0.11%) were included for analysis, including five retrospective analyses and five prospective clinical trials.5-14 We did not find any study detailing the development of NINTs post GPRC5D-directed CAR T-cell therapy. Not all studies reported all NINT categories. As a result, denominators vary across analyses, and each pooled proportion is based only on studies that provided extractable data for that specific end-point. Six studies had data on patients who received ciltacabtagene autoleucel (cilta-cel), two studies had data on patients who received idecabtagene vicleucel (ide-cel) and two studies had data on patients receiving ide-cel and cilta-cel respectively. The studies by Cohen et al. and Martin et al. reported on the CARTITUDE-1 trial at different time points; therefore, the information was combined for analysis purposes.8, 10 One study did not report the total incidence of NINTs but only the incidence of individual NINTs and was therefore included in the pooled analysis only for specific NINTs.11 Baseline characteristics of the studies are shown in Table S2. The total number of patients across studies was 3155. The pooled incidence of NINTs across studies was 6.6% (85 patients) (95% CI 1.2–15.2; I2 = 95.9%, p < 0.001, n = 1659) (Figure 1A) (Table 1). The median time to onset of NINTs post-CAR T infusion was reported in six studies, and the weighted median was 24.3 days (range, 11–387). The most common individual NINT reported was cranial nerve palsies, with the most common being facial nerve involvement, with a pooled incidence of 2.7% (98 patients) (95% CI 0.2–7.2, I2 = 93.9%, p < 0.001, n = 3155) followed by MNTs with a pooled incidence of 1.5% (47 patients) (95% CI 0.0–4.4, I2 = 87.8%, p < 0.001, n = 3155) (Figure 1B,C). Other NINTs included neuropathy (pooled incidence 0.3% (14 patients); CI 0.0–1.3; I2 = 70.7%, p = 0.006, n = 3155), cognitive/personality changes (including confusional states, posterior reversible encephalopathy) (pooled incidence 0.9% (six patients); CI 0.0–2.5; I2 = 0.0%, p = 0.507, n = 3155) and some patients with ataxia, diplopia without cranial nerve neuropathy and dysautonomia. The incidence of NINTs was significantly higher post-cilta-cel infusion compared to patients who received ide-cel (12.7% vs. 0.04%, p < 0.001; Figure S2A). This was true for individual NINTs, including cranial nerve palsy (5.55% vs. 0% p < 0.001) and MNTs (2.74% vs. 0.03% p = 0.0034; Figure S2B,C). A difference in the occurrence of NINTs based on other parameters, such as the occurrence of CRS, ICANS, the number of lines of therapy and gender, could not be determined due to the lack of available information in most studies. As there was possible duplication of cases between the reports by Blumenberg et al. and Karschnia et al., an analysis by removing the cilta-cel patients from the Karschnia series was done. The results obtained were largely similar to the ones obtained in the full analysis (pooled incidence of NINT—6.6%; pooled incidence of MNT—1.5%; pooled incidence of cranial neuropathy—2.7%). Treatment for NINTs varied, ranging from steroids, Intravenous Immunoglobulin (IVIg), anakinra and cyclophosphamide. Information regarding the reversal (partial or complete) of NINTs was available in six studies. NINTs improved either entirely or partially on follow-up in the majority of patients, with a pooled incidence of 62.2% (95% CI 50.2–73.6; I2 = 0.0%, p =0.41, n = 82) (Figure S2A). In terms of specific NINTs, cranial nerve palsy resolved in most patients with a pooled incidence of reversibility of 80.4% (95% CI 60.6–95.3; I2 = 30.1%, p = 0.231, n = 40; Figure S3C). However, movement disorders did not respond as well to therapy, with a pooled incidence of reversibility of only 46.2% (95% CI 19.5–73.6; I2 = 0.0%, p = 0.468, n = 21; Figure S3B). Twelve patients (14.1%) with NINTs died during different study follow-up periods—five patients due to progressive neurotoxicity, six patients due to infections, possibly secondary to immunosuppression given for the NINT and one patient due to progressive myeloma. The findings of our study align with a recent Center for International Blood and Marrow Transplant Research (CIBMTR) study comparing real-world outcomes of patients with relapsed/refractory multiple myeloma undergoing CAR T-cell therapy, showing a higher incidence of NINTs with cilta-cel in comparison with ide-cel.15 The reason for this discrepancy between cilta-cel and ide-cel is incompletely understood. A possible hypothesis may be that the dual BCMA binding structure of cilta-cel may lead to stronger binding, and hence not only greater efficacy but also a greater incidence of NINTs and MNTs. While the incidence of MNTs was low (2.74% with cilta-cel and 1.21% overall), fewer than 50% of patients experienced improvement with therapy. Furthermore, 14% of patients with NINTs died during follow-up, with the majority of deaths being linked either directly or indirectly to the NINT. With growing knowledge regarding these syndromes and the development of EBMT guidelines for their management, outcomes may improve in future datasets.3 However, these results may also temper the move to use CAR T-cells in initial lines of therapy. There does not seem to be a correlation between previous lines of treatment and the development of NINTs. Indeed, in a recent International Myeloma Working Group (IMWG) global registry study, 15% of patients who developed NINTs had ≤1 prior line of therapy.4 Encouragingly, most patients with cranial nerve palsies, which were the most common NINT, recovered with treatment with corticosteroids. The patients with MNTs had an inconsistent response to steroids in our review and were often treated with further immunosuppressive therapy including anakinra and cyclophosphamide. Infection seems to be an important factor in the outcome of patients with NINTs, with infection being the most common cause of death in our pooled analysis, possibly secondary to extensive immunosuppression given for the MNT. Our study has some limitations. There was inconsistent reporting of NINTs across studies, with significant heterogeneity among them (I2 = 95.9%). The Newcastle–Ottawa Scale score for most studies was 6, with only three studies having a score above 7 (Table S3). The doi plot revealed a Luis Furuya- Kanamori (LFK) index of 2.96, highlighting potential publication bias. Additionally, individual patient-level data were not available, limiting the elucidation of other risk factors beyond the CAR construct that may be related to the development of NINTs. Additional real-world and translational studies are required to fully understand the factors leading to the development of NINTs. M.S. and F.N.U.D. were responsible for conducting the search, screening potentially eligible studies, extracting and analysing data, and writing the first draft of the manuscript. A.Jandial, A.Jain, G.P. and A.K. were involved in the writing of the manuscript and verification of findings. R.C., P.M. and C.P. conceptualized the paper and designed the protocol. C.P. performed the statistical analysis. P.M. and C.P. wrote and edited the final manuscript. All authors reviewed and finalized the final manuscript. We would like to thank Lauren Tong, Clinician Librarian affiliated with the University of Arkansas for Medical Sciences for her help. R.C.—Consulting/Advisory Board: Janssen, Sanofi, Adaptive Biotech, Alexion, Pfizer; Honoraria: Janssen, Sanofi, Adaptive Biotech, Alexion, Pfizer; Research funding: Genentech, AbbVie. The other authors have no financial or other disclosures to declare. No funding was received for this manuscript and research. Data will be made available after reasonable request by contacting the corresponding author. Figure S1. Figure S2. Figure S3. Table S1. Table S2. Table S3. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.