Age as a moderator of neurocognitive performance in patients with large B-cell lymphoma treated with axicabtagene ciloleucel Authors: Aasha I. Hoogland, Xiaoyin Li, Bihe Hu, Nathaly Irizarry-Arroyo, Taylor Welniak, Yvelise Rodriguez, Michael D. Jain, Laura B. Oswald, Brent J. Small, Julia Thornton Snider, Sally W. Wade, Julio C. Chavez, Farhad Khimani, Aleksandr Lazaryan, Hien D. Liu, Taiga Nishihori, Javier Pinilla-Ibarz, Bijal D. Shah, Joseph A. Pidala, Jennifer Logue, Margaret Booth-Jones, Abu-Sayeef Mirza, Jeffrey Huang, Frederick L. Locke, Heather S. L. Jim Published: 2026 ABSTRACT Chimeric antigen receptor (CAR) T-cell therapy, including axicabtagene ciloleucel (axi-cel), is efficacious for older patients. However, patients aged 65 + are at risk for more severe neurotoxicity after infusion than younger patients, and post-infusion cognitive outcomes in older patients are unknown. This study assessed changes in objective neurocognitive performance in patients age 65 + compared to those younger than 65 in the first year after axi-cel and examined clinical factors influencing these changes. In patients with large B-cell lymphoma, neuropsychological assessments were conducted prior to initiation of axi-cel and at 30, 90, and 360 days after infusion. Mixed effects models were used to investigate age differences in changes in overall cognition (e.g., total neuropsychological performance [TNP]) and specific domains (i.e., attention, executive function, verbal ability, immediate and delayed memory, and visuospatial abilities). Among 155 participants (63% male, 90% White), older adults (n = 69) were more likely than younger adults (48% vs. 34%, p < 0.05) to demonstrate overall neurocognitive impairment before infusion. However, older adults demonstrated greater recovery of cognitive abilities (e.g., immediate verbal memory and delayed memory) over time (ps < 0.01). These results suggest that CAR T-cell therapy may have a less detrimental impact on neurocognitive performance in older patients compared to younger patients.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s00520-026-11169-w. Chimeric antigen receptor (CAR) T-cell therapy is increasingly used to treat older adults with large B-cell lymphoma (LBCL), in part due to its demonstrated efficacy in this population. Subgroup analyses from the ZUMA-1 trial, which evaluated the efficacy of axicabtagene ciloleucel (axi-cel) demonstrated that patients aged 65 + had high clinical (92%) and complete (75%) response rates [1]. Similarly, subgroup analyses from the ZUMA-7 trial demonstrated that patients aged 65 + had significantly longer survival (21.5 vs. 2.5 months) and better objective response rates (88% vs. 52%) compared to standard care at the time the study was conducted (i.e., chemotherapy followed by autologous stem-cell transplantation for responders) [2]. Real-world administration of axi-cel to patients aged 65 + also confers favorable objective (73%) and complete (56%) response rates [3]. However, despite improved clinical outcomes, patients aged 65 + in the ZUMA-1 and ZUMA-7 trials were more likely to present with severe adverse neurologic events (i.e., grade 3 or higher) than younger patients (44% vs. 28%) [1] and older patients treated with standard of care (27% vs. 2%) [2], respectively. Few studies have examined neurocognitive outcomes after axi-cel in particular, and little is known about the effects of age on neurocognitive outcomes in patients treated with axi-cel. Neurocognitive impairment is common after CAR T-cell therapy, including axi-cel, regardless of age. At 90 days and one year after treatment, 48–69% and 35–57% of patients treated with axi-cel, or in cohorts predominantly treated with axi-cel, exhibit neurocognitive impairment, respectively [4, 5]. Further, one out of every four patients predominantly treated with axi-cel report deterioration in cognitive abilities in the first year after treatment [6]. Only one study (i.e., one of the studies from which the data used in this project were derived) has evaluated age (and other sociodemographic and clinical factors) as a moderator of neurocognitive performance after axi-cel, finding no association between age and total neuropsychological performance (TNP) [4]. Age-related changes in neurocognitive domains and associated risk factors were not explored. There remains a need to investigate associations between age and neurocognitive performance in LBCL patients treated with CAR T-cell therapy. Several clinical factors may contribute to neurocognitive impairment after infusion. Patients predominantly treated with axi-cel who experience grade 2 + neurotoxicity demonstrate worse neurocognitive performance post-infusion than those with grade 1 (or no) neurotoxicity [4]. Additionally, patients with 4 + prior lines of therapy demonstrate worse neurocognitive performance both before and after infusion [4]. Early responders (i.e., patients with a complete or partial response to treatment by 90 days) also demonstrate better neurocognitive functioning by one year after infusion than non-responders [4]. Finally, while cytokine release syndrome (CRS) is a common consequence of CAR T-cell therapy, no published study has explored the role of CRS in neurocognitive functioning in older patients treated with CAR T-cell therapy. Thus, CRS, neurotoxicity, prior therapy, and disease response are all potentially related to neurocognitive functioning. To examine the association between age and cognition among LBCL patients treated with axi-cel, we pooled data from two existing longitudinal studies [4, 5]. The goals were to: 1) compare trajectories of neurocognitive performance in older (aged 65 + years) vs. younger patients (aged < 65 years) treated with axi-cel, and 2) explore clinical risk factors (i.e., disease status at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) that could influence these changes over time. We hypothesized that older patients would demonstrate worse neurocognitive performance than younger patients before infusion and over time. We also hypothesized that active disease (at days 90 and/or 360), high grade CRS, high grade neurotoxicity, and more lines of therapy would be associated with worse neurocognitive performance. Methods Procedure Participants were recruited as part of two studies examining neurocognitive performance in patients treated with CAR T-cell therapy. Eligibility criteria have been described previously [4–8]. Pooled analyses were restricted to patients diagnosed with LBCL and treated with axi-cel. Potential participants were identified in a HIPAA-compliant manner in consultation with the treating medical oncologist and through medical chart review. Neurocognitive assessments were administered either online or in person prior to receiving conditioning therapy (at baseline), and again at 30, 90, and 360 days after infusion by a member of the research team supervised by a clinical neuropsychologist. Clinical trial number: not applicable. Measures Sociodemographic and clinical characteristics At baseline, participants self-reported age, sex, race, ethnicity, marital status, and education level. Clinical variables were abstracted from medical records or collected as part of a clinical trial. These included: Karnofsky Performance Status (KPS), disease status at days 90 and 360, prior lines of therapy, and highest grades of post-infusion CRS and neurotoxicity. As reported previously [4, 7, 8], neurotoxicity was initially graded using a CAR-related encephalopathy syndrome grade as defined by the CAR T-cell Therapy-Associated Toxicity (CARTOX) working group [9] and later using the immune effector cell-associated neurotoxicity syndrome (ICANS) grade [10]. Neurotoxicity grades were scored on a 4-point scale (1 to 4), with higher grades indicating worse toxicity. Neurocognitive performance Pre-treatment intellectual ability (i.e., IQ) was estimated using the Wechsler Test of Adult Reading [11]. Neurocognitive tests were selected [4] based on existing research in patients with hematologic cancers [12, 13] and recommendations from the International Cancer and Cognition Task Force[14] (see Table 1). Parallel forms were used when possible to reduce practice effects. When patients were unable to complete the full battery of tests in-person, selected subtests of the RBANS (i.e., list learning, story memory, semantic fluency, digit span, list recall, list recognition, and story recall) were administered over the phone. For each test, raw scores were standardized as T-scores based on age-adjusted population norms [15], and averaged to create domain scores (i.e., attention, executive functioning, verbal ability, immediate verbal memory, visuospatial abilities, and delayed memory) in which higher scores indicate better performance. The average of all domain scores was interpreted as a TNP score. Table 1Neurocognitive domainsAttentionColor Trails Test (part 1)CPT-3 omissionsCPT-3 reaction timeRBANS digital spanExecutive functioningColor Trails Test (part 2)Stroop interferenceRBANS codingVerbal abilityRBANS picture namingRBANS semantic fluencyImmediate verbal memoryRBANS list learningRBANS story memoryVisuospatial abilitiesRBANS figure copyRBANS line orientationDelayed memoryRBANS list recallRBANS list recognitionRBANS story recallRBANS figure recallCPT-3 Conners Continuous Performance Test Third Edition, RBANS Repeatable Battery for the Assessment of Neuropsychological Status Table 1Neurocognitive domainsAttentionColor Trails Test (part 1)CPT-3 omissionsCPT-3 reaction timeRBANS digital spanExecutive functioningColor Trails Test (part 2)Stroop interferenceRBANS codingVerbal abilityRBANS picture namingRBANS semantic fluencyImmediate verbal memoryRBANS list learningRBANS story memoryVisuospatial abilitiesRBANS figure copyRBANS line orientationDelayed memoryRBANS list recallRBANS list recognitionRBANS story recallRBANS figure recallCPT-3 Conners Continuous Performance Test Third Edition, RBANS Repeatable Battery for the Assessment of Neuropsychological Status Neurocognitive domains CPT-3 Conners Continuous Performance Test Third Edition, RBANS Repeatable Battery for the Assessment of Neuropsychological Status Neurocognitive test and domain performance was considered to be impaired if patients’ scores were 2 + standard deviations (SDs) below the mean (e.g., T-scores of 30 or less [14]). Overall cognitive impairment was determined by identifying when participants scored 2 + SDs below the mean on one test or 1.5 + SDs below the mean on two tests [14]. Age group differences in neurocognitive performance were considered clinically significant if they exceeded 5 points (i.e., 0.5 SD), reflecting a medium effect size [16]. Statistical analyses Participants were included in analyses if they were diagnosed with LBCL and if they completed at least one neurocognitive assessment. Means, SDs, frequencies, and percentages were used to describe sociodemographic and clinical characteristics of participants. Rates of neurocognitive impairment were described using percentages for each age group, and depicted graphically with 95% confidence limits using Wald intervals. The effects of age (< 65 years vs. 65 + years) on changes in neurocognitive performance over time were evaluated using random-effects mixed models with linear and quadratic effects of time. The advantage of these models, over traditional repeated measures analyses, is that they estimate within-person changes in performance and allow for unequal assessment periods and missing data across the follow-up period. Significant age by time interactions were decomposed using within-group linear mixed models to describe within-group changes in neurocognitive performance, and between-subjects comparisons of least squares means were used to describe cross-sectional differences in neurocognitive performance by age group. Clinical associations (i.e., disease status at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) with age and neurocognitive performance were examined using linear mixed models. These models included three-way age by time by covariate interactions to examine the effect of age group and clinical moderators on change in neurocognitive performance. Significant age by time by clinical moderator interactions were decomposed using least squares means. All analyses were completed using SAS, version 9.4 (SAS Institute, Cary, NC; RRID:SCR_008567). Results Participant characteristics and the flow of participants through each study have been published previously [4, 5]. As shown in Table 2, participants (n = 155) ranged in age from 19 to 82 (Mean age = 61, SD = 12), with adults aged 65 + (n = 69) representing 45% of participants. At baseline, participants 65 years of age or older (i.e., older adults) were more likely to have a higher IQ (p < 0.01) and a KPS less than 80 (p = 0.02) than younger participants. Variables considered for inclusion as covariates included sex [4, 17], baseline IQ, and KPS. Spearman’s Rho correlations revealed that sex was not associated with TNP (p > 0.05), and KPS covaried with age (p = 0.04). However, to examine the differential effects of age and functional status on changes in neurocognitive performance, sensitivity analyses were conducted with and without KPS. All longitudinal analyses included IQ as a covariate. Table 2Participant characteristicsYounger Adults (n = 86)Older Adults (n = 69)p-valueAge: M [range]53.22 [19–64]71.56 [65–82] < 0.0001Female: n (%)35 (41)22 (32)0.26Non-Hispanic: n (%)78 (91)65 (97)0.19White: n (%)77 (90)62 (91)0.79Married: n (%)51 (62)51 (76)0.07College graduate: n (%)43 (52)41 (61)0.25Baseline IQ: M (SD)107.40 (9.84)111.70 (9.69) < 0.01KPS: n (%)0.02 Less than 806 (7)13 (19) 80 or higher80 (93)56 (81)Disease response at 3 months: n (%)55 (65)50 (77)0.13Disease response at 12 months: n (%)41 (71)35 (78)0.42CRS grade: n (%)0.11 Low grade (0, 1, 2)83 (97)62 (90) High grade (3 or 4)3 (3)7 (10)Neurotoxicity grade: n (%)0.12 Low grade (0, 1, 2)72 (84)50 (74) High grade (3 or 4)14 (16)18 (26)Lines of prior therapy: n (%)0.41 3 or less61 (71)53 (77) 4 or more25 (29)16 (23)KPS Karnofsky Performance Status, CRS cytokine release syndrome, Disease response = complete response or partial response. N number of participants, M mean, SD standard deviation Table 2Participant characteristicsYounger Adults (n = 86)Older Adults (n = 69)p-valueAge: M [range]53.22 [19–64]71.56 [65–82] < 0.0001Female: n (%)35 (41)22 (32)0.26Non-Hispanic: n (%)78 (91)65 (97)0.19White: n (%)77 (90)62 (91)0.79Married: n (%)51 (62)51 (76)0.07College graduate: n (%)43 (52)41 (61)0.25Baseline IQ: M (SD)107.40 (9.84)111.70 (9.69) < 0.01KPS: n (%)0.02 Less than 806 (7)13 (19) 80 or higher80 (93)56 (81)Disease response at 3 months: n (%)55 (65)50 (77)0.13Disease response at 12 months: n (%)41 (71)35 (78)0.42CRS grade: n (%)0.11 Low grade (0, 1, 2)83 (97)62 (90) High grade (3 or 4)3 (3)7 (10)Neurotoxicity grade: n (%)0.12 Low grade (0, 1, 2)72 (84)50 (74) High grade (3 or 4)14 (16)18 (26)Lines of prior therapy: n (%)0.41 3 or less61 (71)53 (77) 4 or more25 (29)16 (23)KPS Karnofsky Performance Status, CRS cytokine release syndrome, Disease response = complete response or partial response. N number of participants, M mean, SD standard deviation Participant characteristics KPS Karnofsky Performance Status, CRS cytokine release syndrome, Disease response = complete response or partial response. N number of participants, M mean, SD standard deviation Change in neurocognitive performance over time Rates of neurocognitive impairment and performance across age groups have been described previously [4, 5]. Rates of neurocognitive impairment by age group are presented in Supplemental Table 1. Figure 1 shows overall rates of impairment with 95% confidence limits by age group over time. Before infusion, older adults were significantly more impaired than younger adults (48% vs. 34%, p < 0.05), but this group difference was not evident after infusion. Neurocognitive deficits were sporadic, and a large proportion of participants in each age group demonstrated impairment on multiple tests. Among older adults not impaired before treatment, 15% (5/34) demonstrated newly onset cognitive impairment by day 30, 32% (11/34) by day 90, and 35% (12/34) by day 360. Neurocognitive impairment persisted by day 360 for 50% (3/6) of participants who completed neuropsychological assessments at that timepoint (6/12 older adults with newly onset cognitive impairment were missing neurocognitive data at day 360, thus persistent impairment at day 360 was only evaluable for 6 participants). Among younger adults not impaired before treatment, 30% (16/54) demonstrated newly onset cognitive impairment by day 30, 41% (22/54) by day 90, and 50% (27/54) by day 360. Neurocognitive impairment persisted by day 360 for 61% (11/18) of participants who completed neuropsychological assessments at that timepoint (9/27 younger adults with newly onset cognitive impairment were missing neurocognitive data at day 360, thus persistent impairment was only evaluable for 18 participants).Fig. 1Rates of overall neurocognitive impairment over time with 95% confidence limits for younger adults (n = 86) and older adults (n = 69). Note: Numbers inside columns represent sample size for each group at each timepoint Rates of overall neurocognitive impairment over time with 95% confidence limits for younger adults (n = 86) and older adults (n = 69). Note: Numbers inside columns represent sample size for each group at each timepoint Age as a moderator of neurocognitive performance over time Figure 2 shows estimated means for TNP over time by age group, and Supplemental Table 2 shows raw means and 95% confidence limits for all domains over time by age group. Results of random effects mixed models analyzing age group (i.e., < 65 years vs. 65 + years) as a moderator of neurocognitive performance are presented in Table 3. Models including the quadratic effect of time did not reveal significant age by time interactions (p-values > 0.05), thus parameter estimates are presented for linear models only. At baseline, older participants demonstrated worse performance than younger participants on all neurocognitive domains (p-values < 0.05). There were significant age by time interactions for immediate verbal memory (p < 0.01; Fig. 3) and delayed memory (p < 0.01; Fig. 4). These interactions remained statistically significant when controlling for KPS (< 80 vs. 80 +), and the interaction between time and KPS, for both immediate verbal memory (p = 0.02) and delayed memory (p < 0.01). Follow-up within-group analyses indicated that both immediate verbal memory (p < 0.0001) and delayed memory (p < 0.01) improved for older adults over time but remained stable for younger adults (p = 0.10 and p = 0.23, respectively). The improvement in immediate verbal memory was clinically significant for older adults only. Cross-sectional comparisons revealed that older adults demonstrated significantly worse immediate verbal memory and delayed memory than younger adults prior to infusion and at 30 days post-infusion (p-values < 0.01). Older adults also demonstrated significantly worse immediate verbal memory than younger adults at 90 days post-infusion (p = 0.04). These cross-sectional differences in estimated means ranged from 2.5 to 3.8 points, reflecting medium effect sizes; none were clinically significant. By one year after infusion, there were no statistically significant differences in immediate verbal memory (p = 0.47) or delayed memory (p = 0.37) by age group.Fig. 2Raw means for Total Neuropsychological Performance (TNP) over time by age group (< 65 years vs. 65 + years). Note: Sample sizes over time for younger adults and older adults were 82, 74, 47, and 45, and 65, 40, 40, and 30, respectivelyTable 3Parameter estimates for mixed models of age group (< 65 years old vs. 65 + years old) as a moderator of neurocognitive performanceTNPAttentionExecutive FunctioningVerbal AbilityImmediate Verbal MemoryVisuospatial abilityDelayed MemoryIntercept49.78***50.72***50.46***49.53***47.44***52.23***48.99***Time0.040.32***0.160.120.15−0.57***−0.10Age group−2.95***−1.95*−4.00***−2.31*−3.82**−2.52*−3.30**Age group*Time0.170.02−0.180.010.44**0.130.41**Time indicates T-score change over one month. Models adjusted for IQFig. 3Estimated means for immediate verbal memory by age group (< 65 years vs. 65 + years), adjusted for baseline IQ. Note: Sample sizes over time for younger adults and older adults were 82, 74, 47, and 45, and 65, 40, 40, and 30, respectivelyFig. 4Estimated means for delayed memory by age group (< 65 years vs. 65 + years), adjusted for baseline IQ. Note: Sample sizes over time for younger adults and older adults were 82, 74, 44, and 45, and 64, 40, 39, and 30, respectively Raw means for Total Neuropsychological Performance (TNP) over time by age group (< 65 years vs. 65 + years). Note: Sample sizes over time for younger adults and older adults were 82, 74, 47, and 45, and 65, 40, 40, and 30, respectively Table 3Parameter estimates for mixed models of age group (< 65 years old vs. 65 + years old) as a moderator of neurocognitive performanceTNPAttentionExecutive FunctioningVerbal AbilityImmediate Verbal MemoryVisuospatial abilityDelayed MemoryIntercept49.78***50.72***50.46***49.53***47.44***52.23***48.99***Time0.040.32***0.160.120.15−0.57***−0.10Age group−2.95***−1.95*−4.00***−2.31*−3.82**−2.52*−3.30**Age group*Time0.170.02−0.180.010.44**0.130.41**Time indicates T-score change over one month. Models adjusted for IQ Parameter estimates for mixed models of age group (< 65 years old vs. 65 + years old) as a moderator of neurocognitive performance Time indicates T-score change over one month. Models adjusted for IQ Estimated means for immediate verbal memory by age group (< 65 years vs. 65 + years), adjusted for baseline IQ. Note: Sample sizes over time for younger adults and older adults were 82, 74, 47, and 45, and 65, 40, 40, and 30, respectively Estimated means for delayed memory by age group (< 65 years vs. 65 + years), adjusted for baseline IQ. Note: Sample sizes over time for younger adults and older adults were 82, 74, 44, and 45, and 64, 40, 39, and 30, respectively Clinical risk factors for worse neurocognitive performance by age There were significant interactions between lines of prior therapy and age group on change in TNP (p = 0.03) and verbal ability (p = 0.02), driven by improvements in cognition among older adults who received 4 + prior lines of therapy (Supplemental Figs. 1 and 2). By 12 months post-infusion, older adults who received 4 + prior lines of therapy demonstrated clinically meaningful improvements in TNP and verbal ability compared to younger adults with the same treatment history at 12 months post-infusion. There was also a significant interaction between age group and disease status at 3 months on change in visuospatial ability (p < 0.01) (Supplemental Fig. 3), driven by rapid deterioration in cognition among older adults with active disease (i.e., those who did not respond to treatment) at 3 months. Older adults with active disease demonstrated significantly worse visuospatial abilities than younger adults with active disease at all post-infusion timepoints, with clinically meaningful differences observed at 3 and 12 months. However, these interactions should be interpreted with caution, as few older adults with 4 + prior lines of therapy or older adults with active disease at 3 months provided neurocognitive data after infusion; by 12 months post-infusion, only four older adults in each subgroup provided neurocognitive data. Discussion In this study of LBCL patients treated with axi-cel, older adults were more likely than younger adults to demonstrate neurocognitive impairment before infusion. This age difference dissipated by one year after infusion. Contrary to our hypothesis, younger adults were particularly susceptible to post-infusion deficits in immediate verbal memory and delayed memory. Further, clinical variables such as fewer lines of prior therapy and active disease after infusion were associated with cognitive dysfunction in older adults. Practice effects (i.e., improvement over time) of neurocognitive testing are expected in cognitively normal individuals due to participants developing greater comfort with the setting and test material, as well as improving their understanding of how to respond to questions by the test administrator [18]. A lack of practice effects, particularly between the first and second test administrations, connotes cognitive abnormality. In this study, immediate verbal memory and delayed memory improved for older adults but remained unchanged for younger adults. This lack of improvement for younger adults suggests cognitive abnormality not explained by clinical factors (e.g., disease status, CRS, neurotoxicity, or prior lines of therapy). Further, across all neurocognitive domains, half of younger adults with normal cognitive scores pre-infusion later developed neurocognitive impairment (compared to 1/3 of older adults). These findings are consistent with a systematic review and meta-analysis of neurocognitive performance after hematopoietic cell transplant (HCT) [19], another cellular therapy for hematologic malignancies, that found a lack of expected practice effects over time (median age = 43). However, the age-related findings in this study are novel, as little is known about age-related differences in neurocognitive trajectories after treatment. In studies of symptomatology in cancer patients, young adults self-report higher rates of cancer/treatment-related toxicities (e.g., nausea, vomiting, anxiety, depression) than adults aged 40 and older [20, 21]. Among individuals without cancer, there is also evidence that verbal working memory peaks in young adulthood[22] (i.e., around age 30), suggesting a window of vulnerability for cognitive domains like verbal memory among younger adults as compared to older adults. In contrast, previous work from our research group on the effects of age on neurocognitive performance in patients treated with HCT indicates that older age is associated with worsening verbal memory and fluency after transplantation [23]. Considering the dearth of literature on age and neurocognitive performance in patients treated with CAR T-cell therapy, further research is needed to clarify the nature of the association between age and cognitive abilities in this patient population. In this study, we explored the effects of various clinical variables on cognitive outcomes. The results indicated that neurocognitive performance improved the most in older adults previously treated with 4 + lines of therapy, as compared to all other groups (i.e., younger adults treated 4 + previous lines of therapy, and both older and younger adults treated with fewer lines of therapy). This finding merits additional discussion, as our prior work has shown that patients of any age treated with 4 + prior lines of therapy demonstrated worse TNP than patients treated with 3 or fewer prior lines of therapy [4]. Results from these pooled studies suggest that the “multiple hit” hypothesis[24, 25] as it pertains to cancer treatment (in which repeated treatments lead to increasing cognitive impairment), may not hold true for older adults treated with CAR T-cell therapy. However, these results should be interpreted with caution given the small cell sizes after infusion for older adults previously treated with 4 + lines of therapy and with active disease at 3 months that limit statistical power. Reductions in visuospatial abilities in the first year after axi-cel were evident in both age groups, consistent with studies of patients with other cancer types that show declining visuospatial abilities over time [26, 27]. In our exploratory analyses, reductions in visuospatial abilities were particularly evident in older adults who did not respond to treatment and still had active disease at the 3 month follow-up. Despite small numbers of patients in clinical subgroups, it is possible that, in this group, visuospatial decline was influenced by emotional factors such as stress or disappointment due to the lack of treatment efficacy. However, as indicated above, few older adults with active disease at the 3 month follow-up provided neurocognitive data after infusion, so conclusions about this subgroup may not be reliable. Few studies have examined factors related to changes in cognition after CAR T-cell therapy (beyond just axi-cel), and even fewer have examined effects of clinical variables on neurocognition [28], thus additional studies are needed to determine whether these findings are also observed in other samples and patient populations. This study had several strengths, including a longitudinal study design that assessed neurocognitive performance through the first year after infusion, a focus on an understudied patient population of older adults treated with CAR T-cell therapy, the use of validated neurocognitive measures, and the use of advanced statistical techniques that account for missing data over time. Limitations include a sample predominantly comprised of male, White, non-Hispanic, married, college graduates which limits generalizability of findings. Further, clinical outcomes can vary between patients treated with axi-cel commercially vs. on a clinical trial [29], and our sample size precluded comparisons of neurocognitive performance based on axi-cel administration. In addition, over 1/3 of patients with baseline data did not provide data at one year (due to morbidity and/or mortality) [4, 5], but study attrition is not known to be associated with neurocognitive outcomes in patients treated with HCT[23, 30] or CAR T-cell therapy. In conclusion, few studies have explored the relationship between age and objective neurocognitive performance in patients treated with axi-cel. Results from this study suggest that older patients demonstrate worse neurocognitive functioning before infusion, but younger patients may be more susceptible to memory deficits after infusion. Future research on cognitive abilities, including related clinical factors, in larger samples of older adults treated with CAR T-cell therapy is warranted. In addition, research is needed to examine neurocognitive performance (e.g., memory) in younger patients with axi-cel and other cellular therapies. Improved understanding of the neurocognitive effects of CAR T-cell therapy by age, along with clinical benefits, would help inform patients of the possible side effects and/or benefits of this treatment. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 43 KB) Supplementary file1 (DOCX 43 KB)