Change in patients’ perceived cognition following chimeric antigen receptor T-cell therapy for lymphoma

Abstract

Background:

Chimeric antigen receptor (CAR) T-cell therapy can lead to durable responses in patients with relapsed/refractory hematologic malignancies. Immune effector cell-associated neurotoxicity syndrome (ICANS) and cytokine release syndrome (CRS) are common and may place patients at risk for longer-term cognitive impairment.

Objective:

This study examined changes in cognition in the first year after CD19-directed CAR-T-cell therapy for lymphoma, as well as CAR T-cell therapy-specific risk-factors (e.g., ICANS, CRS) and non-specific risk factors (e.g., baseline quality of life, frailty) for worsening cognition.

Study design:

Patients’ perceived cognition was assessed at baseline and at days 90 and 360. Clinical variables were abstracted from medical records. Piecewise mixed models were used to examine acute change (i.e., within 90 days) and longer-term change (i.e., from 90 days to 360 days) in cognition as well as to explore risk factors for worsening cognition.

Results:

Among 118 participants (mean age 61, 59% male), mean levels of perceived cognition did not change from baseline to day 90 (p values>0.05) but worsened from day 90 to day 360 in global cognition and in the domains of memory, language, organization, and divided attention (p values<0.05). Although statistically significant, changes were small (d values 0.15-0.28). Greater baseline fatigue, anxiety, and depression were associated with worse global cognition at day 90 (p values<0.01). Patients with more severe ICANS post-CART reported worse global cognition at day 360 (p<0.05), although there were no differences in perceived cognition by severity of CRS (p values>0.05). Other putative risk factors were not associated with acute or longer-term changes in perceived cognition (p values>0.05).

Conclusions:

CAR T-cell therapy recipients reported delayed deterioration in several cognitive domains, although changes were small. These findings may be useful when educating future patients on what to expect when receiving CAR T-cell therapy.

Chimeric antigen receptor (CAR) T-cell therapy (e.g., axicabtagene ciloleucel, tisagenlecleucel, lisocabtagene maraleucel) can induce durable disease response in approximately 40% of patients with relapsed or refractory diffuse large B cell lymphoma.^1-4 However, up to 50% of patients experience severe side effects in the first 60 days, such as immune effector cell-associated neurotoxicity syndrome (ICANS) or cytokine release syndrome (CRS).^5-7 ICANS is often characterized by encephalopathy, delirium, diminished attention, language disturbance, and impaired handwriting. Confusion, disorientation, agitation, aphasia, somnolence, and tremors may also be present.^{5, 7} CRS results when cancer cells are recognized by CAR T-cells, causing a “cytokine storm” of pro-inflammatory signaling molecules that may lead to fever, hypotension, and hypoxia.^{3, 8} The neurotoxic and inflammatory effects of CAR T-cell therapy raise concerns that patients may be at risk for worse cognitive outcomes. To date, only one study to our knowledge has prospectively examined cognitive outcomes among CAR T-cell therapy recipients. In a cross-sectional study of 40 patients 1-5 years post-CAR T-cell therapy, Ruark and colleagues⁹ found that 38% reported cognitive concerns. Longitudinal studies of the cognitive effects of CAR T cell therapy are needed,^{10, 11} including those focused on patients’ own perceptions of their cognition, which are strongly associated with well-being.¹²

Cancer-related cognitive impairment (CRCI) is common among cancer survivors who received treatments other than CAR T-cell therapy, with 50-75% of patients treated with hematologic or solid malignancies reporting cognitive concerns at some point after cancer diagnosis and treatment (e.g., chemotherapy, radiation, hormonal therapy).^13-15 Objective neuropsychological testing indicates that patients tend to perform worse over time than similar non-cancer controls,^16-18 with up to 36% of patients demonstrating impairment in one or more cognitive domain.¹⁹ Highly relevant to CAR T-cell therapy, a sizable literature suggests that circulating markers of inflammation are associated with worse perceived and objective cognition.^20-24 In addition, inpatient delirium during other adoptive cellular therapies, such as hematopoietic cell transplant (HCT), is a risk factor for worse perceived and objective cognition among hematologic cancer patients.^{25, 26} Additional risk factors for cognitive concerns among cancer survivors are demographic (e.g., older age, less education) and clinical characteristics (e.g., worse frailty) as well as poor quality of life.^{15, 18, 27, 28} CRCI is highly relevant to this new and growing population of cancer survivors treated with CAR T-cell therapy, as it is associated with reduced ability to return to work,²⁹ confidence in the ability to manage treatment-related symptoms³⁰ and quality of life.³¹

The goals of the current study were twofold. First, we sought to quantify changes in cognition from prior to CAR T-cell therapy through 360 days post-treatment in patients diagnosed with non-Hodgkin lymphoma. Second, we sought to explore associations between CAR T-cell therapy-specific risk factors (e.g., ICANS, CRS, response to treatment), non-specific baseline risk factors (e.g., age, education, frailty, quality of life) and changes in cognition at day 90 and 360 post-treatment. It was hypothesized that perceived cognition would worsen over time and be associated with CAR T-cell specific toxicity and response, sociodemographic variables, and baseline quality of life. Because perceived cognition after CAR T-cell therapy could change for a complex variety of reasons, we did not attempt attribute changes to particular causal factors. Instead, the study was intended to provide information to help educate patients and family members about what to expect after CAR T-cell therapy and identify potential supportive care needs.

Materials and Methods

Participants

Participants were recruited prospectively as part of a larger study assessing cognitive and patient-reported outcomes in adults with hematologic malignancies scheduled to receive CAR T-cell therapy. Eligible patients were: 1) ≥18 years old, 2) diagnosed with hematologic cancer, 3) scheduled to receive CAR T-cell therapy as part of standard care or a clinical trial at Moffitt Cancer Center, 4) able to speak and read standard English, 5) without documented or observable psychiatric or neurological disease that could interfere with study participation, 6) without a history of stroke or head trauma with loss of consciousness within the past 5 years, and 7) able to provide informed consent.

Procedure

Data were collected between October 2016 and February 2021. Patients were identified in a HIPAA-compliant manner in consultation with the treating medical oncologist and through medical chart review. Patients were approached, provided informed consent, and completed a baseline assessment prior to receiving conditioning chemotherapy before CAR T-cell therapy. Follow-up assessments for perceived cognition occurred at 90 and 360 days post-treatment, consistent with previous studies of cognition in patients with hematologic cancer treated with HCT.^{32, 33} The study was approved by the Advarra Institutional Review Board.

Measures

Participants provided sociodemographic information at baseline (i.e., age, sex, race, ethnicity, marital status, education). Clinical data were abstracted from medical records (i.e., cancer diagnosis, previous lines of treatment, baseline Karnofsky Performance Status [KPS], baseline Hematopoietic Cell Transplantation Comorbidity Index (HCT-CI),³⁴ CAR T-cell therapy agent, CAR T-cell therapy administration (clinical trial vs standard of care), highest grade of CRS, highest grade of ICANS, admission to the intensive care unit, duration of hospitalization in the first 100 days, disease response at days 90 and 360). ICANS and CRS were graded by the provider on a 5-point scale with higher scores indicating worse toxicity. Neurotoxicity grading was first conducted in accordance with the CAR-T cell-therapy associated TOXicity (CARTOX) classification system,⁸ and then with the American Society of Transplant and Cellular Therapy (ASTCT) Consensus Grading for Cytokine Release Syndrome (CRS) and Neurologic Toxicity Associated with Immune Effector Cells grading systems, based on updated practice guidelines published in 2018.⁷ Frailty was assessed prospectively by a trained research assistant or doctoral-level psychology graduate student through hand grip strength; participants were asked to squeeze a standard hand dynamometer twice using each hand. Grip strength data were normed by age, gender, race and education.³⁵ Participants scoring 1 standard deviation below the normative mean on either hand were considered frail.³²

Current perceived cognition compared to ten years ago was assessed with the 40-item Everyday Cognition Questionnaire (ECog)³⁶ at baseline and days 90 and 360. ECog items are scored on a 5-point scale (1=better or no change, 4=consistently much worse, 5=don’t know); scores of 5 were recoded to missing prior to statistical analysis. The ECog yields a global perceived cognition score and six scores for the following cognitive domains: memory (e.g., remembering items on a shopping list), language (e.g., wordfinding), visuospatial abilities (e.g., reading a map), planning (e.g., thinking ahead), organization (e.g., managing medication), and divided attention (e.g., returning to a task after being interrupted). Satisfaction with cognition was evaluated with one question: “At the present time, how satisfied are you with your thinking abilities (for example, memory and attention)?” (1=completely dissatisfied, 5=completely satisfied). Scores were recoded as necessary so that higher scores indicated better cognition or satisfaction. Clinically meaningful change was defined as one-half standard deviation change.³⁷

Patient-reported quality of life at baseline was initially assessed with the 36-item Medical Outcomes Study-Short Form-36 (SF-36).³⁸ Following the patient-reported outcomes recommendations included in Centers for Medicare and Medicaid Service coverage decision for CAR T-cell therapy in February 2019,³⁹ the SF-36 was replaced by the Patient-Reported Outcomes Measurement Information System-29 Profile v2.1 (PROMIS®-29).⁴⁰ Using the established and well-validated PROsetta Stone® data harmonization methodology,⁴¹ SF-36 scores were converted to PROMIS®-29 T-scores. The PROsetta Stone® conversion yields five outcomes: physical functioning, pain interference, fatigue, anxiety symptoms, and depression symptoms. Scores range from 0 to 100 with higher scores indicating more of the construct being measured.

Statistical Analysis

Participants were included in the analysis if they provided perceived cognition data at any time point. To limit sample heterogeneity, participants diagnosed with a cancer other than non-Hodgkin’s lymphoma and those treated with allogeneic CAR T-cell therapy were not included in analyses. Demographic and clinical characteristics were described using means, standard deviations, frequencies, and percentages. A composite score for frailty was created by including participants who were fragile on either the dominant or non-dominant hand. Percentages of frail participants at baseline as well as those reporting clinically significant worsening, clinically significant improvement, and unchanged cognition over time were calculated. Piecewise mixed models were used to examine changes in global cognition z-scores and cognitive domains. These models included two distinct time periods to examine acute (baseline to day 90) and longer-term (day 90 to day 360) effects.⁴² Cohen’s d effect sizes examining the magnitude of change in cognition were calculated; results were interpreted as follows: small (0.2-0.5), medium (0.5-0.8), or large (>0.8).⁴³ Spearman correlations were used to explore concurrent relationships among global cognition and putative baseline risk factors. Piecewise mixed models were used to explore potential risk factors for change in global cognition z-scores over time. Demographic risk factors (i.e., age, education), clinical risk factors (i.e., previous lines of treatment≥3, grade≥2 ICANS, grade≥2 CRS, disease response at day 90 and at 360, baseline frailty), and baseline quality of life (i.e., fatigue, physical function, pain, depression, anxiety) were explored. Alpha was set at p<0.05. Statistical tests were 2-sided. Analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC).

Results

Participants

A total of 163 participants signed consent. In total, 118 received CAR T-cell therapy for NHL, provided perceived cognitive data at any time point, and were included in the current analyses (see Figure 1). Participant demographic and clinical characteristics are described in Table 1. On average, participants were 61 years of age and predominantly male, White, non-Hispanic, and college graduates. Most participants were treated with axicabtagene ciloleucel (85%).

Figure 1.

Diagram showing participant flow through study

Table 1.

Participant characteristics (N=118)

Age: Mean (SD)	61 (12)
Sex: n (%) male	70 (59)
Race: n (%) white	105 (89)
Ethnicity: n (%) non-Hispanic	110 (94)
Marital status: n (%) married	83 (72)
Education: n (%) college graduate	61 (53)
Diagnosis1
Large B cell lymphomas	102 (86)
Other diagnosis	16 (14)
CAR T-cell therapy agent: n (%)
Axicabtagene ciloleucel	101 (85)
Tisagenlecleucel	15 (13)
Brexucabtagene autoleucel	2 (2)
CAR-T cell therapy administration:
Clinical trial	50 (42)
Standard of care	68 (58)
Previous lines of treatment: n (%) ≥3	74 (63)
Baseline KPS: n (%) ≥80	107 (91)
Baseline HCT-CI: n (%) ≥3	48 (41)
Baseline frailty: n (%)	20 (43%)
Maximum CRS2: n (%) grade ≥2	51 (43)
Maximum ICANS2: n (%) grade ≥2	26 (22)
ICU admission: n (%) yes	16 (14)
Disease response3 at day 90: n (%) yes	83 (74)
Disease response3 at day 360: n (%) yes	47 (67)
Inpatient days until day 100: Mean (SD)	16 (10)

Note:

¹Large B-cell lymphomas include diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma and transformed follicular lymphoma; Other diagnosis include follicular lymphoma, mantle cell lymphoma.

²CRS and ICANS were dichotomized at grade 2 to make groups fairly equal.

³Disease response is defined as complete and partial responses, KPS: Karnofsky Performance Status, HCT-CI: Hematopoietic Cell Transplant-Comorbidity Index, CRS: cytokine release, ICANS: immune effector cell-associated neurotoxicity syndrome, ICU: intensive care unit.

Regarding changes in perceived cognition over time, results of the piecewise mixed models are presented in Table 2. From baseline to day 90, there were no statistically significant mean-level changes in global cognition or in any cognitive domain (p values>0.05). In contrast, from day 90 to 360, participants reported worsening mean-level global cognition (p=0.01 d=0.18), memory (p=0.04 d=0.15), language (p=0.04 d=0.15), organization (p=0.03, d=0.17), and divided attention (p=0.001 d=0.28). Figure 2 displays clinically percent of participants reporting significant change in cognition. At day 90, 12% reported clinically significant worsening relative to baseline, 10% reported clinically significant improvement, and 78% reported no change. At day 360, 25% reported clinically significant worsening in global cognition relative to baseline, 9% reported clinically significant improvement, and 66% reported no change.

Table 2.

Parameter estimates for piecewise mixed models examining perceived cognition over time

	Global	Memory	Language	Visuospatial Abilities	Planning	Organization	Divided Attention	Satisfaction
Intercept	−0.02	−0.03	−0.02	0.11	−0.01	−0.08	0.08	−0.11
Baseline today 90 change	0.00	−0.02	−0.01	0.12	−0.01	−0.05	0.08	−0.10
Day 90 to day 360 change	−0.18*	−0.15*	−0.15*	−0.12	−0.15	−0.17*	−0.28**	−0.12

Note: Change estimates indicate z score change. Positive indicates improvement, negative indicates worsening.

^*p<.05

^**p<.01.

Figure 2.

Percentage of patients reporting worsened, improved and unchanged cognition at day 90 and 360 relative to prior to CAR T-cell therapy

Regarding risk factors for worse global cognition prior to CAR T-cell therapy, worse baseline physical functioning (p<0.01), fatigue (p<0.0001), anxiety (p<0.0001), and depression (p<0.0001) were significant risk factors for worse baseline cognition; number of previous lines of treatment was not associated with baseline cognition (p=0.97). Results of the piecewise mixed models examining potential demographic and clinical risk factors for worsening global cognition are presented in Table 3. Greater baseline fatigue (p<0.01), anxiety (p<0.01), and depression (p<0.01) were significantly associated with worse global cognition at day 90 but not at day 360 (p values>0.30). No variables were associated with changes in global cognition from baseline to day 90 or from day 90 to day 360 (p values>0.05).

Table 3.

Associations of Demographic and Clinical Variables with Change in Cognition

	Age	College graduate	Lines previous treatment ≥3	Frailty	ICANS grade ≥2	CRS grade ≥2	Disease response at day 90	Disease response at day 360	Physical Function	Pain	Fatigue	Anxiety	Depression
Intercept	−0.22	−0.17	0.03	0.16	−0.02	−0.09	−0.06	0.01	−0.64	0.10	1.46	1.90***	2.07***
Baseline to day 90	−0.22	0.04	0.05	0.02	−0.02	0.03	−0.05	−0.09	0.71	−0.59	−0.62	−0.42	−0.26
Day 90 to day 360	−0.20	−0.05	−0.02	0.00	−0.05	−0.04	−0.16**	−0.10*	−0.20	−0.09	0.06	−0.02	−0.03
Risk factor	0.00	0.29	−0.07	0.19	0.02	0.18	0.04	−0.17	0.01	−0.00	−0.03**	−0.04***	−0.04***
Baseline to day 90* risk factor	0.00	−0.07	−0.08	−0.07	0.12	−0.06	0.10	0.17	−0.02	0.01	0.01	0.01	0.01
Day 90 to day 360 *risk factor	0.00	−0.01	−0.07	−0.02	−0.08	−0.04	0.11	0.07	0.00	0.00	−0.00	−0.00	−0.00

Note:

^*p<.05

^**p<.01

Post-hoc analyses explored mean differences at each time point by ICANS, CRS, and disease response, which were not captured in piecewise mixed models. Results indicated that patients with higher ICANS (i.e., Grade 2 or above) reported significantly worse cognition at day 360 relative to those with low ICANS (i.e., Grade 0 or 1) (p<0.05, d=0.20) (see Figure 3). There were no differences in cognition by CRS at any time point or disease status (response versus progression) at day 360 (p values>0.05).

Figure 3.

Change in global cognition by ICANS grade. Patients with ICANS grade≥ 2 reported worse cognition at day 360 relative to patients with ICANS grade <2

Discussion

This study prospectively examined acute and longer-term changes in cognition among CAR T-cell therapy recipients from prior to treatment to day 360. Results indicate no changes in perceived cognition on average in the acute phase of treatment (i.e., within 90 days). This finding may be because there are other medical issues during the first 90 days,²¹ which may limit patient insight regarding cognition. In contrast, patients reported significant worsening on average in several cognitive domains beyond the acute treatment period (i.e., from 90 days to 360 days). These domains included global cognition, memory, language, organization, and divided attention. Changes were small (0.15 to 0.28 SD) and were on average below the generally-accepted criterion of 0.5 SD indicating clinically meaningful differences in patient-reported outcomes.³⁷ In comparison, Janelsins and colleagues¹⁶ reported large deteriorations of .8 SD in perceived cognition in lymphoma patients from before to after their first course of chemotherapy. Patients in the current study were heavily pretreated, with 63% having received at least three lines of therapy before CAR T-cell therapy. Lines of previous treatment was not associated with cognitive outcomes in this study. Taken together with the Janelsins study,¹⁶ these data suggest that the greatest deterioration in perceived cognition may occur during first-line therapy.

Exploratory analyses of risk factors for worsening perceived cognition indicated that none were associated with mean differences in change across time (i.e., age, education, previous lines of treatment, ICANS, CRS, disease response at day 90 and at 360, and baseline frailty and quality of life). However, baseline fatigue, anxiety, and depression were associated with significantly worse global perceived cognition at baseline and day 90. These findings are consistent with a large body of literature suggesting that perceived cognition is associated with fatigue and distress in cancer patients,^{28, 44} including those treated with adoptive cellular therapies such as HCT.^{31, 45} Some data suggest that improving these outcomes can also improve perceived cognition in cancer patients.⁴⁶ Randomized controlled trials have shown the potential of physical activity and cognitive behavioral interventions to improve fatigue, depression, and perceived cognition,^{15, 46-48} suggesting these interventions may prevent or mitigate cognitive concerns among cancer patients.

Due to the hypothesized relationship between ICANS, CRS, disease response, and perceived cognition, we conducted post-hoc analyses to explore whether there were group differences in means across time that would not have been captured with piecewise regression. We found that patients with ICANS Grade 2 or above reported worse cognition at all time points than those with ICANS Grade 0 or 1; this difference reached statistical significance at day 360 but was small in magnitude (d=0.20). Interestingly, the same pattern of results was not observed for CRS, indicating that ICANS may have a unique relationship with perceived cognition. Larger studies are needed to confirm these findings.

The current study is characterized by several strengths. It is among the first to prospectively assess changes in perceived cognition through day 360 by means of standardized measures. Results highlight the importance of the patient’s perspective of treatment, since data on CAR T-cell therapy late effects is limited and mainly come from clinician-rated adverse events. Findings are also clinically relevant considering the recommendation to prioritize perceived cognition over objective cognition as primary indicator of cognitive impairment in patients with cancer due to their negative influence on patients’ well-being.¹² The use of advanced statistical analyses, such as mixed models, allowed us to use all available data. We have previously shown that attrition causes minimal bias in patient-reported and neurocognitive estimates in patients receiving HCT and CAR T-cell therapy.^{49, 50} Study limitations should also be noted. Because CRS and ICANS generally only occur during the first 30 days,⁵¹ a first post-treatment assessment at 90 days may have missed the impact of these side effects on cognition. Similarly, the low number of participants admitted at the ICU (n=16) precluded to examine whether ICU admission was associated with perceived cognition over time. The study began before the first FDA approval of CAR T-cell therapy for adults, and some study procedures were modified as new recommendations were published (e.g., ICANS grading, quality of life assessment). In addition, CRS and ICANS treatment guidelines changed throughout the study. Since we can’t rule out that this change could have affected the study outcomes, future studies should include patients homogenously treated for CRS and ICANS. In addition, the study was conducted at a single institution at the United States and mainly include white, non-Hispanic patients with good performance status, treated with axicabtagene ciloleucel who did not develop grade≥3 ICANS or CRS, which limits the generalizability of the findings to other populations and CAR T-cell therapy recipients. Future prospective studies should also examine the association between baseline risk factors for toxicity (e.g., ferritin levels and LDH) with perceived cognition, as well as include larger, heterogeneous samples sizes to examine cognition in the first year after CAR T-cell therapy and beyond.

In summary, the current study suggests that perceived cognition worsens beyond the acute phase of CAR T-cell therapy in several cognitive domains, with patients with greater baseline fatigue, anxiety, and depression at risk for worse perceived cognition at day 90 and patients with higher ICANS at risk for worse perceived cognition at day 360. However, these changes were small in magnitude. If confirmed, these findings could be incorporated when educating future patients on what to expect when receiving CAR T-cell therapy. Further prospective studies should examine perceived cognition and objective neurocognitive performance in the first year after CAR T-cell therapy and beyond.

Highlights.

• Cognition did not change from baseline to day 90 but worsened from day 90 to 360.

• Domains that worsened were memory, language, organization and divided attention.

• Patients with more severe neurotoxicity reported worse cognition at day 360.

• There were no differences in cognition by severity of cytokine release syndrome.

Data availability:

Deidentified data upon which the manuscript is based are available upon request to the corresponding author.