Do PROs Tell the Whole Story? Differential Outcomes Based on Patient-Reported Outcomes (PROs) Versus Performance-Based Metrics (PBM) on Cognition for Patients Receiving Chimeric Antigen Receptor (CAR)-T Cell Therapy

Introduction: As the use of CAR-T cell therapy grows, there is an increased need to understand its impact on the patient experience, especially symptom burden and cognitive function. While the immediate side-effects of CAR-T therapy have been reported, our study aims to describe the longitudinal impact of CAR-T therapy on patients' quality of life (QoL), including patient-reported cognitive function and performance-based cognition, which are not well understood.

Methods: Patients with hematologic malignancies undergoing CAR-T therapy were prospectively recruited from two academic centers. The primary endpoint was feasibility of completing longitudinal PRO assessments and PBM of cognition. NIH PROMIS measures assessed physical, mental, cognitive, and social health. PROMIS measures use the t-score metric, where 50 is the average in the U.S. population and a 5 point (0.5 SD) change was considered clinically meaningful. The NIH Toolbox Cognition Battery measured 6 constructs of cognition, scored on the t-score metric (10 point = 1SD, change considered clinically meaningful). Exploratory analyses described change from baseline. PROMIS measures were completed at baseline, 7 and 14 days, 1, 3, 6, and 12 months (mo) after CAR-T. The Toolbox was assessed at baseline, 1 month, and 12 months. Due to COVID restrictions on in person research, the Toolbox could not be assessed for the first 13 patients.

Results: From 8/2020 to 6/2021, 28 patients have been enrolled. Baseline, day 7, day 14, 1 mo, 3 mo, and 6 mo data were available in 27, 20, 21, 23, 15 (10 not yet reached), and 9 (16 not yet reached) patients, respectively. The mean age was 57 years (range 27-78); 44% were female. Race distribution was: Caucasian 75%, Asian 8%, Hawaiian/Pacific Islander 4% and other race 8%; 21% were Hispanic ethnicity. Patients received CAR-T for diagnoses of NHL (75%), MM (17%), and ALL (13%). CRS was seen in 86% (all grade 1-2), neurotoxicity (ICANS) in 34% (grade 1-2: N=5 and grade > 3: N=5).

PROMIS questionnaires were completed in >70% of patients across all timepoints with current follow-up; thus it was feasible to collect these data at frequent intervals after CAR-T. Mean baseline PROMIS t-scores (N=27) were similar to the average US population in all domains (fatigue: 53, sleep: 52, pain: 52, anxiety: 53, depression: 49) except for decreased physical function (44) among patients (Fig 1a-b). Physical function, fatigue, and pain interference worsened during the first month but returned to baseline by month 3 (Fig 1a-b).

PBM of cognition (NIH Toolbox) were assessed at baseline in 15 pts and 1 mo in 8 patients (4 incomplete, 3 not reached timepoint). The toolbox requires in-person administration and takes 35 minutes, but has been completed in 75% of evaluable patients. At baseline, the mean total composite score was 65 ^th percentile and t-score was 57; mean fluid composite score was 50 ^th percentile and t-score was 50; mean crystallized composite score was 69 ^th percentile and t-score was 58 (fluid composite score measures ability to reason, crystallized composite score measures accrual of knowledge over time, Weintraub et al Neurology 2013). Little change in scores was seen in language domains and some increase (not clinically significant) was seen in constructs on attention, executive function, and episodic memory. While not significant, a trend towards worsening working memory and processing speed and a trend towards worsening t-scores for all composite scores was seen (Figure 1c). 2 patients with neurotoxicity grade 3 and available baseline and 1-mo Toolboxes were noted to have decreases in all composite scores (clinically significant in 1). Patients did not self-report changes in cognitive function over 6 months (Fig 1d).

Conclusion:

This study reports early data from longitudinal neurocognitive assessments and PROs in patients undergoing CAR-T. It is feasible for patients undergoing CAR-T to complete PROMIS surveys (PROs) and NIH cognitive Toolboxes (performance-based test). Early and frequent PRO surveys captured initial worsening in physical function, fatigue, and pain interference that returned to baseline by month 3. There was no change in patient-reported cognitive function over time, but using PBM cognition testing, we noted a trend towards worsening cognition in some domains. Continued patient accrual and longer follow up will allow assessment of degree and persistence of worsened PBM cognition associated with CAR-T.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal