Longitudinal Patient Reported Outcomes with CAR-T Cell Therapy vs Autologous and Allogeneic Stem Cell Transplant

Abstract

Background:

There is limited data on patient experience after chimeric antigen receptor (CAR) T cell therapy, especially in comparison to autologous and allogeneic transplant, which are more established forms of cellular therapy.

Objective:

We prospectively evaluated longitudinal patient-reported quality of life (QoL), symptom burden and cognition after CAR-T cell therapy and compared it with prospective cohorts of patients undergoing autologous stem cell transplant (autoSCT) and allogeneic SCT (alloSCT).

Study Design:

This was a single center study. The primary endpoint was change in QoL. Secondary endpoints were patient-reported adverse events (PRO-AEs) measured by PRO-CTCAE and cognitive function (NeuroQOLv2 questionnaire). Time profile of PRO-AEs was evaluated using longitudinal analysis Toxicity over Time (ToxT). Patients completed questionnaires at baseline, week 2 and monthly for 6 months.

Results:

104 patients were evaluable (CAR-T: 34, autoSCT: 33, alloSCT: 37). Baseline QoL was similar across groups. We observed a short-term decline in QoL in all groups that gradually returned to baseline. The nadir in QoL was at week 2 and coincided with peak in symptom burden. The decline in overall QoL, physical and functional well-being was significantly less with CAR-T vs. SCT groups and returned to baseline faster. Patients in the alloSCT group experienced the greatest symptom burden, greater decrease in performance status, largest short-term decline in QoL and slowest recovery.

Conclusion:

This study provides comprehensive data comparing QoL, PRO-AEs and cognition in CAR-T cell therapy vs. autoSCT and alloSCT, and the first application of ToxT to PRO-CTCAE data. Short-term QOL, including physical and functional domains was better in the CAR-T group vs. SCT groups, although all groups experienced an initial decline coinciding with peak symptoms. These data can serve as a guide for patient education, symptom management, and future studies in CAR-T cell therapy.

INTRODUCTION

Chimeric antigen receptor-T (CAR-T) cell therapy has resulted in deep and often durable responses in patients with relapsed-refractory hematologic malignancies, including acute lymphoblastic leukemia (ALL), non-Hodgkin’s lymphoma (NHL), and multiple myeloma (MM).^1–3 However, CAR-T cell therapy is associated with unique and serious adverse events (AEs) including cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS).⁴ Other complications, such as long-term cytopenias and high-risk of infection, are also being increasingly recognized.⁵ As this is a relatively new therapy with limited follow-up of patients, side effects with later onset are yet to be determined.

High grade clinically-observable AEs such as CRS, ICANS, etc. have been documented in clinical trials and real world practice, but lower-grade symptomatic AEs, cognitive effects, and the impact of AEs on quality of life (QoL) has not been explored in-depth. Incorporating patient reported outcomes (PROs) and QoL assessment at regular and frequent intervals throughout treatment and then comparing them to established approaches of cellular therapy (i.e., stem cell transplantation, SCT) is the best way to comprehensively measure the toxicity associated with CAR-T therapy. Although AEs reported in clinical trials provide an overview of patient symptoms, it is important to measure symptoms directly reported by patients, as prior studies in patients undergoing chemotherapy and stem cell transplant (SCT) have shown that AE reporting by patients demonstrates greater symptom burden than when AEs are reported by health care providers.^6–8 Moreover, conventional methods of toxicity analysis narrowly focus on maximum grade AEs, failing to describe the time course of AEs or capture chronic low grade AEs that may be burdensome to patients over time. It is also not known whether there is any short or long-term impact of neurotoxicity on cognitive function of patients following CAR-T therapy.

The overall goal of our study was to gain a comprehensive understanding of the patient experience after CAR-T therapy by evaluating QoL and symptom burden over six months in patients receiving CAR-T therapy and comparing them to contemporary cohorts of patients undergoing autologous and allogeneic SCT, which are well established forms of cellular therapy for hematologic malignancies.

METHODS

This was a prospective, longitudinal, observational single-center study. Adult patients with hematologic malignancies at Mayo Clinic, Rochester, MN were recruited in three cohorts depending on treatment: (1) CAR-T therapy, (2) autologous SCT and (3) allogeneic SCT. Patients could be receiving both standard of care or investigational CAR-T cell therapy. This study was approved by the institutional review board. The primary endpoint was change in QoL from baseline, as measured by the FACT-G (Functional Assessment of Cancer Therapy- General) questionnaire. Secondary endpoints were patient-reported AEs (PRO-AEs) using items from the PRO-CTCAE (Patient Reported Outcome Common Terminology Criteria for Adverse Events) and assessment of cognition/memory using the NeuroQoLv2 questionnaire.

Study Procedures:

Adult patients with hematologic malignancies undergoing CAR-T therapy, autologous or allogeneic transplant who were willing to provide consent and able to complete the questionnaire in English themselves or with assistance were eligible. There were no exclusion criteria. Potential patients were prospectively identified and offered participation in this study at the time of a clinical visit prior to the start of CAR-T therapy or transplant related treatment. After obtaining informed consent, patients were given a self-administered web-based questionnaire at baseline, 2 weeks after treatment (defined as 2 weeks after cell infusion for CAR-T therapy / transplant) and then monthly for six months. Links to monthly self-administered questionnaires were either emailed to patients or completed on an iPad during an in-person follow up visit depending on each patient’s clinical appointments. Baseline assessment could be completed any time prior to start of cellular therapy. There was no pre-specified protocol mandated requirement for repeating baseline assessment if there was a delay in cellular therapy. Criteria for early study exit included disease progression, change in treatment, patient’s death and withdrawal of consent.

Questionnaire domains:

QoL was assessed with the Functional Assessment of Cancer Therapy – General (FACT-G) questionnaire⁹, which is a validated tool that has been used extensively in clinical studies of patients with hematologic malignancies.¹⁰ The FACT-G questionnaire consists of 27 questions assessing four domains: physical well-being (WB; seven items, score range 0–28), social WB (seven items, score range 0–28), emotional WB (six items, score range 0–24), and functional WB (seven items, score range 0–28). An overall score is generated for each patient and each subscale is scored individually, provided that more than 50% of the items are answered for that subscale. The physical WB scale includes a question about side effect bother ¹¹. Symptom burden was assessed by using the Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE™), a self-report tool that has been previously validated in patients undergoing treatment for hematologic malignancies.^6,7,12 It assesses the severity, frequency, and interference on daily life of various symptoms. The following 8 symptomatic AEs (between 8 and 18 items, depending on number of symptoms endorsed) were assessed: mouth sores, decreased appetite, diarrhea, numbness and tingling, general pain, fatigue, anxiety and sadness. Memory and cognition were assessed with the validated 8 item Neuro-QOL v2 questionnaire (short form).¹³ Self-reported performance status, adapted from ECOG performance status was also collected.¹⁴

Sample Size:

Sample size and power calculations are provided in supplementary data. A sample size of 111 patients across all three arms was determined a priori, which accounted for 10% attrition rate and provided between 77% and 99% power to detect 0.5 to 0.8 SD differences between arms (moderate to large effect sizes by Cohen) for between-group comparisons in longitudinal modeling.

Statistical Analysis:

Individual scores for each PRO at each timepoint were calculated. Longitudinal change was determined by evaluating the difference in scores for each time point vs. baseline for both positive and negative changes. Differences were evaluated both for statistical significance and clinically significant changes. Minimal clinically important differences (MCID) for moderate effect size have been established in previous research. They include: FACT-G total, 9 points; FACT-G physical and functional WB, 3 points; FACT-G emotional and social WB, 2 points; NeuroQoLv2 questionnaire, 8 points. ^15,16 Composite PRO-CTCAE scores were calculated (combining frequency, severity, and interference)for each AE¹⁷ using a method adjusting for pre-existing baseline symptoms. We also evaluated the time profile of PRO-AEs using the Toxicity over Time (ToxT) approach. The ToxT approach is a validated longitudinal approach to AE analysis,¹⁸ but has not been used with patient reported data before. This method should be equally applicable to patient reported AEs, but this is the first application of this method in patient reported data. Data were analyzed in SAS software using mixed model methods. Continuous variables were summarized as median (range) or mean (range). Differences across groups or longitudinal differences within the same group for continuous variables were compared using Kruskal-Wallis and t-tests. Categorical variables were summarized as proportions and compared using chi-square tests or Fischer’s exact tests.

RESULTS

As shown in the consort flow diagram in supplementary Figure S1, 114 patients consented to the study. Three patients were screen failures as they ultimately did not undergo CAR-T therapy/transplant, and 7 patients did not have any follow-up due to change in plan for treatment (n=4) or withdrawal of consent before any follow-up (n=3). Overall, 104 patients were evaluable and comprise the study cohort.

Table 1 describes the baseline characteristics of the study cohort. Age (median: 60–62 years, p=0.34) and sex distribution (females: 39–47%, p=0.79) was similar across the three groups. The majority of patients were White (88–97%). As expected, the underlying hematologic malignancy diagnosis varied considerably in the three groups (p<0.001). In the CAR-T group, the diagnoses were lymphoma (68%), myeloma (24%) and leukemias/other myeloid neoplasms (9%). The autologous SCT cohort comprised predominantly of patients with myeloma (94%) and a few patients with lymphoma (6%). In the allogeneic SCT cohort, the majority of patients had leukemias or other myeloid neoplasms (89%), whereas about 5% of patients each had myeloma and lymphoma, respectively.

Table 1.

Patient Characteristics and Baseline Patient Reported Outcomes in Patients undergoing CAR-T Therapy, Autologous and Allogeneic Stem Cell Transplant for Hematologic Malignancies

		CAR-T Therapy (N=34)	AUTO SCT (N=33)	ALLO SCT (N=37)	P-value
Age, median (range)		62 years (26–77)	62 years (42–74)	60 years (23–75)	0.34
Sex, female, n (%)		16 (47%)	13 (39%)	17 (46%)	0.79
Ethnicity, Non-Hispanic, n (%)		33 (97%)	33 (100%)	36 (97%)	-
Race: Caucasian, n (%)		33 (97%)	29 (88%)	26 (97%)	0.19
Diagnosis N (%)
Myeloma		8 (24%)	31 (94%)	2 (5%)	<0.001
Lymphoma		23 (68%)	2 (6%)	2 (5%)
Leukemias/Other Myeloid Neoplasms		3 (9%)	-	33 (89%)
Baseline Patient Reported Outcomes	Range of scores *	Median (range)	Median (range)	Median (range)
Quality of Life (FACT-G)
Total Score	0–108	84.0 (57.2–106.0)	84.0 (39.0–106.0)	83.0 (43.0–105.0)	0.86
Physical Well-Being	0–28	23.0 (15.0–27.0)	22.0 (9.0–28.0)	26.0 (9.0–28.0)	0.32
Social/Family Well-Being	0–28	24.8 (18.7–28.0)	25.0 (17.0–28.0)	25.0 (18.0–28.0)	0.99
Emotional Well-Being	0–24	19.0 (7.2–24.0)	19.0 (8.0–24.0)	19.0 (11.0–23.0)	0.89
Functional Well-Being	0–28	18.0 (8.0–28.0)	18.0 (3.0–28.0)	18.0 (4.0–28.0)	0.54
Neurocognitive Function
NEUROQOL v2.0 T-Score **	17–64	51.7 (38.9–64.2)	49.6 (35.0–64.2)	50.9 (37.0–64.2)	0.58
Side Effect Bother	-				0.59
Quite a bit		2 (6%)	3 (9%)	3 (8%)
Somewhat’ or ‘A little bit’		17 (50%)	18 (55%)	12 (32%)
Not at all		15 (44%)	12 (36%)	22 (60%)
Activities and Function	-				0.14
Spend most of the day in bed/chair		0 (0%)	1 (3%)	3 (8.1%)
Not feeling up to most things		6 (17.6%)	11 (33.3%)	4 (10.8%)
Not normal self but fairly normal activities		21 (61.8%)	14 (42.4%)	19 (51.4%)
No limitations		7 (20.6%)	7 (21.2%)	11 (29.7%)
Activities and Function Score (Performance Status)	1–5	4.0 (3.0–5.0)	4.0 (2.0–5.0)	4.0 (2.0–5.0)	0.44

^*Higher scores represent better outcome

^**NeuroQoL uses a T-score with mean score in general US population of 50 with SD of 10

Acute CAR-T Therapy Side Effects:

In the CAR-T group, CRS was observed in 77% (n=26) of patients, with grade 2 CRS in 62% of patients, respectively. Median duration of CRS was 5 days (range 0–15). ICANS was observed in 35% (n=12) of patients, with grade ≥ 2 ICANS in 21% (n= 7) of patients, respectively. Median duration of ICANS was 6 days (range: 0–16).

Questionnaire Completion Rate:

Longitudinal questionnaire completion rate is shown in supplementary data, Table S1A and the reasons for missing data are described in Table S1B. The completion rate was as follows- baseline: 100%, week 2: 95%, month 1: 86%, month 3: 73% and month 6: 60%, respectively. Overall, 48% of patients completed all timepoints, while one or more timepoints were not completed in the remaining patients because of (1) early study exit due to disease progression/death/change in therapy: 25%, (2) patient choice: 24%, (3) other/multiple reasons: 3%. Completion rates in the CAR-T group were- baseline: 100%, week 2: 100%, month 1: 91%, month 3: 71% and month 6: 44%, respectively. All timepoints were completed by 38% of CAR-T patients, while one or more timepoints were missing because of (1) early study exit due disease progression/death/ change in therapy: 41% and (2) by patient choice: 21%, respectively.

Baseline Patient Reported Outcomes:

Baseline scores for overall QoL and subdomains of QoL (FACT-G), and neurocognitive function (NeuroQoLv2) were similar across the three groups. The median time from baseline evaluation to cellular therapy was 11 days (interquartile range: 7–24 days), with 12% of patients having a baseline evaluation > 30 days before cellular therapy. Interestingly, baseline physical WB was less affected in all three groups (median scores 23–26, of maximum 28) compared to functional WB (median score 18, of a maximum of 28). At baseline, patients reported similar level of impact from side effects. There was no difference in baseline patient reported performance status and 63–82% patients across all groups had no limitations or were able to continue fairly normal activities, corresponding to an ECOG performance score of 0–1 (Table 1).

Trajectory of QoL after CAR-T Therapy:

Longitudinal trajectory of QoL, depicted as change from baseline including overall QoL and subdomains is shown in Figure 1, with actual values shown in Supplementary Table S2. There was a slight downtrend in overall QoL at week 2 (mean decrease: 4.2) following CAR-T therapy, which was statistically significant vs baseline, but did not meet the criteria for MCID (9 points). QoL improved by month 1 and remained similar to baseline line from month 1 until last follow-up at 6 months. Amongst sub-domains of QoL, there was a decline in physical WB at week 2 (mean decrease: 2.6), which was statistically significant vs baseline, but did not meet criteria for MCID (3 points). Physical WB returned to baseline starting month 1. The drop in functional WB was more dramatic (median decrease at week 2: 7 points). It was both clinically and statistically significant compared to baseline, with MCID being 3 points. Functional WB continued to be lower than baseline at the end of month 1 and returned to baseline starting month 2. On the contrary, patients had improvements in emotional WB over baseline starting at week 2 and continuing until 4 months, though the mean improvement in scores was 1.4 to 1.8 vs baseline and did not meet the MCID (2 points).

Figure 1. Change in quality of life from baseline to six months following therapy in patients undergoing CAR-T therapy, autologous and allogeneic stem cell transplant using FACT-G (Functional Assessment of Cancer Therapy-General) questionnaire.

Shown are overall quality of life score trajectory with FACT-G questionnaire from baseline as well as trajectories of sub-domains of quality of life including (b) physical, (c) functional, (d) social and emotional well-being.

QoL with SCT and Comparison of CAR-T vs. SCT:

Patients undergoing SCT also experienced a short term decline in QoL and gradual improvement, although the degree of decline in QoL and sub-domains of physical and functional WB was more significant and the recovery to baseline started at later timepoints. (Figure 1 and supplementary data Table S2)

Autologous SCT:

Patients undergoing autologous SCT experienced a decline in overall QoL score at week 2 and month 1 (mean decrease. week 2: 14.2 and month 1: 9.7 points). When compared to patients’ own baseline, this decline met the MCID of 9 points. QoL nadired at week 2 and gradually improved and returned to baseline by month 2–3. When comparing QoL in patients in the CAR-T vs autologous SCT group, patients in the CAR-T group had a better QoL at week 2 and month 1, p <0.01 and a trend towards better QoL at month 2. Similar results were seen in the physical and functional WB domains, with nadir at week 2 and gradual improvement thereafter. Compared to patients’ baseline, the decline after autologous SCT was greater than the MCID of 3 points. Physical WB was better in the CAR-T group vs autologous SCT group at week 2 and month 1 and functional WB was better in the CAR-T group at week 2 and a trend towards superiority was seen at month 1 as well.

Allogeneic SCT:

Patients undergoing allogeneic transplant had the greatest decline in QoL amongst the three groups, which nadired at week 2 and although gradual improvement was seen starting at month 1, QoL remained significantly below baseline until month 3 (mean decrease from baseline, week 2: 21.4, month 1: 15, month 2: 8.6, month 3: 5). Patients in the CAR-T group had better QoL than those in the allogeneic SCT group starting at week 2 until month 3 (p<0.01 until month 2, p=0.02 at month 3). Patients in the CAR-T group had significantly better physical and functional WB from week 2 until month 3.

Patient Reported performance Status and Side Effect Bother:

As shown in Figure 2 and supplementary Table S3, patient reported performance status followed a trend similar to QoL, with short term decline and nadir at week 2, and gradual return to baseline in all groups. The decline was the least in the CAR-T group and the most in the allogeneic SCT group. Performance status is scored from 1–5 and mean decrease in the CAR-T group from baseline to week 2 (decrease 0.9) and month 1 (decrease: 0.3) was statistically significantly and returned to baseline by month 2. A similar trend was observed in the autologous SCT group, with mean decrease being significant at week 2 (decrease: 1.2) and month 1 (decrease: 0.9) compared to baseline. On the other hand, the decline was steeper in the allogeneic SCT group and remained lower than baseline up to month 4 (mean decrease, week 2: 1.6, month 1: 1, month 2: 0.6, month 3: 0.5, and month 4: 0.4) Patients in the CAR-T group had better performance status vs. autologous SCT group at month 1 and better performance status vs allogeneic SCT group at all timepoints starting week 2 until month 4. When evaluating patient-reported side effect bother (scored from 0–4), patients in the CAR-T group had no significant change from baseline, whereas there was a significant change noted from baseline up to month 1 in the autologous SCT group and up to month 2 in the allogeneic SCT group. Consequently, patients fared better in the CAR-T group compared to both autologous and allogeneic SCT groups. (Supplementary Table S3)

Figure 2. Change in performance status (self-reported activities and function), side effect bother and cognitive functioning (self-reported using NeuroQoL version 2) from baseline to six months following CAR-T therapy, autologous and allogeneic stem cell transplant.

Patient Reported Individual Symptom Burden with PRO-CTCAE:

It was feasible to collect longitudinal patient symptoms with PRO-CTCAE in patients undergoing complex cellular therapies. The most common treatment emergent AEs in the CAR-T group were decreased appetite (65%), fatigue (55%), pain (47%), mouth sores (44%) and diarrhea (44%) (Table 2). These AEs were mostly grade 1 or 2. The most common grade 3 AEs were fatigue (18%) and decreased appetite (15%). Patients in the CAR-T group had fewer overall AEs compared to the autologous SCT group, including grade 3 AEs of diarrhea (3% vs 21%, p=0.03), fatigue (18% vs. 44%, p=0.03) and pain (6% vs 33%, p=0.006). Patients undergoing allogeneic SCT had the highest proportion of patients experiencing AEs. Patients in the CAR-T group had fewer AEs compared to the allogeneic SCT group, including grade 3 AEs of decreased appetite (15% vs 53%, p=0.002), diarrhea (3% vs 24%, p=0.01), mouth sores (3% vs 50%, p<0.001), fatigue (18% vs 61%, p<0.001) and sad/unhappy feelings (0% vs 18%, p=0.03).

Table 2.

Patient Reported Adverse Events using PRO-CTCAE grading.

	N		Any Grade (adjusted for baseline)			Grade 3 (adjusted for baseline)
PRO-CTCAE term	AUTO	CART	AUTO, n(%)	CART, n(%)	Fisher P	AUTO, n(%)	CART, n(%)	Fisher P
Mouth or Throat Sores	33	34	11 (33%)	15 (44%)	0.46	3 (9%)	1 (3%)	0.36
Decreased Appetite	33	34	28 (85%)	22 (65%)	0.09	10 (30%)	5 (15%)	0.15
Diarrhea	33	34	25 (76%)	15 (44%)	0.01	7 (21%)	1 (3%)	0.03
Numbness/Tingling in Hands/Feet	33	34	16 (48%)	8 (24%)	0.04	6 (18%)	1 (3%)	0.054
Pain	33	34	18 (55%)	16 (47%)	0.63	11 (33%)	2 (6%)	0.006
Fatigue	32	33	22 (69%)	18 (55%)	0.31	14 (44%)	6 (18%)	0.03
Anxiety	33	33	11 (33%)	13 (39%)	0.80	2 (6%)	2 (6%)	1.00
Sad/Unhappy Feelings	33	33	11 (33%)	9 (27%)	0.79	2 (6%)	0 (0%)	0.49
	N		Any Grade (adjusted for baseline)			Grade 3 (adjusted for baseline)
PRO-CTCAE term	ALLO	CART	ALLO, n(%)	CART, n(%)	Fisher P	ALLO, n(%)	CART, n(%)	Fisher P
Mouth or Throat Sores	34	34	27 (79%)	15 (44%)	0.006	17 (50%)	1 (3%)	<0.001
Decreased Appetite	34	34	33 (97%)	22 (65%)	0.001	18 (53%)	5 (15%)	0.002
Diarrhea	33	34	28 (85%)	15 (44%)	<0.001	8 (24%)	1 (3%)	0.01
Numbness/Tingling in Hands/Feet	33	34	17 (52%)	8 (24%)	0.02	1 (3%)	1 (3%)	1.00
Pain	34	34	31 (91%)	16 (47%)	<0.001	23 (68%)	2 (6%)	<0.001
Fatigue	33	33	30 (91%)	18 (55%)	0.002	20 (61%)	6 (18%)	<0.001
Anxiety	34	33	16 (47%)	13 (39%)	0.62	5 (15%)	2 (6%)	0.43
Sad/Unhappy Feelings	34	33	19 (56%)	9 (27%)	0.03	6 (18%)	0 (0%)	0.03

While maximum grade events described above can provide information about peak toxicity, they do not provide insight into longitudinal symptom burden. To evaluate this, we used a ToxT approach. The bar graphs in Figure 3 compare the PRO-CTCAE composite scores at baseline and each timepoint thereafter. In the CAR-T group, symptoms peaked at week 2, especially fatigue, pain, mouth sores and low appetite, coinciding with nadir in QoL. The line plot in supplementary Figure S3 shows mean change from baseline. After a peak at week 2, symptoms return to baseline by month 1 for the majority of symptoms. Symptoms such as pain, which are likely related to underlying malignancy were noted to have an overall improvement in trajectory after CAR-T therapy. Fatigue was also noted to show an improvement in trajectory after mild initial worsening. The peak symptom burden was less in the CAR-T group compared to the SCT groups and there was faster recovery. Patients undergoing allogeneic SCT were observed to have the highest symptom peak and slowest recovery.

Figure 3. Composite grades of patient reported adverse effects using Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) from baseline to six months following CAR-T therapy, autologous and allogeneic stem cell transplant.

Cognitive Function:

Self-reported cognitive function was assessed by NeuroQoLv2. There was no significant decline in self-reported cognitive function after CAR-T therapy as shown in Figure 2 and Supplemental Table S3. In fact, the overall trajectory was positive compared to baseline. In contrast there was a decline in cognitive function in the autologous SCT group at week 2 (decrease 3.7) and in the allogeneic SCT group at week 2 (decrease 5.7) and month 1 (decrease 2.8), although none of these differences met the criteria for MCID (8 points).

DISCUSSION

This study provides comprehensive patient reported data to describe the lived experience of patients undergoing CAR-T therapy and compare it to patients undergoing established forms of cellular therapy i.e autologous and allogeneic SCT.

Baseline QoL and performance status and were similar in patients undergoing CAR-T therapy and SCT. It is interesting to note that functional WB was more impacted at baseline than physical WB across all groups. This is not surprising as patients proceeding to intensive therapies like CAR-T/SCT generally have to be in good physical state. However, they may not be able to fulfil their functional roles at work or at home due to ongoing treatment. Overall, we observed a short-term decline in QoL in patients in all groups, that gradually returned to baseline. The nadir in QoL coincided with peak in AEs. The decline in QoL, physical and functional WB was significantly smaller in the CAR-T group vs. SCT groups and returned to baseline faster. Not surprisingly, patients in the allogeneic SCT group experienced the maximum symptom burden, greatest decrease in performance status and the largest short-term decline in QoL and slowest recovery. The differences may be explained by the nature of cellular therapy and conditioning regimen, as well as the aggressive nature of induction therapy received by patients with acute leukemias prior to alloSCT, which may increase the risk of side effects during post-transplant course. However, it is encouraging to see that QoL in patients in all groups, including the allogeneic SCT group, recovered to baseline in a few months.

There is significant concern about the impact of neurotoxicity on cognition in patients undergoing CAR-T therapy. We observed that patient-reported cognitive function was not impacted in the CAR-T group. However, self-reporting of cognitive function may be limited by patients’ lack of insight into their cognitive deficits. Impact of neurotoxicity on cognitive function after CAR-T therapy should be explored further with objective neurocognitive testing including formal neuropsychiatric testing or computer assisted applications like NIH toolbox¹⁹, which can be administered more widely.

There is limited published data on QoL and other patient reported outcomes in adult patients undergoing CAR-T therapy. In 58 pediatric/adolescent patients undergoing CAR-T therapy for relapsed acute lymphoblastic leukemia in the ELIANA trial, clinically meaningful improvement in QoL was observed over baseline starting at 3 months post-therapy.²⁰ Patients with severe CRS/ICANS also reported improvement in QoL, although the improvement was delayed and lower. Improvement was greatest for physical function. In adult patients receiving CAR-T therapy for large B-cell lymphoma in the JULIET study, patients who responded to CAR-T therapy experienced clinically meaningful and sustained improvement in QoL.²¹ There were somewhat greater improvements in QoL reported in the JULIET study than seen in our study where QoL returned to baseline, but did not improve compared to baseline. For example, mean improvement in functional WB by month 3 and 6 was 1.6 and 1.7 points in the JULIET study (vs. 1.2 and 0.7 in our study); mean improvement in physical WB was 1.5 and 2 points (compared to 1.1 and 0.7 in our study). These differences may be explained by variability in patient population and baseline symptom burden. The JULIET trial did not start assessing post CAR-T QoL until 3 months post CAR-T therapy. There is a lack of published data on short-term QoL in the first few weeks after CAR-T therapy when patients experience peak AEs. Our data show a clear short-term drop in QoL, followed by recovery. Short-term evaluation of QoL is important to inform patient expectations and supportive care options. Many clinical trials of CAR-T therapy are including measures of QoL, but earlier timepoints, that correspond to peak toxicity are often excluded. It is important that these trials capture both short and long-term QoL data and that they are published at the same time as efficacy analysis to allow for an in-depth understanding the patient experience and provide context for the efficacy data.

While there is some data on overall QoL trajectory after CAR-T therapy, none of the previously published studies in CAR-T therapy have comprehensively evaluated patient-reported symptom burden or cognitive assessment in a longitudinal manner. Administration of PRO-CTCAE was feasible in patients undergoing complex cellular therapies and provides important insights into patient symptom burden. Feasibility of PRO-CTCAE has been recently demonstrated in SCT, but not in CAR-T therapy.²² It is clear that patients experience a significant symptom burden with CAR-T therapy, which peaks in the first few weeks after treatment and gradually improves. However, patients continue to have significant symptoms even at 3–6 months after CAR-T therapy. Our findings provide valuable information on key symptoms that can be addressed during follow-up survivorship care. Interventions directed to these symptoms may further improve QoL of these patients. Ruark et al conducted a cross sectional survey with patients who had received CAR-T therapy at least one year prior.²³ About half of the patients reported significant neuropsychiatric concerns including cognitive difficulty or clinically meaningful depression/and or anxiety.²³ In our study, using PRO-CTCAE data, 43% patients in the CAR-T group reported having sad or unhappy feelings at 6 months, although most patients only experienced them rarely and it was similar to the SCT groups (48–50%).

Our study is innovative for several reasons. We evaluated longitudinal QoL and symptom burden in patients undergoing CAR-T cell therapy and compared it to established forms of cellular therapy and there have been no prior direct comparison amongst these groups. If the responses with CAR-T cell therapy remain durable and if it is found to be less toxic in the long-term, it may be used earlier in the treatment course in future. Clinical trials evaluating earlier use of CAR-T therapy in lieu of transplant are already underway. This is also the first study to our knowledge to assess patient reported AEs with PRO-CTCAE and self-reported cognitive function in this population. It also the first application of the ToxT approach to PRO-CTCAE data.

Limitations of our study include its single-center design, different attrition rate across treatments and variable distribution of patients with different hematologic malignancies across the three arms. It is to be noted that the majority of the survey non-completion in the CAR-T group was due to early study exit because of disease progression/death and only 21% patients in the CAR-T group elected not to complete one or more questionnaires. This questionnaire completion rate in eligible patients is quite good for survey studies and comparable to published QoL data from CAR-T clinical trials, but still differs significantly from the other arms. 21 Importantly, we were able to capture 90–100% patients in the first month across all groups, giving an accurate reflection of QoL at the time of peak toxicity. Even though we included a heterogenous disease population, patient demographics and baseline QoL were similar across groups. The difference in diagnosis is to be expected with a study comparing CAR-T and SCT, as common indications for different cellular therapies vary by diagnosis. While underlying diagnosis can impact QoL/symptoms in the long-term, short to intermediate change from baseline is likely due to the type of therapy and not disease related. This study was designed when CAR-T therapy was still early in clinical use. As we now have wider availability of these therapies, future studies should focus on disease specific and CAR-T construct specific evaluation of patient experience. QoL in the allogeneic SCT arm can also be impacted by the conditioning regimen, graft source and development of complications, which was not captured in the study and is a limitation. Disease related outcomes were also not captured.

Understanding and improving patient experience should continue to remain a key focus of clinical trials of CAR-T therapy and hematology practice for several reasons.^7,24 It has been well-described that traditional AE assessment often under-estimates toxicity and patient-reported AEs can provide vital information that is complementary to clinician assessment of AEs.²⁵ Our study provides data for patient education, expectations and shared decision making in CAR-T cell therapy to improve the overall patient experience. Our results can also serve as reference for future studies of patient experience with CAR-T therapy, including interventional studies to improve QoL ²⁶ and for the creation of survivorship plans as disease related outcomes continue to improve in this population. Future areas of study should focus on understanding the long-term symptoms and impact on QoL, cognitive function and ability to return to work and other roles in society. CAR-T therapy is a very expensive treatment. This can potentially limit access to these therapies and place a significant burden on patients and their families.²⁷ Financial burden after CAR-T therapy should be evaluated by using qualitative interviewing and established tools.

In summary, our study provides comprehensive data on symptom burden and QoL over 6 months following CAR-T therapy and compares it to patients undergoing autologous and allogeneic SCT. The results of our study can be used to inform all stakeholders and also identify opportunities for therapeutic intervention and alleviation of symptoms.

Highlights.

• We describe comprehensive longitudinal patient reported outcomes, adverse events and QoL with CAR-T cell therapy vs. autoSCT and alloSCT.

• Short-term QOL, including physical and functional domains was better in the CAR-T group vs. SCT groups.