Prospective Assessment of Quality of Life and Patient-Reported Toxicities Over the First Year After Chimeric Antigen Receptor T-Cell Therapy

Abstract

Chimeric antigen receptor (CAR) T-cell therapy has transformed survival outcomes in patients with relapsed and refractory large B-cell lymphoma (LBCL), but it is associated with a variety of side effects. This study examined changes in patient-reported quality of life (QoL) and toxicities, as well as risk factors for worse QoL and toxicities, in the first year after treatment. Patients with LBCL completed questionnaires assessing QoL and toxicity severity before infusion, and 90, 180, and 360 days after infusion. Mixed models were used to examine changes in QoL and toxicities over time, and clinical moderators of change in QoL and toxicities. Patients reported improvements in physical functioning and fatigue in the year after treatment (P values <.01), but there were no changes in pain, anxiety, or depression over time. Patients with active disease at day 90 reported more physical dysfunction at all postinfusion timepoints (Ps ≤ .01) compared to patients who responded to treatment. Similarly, patients with active disease at day 90 reported worsening depression over time, such that at day 360, depressive symptoms were worse for patients with active disease than patients without active disease (P = .02). Patients treated with 4+ lines of prior therapy reported worsening pain and anxiety over time, such that at day 360, both pain and anxiety were significantly worse for patients previously treated with 4 of more lines of therapy than patients treated with fewer lines of therapy (Ps ≤ .01). Regarding toxicities, patients reported decreasing overall toxicity burden up to day 180, with subsequent worsening at day 360 (P = .02). Most patients reported at least one or two grade 2 toxicities at each timepoint. Patients demonstrated unchanging or improved QoL after treatment with CAR T-cell therapy, but active disease and greater prior lines of therapy were associated with worse QoL outcomes over time. Toxicity severity also improved during the first 6 months post-treatment, but worsened thereafter, particularly among patients with active disease after treatment.

INTRODUCTION

Chimeric antigen receptor (CAR) T-cell therapy has radically improved survival for patients diagnosed with large B-cell lymphoma (LBCL). For example, axicabtagene ciloleucel was FDA-approved for relapsed and refractory LBCL in 2017 and is associated with 41% overall survival and 51% disease-specific survival at 5 years post-treatment [1], underscoring the curative potential of this therapy [2]. In recent years, three additional CAR T-cell therapy products have been approved for relapsed and refractory LBCL (i.e., tisagenlecleucel, brexucabtagene autoleucel, and lisocabtagene maraleucel) [1,3–5].

Despite the potential to cure patients with advanced LBCL, CAR T-cell therapy is associated with a unique toxicity profile. Side effects such as cytokine release syndrome (CRS) and neurotoxicity are well-described [6], along with the risk for B-cell aplasia, long-term cytopenia and infections, and late neurologic and psychiatric effects [7,8]. These toxicities make it critical to address the patient perspective when receiving CAR T-cell therapy. Available evidence suggests patient-reported toxicities peak 7 to 14 days after infusion (eg, pain, fatigue, chills, lack of appetite) [9,10], and generally decrease by day 90 [9–11]. However, some side effects (eg, aching muscles) may not resolve [9] and others (eg, perceived cognition) may worsen beyond day 90 [12]. In addition, in a cross-sectional study of long-term survivors, 20% reported clinically significant anxiety or depression [13]. Initial studies during the first 90 days post-CAR T-cell therapy suggest that quality of life (QoL) tends to improve or remain the same relative to pretreatment [9], and progressive improvements may occur thereafter [14–17]. However, patients’ QoL has been often evaluated in the context of clinical trials [14–16], and QoL among CAR T-cell therapy recipients beyond the first 90 days postinfusion remains understudied. Additional research examining changes in PROs after CAR T-cell therapy is needed.

The current study was designed to examine patient-reported QoL and toxicities in the first year after CAR T-cell therapy. The first aim was to describe changes in patient-reported QoL during the first year after infusion in patients diagnosed with LBCL. The second aim was to describe changes in patient-reported toxicities during the first year after infusion. Post hoc analyses explored whether clinical variables (ie, active disease at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) were associated with patient-reported QoL or toxicities.

MATERIALS AND METHODS

Participants

Patients scheduled to receive treatment with CAR T-cell therapy as part of a clinical trial or standard of care were recruited from October 2016 to February 2020 as part of a study describing patient-reported QoL. To limit sample heterogeneity due to disease type, patients were included in analyses only if they were diagnosed with LBCL and treated with a CD19-directed CAR T-cell therapy. The study was approved by the Advarra Institutional Review Board. Patients were eligible to participate if they were at least 18 years of age, able to speak and read English, able to provide written informed consent, diagnosed with hematologic cancer, scheduled to receive CAR T-cell therapy at Moffitt Cancer Center, and without documented or observable psychiatric or neurological diagnoses that could interfere with study participation (eg, schizophrenia).

Study Procedure

Participants provided written informed consent prior to study initiation, and completed assessments for the larger study prior to receiving conditioning chemotherapy for CAR T-cell therapy (ie, baseline), and at 14, 30, 60, 90, 180, and 360 days later. Analyses of the early recovery period (ie, the first 90 days after infusion) were previously published [9].

Measures

Demographic and clinical characteristics.

At baseline, participants provided sociodemographic data (eg, date of birth, gender, marital status, race, ethnicity, education). Clinical data were collected via medical chart review (eg, CAR T-cell therapy type, trial type, inpatient days before day 100, prior lines of therapy, comorbidities, and highest grades of CRS and neurotoxicity). Disease status at approximately 90 and 360 days after infusion was evaluated using data collected on participants from the Center for International Blood and Marrow Transplant Research. In instances where disease status was unavailable through the Center for International Blood and Marrow Transplant Research (eg, recency of patient treatment), these data were manually abstracted from patients’ electronic medical charts by the study team. As reported previously [9,18], neurotoxicity was initially graded using the CAR T-cell-related encephalopathy syndrome grade as defined by the CAR T-Cell Therapy-Associated Toxicity (CARTOX) working group [19]. Later in the study, neurologic toxicity was graded using the Immune Effector Cell-Associated Neurotoxicity Syndrome [6]. For both measures, higher grades indicate worse toxicity.

Quality of life.

Participants completed assessments at baseline, and days 90, 180, and 360. QoL was initially measured using the Medical Outcomes Study Short Form-36 version 1 (SF-36v1) [20], and, later, by the PROs Measurement Information System-29 version 2.1 (PROMIS-29) [21] after publication of the proposed Center for Medicare and Medicaid Services coverage decision for CAR-T in early 2019 [22]. The SF-36 is a multi-purpose, short-form health survey with 36 items measuring functional health and well-being. The PROMIS-29 was developed to evaluate eight domains of QoL: physical function, anxiety, depression, fatigue, sleep disturbance, ability to participate in social roles and activities, pain interference, and pain intensity. Using linked crosswalk tables created by PROsetta Stone [23], individual SF-36 raw scores were converted to PROMIS T-scores (SF-36v1 items were reverse coded as needed if they were in the opposite direction of the PROMIS-29 QoL subscales, and the summed SF-36 scales were converted to PROMIS-29 T-scores using the linked crosswalk tables). These converted values yielded five QoL subscales: physical functioning, anxiety, depression, fatigue, and pain interference [24–26]. For each subscale, the general population mean is 50 with a standard deviation of 10. Higher scores indicate more of the attribute being measured. A five-point difference in PROMIS scores (ie, one-half a standard deviation) is considered clinically meaningful [27].

Toxicities.

Participants completed items from the National Cancer Institute-developed PRO version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) at baseline, and days 90, 180, and 360 [28,29]. The PRO-CTCAE was designed for investigators to select individual items relevant to a specific cancer treatment and/or diagnosis. Using 5-point Likert-type scales, participants rated the frequency (0 = “never,” 1 = “rarely,” 2 = “occasionally,” 3 = “frequently,” 4 = “almost constantly”) and severity (0 = “none,” 1 = “mild,” 2 = “moderate,” 3 = “severe,” 4 = “very severe”) of each toxicity. Severity was considered moderate-severe if patients selected “moderate,” “severe,” or “very severe” for each toxicity. A total of 21 common toxicities were assessed: abdominal pain, constipation, cough, decreased appetite, diarrhea, dry mouth, fatigue, feeling sad, hair loss (yes/no), hand–foot syndrome, headache, insomnia, itchy skin, joint aches, muscle aches, nausea, problems with concentration, problems with memory, rash (yes/no), shortness of breath, and wheezing. These 21 toxicities were selected based on adverse events reported previously in patients treated with axicabtagene ciloleucel, hematopoietic cell transplantation (HCT), and other immunotherapies.

Using the 18 severity items (see Supplementary Table 1), a toxicity index was calculated as an indicator of overall toxicity burden according to the following equation: [30,31]

$$\mathit{\text{Toxicity}}\mathit{\text{Index}}=\sum _{i=1}^n\frac{X_i}{{\prod }_{j=1}^{i-1}(1+X_j)}$$

In this equation, toxicities are ranked in order of toxicity severity, assigned decreasing weights, and then summed. This approach accommodates the differential impact of multiple toxicities and yields an easily interpretable score, wherein the number before the decimal point indicates the highest toxicity grade reported by a participant and the numbers after the decimal point indicate other reported toxicities beyond the highest grade, with lower grade toxicities contributing less and less to the final score. For example, a participant reporting only one grade 3 toxicity would have a toxicity index of 3.0, a participant reporting one grade 3 and two grade 2 toxicities would have a toxicity index of 3.67, and a participant reporting two grade 3 toxicities would have a toxicity index of 3.75, indicating worse symptom burden.

DATA ANALYSES

Participants were included in analyses if they completed at least one assessment. Means and frequencies were used to describe participant characteristics and outcomes over time. Linear and quadratic changes in QoL over time were examined using mixed models, which allow for the use of all available data. Associations between clinical moderators (ie, active disease at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) and QoL were examined using linear mixed models. Because relatively few participants experienced maximum CRS and neurotoxicity of grade 3 or above, these variables were dichotomized as grade 0 to 1 versus 2 or higher to ensure adequate statistical power to compare groups. Because disease status was assessed at days 90 and 360, mixed models examining disease status included intercepts set to days 90 and 360, respectively.

Changes in the toxicity index were examined using linear and quadratic mixed models using all available data. Differences in the toxicity index by disease status (ie, active disease versus no active disease) over time were analyzed using independent samples t-tests. Associations between clinical moderators (ie, active disease at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) and toxicity (ie, the toxicity index) were examined using linear mixed models. Significant moderator analyses were decomposed using between-subject comparisons of least-squared means from the linear mixed models. All analyses were conducted in SAS, version 9.4 (Cary, NC).

RESULTS

A participant flow diagram is included in Supplementary Figure 1. Participant characteristics are shown in Table 1. On average, participants (n = 120) were 61 years of age (SD = 12) and majorities were male (59%), married (71%), and/or White (89%). Most participants were treated with axicabtagene ciloleucel (86%) and were without active disease at days 90 (72%) and 360 (69%).

Table 1.

Participant Characteristics (n = 120)

	M (SD)/n (%)
Age: M (SD), [range]	61.18 (11.90), [19.5–78.4]
Gender: n (%) female	49 (41)
Marital status: n (%) married	84 (71)
Race: n (%) White	107 (89)
Ethnicity: n (%) non-Hispanic	112 (94)
Education: n (%) college graduate	62 (53)
Diagnosis: n (%)
Diffuse large B-cell lymphoma	76 (63)
Follicular lymphoma	39 (33)
Other	5 (4)
CAR T-cell therapy: n (%)
Axicabtagene ciloleucel	104 (86)
Tisagenlecleucel	17 (14)
Trial type: n (%)
Clinical trial	50 (42)
Commercial/standard of care	70 (58)
KPS: n (%)
<80	12 (10)
80+	108 (90)
Inpatient days before day 100: M (SD)	15.43 (9.71)
Prior lines of therapy: n (%)
1–3	72 (60)
≥4	48 (40)
Comorbidities: n (%)
<3	71 (59)
≥3	49 (41)
Baseline remission status: n (%)
Complete response	2 (3)
Partial response	7 (9)
Resistant	66 (88)
Without active disease at day 90: n (%)	83 (72)
Without active disease at day 360: n (%)	60 (69)
Maximum CRS grade: n (%)
Grades 0 or 1	69 (58)
Grades 2–4	51 (43)
Maximum neurologic toxicity grade: n (%)
Grades 0 or 1	92 (78)
Grades 2–4	26 (22)

Comorbidities were assessed using the Hematopoietic Cell Transplant-Comorbidity Index. Baseline remission status was defined as partial response when inclusive of sensitive disease after either primary induction failure or relapse, and resistant when patients had stable disease/progressive disease after primary induction failure or relapse. Without active disease was defined as either a complete or partial response to treatment.

M indicates mean; SD, standard deviation; KPS, Karnofsky Performance Status; CRS, cytokine release syndrome.

Quality of Life

Raw means for QoL by timepoint are presented in Supplementary Table 2. Parameter estimates from mixed models examining change in QoL over time are presented in Table 2. Estimated means from the mixed models are displayed in Figure 1. For physical functioning, there was a significant linear change such that patients reported improvements over time (P < .01) (Figure 1A). Prior to infusion, and at day 90, physical functioning was more than one-half a standard deviation below the general population norm (ie, the estimated T-scores were below 45), indicating a clinically meaningful level of physical dysfunction on average. Pain was similar to population norms at all timepoints and did not change over time (P = .71) (Figure 1B). For fatigue, there was a significant quadratic effect of time such that patients reported initial reductions in fatigue after infusion, with little change thereafter (P < .01) (Figure 1C). However, fatigue scores were similar to population norms at all timepoints. For anxiety and depression, scores were similar to population norms at all timepoints and did not change significantly over time (Ps > .37) (Figure 1D,E).

Table 2.

Parameter Estimates From Mixed Models of Quality of Life, as Measured by the PROMIS-29

	Physical Functioning	Pain	Fatigue	Anxiety	Depression
Intercept	46.96*	49.90*	49.64*	48.75*	48.42*
Time	0.67†	−0.10	1.61	−0.19	0.21
Time × Time	-	-	0.59†	-	-

The intercept was set to day 360; a main effect of time indicates that there was a significant linear change in QoL over time. A time-by-time interaction indicates a significant quadratic change in QoL over time. Estimates from quadratic mixed models are shown when the quadratic effect of time was significant.

^*P < .001.

^†P <.01.

Figure 1.

Estimated means and 95% confidence intervals over time, as measured by the PROMIS-29 for (A) physical functioning, (B) pain, (C) fatigue, (D) anxiety, and (E) depression.

Moderator analyses indicated no significant interactions between clinical variables and time for physical functioning (P values >.05), but there were main effects of day 90 disease status (P < .01) and neurotoxicity (P = .02) on physical functioning. Between-subjects comparisons of least squares means indicated that patients who responded to treatment by day 90 reported significantly better physical functioning at baseline (P = .01), and at all postinfusion timepoints (Ps < .01) compared to patients with active disease at day 90. At all postinfusion timepoints, this group difference was also clinically significant (ie, the difference in estimated T-scores was 5+ points). Similarly, patients with grade 0 or 1 neurotoxicity reported significantly better physical functioning than patients with grade 2 to 4 neurotoxicity at baseline (P = .02) and day 90 (P = .03), although the differences in estimated scores over time were not clinically significant between the two groups at any timepoint.

For pain, there was a significant main effect of disease status at day 90 (P = .01), such that patients who responded to treatment by day 90 reported significantly less pain at baseline (P = .01) and days 90 (P = .01) and 180 (n = 0.03) than patients with active disease at day 90. However, the differences in estimated scores over time were not clinically significant between the two groups at any timepoint. There was also a significant interaction between time and lines of prior therapy (P < .01) such that pain increased over time in patients with 4 or more lines of prior therapy (P = .03) but decreased in patients with 3 or fewer lines of prior therapy (P = .04). Patients with 4 or more lines of prior therapy reported significantly more pain at day 360 than patients with 3 or fewer lines of prior therapy (P < .01), a difference that was also clinically significant (see Supplementary Figure 2).

For fatigue, there was a significant interaction between time and day 360 disease status (P = .04), such that patients who responded to treatment by day 360 reported reductions in fatigue over time (P < .01), whereas patients with active disease at day 360 reported unchanging fatigue prior to day 360 (P = .74). However, there were no statistically significant (Ps > .06) or clinically significant cross-sectional differences in fatigue between patients with active disease at day 360 versus patients without active disease at day 360 (see Supplementary Figure 3).

For anxiety, there was a significant interaction between time and lines of prior therapy (P = .03). Within-group analyses revealed no significant changes over time in anxiety for patients with 4 or more lines of prior therapy or patients with fewer prior lines of therapy (Ps > .10), but patients with 4 or more lines of prior therapy reported significantly more anxiety at day 360 than patients with fewer lines of prior therapy (P = .01) (see Supplementary Figure 4). This difference at 360 days was clinically significant.

For depression, there was a significant interaction between time and day 90 disease status (P = .03) whereby patients who responded to treatment by day 90 reported unchanging depressive symptoms over time (P = .98), but patients with active disease at day 90 reported worsening depression over time (P = .02). By day 360, patients with active disease reported significantly worse depressive symptoms than patients without active disease (P = .02) (see Supplementary Figure 5). This difference at 360 days was clinically significant. There were no other significant main effects of clinical moderators (ie, maximum CRS grade) or interactions with time (P values >.05).

Toxicities

There were significant quadratic changes in the toxicity index among patients over time (P = .02), such that average toxicity levels declined after infusion until day 180 and increased thereafter. Specifically, the mean toxicity index was 2.89 at baseline (SD = 1.23, range=0 to 4.99), 2.65 at day 90 (SD = 1.09, range = 0 to 4.99), 2.59 at day 180 (SD = 1.22, range = 0 to 5.00), and 2.76 at day 360 (SD = 1.15, range = 0 to 5.00). Thus, at all timepoints, patients reported on average at least one or two grade 2 toxicities. Results of independent samples t-tests indicated no significant differences in the toxicity index by disease status at day 90 (P = .63), day 180 (using the disease status at day 90) (P = .49), or day 360 (P = 1.00). Linear mixed models of clinical moderators (ie, active disease at days 90 and 360, maximum CRS grade, maximum neurotoxicity grade, and prior lines of therapy) indicated no significant main effects or interactions with time (Ps > .05).

Prior to receiving CAR T-cell therapy, 72% (n = 85) of patients providing toxicity data reported at least one symptom as moderate-severe. For brevity, we report on the 6 to 7 most frequently reported moderate-severe symptoms. The most frequently reported moderate-severe symptoms were fatigue (42%), insomnia (35%), decreased appetite (27%), dry mouth (26%), shortness of breath (22%), and aching muscles (19%).

At day 90, 69% (n = 59) of patients providing toxicity data reported at least one symptom as moderate-severe. Among patients who responded to treatment at day 90 and provided toxicity data (n = 68, or 79% of patients who provided data at day 90), 71% (n = 48) reported at least one moderate-severe symptom. The most frequently reported moderate-severe symptoms were aching muscles (27%), fatigue (27%), aching joints (22%), insomnia (21%), dry mouth (15%), and headache (13%). Similarly, at day 90, patients with active disease (n = 18, or 21% of patients who provided data at day 90; 61% of whom reported at least one symptom as moderate-severe) frequently reported moderate-severe aching muscles (22%), fatigue (28%), aching joints (33%), insomnia (11%), and headache (11%). In addition, patients with active disease at day 90 reported moderate-severe problems with abdominal pain (17%).

At day 180, 59% (n = 45) of patients providing toxicity data reported at least one symptom as moderate-severe. Among patients who responded to treatment (at day 90) and provided toxicity data at day 180 (n = 62, or 82% of patients who provided data at day 180), 58% (n = 36) reported at least one moderate-severe symptom. The most frequently reported moderate-severe symptoms were fatigue (26%), aching joints (26%), insomnia (24%), shortness of breath (21%), dry mouth (18%), sad or unhappy feelings (16%), and aching muscles (16%). Similarly, at day 180, patients who had active disease (at day 90) (n = 14, or 18% of patients who provided data at day 180; 64% of whom reported at least one symptom as moderate-severe) also frequently reported moderate-severe fatigue (36%), insomnia (36%), dry mouth (36%), sad or unhappy feelings (36%), and aching muscles (29%). In addition, patients with active disease at day 90 reported moderate-severe problems with concentration (29%) and decreased appetite (29%).

At day 360, 68% of participants (n = 48) providing toxicity data reported at least one symptom as moderate-severe. Among patients who responded to treatment at day 360 and provided toxicity data (n = 51, or 72% of patients who provided data at day 360), 63% (n = 32) reported at least one moderate-severe symptom. The most frequently reported moderate-severe symptoms were insomnia (25%), fatigue (25%), shortness of breath (22%), aching joints (22%), aching muscles (18%), sad or unhappy feelings (16%), and cough (16%). Similarly, at day 360, patients with active disease (n = 20, or 28% of patients who provided data at day 360; 80% of whom reported at least one symptom as moderate-severe) also frequently reported moderate-severe insomnia (37%), fatigue (53%), and sad or unhappy feelings (21%). In addition, patients with active disease at day 360 reported moderate-severe dry mouth (32%), nausea (21%), and constipation (21%). Figure 2 shows the proportions of participants reporting none, mild, or moderate-severe toxicity severity at each timepoint for the most frequently reported moderate-severe symptoms at day 360 by patients who responded to treatment at day 360.

Figure 2.

Proportions of participants with no (green), mild (yellow), or moderate-severe (red) toxicities, as measured by the PRO-CTCAE, at each timepoint by disease status (no active disease [No AD] versus active disease [AD]) for (A) insomnia, (B) fatigue, (C) shortness of breath, (D) aching joints, (E) aching muscles, (F) sad or unhappy feelings, and (G) cough.

DISCUSSION

This study describes patient-reported QoL and toxicities among patients with LBCL treated with CAR T-cell therapy in the first year after treatment, building on previous research on acute PROs in the first 90 days after CAR T-cell therapy in the same sample [9]. At baseline, patient-reported physical functioning was worse than population norms but improved over time by day 360. Improvements in patient-reported physical functioning have previously been reported in the context of some clinical trials of CAR T-cell therapy (tisagenlecleucel, idecabtagene vicleucel) [15] but not others (lisocabtagene maraleucel) [16]. Further, studies of CAR T-cell therapy recipients suggest that patient-reported physical functioning is better in this patient population relative to patients receiving standard of care [32], including autologous or allogeneic HCT [11]. Of note, results from this study suggest significant main effects of clinical moderators on physical functioning, whereby patients with active disease at 90 days postinfusion reported worse physical functioning than patients who responded to treatment at 90 days. Similarly, our prior research with the same sample showed that patients with active disease at 90 days demonstrate worse objectively measured cognition at 360 days [18].

For fatigue, patients reported improvements from baseline to day 360, but this change was not clinically significant. Previous research has demonstrated improvements in fatigue after CAR T-cell therapy [14,16,33] such that fatigue in long-term survivors is comparable to the general population [13]. Indeed, in patients who responded to treatment by day 360, fatigue improved over time postinfusion. In contrast, patients who did not respond to treatment and had active disease by day 360 reported no improvements in fatigue postinfusion.

Regarding the other QoL domains (pain, anxiety, and depression), there were no significant mean-level changes over time, and patient-reported scores were comparable to population norms at all timepoints. However, there were meaningful associations with clinical variables such as disease status and lines of prior therapy. For example, patients who ultimately responded to treatment at day 90 reported less pain both before and after infusion than patients who did not respond to treatment. For depression, patients who did not respond to treatment at day 90 reported worsening depression over time, and by day 360, this difference in depressive symptoms was clinically significant. In addition, patients previously treated with 4 or more lines of therapy reported worse pain and anxiety at day 360 as compared to patients who received fewer lines of therapy. These findings add to our previously published results highlighting that receiving more prior lines of therapy is associated with worse outcomes after treatment (eg, worse neurocognitive performance) [18]. Further, even though there was no significant mean-level change in depression or anxiety over time, mental health concerns are common across all stages of survivorship [9,13], particularly symptoms of anxiety, depression, and cancer-related distress [9,13,17,33], and many patients would benefit from mental health services. For example, research examining distress in the first 6 months after treatment with CAR T-cell therapy suggests that 29% of patients still report clinically significant depression at 6 months postinfusion [17]. Additional work with larger samples is needed to identify mental health concerns and comprehensive risk factors for poor QoL after CAR T-cell therapy.

Prior to CAR T-cell therapy, more than two-thirds of participants reported at least one moderate-severe toxicity, likely due to effects of previous therapies. This percentage was similar at each subsequent timepoint among participants who responded to treatment. Conversely, the percentage of patients with active disease who reported at least one moderate-severe toxicity generally increased from day 90 to day 360 such that 80% of patients with active disease reported at least one moderate-severe toxicity at day 360. Thus, among patients with active disease in particular, the percentage of patients reporting at least one moderate-severe symptom increased after the acute period and up to 1-year postinfusion. In terms of overall toxicity burden, participants reported at least one or two grade 2 toxicities at each timepoint, indicating that toxicities were bothersome and interfered their daily activities. Further, there were quadratic changes in overall toxicity burden such that toxicity improved up to 180 days and worsened thereafter. These changes in overall toxicity burden occurred regardless of patients’ disease status suggesting that even patients who respond to treatment may experience worsening toxicities after 180 days post-treatment. Post-hoc stratified analyses suggest, however, that quadratic changes in toxicity burden may be most evident in patients with active disease (ie, those who do not respond to treatment) by day 360 (P = .0545). Despite this finding, the toxicity profile of CAR T-cell therapy compares favorably to other forms of cellular therapy, such as autologous and allogeneic HCT, considering that CAR T-cell therapy survivors reported less frequent and severe side effects than HCT recipients [11].

The current study has several strengths, including a longitudinal design and the inclusion of a toxicity index which provides a comprehensive measure of patient-reported side effects used to examine change in toxicity burden over time. This study also utilized advanced statistical methods such as mixed models that account for missing data over time. However, the study also has some limitations. The sample was primarily well-educated, White, and married, which may limit the generalizability of our results to patients with different demographic characteristics. Further, the majority of patients received axicabtagene ciloleucel, and only a few patients were treated with tisagenlecleucel, thus we did not have the statistical power compare patient-reported toxicities and QoL by CAR T-cell therapy type, even though tisagenlecleucel has a different toxicity profile [34]. In addition, patients with active disease at the postinfusion timepoints may have received additional treatments that may have influenced their QoL or symptom burden over time. Future, prospective studies of QoL among patients with active disease (ie, who do not respond to treatment) are needed to clarify the effects of later-line therapies on QoL in this patient population. However, despite these drawbacks, findings from this study improve our understanding of patients’ QoL after CAR T-cell therapy and can be used to help educate patients about what to expect before starting treatment.

Overall, results from this study indicate that QoL in patients treated with CAR T-cell therapy may improve or remain similar to pretreatment levels during the first year after treatment. However, risk factors such as non-response to treatment and more prior lines of therapy may worsen patients’ QoL over time. Further, toxicities may improve during the first 6 months after infusion but worsen thereafter. Future research employing qualitative methods with disease-free survivors of CART is needed to better understand patients’ reports of symptom burden and impairment in QoL. Supportive care interventions to prevent or address worsening QoL and toxicities (eg, exercise or nutritional prehabilitation prior CAR T-cell therapy, post-treatment symptom management) have potential to significantly improve patient well-being over time.