Patient-reported outcomes of chimeric antigen receptor T-cell therapy in hematologic malignancies: a systematic review and meta-analysis

Abstract

Few studies evaluated patient-reported outcomes (PROs) for patients with hematologic malignancies receiving with Chimeric Antigen Receptor T (CAR-T) cell therapy. We performed a systematic review and meta-analysis to evaluate the benefits of CAR-T cell therapies focused on PROs. A systematic literature searched from PubMed, Cochrane, and the Web of Science from inception to September 2023. Study selection and data extraction were conducted independently by two reviewers based on pre-specified criteria. The COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) Risk of Bias checklist was used to evaluate the methodological quality of the included studies. The random-effects model was employed to calculate the combined effect and 95% Confidence intervals. This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline and the protocol was registered with PROSPERO (CRD42024586706). We identified 20,110 studies. Of those, 15 studies with 16 different PRO measures (PROMs) were included in the meta-analysis. CAR-T cell therapy improved PROs in the six domains of general health status, pain, fatigue, depression, social function, and cognitive function: from the general health status (SMD: 0.57, 95% CI: 0.34 to 0.81) to cognitive function (SMD: 0.25, 95% CI: 0.14 to 0.37). The current meta-analysis shows that CAR-T cell therapy produces clinically meaningful differences in PROs. These results suggest that the professional perspective and patient values and preferences should be weighed equally when considering CAR-T cell therapy for hematologic malignancies.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-77210-2.

Introduction

Management of patients with relapsed/refractory hematologic malignancies has consistently been a major challenge in clinical oncology. Despite the advances in treatment over the past century, the prognosis for patients with relapsed/refractory hematologic malignancies remains poor. Chimeric antigen receptor T (CAR-T) cell therapy has recently become a viable treatment option and has achieved remarkable results in hematologic malignancies¹. To date, the six CAR-T cell products for treatment of hematologic malignancies have been approved by the Food and Drug Administration (FDA). CAR-T cell therapy has a good therapeutic response in hematologic malignancies, but it has the commonly observed toxicities related to activation of the immune system such as cytokine release syndrome (CRS) or immune effector cell-associated neurotoxicity syndrome (ICANS)². Moreover, there is controversy over its cost-effectiveness due to high prices of CAR-T³. Decision-makers and healthcare systems responsible for discussing these treatment options with patients suffer from insufficient information and uncertainty. Although these uncertainties are inevitable during the technology development phase, they can be reduced through clinical trials and relevant data collection⁴. Among relevant data collection, the patient’s perspectives and experiences during treatments have been considered an important factor in providing real-world data on treatment safety and supporting decisions about treatment options.

Patient-reported outcomes (PROs) are a method of measuring health status based on information self-reported by patients rather than assessed by clinicians. PRO measures (PROMs) are tools used to report PROs and allows assessing symptom burden, activity limitations, health-related quality of life (HRQoL), satisfaction or adherence to treatment and quality of care⁵. PROs can provide important indicators to assess treatment efficacy that are not confirmed by objective tests or clinical assessments⁶. Furthermore the FDA and the European Medicines Agency (EMA) have recognized PROs as a measure of treatment efficacy⁷. Currently, those outcomes are used to assess the effects of treatment and quality of care, to evaluate policies, and to inform health economics. It is important to combine PROMs with functional outcomes to gain insight into both physiological effect and patient well-being.

The concept of a minimal clinically important difference (MCID) for PROMs has become a standard approach to interpreting the clinical relevance of PRO changes. The MCID is defined as the smallest difference in score in an area of interest that requires a change in management that is perceived by the patient as beneficial and without problematic side effects and excessive costs⁸. When interpreting PROs, it is important to understand whether the results represent clinically significant changes rather than statistically significant changes.

A review of PROs in patients with hematologic malignancies receiving CAR-T cell therapy was published by Kamal et al.⁹, but only three studies were included. Another review including 14 CAR-T cell therapy studies on hematologic malignancies also evaluated PROs and was published from 2015 to July 2022 ¹⁰. However, no prior study has meta-analyzed the patients’ perspective of toxicity and efficacy of CAR-T using PROs. Given the increasing interest in this area of research, we performed a systematic review and meta-analysis to evaluate the benefits of CAR-T cell therapies focused on PROs. In addition, we examined whether the results of PROs were clinically meaningful changes through application of MCID.

Materials and methods

This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines¹¹ (supplementary 1) and the PRISMA abstract checklist (supplementary 2).

Identification of studies

We conducted a systematic literature search in PubMed, Cochrane, and Web of Science databases to identify CAR-T cell therapy studies in patients with hematologic malignancies that assessed PROs and were published up to September 2023. The following terms were used: (“chimeric antigen receptor” or “CAR-T” or “axicabtagene ciloleucel (axi-cel)” or “tisagenlecleucel (tisa-cel)” or “lisocabtagene maraleucel (liso-cel)” or “brexucabtagene autoleucel (brexu-cel)” or “idecabtagene vicleucel (ide-cel)” or “ciltacabtagene autoleucel (cilta-cel)”) and (“hematologic malignancy” or “hematologic neoplasm” or “leukemia” or “lymphoma” or “myeloma”) and (“quality of life” or “health related quality of life” or “patient reported outcomes” or “PRO” or “PROs” or “HRQoL” or “QOL”). The full search procedure is documented in Supplementary 3. Additional publications were identified by hand searching the reference lists of these articles or consulting relevant study publication lists at http://clinicaltrials.gov. The protocol was registered with PROSPERO (CRD42024586706).

Eligibility criteria

The inclusion criteria were articles suitable for PICO-SD, as follows: participants diagnosed with any hematologic malignancy regardless of disease stage, patient age, or sex; intervention involving any kind of CAR T-cell therapy used alone regardless of its approval from regulatory agencies; and studies incorporating any kind of PRO measurement(s). Intervention studies involving more than five people were included if longitudinal PROs were reported, regardless of their research design. Conference abstracts or studies in which longitudinal PRO results could be identified were also included.

The exclusion criteria were as follows: the use of any kind of combination therapy in addition to CAR-T cell, studies reporting cost-utility analyses, individual case reports involving fewer than five people, studies that were not written in the English language, case reports and reviews, and basic science articles (such as in vitro research). If a selected study involved multiple publications, information from all related papers was extracted to maximize the quality of the information in our review.

Study selection

Two independent investigators (HJP and HJ) reviewed the database search results using abstracts as the initial screening. Duplicate studies were removed before the initial screening. During the initial screening phase, we included original studies such as clinical trial, case reports, observational studies, as well as cases published in abstract format. Studies were considered eligible if they addressed PROs and involved patients of any age or gender who had received CAR-T cell for hematological malignancies. In the subsequent screening phase, full-text manuscripts that appeared potentially eligible were retrieved and reviewed to select the final eligible studies. Disagreements between the two investigators were adjudicated by a third independent reviewer (HWY) who confirmed the adequacy of studies based on eligible criteria.

Data extraction

A predefined data extraction form was used to collect the following information from each eligible study by two investigators: (1) baseline characteristics of the included studies (i.e., first author, year of publication, study acronyms if available, registration number, trial phase); (2) details of clinical information (i.e., type of hematologic disease, type of CAR T-cell product, number of patients); and (3) PRO assessment characteristics (i.e., PRO measures and follow-up period). The maximum time point at which the results of PROM were reported was collected as a follow-up period. Data were extracted at baseline, one month, three months, and more than six months after therapy. If the data were collected for more than six months, the data at the most recent follow up time were extracted by classifying the results of PROs into the following seven domains: general health status, pain, fatigue, anxiety, depression, social function, and cognitive function. When the results of two or more PROMs were reported in each domain (general health status, pain, fatigue, anxiety, depression, social function, and cognitive function), we selected the result of one for analysis according to the following criteria: the PROM measured in all patients and reported as mean and standard error (SE) or standard deviation (SD) at the time of measurement, the PROM with a higher COSMIN score, the PROM with more patient reports, or the PROM that was preferred with regard to cancer patients.

When studies presented the result data with SE, we changed the data from SE to SD. When the data were presented graphically, we extracted the numerical data from the graph using a WebPlotDigitizer version 4.6 (https://automeris.io/WebPlotDigitizer). If no available data were found in the journal, we attempted to contact the authors for raw data. If available data were found in the abstract or poster for the same studies, the data were extracted for our study.

Risk of bias assessment

The COSMIN Risk of Bias checklist was used to evaluate the methodological quality of the included studies¹². We assessed the measurement properties of the five domains of structural validity, internal consistency, reliability, construct validity, and responsiveness for each PROM development or validation study. The PROMs used in the eligible studies were listed. The measurement properties of the five domains were evaluated based on the original validation journal of PROMs developed for hematological cancer or cancer patients. The results of each domain were evaluated as sufficient, insufficient, or indeterminate with 1 point for “sufficient” and 0 points for “insufficient” or “indeterminate.” The total overall quality score of each PROM was calculated as the sum of scores in the five domains.

Statistical analysis

The PROM outcome data were analyzed using R software and RStudio (version 2022.12.0 + 353; R Core Team, Vienna, Austria) and the Comprehensive Metaanalysis 2.0 ¹³. To standardize the results measured by different tools, we calculated the standardized mean differences (SMDs) and 95% confidence intervals (CIs) for each study. The SMD expresses a difference in the mean outcome between groups in units of pooled standard deviation in each study¹⁴. The mean changes of general health status, pain, fatigue, anxiety, depression, social function, and cognitive function were evaluated at one, three, and six months from baseline. Random-effects modeling was used for analysis if the data came from varied populations with different distributions. The DerSimonian and Laird’s random-effects model was employed to calculate the combined effect and 95% CI considering the existence of heterogeneity between studies. The results were presented as pooled analyses in forest plots¹⁵. We analyzed contour-enhanced funnel plots to evaluate publication bias in the general health status outcome using the inverse variance method. The trim-and-fill method was applied to adjust for funnel plot asymmetry. All procedures were conducted in accordance with the Cochrane Handbook¹⁶. The step-by-step R code was included in the supplementary 4.

Results

Search and study selection

We identified 20,110 studies published through September 2023 that were screened for eligibility. Of these, we retrieved a total of 15 eligible studies that reported PRO results. Five posters that had additional PRO information for the 15 eligible studies were included. Details of the search strategy and selection process of the studies included in this review were documented according to the PRISMA 2020 flowchart (Fig. 1).

Fig. 1.

Flow diagram of the study selection process.

Study characteristics

Fifteen eligible studies enrolled patients with relapsed/refractory hematologic diseases^17–31. Ten studies examined patients with B-cell lymphoproliferative malignancies treated with CD19-directed CAR T cells (two studies with axi-cel, two studies with tisa-cel, two studies with liso-cel, two studies with brexu-cel, and two studies with axi-cel or tisa-cel). One study examined patients with B-cell lymphoproliferative malignancies treated with bispecific CD20- and CD19-directed CAR T cells. Two studies examined patients with multiple myeloma treated with BCMA (B cell mutation antigen)-directed CAR T cells (one study with ide-cel, one study with cilta-cel). Finally, two studies examined patients with either B-cell lymphoproliferative malignancy or multiple myeloma treated with any type of CAR T-cell therapy.

Eight studies were early-phase trials (six phase 2 trials, two phase 1 trials), two were phase 3 trials, and five were observational. Overall, 1157 (range 12–181) patients treated with CAR-T cell therapy were included in the selected studies. Most of the studies (n = 10) enrolled fewer than 100 patients. The follow-up periods were 3 to 24 months (Table 1).

Table 1.

Characteristics of the included studies.

First author (year)	CAR-T producta	Target antigen	Study acronym	Registration number	Phase	Multi-center	Type of disease	Age	Subject (n)	Instrument	Maximum Follow-up (month)
Hoogland 2020 28	Axi-cel	CD 19	NA	NA	NA	No	R/R LBCL	≥ 18	103	SF-36, PROMIS-29	3
Elsawy 2022 22	Axi-cel	CD 19	ZUMA-7	NCT03391466	3	Yes	R/R LBCL	≥ 18	165	EORTC QLQ-C30, EQ-5D-5 L	15
Laetsch 2019 31	Tisa-cel	CD 19	ELIANA	NCT02435849	2	Yes	R/R BALL	≤ 21	58	PedsQL, EQ-5D-Y	12
Maziarz 2019 30	Tisa-cel	CD 19	JULIET	NCT02445248	2	Yes	R/R DLBCL	≥ 18	115	FACT-Lym, SF-36	18
Patrick 2021 26	Liso-cel	CD 19	TRANSCEND-NHL-001	NCT02631044	1	Yes	R/R LBCL	≥ 18	181	EORTC QLQ-C30, EQ-5D-5 L	12
Abramson 2022 24	Liso-cel	CD 19	TRANSFORM	NCT0357531	3	Yes	R/R LBCL	≤ 75	92	EORTC QLQ-C30, FACT-Lym	6
Wang 2020 29	Brex-cel	CD 19	ZUMA-2	NCT02601313	2	Yes	R/R MCL	≥ 18	65	EQ-5D-5 L	6
Shah 2021 25	Brex-cel	CD 19	ZUMA-3	NCT02614066	2	Yes	R/R BALL	≥ 18	55	EQ-5D-5 L	12
Delforge 2021 23	Ide-cel	BCMA	KarMMa	NCT03361748	2	Yes	R/R MM	≥ 18	126	EORTC QLQ-C30, QLQ-MY20, EQ-5D-5 L	18
Knight 2022 21	Bispecific LV20.19	CD 19, 20	NA	NCT03019055	1	No	R/R CLL, DLBCL, FL, MCL	≥ 18	15	IDAS, FSI, PSQI, BPI	3
Maillet 2021 27	Axi, Tisa	CD 19	NA	NA	NA	No	R/R DLBCL	≥ 18	27	HADS, QMRP, MMSE	7.6
Oswald 2022 19	CAR-T(NR)	CD 19, BCMA	NA	NA	NA	No	R/R MM, MCL, CLL, DLBCL	≥ 18	12	FACT-G, PROMIS-29 + 2 profile	3
Ram 2022 18	Axi, Tisa	CD 19	NA	NA	NA	Yes	R/R DLBCL	≥ 18	41	EORTC QLQ-C30	3
Martin 2022 20	Cilta-cel	BCMA	CARTITUDE-1	NCT03548207	2	Yes	R/R MM	≥ 18	68	EORTC QLQ-C30, EORTC QLQ-MY20, EQ-5D-5 L	16
Sidana 2022 17	CAR-T(NR)	CD 19, BCMA	NA	NA	NA	No	R/R myeloma, lymphoma, leukemia	≥ 18	34	FACT-G, NEUROQOL	6

Axi-cel, Axicabtagene ciloleucel; Tisa-cel, Tisagenlecleucel; Liso-cel, Lisocabtagene maraleucel; Brex-cel, Brexucabtagene autoleucel; Ide-cel, Idecabtagene vicleucel; Cilta-cel, Ciltacabtagene autoleucel.

NR, Not reported; NA, Not applicable; R/R, relpased/refractroy; LBCL, large B-cell lymphoma; BALL, B-cell acute lymphoblastic leukaemia; DLBCL, diffuse large B-cell lymphoma; MCL, mantle-cell lymphoma; MM, multiple myeloma; CLL, chronic lymphocytic leukemia; FL, follicular lymphoma.

^aSome studies used more than one CAR-T product.

PROM characteristics

Sixteen PROMs were identified, as follows: EuroQoL 5-Dimensional instrument 5-Level (EQ-5D-5 L), Youth friendly version of the EQ-5D (EQ-5D-Y), European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ), PROs Measurement Information System-29 (PROMIS-29), PROs Measurement Information System-29 and 2 items cognitive function (PROMIS-29 + 2), Short Form-36 (SF-36), Functional Assessment of Cancer Therapy-General (FACT-G), Functional Assessment of Cancer Therapy-Lymphoma (FACT-Lym), Pediatric Quality of Life (PedsQL), Fatigue Symptom Inventory (FSI), Brief Pain Inventory (BPI), Inventory of Depression and Anxiety Symptoms (IDAS), Hospital Anxiety and Depression Scale (HADS), Prospective and Retrospective Memory Questionnaire (PRMQ), Mini-Mental State Examination (MMSE), and Quality of Life in Neurological disorders-Short Form (Neuro-QoL-SF) (Table 2).

Table 2.

Characteristics of the included PROMs.

PROM	Full name	Study	Construct	Target population	Reference period	Subscale, No. of items	Response options	Score Range	Extracted data	Interpretation of extracted data score
EQ-5D-5 L	EuroQoL 5-dimensional instrument 5-level	Wang 2020 Delforge 2021 Shah 2021 Patrick 2021 Elsawy 2022 Martin 2022	Quality of life	Adults with hematologic cancer	Today	mobility(1), self care(1), usual activity(1), pain(1), anxiety(1), health status(1)	5-point adjectival scale	1–5 for each subscale 0-100 for health status	general health status	Higher score indicates better outcome
EQ-5D-Y	Youth friendly version of the EQ-5D	Laetsch 2019	Quality of life	Children and adolescents with hematologic cancer	Today	mobility(1), self care(1), usual activity(1), pain(1), anxiety(1), health status(1)	3-point adjectival scale	1–3 for each subscale 0-100 for health status	general health status	Higher score indicates better outcome
EORTC QLQ-C30	European Organization for Research and Treatment of Cancer Quality of Life Questionnaire	Elsawy 2022 Ram 2022 Delforge 2021 Patrick 2021 Abramson 2022 Martin 2022	Cancer-related quality of life	Adults with hematologic cancer	Past week	physical function(5), role function(2), cognitive function(2), emotional function(4), social function(2), financial impact of the disease(1), cancer symptom(12), global health status(2)	4-point adjectival scale	0-100	general health status pain fatigue depression social function cognitive function	Higher score indicates a higher response level
PROMIS-29	PROs Measurement Information System 29	Hoogland 2020 Barata 2022 Ruark 2019 Wang 2021	Quality of life	Adults with hematologic cancer	Past week	physical function(4), anxiety(4), depression(4), fatigue(4), sleep disturbance(4), social role(4), pain interference(4), pain intensity(1)	Measure-specific subscales: 5-point adjectival scale; pain intensity items: 10-point adjectival scale	1–5 for each subscale; 0–10 for pain intensity	pain fatigue anxiety depression social function	Higher score indicates a higher response level
PROMIS-29 + 2 profile v2.1	PROs Measurement Information System 29 and 2 items from cognitive gunction	Oswald 2022	Quality of life	Adults with hematologic cancer	Past week	physical function(4), anxiety(4), depression(4), fatigue(4), sleep disturbance(4), social role(4), pain interference(4), pain intensity(1), cognitive function(2)	Measure-specific subscales: 5-point adjectival scale; pain intensity items: 10-point adjectival scale	1–5 for each subscale; 0–10 for pain intensity	pain fatigue anxiety depression social function cognitive function	Higher score indicates a higher response level
SF-36	Short Fom-36	Hoogland 2020 Barata 2022 Maziarz 2019	Quality of life	Adults with hematologic cancer	Past 4 weeks	physical function(10), physical role(4), pain(2), general health(5), vitality(4), social function(2), emotional role(3), mental health(5), health change(1)	Variable adjectival scales	0-100	social function	Higher score indicates better outcome
FACT-G	Functional Assessment of Cancer Therapy-General	Oswald 2022 Sidana 2022 Maziarz 2019	Cancer-related quality of life	Adults with hematologic cancer	Past week	physical well-being(7), social well-being(7), emotional well-being(6), functional well-being(7)	5-point adjectival scale	0–4 for each subscale	general health status social function	Higher score indicates better outcome
FACT-Lym	Functional Assessment of Cancer Therapy-Lymphoma	Abramson 2022	Lymphoma-related quality of life	Adults with hematologic cancer	Past week	physical well-being(7), social well-being(7), emotional well-being(6), functional well-being(7), cancer symptom(15)	5-point adjectival scale	0–4 for each subscale	general health status	Higher score indicates better outcome
PedsQL	Pediatric Quality of Life Inventory	Laetsch 2019	Qulity of life in children and adolescents	Children and adolescents with hematologic cancer	Past 4 weeks	physical function(8), emotional function(5), social function(5), school function(5)	5-point adjectival scale	0–4 for each subscale	general health status	Higher score indicates better outcome
FSI	Fatigue Symptom Inventory	Knight 2022	Fatigue symptom	Adults with hematologic cancer	Past week	Severity (4) Frequency (2) Perceived interference (7) qualitative information(1)	11-point adjectival scale	0–10 for each subscale	fatigue	A high score represents a high level of fatigue
BPI	Brief Pain Inventory	Knight 2022	Pain	Adults with hematologic cancer	Past 1 day	pain severity(8), pain interference(7)	11-point adjectival scale	0–10 for each subscale	pain	A high score represents a high level of pain
IDAS	Inventory of Depression and Anxiety Symptoms	Knight 2022	Depression, anxiety	Adults with hematologic cancer	Past 2 week	dysphoria(10), well-bing(8), panic(8), lassitude(6), insomnia(6), suicidality(6), social anxiety(5), temper(5), traumatic intrusions(4), appetitie loss(3), appetite gain(3)	5-point adjectival scale	20–100	depression anxiety	A high score represents a high level of depression and anxiety
HADS	Hospital Anxiety and Depression Scale	Maillet 2021	Depression, anxiety	Adults with hematologic cancer	Past week	depression(7), anxiety(7)	4-point adjectival scale	0–3 for each subscale	depression anxiety	A high score represents a high level of depression and anxiety
PRMQ	Prospective and Retrospective Memory Questionnaire	Maillet 2021	Cognitive function	Adults with hematologic cancer	None	prospective memory(8), retrospective memory(8)	5-point adjectival scale	1–5 for each subscale	cognitive function	Higher score represents greater frequency of memory failures
MMSE	Mini-Mental State Examination	Maillet 2021	Cognitive function	Adults with hematologic cancer	None	orientation(10), registration(3), attention and calculation(5), recall(3), language(9)	1-point adjectival scale	0–30	cognitive function	Higher score represents a high level of cognitive function but a score of 23 or lower is indicative of cognitive impairment
Neuro-QOL-SF	Quality of life in neurological disorders-short form	Sidana 2022	Cognitive function	Adults with hematologic cancer	Past week	anxiety(8), depression(8), fatigue(8), extremity function(16), cognition(16), emotional and behavioral dyscontrol(8), well-being(9), sleep disturbance(8), social role(8), satisfaction(8), stigma(8)	5-point adjectival scale	1–5 for each subscale	cognitive function	Higher score indicates a higher response level

Data from the PRO measurements for each subscale assessment were as follows: (1) General health status (global health status of EORTC QLQ-C30, EQ-visual analogue scale (EQ-VAS), total QoL of FACT, general health of SF-36, total score of HRQol, total score of PedsQL); (2) Pain (pain of EORTC QLQ, pain of PROMIS-29, pain intensity of BPI, pain of MDASI); (3) Fatigue (Fatigue of EORTC-QLQ, fatigue of PROMIS-29, fatigue intensity of FSI, fatigue of MDASI); (4) Anxiety (anxiety of PROMIS-29, anxiety of IDAS, anxiety of HADS); (5) Depression (depression of PROMIS-29, depression of IDAS, depression of HADS); (6) Social function (social function of EORTC-QLQ, social well-being of FACT-G, social roles and activities of PROMIS-29, social function of SF-36); and (7) Cognitive function (cognitive function of EORTC-QLQ, cognitive function of PROMIS-29 + 2 profile, total score of PRMQ, total score of MMSE, cognition of Neuro-QoL-SF).

In most studies (n = 12), the PROs were assessed using more than one measurement. The EQ-5D-5 L (n = 6) and the EORTC-QLQ (n = 6) were the most frequently used PROMs.

Risk of bias evaluation using the COSMIN guideline

Of the 16 PROMs, 14 were evaluated using the Criteria for Good Measurement Properties proposed by the COSMIN guidelines (Table 3). Although no specific information on the validity and reliability of FACT-Lym has been provided, the finding is consistent with results from the validation of FACT-G³². Therefore, evaluation of FACT-Lym was not conducted, assuming that the quality values of FACT-G and FACT-Lym were similar. The PROMIS-29 + 2 profile was added to PROMIS-29 with two items of cognitive function; therefore, evaluation of the PROMIS-29 + 2 profile was not conducted.

Table 3.

Summary of the PROM Measurement Properties.

Instrument	Structural validity	Internal consistency	Reliability	Construct validity	Responsiveness	Score
EQ-5D-5L	indeterminate	indeterminate	sufficient	sufficient	sufficient	3
EQ-5D-Y	indeterminate	indeterminate	indeterminate	sufficient	sufficient	2
EORTC QLQ-C30	sufficient	indeterminate	sufficient	indeterminate	indeterminate	2
PROMIS-29	sufficient	indeterminate	insufficient	sufficient	sufficient	3
SF-36	indeterminate	sufficient	sufficient	indeterminate	sufficient	3
FACT-G	sufficient	sufficient	sufficient	indeterminate	indeterminate	3
PedsQL	sufficient	indeterminate	indeterminate	sufficient	sufficient	3
FSI	sufficient	sufficient	sufficient	sufficient	sufficient	5
BPI	indeterminate	sufficient	sufficient	sufficient	sufficient	4
IDAS	indeterminate	sufficient	sufficient	indeterminate	sufficient	3
HADS	indeterminate	sufficient	sufficient	sufficient	sufficient	4
PRMQ	sufficient	sufficient	sufficient	sufficient	sufficient	5
MMSE	indeterminate	indeterminate	sufficient	indeterminate	sufficient	2
Neuro-QOL-SF	sufficient	sufficient	sufficient	sufficient	sufficient	5

EQ-5D-5L, EuroQoL 5-dimensional instrument 5-level; EQ-5D-Y, Youth friendly version of the EQ-5D; EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire; PROMIS-29, PROs Measurement Information System 29; SF-36, Short Fom-36; FACT-G, Functional Assessment of Cancer Therapy-General; FSI, Fatigue Symptom Inventory; BPI, Brief Pain Inventory; IDAS, Inventory of Depression and Anxiety Symptoms; HADS, Hospital Anxiety and Depression Scale; PRMQ, Prospective and Retrospective Memory Questionnaire; MMSE, Mini-Mental State Examination; Neuro-QOL-SF, Quality of life in neurological disorders-short form

The three PROMs FSI, PRMQ, and Neuro-QoL-SF were rated at five points of “sufficient” in all the measurement properties. The BPI and HADS were rated at four points because structural validity was evaluated as “indeterminate.” The six PROMs EQ-5D-5 L, PROMIS-29, SF-36, FACT-G, PedsQL, and IDAS were rated at three points. The EQ-5D-Y, EORTC QLQ-C30, and MMSE were rated at the lowest points of two. Except for the EQ-5D-Y, EORTC QLQ-C30, and MMSE, 11 PROMs were evaluated as “sufficient” in more than three of the five measurement properties (structural validity, internal consistency, reliability, construct validity, and responsiveness). Therefore, we found that 11 PROMs had good reliability and validity. Twelve of 14 PROMs (78%) had adequate responsiveness. Of the total 14 PROMs, 7 (50%), 8 (57%), 11 (78%) and 9 (64%) were evaluated as adequate in structural validity, internal consistency, reliability, and construct validity, respectively.

Overall analysis

Overall, pain was significantly reduced at 1, 3, and 6 months compared to baseline (SMDs and 95% CIs: -0.24 (-0.32 to -0.16), -0.29 (-0.45 to -0.13), and − 0.43 (-0.71 to -0.15), respectively). General health status was no change at 1 month, but there was significant improvement at 3 months and more than 6 months (SMDs and 95% CIs: 0.57 (0.21 to 0.92) and 0.57 (0.34 to 0.81), respectively). Fatigue were no change at 1 month, but significantly decreased at 3 months and more than 6 months (SMDs and 95% CIs: -0.28 (-0.43 to -0.14) and − 0.66 (-0.91 to -0.41), respectively). Social and cognitive function remained unchanged for up to 3 months, but significantly improved after a prolonged period of 6 months (SMDs and 95% CIs: 0.32 (0.03 to 0.61) and 0.25 (0.14 to 0.37), respectively). Depression was no change at 1 month, but significantly decreased at 3 months (SMDs and 95% CIs: -0.31 (-0.55 to -0.07)). Anxiety did not change even for up to 3 months. Long-term results beyond 6 months of depression and anxiety were not available due to lack of data.

General health status

A total of 12 studies^{17–20,22–26,29–31}, including 924 patients, was included to analyze the effect of CAR-T on general health status. There was no difference in change from baseline for general health status at one month (SMD: 0.03, 95% CI: -0.12 to 0.17). However, this status improved significantly at three months (SMD: 0.57, 95% CI: 0.21 to 0.92) and at more than six months after therapy (or the long-term period) (SMD: 0.57, 95% CI: 0.34 to 0.81) (Fig. 2).

Fig. 2.

Forrest plots with standard mean differences and corresponding 95% confidence intervals in seven domains of PRO in patients with hematologic malignancies treated with CAR-T.

Pain

A total of 8 studies^{19–24,26,28}, including 712 patients, was included to analyze the effect of CAR-T on pain. Pain significantly decreased one month after therapy (SMD: -0.24, 95% CI: -0.32 to -0.16), and decreased even more significantly at three and six months after therapy (SMD: -0.29, 95% CI: -0.45 to -0.13; SMD: -0.43, 95% CI: -0.71 to -0.15, respectively) (Fig. 2).

Fatigue

A total of 8 studies^{19–24,26,28}, including 712 patients, was included to analyze the effect of CAR-T on fatigue. There was no difference in fatigue at one month compared to baseline (SMD: 0.16, 95% CI: -0.02 to 0.34), but it significantly decreased at three months and during the long-term period more than six months after therapy (SMD: -0.28, 95% CI: -0.43 to -0.14; SMD: -0.66, 95% CI: -0.91 to -0.41, respectively) (Fig. 2).

Anxiety

A total of 4 studies^19,21,27,28, including 157 patients, was included to analyze the effect of CAR-T on anxiety. There was no difference in anxiety at one and three months compared to baseline (SMD: 0.25, 95% CI: -0.11 to 0.61; SMD: -0.09, 95% CI: -0.35 to 0.18, respectively). Due to the lack of data, anxiety could not be analyzed during the long-term period more than six months after therapy (Fig. 2).

Depression

A total of 8 studies^{19–23,26–28}, including 689 patients, was included to analyze changes in depression symptoms after CAR-T. There was no difference in depression at one month compared to baseline (SMD: -0.03, 95% CI: -0.39 to 0.32), but it improved significantly three months after therapy (SMD: -0.31, 95% CI: -0.55 to -0.07). Due to the lack of data, depression could not be analyzed during the long-term period more than six months after therapy (Fig. 2).

Social function

A total of 7 studies^{17,19,20,22,23,26,30}, including 686 patients, was included to analyze the effect of CAR-T on social function. There was no difference in social function at one and three months compared to baseline (SMD: -0.16, 95% CI: -0.33 to 0.01; SMD: 0.12, 95% CI: -0.04 to 0.28, respectively), but it improved significantly during the long-term period more than six months after therapy (SMD: 0.32, 95% CI: 0.03 to 0.61) (Fig. 2).

Cognitive function

A total of 8 studies^{17,19,20,22–24,26,27}, including 652 patients, was included to analyze the effect of CAR-T on cognitive function. There was no difference in the change from baseline for cognitive function at one and three months (SMD: 0.05, 95% CI: -0.06 to 0.16; SMD: 0.17, 95% CI: -0.01 to 0.34, respectively), but it improved significantly during the long-term period more than six months after therapy (SMD: 0.25, 95% CI: 0.14 to 0.37) (Fig. 2).

Publication bias

There was no difference between the contour-enhanced funnel plot corrected by the trill and fill method regarding the general health status outcome (Supplementary 5). Publication bias was not suspected.

Discussion

CAR-T cells have become an established treatment for patients with relapsed and/or refractory B cell lymphomas, B cell acute lymphoblastic leukemia, or multiple myeloma. CAR-T cell therapy can be a revolutionary treatment for metastatic cancer patients with limited treatment options. However, CAR-T cell therapies are among the most expensive cancer therapies to date, with list prices of approximately $373,000 US dollars and 320,000€ in Europe³³. Moreover, they are also associated with substantial risks and psychological and financial burdens for patients and caregivers given their complex clinical and infrastructural challenges³⁴. Therefore, it is important to better understand the impact of CAR-T cell therapies from the patient’s standpoint, alongside its clinical efficacy and safety.

PROMs are crucial in pharmacological research for evaluating treatment effects in conditions such as cancer, where quality of life is a major concern. Well-developed PROMs can capture various treatment effects across multiple domains, providing valuable insights into disease-specific outcomes. A wide variety of PROMs is available for measuring PROs, which can be categorized into two main types: generic and disease-specific measures. Generic measurements, such as EQ-5D, SF-36 and PROMIS, are used across various conditions, while disease-specific measurements, like EORTC-QLQ and FACT-G, are tailored to particular diseases, such as cancer. These measurements are well-developed and validated. In our meta-analysis, EQ-5D and EORTC-QLQ were the most commonly utilized.

The current meta-analysis included 15 studies and categorized symptom changes after receiving CAR-T into general health status, pain, fatigue, anxiety, depression, social function, and cognitive function. The findings showed that pain significantly improved from one month after CAR-T cell therapy. In addition, pain tended to improve more over time. Pain is frequently experienced by patients with hematological cancer and is closely related to quality of life. Therefore, a significant reduction in pain one month after CAR-T and a further reduction in pain over time are very meaningful changes with regard to quality of life.

Although the patients’ general health status did not improve one month after CAR-T, it improved significantly thereafter. Similarly, fatigue worsened at one month but then improved over time. This finding aligns with the previous review by Efficace et al.¹⁰, which reported improvements in physical functioning and reductions in fatigue over time following CAR-T cell infusion. This result might reflect the toxicity of CAR-T. The two most frequently observed toxicities of CAR-T are CRS and ICANS. Emma et al. reported that 55.3% of 594 patients treated with CAR-T cell therapy in 18 hematological cancer studies experienced CRS, and 37.2% of 296 patients treated with CAR-T cell therapy in 14 hematological cancer studies experienced neurotoxicity³⁵. The median time to develop CRS is within the first week following CAR T-cell infusion, and it typically resolves within 7–8 days. Most neurotoxicity resolves within four weeks of CD19 CAR T-cell therapy³⁶. This is consistent with the significant improvements in general health status and fatigue from one month after CAR-T.

Anxiety and depression remained unchanged or slightly worsened at the early stage of CAR-T but improved significantly three months after therapy. It is possible that patient anxiety increases within one month of therapy due to the toxicities of CAR-T. Anxiety was only analyzed at one and three months due to a lack of data, and there was no significant difference compared to baseline. It is necessary to evaluate changes in anxiety symptoms in hematological cancer patients receiving CAR-T through further research. Functional aspects, such as social function and cognitive function, did not recover during the early stage of CAR-T but improved over time.

The MCID represents the smallest improvement considered worthwhile by the patient. Another definition of MCID is a difference score that is large enough to have an implication for the patient’s treatment or care³⁷. The MCID of the EORTC QLQ-C30 for cancer patients was reported to be 10 points or more from baseline^38,39. According to the current findings, the weighted mean difference of the EORTC QLQ-C30 before and after CAR-T cell therapy was 10 with a 95% CI of 3 to 17. The MCID of EQ-VAS has ranged from 5.93 to 6.99 in patients with cervical intraepithelial neoplasia⁴⁰. The weighted mean difference of EQ-VAS before and after CAR-T cell therapy was 10 with a 95% CI of 4 to 17 in the current findings. These results can be interpreted as a minimal clinically meaningful improvement in the general health status from three months after CAR-T cell treatment in patients with hematologic malignancies.

Our study has several limitations. First, the analysis was conducted using data from patients who were followed, because the primary outcome of all studies included in the analysis was not the PRO. This methodology may have introduced selection bias because it is possible that only the PRO of patients who responded to CAR-T were collected. Second, because the sample size of the studies was calculated to confirm the clinical efficacy of CAR-T, the power of the meta-analysis using PRO might be weakened. Nevertheless, the MCIDs of EORTC QLQ-C30 and EQ-VAS showed that CAR-T cell therapy resulted in clinically meaningful improvements. Therefore, if a meta-analysis is conducted with studies in which PRO is the primary outcome, it is possible that the estimated effect size would be larger. Third, among the studies included in the analysis, the longest FU period was 18 months, and most of them were less than one year. The FDA recommends long-term follow-up for up to 15 years after CAR-T administration. Therefore, further analysis is needed to determine the effect of CAR-T through long-term follow-up. Fourth, there are many single-arm studies of patients with relapsing or refractory hematologic malignancies, and placebo-controlled trials are insufficient. Therefore, it is inevitable to compare the results before and after the intervention, not the control group. To date, no meta-analysis studies have used PRO and this study fills this evidence gap by quantifying the effect of CAR-T cell therapy using PRO.

Conclusion

The current meta-analysis indicates that CAR-T cell therapy for hematologic malignancies improves PROs in six domains including general health status, pain, fatigue, depression, social function, and cognitive function. General health status might be a minimal clinically meaningful improvement from three months after CAR-T cell therapy in patients with hematologic malignancies. Although there is controversy as to whether CAT-T cell therapy is cost-effective, it has reported efficacy in patients with hematologic malignancies with unmet medical need. The current meta-analysis adds evidence that CAR-T cell therapy produces clinically meaningful differences in patient-reported outcomes. These results suggest that the professional’s perspective and patient’s values and preferences should be balanced in determining to use CAR-T cell therapy for patients with hematologic malignancies.

Authors’ contributions

HP and HJ participated in design and implementation of search strategies, eligibility screening, protocol preparation, data extraction, quality assessment, data analysis and interpretation, and writing the manuscript; HJ participated in the scientific concept, eligibility screening, data extraction, quality assessment, interpretation of data, scientific guidance, and critical review of the manuscript; HWY participated in clinical guidance, comments on protocol, and critical review of scientific content and edited the manuscript; NJK implementation of search strategies and searched data base as an expert of medical librarian. All authors read, edited, and approved the final manuscript.

Funding

None.

Data availability

Please contact the corresponding author if you would like to see any data that are not included in the Article or the Appendix.

Declarations

Ethics approval and consent to participate

The Catholic University of Korea institutional review board (IRB) determined that this study has been granted an exemption from requiring ethics approval by the IRB as this literature review did not involve the participation of human subjects.

Consent for publication

All authors consent for publication.

Competing interests

The authors declare no competing interests.