FormalPara Key Points for Decision Makers

While the systematic literature review (SLR) assessments showed a wide range of reporting parameters and conventions, chimeric antigen receptor (CAR) T cell therapies, novel agents, rituximab, and a variety of salvage chemotherapies were collated. The most consistent and broadest range of health-related quality of life (HRQOL) outcomes and health utility values were for CAR T cell therapies, which showed overall improvements/benefits.

The perpetually evolving treatment landscape requires periodic consolidation of HRQOL and health utility value data to inform treatment decision-making. Consistency in reporting of these studies would be more useful for SLRs such as this one.

1 Introduction

Non-Hodgkin lymphoma (NHL) is one of the most common forms of cancer worldwide, with incidence rates in 2020 of 6.9 and 4.8 per 100,000 in males and females, respectively [1]. Diffuse large B cell lymphoma (DLBCL) represents the most common NHL subtype of large B cell lymphoma (LBCL) worldwide [2]. First-line treatment for large B cell lymphoma (LBCL) generally comprises rituximab-containing immunochemotherapy regimens, most commonly rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) [3, 4]. Standard first-line therapy with immunochemotherapy regimens such as R-CHOP is considered highly effective. Approximately 70% of patients with DLBCL will be event-free at 2 years and have overall survival (OS) equivalent to the general population from that point forward [5].

For those patients with relapsed or refractory (R/R) LBCL (~30% with DLBCL), second-line therapy generally entails a rituximab-based multiagent regimen to induce a complete response, followed by high-dose chemotherapy and autologous hematopoietic stem cell transplantation (HSCT) with or without involved site radiation therapy [3, 4]. Recently, chimeric antigen receptor (CAR) T cell therapies have been shown to be new assets in the second-line or later treatment armamentarium [6, 7]. The third-line or later setting utilizes a variety of small-molecule (conventional), cellular, and targeted molecular therapies [3, 8, 9].

Health-related quality of life (HRQOL) instruments are useful in monitoring patients’ experiences of both disease and treatment. Patients with LBCL have a decreased HRQOL [10, 11]. The impact of treatment on HRQOL varies by therapeutic modality and has been highlighted as an important factor to consider in treatment decision-making in LBCL [12, 13]. In the R/R treatment setting, patients with LBCL may have cumulative toxicities from prior therapies [14], which may also contribute to HRQOL burden.

The HRQOL instruments used in LBCL are varied, and include the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30), the health state index score and EuroQol visual analog scale (VAS) from the EuroQol 5-dimension 5-level (EQ-5D-5L), the Functional Assessment of Cancer Therapy-General (FACT-G), the FACT lymphoma (FACT-Lym) subscale, and the 36-item Short Form Health Survey (SF-36) [10, 11, 15,16,17,18,19,20].

Health state utility measures of HRQOL are patient derived using multi-attribute, preference-based values that indicate the effect on the patient’s overall health status. In clinical trials, utility measures summarize both positive and negative effects of an intervention, on a scale of 0 (dead) to 1 (full health). These measures allow for comparison of overall patient outcomes across different diseases, and for comparison between various health care interventions. Utility values are used by decision-makers to determine reimbursement, which may have an impact on the availability of certain treatments [21].

A number of new agents, combinations, and regimens are currently under investigation for the treatment of R/R LBCL [4, 5, 19, 22,23,24,25,26,27,28,29]. In this ever-expanding treatment landscape, there is need for consolidated HRQOL evidence in R/R LBCL to help patients and physicians make treatment decisions. The current systematic literature review (SLR) was performed to understand the HRQOL associated with treatments in R/R LBCL and identify associated health utility values.

2 Methods

2.1 Study Design and Search Process

This unregistered SLR was designed, conducted, and reported using best practices in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and the Preferred Reporting Items for Systematic reviews and Meta-Analyses statement [30, 31]. The initial search was conducted on 5 February 2021, and updated on 2 May 2022. The search and modifications to the protocol aligned with Population, Intervention, Comparator, Outcome, Study design (PICOS) criteria [32] to investigate the research question. The database search was developed in Ovid MEDLINE (Supplementary Table S1) by an experienced information specialist and underwent a Peer Review of Electronic Search Strategies (PRESS) [33] analysis by a second information specialist. The search syntax was then adapted in Ovid across Embase, the Cochrane Database of Systematic Reviews, the Centre for Reviews and Dissemination Health Technology Assessment (HTA) database, and the National Health Service Economic Evaluation Database. Single searches of conferences and gray literature sources were also conducted (Supplementary Table S2). Protocol requests can be requested using the following website: https://www.bms.com/researchers-and-partners/independent-research/data-sharing-request-process.html.

2.2 Eligibility Criteria

Studies were screened based on PICOS criteria established a priori (Table 1) and were assessed by two independent reviewers. Study screening of the database search was conducted in the following two stages in DistillerSR (Evidence Partners Inc., Ottawa, ON, Canada): (1) review of titles and abstracts and (2) review of full-text articles. Exclusion reasons were recorded in detail during the full-text screening stage. Searches for gray literature including conference abstracts and hand searches of bibliographies of published SLRs were conducted by a single reviewer and verified by a second reviewer. Conflicts for both the PICOS and gray literature reviews were resolved by consensus through discussion or a third reviewer.

Table 1 PICOS criteria
Full size table

Randomized and nonrandomized studies and economic evaluations reporting health state utility values or HRQOL measures were eligible. Assessments from HTA agencies with full reviewer’s reports available from the National Institute for Health and Care Excellence (NICE; UK), Canadian Agency for Drugs and Technologies in Health (CADTH), Pharmaceutical Benefits Advisory Committee (PBAC, Australia), and Scottish Medicines Consortium (SMC) were also eligible to ensure the most complete set of analysis results supporting HTA recommendations were captured. Studies were not limited by sample size.

2.3 Data Extraction and Quality Assessment

Data extraction was performed using a standardized form implemented in Microsoft Excel. For each study, data were extracted by one reviewer and verified by a second independent reviewer.

The quality of published trials was assessed using a NICE-recommended checklist [34] for appraising review articles to ensure search and resulting analyses are unbiased and of sufficient quality to guide policy and practice. Quality assessments were conducted by a single reviewer and validated by a second reviewer. Conflicts were resolved by a third reviewer when the two reviewers did not reach an agreement. Studies were not excluded on the basis of the quality assessment. Quality assessments for the risk of bias are summarized in Supplementary Table S3.

3 Results

3.1 SLR Search Results

The initial SLR identified 1072 records and an additional 84 when updated (Fig. 1 and Supplementary Fig. S1). The initial gray literature search identified 8244 additional records for consideration and 6476 more when updated. After removing duplicates, screening, and searching reference lists and Supplementary Material, a total of 33 unique studies reporting HRQOL outcomes were included in the qualitative synthesis. Studies included in this analysis that reported HRQOL measures or health state utility values are summarized in Supplementary Table S4.

Fig. 1
figure 1

PRISMA flow diagram. CDC Centers for Disease Control and Prevention; CEVR Center for Evaluation of Value and Risk in Health; EMA European Medicines Agency; Embase Excerpta Medica dataBASE; EQ-5D EuroQol 5-dimensional; EudraCT European Union Drug Regulating Authorities Clinical Trials Database; FDA United States Food and Drug Administration; HERC Health Economics Research Centre; HRQOL health-related quality of life; HTA health technology assessment; MA meta-analysis; MEDLINE Medical Literature Analysis and Retrieval System Online; NMA network meta-analysis; NHS EED National Health Service Economic Evaluation Database; PRISMA Preferred Reporting Items for Systematic reviews and Meta-Analyses; SLR systematic literature review; WHO CT World Health Organization Clinical Trials. aConferences searched included American Association for Cancer Research (n = 426), American Society of Clinical Oncology (n = 313), American Society of Hematology (n = 745), European Hematology Association (n = 178), European Society for Medical Oncology (n = 33), International Society for Pharmacoeconomics and Outcomes Research (n = 50). bSources of HTAs searched included Canadian Agency for Drugs and Technologies in Health (n = 22), Health Technology Assessment International (n = 0), Institut National d’Excellence en Santé et en Services Sociaux (n = 64), National Institute for Health and Care Excellence (n = 73), Pharmaceutical Benefits Advisory Committee (n = 13), and Scottish Medicines Consortium (n = 132)

Full size image

3.2 HRQOL

Seven studies were identified that assessed and reported disease-specific HRQOL measures (Table 2).

Table 2 Overview of included studies on HRQOL measures
Full size table

3.2.1 CAR T Cell Therapies

Three studies reported HRQOL data in patients with LBCL treated with CAR T cell therapies, including lisocabtagene maraleucel (liso-cel; n = 186 [17] and 184 [18] evaluable patients) and tisagenlecleucel (n = 108 [19]). In the phase 1, seamless design, single-arm trial of liso-cel (TRANSCEND NHL 001 [17]), the EORTC QLQ-C30 global health status instrument was used to assess HRQOL in patients with third-line or later LBCL who received liso-cel [35,36,37]. Baseline HRQOL scores improved by + 17.5 points at the month-12 follow-up after treatment.

The TRANSFORM trial [18] used the FACT-Lym subscale and EORTC QLQ-C30 to evaluate HRQOL in patients with LBCL treated with liso-cel or standard of care (SOC; platinum-based immunochemotherapy followed by carmustine, etoposide, cytarabine, and melphalan, and autologous HSCT in responders) as second-line therapy [18]. Mean change at day 126 (2 months after treatment) from baseline was reported. Improvements in mean scores were observed for both instruments for liso-cel and SOC. For liso–cel–treated patients, the mean [95% confidence interval (CI)] FACT-Lym increase was + 1.48 (0.30‒3.26) and the mean EORTC QLQ-C30 increase was + 3.08 (− 1.83 to 7.99); for SOC-treated patients, the mean (95% CI) FACT-Lym increase was + 1.63 (0.41‒3.68) and the mean (95% CI) EORTC QLQ-C30 increase was + 0.04 (− 5.24 to 5.31).

In the JULIET trial, change in HRQOL in patients with LBCL after treatment with tisagenlecleucel in the third-line or later setting was measured using the FACT-G, FACT-Lym, and SF-36 instruments [19]. HRQOL improved from baseline to 18-month follow-up across all instrument subscales, with mean change scores ranging from + 3.1 to + 13.1 points in the FACT instrument scales and + 2.3 to + 4.3 for the SF-36 subscales.

3.2.2 Novel Therapies

The phase 2 single-arm SADAL trial (selinexor) assessed HRQOL in responders and nonresponders using the FACT-G treatment satisfaction, FACT-Lym, and FACT-Lym Trial Outcome Index (TOI) [38]. Outcomes were assessed at baseline, treatment cycles 2‒7, and the end of treatment. All patients reported decreased HRQOL at last follow-up, with responders reporting mean change scores from baseline of − 9.9 to − 6, and nonresponders reporting mean change scores of − 9.2 to − 15.7.

In the phase 3 ORCHARRD trial of ofatumumab plus dexamethasone, cytarabine, and cisplatin (DHAP) chemotherapy versus rituximab plus DHAP (R-DHAP) therapy for second-line LBCL, HRQOL was assessed using the FACT-G and FACT-Lym TOI instruments [39]. In patients treated with ofatumumab, the mean [standard error (SE)] FACT-G total score and FACT-Lym TOI decrease was − 2.561 (0.7671) and − 2.028 (0.9196), respectively. In patients treated with R-DHAP, the mean (SE) FACT-G total score and FACT-Lym TOI decrease was − 2.591 (0.7696) and − 3.156 (0.9204), respectively.

3.2.3 Salvage Therapies

One phase 1/2, single-arm trial evaluating the clinical efficacy of rituximab plus cyclophosphamide, etoposide, and prednisone (R-CVEP) used the FACT-G and FACT-Lym instruments to measure HRQOL [40]. HRQOL improved from baseline to 12-month follow-up across all scales, with a mean change in FACT-G total score of + 11.72, FACT-Lym total score of + 18.61, and FACT-Lym TOI of + 12.29.

3.2.4 EORTC QLQ-C30 by Line of Therapy

One identified study reported HRQOL by line of therapy in a real-world setting for patients with DLBCL via a self-completed point-in-time survey using the EORTC QLQ-C30. The reported EORTC QLQ-C30 global health status score was higher for patients on second-line therapies (55.48 [18.02]) than for those on third-line and beyond therapies (49.93 [21.07]) [41].

3.3 Utilities

Six studies were identified that assessed and reported health state utility values (Table 3).

Table 3 Overview of included studies on health statue utility values
Full size table

3.3.1 CAR T Cell Therapies

Three studies reported HRQOL data in patients with LBCL treated with CAR T cell therapies as follows: liso-cel (TRANSCEND NHL 001; n = 186 evaluable patients [17]), axicabtagene ciloleucel (axi-cel; ZUMA-1; n = 34 [42]), and a product-agnostic vignette-based time tradeoff study [43, 44]). In TRANSCEND NHL 001, the health utility impact of liso-cel was evaluated in eligible patients with DLBCL in the third-line or later setting using the validated EQ-5D-5L questionnaire [17, 35,36,37]. The questionnaire includes the EQ-5D-5L descriptive system and EQ-VAS. Health utilities were assessed at baseline and months 1, 2, 3, 6, 9, 12, and 18 after infusion. At baseline, the mean [standard deviation (SD)] EQ-5D-5L health state score was 0.817 (0.120). This decreased slightly at month 1 (− 0.023) and then increased from months 2 to 12 (+ 0.020 to + 0.031). At month 18, the increase from baseline in the EQ-5D-5L index score was + 0.061 (0.013). The mean (SD) score for EQ-5D-5L VAS at baseline was 68.3 (19.5), and the increase from baseline was + 10.4 (5.4) at month 18.

A phase 2 ZUMA-1 safety management study ad hoc analysis investigated the impact of axi-cel treatment on health utility patients with R/R LBCL in the third-line or later setting, as measured by the EQ-5D-5L [42]. Health utilities were assessed at screening, week 4, and months 3 and 6 after infusion. Mean (SD) EQ-5D-5L score at baseline was 0.80 (0.17). Scores at week 4 decreased slightly [0.74 (0.15); − 0.06] and then increased by month 3 [0.80 (0.13); + 0.00] and month 6 [0.82 (0.21); + 0.02]. Mean (SD) utility values by health state were 0.80 (0.14) for progression-free disease and 0.72 (0.17) for progressed disease. A disutility of 0.05 (SE, 0.04) was observed at 4 weeks, and the authors noted this was associated with the timing of CAR T cell–related adverse events (AE).

In a vignette-based time tradeoff study [43, 44], adults from the general population in the UK were surveyed to estimate the health utility impact of all-grade cytokine release syndrome (CRS) and neurological events (NE) related to CAR T cell treatment for R/R LBCL. The EQ-5D-5L was self-administered, and participants were also asked to value the following six health states: one state describing CAR T cell therapy for LBCL with no AEs, three states describing CAR T cell therapy for LBCL with varying grades of CRS, and two states describing CAR T cell therapy for LBCL with varying grades of NEs. A total of 218 participants completed the interview. The mean (SD) EQ-5D-5L index score was 0.86 (0.17) and the mean (SD) EQ-5D-5L VAS score was 82.15 (13.54) for the general population. Among the six health state vignettes valued by the general population respondents, the highest mean utility was for the health state representing CAR T cell therapy with no AEs. Disutilities per AE included grade 1 CRS (− 0.01), grade 1/2 NEs (− 0.04), grade 2 CRS (− 0.05), grade 3/4 NEs (− 0.18), and grade 3/4 CRS (− 0.23).

3.3.2 Novel Therapies

The phase 2 single-arm SADAL trial evaluated the health utility impact of selinexor in the third-line or later setting in eligible patients with DLBCL using the validated EQ-5D-5L questionnaire [38]. Health utilities were assessed in responders and nonresponders at baseline and at cycles 2‒7. At baseline (n = 89 evaluable patients), the mean (95% CI) EQ-5D health state score was 0.789 (0.767‒0.811) for responders and 0.801 (0.781‒0.822) for nonresponders. Responders reported a mean change from baseline of − 0.010 at last follow-up, whereas nonresponders reported a mean change from baseline of − 0.274.

In the phase 1/2 CheckMate 436 trial, HRQOL was reported as an exploratory endpoint in patients with primary mediastinal B cell lymphoma treated with nivolumab combined with brentuximab vedotin in the third-line or later setting using the EQ-5D-3L [45]. HRQOL was reported at baseline, cycle 5, and for the first follow-up visit. The study reported the proportion of patients experiencing “some problems” or “extreme problems” for each dimension but did not map the results to preference-based health state utilities. The percentage of patients experiencing “some problems” with activity and pain appeared to decrease over time (− 18% and − 50%, respectively), and no clear pattern could be seen for the remaining dimensions.

The impact of naratuximab emtansine and rituximab in patients with R/R DLBCL was evaluated in a phase 2 single-arm trial using the FACT-Lym questionnaire mapped to EQ-5D index values [46]. At baseline, the mean estimated EQ-5D index score was 0.78 for responders and 0.73 for nonresponders. At the end of treatment, mean (difference) index scores were 0.77 (− 0.01) for responders and 0.67 (− 0.06) for nonresponders.

3.4 Health State Utility Values Reported in Economic Studies and HTA Submissions

A total of 15 economic studies and seven HTA reports were identified that reported health state utility values used to inform comparative analyses of therapies for patients with R/R LBCL (Table 4). Reporting of utility values varied among HTA reports (n = 7), with six reporting values associated with progression-free and progressed states and one reporting disutility values for specified AEs (Table 5). Sources for utility values were typically clinical trials, as well as published literature. Two reports (both from the SMC) reported mapping SF-36 data to EQ-5D.

Table 4 Overview of economic studies reporting health state utility values
Full size table
Table 5 Health state utility values included in HTA reports
Full size table

3.4.1 CAR T Cell Therapies

Eleven studies reported health state utility values used to inform cost-effectiveness models comparing CAR T cell therapies with other CAR T cell therapies (three studies) [47,48,49] or salvage chemotherapy (eight studies) [50,51,52,53,54,55,56,57] in patients with DLBCL in the third-line or later setting. These studies were conducted from a number of different perspectives, including US payer perspective [50, 51, 56, 57], Spanish National Health Service perspective [47], Singapore health care system perspective [53], Singapore private insurance payer perspective [58], Chinese health care system perspective [54], and Japanese public health care payer perspective [55]. Across these 11 studies, utility values for remission/progression-free survival ranged from 0.70 to 0.90 [50, 58], and values for disease progression ranged from 0.39 to 0.59 [47, 48, 51, 53,54,55,56,57, 59, 60]. Treatment-related utility values for CAR T cell therapies ranged from 0.50 to 0.740 [47, 48, 50, 51]. Disutilities related to treatment and AEs were reported for tisagenlecleucel: two studies [55, 57] reported treatment disutility of − 0.15 over a duration of 28 days, while disutilities related to intensive care unit stays for CRS ranged from − 0.70 to − 0.90 across three studies [53, 57, 58].

3.4.2 Novel Therapies

Among three cost-effectiveness analyses conducted from a US payer perspective that compared polatuzumab vedotin, bendamustine, and rituximab (pola-BR) with BR (two studies) [60, 61] or tafasitamab plus lenalidomide (one study) [59], utility values were 0.83 for progression-free survival [47, 48, 50, 51, 53,54,55, 57,58,59,60,61,62] and ranged from 0.39 to 0.71 for disease progression [47, 48, 50, 51, 53,54,55, 57,58,59,60,61,62]. AE-related disutility values were − 0.02 for pola-BR and − 0.01 for tafasitamab plus lenalidomide [59].

3.4.3 Salvage Therapies

One study reported health state utility values for pixantrone compared with current clinical practice (vinorelbine, oxaliplatin, ifosfamide, etoposide, mitoxantrone, and gemcitabine) in patients with aggressive R/R non-Hodgkin lymphoma [62]. Utility values were reported for preprogression (0.76) and progressive disease (0.68). For pixantrone-treated patients, disutilities were reported for grade 2 AEs (− 0.0075) and grade 3/4 AEs (− 0.0078). For patients treated with current clinical practice, disutilities were reported for grade 2 AEs (− 0.0066) and grade 3/4 AEs (− 0.0073).

Treatment-related utility and disutility values were reported for salvage chemotherapy in eight studies [50, 51, 53,54,55,56,57,58] and for HSCT in five studies [50, 54,55,56, 58]. On-treatment utility values for salvage chemotherapy ranged from 0.63 to 0.67 [50, 51]. Disutilities for salvage chemotherapy ranged from − 0.42 to − 0.15 [53,54,55, 57, 58], while reported disutility values for HSCT ranged from − 0.57 to − 0.30 [55, 56].

3.5 Quality Assessment of HRQOL and Utility Evidence

Six studies (five clinical trials and one vignette-based study) were assessed using the NICE quality assessment criteria for health state utility values. The results are summarized in Supplementary Table S2. Overall, sources of potential bias in the six studies centered around study population (e.g., self-selected participants [41]), small sample size [38, 40, 48], and the presence and handling of missing data (e.g., reasons for loss to follow-up not reported [19, 37, 38] or loss to follow-up unaccounted for in analysis [19, 37]).

4 Discussion

Overall, the evidence synthesized in this review provides a comprehensive synthesis of HRQOL and health state utility evidence for treatments of aggressive R/R LBCL. Analyses were identified for CAR T cell therapies; novel agents such as pola-BR, ofatumumab, and selinexor; and salvage therapies. However, studies were heterogeneous in terms of methods and outcomes reported.

All studies reporting health state utility values reported data for the EQ-5D. Treatment of R/R LBCL with CAR T cell therapies liso-cel, axi-cel, and tisagenlecleucel in the third-line or later setting was associated with improvement in EQ-5D utility values from baseline. Similar improvements in general and lymphoma-specific HRQOL measures (i.e., FACT-G, FACT-Lym, SF-36, and EORTC QLQ-C30 scales) were observed with liso-cel in both the second- and third-line or later settings and for tisagenlecleucel in the third-line or later setting. HTAs varied in their recommendations, sometimes for the same product (e.g., pola-BR).

This SLR has several strengths. It was designed, conducted, and reported using best practices in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and the Preferred Reporting Items for Systematic reviews and Meta-Analyses statement [30, 31]. This includes the use of a prespecified PICOS question design, comprehensive database literature search, supplementary searches of bibliographies and gray literature, standardized approach to study selection and data extraction with screening by two independent reviewers, and a rigorous quality assessment process. It was designed to include not only comparative HRQOL evaluations but also utility studies for a more complete picture of the available evidence for therapies in aggressive R/R LBCL.

The scope of this SLR was designed to collect the most relevant evidence for R/R LBCL. The time limitation of 2003 was chosen because the first trial for rituximab, the SOC in newly diagnosed LBCLs, was published in 2002 [3, 63]. As rituximab was not yet established as the SOC for newly diagnosed lymphomas, studies published before 2003 were not likely to include patients with third-line or later lymphoma who were treated with both an anthracycline-containing regimen and a rituximab (or other CD20-targeted agent)-containing regimen.

Challenges encountered during the SLR that limited direct comparison of findings across studies included between-study heterogeneity of populations and methodologies, as well as inconsistent outcome reporting (i.e., outcome definitions, HRQOL measures, follow-up periods, etc.). Despite these limitations, the evidence synthesized in this SLR provides a comprehensive understanding of the HRQOL evidence in R/R LBCL and has identified several sources for utility values in the published literature.

Of note, studies published before and after the 2016 revision of the World Health Organization classification of lymphoid neoplasms may not be directly comparable [64]. Before this revision, LBCLs with MYC and BCL2 and/or BCL6 rearrangements were considered as “double-/triple-hit lymphomas” and categorized under DLBCL not otherwise specified (NOS) [64, 65]. The 2016 revision reclassified all LBCLs with MYC and BCL2 and/or BCL6 rearrangements in a single category of “high-grade B cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements,” a distinct entity from DLBCL NOS [64]. Therefore, it is unknown whether studies conducted with patients with DLBCL NOS before this revision included patients with these high-grade B cell lymphomas, which may have impacted the results.

Overall, the evidence synthesized in this SLR provides a comprehensive understanding of the HRQOL evidence in R/R LBCL. A number of cost utility analyses were identified for CAR T cell therapies; novel agents such as pola-BR, ofatumumab, and selinexor; and salvage therapies, along with numerous sources for utility values. Consistency in reporting would be beneficial for analyses such as this one, and more HRQOL studies in R/R LBCL are needed to better understand the impact of new therapies on HRQOL.