Evaluating process utilities for the treatment burden of chemotherapy in multiple myeloma in Japan: a time trade-off valuation study

Abstract

Aims

This study estimated the “process utilities” of treatment options for patients with relapsed/refractory multiple myeloma (RRMM) in Japan using the time trade-off (TTO) method. Chimeric antigen receptor (CAR) T cell immunotherapy is available for patients with RRMM who are triple-class exposed (TCE) after treatment with immunomodulatory agents, proteasome inhibitors, and anti-CD38 monoclonal antibodies. However, the impact of available treatment options on health state utilities has not been well characterized, particularly in relation to process utilities.

Methods

Eight vignettes of health states and daily activity restrictions related to each of the following RRMM therapies were prepared: no treatment, CAR T cell therapy with idecabtagene vicleucel (ide-cel), regular intravenous infusion, and oral administration. A face-to-face survey of healthy Japanese adults who were representative of the general population was conducted. The TTO method was used to evaluate each vignette and to generate utility scores for each treatment regimen.

Results

Three hundred and nineteen respondents participated in the survey (mean age: 44 years [range: 20–64]; female: 50%). Utility scores for no treatment, ide-cel, oral pomalidomide, and dexamethasone (Pd) therapy ranged from ∼0.7 to 0.8. Utility scores for regular intravenous infusion regimens ranged from 0.50 to 0.56. There was a difference of ∼0.2 between the utility scores for no treatment/ide-cel/oral administration and regular intravenous infusions.

Conclusions

Differences in treatment administration across RRMM therapies showed a substantial impact on health state utilities. When quantifying the value of treatments, process utility gains should be considered as an independent factor in health technology assessments.

Keywords:

• Relapsed/refractory multiple myeloma
• time trade-off
• health state utilities
• triple-class exposed
• CART cell therapy

JEL CLASSIFICATION CODES:

• C83
• C8
• C
• O53
• O5
• O

Introduction

Multiple myeloma (MM) accounts for 10–15% of hematopoietic tumors and ∼1% of all malignant tumors in Japan¹. MM is the third most prevalent hematopoietic tumor in Japan, after malignant lymphoma and leukemia. In Japan, the incidence of MM is ∼6.1 per 100,000 persons, and both incidence and mortality rates have increased annually as the population has aged^1–3.

Chemotherapy with autologous peripheral blood stem cell transplantation is indicated for patients with MM aged 65 years or younger, but not all patients can undergo transplantation¹. The introduction of immunomodulating agents, such as IMiDs, proteasome inhibitors, and anti-CD38 monoclonal antibodies has helped to improve survival while maintaining patients’ quality of life (QoL), even among those receiving systemic therapy⁴; however, most patients relapse. Patients with relapsed and refractory MM (RRMM) who are triple-class exposed (TCE) to immunomodulatory drugs, proteasome inhibitors, and anti-CD38 monoclonal antibodies have particularly poor outcomes and health-related QoL. As there are limited treatment options for TCE patients with RRMM, the median progression-free survival (PFS) is only about 3–4 months^4,5 and patients have a poor prognosis⁶.

Chimeric antigen receptor (CAR) T cell therapy is an immunotherapy that has recently been approved for patients with TCE RRMM in Japan. During CAR T cell therapy, T cells are extracted from patients, genetically modified to express CARs, and infused into the patient; the altered T cells are able to proliferate and attack cancer cells⁴. Idecabtagene vicleucel (ide-cel), a CAR T cell therapy that targets B-cell maturation antigen (BCMA), is indicated for patients with TCE RRMM. In the pivotal single arm phase 2 clinical study, KarMMa (bb2121 MM-001), a single infusion of ide-cel led to deep, and durable responses in TCE patients with RRMM (PFS for ide-cel [450 × 10⁶ CAR T cells], 12.1 months; data cutoff 14 January 2020) and was well tolerated, with a low incidence of cytokine release syndrome and neurotoxicity⁵. Conventional standard MM treatment requires regular hospital visits and ongoing drug administrations. Ide-cel requires a single infusion, which may improve patients’ QoL by reducing treatment administration-related burden. “Process utility” is a concept that considers the impact of treatment administration experience on QoL is distinct from the impact of clinical outcomes on QoL⁷.

Traditional cost-effectiveness analyses use outcome-based health state utilities to quantify the value of new healthcare technologies in terms of quality-adjusted life years (QALYs) associated with their use. In their guide entitled NICE health technology evaluations: the manual, the National Institute for Health and Care Excellence (NICE) recommends that benefits associated with process characteristics, such as convenience or improved speed of diagnosis, should be quantified, if possible⁸. A systematic review of studies evaluating process utility also indicated that process utility exists as an independent concept distinct from the health outcome⁹. Furthermore, it has been reported that patients themselves have experienced substantial differences in QoL during the treatment process¹⁰.

While commonly used preference-based measures (such as EQ-5D) may implicitly involve aspects that are influenced by process utility (especially regarding the measurement of daily activities), it is difficult to quantitatively separate the influence of process utility from that of health outcome-related QoL when using these measures. For example, if patients feel that hospital visits for treatment interfere with their daily activities, they may not respond to the assessment positively regardless of their clinical symptoms. Therefore, such preference-based measures do not fully reveal the impact of changes in process utilities.

This study aimed to evaluate the impact on process utility of treatment options for TCE patients with RRMM using the vignette-based time trade-off (TTO) method.

Methods

Study design

This study estimated utility scores for eight RRMM treatment options for TCE patients using the TTO method. The RRMM treatment options evaluated were no treatment, ide-cel, and the following six treatment regimens that are widely used as conventional treatment in Japan: daratumumab, lenalidomide, and dexamethasone (DRd); isatuximab, pomalidomide, and dexamethasone (IsaPd); daratumumab, carfilzomib, and dexamethasone (DKd); carfilzomib and dexamethasone (Kd); elotuzumab, pomalidomide, and dexamethasone (EPd); and pomalidomide and dexamethasone (Pd) therapy. Utilities were estimated assuming survival from a time point after the start of treatment with ide-cel and after the patient’s condition had stabilized following any early adverse events, such as cytokine release syndrome, and before any disease progression.

A questionnaire was developed and data were collected according to the methodology recommended by Matza et al.⁷. Based on previous TTO studies, the target number of responses was 150–300 to ensure a sufficient sample size for utility estimation^11–13. Eligible survey respondents were healthy adults ≥20 years of age living in Japan. Demographic characteristics were also collected, and the current utility score of the sample was measured using the EQ-5D-5L. Eight virtual health states were randomly presented to respondents using the TTO method, and the number of years in full health that were considered equivalent to 10 years in the corresponding health state was estimated using the ping-pong titration procedure (in which a survival period in the health state is increased or decreased incrementally by 1 year to a maximum of 10 years; see Supplemental Figure 1)¹⁴. To enhance the comparability of results from the present study with EQ-5D in future economic modeling, a standard survival length of 10 years was adopted in the TTO method; this is to ensure consistency with the time periods used to derive scoring tariffs for the EQ-5D, as recommended by Matza et al.⁷. The lead-time TTO method was used to evaluate health states defined as “worse than death”¹⁴. The question statements and the figures representing the years of survival in the compared health states that were presented to respondents, as well as the logic of the ping-pong titration method, are shown in Supplemental Figures 1–3. To ensure they had sufficient understanding of the disease and the TTO concept, respondents were asked the following before the main TTO tasks: (1) to give their prior impression of the study attributes of the health state description (i.e. duration of survival and dose frequency, time for the hospital visit, symptom, impact on daily activity), (2) to complete a trade-off task for the duration of survival and dosing frequency, and (3) to complete TTO training using the example of living in a wheelchair.

Health state development and validation

Vignettes defining health states and daily activity limitations were prepared based on previous studies, clinical trial data, and the expert opinions of MM specialists Dr. Tadao Ishida and Dr. Shinsuke Iida. Vignettes are descriptions of health states used in preference elicitation tasks to obtain utility estimates and are widely used in economic modeling⁷. Vignettes facilitate a consistent description of the condition and its treatment, as well as their impact on various components of QoL. They are also referred to as “scenarios” or “descriptions of the health state,” and have long been used to estimate utilities for medical and psychiatric conditions^15–17.

The purpose of this study was to estimate process utility improvements achieved by reducing the burden of treatment administration. Therefore, the eight treatment vignettes all used the same descriptions of the disease and clinical symptoms (Supplemental Tables 1–8); only the descriptions of treatment schedules differed across the eight treatment regimens. The clinical MM specialists provided their expert opinion to ensure that the daily activity and health states were clinically appropriate and were accurate reflections of the experiences of patients with MM. Vignettes of the treatment schedules for no treatment, ide-cel, treatment with regular intravenous infusion, and treatment by oral administration are provided in Table 1. Vignettes of the magnitude of other symptoms and restrictions on daily activity are provided in Table 2. The vignettes prepared for all treatments are presented in the Supplemental Data.

Table 1. Vignettes for treatment schedule.

Name of treatment	Descriptive contents of treatment schedules
No treatment	There is no treatment- or hospital-visit-related burden.
Ide-cel	After a single session of treatment by drip infusion, monthly hospital visits are necessary. You take oral drugs every day to relieve or prevent symptoms or complications. During a monthly hospital visit, follow-up examinations are performed for about 3 h.
DRd therapy	During a 4-week period, patients receive drug administration by drip infusion on 4 hospital visits. They take an oral drug every day for 21 days and another one on Day 2 of every week. They also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated twice. During the next 4 weeks, you receive drug administration by drip infusion on 2 hospital visits. You take an oral drug every day for 21 days and another one for 6 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated 4 times. During the subsequent 4 weeks, you receive drug administration by drip infusion on a single hospital visit. You take an oral drug every day for 21 days and another one for 7 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated all the time. On every hospital visit, drip infusion and follow-up examinations are performed for about 3–4 h.^a
IsaPd therapy	During the first 4 weeks, you receive drug administration by drip infusion on 4 weekly hospital visits. You take an oral drug every day for 21 days and another one on Day 1 of every week. You also take oral drugs every day to relieve or prevent symptoms or complications. During the subsequent 4 weeks, you receive drug administration by drip infusion on two hospital visits. You take an oral drug every day for 21 days and another one on Day 1 of every week. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated indefinitely. On every hospital visit, drip infusion and follow-up examinations are performed for about 3 to 4 h.
DKd therapy	During a 4-week period, you receive drug administration by drip infusion on 7 hospital visits. You take an oral drug for 7 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated twice. During the subsequent 4 weeks, you receive drug administration by drip infusion on 6 hospital visits. You take an oral drug for 7 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated indefinitely. On every hospital visit, drip infusion and follow-up examinations are performed for about 3–4 h.^a
Kd therapy	During a 3-week period, you receive drug administration by drip infusion on 6 hospital visits. You take an oral drug on Days 1 and 2 of every week. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated indefinitely. On every hospital visit, drip infusion and follow-up examinations are performed for about 3–4 h.
EPd therapy	During a 4-week period, you receive drug administration by drip infusion on 4 hospital visits. You take an oral drug on Day 1 of every week and another one every day for 21 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated twice. During the subsequent 4 weeks, you receive drug administration by drip infusion on two hospital visits.^b You take an oral drug on Day 1 of every week and another one every day for 21 days. You also take oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated indefinitely. On every hospital visit, drip infusion and follow-up examinations are performed for about 3–4 h.
Pd therapy	You take an oral drug every day for 21 days, and discontinue it for 7 days. You also take another one on Day 1 of every week, as well as oral drugs every day to relieve or prevent symptoms or complications. The above 4-week schedule is repeated all the time. During a monthly hospital visit, follow-up examinations are performed for about 3 h.

Abbreviations. DKd, daratumumab, carfilzomib, and dexamethasone; DRd, daratumumab, lenalidomide, and dexamethasone; EPd, elotuzumab, pomalidomide, and dexamethasone; Ide-cel, idecabtagene vicleucel; IsaPd, isatuximab, pomalidomide, and dexamethasone; Kd, carfilzomib and dexamethasone; Pd, pomalidomide and dexamethasone.

^aIn practice, drip infusion alone takes about 5 h, but the time required for hospital visit was standardized to 3–4 h for all treatment.

^bIn practice, the number of hospital visits is 1, but the survey was conducted with the description of two visits.

Table 2. Vignettes for disease, symptoms, and influence on daily activity.

Item	Descriptive contents
Disease	You have a life-threatening disease related to a chronic disease. If you do not receive treatment, the disease may worsen. However, the number of treatment options for cure is limited.
Symptoms	There is no fatigue, or, even if it is present, its degree may be mild. Pain of the backbone or breastbone is present, sometimes requiring analgesic drugs. Appetite is slightly reduced. Numbness of the legs is sometimes noted.
Influence on daily activity	Due to symptoms of the disease or knowledge about the disease,  – Anxiety or depression is noted.  – You can work, but the kind of work is restricted to light work. As infectious diseases or fracture may occur, many everyday activities, such as housekeeping, exercise, and social activities with other persons, are restricted.

Table 3. Social background information.

Characteristics^a	Respondents (n = 319)	General population^b
Age, mean (SD), years	44.0 (13.57)	46.4
Residence
Saitama Prefecture	32 (10.0%)
Chiba Prefecture	38 (11.9%)
Tokyo	208 (65.2%)
Kanagawa Prefecture	41 (12.9%)
Prefectures other than the ones above	0 (0.0%)
Other than Japan	0 (0.0%)
Female	160 (50.2%)	(51.3%)
Final educational background
Completion of graduate school	7 (2.2%)	(2.2%)
Graduation from university (bachelor’s degree)	158 (49.5%)	(22.1%)
Graduation from college (associate’s degree)	38 (11.9%)	(7.8%)
Graduation from technical school	48 (15.0%)	(12.7%)
Graduation from senior high school	67 (21.0%)	(39.7%)
Graduation from junior high school	0 (0.0%)	(14.8%)
Others	1 (0.3%)
Employment status
Working
Regular staff/employees	163 (51.1%)	(31.6%)
Part-time staff/employees	34 (10.7%)	(9.4%)
Part-time workers	9 (2.8%)	(4.3%)
Temporary workers from worker dispatching offices	8 (2.5%)	(1.3%)
Contract employees	16 (5.0%)	(2.7%)
Commission	2 (0.6%)	(1.1%)
Company executives	3 (0.9%)	(3.0%)
Self-employed workers (employees: present)	4 (1.3%)	(1.1%)
Self-employed workers (employees: absent)	17 (5.3%)	(3.6%)
Helping with the family business	0 (0.0%)	(1.3%)
Side job	0 (0.0%)	(0.1%)
Others	3 (0.9%)	(0.8%)
Not working
Looking for a job	2 (0.6%)	(1.5%)
Attending school	17 (5.3%)	(5.4%)
Housekeeping	41 (12.9%)	(12.0%)
Others (e.g. retired)	0 (0.0%)	(20.5%)
Annual household income, JPY
<2,000,000	8 (2.5%)	(19.0%)
2,000,000 to <3,000,000	21 (6.6%)	(13.6%)
3,000,000 to <4,000,000	46 (14.4%)	(12.8%)
4,000,000 to <5,000,000	33 (10.3%)	(10.5%)
5,000,000 to <6,000,000	42 (13.2%)	(8.7%)
6,000,000 to <7,000,000	36 (11.3%)	(8.0%)
7,000,000 to <8,000,000	24 (7.5%)	(6.2%)
8,000,000 to <9,000,000	21 (6.6%)	(4.9%)
9,000,000 to <10,000,000	21 (6.6%)	(4.1%)
≥10,000,000	56 (17.6%)	(12.1%)
I don’t want to answer/don’t know.	11 (3.4%)

Abbreviations. JPY: Japanese yen; SD. standard deviation.

Percentages may not add up to 100% owing to rounding.

^aData are n (%) unless otherwise indicated.

^bSources^18–20.

Table 4. Clinical background information.

Characteristic^a	Respondents (n = 319)	General population^b
Status of hospital visit
Not consulting the hospital	224 (70.2%)	(59.6%)
I was previously diagnosed with the following diseases
Multiple myeloma	0 (0.0%)
Hematologic cancer	0 (0.0%)
I have not been diagnosed with these diseases	319 (100.0%)
I don’t know	0 (0.0%)
I know persons diagnosed with the following diseases
Multiple myeloma	5 (1.6%)
Hematologic cancer	40 (12.5%)
I don’t know any person diagnosed with these diseases	269 (84.3%)
I don’t know	7 (2.2%)
I have received chemotherapy
Yes	1 (0.3%)
No	318 (99.7%)
I don’t know	0 (0.0%)
I know persons who have received chemotherapy
Yes	167 (52.4%)
No	143 (44.8%)
I don’t know	9 (2.8%)
EQ-5D-5L index score, mean (SD)	0.97 (0.07)	0.91
EQ-5D-5L index score, ceiling	247 (77.4%)	(57.3%)
VAS, mean (SD)	86.1 (10.9)	77.5

Abbreviations. COPD, chronic obstructive pulmonary disease; SD, standard deviation; VAS, visual analog scale.

Percentages may not add up to 100% owing to rounding.

^aData are n (%) unless otherwise indicated.

^bSource²⁰.

Table 5. TTO utility values.

Health state^a	Mean	SD	Median	Range	95% CI	Floor, n (%)	Ceiling, n (%)
No treatment	0.758	0.336	0.900	−0.950–1.000	0.721–0.795	5 (1.6%)	118 (37.0%)
Ide-cel	0.729	0.343	0.850	−0.950–1.000	0.692–0.767	5 (1.6%)	100 (31.3%)
DRd therapy	0.501	0.461	0.600	−0.950–0.950	0.450–0.552	14 (4.4%)	38 (11.9%)
IsaPd therapy	0.553	0.428	0.700	−0.950–0.950	0.506–0.601	9 (2.8%)	46 (14.4%)
DKd therapy	0.507	0.446	0.600	−0.950–0.950	0.458–0.556	13 (4.1%)	38 (11.9%)
Kd therapy	0.552	0.428	0.700	−0.950–0.950	0.505–0.599	12 (3.8%)	39 (12.2%)
EPd therapy	0.547	0.425	0.700	−0.950–0.950	0.501–0.594	10 (3.1%)	45 (14.1%)
Pd therapy	0.722	0.350	0.850	−0.950–0.950	0.684–0.761	6 (1.9%)	85 (26.6%)

Abbreviations. CI, confidence interval; DKd, daratumumab, carfilzomib, and dexamethasone; DRd, daratumumab, lenalidomide, and dexamethasone; EPd, elotuzumab, pomalidomide, and dexamethasone; Ide-cel, idecabtagene vicleucel; IsaPd, isatuximab, pomalidomide, and dexamethasone; Kd, carfilzomib and dexamethasone; Pd, pomalidomide and dexamethasone; SD, standard deviation; TTO, time trade-off.

^aNumber of samples = 319.

Table 6. Comparison of TTO utility values.

Health state^a	Mean	SD	Median	Range	95% CI	p-Value^b
No treatment—Ide-cel	0.028	0.204	0.000	−1.400–0.950	0.006–0.051	.014
No treatment—DRd therapy	0.257	0.336	0.200	−0.400–1.950	0.220–0.294	<.001
No treatment—IsaPd therapy	0.204	0.291	0.100	−0.500–1.650	0.172–0.236	<.001
No treatment—DKd therapy	0.251	0.323	0.200	−0.500–1.900	0.215–0.286	<.001
No treatment—Kd therapy	0.206	0.285	0.100	−0.450–1.900	0.174–0.237	<.001
No treatment—EPd therapy	0.210	0.290	0.150	−0.500–1.650	0.178–0.242	<.001
No treatment—Pd therapy	0.036	0.247	0.000	−1.500–1.650	0.008–0.063	.010
Ide-cel—DRd therapy	0.229	0.310	0.150	−0.900–1.650	0.194–0.263	<.001
Ide-cel—IsaPd therapy	0.176	0.282	0.100	−0.900–1.650	0.145–0.207	<.001
Ide-cel—DKd therapy	0.223	0.304	0.150	−0.900–1.900	0.189–0.256	<.001
Ide-cel—Kd therapy	0.177	0.266	0.100	−0.900–1.700	0.148–0.207	<.001
Ide-cel—EPd therapy	0.182	0.267	0.100	−0.900–1.650	0.153–0.211	<.001
Ide-cel—Pd therapy	0.007	0.213	0.000	−1.100–1.650	−0.016–0.031	.537

Abbreviations. CI, confidence interval; DKd, daratumumab, carfilzomib, and dexamethasone; DRd, daratumumab, lenalidomide, and dexamethasone; EPd, elotuzumab, pomalidomide, and dexamethasone; Ide-cel, idecabtagene vicleucel; IsaPd, isatuximab, pomalidomide, and dexamethasone; Kd, carfilzomib and dexamethasone; Pd, pomalidomide and dexamethasone; SD, standard deviation; TTO, time trade-off.

^aNumber of samples = 319.

^bPaired t-test.

A pilot survey involving 20 healthy volunteers was conducted to test the instrument before finalization, as well as to estimate the number of eligible respondents expected and respondents’ comprehension of the questionnaire. The implementation procedures for the survey are presented below:

1. Candidates were identified by snowball sampling by recruiters registered with a research firm and were requested to participate in a questionnaire survey by email.

2. Of those candidates who consented to participate in the pilot survey, potential respondents were selected according to the eligibility criteria.

3. Eligible respondents completed a face-to-face interview conducted by trained interviewers from INTAGE Healthcare Inc. at a designated venue. After completion of the pilot survey, respondents provided feedback about their comprehension of the health states and survey procedure descriptions.

The questionnaire was revised based on the results of the pilot survey. Describing the disease as “a life-threatening disease” led to mixed responses; therefore, the word “cancer” was added. In some cases, the expression “no need for drugs or hospital visits” had unintended negative connotations; therefore, the sentence was revised to “there is no treatment- or hospital-visit-related burden”. The time required for a hospital visit was amended to include the waiting time and time spent paying. Details of specific examinations performed during hospital visits (hematology tests, imaging, urinalysis) were added.

The candidates of the main survey were also selected by snowball sampling. Of those candidates who consented to participate in the survey, potential respondents were selected according to the eligibility criteria, as in the pilot study. Eligible respondents completed a face-to-face interview conducted by trained interviewers from INTAGE Healthcare Inc. at a designated venue.

Ethical approval

The protocol, questionnaires, and informed consent forms were reviewed and approved by the non-profit organization MINS research ethics committee. This study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Biological Research Involving Human Subjects. All respondents meeting the eligibility criteria were provided with a description of the study to ensure they understood the study, and all participants provided informed consent.

Statistical analyses

The health state utility scores based on the survey results were calculated as follows. The utility for a health state that was not evaluated as “worse than death” was calculated based on the following formula using death (0) to full health (1) as an anchor: $$u=x/y$$

• u: estimated utility value

• x: survival duration in full health

• y: survival duration in health states to be evaluated

“x” and “y” were the survival durations for each health state at the point when the health states were judged to be equivalent by respondents using the ping-pong titration method, or the mean of the survival durations for each health state at the point when the superiority or inferiority of the health states were reversed.

The utility for a health state that was evaluated as “worse than death” was calculated based on the following formula. In this case, the utility ranged from 0 to −1. $$u^{\prime }=\frac{x^{\prime }}{10}-1$$

• u′: estimated utility value

• x′: survival duration only in full health.

In the lead-time TTO, full health preceding the health state under evaluation is fixed at 10 years, and the survival time of the compared full health-only health state varies. “x′” was the survival duration for full health only at the point when the two health states were judged to be equivalent by respondents using the ping-pong titration method, or the mean of the survival duration in full health only at two points when the superiority or inferiority of two health states were reversed.

Summary statistics (including mean, standard deviation [SD], median, range, floor, ceiling, 95% confidence interval, and frequency) were used to evaluate health state utility scores, EQ-5D-5L index score, visual analog scale (VAS), and social/clinical background information. Summary statistics were also calculated for differences in the utility scores among health states, and these were compared using paired t-tests.

Subgroup analyses of health state utility scores were performed based on age, sex, educational background, and annual income, and these were compared using two-sample t-tests. Health state utility was also estimated using regression analysis as described in previous studies¹³. In addition to estimation with ordinary least squares (OLS), the utility was estimated using a mixed-effects model that accounted for random effects among the samples as well as a Tobit model in which the data characteristics (upper limit: 1, lower limit: −1) were considered. In the case of the mixed-effects model, an error term (${\epsilon }_{ij}=u_j+e_{ij}$) was decomposed into a random effect, $u_j,$ by the respondents and a random error, $e_{ij}.$ In addition to analysis using the health state alone as an explanatory variable, patient characteristics were used as covariates, and analysis involving variable selection was conducted (health state was always included as an explanatory variable). The stepwise-forward selection method with an OLS model to select covariates was used. The selected covariates were also used in the mixed-effects and Tobit models. The goodness of fit for each regression model was evaluated using the mean absolute error.

All analyses were performed using SAS Version 9.4 (SAS Institute, Cary, NC, USA). A p-value of <.05 on both sides was regarded as statistically significant. Missing values were not imputed.

Results

Demographic characteristics

The pilot survey and final survey were conducted on 28 and 29 August 2021, and between 26 and 30 September 2021, respectively. Twenty responses to the pilot survey and 319 responses to the final survey were collected. Half (50.2%) of the respondents were female. The mean age was 44.0 years, and 42.6% of the respondents were unmarried (Table 3 and Supplemental Table 9)^18–20. The educational level and annual household income of respondents were higher than the general Japanese population, and respondents were more likely to be regular, part-time, or contract staff/employees. Dental diseases (5.0%), hypertension (4.4%), lower back pain (3.1%), dyslipidemia (2.8%), ocular diseases (2.8%), allergic rhinitis (2.8%), and other skin diseases (2.8%) were the most frequent concomitant diseases²⁰. None of the respondents reported a previous diagnosis of hematologic cancer, including MM (Table 4 and Supplemental Table 10). The mean EQ-5D-5L index score was 0.97, and the mean VAS score was 86.1, which was slightly higher than that reported in the general Japanese population (the mean values in adults 20–89 years of age were 0.91 and 77.5, respectively, according to a report by Shiroiwa et al.²¹).

Differential utilities of RRMM treatment regimens

The utility scores for no treatment, ide-cel, and Pd therapy ranged from 0.722 to 0.758. The utility score for the intravenous infusion regimens therapy ranged from 0.501 to 0.553 (Table 5). The utility scores for the regular intravenous infusion regimens were consistent with that of others. There were statistically significant differences between utility scores for no treatment and ide-cel or Pd therapy (Table 6). There was no statistically significant difference in utility score between ide-cel and Pd therapy (mean: 0.007 [SD, 0.213]; p = .537). The approximate utility score difference between no treatment/ide-cel/Pd therapy, and the intravenous infusion regimens was 0.2 (p < .001). Histograms of the utility scores for each health state and differences in the utility scores between health states are shown in Supplemental Figures 4–6.

Subgroup comparisons and regression analysis

No major differences were observed between utility scores when evaluated by age, sex, educational level, and annual income (Supplemental Tables 11–14).

The results of the utility estimation for each health state using regression analysis with and without covariates are presented in Supplemental Tables 15 and 16. In all statistical models without covariates, the partial regression coefficient of each health state for no treatment was similar to differences in the utility scores across health states (Table 6). This was also consistent when considering covariates.

Discussion

This vignette-based TTO study found differences in the process utilities for various treatment options available for TCE patients with RRMM in Japan. Utility scores before disease progression were consistent with those obtained directly from EQ-5D scores from clinical trials²²; the vignettes adopted in this study may therefore be considered appropriate. The utility scores related to the burden of drug administration across treatment regimens ranged from ∼0.5 to 0.8. Utility scores for no treatment, ide-cel, and Pd therapy were analogous (ranging from 0.72 to 0.76), while the utility score for the intravenous infusion regimens, as a representative of intravenous infusion regimens, was substantially lower (0.55 ranging from 0.50 to 0.55). Subgroup analyses based on key demographic characteristics showed findings consistent with the primary analysis.

When compared with vignettes for no treatment, ide-cel, and Pd therapy, the vignette for the intravenous infusion regimens treatment health state differed only in terms of the descriptions of burdens related to drug administration (frequency of hospital visits and time required for treatment on a hospital visit), with identical descriptions being used for the magnitude of other symptoms or daily activity restrictions. Therefore, a difference of ∼0.2 can be interpreted as a process utility gain in favor of no treatment, ide-cel, or Pd therapy. The burden related to treatment with intravenous infusion (utilities between 0.7–0.8 and 0.55) was similar to that of disease-related symptoms or daily activity restrictions (differences between 1.0 and 0.7–0.8), suggesting that the burden of the treatment process for an intravenous infusion regimen is substantial, and is similar to the burden associated with disease development according to an estimation of value from a general population sample.

Furthermore, the estimated utility of a health state for an intravenous infusion regimen was lower than that observed before disease progression, as obtained from clinical trials using the EQ-5D. This is probably because the EQ-5D does not include characteristics directly reflecting an increasing burden that is specifically attributable to the treatment process. The EQ-5D evaluates a health state “today”; naturally, the size of the problem may be measured in comparison with the health state on days other than today, and burdens that increase routinely over a long period may not be quantified.

To our knowledge, this is the first TTO study evaluating utility or process utility in patients with MM specifically. Several studies have evaluated treatment utilities using the TTO method in patients with other conditions in the field of hematologic cancer. Aristides et al. reported that the utility under the complete remission of relapsed or refractory B-precursor acute lymphoblastic leukemia ranged from ∼0.75 to 0.86²³, and Guest et al. reported that the utility associated with molecular, cytogenetic, and hematologic responses to treatment of chronic myelogenous leukemia ranged from ∼0.80 to 0.94²⁴. Howell et al. reported that the utility in patients with diffuse large B-cell lymphoma during treatment with CAR T cell therapy without adverse events was ∼0.73²⁵. As the vignettes used in the referenced studies did not include detailed descriptions of the burden related to drug administration (frequency of hospital visits or time required for treatment during a hospital visit), their results may be considered analogous to those observed in this study for no treatment, ide-cel, and Pd therapy.

Process utility with respect to treatment administration using the vignette-based approach has also been investigated in other conditions. Johnson et al. reported that the utility gain related to oral administration of antiviral drugs for patients with HIV was 0.362 when compared with administration by intravenous injection²⁶. Osborne et al. reported the utility gain related to oral administration of treatment for chronic iron overload was 0.24 when compared with subcutaneous injection²⁷. Both studies used the TTO method and the findings were consistent with those of the current study despite different conditions and patient populations. Several studies have also measured process utility using the standard gamble method. Boye et al. reported a utility gain of 0.023 for weekly insulin injection compared with daily injections²⁸, and Chancellor et al. reported a utility gain of ∼0.01–0.08 for insulin inhalation vs. injection²⁹. These estimated values are lower than the process utility gains estimated using the TTO method, which may suggest systematic differences between these methodologies, despite other differences in condition and treatment scenarios.

These findings should be considered in the context of certain study limitations. In the questionnaire introduction guide, the target condition of the vignettes was described as a “life-threatening disease like cancer,” but the psychological fear caused by this description may have led to an overestimation of the impact of the clinical burden. On the other hand, when the word “cancer” was not included in the cognitive debriefing process, the respondents envisioned varied conditions, suggesting that omitting the name of the condition likely led to uncertainty in respondents’ interpretations. Although these limitations may influence the results as biases, the condition was expressed relatively vaguely in this study to reduce uncertainty while also minimizing potential bias related to the overestimation of fear. Furthermore, the utilities of no treatment, ide-cel, and Pd therapy, in which burdens related to drug administration were minor, did not differ from the utility before disease progression obtained directly from clinical trials using a preference-based measure; therefore, the influence of this limitation on the results is likely to be minimal. An additional limitation concerns the fact that, in reality, treatment with intravenous infusions does not continue for the entire survival period in the TTO task (i.e. 10 years), but rather for several months. Although we prioritized the comparability with EQ-5D in this study, the estimates would potentially be affected by the length of the survival period in the TTO task.

The educational level and annual income in the sample population were slightly higher than those of the general Japanese population. However, previous studies estimating utility in the general population have shown a similar trend. Furthermore, while the age of the respondents was lower than the peak age of MM onset, for example, the impact of age by subgroup analysis was also minimal. Therefore, the impact of this limitation on the findings may be limited, as affirmed by the results of the subgroup analyses.

Although ide-cel requires a certain period of hospitalization and pretreatment, including lymphodepleting chemotherapy, for the administration of CAR T cells, hospitalization, and pretreatment were not included in the description of the health state. The purpose of this study was to estimate the health state utility with ide-cel from the stable condition of the patient after administration of CAR T cells until disease progression. In addition, because the study focused on process utilities across treatments, it did not consider potential adverse events associated with each treatment.

Administration of DRd therapy by subcutaneous injection was approved in May 2021 as an alternative to intravenous injection but was not included in this study. The burden related to subcutaneous injection may be lower than that of intravenous infusion, and therefore the process utility gain on assuming subcutaneous injection may be smaller than the improvements observed with reduced burden related to CAR T cell therapy or oral administration as estimated in this study.

As we discussed in the previous section, there would be systematic differences between the valuation methodologies, such as TTO method and the standard gamble method, but we were not able to compare our results with those of recent studies. We hope to see more research on process utility in the future using various methods, including the discrete choice experiment (DCE) method.

Finally, this study used respondents drawn from the general population; it is difficult to say whether the preferences observed in this study would be consistent with the value perceptions of patients with MM.

Conclusions

This is the first study evaluating process utilities related to the administration of different RRMM treatments. Our results suggest that the burden related to the treatment process for intravenous infusions is substantial, suggesting that CAR T cell therapy has a reduced treatment process burden when compared with other RRMM treatment regimens. When quantifying the value of a treatment that may reduce the burden of drug administration, like CAR T cell therapy, process utility gains should be considered as an independent factor in health technology assessments. These findings may inform future methodological development, particularly with regard to how process utility is considered in health technology assessments.

Transparency

Declaration of funding

This work was supported by Bristol Myers Squibb K.K. The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Declaration of financial/other relationships

TI received grants from Bristol Myers Squibb, Janssen, Pfizer, and Takeda, and honoraria from Bristol Myers Squibb, Janssen, Ono, Sanofi, and Takeda outside of this work. MN is an employee of and shareholder in Bristol Myers Squibb K.K. TM is an employee of CRECON Medical Assessment Inc., which received payment for these works from Bristol Myers Squibb K.K. FM received consultation fees from Bristol Myers Squibb K.K. SI received honoraria from Bristol Myers Squibb, Celgene, Janssen, Ono, Sanofi, and Takeda, and research grants from AbbVie, Amgen, Bristol Myers Squibb, Caelum, Chugai, Daiichi Sankyo, Janssen, Ono, Otsuka, Pfizer, and Takeda, outside of this work.

Author contributions

All authors contributed to the conception and design of the study and the data analysis and interpretation. Acquisition of data was conducted by MN, TM, and FM. All authors critically revised the manuscript for intellectual content and gave final approval of the version to be published. All authors agree to be accountable for all aspects of the work.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgements

The authors received writing and editorial assistance in the preparation of this manuscript from Eilish McBurnie, of Excerpta Medica, funded by Bristol Myers Squibb.

Data availability statement

BMS policy on data sharing may be found at https://www.bms.com/researchers-and-partners/independent-research/data-sharing-request-process.html.