Economic evaluations of cancer immunotherapy: a systematic review and quality evaluation

Abstract

Objectives

Scientific advances in the last decade have highlighted the use of immunotherapy, especially immune checkpoint inhibitors, to be an effective strategy in cancer therapy. However, these immunotherapeutic agents are expensive, and their use must take into account economic criteria. Thus, the objective of the present study was to systematically identify and review published EE related to the use of ipilimumab, nivolumab or pembrolizumab in melanoma, lung cancer, head and neck cancer or renal cell carcinoma, and to assess their quality.

Methods

The systematic literature research was conducted on Medline via PubMed and the Cochrane Central Register of Controlled Trials to identify economic evaluations published before July 2018. The quality of each selected economic evaluation was assessed by two independent reviewers using the Drummond checklist.

Results

Our systematic review was based on 32 economic evaluations using different methodological approaches, different perspectives and different time horizons. Three-quarters of the economic evaluations are full (n = 24) with a Drummond score ≥ 7, synonymous of “high quality”. Among them, 66% reported a strategy that was cost-effective. The most assessed immunotherapeutic agent was nivolumab. In patients with renal cell carcinoma or head and neck cancer, it was less likely to be cost-effective than in patients with melanoma or lung cancer.

Conclusions

Whether or not these findings will be confirmed remains to be seen when market approval to cover more indications is extended and new effective immunotherapeutic agents become available.

Electronic supplementary material

The online version of this article (10.1007/s00262-020-02646-0) contains supplementary material, which is available to authorized users.

Introduction

Scientific advances in the last decade have highlighted the significant role of the immune system in the battle against cancer [1]. The development of immunotherapy, especially the immune checkpoint inhibitors, such as anti-programmed death (PD)-1 (nivolumab, pembrolizumab) and anti-cytotoxic T lymphocyte-associated protein (CTLA)-4 (ipilimumab) agents, has been shown to restore the anti-tumor activity of T-lymphocytes, thus representing an effective strategy in cancer therapy [2–4]. The efficacy and safety of these immunotherapeutic agents has been demonstrated in patients with melanoma [5–8], lung cancer [9–12], renal cell carcinoma [13], head and neck squamous-cell carcinoma [14], and urothelial cancer [15], with improved overall survival (OS).

In ipilimumab-naive patients with advanced melanoma, an OS of 10.1 months (95% Confidence Interval (CI) [8.0–13.8], Hazard ratio (HR) 0.66 [0.51–0.87]) has been reported using ipilimumab versus an OS of 6.4 months [5.5–8.7] using gp100 [5, 6]. In patients with advanced melanoma no longer responding to the drug previously used to treat it, a progression-free survival (PFS) of 5.5 months (95% CI [3.4–6.9]) has been reported using pembrolizumab 10 mg/kg every 2 weeks versus a PFS of 4.1 months [2.9–6.9] using pembrolizumab 10 mg/kg every three weeks versus a PFS of 2.8 months (95% CI [2.8–2.9]) using ipilimumab [7]. In patients with advanced melanoma who progressed after ipilimumab, 31.7% (95% CI [23.5–40.8]) of patients treated with nivolumab were responding versus 10.6% (95% CI [3.5–23.1]) of patients treated with paclitaxel or dacarbazine [8].

In patients with advanced squamous-cell non-small-cell lung cancer, previously treated with platinum-based chemotherapy, the median overall survival was 9.2 months using nivolumab, versus 6.0 months using docetaxel (HR 0.59 [0.43–0.81] [9]. In previously treated patients with advanced non-small-cell lung cancer and PD-L1 expression on at least 1% of tumor cells, OS was significantly longer for pembrolizumab 2 mg/kg versus docetaxel (HR 0.71, 95% CI [0.58–0.88]) and for pembrolizumab 10 mg/kg versus docetaxel (HR 0.61, 95% CI [0.49–0.75]) [12].

In previously treated patients with advanced renal-cell carcinoma, the median OS was 25 months (95% CI [21.8–not estimated]) with nivolumab versus 19.6 months (95% CI [17.6–23.1]) with everolimus [13].

In patients with recurrent squamous cell carcinoma of the head and neck whose disease had progressed after platinum-based chemotherapy, a median OS was 7.5 months (95% CI [5.5–9.1]) in the nivolumab group versus 5.1 months (95% CI [4.0–6.0]) in the group that received standard therapy (methotrexate, docetaxel or cetuximab) [14].

In previously treated patients with advanced urothelial cancer, the median OS was 10.3 months (95% CI [8.0–11.8]) versus 7.4 months (95% CI [6.1–8.3]) with pembrolizumab and standard therapy, respectively (investigator’s choice of paclitaxel, docetaxel or vinflunine) [15].

In Europe, immune checkpoint inhibitors have been used in clinical practice since 2011 for ipilimumab and since 2015 for nivolumab and pembrolizumab [16]. The first marketing authorizations were granted for previously treated patients with advanced melanoma. Thus, in France, in 2016, anticancer drugs cost the health care system 5.9 billion euros, with ipilimumab being the most expensive drug (19,511 € per month) [17]. Of course, this subsequently raises real concerns about the affordability and cost-effectiveness of these new innovative drugs [18].

In the context of rational decision-making in health care, which is an integral part of health technology assessment (HTA), the economic evaluation (EE) of health care products has become a necessity. One of the major challenges is to provide cost-effectiveness data that are relevant to daily practices and may be required to optimize consumption of health care resources. Increasingly, decision-making for coverage and reimbursement of new drugs is being supported by EE in many countries including Australia, Canada and the United Kingdom (UK). These EE must be high-quality assessments to trust the results and make informed decisions. There are several different methods of economic evaluation [19]. According to Drummond et al., the EE classification is usually based on the ability of studies to answer both of the following questions: Is there a comparison of two or more strategies? Are both costs (inputs) and consequences (outputs) of alternative interventions examined? [20]. If both the costs and the consequences of two or more strategies are compared, the EE is considered as full and partial otherwise. Recommended EE methods include cost-effectiveness analyses (CEA) and cost-utility analyses (CUA). The choice of method also depends on the nature of the expected health effects of the interventions under study. CEA are required if health-related quality of life (HRQOL) is not identified as a relevant health effect of the interventions studied; health outcome is measured by the length of life expressed in life years (LY). CUA is the preferred method when HRQOL is identified as an important health effect of interventions; health outcome is then measured by the length of life weighted by the valuation of the HRQOL, represented by health-state utility values (HSUV), to produce quality-adjusted life year (QALY). The HSUV is on a scale anchored by 1 (best imaginable health state, i.e. perfect health) and 0 (worst imaginable health state, i.e. death) using patient preference-based measures [21]. For example, one QALY represents one life-year spent in perfect health and 8 QALY is equivalent to living 10 years with a chronic disease with an HSUV of 0.8. CUA should be preferred to CEA, especially for the EE of drug therapies, due to the impact that cancer treatments have on patients’ HRQOL. However, the quality of these EE has been given little attention thus far.

Therefore, the objective of the present study was to systematically identify and review published EE related to the use of ipilimumab, nivolumab or pembrolizumab in melanoma, lung cancer, head and neck cancer, renal cell carcinoma and urothelial cancer, and to assess their quality.

Methods

Search strategy

Based on the PRISMA checklist, a systematic literature search was conducted on Medline via PubMed and the Cochrane Central Register of Controlled Trials to identify published economic evaluations. Articles were included if they:

• reported full or partial economic evaluations;

• involved ipilimumab, nivolumab or pembrolizumab;

• melanoma, lung cancer, renal cell carcinoma, head and neck cancer or urothelial cancer;

• were written in English or French and published before July 2018.

Articles were excluded if they:

• did not report economic evaluations;

• did not involve ipilimumab, nivolumab or pembrolizumab;

• concerned cancers other than melanoma, lung cancer, renal cell carcinoma, head and neck cancer or urothelial cancer;

• were systematic reviews, editorials, comments, or letters to the editor;

• appeared in the National Health Service Economic Evaluation database and National Health Service Health Technology Assessment database, corresponding to appraisals and guidelines produced by the National Institute for Health and Clinical Excellence (NICE).

Medical Subject Heading (MeSH) terms were individually selected by means of the National Library of Medicine controlled vocabulary thesaurus used for indexing articles for PubMed, i.e. MeSH database, before being combined: (cost–benefit analysis OR cost OR costs OR cost-effectiveness analysis OR cost-utility analysis OR quality-adjusted life years) AND (ipilimumab OR yervoy OR nivolumab OR opdivo OR pembrolizumab OR keytruda).

Article selection

The title and abstract of each identified article were screened and independently analyzed by both reviewers (CC, PF) to determine whether the article presented an original EE that matched the previously defined topic. Secondly, the full text of each selected article was independently analyzed by both reviewers. Additionally, the reference lists of all selected EE were screened to identify other potentially relevant articles that had not been identified by means of the electronic source. A consensus was found in the case of disagreements between the two reviewers. The main reason for each article exclusion was recorded.

Data extraction

Based on the CHEERS checklist, the following data were identified and extracted from each selected article:

• year of publication;

• journal;

• main location of the first author;

• sponsor;

• conflict(s) of interest;

• stage and characteristics of cancer and line of treatment [22].

Data about the economic analysis were also extracted:

• aim(s) of economic analysis;

• examination of both the costs and consequences of alternative strategies;

• examination of at least two strategies;

• type of EE (full, including cost-effectiveness analysis, cost-minimization analysis, cost-utility analysis, cost–benefit analysis, or partial, including cost description (i.e. cost-of-illness), cost analysis and cost-outcome description).

For each article, more specific information about the analysis was extracted, such as:

• number of interventions compared ≥ 2;

• compared interventions, and populations analyzed (hypothetical cohort of patients, patients included in a clinical trial, patients of a national or a local institution);

• perspective (i.e. the point of view from which the costs and health effects are recorded and assessed: society, health care system);

• country of origin, time horizon (i.e. the period during which patients were followed with measure of costs and health effects, long enough to reflect all expected differences);

• main data source of costs (derived from local databases(s), national database(s), the literature, etc.);

• year reference of costs;

• main data source of effectiveness (derived from the results of clinical trial(s), the literature, original study);

• especially health-state utility values if applicable.

Other data were analyzed:

• decision-making analysis model (yes or no);

• decision-making analysis Markov model (yes or no);

• discounting of costs and effectiveness (i.e. the reflection of the present value of future costs and health effects as soon as the time horizon exceeds 12 months) and discount rate;

• results in terms of costs, effectiveness (LY, QALY, etc.) and cost-effectiveness (incremental cost-effectiveness ratio (ICER)) or cost-utility (incremental cost-utility ratio (ICUR));

• difference between ICER and ICUR according to the threshold, deterministic and probabilistic sensitivity analyses (i.e. the characterization of the uncertainty; yes (complete), yes (partial), no, not assessable).

To allow direct comparisons across countries, all costs were converted to US dollars, then inflated to the reference year 2018 [23].

Thus, the quality assessment of each selected EE was performed by two independent reviewers (CC and PF) using the Drummond checklist of 10 questions. It provides a framework for assessing the methodological quality and the results of any EE (supplementary Table 1). For each question, there are four possible responses: yes, no, not clear and not appropriate. One point is assigned for each “yes” response. The lowest and highest possible scores were thus 0 and 10, respectively. A score ≥ 7 is assigned to a “high quality” [20]. A consensus was found whenever the two reviewers were unable to come to an agreement.

Results

Article selection

The literature search conducted on PubMed and the Cochrane Central Register of Controlled Trials identified 236 articles, among which 20 were duplicates and 149 were excluded after reviewing the titles and abstracts that did not match the eligibility criteria (Fig. 1). A total of 67 articles were included for full text review, among which 35 were excluded for several reasons: they were not EE; they did not concern melanoma, lung cancer, head and neck cancer, renal cell carcinoma or urothelial cancer; they were systematic reviews, letters to editors, abstracts, appraisals or guidelines produced by the NICE. In the end, 32 EE were eligible for the present systematic review [24–55]. All of the 32 selected and analyzed EEs are presented in supplementary Table 2.

Fig. 1.

CONSORT flow diagram of literature review. a Not economic evaluation; b not melanoma, lung cancer, head and neck cancer or renal cell carcinoma; c systematic review, editorial, comment, letter to the editor, practice point. EE economic evaluation

Characteristics of the economic evaluations

The characteristics of the 32 EEs are summarized in Table 1. EEs were first published in 2013, and there has been a considerable increase in published EEs ever since that time (since 2017: n = 20). EEs were published in a clinical journal in 53% of cases (n = 17). One author published at least two articles on this subject. The most common type of cancer was melanoma (n = 16, 50%). With some exceptions, all EEs involved an unresectable, advanced or metastatic stage of cancer, and a first-, second- and/or third-line of treatment.

Table 1.

Characteristics of thirty-two included economic evaluations

	Number	Percentage
Selected articles	32	100
Year of publication
2009–2014	3	9.4
2015	3	9.4
2016	6	19.7
2017	12	37.5
2018	8	25.0
Journal
Clinical	17	53.2
Economic	9	28.1
Pharmaceutical	6	18.7
Journal of Medical Economics	3	9.4
Journal of Managed Care and Specialty Pharmacy	3	9.4
Melanoma research	2	6.2
Value health	2	6.2
Main location of first author
University	6	18.8
University and hospital	5	15.6
University and consulting group	1	3.1
University and research group	1	3.1
Hospital	5	15.6
Economic institute	8	25.0
Pharmaceutical industry	3	9.4
Consulting group	3	9.4
Sponsor
No	8	25.0
Yes	19	59.4
Not reported	5	15.6
Conflicts of interest
No	13	40.6
Yes	19	59.4
Characteristics of cancer
BRAF wild-type melanoma	2	6.2
BRAF-mutated melanoma	1	3.1
BRAF-mutated melanoma with V600E or V600K mutation	1	3.1
BRAF wild-type and BRAF-mutated melanoma	1	3.1
Melanoma	11	34.4
Squamous and non squamous non-small-cell lung cancer	6	18.8
Renal cell carcinoma	6	18.8
Head and neck squamous-cell carcinoma	3	9.4
Bladder cancer	1	3.1
Stage of cancer
Advanced	13	40.6
Locally advanced or metastatic	1	3.1
Metastatic	10	31.3
Unresectable and/or metastatic	4	12.5
Recurrent or metastatic	3	9.4
Not reported	1	3.1
Line of treatment
All	2	6.2
First-line treatment	6	18.8
First ± second-line treatment	2	6.2
First-, second and third-line treatment	1	3.1
Second-line treatment	14	43.7
Second- or third-line treatment	1	3.1
Previously treated	1	3.1
Not reported	5	15.7

Synthesis of the basic elements of economic evaluations

The synthesis of the basic elements of the 32 EEs is summarized in Table 2. Three-quarters of the EEs are full (n = 24) and are cost-utility analyses. More than 84% of EEs compared at least 2 interventions (n = 27) and more than 81% of EEs used a hypothetical cohort of patients. In four out of five cases, authors conducted the analysis from the perspective of the health care system (n = 26). The main country of origin was the United States (n = 19, 59%). The time horizon ranged from a line of treatment to lifetime, with 25% (n = 8) of EEs with a “lifetime” perspective, and only three EEs did not report it. In most cases, costs were derived from local or national database(s) and/or the literature (n = 25, 78%). Effectiveness was derived from the original study in only 2 EEs and was mainly derived from the results of clinical trials completed by more or less others sources (n = 24). HSUV were mainly derived only from the results of clinical trial(s) in 50% of cases (n = 12) when applicable. In one quarter of cases (n = 6), HSUV were derived from the results of clinical trial(s) and other sources (the literature or assumptions). Authors of EEs used a model in more than 75% of EEs (n = 25). A discounting of costs and/or effectiveness was made in more than 50% of cases (n = 17). A complete sensitivity analysis (deterministic and probabilistic) was conducted in more than 50% of EEs (n = 17).

Table 2.

Synthesis of basic elements of thirty-two included economic evaluations

	Number		Percentage
Selected articles	32	100.0
Examination of both the costs and effectiveness of alternative interventions
Yes	27		84.4
No	5		15.6
Examination of at least two interventions
Yes	27		84.4
No	5		15.6
Full or partial economic evaluation
Full	24		75.0
Partial	8		25.0
Type of economic evaluation
Cost analysis	3		9.4
Cost-outcome description	3		9.4
Cost description (i.e. cost-of-illness)	2		6.2
Cost-utility analysis and cost-effectiveness analysis	14		43.8
Cost-utility analysis	10		31.2
≥ 2 compared interventions
Yes	27		84.4
No	5		15.6
Hypothetical cohort of patients as population analysed
Yes	26		81.3
No	6		18.7
Perspective
Health care system	23		71.9
Health care system and societal	3		9.4
Societal	2		6.2
Hospital	1		3.1
Not reported	3		9.4
Country of origin
Australia	1		3.1
Canada	4		12.6
Italy	1		3.1
Portugal	1		3.1
Spain	1		3.1
Swiss	1		3.1
United Kingdom	4		12.6
United States	17		53.1
United States and other(s) country(ies)	2		6.2
Time-horizon
Line(s) of treatment	3		9.4
1 or 2 years	1		3.1
3 years	1		3.1
5 years	3		9.4
10 years	4		12.5
20 years	4		12.5
30 years	2		6.2
40 years	1		3.1
Lifetime	8		25.0
Until progression or death	1		3.1
Until the earliest of the end of eligibility or the end of data availability	1		3.1
Not reported	3		9.4
Main data source of costs
Derived from local or national database(s)	16		50
Derived from local and/or national database(s) and the literature	9		28.3
Derived from local and/or national database(s), the literature and the results of clinical trial(s)	1		3.1
Derived from local or national database(s), the literature and expert opinion	2		6.2
Derived from local or national database(s) and expert opinion	1		3.1
Derived from the literature	1		3.1
Derived from the literature and the results of clinical trial(s)	1		3.1
Not reported	1		3.1
Main data source of effectiveness (n = 27)
Derived from the results of clinical trial(s)	15		55.6
Derived from the results of clinical trial(s) and local and/or national database(s)	1		3.7
Derived from the results of clinical trial(s) and local and/or national database(s) and assumptions	1		3.7
Derived from the results of clinical trial(s), the literature and assumptions	2		7.4
Derived from the results of clinical trial(s) and assumptions	4		14.8
Derived from the results of clinical trial(s), assumptions and expert opinion	1		3.7
Derived from the literature and assumptions	1		3.7
Original study	2		7.4
Main data source of health-state utility values if cost-utility analysis (n = 24)
Derived from the results of clinical trial(s)	12		50.0
Derived from the results of clinical trial(s) and the literature	3		12.5
Derived from the results of clinical trial(s) and assumptions	3		12.5
Derived from the literature	6		25.0
Model
Yes	25		78.1
Markov model	13		40.6
No	7		21.9
Discounting of costs and effectiveness
Yes	17		53.1
No	2		6.2
No, only costs	3		9.4
Not reported	10		31.2
Sensitivity analysis
Yes, complete	17		53.1
Yes, partial	10		31.2
No	5		15.6
Deterministic sensitivity analysis
Yes, complete	19		59.4
Yes, partial	7		21.9
No	6		18.7
Probabilistic sensitivity analysis
Yes, complete	21		65.6
Yes, partial	2		6.2
No	9		28.1

Quality assessment of the economic evaluations

The median Drummond score of the 32 selected EEs was 9 [range: 1–10; first quartile: 6; third quartile: 10] (Fig. 2 and supplementary Table 3). Three-quarters of the economic evaluations (n = 25) were high quality EEs with a Drummond score ≥ 7.

Fig. 2.

Percentage of “Yes” for each question of Drummond’s 10-point checklist for assessing economic evaluations. The 10 questions of Drummond’s checklist are presented on the left side of the figure. For each question, corresponding percentages of “yes” are specified on the right side

Discussion

Over the past 15 years, interest in the economic evaluation of health care interventions has risen. Results of EE have become increasingly important as criteria for the allocation of health care resources. To inform health care decision-making, the quality of EE must be high. Our systematic review identified 32 published EEs involving ipilimumab, nivolumab and pembrolizumab in an approved indication [24–55]. An optimal method implemented to enhance the quality of the systematic review based on modified AMSTAR, a reliable and valid measurement tool consisting of 11 items [56]. Three-quarters of the EEs were high-quality assessments, with a Drummond score ≥ 7.

Seventy-five percent of the 32 EEs were full, and all of them were CUA. CUA are recommended by the Canadian Agency for Drugs and Technologies in Health (CADTH), the National Institute for Health and Clinical Excellence (NICE), and the French National Authority for Health (HAS) when HRQOL is identified as a relevant health effect of the interventions studied and compared, such as in oncology or for chronic diseases. Unsurprisingly, the United States and Canada, for which CUA are recommended, are the countries where EE originated. The number of EEs in UK is underestimated because CUA piloted by the NICE were excluded so as not to introduce biases regarding the quality of EE. More than 50% of the EEs were published in clinical journals whereas 25% were published in economic journals, thus giving “high quality” EEs (1 EE had a Drummond score of 9 and 5 EEs had a Drummond score of 10).

To provide EE results that are relevant to daily practices and optimize the consumption of health care resources, it is recommended to develop full EEs (CEA, CUA) of high quality. For this reason, and to help the potential reader, it is important to assess the quality of selected EEs. The overall quality of the 32 EEs in our study appears to be high with a median Drummond score equal to 9. This can be explained by the number of full EEs, for which the Drummond score is always ≥ 7. Even 1 partial EE managed to have a Drummond score ≥ 7. The critical appraisal of the methodological quality of health economic evaluations is not easy, may be subjective and may depend on the evaluator. Several checklists have been developed to guide critical appraisal of full health economic evaluations [21, 22]. Both the Drummond and Evers checklists are recommended in Cochrane reviews [19]. The use of the Drummond checklist as a scoring system has the advantage of allowing comparisons between various EEs and their results [20].

The majority of the EE involved melanoma (n = 16). As melanoma is the oldest indication of checkpoint inhibitors since 2011, and all three molecules have been approved for this indication, it is consistent with a higher number of EEs than other indications. These EEs investigated different aims, including whether a strategy is cost-effective, the cost of adverse effects, the cost-containment strategies etc. Among them, more than 77% reported a strategy with immunotherapy that was cost-effective at the chosen threshold. However, two EEs focused on strategies with a combination of two immunotherapies (ipilimumab plus nivolumab) and a combination of immunotherapy and a BRAF inhibitor (ipilimumab plus vemurafenib) that were not cost-effective [34, 42].

Six EEs concerned non-small-cell lung cancer. In four of them, authors reported that nivolumab or pembrolizumab are cost-effective at the chosen threshold, with a lower cost when the positive status of PD-L1 is taken into account. One EE examined how the perspective of a study can influence its conclusion: from a payer perspective, nivolumab is not cost-effective, but from a societal perspective, using the same threshold, it is cost-effective. Depending on the type of costs taken into account, the conclusion of an EE may change. Two EEs reported nivolumab or pembrolizumab to be cost-effective at the chosen threshold.

Six EEs involved renal-cell carcinoma and all of them were mainly based on results of the CheckMate-025 clinical trial [13]. Among 5 full EEs, only 2 of them reported that nivolumab versus everolimus was cost-effective at a willingness-to-pay threshold of US $100,000 to US $150,000. However, the aims of these EE were different: to assess the cost-effectiveness of nivolumab from an American, English or Canadian perspective, to apply the results to a Chinese context, or to compare the costs and benefits over a 1 or 2-year horizon.

Three EEs concerned head and neck squamous-cell carcinoma [24–26]. All of them were full EEs published in 2017 and based on the same clinical trial results, but the aims of the three studies were different [14]. Two studies were conducted from an American perspective, like the clinical trial, and one EE was conducted from a Canadian perspective. The sources of costs, effectiveness and utility were different for the three studies; so, as a result, the calculated cost and effectiveness of interventions ranged from about US $45,000 to US $175,000 and from 0.248 to 0.796 QALY for nivolumab. Only 1 of the 3 studies found nivolumab to be cost-effective at a threshold of US $150,000. Nevertheless, all three studies were considered “high quality” EE. A previous systematic review concluded that nivolumab was not cost-effective for head and neck cancer [57].

Since 2016, ipilimumab and nivolumab have had the approval of the American Food and Drug Administration as combination therapy in advanced melanoma. With the development of new immunotherapies, it appears likely that more and more treatment strategies will include more than one immunotherapy, either as a combination of drugs or as a sequence treatment. Two recent studies have reported the combination of ipilimumab and nivolumab to be cost-effective [53, 58]. The first study examined only one line of treatment and compared the combination of ipilimumab and nivolumab versus ipilimumab, whereas the second study integrated the combination in a sequential treatment. Both of the EEs found the immunotherapies to be cost-effective at their threshold. So it is quite probable that although the cost of treatment using a combination of immunotherapies or a sequence treatment may rise, the drugs used can remain cost-effective at the chosen threshold.

To our knowledge, the present study is the first systematic review about full or partial EE of ipilimumab, nivolumab and pembrolizumab in their marketed indications. However, it has some limitations. First, our systematic review included only EEs written in English and published before July 2018, excluding EEs published in guidelines. Secondly, unpublished EEs (grey literature) and EEs from other electronic sources, such as LILACS, were not included in our research, which may also introduce some bias.

In conclusion, our systematic review identified 32 EEs published over the past 6 years, in the current context of rational decision-making in health care. Three-quarters of these identified EE were high-quality assessments. This finding tends to encourage the growing interest in developing other effective innovative immunotherapeutic agents. Nevertheless, it reveals a wide variety of methodological approaches, including differences in the perspective used for analysis, time horizon, study design, cost categories, discounting rates, measurement of cost and effectiveness, etc. Two-thirds of the 24 full EEs published before July 2018 showed that a strategy with immunotherapy was cost-effective at the willingness-to-pay threshold ranging from US $30,000 to US $150,000. Whether or not these findings will be confirmed remains to be seen when market approval to cover more indications is extended and new effective immunotherapeutic agents such as atezolizumab, durvalumab, cemiplimab, etc. become available.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CEA

Cost-effectiveness analyses

CUA

Cost-utility analyses

EE

Economic evaluation

HRQOL

Health-related quality of life

HSUV

Health-state utility values

HTA

Health technology assessment

LY

Life years

MeSH

Medical subject headings

QALY

Quality-adjusted life year

UK

United Kingdom

Author contributions

Conceptualization: VN, PF; Methodology: VN, PF, SL; Writing: CC, FP, MK, VN; Literature search: CC, PF; Critically revised work: FA, VW, TM, ATV, CB, MK; Supervision: VN.

Funding

No sources of funding were used to assist in the preparation of this study.

Compliance with ethical standards

Conflict of interest

C. Couchoud, P. Fagnoni, C. Gérard, M. Kroemer and S. Limat have declared no conflicts of interest. F. Aubin is a consultant or investigator for BMS, MSD, Roche, GSK, Novartis, Amgen, and Pierre Fabre and has received support for congresses from BMS, MSD, Novartis and Roche. V. Westeel has received honoraria from Astra Zeneca, BMS, MSD and Roche. T. Maurina has received honoraria from Ipsen, Roche, Janssen and Sanofi. A. Thiery-Vuillemin has received honoraria from Pfizer, Astra Zeneca, Roche, BMS and MSD. C. Borg is an expert for Sanofi, Byer, Servier and MSD. V. Nerich has received honoraria from BMS, Pfizer, Roche, and Sanofi.

Research involving human participants and/or animals

This article does not contain any studies with human participants or animals performed by any of the authors.