Steroid Refractory Chronic Graft-Versus-Host Disease: Cost-Effectiveness Analysis

Abstract

Given the increasing incidence of chronic graft-versus-host disease (cGVHD) and its rapidly escalating costs due to many lines of drug treatments, we aimed to perform a meta-analysis to assess the comparative effectiveness of various treatment options. Using these results, we then conducted a cost-effectiveness analysis for the frequently utilized agents in steroid-refractory cGVHD. We searched for studies examining tacrolimus, sirolimus, rituximab, ruxolitinib, hydroxychloroquine, imatinib, bortezomib, ibrutinib, extracorporeal photopheresis, pomalidomide, and methotrexate. Studies with a median follow-up period shorter than 6 months and enrolling fewer than 5 patients were excluded. Meta-analysis for overall and organ system-specific GVHD response (overall response [ORR], complete response [CR], and partial response [PR]) was conducted for each intervention. Cost per CR and cost per CR + PR were calculated as the quotient of the 6-month direct treatment cost by CR and CR + PR. Forty-one studies involving 1047 patients were included. CR rates ranged from 7% to 30% with rituximab and methotrexate, respectively, and ORR ranged from 30% to 85% with tacrolimus and ruxolitinib, respectively. Cost per CR ranged from US$1,187,657 with ruxolitinib to US$680 with methotrexate. Cost per ORR ranged from US$453 for methotrexate to US$242,236 for ibrutinib. The most cost-effective strategy was methotrexate for all of the organ systems. Pomalidomide was found to be the least cost-effective treatment for eye, gastrointestinal, fascia/joint, skin, and oral GVHD, and imatinib was found to be the least cost-effective treatment for liver and extracorporeal photopheresis for lung GVHD. We observed huge cost-effectiveness differences among available agents. Attention to economic issues when treating cGVHD is important to recommend how treatments should be sequenced, knowing that many patients will cycle through available agents.

INTRODUCTION

Hematopoietic cell transplantation (HCT) is an established procedure for many acquired and congenital disorders of the hematopoietic and the immune system. More than 25,000 allogeneic HCTs are carried out annually worldwide and are increasing each year [1,2]. However, HCT is an expensive procedure and HCT-related costs become increasingly important because of the widespread application of this treatment. According to an Agency for Healthcare Research and Quality report, HCT had the most rapidly growing expenditure among medical procedures between 2004 and 2007, with an increase of 84.9% and a total of US$1.3 billion spent on HCT in 2007 [3]. Most of the studies looking at cost drivers of HCT have focused on early post-transplant costs of HCT [4-9]. The economic impact of long-term costs and chronic graft-versus-host disease (cGVHD) treatments is not clear [10,11].

cGVHD remains the most serious complication affecting long-term survivors of allogeneic HCT [12,13]. Increasing HCT numbers, use of unrelated donors, peripheral blood stem cell utilization, and the inclusion of older recipients utilizing nonmyeloablative regimens have led to an increased incidence of cGVHD [14]. Currently, the therapeutic mainstay for cGVHD is steroids; however, it fails to produce complete response in approximately one-half of the patients and there is no established consensus on the treatment of steroid-refractory cGVHD (SR-cGVHD) patients [15-17]. The available options vary substantially in both cost and effectiveness [18].

Several advances in transplantation techniques and supportive care practices over the last 2 decades have significantly improved survival of patients undergoing HCT [19]. However, along with several other factors (eg, increase in peripheral blood stem cells for HCT), this improved early survival has also contributed to an increase in the incidence of cGVHD. More than 40,000 cGVHD patients are estimated from the last 5 years of trends reported in the registries to require treatment over the next decade [20].

The increasing numbers of patients with cGVHD, and the increasing use of marketed and off-label therapies for them will cause a significant increase in the expenditures of both the private insurers and the Centers for Medicare and Medicaid Services in the future. To delineate the current landscape of cGVHD drug therapy, we aimed to perform a cost-effectiveness analysis (CEA) for the frequently utilized agents in adult patients with SR-cGVHD; before pharmacoeconomic analysis, we performed a comprehensive meta-analysis of the response rates for each drug/extracorporeal photopheresis (ECP) treatment and also for each organ system separately.

MATERIALS AND METHODS

Literature Search Methods

A comprehensive search of several databases for studies published from January 2000 to May 2016 was conducted by an expert medical librarian. The databases searched were: Ovid Medline In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, and Scopus. The searched MeSH terms were graft vs host disease augmented by text words such as chronic, second line, steroid refractory, treatment. The search terms were then translated to the preferred terms of the other databases. This terminology was utilized for management with each one of the following therapies: tacrolimus, sirolimus, rituximab, ruxolitinib, hydroxychloroquine, imatinib, bortezomib, ibrutinib, ECP, pomalidomide, and methotrexate. This study was conducted before the approval of ibrutinib by the U.S. Food and Drug Administration for treatment of SR-cGVHD.

Eligibility Criteria

We included prospective and retrospective studies examining tacrolimus, sirolimus, rituximab, ruxolitinib, hydroxychloroquine, imatinib, bortezomib, ibrutinib, ECP, pomalidomide, and low-dose methotrexate as a sole therapy for SR-cGVHD, which was defined as any disease that failed to respond to previous immunosuppressive therapy with steroids at a dose of ≥.5 mg/kg/day for at least 4 weeks or inability to taper it with or without additional immunosuppressive drugs. There were insufficient publications to analyze mycophenolate mofetil, mesenchymal stem cells azathioprine, daclizumab, basiliximab, thalidomide, anti-TNF antibodies, daclizumab, total nodal irradiation, everolimus, and cyclophosphamide. We excluded studies evaluating any of the aforementioned drugs in combination with another drug when introduced concurrently. Review articles and meta-analysis were excluded. We excluded studies enrolling fewer than 5 patients or with <6 months of follow-up, publications in languages other than English, and those that included any pediatric patients unless they had a separate data analysis for the adult patients in the study.

Study Selection

Two reviewers (S.H. and F.F.Y.) independently considered the potential eligibility of each of the abstracts and titles from the retrieved citations and requested full text versions for these potentially eligible studies. Working separately, reviewers assessed the full text of reports to confirm eligibility.

Data Collection and Extraction

Data were extracted using a predefined data extraction form, including general information about the included studies, participants, and outcomes (all measured outcomes, including overall cGVHD and organ system–specific cGVHD responses). The outcomes measured were complete response (CR), partial response (PR), and overall response (ORR) for overall cGVHD and organ-specific cGVHD as defined in the reports. ORR was defined as the sum of CR and PR. Of note, the response rates were rarely based on formal National Institutes of Health consensus criteria and were usually physician-reported responses [21,22]. The response rate represented the cumulative incidence of the outcome variable in the meta-analysis.

Risk of Bias Assessment

We conducted a quality assessment of each study. Randomized controlled trials were assessed using the Cochrane risk of bias tool [23]. The Newcastle Ottawa Scale [24] was used to assess the methodological quality of nonrandomized studies.

Statistical Methods and Analysis

Meta-analysis

For each study, we estimated the response rates (cumulative incidence) and the associated 95% confidence interval. Response rates were pooled across studies using the random effects model. Meta-analysis was conducted using the STATA statistical software package release 14 (StataCorp, College Station, TX).

Cost-effectiveness analysis

Drug prices were obtained from uptodate.com, representing the average wholesale price. Regarding ECP, cost per session was obtained from the manufacturer’s (Therakos, Mallinckrodt plc, Staines-upon-Thames, United Kingdom) internal data as used previously [25].

Treatment protocols for each drug were determined based on the published reports and if different doses were used in different trials for the same drug, then the standard doses (based on consensus of ≥2 practicing hematologists) for a specific drug were utilized for input for analysis. Direct medical costs of each drug for 6 months of treatment were calculated (Table 1). We did not include the direct nonmedical and indirect costs. Cost per CR and cost per CR + PR were calculated as the quotient of the 6-month cost by CR and CR + PR of each drug respectively (Table 2). Costs per CR + PR for each of the organ system were calculated as the quotient of the 6-month cost by CR + PR of each organ system for each of the drug (Table 3).

Table 1.

Treatment Protocols and Costs for 6 Months of Treatment (United States Only)

Active Ingredient	Brand Names	Dose/Form	Price	Treatment Protocol	6-Month Drug Cost*
Tacrolimus	Tacrolimus	0.5 mg (100 capsules)	$222	.12 mg/kg/d, 6 mo	$6815
		1 mg (100 capsules)	$445
		5 mg (100 capsules)	$2229
Sirolimus	Sirolimus	2 mg (100 tablets)	$3149	6 mg loading, 2 mg/d, 6 mo	$5731
Rituximab	Rituxan	100 mg/10 mL (10 mL)	$1042	375 mg/m2/wk, 4 wk	$29,184
		500 mg/50 mL (50 mL)	$5212
Ruxolitinib	Jakafi	5 mg (60 tablets)	$13,856	10–20 mg twice daily, 6 mo	$83,136
		10 mg (60 tablets)	$13,856
HCQ		200 mg (100 tablets)	$408	800 mg/d, 6 mo	$2938
Imatinib	Gleevec	100 mg (90 tablets)	$10,112	100 mg/d, 6 mo	$20,224
Bortezomib	Velcade	3.5 mg (1 vial)	$1923	.2 mg/m2/week, 6 months	$46,152
Ibrutinib	Imbruvica	140 mg (90 capsules)	$13,323	420 mg/d, 6 mo	$79,938
ECP			$1348	†	$41,788
Pomalidomide	Pomalyst	1 mg (21 capsules)	$17,430	1–4 mg/d, 21 of 28 d per course, 6 courses	$104,580
		2 mg (21 capsules)	$17,430
		3 mg (21 capsules)	$17,430
		4 mg (21 capsules)	$17,430
Methotrexate	Methotrexate Sodium injection	25 mg/mL (2 mL)	$8.5	7.5 mg/m2/wk, 6 mo	$204

HCQ indicates hydroxychloroquine.

^*Cost calculation based on 170 cm height and 70 kg weight and only includes direct costs calculated per described protocol.

^†Three times during week 1, and then twice weekly on consecutive days during weeks 2–12. Responding patients in the ECP group could continue 2 ECP treatments every 4 weeks until week 24.

Table 2.

Response Rates and Costs Per Response Types

Active ingredient	Brand Name	CR Rate (95% CI)	Cost per CR	CR + PR Rate (95% CI)	Cost per CR + PR*
Tacrolimus	Tacrolimus	.17 (.08–.29)	$40,088	.30 (.16–.44)	$22,717
Sirolimus	Sirolimus	.17 (.08–.29)	$33,712	.77 (.57–.92)	$7443
Rituximab	Rituxan	.07 (.02–.14)	$416,914	.62 (.53–.71)	$47,071
Ruxolitinib	Jakafi	.07 (.003–.19)	$1,187,657	.85 (.72–.93)	$97,807
HCQ		.05 (.01–.25)	$58,760	.32 (.15–.53)	$9181
Imatinib	Gleevec	.03 (.00–.13)	$674,133	.46 (.32–.62)	$43,965
Bortezomib	Velcade	.00 (.00–.28)		.50 (.24–.76)	$92,304
Ibrutinib	Imbruvıca	.00 (.00–.39)		.33 (.10–.70)	$242,236
ECP		.11 (.06–.18)	$379,891	.62 (.54–.69)	$67,400
Pomalidomide	Pomalyst	.00 (.00–.30)		.78 (.45–.94)	$134,077
Methotrexate	Methotrexate Sodium injection	.30 (.14–.49)	$680	.45 (.26–.64)	$453

CI indicates confidence interval.

^*Cost calculation based on 170 cm height and 70 kg weight and only includes direct costs calculated per described protocol.

Table 3.

Organ System–Specific Response Rates

Active Ingredient	Eye Rate (95% CI) Cost per CR + PR	GI Rate (95% CI) Cost per CR + PR	Fasciae/Joint Rate (95% CI) Cost per CR + PR	Liver Rate (95% CI) Cost per CR + PR	Skin Rate (95% CI) Cost per CR + PR	Lung Rate (95% CI) Cost /perCR + PR	Mouth Rate (95% CI) Cost per CR + PR
Tacrolimus	NA	NA	NA	NA	NA	NA	NA
Sirolimus	.59 (.33–83) $9714	.57 (.10–98) $10,054	.33 (.06–79) $17,367	.50 (.23–78) $11,462	.64 (.48–78) $8955	NA	.62 (.39–83) $9244
Rituximab	.28 (.09–52) $104,229	.33 (.01–78) $88,436	.77 (.47–98) $37,901	.44 (.19–70) $66,327	.61 (.51–71) $47,843	NA	.52 (.22–81) $56,123
Ruxolitinib	NA	NA	NA	NA	NA	NA	NA
HCQ	.00 (.00–43)	.00 (.00–49)	.00 (.00–66)	.18 (.05–48) $16,322	.25 (.09–53) $11,752	NA	.22 (.06–55) $13,355
Imatinib	.58 (.33–81) $34,869	.72 (.47–92) $28,089	.55 (.30–79) $36,771	.30 (.00–74) $67,413	.56 (.29–81) $36,114	.33 (.14–61) $ 61,285	.30 (.07–58) $67,413
Bortezomib	.67 (.30–90) $68,884	.50 (.09–91) $92,304	.50 (.09–91) $92,304	.00 (.00–79)	.50 (022–78) $92,304	NA	.56 (.27–81) $82,414
Ibrutinib	NA	NA	NA	NA	NA	NA	NA
ECP	.65 (.44–84) $64,289	.68 (.20, 1.00) $61,453	.52 (.31–72) $80,362	.63 (.34–87) $66,330	.65 (.51–78) $64,289	.67 (.35–94) $ 62,370	.69 (.51–84) $60,562
Pomalidomide	.33 (.10–70) $316,909	.50 (.15–85) $209,160	.17 (.03–56) $615,176	.00 (.00–79)	.63 (.31–86) $166,000	NA	.29 (.08–64) $360,621
Methotrexate	.56 (.11, 0,97) $364	.80 (.38–96) $255	.33 (.10–70) $618	.67 (.40–90) $304	.78 (.30, 1.00) $262	NA	.40 (.06–79) $510

NA indicates not available; GI, gastrointestinal.

^*Cost calculation based on 170 cm height and 70 kg weight and only includes direct costs calculated per described protocol.

RESULTS

Selected Studies

The initial search of electronic databases based on MeSH terms identified studies for tacrolimus (189), sirolimus (82), rituximab (122), ruxolitinib (3), hydroxychloroquine (65), imatinib (78), bortezomib (6), ibrutinib (1), ECP (63), pomalidomide (3), and methotrexate (98). Following the review of these, based on the selection criteria, we included studies of tacrolimus (2) [26,27], sirolimus (3) [28-30], rituximab (7) [31-37], ruxolitinib (1) [38], hydroxychloroquine (1) [39], imatinib (6) [40-45], bortezomib (1) [46], ibrutinib (1) [47], ECP (15) [48-62], pomalidomide (1) [63], and methotrexate (3) [64-66] (eTable 1 in the Supplement). Among these, only 1 randomized controlled trial was identified [52]. The quality appraisal for this clinical trial using the Cochrane Collaboration’s tool indicated a low risk of bias considering adequate methodology described for sequence generation, allocation concealment, and for selected outcome reporting.

Pooled Overall cGVHD Responses

Pooled CR and ORR for the studied agents are shown in Figures 1 and 2. In general, CR rates were low with a median of 7% and range of 0% to 30%. ORR rates were higher with a median of 50% and range of 30% to 85%. Response rates were not statistically significant between the agents, supporting presentation of results as cost per response.

Figure 1.

Forest plot of the CR rates following treatments for SR-cGVHD.

Figure 2.

Forest plot of the ORR rates following treatments for SR-cGVHD.

Cost per CR and ORR

Cost per CR and cost per ORR are shown in Table 2 and ranged from US$453 per ORR for methotrexate to US$242,236 per ORR for ibrutinib. The median was US$47,071 per CR + PR. As rates of CR and ORR were fairly similar between agents, cost-effectiveness differences were largely driven by the costs of the agents/treatments.

Pooled Organ System–Specific Responses

Organ system–specific responses were reported for sirolimus, rituximab, hydroxychloroquine, imatinib, bortezomib, ECP, pomalidomide, and methotrexate.

More than half of the patients experienced an ocular ORR with sirolimus, imatinib, bortezomib, ECP, and methotrexate. The highest response rates are reported with bortezomib and ECP, with 67% and 65%, respectively. Hydroxychloroquine was not effective with 0% reported ORR for ocular cGVHD (eFigure 1 in the Supplement).

Gastrointestinal GVHD ORR are shown in eFigure 2 in the supplement and ranged from 0% and 80%, with a median of 50%. Methotrexate is reported as the most effective agent while hydroxychloroquine was not effective.

The best responses to fascia and joint GVHD are observed with rituximab followed by imatinib, ECP, and bortezomib. The ORR was <50% in patients treated with hydroxychloroquine, pomalidomide, sirolimus, and methotrexate (eFigure 3 in the Supplement).

For liver GVHD, more than half of the patients treated with sirolimus, ECP, and methotrexate responded. The highest reported response 67% was with methotrexate. Bortezomib and pomalidomide were not effective with reported response of 0%. (eFigure 4 in the Supplement).

Skin GVHD ORR are shown in eFigure 5 in the supplement and ranged from 25% and 78%. Methotrexate is reported as the most effective agent while hydroxychloroquine was not found to be effective.

The median ORR for oral GVHD is 46% with the highest responses seen with ECP (eFigure 6 in the Supplement). Less than half of the patients treated with hydroxychloroquine, pomalidomide, imatinib, and methotrexate had a response.

For lung GVHD, the information was not available for majority of treatments. The observed pooled ORR was 33% for imatinib, and 67% for ECP (eFigure 7 in the Supplement).

Cost per Organ System Responses

The CEA per organ responses are summarized in Table 3. The most cost-effective strategy was methotrexate for all of the organs systems. On the other hand, pomalidomide was found to be the least cost-effective treatment for eye, gastrointestinal, fascia/joint, skin, and oral GVHD and imatinib was found to be the least cost-effective treatment for liver and ECP for lung GVHD.

DISCUSSION

The national cost of cancer care in 2010 was estimated to be US$124.57 billion, and this cost in 2020 is estimated to rise to US$157.77 billion, representing a 27% increase from 2010. Taken together, leukemia and lymphoma are the third most expensive cancers in women and second most expensive in men from the perspective of management costs [67]. The costs of HCT within the first 100 days or 1 year are quite high, and key cost drivers in management of these diseases. To date, the vast majority of HCT cost-identification studies have focused on early post-transplant costs of HCT (typically the first few months of HCT) so the economic impact of late complications as well as cGVHD remains unclear [6-10,68].

We calculated 6-month direct drug costs for cGVHD in adult patients for the most frequently studied therapies. Enormous differences were observed between various treatments in the CEA with ruxolitinib associated with a cost per CR of US$1,187,657 and methotrexate with a cost per CR of US$680.Furthermore, for organ system–specific responses, a clear signal of cost effectiveness of a particular drug was observed (eg, pomalidomide was found to be the least cost-effective treatment for eye, gastrointestinal, fascia/joint, skin, and oral cGVHD) (Table 3).

There have been some economic evaluations for GVHD reported in the literature, but there are no reported economic evaluations comparing different drugs in SR-cGVHD. Crespo et al. [69] conducted an excellent pharmacoeconomic evaluation of selected cGVHD treatments. They assessed the cost effectiveness of ECP, rituximab, and imatinib in patients with cGVHD using local cost data (consumer price index; Spain) for the aforementioned agents [69]. Unlike our results, they found ECP to be more cost effective than imatinib and rituximab. The main differences are that they used microsimulation techniques for 1000 hypothetical patients and report the cost effectiveness at a 5-year time horizon, whereas we utilized the base case of 6 months of treatment with effectiveness measured by variables including CR and ORR. Another difference is the cost of the reviewed treatments. In another study Jones et al. [20] evaluated the cost burden of cGVHD by the summation of direct and indirect costs from prior studies. They estimated the total 10-year cGVHD cost burden as US$30.2 billion [20].

Our analysis was limited to direct drug costs for cGVHD and did not include costs necessary to administer a drug. Apart from the cost of medication, the direct costs in various pharmacoeconomic evaluations generally include the costs for the medical services including hospital services, physician and nurse services, medical supplies and laboratory monitoring tests, infusion unit costs (where applicable), and concomitant medications. Direct nonmedical costs were also excluded such as parking, tolls, childcare, and other costs related to receipt of treatment. We also did not include indirect nonmedical costs. These costs include years of labor lost attributable to the disease or its treatment as well. In this study, we did not include the direct nonmedical and indirect costs because an accurate measurement of these costs is limited due to the absence of data in published reports. A large randomized trial comparing cord blood to haploidentical HCT is currently ongoing (BMT-CTN 1101 CEA), which has a formal cost-effectiveness substudy built into the protocol to evaluate the most important determinants of both direct and indirect costs. This study is also restricted to 6-month treatment cost. As well known, cGVHD requires systemic immunosuppressive treatment for a median duration of 2 to 3 years depending on the site of the involvement. However, most of the studies have a short follow-up and did not include the proportions of patients still receiving treatment.

Another limitation of our study is the rapidly evolving literature on the effectiveness of agents for SR-cGVHD and our strict inclusion criteria. Based on our inclusion criteria, at the time of the electronic search, only 1 study existed for 4 drugs—pomalidomide, ruxolitinib, ibrutinib, and bortezomib. As new studies are published, the response rates based on meta-analysis for these agents may change, which will likely result in a change of CEA for each treatment. As an example, a recent study exploring the role of ibrutinib in SR-cGVHD demonstrated an ORR of 67% (28 of 42 patients) [70]. Based on this finding, the U.S. Food and Drug Administration approved ibrutinib for the treatment of adult patients with cGVHD after failure of 1 or more lines of systemic therapy. This is the first Food and Drug Administration–approved therapy for the treatment of cGVHD.

Organ-specific cost rates should be interpreted with caution, especially for some fibrotic manifestations such as contractures, bronchiolitis obliterans, and sicca syndrome, where responses are difficult to achieve. Also it should be kept in mind that, there is a proportion of patients that remain with some deficits but off immunosuppression which also implicates “financial CR.”

In conclusion, this CEA illustrates significant variability in costs associated with various treatment modalities for SR-cGVHD, accompanied by heterogeneous effectiveness results. The lack of clinical trials directly comparing these agents and heterogeneity of trial designs and study populations limit firm conclusions about incremental cost effectiveness. Unfortunately, the majority of the clinical trials being conducted for SR-cGVHD in the current decade are single-arm studies, which makes this pharmacoeconomic analysis important to compare the currently available agents with respect to both their effectiveness and costs for use in the clinic. As patients with cGVHD are treated with several different agents over the course of their disease, attention to economic issues when treating cGVHD can help guide how treatments should be sequenced, knowing that many patients will cycle through the currently available agents. When third-party payers reimburse expensive off-label drugs for cGVHD without sufficient evidence of safety and efficacy, they could be hampering the development of high quality data from randomized clinical trials that could help ensure wise use of resources. Enrollment of patients in clinical trials is encouraged to help advance the field and allow evidence-based treatment decisions.