Abstract
Background:
Tyrosine kinase inhibitors (TKIs) such as imatinib mesylate have revolutionized the treatment of primary unresectable and/or metastatic gastrointestinal stromal tumors (GISTs), providing durable disease control and extended survival. Although most patients eventually progress on therapy, dose escalation has been shown to benefit some patients. Sunitinib, a multitargeted kinase inhibitor is effective against imatinib-resistant or intolerant GIST patients. Although the cost of TKI therapy in GIST is high, no other effective systemic treatment options exist.
Objective:
Review pharmacoeconomic studies to determine the cost effectiveness (CE) of 1st- and 2nd-line TKI therapies in GIST.
Methods:
A literature review using Medline and PubMed databases was conducted to identify published economic analyses of TKI therapy in GIST. Key results from these studies were analyzed.
Results:
Six pharmacoeconomic studies were identified, including three analyses of 1st-line imatinib and three analyses of 2nd-line sunitinib. These studies employed various time horizons and discount rates and modeled CE from a number of different perspectives. Most of the pharmacoeconomic studies reviewed used survival as their efficacy endpoint, projecting outcomes beyond available data to model CE. Analyses of 2nd-line sunitinib using survival additionally faced the challenge of adjusting for the effect of placebo crossover to active treatment in the pivotal phase III study. Most studies used Markov techniques with a range of transition probabilities.
Conclusions:
Published pharmacoeconomic studies of 1st- and 2nd-line TKI therapy for advanced GIST employ various time horizons, discount rates, and different CE models. Consequently, these differences make comparisons between studies difficult. Studies of 1st-line imatinib concluded that imatinib was cost effective in advanced, metastatic GIST. Likewise, based on data reviewed here, 2nd-line sunitinib appears to be cost effective in patients with advanced GIST who are intolerant/resistant to imatinib. Key limitations of this review included inconsistency among the studies evaluated with regard to methodologies, countries of origination (currency and healthcare systems), and patient demographics.
Introduction
Gastrointestinal stromal tumors (GIST) are the most common mesenchymal tumors of the gastrointestinal tract, occurring in an estimated 7 to 14 persons per 1 million in the general populationCitation1–7. Initial treatment for localized GIST is usually surgical resection, but recurrence and metastasis following surgery is commonCitation8–10. The discovery of the central oncogenic role of gain-of-function mutations of KIT and PDGFRA in GIST prompted the development of the tyrosine kinase inhibitor (TKI) imatinib mesylate (Glivec) for treatment of advanced, metastatic diseaseCitation11–14. Introduction of imatinib has transformed the way advanced GIST is managed, increasing the median overall survival (OS) from 19 months prior to the availability of imatinib to 57 months in the post-imatinib eraCitation9,Citation15. Imatinib is now recognized as the standard of care in untreated unresectable and/or metastatic GISTCitation8,Citation16,Citation17.
The standard recommended dose of imatinib in unresectable and/or metastatic GIST is 400 mg/dayCitation8,Citation17. However, recent results have demonstrated the benefits of high-dose imatinib (800 mg/day) in patients with resistant disease, as well as in those with specific high-risk genotypes (e.g., KIT exon 9). The phase III randomized European Organisation for Research and Treatment of Cancer (EORTC) 62005 study, which assessed the activity of imatinib at two dosing levels (400 and 800 mg/day), found that dose escalation with imatinib appeared to restore tumor control in a substantial proportion of patients who developed resistant diseaseCitation18–20. Among patients who were randomized to standard-dose treatment and crossed over to imatinib 800 mg/day upon progression (n = 133), 29% achieved either partial response (PR) or stable disease (SD). Median progression-free survival (PFS) following crossover was 81 days and 18.1% of patients were alive and progression-free after 1 yearCitation20. High-dose imatinib also appeared to benefit high-risk patients with KIT exon 9 mutations. This group had significantly longer PFS with 800 mg/day than with 400 mg/day.
Results from the EORTC 62005 study with regard to the efficacy of high-dose imatinib in primary tumors with KIT exon 9 mutations were supported by findings from the MetaGIST analysis of combined data from the EORTC 62005 and the Southwest Oncology Group (SWOG) S0033 phase III studiesCitation21,Citation22. This analysis showed that dose had a significant effect on overall PFS and strong effect on PFS in patients with KIT exon 9 mutations (the analysis showed no effect on OS). Among patients with KIT exon 9 mutations, median PFS was significantly higher in the 800 mg/day group (1.59 years) than in the 400 mg/day group (0.5 year) (p = 0.017)Citation22. Based on these results, both the European Society for Medical Oncology (ESMO) and National Comprehensive Cancer Network (NCCN) have recognized the usefulness of imatinib dose escalation in patients with progressive disease (PD) and those with high-risk mutationsCitation8,Citation17,Citation22.
Although long-term disease control is observed in most imatinib-treated patients, some patients show intolerance or primary resistance to imatinib and acquired, secondary resistance is observed in most responding patients over timeCitation15,Citation23. For patients who are intolerant/resistant to imatinib, the 2nd-generation TKI inhibitor sunitinib (Sutent) has been approved and endorsed by worldwide guidelines as 2nd-line treatmentCitation8,Citation17,Citation24. Sunitinib is a multitargeted TKI with in vitro activity against known imatinib-resistant clones, including secondary mutations in exons 13 and 14Citation25,Citation26. This activity has been shown to correlate with clinical benefit. Among patients with imatinib-resistant disease, rates of clinical benefit (PR or SD ≥6 months) were 58% for patients with primary KIT exon 9 mutations, 34% for those with KIT exon 11 mutations, and 56% for those with wild-type KIT or PDGFRACitation25. The efficacy and safety of sunitinib in patients intolerant/resistant to imatinib was demonstrated in a large phase III randomized, controlled trial, in which sunitinib 50 mg/day (schedule 4 weeks on, 2 weeks off [4/2]) was effective in extending both time to progression and OSCitation27,Citation28.
Two other newer-generation TKIs, nilotinib (Tasigna) and sorafenib (Nexavar) have demonstrated activity in imatinib- and sunitinib-resistant disease and may be potentially considered as 3rd-line therapy optionsCitation8. Both agents are multitargeted TKIs and both have been shown to be modestly effective in achieving prolonged PFS in patients who have failed treatment with imatinib and sunitinibCitation29–32. Several other systemic treatments, including TKIs and agents that target other known oncogenic mechanisms in GIST, are currently under clinical investigation in 2nd- and 3rd-line settingsCitation26.
Results from clinical studies examining long-term efficacy of TKI therapy in GIST have demonstrated that major gains in survival require continuous, long-term administration at adequate dosing levels. Interruption of TKI therapy often results in rapid tumor progression. The ongoing phase III French Sarcoma Group BFR14 study evaluated the effects of interrupting versus continuing imatinib treatment following predefined positive responsesCitation33–35. Following a treatment period of either 1 or 3 years, groups of responding patients (those with complete response [CR], PR, or SD) were then randomized to subsequent continuous or interrupted imatinib treatment. In each case, interruption of therapy, even after CR was achieved, resulted in higher rates of relapse and decreased PFSCitation33–35.
Cost of TKI therapy in advanced GIST
Like many novel cancer therapeutics, TKIs are quite costly. At average wholesale prices, the normal starting dose of imatinib (400 mg) is $145 per day. If higher doses of imatinib are required, the cost increases proportionately, whereas switching to sunitinib as 2nd-line therapy (50 mg/daily for 4 weeks out of 6) entails an average cost of $203 per day. While actual reimbursements are typically discounted from average wholesale prices, these figures suggest that private and public sector payers for healthcare will want evidence that the benefits of treatment are commensurate with cost.
Wide variations in CE estimates for a given treatment often arise due to differences in healthcare markets, sources of efficacy and safety data, and study methodology. In the present article, the authors review existing published pharmacoeconomic studies of TKI therapy in advanced GIST, provide an overview of the methods used, and discuss major findings. Important study limitations are also highlighted and how these may impact estimates of CE discussed.
Patients and methods
To identify existing pharmacoeconomic studies of TKI therapy in advanced GIST, Medline and PubMed searches for articles published in English were conducted between August 15, 2002, and June 30, 2009, using the following terms, alone or in combination: cost effectiveness, tyrosine kinase inhibitor, imatinib, sunitinib, gastrointestinal stromal tumor, GIST, Markov models, reimbursement, and cost utility. Bibliographies of identified published articles were also reviewed for additional publications. Studies were included that involved some form of economic analysis (e.g., budget impact, cost minimization, and CE). In the critical appraisal of the studies, several key variables were considered, including model characteristics and assumptions and methods used to synthesize efficacy evidence, estimate cost associated with disease, and handle uncertainty in model parameters.
Results
The literature review identified six published pharmacoeconomic studies of TKI therapy in advanced GIST (). These included three analyses of 1st-line therapy with imatinibCitation36–38 and three analyses of 2nd-line therapy with sunitinib, two of these comparing sunitinib versus best supportive care (BSC)Citation39,Citation40 and one analysis comparing sunitinib versus high-dose imatinibCitation41.
Table 1. Pharmacoeconomic studies of 1st- and 2nd-line therapies in advanced GIST.
Pharmacoeconomic studies of 1st-line therapy
Wilson and colleagues (2005) evaluated the CE of imatinib in advanced GIST relative to other treatment approaches, using a Novartis-developed state-transition model with two arms: imatinib and controlsCitation38. In the model, controls had two states, PD or death, while patients in the imatinib arm had three states, PD, death, or response (PR or SD). Because survival estimates for imatinib were deemed too favorable upon review by the National Institute for Clinical Excellence (NICE), the Novartis model was modified both structurally and in terms of the survival curve used for controls with PD. The structural modification (Modified-A) applied the same survival curve from the unpublished report of the B2222 study that served as the data source for imatinib to patients in the control arm. Because survival among controls was still judged to be underestimated based on B2222 data, the model was further modified (Modified-B) to include the survival curve for all patients with metastatic or recurrent GIST, as well as the exponential time to treatment failure curve and imatinib dose, based on available data from the B2222 study. CE results were published for the original Novartis model as well as the two versions of the modified Novartis model, with results from Modified-B deemed adequate for NICE requirements. To provide a comparison for results from the Novartis model, a second analysis (referred to as the Birmingham model) was performed using a four-state probability Markov methodology. The perspective of these analyses was the United Kingdom National Health Service (UK NHS) and the time horizon was 10 years. They compared imatinib (400–600 mg/day) with no imatinib (historical controls). The source for economic data was UK NHS system data, including cost of drug acquisition, outpatient visits, computed tomography scans, general practitioner visits, and management of adverse events (AEs). The sources for data on the efficacy of imatinib consisted of an open-label, multicenter trial conducted in 147 patients comparing two doses of imatinib (400 and 600 mg/day), with a median follow-up of 25 months. Efficacy data for controls was taken from two studies of conventional therapies. Rates of discounting were 6% for costs and 1.5% for health benefitsCitation38.
According to the modified Novartis model (Modified-B), the estimated cost per quality-adjusted life-year (QALY) was UK£85,224 (range: UK£51,515–98,889) after 2 years, UK£41,219 (UK£27,331–44,236) after 5 years, and UK£29,789 (UK£21,404–33,976) after 10 years. Results from the Birmingham model were within the range of estimates from the Modified-B Novartis model.
Huse and colleagues (2007) estimated the CE of imatinib in advanced GIST projecting long-term survival and duration of imatinib benefit based on data from a phase II open-label trial of imatinibCitation36. The perspective of the study was the US healthcare market and the time horizon was 10 years. Costs, life-years, and QALYs were discounted to treatment initiation at an annual rate of 3%. Weekly cost estimates made in 2005 US dollars, including the cost of imatinib 400 mg/day (US$685), other medical services for imatinib patients (US$359), and palliative care for patients in the end stage of GIST (US$2575), were based on recent study of costs of care in pancreatic cancer. The source for efficacy and safety data were 52-month results from a pivotal trial (uncontrolled) of imatinib in advanced GIST (N = 147)Citation15,Citation23. Weibull curves were fitted to 52-month data to estimate probability of survival and time to discontinuation. Utility associated with successful treatment was estimated at 0.935 and treatment failure and PD at 0.875. Sensitivity analysis (univariate) of CE was conducted in variation to each model parameter across a range of values. The study found that imatinib was projected to increase life expectancy to 5.8 years, an increase of 2.7 years over the control group (102 patients who withdrew due to treatment failure). This translated into an increase of 1.9 QALYs at a marginal cost of US$74,369, yielding a CE ratio of US$38,723 per QALY. The CE of imatinib was somewhat sensitive to changes in model parameters, particularly drug costCitation36.
Mabasa and colleagues (2008) evaluated the CE of imatinib in advanced GIST in British Columbia Cancer Agency (BCCA) patientsCitation37. The study compared the CE of imatinib 400 mg/day, with increase to 600–800 mg/day on evidence of progression (n = 46), with no imatinib (BCCA historical controls before availability of imatinib: n = 47). Economic data were derived from the BCCA Systemic Therapy Database using a previously published costing template based on total direct costs (drug, labor, and supplies) in 2006 Canadian dollars. Efficacy and safety data were derived from a retrospective analysis of BCCA medical records. For treatment continued beyond 1 year, annual discounting rates of 0%, 3%, and 5% were applied. The primary outcome for the study was CE based on median OS. Secondary outcomes included CE based on median PFS and comparison to efficacy from published literature. Median OS and PFS were assessed using Kaplan–Meier analysis with 95% confidence interval (CI), with comparisons between groups made using the chi-square test. Log rank test was used to compare median OS and PFS between groups. Average life expectancy (mean survival) was estimated using area under the curve for Kaplan–Meier curves. Sensitivity analyses varying effectiveness over the 95%CI, cost to its extremes, discounting level at 0%, 3%, and 5%, and substituting life expectancy for median OS were also performed. Median OS for imatinib (n = 46) and historical controls (n = 47) was 66.7 months (95%CI, 61.7 to infinity) and 7.7 (95%CI, 6.0–12.6), respectively (p < 0.001). Median PFS for imatinib and historical controls was 45.3 months (95%CI, 24.4 to infinity) and 5.6 months (95%CI, 3.5–8.5), respectively (p < 0.001). OS at 1 year for imatinib was 95.4% (95%CI, 82.9–99.2) vs. 32.6% (95%CI, 20.0–48.1) for historical controls. PFS at 1 year for imatinib was 81.4% (95%CI, 66.1–91.1) vs. 17.4% (95%CI, 8.3–32.0) for historical controls. Kaplan–Meier area under the curve life expectancy for imatinib was 4.34 vs. 0.83 years for controls, yielding an incremental CE ratio (ICER) of Canadian (CAN) dollars CAN$22,247 per life-years gained (LYG), varying from CAN$15,104 to CAN$37,665 per LYG when life expectancy varied across the 95%CI. Imatinib effectiveness was similar to published interim results for imatinib 400 mg/day from two ongoing phase III studiesCitation42,Citation43. The annual ICER for imatinib was CAN$15,882 per median LYG and CAN$23,603 per median year of PFS. Univariate sensitivity analyses showed that median OS and PFS varied widely across 95%CI and costs were varied to extent of range, but even with varying median OS and PFS, CE results remained robust.
Pharmacoeconomic studies of 2nd-line therapy
Two studies examined the CE of 2nd-line therapy, comparing sunitinib with BSCCitation39,Citation40. Chabot and colleagues (2008) evaluated 2nd-line sunitinib in GIST for the Canadian Health System (Provincial Health Ministry) as a case study in challenges that reimbursement requirements place on pharmacoeconomic analyses of cancer drugsCitation39. The authors used a Markov model to simulate disease progression and death in estimating QALYs and LYG based on interim results from a randomized, controlled phase III pivotal trial of sunitinib in patients who failed imatinib therapyCitation27. The study compared the CE of sunitinib (50 mg/day, 4/2 schedule) with BSC. Estimates of direct costs (cost of sunitinib and BSC, routine follow-up for sunitinib treatment, treatment of clinically significant AEs, and end-of-life costs) were obtained from published literature, Canadian government benefit schedules, and expert opinion. The CE model employed in this study used the following outcomes: PFS, OS, utility (measured by EuroQol5 dimensions [EQ-5D] questionnaire), and treatment-related AEs. Uncertainty in model was evaluated using univariate sensitivity analyses of parameters based on extreme values for each parameter, with the exception of the acquisition of sunitinib. This cost analysis resulted in an ICER for sunitinib versus BSC of CAN$79,884 per QALY gained and CAN$49,826 per LYG. The difference for each patient in total average lifetime cost of treatment between sunitinib and BCS was CAN$34,493. Sunitinib patients spent an average of 0.5 years in progression-free health and 1.1 years with PD (mean survival: 1.6 years) compared with 0.2 years and 0.7 years for BCS (mean survival: 0.9 years). Compared with BCS, treatment with sunitinib resulted in an estimated 0.7 LYG and 0.4 QALYs gained. Sensitivity analyses showed that results were robust to changes in most model parameters, with the cost–utility ratio most sensitive to changes in health-state utility (disease progression) and survival. Paz-Ares and colleagues (2008) evaluated the CE of sunitinib (50 mg/day, 4/2 schedule) compared with BSC in GIST patients intolerant of or resistant to imatinib from the perspective of the Spanish National Health SystemCitation40. The study used a Markov model with transition probabilities between outcomes (PFS, progression, and death) based on interim results from a phase III pivotal trial of sunitinib as 2nd-line therapy in GISTCitation27. The time horizon for the analysis was 6 years, with a discount rate of 3.5% applied to cost and effects after year 1. The ICER was expressed as per month of PFS and per LYG and QALY gained. Quality of life (utility data) was obtained from EQ-5D scores taken from the phase III sunitinib trial. Economic analysis used estimates of direct costs (drugs, medical visits, laboratory and radiology tests, palliative care, and AEs) obtained from an expert panel. All costs are expressed in 2007 euros. Deterministic and probabilistic sensitivity analyses were conducted as part of the cost analysis. Additionally, an alternative cost scenario was performed to take into account that some patients might receive imatinib 400 mg/day as part of palliative care after progression. Therapy with sunitinib was associated with higher projected PFS years, LYG, and QALYs gained compared with BSC, at 0.50 vs. 0.24, 1.59 vs. 0.88, and 1.00 vs. 0.55, respectively. The mean per-patient cost associated with sunitinib was €23,259 compared with €1622 for BSC. Therapy with sunitinib was associated with an ICER of €4090 per month of PFS, €30,242 per LYG, and €49,090 per QALY gained. The most influential variables (univariate analysis) were OS and unit cost of sunitinib. There was a variation of ±25% in OS hazard ratio results in ICER/QALY range of €39,201 and €62,806.
Contreras-Hernandez and colleagues (2008) evaluated 2nd-line treatments recommended by NCCN for advanced GIST, including high-dose imatinib and sunitinib, to determine the cost and CE of each from the perspective of the Mexican insurance systemCitation41. The time horizon of the study was 5 years and a discount rate of 5% was applied to cost and effects after year 1. The study compared the estimate of mean cost, CE, and overall benefit of increased imatinib dosage to 800 mg/day, switching to sunitinib (50 mg/day, 4/2 schedule), and regulating symptoms with palliative care. A Markov model was used. The estimate of mean cost of sunitinib was based on the cost amount not reimbursed at time of study. Sensitivity analyses (based on results from the Markov model) were used to predict the mean cost and likelihood of reimbursement. Survival estimates (5 years) were extrapolated based on Weibull curves. Data on direct costs and therapies associated with 2nd-line GIST were derived from NICE values for reimbursement for advanced GIST. Cost for imatinib and palliative care were based on hospital data for a sample of patients (N = 21). Cost estimates for sunitinib were based on data provided by Pfizer. ICER was based on LYG and progression-free months (PFMs). Sources for the efficacy and safety of sunitinib included interim results from a phase III pivotal studyCitation27 and a review of current clinical data on sunitinib in GIST through 2006Citation44. Efficacy and safety of imatinib and palliative care were based on hospital data for a sample of patients (N = 21) who received 2nd-line imatinib or palliative care. High-dose imatinib was associated with a mean annual cost per patient of US$35,226 (±US$1,253) compared with US$17,806 (±$US695) for sunitinib and US$2,072 (±US$473) for palliative care. The ICER over a 5-year horizon for sunitinib versus palliative care was US$15,734. Mean ICERs for sunitinib, imatinib, and palliative care were US$11,862, US$28,424, and US$1,754, respectively. Sunitinib was cost effective for 38.9% of patients versus palliative care. Mean PFM and LYG for sunitinib were 5.64 and 1.4 (95%CI, 1.3–1.6) compared with 5.28 and 1.31 (95%CI, 1.2–1.4) for imatinib and 2.58 PFM and 1.08 LYG (95%CI, 1–1.3) for palliative care.
Discussion
Among the studies considered in this review, there were three studies of 1st-line imatinib therapyCitation36–38 and three studies of 2nd-line sunitinib therapyCitation39–41. Studies of 1st-line imatinib used various methodologies and different perspectives. In general, 1st-line imatinib appeared to be cost effective, but with a wide range of CE ratios. While lower CE ratios resulted from models using the most mature survival data and the one retrospective study using real-world survival data, valid comparisons of CE between studies are not possible due to differences in healthcare systems between countriesCitation37.
The CE of 2nd-line therapy is less well-studied, given the choice between high-dose imatinib and sunitinib. Chabot and colleagues (2008) and Paz-Ares and colleagues (2008) concluded that sunitinib was cost effective as a 2nd-line therapy compared with BSC for GIST after imatinib failureCitation39,Citation40. The study by Contreras-Hernandez and colleagues (2008)Citation41 is the only existing study of 2nd-line therapy comparing sunitinib and high-dose imatinib. This CE analysis predicts cost savings with the use of sunitinib, which follows logically from a simple cost comparison of sunitinib 50 mg/day (4/2 schedule) versus imatinib 800 mg/day.
In modeling CE, the majority of the pharmacoeconomic studies of 1st- and 2nd-line therapy in advanced GIST reviewed here relied on extrapolating survival outcomes from clinical trials beyond available follow-up data. While this practice contributes a certain level of uncertainty to CE models, current guidelines for CE analysis do not allow use of surrogate endpoints that are increasingly accepted in regulatory review for oncology therapeutics. In fact, clinical studies are often stopped early when it is determined that an agent is effective, often hindering the availability of mature data that healthcare policymakers require, leaving the CE model as the only tool to extrapolate survival outcomes.
Two of the three studies of 1st-line imatinib therapy used CE models relying on estimates of long-term survival based on interim study resultsCitation36,Citation38. These studies differed in maturity of survival data from the B2222 study, with Wilson and colleagues (2005)Citation38 using 25-month follow-up results and Huse and colleagues (2007)Citation36 using 52-month mean follow-up results. Among CE studies of imatinib, only Mabasa and colleagues (2008)Citation37 used survival data based on medical records (Canadian Health System), without relying on projections of survival. All three pharmacoeconomic studies of 2nd-line therapy relied on extrapolation of survival curves based on interim results from the same phase III pivotal study of sunitinibCitation39–41.
While valid comparisons of CE estimates between studies are not possible due to differences in perspective, the influence of underlying survival estimates on CE results is apparent in the several studies reviewed here. In Wilson and colleagues (2005)Citation38 (using 25-month data from the B2222 study), survival curves for the modified Novartis model (Modified-B) were based on survival rates for treated patients at 1 year (88%) and 2 years (78%), adjusted to correct for the disproportion of survival in the imatinib arm. The survival curve for untreated patients with advanced disease was based on estimated median survival of 12 months (range 2–20 months) based on published reports. This curve was adjusted for the modified Novartis model (Modified-B) to correct for bias against long-term survivors resulting in an underestimate of survival in this group. By comparison, Huse and colleagues (2007)Citation36 estimated CE based on more mature survival data (median follow-up 52 months from the B2222 study). Using fitted Weibull functions, this analysis estimated a mean survival of 70 months (median: 55 months) for all treated patients compared with 37 months (median: 20 months) for nonresponders from the same trial. Use of untreated patients from this study as controls resulted in a superior survival compared with historical controls due to exclusion from the study of patients with the worst prognosis. Survival estimates from these studies contrast further with results from Mabasa and colleagues (2008)Citation37, which were based on real-world survival data taken from the Canadian Health System. This study estimated survival based on data from 93 patients and reported a median OS of 66.7 months for imatinib versus 7.7 months for controls. Unlike both Wilson and colleagues (2005) and Huse and colleagues (2007), Mabasa and colleagues (2008) did not collect quality-of-life data and therefore did not report QALYs. Also, it should be noted that survival distributions may be skewed, and the use of models can underestimate or overestimate the cost per life-years.
Pharmacoeconomic studies of 1st-line therapy generally involved thorough and adequate sensitivity analyses, using univariate analyses to examine sensitivity to uncertainty in a range of model parameters including survival, costs, utility, time horizon, and discount rates. Due to the nature of each model used by Wilson and colleaguesCitation38 and Huse and colleaguesCitation36, drug and survival costs were in ranges consistent with those previously reported in CE analyses for oncology interventionsCitation45. In Mabasa and colleagues (2008), the one study based on real-world data, CE estimates remained robust, even with widely varying median OS, PFS, and costsCitation37.
In two of the pharmacoeconomic studies of 2nd-line sunitinib, the methods used to handle uncertainty in cost analysis were adequate, with univariate sensitivity analyses conducted on several model parametersCitation39,Citation40. Paz-Ares and colleagues (2008) acknowledged that utility data were lacking for the Spanish population and that utility parameters may differ substantially in the Spanish National Health System and may have influenced quality-of-life assessment and treatment costsCitation40. In Chabot and colleagues (2008), the CE model was highly sensitive to uncertainty in utility valuation based on EQ-5D resultsCitation39. In Contreras-Hernandez and colleagues (2008), mean values for LYG and PFM were used as the final measures of the effectiveness of treatmentCitation41. There was overlap in the LYG 95%CI for sunitinib and imatinib (95%CI for PFM values were not provided), indicating possible problems of indirect comparison and uncertainty. Additionally, the sensitivity analysis is vague and covers only incremental LYG for sunitinib and palliative care with no univariate analyses. Additionally, no utility analyses were provided.
The majority of pharmacoeconomic studies reviewed here used Markov techniques to model CE. This method is more practical than decision-tree analysis in modeling outcomes in chronic diseases such as GIST. The model tracks patients through a series of cycles or fixed time periods. During each cycle, patients may remain in their current health state or move to another state (response or progression), depending on defined transition probabilities. Wilson and colleagues (2005)Citation38 used a Markov model with constant conditional transition probabilities. Such a model does not allow for change in treatment failure and mortality risk over time. By contrast, Huse and colleagues (2007)Citation36 used Weibull curves, which allow for risks of treatment failure and mortality to change over time. As noted previously, Mabasa and colleagues (2008)Citation37 estimated CE using Kaplan–Meier analysis of OS based on Canadian Health System data. All three pharmacoeconomic studies of 2nd-line therapy employed Markov models in estimating CECitation39–41.
There were various assumptions related to cost data used in CE studies of 1st-line imatinib. For instance, Wilson and colleagues (2005) assumed that monthly costs of treatment for GIST decline sharply after imatinib failureCitation38. By contrast, Huse and colleagues (2007) assumed that end-of-life costs after disease progression increase above those incurred during imatinib treatmentCitation36. This may reflect differences in end-of-life clinical decision-making between UK and US healthcare systems. Huse and colleagues (2007) based palliative care cost estimates on a study of the cost of care in pancreatic cancerCitation36,Citation46. In Mabasa and colleagues (2008), costs were based on Canadian Health System dataCitation37. In two pharmacoeconomic studies of 2nd-line therapy, appropriate resources within the respective health systems for each study served as sources for economic data, with Chabot and colleagues (2008)Citation39 taking cost data from the Canadian Health System and Paz-Ares and colleagues (2008)Citation40 from the Spanish National Health System. Increased cost estimates in Chabot and colleagues (2008) versus the two other studies are attributable to increased acquisition cost of sunitinib in Canada versus Spain. Contreras-Hernandez and colleagues (2008) provided little information on basis of cost for sunitinib and imatinibCitation41.
Several approaches were used to discount costs and outcomes after the first year of treatment with 1st-line imatinib. Wilson and colleagues (2005) used a higher rate of discount (6%) for costs and a lower rate for outcomes (1.5%)Citation38. This approach tends to reduce the CE ratio and makes comparisons with other studies difficult. By contrast, Huse and colleagues (2007) used rates of 3% and 5% for both costs and utilitiesCitation36,Citation46. In pharmacoeconomic studies of 2nd-line therapy, annual discount rates for benefits and outcomes after year 1 varied, with Chabot and colleagues (2008)Citation39 and Contreras-Hernandez and colleagues (2008)Citation41 using a discount rate of 5% and Paz-Ares and colleagues (2008)Citation40 using a discount rate of 3.5%.
In cost analyses, as in other types of studies, the selection of a comparison (nontreatment) group has important implications for study outcomes. Ideally, comparison groups should be similar in terms of baseline demographics and should not reflect bias in terms of response to therapy. In their analysis of 1st-line imatinib, Huse and colleagues (2007)Citation36 used nontreatment groups made up of patients who discontinued imatinib treatment due to progression or intolerance. Use of such controls may have resulted in an underestimate of effect of treatment, as imatinib may have slowed the progression of disease in patients who did not exhibit objective response. Wilson and colleagues (2005)Citation38 used historical controls from two studies of conventional therapies and Mabasa and colleagues (2008)Citation37,Citation38 used real-world patient data from before the availability of imatinib. With respect to both of these studies, questions can be raised concerning the similarity of baseline characteristics between active and control groups.
The single pharmacoeconomic study that compared 2nd-line sunitinib and high-dose imatinib was the study by Contreras-Hernandez and colleagues (2008)Citation38,Citation41. Although CE of standard-dose imatinib has been established, the CE of high-dose imatinib or sunitinib in the 2nd-line setting remains to be determined. A major limitation inherent in the study is the lack of head-to-head comparisons between the treatments included in the analysis, although this represents a limitation associated with most CE studies. Sunitinib patients (N = 321) are ostensibly taken from the randomized, controlled, phase III pivotal study. The institutional sample of patients who received high-dose imatinib or palliative care was much smaller (N = 21), with no assurance that it was comparable in terms of baseline demographics. The sunitinib phase III pivotal study was conducted in imatinib-resistant patients in whom the median dose of previous imatinib therapy was 800 mg. By contrast, no determination of level of PD was made in the small institutional sample of high-dose imatinib patients used in this comparison. None of the various strategies (e.g., mixed treatment comparison) recommended in the literature was used to address this fundamental weaknessCitation47,Citation48.
Conclusion
Results from pharmacoeconomic studies support the CE of 1st-line imatinib in advanced, metastatic GIST. In addition, studies of 2nd-line sunitinib demonstrate the CE of this agent as 2nd-line therapy for advanced GIST after failure of imatinib. Further studies are needed to adequately determine the CE of high-dose imatinib in comparison with sunitinib in the 2nd-line setting. Moreover, because these treatments are used in resistant disease and may have benefits for different patient populations, based on risk level and tumor genotyping, further studies of 2nd-line options are needed in patients with specific KIT and PDGFRA mutations to determine their use in these patient populations. Overall, greater consistency in methods would improve the comparability and interpretability of CE estimates.
Transparency
Declaration of funding
The authors have not received financial support for the research reported in this article.
Declaration of financial/other relationships
All authors have completed the disclosure declaration. C.D.B. is chief of the Division of Medical Oncology at the University of British Columbia and has been a consultant/advisor for both Novartis Pharmaceuticals and Pfizer Pharmaceuticals. D.M.H. is vice president of the healthcare division at Thomson Reuters and has no relationships to declare.
Acknowledgments
Financial support for medical editorial assistance was provided by Novartis Pharmaceuticals. The authors thank Angelo Russo, PhD, for his medical editorial assistance with this manuscript.
Notes
* Glivec is a registered trade name of Novartis Pharmaceuticals, Basel, Switzerland.
† Sutent, is a registered tradename of Pfizer, New York, NY, USA.
‡ Tasigna is a registered tradename of Novartis Pharmaceuticals, Basel, Switzerland.
§ Nexavar is a registered tradename of Bayer HealthCare Pharmaceuticals, Montville, NJ, USA.
References
- Goettsch WG, Bos SD, Breekveldt-Postma N, et al. Incidence of gastrointestinal stromal tumours is underestimated: results of a nation-wide study. Eur J Cancer 2005;41:2868-72
- Mucciarini C, Rossi G, Bertolini F, et al. Incidence and clinicopathologic features of gastrointestinal stromal tumors. A population-based study. BMC Cancer 2007;7:230
- Nilsson B, Bumming P, Meis-Kindblom JM, et al. Gastrointestinal stromal tumors: the incidence, prevalence, clinical course, and prognostication in the preimatinib mesylate era – a population-based study in western Sweden. Cancer 2005;103:821-9
- Nishida T, Hirota S. Biological and clinical review of stromal tumors in the gastrointestinal tract. Histol Histopathol 2000;15:1293-301
- Rubio J, Marcos-Gragera R, Ortiz MR, et al. Population-based incidence and survival of gastrointestinal stromal tumours (GIST) in Girona, Spain. Eur J Cancer 2007;43:144-8
- Tran T, Davila JA, El Serag HB. The epidemiology of malignant gastrointestinal stromal tumors: an analysis of 1,458 cases from 1992 to 2000. Am J Gastroenterol 2005;100:162-8
- Tryggvason G, Gislason HG, Magnusson MK, et al. Gastrointestinal stromal tumors in Iceland, 1990-2003: the Icelandic GIST study, a population-based incidence and pathologic risk stratification study. Int J Cancer 2005;117:289-93
- NCCN Clinical Practice Guidelines in Oncology. Soft Tissue Sarcoma. V2.2010. Available at: http://www.nccn.org. Last access April 14, 2010
- Dematteo RP, Lewis JJ, Leung D, et al. Two hundred gastrointestinal stromal tumors: recurrence patterns and prognostic factors for survival. Ann Surg 2000;231:51-8
- Demetri GD, Benjamin RS, Blanke CD, et al. NCCN Task Force report: update on the management of patients with gastrointestinal stromal tumor (GIST) – update of the NCCN clinical practice guidelines. J Natl Compr Canc Netw 2010;8(Suppl 2):S1-S41
- Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating mutations in gastrointestinal stromal tumors. Science 2003;299:708-10
- Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998;279:577-80
- Hirota S, Ohashi A, Nishida T, et al. Gain-of-function mutations of platelet-derived growth factor receptor alpha gene in gastrointestinal stromal tumors. Gastroenterology 2003;125:660-7
- Joensuu H, Roberts PJ, Sarlomo-Rikala M, et al. Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. N Engl J Med 2001;344:1052-6
- Blanke CD, Demetri GD, von Mehren M, et al. Long-term results from a randomized phase II trial of standard- versus higher-dose imatinib mesylate for patients with unresectable or metastatic gastrointestinal stromal tumors expressing KIT. J Clin Oncol 2008;26:620-5
- Blackstein ME, Blay JY, Corless C, et al. Gastrointestinal stromal tumours: consensus statement on diagnosis and treatment. Can J Gastroenterol 2006;20:157-63
- Casali PG, Blay JY. On behalf of the ESMO Guidelines Working Group. Gastrointestinal stromal tumours: ESMO Clinical recommendations for diagnosis, treatment and follow-up. Ann Oncol 2010;21(Suppl 5): v98-102
- Debiec-Rychter M, Sciot R, Le Cesne A, et al. KIT mutations and dose selection for imatinib in patients with advanced gastrointestinal stromal tumours. Eur J Cancer 2006;42:1093-103
- Verweij J, Casali PG, Zalcberg J, et al. Progression-free survival in gastrointestinal stromal tumours with high-dose imatinib: randomised trial. Lancet 2004;364:1127-34
- Zalcberg JR, Verweij J, Casali PG, et al. Outcome of patients with advanced gastro-intestinal stromal tumours crossing over to a daily imatinib dose of 800 mg after progression on 400 mg. Eur J Cancer 2005;41:1751-7
- Judson IR. Prognosis, imatinib dose, and benefit of sunitinib in GIST: knowing the genotype. J Clin Oncol 2008;26:5322-5
- Van Glabbeke M. Comparison of two doses of imatinib for the treatment of unresectable or metastatic gastrointestinal stromal tumors (GIST): a meta-analysis based on 1640 patients. The GIST Meta-Analysis group (MetaGIST). J Clin Oncol 2010;28:1247-53
- Demetri GD, von Mehren M, Blanke CD, et al. Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med 2002;347:472-80
- Hopkins TG, Marples M, Stark D. Sunitinib in the management of gastrointestinal stromal tumours (GISTs). Eur J Surg Oncol 2008;34:844-50
- Heinrich MC, Maki RG, Corless CL, et al. Primary and secondary kinase genotypes correlate with the biological and clinical activity of sunitinib in imatinib-resistant gastrointestinal stromal tumor. J Clin Oncol 2008;26:5352-9
- Lasota J, Miettinen M. Clinical significance of oncogenic KIT and PDGFRA mutations in gastrointestinal stromal tumours. Histopathology 2008;53:245-66
- Demetri GD, van Oosterom AT, Garrett CR, et al. Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumour after failure of imatinib: a randomised controlled trial. Lancet 2006;368:1329-38
- Demetri GD, Huang X, Garrett CR, et al. Novel statistical analysis of long-term survival to account for crossover in a phase III trial of sunitinib (SU) vs. placebo (PL) in advanced GIST after imatinib (IM) failure. J Clin Oncol (Meeting Abstracts) 2008;26:10524
- Demetri GD, Casali PG, Blay JY, et al. A phase I study of single-agent nilotinib or in combination with imatinib in patients with imatinib-resistant gastrointestinal stromal tumors. Clin Cancer Res 2009;15:5910-16
- Weisberg E, Manley PW, Breitenstein W, et al. Characterization of AMN107, a selective inhibitor of native and mutant Bcr-Abl. Cancer Cell 2005;7:129-41
- Wiebe L, Kasza KE, Maki RG, et al. Activity of sorafenib (SOR) in patients (pts) with imatinib (IM) and sunitinib (SU)-resistant (RES) gastrointestinal stromal tumors (GIST): a phase II trial of the University of Chicago Phase II Consortium. J Clin Oncol (Meeting Abstracts) 2008;26:10502
- Wilhelm S, Carter C, Lynch M, et al. Discovery and development of sorafenib: a multikinase inhibitor for treating cancer. Nat Rev Drug Discov 2006;5:835-44
- Adenis A, Cassier PA, Bui BN, et al. Does interruption of imatinib (IM) in responding patients after three years of treatment influence outcome of patients with advanced GIST included in the BFR14 trial? J Clin Oncol 2008;26(Suppl): abstract 10522
- Blay JY, Le Cesne A, Ray-Coquard I, et al. Prospective multicentric randomized phase III study of imatinib in patients with advanced gastrointestinal stromal tumors comparing interruption versus continuation of treatment beyond 1 year: the French Sarcoma Group. J Clin Oncol 2007;25:1107-13
- Rios M, LeCesne A, Bui B, et al. Interruption of imatinib (IM) in GIST patients with advanced disease after one year of treatment: Updated results of the prospective French Sarcoma Group randomized phase III trial on long term survival. J Clin Oncol (Meeting Abstracts) 2007;25:10016
- Huse DM, von Mehren M, Lenhart G, et al. Cost effectiveness of imatinib mesylate in the treatment of advanced gastrointestinal stromal tumours. Clin Drug Investig 2007;27:85-93
- Mabasa VH, Taylor SC, Chu CC, et al. Verification of imatinib cost-effectiveness in advanced gastrointestinal stromal tumor in British Columbia (VINCE-BC study). J Oncol Pharm Pract 2008;14:105-12
- Wilson J, Connock M, Song F, et al. Imatinib for the treatment of patients with unresectable and/or metastatic gastrointestinal stromal tumours: systematic review and economic evaluation. Health Technol Assess 2005;9:1-142
- Chabot I, LeLorier J, Blackstein ME. The challenge of conducting pharmacoeconomic evaluations in oncology using crossover trials: the example of sunitinib for gastrointestinal stromal tumour. Eur J Cancer 2008;44:972-7
- Paz-Ares L, Garcia dMX, Grande E, et al. Cost-effectiveness analysis of sunitinib in patients with metastatic and/or unresectable gastrointestinal stroma tumours (GIST) after progression or intolerance with imatinib. Clin Transl Oncol 2008;10:831-9
- Contreras-Hernandez I, Mould-Quevedo JF, Silva A, et al. A pharmaco-economic analysis of second-line treatment with imatinib or sunitinib in patients with advanced gastrointestinal stromal tumours. Br J Cancer 2008;98:1762-8
- Benjamin RS, Rankin C, Fletcher CD. Phase III dose-randomized study of imatinib mesylate (STI571) for GIST: Interrgroup S0033 early results. Proc Am Soc Clin Oncol 2003;22:(abstract 3271)
- Verweij J, Casali PG, Zalcberg J. Early efficacy comparison of two doses of imatinib for the treatment of advanced gastro-intestinal stromal tumors (GIST): interim results of a randomized Phase III trial from the EORTCSTBSG, ISG and AGITG. Proc Am Soc Clin Oncol 2003;22:(abstract 3272)
- Motzer RJ, Hoosen S, Bello CL, et al. Sunitinib malate for the treatment of solid tumours: a review of current clinical data. Expert Opin Investig Drugs 2006;15:553-61
- Earle CC, Chapman RH, Baker CS, et al. Systematic overview of cost-utility assessments in oncology. J Clin Oncol 2000;18:3302-17
- El Ouagari K, Huse DM. Cost-effectiveness of imatinib in the treatment of advanced gastrointestinal stromal tumors (GIST): Canadian perspective. 33rd ESMO Congress, Stockholm, Sweden, September 12-16, 2008. Abstract 722P
- Ades AE, Sculpher M, Sutton A, et al. Bayesian methods for evidence synthesis in cost-effectiveness analysis. Pharmacoeconomics 2006;24:1-19
- Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med 2004;23:3105-24