Objective To evaluate the cost effectiveness of trabectedin plus pegylated liposomal doxorubicin (PLD) versus PLD monotherapy in patients with platinum-sensitive relapsed ovarian cancer from the perspective of the Spanish National Health Service.
Methods A decision analytical model was developed to estimate the total treatment-related costs and clinical benefits of trabectedin plus PLD and PLD alone. Patient data and utilities were obtained from the OVA-301 study. Adverse event management costs were obtained from the literature. Other unitary costs (euros, 2011) were obtained from a Spanish healthcare cost database and the Catalogue of Medicines. Costs and benefits (quality adjusted life years (QALYs)) were discounted at 3%. Sensitivity analyses were performed.
Results Trabectedin plus PLD yielded greater health benefits (2.35 QALYs) than PLD alone (1.86 QALYs). The global costs (treatment, adverse events management, and medical management) per patient were €45 573 and €23 072 for trabectedin plus PLD and PLD alone, respectively. The incremental cost-effectiveness ratio of trabectedin plus PLD versus PLD was €45 592/QALY gained.
Conclusions Compared with PLD alone, the trabectedin plus PLD combination is an effective therapy that slightly exceeds the common threshold (€45 000/QALY gained).
Statistics from Altmetric.com
Ovarian cancer is one of the most common gynaecologic malignancies and a frequent cause of cancer deaths among women. The mortality rate in Europe is 12 per 100 000 women/year and the crude incidence is 18 per 100 000 women/year.1 In Spain, where the mortality rate is 4 per 100 000 women,2 ovarian cancer was responsible for 1760 deaths in 2006.3
Approximately 50% of cases occur in older women (>65 years). Early cancer has no clear symptoms, and the lack of effective screening leads to the majority of patients being diagnosed at an advanced stage.4
The management of ovarian carcinoma depends on the disease's extent and the prior therapy received. The standard of care for advanced disease consists of surgical cytoreduction followed by chemotherapy with paclitaxel and a platinum compound.1 ,5
Most women will develop recurrent disease following initial treatment. The optimal treatment for relapsed ovarian cancer is evolving, and despite treatment advances, improvements in survival rates have been modest: the median length of survival after recurrence is 2 years,6 and approximately 50% of patients die.4
The time that elapses between the last dose of platinum-based chemotherapy and disease recurrence determines the subsequent treatment. Patients with ovarian cancer who experience a late relapse after first-line platinum-based chemotherapy (ie, platinum-free interval (PFI) >6 months) are considered to be platinum sensitive, and usually re-treated with the first-line platinum therapy.
A longer PFI increases the chances of benefiting from a platinum rechallenge, especially for patients with a PFI >12 months.7 8 Patients whose disease recurs after 6–12 months (partially platinum sensitive (PPS)) show a more modest response to further platinum treatment. However, there are a significant number of patients with platinum-sensitive disease for whom retreatment with platinum-based agents is not recommended.9
Trabectedin (Yondelis, Pharma Mar, Spain), a cytotoxic agent originally isolated from the tunicate Ecteinascidia turbinate, is currently produced synthetically. The pivotal study OVA-301 was conducted in 672 patients with relapsed ovarian cancer that progressed after one platinum-based regimen.10 Pegylated liposomal doxorubicin hydrochloride (PLD) is a standard therapy used for relapsed ovarian cancer that has been shown to have a superior efficacy over other substances, including topotecan11 and paclitaxel.12
OVA-301, a randomised, controlled, multicentre, open-label, phase-III clinical trial, compared the efficacy and safety of trabectedin combined with PLD against PLD alone.10 The study was stratified according to platinum sensitivity (platinum-resistant disease (PFI <6 months) versus platinum-sensitive disease (PFI ≥6 months)) and Eastern Cooperative Oncology Group performance status (0–1 vs 2). The primary endpoint was progression free survival (PFS) and the secondary endpoints included overall response rate, duration of response, safety and overall survival (OS).
Based on the results from OVA-301, the combination of trabectedin plus PLD was approved in the EU in 2009 for the treatment of patients with relapsed, platinum-sensitive ovarian cancer.13
The final protocol-specified OS analysis was conducted when 522 deaths were observed (median follow-up period 47.4 months). The final survival (median OS) was 22.2 months for the trabectedin plus PLD combination and 18.9 months for the PLD arm (HR 0.86; 95% CI 0.72 to 1.02; p=0.0835; unstratified log-rank test). Among the 430 patients who were platinum sensitive, the median OS according to the Kaplan–Meier method was 27.0 months in the trabectedin plus PLD arm and 24.1 months in the PLD arm (HR 0.83; 95% CI 0.67 to 1.04; p=0.1056).14
Despite study stratification, an unanticipated but significant overall imbalance in the PFI between the two arms favouring the PLD arm was observed.14 To provide an appropriate and reliable estimate of the treatment effect, a multivariate analysis was performed. When prognostic factors (including PFI) were included in the Cox regression model, the trabectedin plus PLD treatment for the platinum-sensitive population resulted in a 22% decrease in the risk of death compared with PLD alone (HR 0.78; p=0.0319).
In patients with PPS disease, the PFI imbalance was not statistically significant. The results of the log-rank Kaplan–Meier test (HR 0.64; 22.4 months vs 16.4 months; p=0.0027) found a 36% decrease in the risk of death.
The objective of this study was to estimate the cost-effectiveness ratio of trabectedin plus PLD compared with PLD alone for the treatment of platinum-sensitive relapsed ovarian cancer from the point of view of the Spanish National Health System.
Materials and methods
A decision model was developed with Microsoft Excel to simulate a patient's clinical course and to assess the cost effectiveness of trabectedin plus PLD (intravenously administered) compared with intravenous PLD monotherapy.
The underlying assumptions, missing information and resource use parameters were derived in consultation with a clinical advisory group comprising oncology and pharmacy specialists, and three healthcare economists. Because the model was based on the perspective of the Spanish National Health System, only direct healthcare resources and costs were included. The measures of benefits used were life years gained (LYG) and quality-adjusted life years (QALYs) gained.
The structure of the model is shown in figure 1. This structure reflects the major clinical and resource-generating events that patients may experience over the course of their remaining life, segregated into three potential health-defining conditions of the patient: stable disease, progressive disease and death.
The choice of model was based on the approach used by the National Institute for Health and Clinical Excellence (NICE) for the multiple technology appraisal of topotecan, PLD and paclitaxel for relapsed ovarian cancer.15
Patients began in the stable state and transitioned to the progressive state after the mean PFS period; subsequently, patients transitioned to death after the mean (OS–PFS) period. Therefore, OS was calculated in relation to two periods: the progression-free period (time in the stable state or PFS) and the time from progression to death (time in the progressive state was estimated as the difference between OS and PFS).
The simulation also accounted for the occurrence of adverse events (AEs).
Time horizon and discount rate
In the base case, the model evaluated the cost effectiveness over the patient lifetime, meaning that every patient remained in the model until death. The lifetime horizon was chosen because patients remain on treatment and accumulate costs and QALYs until death, allowing the differences in costs and outcomes between treatments to be captured. Costs and health benefits were both discounted at 3.0% annually, as recommended by the latest Spanish recommendations.16
In the base-case deterministic analysis, the expected total costs and QALYs were calculated for the treatment arms. With this information, the incremental cost-effectiveness ratio (ICER) of trabectedin plus PLD and PLD alone was calculated.
Clinical effectiveness data
The clinical effectiveness evidence was based on the OVA-301 study.10 14 The methodology used to calculate the mean PFS and OS used final survival data14 and was based on the methods applied by the NICE Evidence Review Group (ERG) in its evaluation of trabectedin plus PLD.17 Only data for patients with platinum-sensitive disease were included in the analysis. As proposed by the ERG, patients with no measurable disease at baseline (n=12) were excluded from the PFS analysis.
Because of the lifetime approach used, the survival results from OVA-301 must be extrapolated over time for the global simulation, and survival distributions used. Five parametric survival distributions (exponential, Weibull, Gompertz, log-logistic and lognormal) were explored for the PFS and OS data separately to identify the best distribution fitting. Goodness-of-fit criteria based on the Akaike information criterion and Bayesian information criterion of each distribution were used to select the preferred distribution for PFS and OS by treatment arm.
A Weibull distribution was chosen for the PFS calculations for both treatment arms, and a log-logistic distribution was applied for OS. The mean PFS was 11.26 and 8.25 months for trabectedin plus PLD and PLD alone, respectively. The mean OS was 44.69 and 34.97 months for trabectedin plus PLD and PLD alone, respectively.
Based on the OVA-301 trial,10 the incidence of anaemia, diarrhoea, nausea/vomiting, febrile neutropenia, neutropenic infection, neutropenic sepsis, neutropenia, palmar-plantar erythrodysaesthesia syndrome, thrombocytopenia and stomatitis were incorporated into the model (table 1).
The type of chemotherapy administered to patients with advanced ovarian cancer impacts survival, progression and quality of life. Therefore, health-related quality of life data were considered as a cost–utility analysis incorporating utility values.
The utility of the different treatment arms is affected by the AE profile and the rates of response and progression. Based on the EQ-5D data collected in the OVA-301 study, the mean utility whilst in the stable and progressive states was estimated to be 0.72 and 0.65, respectively. Utility was assumed to be constant over time, as repeated measurements yielded no systematic changes over time. The utility for each disease state was multiplied by the mean time spent in each state to obtain QALYs.
Resources and costs
The model is run from a third-party payer perspective, including the following direct healthcare resources:
drug costs, including pharmaceutical and administration costs
medical management costs
AE management costs.
The median body surface area for patients with platinum-sensitive ovarian cancer recruited in European centres and included in OVA-301 (1.7 m2 per patient) was used to calculate the average treatment cost per cycle for both regimens. The mean cumulative doses by cycle for the European patients with platinum-sensitive ovarian cancer were also considered. Patients in the trabectedin plus PLD arm were assumed to receive 1.04 mg/m2 of trabectedin and 28.43 mg/m2 of PLD. Patients in the PLD monotherapy arm received doses equivalent to 44.40 mg/m2.
Drug costs were calculated based on the total milligrams administered, assuming no drug wastage. There are two available packages for trabectedin (1 mg and 250 µg), and the average cost per mg was used. Administration costs were added to pharmaceutical costs to obtain the total cost per treatment cycle. Administration was assumed to be equivalent to a hospital day visit for each infusion. For trabectedin, the required use of a central catheter was also considered. The total treatment cost per cycle is the sum of the pharmaceutical and the administration cost. The total drug cost per treatment arm represents the cost of six cycles.
Medical management costs depended on whether a patient was in a stable or progressive disease state. During the progression-free period, it was assumed that patients attended one outpatient visit to see the specialist and underwent one CT scan every 2 months.
During the time from progression to death, it was assumed that patients received palliative care; the cost of palliative care in patients with advanced cancer was collected from the medical literature.18 The costs of the treatment of related AEs, table 1, were incorporated into the model. AEs were considered for cost purposes within the health economic model if they met one or more of the following criteria: grade 3 or 4; frequency ≥10% in either of the arms; a significant associated cost. The costs for AE management in Spain were obtained from the literature.19 ,20
Unitary resource costs (table 2) were obtained from a healthcare costs database,21 and drug costs (ex-factory prices) were taken from the General Council of Official Pharmaceutical Colleges.22 All costs are presented in euros (2011).
To test the uncertainty and the robustness of the results of the base case, deterministic one-way and probabilistic analyses were performed. To identify the key model parameters, the following parameters were varied in the one-way analyses:
Drug costs; pharmaceutical costs were varied by means of three different analyses:
number of cycles:
a maximum of eight cycles and a minimum of four cycles
the mean number of cycles in the OVA-301 trial was applied (6.9 cycles for trabectedin plus PLD and 5.7 cycles for PLD alone)
the impact of considering the median doses (1.10 mg/m2 for trabectedin plus 29.94 mg/m2 for PLD and 49.30 mg/m2 for PLD in monotherapy) from OVA-30110 rather than the mean doses was explored
the full number of vials per drug: although the maximum use of any drug should be pursued, especially in a global cost restriction context, an analysis based on the number of vials of medicine required to administer the required dose of each treatment was conducted.
Utility values were varied ±10%.
Costs of AE management were varied ±50%.
Probabilistic sensitivity analysis (PSA) was conducted through a Monte Carlo simulation with 1000 runs.23 Patient proportions and utilities were assumed to follow a β distribution. HRs for PFS and OS were represented using log-normal distributions,24 and healthcare resource utilisation parameters and costs were assumed to follow γ distributions. PSAs were plotted on a cost-effectiveness plane23 based on a tentative threshold in Spain of €45 000/QALY gained.25
The cost-effectiveness plane reflects the relationship between incremental cost and incremental health benefit and includes a line representing the acceptable threshold. Results below this line indicate that the strategies are cost effective.
For a lifetime horizon, the trabectedin plus PLD combination yields more health benefits than PLD alone: 2.80 vs 2.13, respectively. When quality of life is included, multiplying LYG by utilities, the gain in QALYs was 2.35 versus 1.86 (trabectedin plus PLD vs PLD).
The total cost was €45 573 for trabectedin plus PLD and €23 072 for PLD alone.
In the base case, the ICER of trabectedin plus PLD compared with PLD alone resulted in €33 335/LYG and €45 592/QALY gained (table 3).
The deterministic results are represented as a tornado diagram (figure 2).
Drug cost, reflected in cost per vial and number of cycles administered, was a key parameter for ICERs. Differences compared with the base case were 12.7% when cost per vial instead of cost per mg was considered, 27.4% with eight drug cycles instead of six cycles and −27.4% with four cycles of treatment. A lower ICER (€41 447/QALY gained) than that for the base case was obtained when utilities were decreased by 10%, showing that quality of life is also an important factor to be taken into account. However, variation up to ±50% of the AE management costs was not associated with the main differences versus the base case (±0.5%).
The PSA results are represented as a cost-effectiveness plane (figure 3).
The OVA-301 trial demonstrated that the combination of trabectedin plus PLD provides a statistically significant and clinically relevant patient benefit. In the primary analysis, the combination was associated with improved OS in all patients with platinum-sensitive ovarian cancer, particularly in those with PPS disease.
An accurate assessment of the costs and the impact of the available therapies on the disease, in terms of clinical effectiveness and health-related quality of life of patients, is necessary to obtain an optimal, cost-effective treatment strategy. Therefore, this assessment can help decision-makers prioritise resource allocation and maximise benefit in cancer management in times of limited resources.
The current assessment is the first study to evaluate the long-term cost effectiveness of trabectedin plus PLD compared with PLD alone for the treatment of relapsed platinum-sensitive ovarian cancer from the perspective of the Spanish National Health Service. The results of our analysis show that, compared with PLD monotherapy, trabectedin plus PLD is associated with €45 592/QALY gained. The PSA indicated that these results are quite stable (figure 3).
According to the economic theories related to the cost-effectiveness relationship, the threshold should reflect the opportunity costs, although this rarely occurs in practice.
The first threshold may have been developed for the US government funding of renal dialysis, which was introduced in 1973 and was roughly $50 000 (£30 000) per patient for each LYG. A similar justification established the basis for the £30 000/QALY gained threshold used by NICE. Controversy persists regarding the criteria that should be assessed in the process. NICE suggests that the evaluation of drugs that have demonstrated benefits for survival in a reduced terminal patient target (survival <24 months) should be flexible in terms of the common threshold, and alternatives above £30 000/QALY gained could be funded by the public system.
The recent creation of the Cancer Drugs Fund intends to improve patient access to drugs pending NICE evaluation and drugs that have been refused based on cost-effectiveness criteria.26 Similar strategies with more flexible criteria for reimbursement for any innovative options, mainly oncology, have been proposed in other settings.27–29
Although most publications use a reference value between €30 000 and €45 000/LYG,25 there is no public consensus in Spain about willingness to pay for each LYG or QALY gained. In the present work, the ICER for trabectedin plus PLD compared with PLD alone is €45 592/QALY gained, which slightly exceeds the values cited in the literature.
Economic evaluations are one of the many criteria used to inform decisions made by healthcare payers. Threshold ICERs are only applied as reference criteria in some countries and are likely to be replaced over time by the value-based pricing model. The extended absence of innovative and effective therapeutic options for patients with cancer must drive decisions about whether the process of adding new oncology drugs to the public health system should incorporate arguments similar to the one previously described. There are some limitations that need to be considered when interpreting the results. Prospective data over a lifetime of treatment are preferred for making decisions regarding the adoption of new therapies. However, when such data are unavailable, a model analysis based on clinical trial data is necessary, despite its limitations.
The efficacy data for trabectedin plus PLD in this study were based on the OVA-301 trial.10
A model based on a cohort population was designed for this evaluation. As cohort models cannot account for individual characteristics, the typical patient in the model is based on the overall characteristics of the patient cohort that reflects the OVA-301 population.
Another limitation of the model is that the AE management costs were not calculated from any prospective study; however, they were obtained from the literature and accepted by an expert panel. A naturalistic design with a prospective economic appraisal to determine resource utilisation would be the ideal research design for the assessment of medical resource utilisation.
The present model was developed from a third-party payer perspective; thus it did not include indirect costs that could be useful for a societal analysis.
Despite these limitations, the assumptions in the model are quite reasonable (or even conservative), and the results of the sensitivity analyses indicate that the findings are robust.
Unlike other evaluations based on clinical studies that use values from the literature, the utilities included in the model were estimated from EQ-5D information directly collected during the trial. The utility estimates obtained directly from OVA-301 differ from those used in other economic evaluations related to ovarian cancer.30 Due to the absence of direct estimates for the progressive disease utility of ovarian cancer, Main et al assumed an estimate based on a breast cancer value. Although assumptions are usually the only way to capture information, the use of direct estimations of utilities, such as those applied in the present work, should be preferred if available. A comparison of cost-effectiveness ratios between studies is difficult because of differences in healthcare system organisations, year cost values and other factors.
In conclusion, for the Spanish National Health System, trabectedin plus PLD for relapsed platinum-sensitive ovarian cancer is an effective option compared with PLD alone, with an ICER of €45 592/QALY gained.
Trabectedin in combination with PLD for the treatment of women with relapsed platinum-sensitive ovarian cancer is an effective option from the Spanish National Health System point of view compared with PLD alone, with an ICER of €45 592/QALY gained.
Contributors JM, MJR, MJC and GC validated local inputs, provided local information, read and approved the final version of the manuscript. IO and MAC performed the local adaptation of the economic evaluation and drafted the manuscript. BGS validated the assumption, read and approved the final version of the manuscript.
Conflict of interest This work was conducted with an unrestricted grant from PharmaMar. Expert opinions and model adaptation were done independently.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.