A decision-analytical model shows that everolimus has a favorable budget impact when introduced into the treatment algorithm for metastatic renal cell carcinoma.
Renal cell carcinoma (RCC) is diagnosed in approximately 58,000 individuals each year in the United States1 and accounts for more than 90% of all primary malignant renal tumors.1,2 These statistics refl ect the steady rise in the incidence rate of kidney cancers over the past 3 decades.3 It has been estimated that 20% to 30% of patients with RCC present with metastatic disease.4 Metastatic renal cell carcinoma (mRCC) is diffi cult to treat, with a 5-year survival rate of less than 10%.5 Additionally, RCC is costly, with US-based studies suggesting the estimated annual cost of RCC to be as high as $4.4 billion. In those with metastatic disease, the estimate ranges from $107 to $556 million per year.6,7
Several targeted chemotherapeutic agents have emerged over the past 4 years. Specifically, vascular endothelial growth factor tyrosine kinase inhibitors such as sunitinib and sorafenib have become the standard of care, replacing systemic therapies such as interferon-alfa. Despite the recent advances in treatment options, disease progression in RCC patients is typically only delayed; the disease is not cured. Trials investigating the effi cacy of sunitinib and sorafenib demonstrated an increase in progression-free survival (PFS) ranging from 2.7 months to 6 months.8,9 Bevacizumab, another drug option, has shown a benefit in PFS in a pivotal phase III trial in combination with interferon-alfa versus interferon-alfa alone; however, overall survival results have not yet been reported.10 While there has been some improvement over prior therapies, the majority of patients will face inevitable disease progression. Therefore, there is a significant unmet need among RCC patients who experience treatment failure on these newer agents. Everolimus (Afinitor; Novartis Pharmaceuticals Corporation, East Hanover, New Jersey) is a once-daily, orally administered agent with documented effi cacy for the treatment of patients with advanced RCC after they fail to respond to or become intolerant to treatment with sunitinib or sorafenib.
Everolimus is an inhibitor of the mammalian target of rapamycin (mTOR), a component of an intracellular signaling pathway that regulates cellular metabolism, growth, proliferation, and angiogenesis.11 A pivotal, randomized, phase III trial of patients with mRCC whose disease had progressed on sunitinib, sorafenib, or both demonstrated a median PFS of 4.9 months for everolimus-treated patients compared with 1.9 months for those receiving best supportive care alone (hazard ratio 0.33, 95% confi dence interval 0.25-0.43; P <.001).
Although novel, targeted oncology treatments have resulted in incremental improvements in survival, they may be viewed as a potential source of substantial economic burden on the healthcare system, requiring payers and physicians to weigh costs and benefits when determining a treatment strategy.13 In addition, payers can expect to incur relatively high costs during the last few months of life, during which time patients are receiving numerous expensive salvage therapies that ultimately may not affect their overall survival. Therefore, health plans will likely benefit from an examination of the incremental budget impact of a given treatment regimen upon its approval. The purpose of this study is to determine the impact that the introduction of everolimus may have on a health plan’s budget by using a health economic model that considers the various treatments currently available to patients with mRCC within the bounds of explicit model assumptions.
STUDY DESIGN AND METHODS
A budget impact analysis was developed using Microsoft Excel (Microsoft Corporation, Redmond, Washington) to evaluate the current cost of therapy for RCC from April 2008 to March 2009 and October 2009 to September 2010 for a hypothetical health plan population of 1 million members. These time points were selected to account for markets before and after the approval of everolimus in mRCC (March 2009). Because there may be a delay associated with the uptake of everolimus immediately following approval, a 6-month period (April 2009 to September 2009) was considered as a lag period and was not included for base case post-everolimus launch market (see
, available at www.ajpblive.com).
Population and Treatments Under Consideration
Medication utilization data and prevalence rate for mRCC were applied to a hypothetical health plan of 1 million insured adults. The analysis considered only pharmacotherapeutic treatments for mRCC (ie, systemic chemotherapy and targeted treatments). The number of eligible patients was calculated based on mRCC epidemiology estimates (Table 1). Market share data for the different treatments were based on recent surveillance studies for RCC and real-world market share data (Novartis Pharmaceuticals Corporation, research market share data on file, proprietary and unpublished data, 2010). The Novartis data were obtained from patient-level drug utilization information, which was collected from 36 states electronically on a monthly basis. Treatment duration as well as the proportion of patients who moved from fi rst-line to subsequent lines was based on a retrospective analysis of claims data performed by Intercontinental Marketing Services (IMS) Health, which has been used previously for other studies.14 Information on treatment durations was available for both pre—and post–everolimus launch periods from IMS Health analysis. However, in order to keep the durations of each treatment consistent across both pre-and post-everolimus periods, treatment durations recorded during the pre-everolimus market were applied to the post-everolimus market. For drugs for which treatment duration data were not available in the pre-everolimus period (eg, everolimus, pazopanib), the treatment duration recorded in the post-everolimus period was applied.
To establish the total cost of treatment with a chemotherapeutic agent, the model considered the wholesale acquisition cost of each treatment, the recommended dose, administration costs, and the duration of treatment. Treatment schedule and dosing assumptions were based on the prescribing information for each product.11,15-20 Treatment administration costs were added to every infused treatment, and were obtained from Current Procedural Terminology (CPT) codes.21 The total administration cost for each regimen was estimated by multiplying unit 12 administrative costs by the mean treatment duration. Details on the treatment schedules, unit costs, costs per dose, cost per week, and administration costs for the treatments considered are provided in Table 2.
The model accounted for the differing safety profi les associated with each treatment. Treatment-related grade 3/4 adverse event incidence estimates were adopted from prescribing information or pivotal trials for each agent.11,15-19,22 The costs of treating the adverse events were based on independent cost analyses.23-26 Adverse events for which no cost literature was available had treatment costs estimated based on National Comprehensive Cancer Network guidelines10,27-29 and unit costs from the Red Book 2010 CPT code book.30,31 Details of specific calculations are available upon request. Costs calculated to treat adverse events were assumed for the entire course of chemotherapeutic treatment within a particular line of treatment. Costs for managing adverse events are shown in
, available at www.ajpblive.com. These costs along with incidence rates from prescribing information and pivotal trials were used to calculate the total cost of treating adverse events for treatments considered in the model (Table 2).11,15-19,22 As underlying best supportive care was assumed to be the same across comparators, the cost of best supportive care was not included in this analysis.
Total cost for each medication across the first, second, and third line of treatment was estimated by applying the drug acquisition, administration, and adverse event costs to the hypothetical mRCC population for the preeverolimus and post-everolimus scenarios. Across both the pre- and post-everolimus scenarios, mRCC incidence, treatment comparators, and costs (drug, administration, and adverse event management) were all assumed to remain the same; only the relative market share for each comparator varied. The eligible relative market share of patients treated with everolimus as second- or thirdline therapy in the postlaunch scenario was 11.63% and 27.73%, respectively.
The analysis focused on evaluating budgetary differences between the pre—and post–everolimus launch markets at 3 levels: (1) the aggregate level, (2) the aggregate budget impact stratified by treatment comparators, and (3) the budget impact per member per year (PMPY) and per member per month (PMPM).
Sensitivity analyses were performed, where (1) thepost-everolimus launch period was changed from October 2009—September 2010 to April 2009–March 2010 (which included the initial postapproval uptake lag period) and (2) the post–everolimus launch period was changed to January 2010–December 2010 (which assumed a 9-month everolimus uptake lag period instead of 6 months) (see eAppendix A). Additional sensitivity analyses were performed. They assumed that (3) administration costs were zero, (4) adverse event costs were zero, and (5) treatment duration for all comparators was the same as that for everolimus within each line of treatment. Market share of everolimus for second- and third-line treatment for the April 2009–March 2010 postlaunch period (sensitivity analysis 1) was 7.77% and 25.12%, respectively. For the January 2010–December 2010 postlaunch period (sensitivity analysis 2), market share was 12.80% and 27.82%, respectively. Sensitivity analyses 3, 4, and 5 were performed primarily to remove the effect of administration costs, adverse event costs, and treatment durations from the aggregate drug budgetary impact.
Supplementary scenario analyses were conducted where market share of everolimus in the postlaunch period was exclusively captured from the market share of sorafenib (scenario 1) and sunitinib (scenario 2). Another scenario analysis, where everolimus market share was captured from temsirolimus (64.0%), bevacizumab (26.4%), and sorafenib (9.6%), was also performed. These analyses were conducted for both second and third lines of treatment. Market share data for the base case and all the scenarios are shown in
. Costs are reported in 2010 US dollars.
With an incidence rate of 0.0203% and a hypothetical plan of 1 million covered individuals, a population of 203 mRCC patients was calculated, of whom 90% (182 patients) received first-line active treatment for mRCC. Longitudinal analysis of IMS Health data for patients in 2008 showed that 28.2% (52 patients) of patients who received first-line therapy moved to second-line therapy; 24.1% (12 patients) of these patients additionally received a third-line therapy. Similarly for 2009, 21.8% (40 patients) of patients receiving first-line therapy also received second- line therapy, while only 13.5% (5 patients) of these patients received third-line therapy (Table 1).
In the pre—everolimus launch market scenario, the total cost of drugs, administration, and adverse event management, across all lines of treatment, was found to be $7,050,157. In the market following the introduction of everolimus, the total cost decreased to $6,741,642, yielding a savings of $308,516.
reports the cost stratified by different therapies within each line of treatment. Based on the model calculations, with the availability of everolimus after October 2009, the budget impact PMPM cost was −$0.03 and the budget impact PMPY cost was—$0.31.
Results for the sensitivity and additional scenario analyses are presented in
. Sensitivity analyses that accounted for the extended postapproval lag period (January 2010 to December 2010) and the assumption that all comparators had the same treatment duration as that of everolimus (analyses 2 and 5, respectively) both resulted in additional savings of more than $70,000 (increased PMPY of >$0.07). However, the sensitivity analyses that examined the inclusion of the postapproval uptake lag period (April 2009 to March 2010) and set the adverse event management costs to zero dollars (analyses 1 and 4, respectively) resulted in less savings than the base case analysis. No substantial differences were found across all additional scenario analyses (scenarios 1, 2, and 3), where total savings ranged from $233,237 to $254,274. Both PMPY and PMPM costs for all the 3 additional scenarios appeared to be similar to those of the base case analysis.
This model estimated the annual budget impact of treating patients with mRCC in order to show the range of possible budget impacts in a market without everolimus compared with a market where everolimus was used in different amounts by different types of patients. The model results demonstrate the potentially favorable impact that the introduction of everolimus could have on a health plan’s budget.
Additional analyses were conducted to assess scenarios whereby the introduction of everolimus as a next-line therapy to sunitinib (scenario 1), sorafenib (scenario 2), or temsirolimus, bevacizumab, and sorafenib (scenario 3) resulted in a favorable aggregate budget impact compared with a treatment algorithm that excluded everolimus altogether. At the PMPM level, including everolimus in the treatment algorithm is cost neutral. The difference in total cost is mostly attributed to the difference in the cost of anticancer drugs.
There are several limitations to this analysis. This model did not take into account or correct for differences in populations from study to study. Consequently, results might have been different if data had been extracted from a unified study. For instance, because adverse event rates were taken from a series of different studies, variation in the adverse event rates assigned to each drug might have been more directly attributable to differences in patient populations. However, a sensitivity analysis was undertaken to assess the budget impact of everolimus when adverse event costs were excluded. The findings suggest an incremental PMPM budget impact of −$0.02, which is in line with the base case results.
Also, it was assumed that treatment durations recorded for comparator agents in the pre-everolimus market would be the same in the post-everolimus market. This assumption was made in order to prevent potential differences in treatment duration from biasing the model results. However, it should be acknowledged that this assumption could have led to either an overestimation or an underestimation of overall costs. To address this issue, a sensitivity analysis was performed, whereby all agents were assigned the duration of treatment recorded for everolimus. The results of this sensitivity analysis show that even with this adjustment, the budget impact results are similar to the base case —$0.04 incremental PMPM for the sensitivity analysis versus –$0.03 for the base case. Additionally, costs for best supportive care and palliative care were not taken into account in this model; therefore, the change in proportion of overall costs is likely overestimated in each scenario.
Of note, the market share data were collected on a monthly basis only, and patients were not followed longitudinally. Therefore, the analysis was performed using patient-months to account for patient entry/dropout during the course of the data collection. To take into account the lag time involved in drug uptake immediately following approval and to avoid underestimating everolimus use in the current scenario, market share data 6 months following approval of everolimus were used for the base case analysis. However, sensitivity analyses that included the 6-month lag period and an extended lag period (9 months) resulted in findings comparable to those in the base case. Moreover, with increased market share of everolimus (relative to the base case) in the extended lag period analysis, the savings incurred were more than those observed in the base case.
The analysis did not consider surgical options for 2 main reasons. Because the target population of this analysis was patients with mRCC, only a small proportion would have qualified for surgery.27 Additionally, surgical procedures do not directly compete with everolimus therapy because the proportion of patients receiving surgery would remain the same before and after the introduction of everolimus to this disease landscape. If surgery were to be included as a comparator, the results would remain the same.
Despite the potential generalization of some of the inputs given the only available data, it is important to consider the relatively small overall impact that mRCC has on managed care budgets compared with other types of cancer or other chronic diseases. Even if the results of this study were underestimated or overestimated, the analysis shows that providing patients with additional treatment options such as everolimus would not result in a substantial increase in a health plan’s overall budget.
The future management of oncology therapy in mRCC may require quantifi cation of value through evidencebased medicine, justifying the margin of benefit with each cycle. It is an opportunity to develop clinical pathways based on the efficacy, safety, and cost of second- or third-line treatment options. With new treatment options available in the future, it will become important to understand who can benefit with meaningful improvement in overall survival and to identify through biomarkers the sequence of therapies needed to maximize value.