Lifetime cost of everolimus vs axitinib in patients with advanced renal cell carcinoma who failed prior sunitinib therapy in the US

Abstract

Objective:

Everolimus and axitinib are approved in the US to treat patients with advanced renal cell carcinoma (RCC) after failure on sunitinib or sorafenib, and one prior systemic therapy (e.g., sunitinib), respectively. Two indirect comparisons performed to evaluate progression-free survival in patients treated with everolimus vs axitinib suggested similar efficacy between the two treatments. Therefore, this analysis compares the lifetime costs of these two therapies among sunitinib-refractory advanced RCC patients from a US payer perspective.

Research design and methods:

A Markov model was developed to simulate a cohort of sunitinib-refractory advanced RCC patients and estimate the cost of treating patients with everolimus vs axitinib. The following health states were included: stable disease without adverse events (AEs), stable disease with AEs, disease progression (PD), and death. The model included the following resources: active treatments, post-progression treatments, adverse events, physician and nurse visits, scans and tests, and palliative care. Resource utilization inputs were derived from a US claims database analysis. Additionally, a 3% annual discount rate was applied to costs, and the robustness of the model results was tested by conducting sensitivity analyses, including those on dosing scheme and post-progression treatment costs.

Results:

Base case results demonstrated that patients treated with everolimus cost an average of $12,985 (11%) less over their lifetimes than patients treated with axitinib. The primary difference in costs was related to active treatment, which was largely driven by axitinib’s higher dose intensity. Results remained consistent across sensitivity analyses for AE and PD treatment costs, as well as dose intensity and discount rates.

Conclusion:

The results suggest that everolimus likely leads to lower lifetime costs than axitinib for sunitinib-refractory advanced RCC patients in the US.

Keywords::

• Renal cell carcinoma
• Everolimus
• Axitinib
• Costs analysis
• Economic models

Introduction

Renal cell carcinoma (RCC) arises from the renal parenchyma and occurs in about nine out of 10 individuals presenting with renal cancer. The worldwide incidence of RCC has been rising at the rate of 1.5–5.9% each year1^,2. The American Cancer Society3 estimates that there were ∼65,150 new cases and 13,680 deaths due to RCC in 2013 in the US. RCC also imposes a substantial economic burden: the average total cost, including direct costs and indirect costs due to lost work productivity, was estimated to be $16,488–$43,805 per patient in 20094. The aggregate, national annual cost of RCC was ∼$4.4 billion in the US in 20055.

Among patients with RCC, 30–42% present with metastasis at the time of diagnosis1^,6. Metastatic renal cell carcinoma (mRCC) is difficult to treat and is among the top ten causes of cancer deaths in the US7. Although survival in mRCC patients has improved in recent years, the clinical challenge remains to identify the optimal sequence of therapies while minimizing treatment toxicities. Currently, patients with advanced RCC have a variety of treatment options approved by the Food and Drug Administration (FDA), including everolimus, interleukin-2, bevacizumab, axitinib, sorafenib, pazopanib, sunitinib, and temsirolimus8. Physicians and patients face a dynamic treatment landscape and uncertain treatment pathways. In the first-line setting, sunitinib, temsirolimus, bevacizumab, pazopanib, high dose interleukin-2, and sorafenib have received National Comprehensive Cancer Network (NCCN) category 1 recommendation for advanced RCC9. A retrospective observational study using US claims data suggests that sunitinib is currently the most used treatment for advanced and/or metastatic RCC in the first-line setting10. Despite the progression-free survival (PFS) improvement from sunitinib therapy11, almost all patients with advanced RCC still ultimately progress within a year and require second-line treatment options. Hence, it is important to address a treatment strategy for patients who experience disease progression while on sunitinib or other first-line therapies.

According to NCCN guidelines, everolimus and axitinib are the only two therapies that have category 1 evidence for second-line treatment of advanced RCC following tyrosine-kinase inhibitor (TKI) therapy9. Everolimus was approved by the FDA in 2009 as a second-line treatment for advanced RCC after failure to respond to vascular endothelial growth factor receptor (VEGFR) TKIs sunitinib or sorafenib12. Everolimus controls disease progression through inhibition of the mammalian target of rapamycin (mTOR) pathway13, which acts as a component of an intracellular signaling pathway that regulates cell growth, proliferation, and angiogenesis. In the RECORD-1 phase III clinical trial, patients with mRCC who had progressed on sunitinib, sorafenib, or both, were treated daily with orally administered everolimus plus best supportive care (BSC) in one arm and BSC alone in the other. Patients demonstrated a significant improvement in PFS when treated with everolimus plus BSC compared to BSC alone, resulting in a median PFS of 4.9 months (95% CI = 4.0–5.5) and 1.9 months (95% CI = 1.8–1.9), respectively. Compared to BSC alone, everolimus plus BSC was associated with a 67% reduction in risk for disease progression (HR = 0.33; p < 0.001)14. In the sub-group of patients previously treated with sunitinib as the only neoplastic therapy, everolimus treatment was associated with a median PFS of 4.6 months vs 1.8 months for BSC patients (HR = 0.22; 95% CI = 0.09–0.55)15.

Axitinib, a tyrosine kinase inhibitor, has also recently been approved to treat advanced RCC after failure of prior systemic therapy16. The AXIS phase III clinical trial enrolled patients who had progressed on a first-line regimen with sunitinib, bevacizumab plus interferon-alfa, temsirolimus, or cytokine therapy. The patients were randomly assigned to receive either axitinib or sorafenib. Patients showed significant improvement with axitinib compared to sorafenib. The median PFS was 6.7 months for axitinib patients and 4.7 months for sorafenib patients. Axitinib was associated with a 33.5% reduction in the risk of disease progression as compared to sorafenib (HR = 0.665; 95% CI = 0.544–0.812; p < 0.0001)17. Patients previously treated with sunitinib demonstrated a median PFS of 4.8 months with axitinib and 3.4 months with sorafenib (HR = 0.741; 95% CI = 0.573–0.958; p = 0.0107)17.

Both everolimus and axitinib therapies appear to provide clinical benefit in the treatment of advanced RCC, although they have not been compared in a head-to-head clinical trial study. However, two separate indirect comparisons of health outcomes for everolimus vs axitinib have suggested similar efficacy between the treatments18^,19. Payers considering their coverage status thus face the question of their relative costs. Currently, there are no studies that examine the treatment cost of everolimus vs axitinib for patients with sunitinib-refractory advanced RCC. Knowing the difference in total treatment costs between the therapies should help payers decide which treatment is more efficient in this patient population. Given the results of the two indirect comparisons, this analysis estimated the likely lifetime costs of everolimus patients compared to axitinib patients in the treatment of sunitinib-refractory advanced RCC from a US payer’s perspective.

Methods

Economic model structure

A Markov model was developed in Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA) to simulate two hypothetical patient cohorts with sunitinib-refractory advanced RCC. One cohort was treated with everolimus and the other with axitinib. In order to allow 99.99% of patients in both treatment arms to reach the absorbing death health state, the cohorts were modeled over a 10.6-year time horizon. A monthly (30.4-day) cycle (i.e., the average number of days per month in a year) was incorporated to model the cohorts. The four health states in the model included: stable disease with no adverse events (AEs), stable disease with AEs, disease progression, and death (Figure 1).

Figure 1. Markov model schematic.

All patients began in the stable disease with no AEs health state and transitioned to the disease progression and death health states according to the PFS estimate for axitinib derived from the AXIS trial publication17 and the overall survival (OS) estimate from the RECORD-1 clinical trial data20. As the results of two separate indirect comparisons suggest that everolimus and axitinib have similar PFS durations among patients with sunitinib-refractory disease18^,19, both cohorts were assumed to have equivalent PFS in the model. Furthermore, as OS data from the AXIS trial had not matured at the time of the analysis, the model assumed that OS between everolimus and axitinib were equivalent based on the demonstrated association between the treatment effects on PFS and the treatment effects on OS in patients with mRCC21. As the same survival curves were used for both treatment arms, membership in each health state was equivalent across treatment arms and cycles; therefore, neither everolimus nor axitinib derived any incremental survival benefit in the model.

Aside from movement between the stable disease without AEs and stable disease with AEs states, transitions through health states were unidirectional. For example, patients who moved to disease progression could not return to stable disease. Health state membership was estimated using a partitioned survival analysis, which calculated the average time spent in a health state from the area under the curve (AUC) approach. This type of analysis has been described in Glasziou et al.22^,23. Various distributions were constructed from OS data for a matched sub-set of the RECORD-1 population based on results from an indirect comparison with axitinib19 (derived in SAS), but Weibull cumulative distribution curves were eventually chosen to model the health-state membership at each cycle. The Weibull distribution was selected based on Akaike information criterion (AIC) and Bayesian information criterion (BIC) fit statistics, as well as visual inspection. Similarly, AIC, BIC, and visual inspection support the use of Weibull cumulative distribution curves to model the PFS curve estimates derived from the AXIS trial. Using these curves, the AUC was calculated to determine the health-state membership at each cycle. As health states were mutually exclusive, membership in the disease progression health state was calculated as the complement of the sum of the membership in the stable disease and death health states. The Weibull distribution is also uniquely relevant in the clinical setting, as it is the only parametric distribution that is simultaneously both proportional and accelerated so that both relative event rates and relative extension in survival time can be estimated24.

The proportion of everolimus patients who experienced a grade 3/4 AE in each cycle was calculated from RECORD-1 data for the matched sub-set. As such granular data was not publically available for axitinib at the time of the analysis, the model assumed that both treatment arms had an equivalent proportion of patients experience any AEs at each cycle. The AEs included in each treatment’s profile were different, however. As the AE profiles for the sub-set of patients previously treated with sunitinib were not known, the AEs included in the analysis were based on grade 3/4 AEs that occurred in at least 2% of everolimus patients in the RECORD-1 trial, as reported in the everolimus prescribing information25, and at least 2% of axitinib patients in the AXIS trial, as reported in the FDA Oncologic Drugs Advisory Committee (ODAC) briefing document26. Adverse event profiles did not affect the likelihood of progression or death, but were used to evaluate the associated increases in costs.

Resource utilization

To inform the economic evaluation, a retrospective database analysis was conducted using the MarketScan Commercial Claims and Encounters and Medicare Supplemental database (Truven Health Analytics, Ann Arbor, MI) to estimate resource utilization in patients with mRCC who received first-line sunitinib therapy before switching to another targeted therapy27. The database analysis included individual-level healthcare claims data from January 1, 2004 to June 30, 2011. Patients were included if they met the following inclusion criteria: had at least one claim for RCC during the study period; treated with sunitinib in the first-line setting; initiated second-line therapy with everolimus; had at least 3 months of follow-up prior to second-line initiation; and were at least 18 years of age. The resources assessed included physician and nurse visits, CT scans, MRIs, CBC blood tests, comprehensive metabolic panels, morphine sulfate claims and administrations, and oxycodone hydrochloride (HCL) claims (Table 1), each of which was modeled separately. Univariate negative binomial regression28^,29 was used to estimate the mean number of resources utilized per patient-month during the second-line setting (i.e., initial progression stage) and in the post-second-line setting (i.e., second progression stage).

Table 1. Resource utilization estimates for PFS† and PD health states.

	Progression-free survival health state (post-sunitinib)*	Progressive disease health state (post-second-line therapy)*
Office visit	0.104	0.188
Community nurse visit	0.000	0.000
CT scan	0.028	0.044
MRI	0.036	0.047
CBC blood test	0.081	0.079
Comprehensive metabolic panel	0.756	0.736
Morphine sulfate claims (oral)	0.019	0.033
Morphine sulfate claims (injection)	0.000	0.001
Morphine sulfate injection administration	0.027	0.058
Oxycodone HCL claims	0.051	0.110

PFS, progression-free survival; PD, disease progression.

†PFS includes stables disease with and without adverse events.

*Per patient month.

In the economic analysis, resource utilization estimates from the initial progression stage were included as the stable disease health state for the RECORD-1 patient population. To capture resource utilization estimates for the disease progression health state in the model, the analysis included results from the database analysis’ second progression stage (Table 1). Analysis was conducted using SAS 9.3 statistical software (Cary, NC).

Cost inputs

In the RECORD-1 trial, patients were randomly assigned to everolimus 10 mg per day plus BSC or placebo plus BSC14. Patients in the AXIS trial were randomized to receive axitinib 5 mg twice daily or sorafenib 400 mg twice daily17. In both trials, patients received treatment until disease progression, unacceptable toxicity, or study discontinuation14^,17. Patient-level data from the RECORD-1 trial was assessed to calculate everolimus dose intensity (88.1%) for patients who failed on first-line sunitinib therapy. Dose intensity was calculated by dividing the total dosage received by the expected dosage received over the PFS duration20. Axitinib dose intensity (102.0%) was obtained from the AXIS trial as reported in the FDA ODAC briefing document26. This document defined dose intensity as the proportion of the total dose administered divided the total dose assigned26^,30. Based on 2014 wholesale acquisition costs (WAC), obtained from RED BOOK Online31, everolimus and axitinib were estimated to be $342 per day and $340 per day (Table 2), respectively31. When incorporating dose intensity, the monthly drug acquisition costs were estimated to be $9174 for everolimus and $10,558 for axitinib (Table 2).

Table 2. Drug acquisition costs.

Drug	Dose and frequency	Drug acquisition cost ($USD)	Dose intensity (%)	Cost per monthly cycle
Everolimus	10 mg q.d.	$342.44 per day	88.1%20	$9173.82
Axitinib	5 mg b.i.d.	$340.31 per day*	102%26	$10,558.15

q.d., once daily; b.i.d., twice daily; $, costs in US dollars.

*Wholesale acquisition costs: $170.16 per 5-mg tablet, $56.72 per 1-mg tablet, accessed in July 2014.31

Additional costs considered in the analysis include general practitioner and nurse visits, imaging and lab tests, hospitalizations, adverse events, post-progression therapies, and palliative care. Costs of medical resources such as general practitioner and nurse visits were derived from current procedural terminology codes, as recorded in an online database of medical billing codes and information32. Adverse event costs were derived from a variety of sources including the Healthcare Cost & Utilization Project (HCUP)33, secondary literature34, and expert opinion. All AE costs calculated from HCUP or secondary literature were validated by a medical oncologist who also provided insight into the treatment setting (i.e., inpatient vs outpatient) for each AE according to severity (Table 3). As AE costs calculated from HCUP represented inpatient costs, AE costs attributed to outpatient stays were determined by applying an inpatient to outpatient cost ratio to the costs derived from HCUP. This cost ratio was calculated from Dial et al.35, which examined the costs of various adverse events, presented separately for inpatient and outpatient stays, in patients with RCC treated with angiogenesis inhibitor therapies.

Table 3. Adverse event costs and treatment setting.

Adverse event	Unit cost	Inpatient	Outpatient
Abdominal pain	$5228.19	100%	0%
Amylase	$0.00	0%	100%
Anemia
Grade 3	$64.44	0%	100%
Grade 4	$5949.20	100%	0%
Arthralgia	$52.88	0%	100%
Asthenia	$5559.34	100%	0%
Back pain
Grade 3	$5278.53	100%	0%
Decreased appetite	$6362.48	100%	0%
Dehydration
Grade 3 (Outpatient)	$44.91	0%	50%
Grade 3 (Inpatient)	$4145.99	50%	0%
Grade 4	$4145.99	100%	0%
Diarrhea	$5859.58	100%	0%
Dyspnea	$4967.93	100%	0%
Fatigue	$5559.34	100%	0%
Hypercholesterolemia
Grade 3	$67.85	0%	100%
Hyperglycemia	$4084.45	100%	0%
Hyperkalemia	$5130.82	100%	0%
Hypertension
Grade 3	$53.29	0%	100%
Grade 4	$4920.16	100%	0%
Hyponatremia
Grade 3	$59.74	0%	100%
Grade 4	$5515.89	100%	0%
Hypophosphatemia
Grade 3	$64.01	0%	100%
Grade 4	$5909.52	100%	0%
Infections/infestations	$5666.68
Lipase	$0.00		100%
Lymphocytopenia	$0.00		100%
Nausea	$5462.17	100%	0%
Palmar-Plantar Erythrodysesthesia
Grade 3	$41.68	0%	100%
Pneumonitis
Grade 3 (Outpatient)	$106.49	0%	50%
Grade 3 (Inpatient)	$9831.75	50%	0%
Proteinuria
Grade 3	$0.00	0%	100%
Stomatitis
Grade 3 (Outpatient)	$109.11	0%	50%
Grade 3 (Inpatient)	$10,073.67	50%	0%
Grade 4	$10,073.67	100%	0%
Vomiting	$3579.97	100%	0%
Weight decreased
Grade 3	$6700.26	100%	0%

Costs of adverse events were based on the Healthcare Cost & Utilization Project33, secondary literature34, and expert opinion.

$, costs in US dollars.

The post-progression treatment profile for everolimus was derived from RECORD-1 trial data based on the sub-group (n = 43) of interest20. Axitinib’s post-progression treatment profile was extracted from the clinical trial ODAC document, which reported information for the prior-sunitinib sub-group from the AXIS trial26. In order to inform the costs associated with these post-progression profiles, drug costs were taken from RED BOOK Online36. Lastly, secondary literature was used to inform end-of-life care costs in the analysis37. Other societal costs, such as patient time or productivity losses, were not considered since this analysis examined the lifetime costs from a third-party payer’s perspective (Table 4).

Table 4. Total health state medical costs.

Health state	Cost per patient ($USD)*
Stable disease with no adverse events	$491.90
GP and nurse visits	$104.69
Tests and scans	$355.85
Palliative care	$31.37
Stable disease with adverse events†	$3698.33 (everolimus) $4115.43 (axitinib)
GP and nurse visits	$22.97
Tests and scans	$78.09
Palliative care	$6.88
Adverse events†	$3590.38 (everolimus) $4007.48 (axitinib)
Disease progression	$27,283.00 (everolimus) $30,297.19 (axitinib)
GP and nurse visits	$225.57
Tests and scans	$551.88
Palliative care	$76.40
Post-progression treatments	$26,429.15 (everolimus) $29,443.34 (axitinib)
Death	$9448.48^A
End-of-life care	$9448.48

*Drug costs were derived from Red Book 201336; Medical resource utilization costs were taken from http://www.findacode.com/32; Costs of adverse events were based on the Healthcare Cost & Utilization Project33, secondary literature34, and expert opinion; End-of-life care costs were calculated from literature.37

†Adverse events included: abdominal pain, amylase, anemia, arthralgia, asthenia, back pain, decreased appetite, dehydration, diarrhea, dyspnea, fatigue, hypercholesterolemia, hyperglycemia, hyperkalemia, hypertension, hyponatremia, hypophosphatemia, infections/infestations, lipase, lymphocytopenia, nausea, palmar-plantar erythrodysesthesia, pneumonitis, proteinuria, stomatitis, vomiting, and weight decreased.

^AThis cost is applied upon entry into the death state.

$, costs in US dollars.

Sensitivity analyses

One-way sensitivity analyses were performed to identify parameters that substantially influence the lifetime costs of treating patients with sunitinib-refractory advanced RCC. One analysis included changes to axitinib’s drug acquisition cost based on real-world dosing data from France, which assumed the following: 5 mg BID for 71% of patients, 7 mg BID for 13% of patients, and 10 mg BID for 16% of patients38. In another scenario, dose intensity was 100% for both therapies. The effect of discounting was also examined by setting cost discount rates at 0%, and at 5%; however, the effect of discounting efficacy was not explored in this analysis since discounting both arms equally would negate the change across both interventions as a result of the equivalent efficacy assumption. In addition, adverse event costs were eliminated from both therapies in another sensitivity analysis. Lastly, monthly post-progression treatment costs for everolimus patients were assumed to be equivalent to that in patients treated with axitinib.

To further assess the uncertainty regarding the parameters in the analysis, a probabilistic sensitivity analysis (PSA) using a Monte Carlo simulation was performed to simulate a cohort of patients for 1000 iterations using the following distributions: cost parameters were varied along a gamma distribution and survival curve parameters and dose intensity along a normal distribution (Table 5).

Table 5. Model parameters varied in the PSA and their distributions.

Parameter	Distribution (range)	Rationale
Costs	Gamma	Standard convention.40
Dose intensity	Normal Note: This range was truncated so that values <0 could not be generated	The normal distribution was truncated in order to prevent generation of negative numbers and to avoid a skew towards higher values; the range of possible values is symmetrical across the mean. A beta distribution cannot be used as mean daily dose may be greater than 100%. A gamma distribution is not recommended as it has a long tail to the right, which is unlikely to reflect variance in mean daily dose.
Weibull parameters: shape, scale	Normal Note: This range was truncated so that values beyond the 95% CI could not be generated	The normal distribution was truncated to ensure that the shape and scale of the Weibull curve remained within the confidence interval bounds. Without truncating the distribution, iterations of the PSA may generate negative scale and shape parameter values, resulting in an error output in the Weibull function.

Results

In the base-case scenario, patients with sunitinib-refractory advanced RCC who were treated with everolimus accrued average lifetime costs of $104,226, compared to $117,211 for patients treated with axitinib. Therefore, patients treated with everolimus cost an average of $12,985 (11%) less over their lifetimes than patients treated with axitinib. This overall cost difference is largely due to the difference in drug costs. On average, drug acquisition costs for patients treated with axitinib were $9555 more than drug acquisition costs for everolimus patients. This difference in costs was primarily influenced by the therapies’ dose intensities. Moreover, post-progression treatment was $3013 more costly for axitinib patients compared to everolimus patients. Since the analysis assumed equivalent other healthcare resource utilization across therapies, and the indirect comparisons suggested similar efficacy, there were no differences in the overall costs of general practitioner and nurse visits, imaging and lab tests, and palliative care (Table 6).

Table 6. Base case results.

	Everolimus	Axitinib	Incremental
Mean months of drug supply (months)	6.16	7.14	(0.97)
Active treatment costs	$63,320	$72,875	($9555)
GP and nurse visits costs	$353	$353	–
Tests and scans costs	$986	$986	–
Palliative care costs	$9559	$9559	–
Adverse events treatment costs	$3590	$4007	($417)
Post-progression treatment costs	$26,418	$29,431	($3013)
Total	$104,226	$117,211	($12,985)

$, costs in US dollars.

The sensitivity analyses further demonstrated that everolimus is likely a less costly treatment option than axitinib for treating patients with sunitinib-refractory advanced RCC. When the dosing scheme for axitinib that reflected real-world data from France38 was incorporated into the analysis, the incremental, lifetime cost difference between everolimus patients and axitinib patients increased to $29,179. Even as dose intensities for everolimus and axitinib were both 100%, treatment with everolimus would still have a projected cost savings of $2983. Discounting costs had a negligible effect on the results, yielding an incremental cost of $13,255 and $12,815 when discounting was 0% and 5%, respectively. Similarly, eliminating adverse event costs from the analysis resulted in a projected incremental cost savings of $12,568, virtually the same as the base-case scenario. On the other hand, the incremental lifetime cost difference between everolimus patients and axitinib patients would decrease to $9972 when assuming that the monthly cost of post-progression treatment in everolimus is the same as that for axitinib patients. Despite any decrease in cost savings, the results of the various sensitivity analyses remained in favor of everolimus (Table 7).

Table 7. One-way sensitivity analysis results.

	Base case	Sensitivity analysis	Incremental cost (axitinib cost – everolimus cost)
Base case	N/A	N/A	($12,985)
General
Discounting (costs)	3.0%	0%	($13,255)
		5%	($12,815)
Costs
AE cost	$3590 (everolimus) $4007 (axitinib)	$0 (both treatment arms)	($12,568)
Post-progression treatment	$26,418 (everolimus) $29,431 (axitinib)	$29,431 (both treatment arms)	($9972)
Dosing
Active treatment dose intensity	88% (everolimus) 102% (axitinib)	100% (for both treatment arms)	($2983)
Axitinib dosing scheme*	5 mg BID	5 mg BID (71%) 7 mg BID (13%) 10 mg BID (16%)	($29,179)

*Axitinib dosing scheme for sensitivity analysis based on real-world data from France.38

N/A, not applicable; AE, adverse event; BID, twice daily; $, costs in US dollars.

The PSA simulated a cohort of patients for 1000 iterations, producing results that were consistent with the base-case analysis. Over the course of 1000 iterations, the average lifetime cost of treating patients with everolimus compared to axitinib was $105,113 and $117,930, respectively. Hence, the average incremental cost was $12,817 in favor of everolimus.

Discussion

With the assumption based on clinical evidence that everolimus and axitinib have similar PFS among second-line patients with advanced RCC who previously failed sunitinib therapy, it becomes important to determine which therapy represents the more cost-efficient option in this setting. The results of this cost comparison analysis suggest that second-line treatment with everolimus is likely to cost less than treatment with axitinib. The strength of this finding, which suggests a $12,985 incremental cost in favor of everolimus, is enhanced by the results of various sensitivity analyses, which consistently demonstrate cost savings for patients treated with everolimus. Despite uncertainties surrounding the parameters used in the analysis, the PSA results also support the finding that everolimus is a less costly treatment option for patients with sunitinib-refractory advanced RCC. As previously mentioned, the difference in projected lifetime costs is largely driven by the difference in drug acquisition costs, which also reflect dose intensity. To our knowledge, this is the first study comparing the lifetime costs of everolimus vs axitinib in patients with sunitinib-refractory advanced RCC; thus, it should be useful to physician and payers who have the responsibility to manage and monitor these clinical pathways. However, both the economic analysis findings and the assumed equivalence in clinical efficacy await further substantiation from a real-world comparison.

As costs vary depending on health states, the length of time that the hypothetical patient cohort modeled in this economic analysis spends in each health state thereby influences the aggregate cost of treatment. Hence, the choice of distribution has a significant impact on the lifetime treatment costs. In this analysis, Weibull cumulative distribution curves were chosen to represent the OS and PFS data in the model because it was determined to be the best-fit distribution in both instances, as well as clinically relevant24. Hence, the use of the Weibull distribution exerts a substantial influence on treatment costs, which must be projected for the patient cohort beyond RECORD-1 and AXIS clinical trials follow-up times.

The lack of granular dosing information regarding the timing and duration of dose titrations provided in the AXIS trial documentation required that additional assumptions be made in the model26. Thus, as information about the appropriate dosing scheme for axitinib patients was incomplete, the analysis assumed that the dose intensity of 102% led to a 2% increase in daily drug acquisition costs26. Since drug pricing for axitinib is not linear, this may not accurately reflect the cost attributable to axitinib.

As post-progression treatment profiles for everolimus and axitinib were based on their respective clinical trials, the analysis may not accurately reflect current real-world, third-line treatment patterns for patients with advanced RCC. Most patients in the RECORD-1 trial were administered sorafenib following progression on second-line everolimus treatment20. In contrast, a wider variety of therapies were administered to patients whose disease progressed after second-line axitinib treatment in the AXIS trial26. A higher proportion of patients who progressed on axitinib therapy ultimately used post-progression therapy, which accounted for a substantial portion of the higher post-progression treatment costs attributed to the axitinib arm in the analysis.

Regarding resource utilization, the MarketScan database analysis presented a set of limitations inherent to any analysis using claims data39. Moreover, as it was a univariate analysis, it did not adjust for potential resource use-driving patient characteristics that may influence the mean resource use estimates obtained and applied in the Markov model. Despite these limitations, the MarketScan analysis also offers the strengths inherent to a claims database analysis39. Rather than deriving resource utilization from secondary literature, or expert opinion that usually presents practice-specific perspectives, a claims database reflects real-world practices. However, since it was assumed that treatment patterns were the same across therapies in this instance, the MarketScan analysis does not capture the potential differences in resource use between everolimus patients and axitinib patients in real-world practice.

As previously mentioned, the AE profile for everolimus was captured from the RECORD-1 trial for the overall population. The prescribing information only captured treatment-emergent AEs with an incidence of ≥10% which occurred at a higher rate in the everolimus arm than in the placebo arm25. In contrast, the AE profile for axitinib was for the overall population, and taken from the AXIS trial ODAC briefing document, which recorded AEs that occurred at a rate of ≥5%26. As a result, AE-related costs for everolimus may be under-estimated. In this case, AE costs could be biased against the axitinib treatment arm. Furthermore, AE costs were initially derived from HCUP33 or secondary literature34. As the AE costing exercise was then validated by a single physician, the results may not be generalizable to all oncology practices. Additionally, the lack of available safety data for sunitinib-refractory axitinib patients precluded an accurate comparison of the AE profiles between the treatment arms. As a result, the analysis assumed equivalent per-cycle AE rates across therapies. Nonetheless, although the overall cost difference in adverse events slightly favored everolimus treatment, this was not a major driver in the overall cost comparison result.

The debate on sequential therapy in advanced RCC is clearly an important clinical and economic issue, and this analysis contributes to the ongoing discussion. While some studies have demonstrated clinical efficacy by administering therapies with different mechanisms of action, e.g., VEGFR-TKI and mTOR, additional studies investigating the range of possible treatment sequences should be conducted for further validation. This analysis suggests that, while both everolimus and axitinib are both beneficial second-line treatment options in the post-sunitinib setting, everolimus is likely to be the less costly option, which is largely driven by the higher dose intensity associated with axitinib as reported in the AXIS trial ODAC briefing document. While further studies are clearly warranted, providers and decision-makers should take the results of this analysis into consideration when treating patients with sunitinib-refractory advanced RCC.

Conclusion

Indirect comparisons suggest that everolimus and axitinib provide similar PFS benefits in patients with sunitinib-refractory advanced RCC; hence, this analysis compared the expected lifetime costs of these two regimens. Everolimus is projected to be a less costly treatment option than axitinib in patients with sunitinib-refractory advanced RCC. Scenario analyses support the robustness of this finding.

Transparency

Declaration of funding

Novartis Pharmaceuticals Corporation provided funding for this study. Novartis employees made clinical trial data available for analysis, and offered analytical suggestions. The sponsored research did not put limits on freedom to publish or the content of publication.

Declaration of financial/other relationships

RC, SS, AP, AC, & MG are employees of LASER Analytica, a consultancy that received compensation for the overall economic study design, the analysis, and preparation of this manuscript. ZL, XW, & KC are employees of and own stock in Novartis Pharmaceuticals Corporation. They also contributed to the analysis and manuscript preparation. SP's potential conflicts of interest include serving on advisory boards for Novartis and Aveo, performing corporate-sponsored research for GSK, and receiving honoraria from Pfizer, Novartis, Astellas and Medivation. LPG has served on an advisory board for Novartis and received compensation for his contributions to the study design, data analysis and preparation of the manuscript. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentations: This study was presented in part (Figure 1 and Tables 2, 4, and 6) at the Kidney Cancer Association, 12th International Kidney Cancer Symposium; October 25–26, 2013, Chicago, IL. The poster was titled, ‘The Lifetime Cost of Everolimus vs Axitinib in Metastatic Renal Cell Carcinoma Patients who have Failed Prior Sunitinib Therapy in the US’.