Number needed to treat and associated incremental costs of treatment with enzalutamide versus abiraterone acetate plus prednisone in chemotherapy-naïve patients with metastatic castration-resistant prostate cancer

Abstract

Objective: Enzalutamide (ENZA) and abiraterone acetate plus prednisone (AA) are approved second-generation hormone therapies for chemotherapy-naïve metastatic castration-resistant prostate cancer (mCRPC). This study compared ENZA with AA in chemotherapy-naïve mCRPC by calculating the number needed to treat (NNT) and associated incremental costs to achieve one additional chemotherapy-naïve patient with mCRPC free of radiographic progression, chemotherapy, or death over a 1-year time horizon.

Methods: Clinical outcomes were obtained from the PREVAIL and COU-AA-302 trials. Three outcomes were evaluated: radiographic progression-free survival, time to cytotoxic chemotherapy initiation, and overall survival at 1 year. NNT was calculated as the reciprocal of the outcome event rate difference for ENZA compared with AA. The incremental costs to achieve one additional outcome at 1 year were calculated as the difference in cost per treated patient multiplied by the NNT. Per-treated-patient costs were considered from a US payer perspective and included medications, monitoring, adverse events, post-progression treatments, and end-of-life care.

Results: Within a 1-year time horizon, the total cost per treated patient for ENZA was $2,666 less than AA. Compared with AA, treating 14 patients with ENZA resulted in one additional patient free of progression or death over 1 year; treating 26 patients with ENZA resulted in one additional patient with chemotherapy delayed over 1 year; and treating 91 patients with ENZA resulted in one additional patient free of death over 1 year. Therefore, ENZA is cost-effective compared with AA for all three outcomes evaluated, and the modeled results suggest ENZA is associated with potentially improved clinical outcomes in delaying chemotherapy initiation and disease progression for chemotherapy-naïve patients. The results are robust in sensitivity analyses, where the effect of changes in key model inputs and assumptions were tested.

Conclusion: The results modeled in the present study suggest ENZA is cost-effective compared with AA for treating chemotherapy-naïve patients with mCRPC.

Keywords:

• Castration-resistant prostate cancer
• Cost-effectiveness
• Number needed to treat
• Cost per outcome
• Enzalutamide
• Abiraterone

Introduction

Prostate cancer is the most commonly diagnosed cancer and the second leading cause of cancer-related death among men in the US¹. Castration-resistant prostate cancer (CRPC) is an advanced form of prostate cancer characterized by disease progression following surgical or pharmacological (e.g. androgen deprivation) castration, a treatment modality aimed at suppressing androgen production that contributes to stimulating growth of prostate cancer cells. CRPC is estimated to account for 10–20% of prostate cancer cases, with over 70% of these cases defined as metastatic CRPC (mCRPC)². The prognosis for patients with mCRPC is generally poor, although the introduction of several novel oral oncolytic agents in recent years has increased the overall survival of these patients^2–4.

Among therapies that have recently gained regulatory approval for the treatment of chemotherapy-naïve mCRPC, enzalutamide (ENZA) and abiraterone acetate plus prednisone (AA) are second-generation hormonal agents that received Category 1 recommendation in the National Comprehensive Cancer Network (NCCN) guidelines for the treatment of mCRPC⁵. The median survival was 35.3 months with ENZA⁶ and 34.7 months with AA⁷ in their respective pivotal clinical trials.

There are currently no published head-to-head clinical trials comparing ENZA with AA for the treatment of chemotherapy-naïve patients with mCRPC. Therefore, indirect comparisons are useful to evaluate the relative efficacy and cost-effectiveness of the two oral oncolytic agents. The efficacy of ENZA and AA, observed in their respective randomized controlled trials with similar study populations, suggests that ENZA is associated with potentially improved clinical outcomes with respect to delaying chemotherapy initiation and delaying disease progression throughout the chemotherapy-naïve settings^3,4. One prior study compared the drug cost per median overall survival month of ENZA to AA in chemotherapy-naïve patients with mCRPC from a US payer perspective⁸. Even though a sensitivity analysis was conducted to include monitoring costs, the study did not consider adverse event-related costs, costs associated with subsequent therapies or incremental costs of disease progression or death.

To evaluate the relative value of these two treatment options from a US third-party payer perspective, the current analysis was established to compare ENZA with AA indirectly with respect to cost-effectiveness by calculating the number needed to treat (NNT) and the associated incremental cost per outcome. The NNT methodology was introduced in 1988 as a way to quantify the clinical usefulness of different treatments⁹. NNT is calculated as the inverse of the event rate difference between two treatment alternatives, with the outcome representing the number of patients who require treatment to prevent one additional event compared to treatment with an alternative therapy. From the NNT calculation, the associated incremental cost per additional outcome can be estimated and interpreted as the cost-effectiveness of alternative treatments^10–16. The use of NNT in health economic evaluations is demonstrated among clinicians and formulary decision-makers. Additionally, there are many literature-based examples of NNT and incremental cost-per-additional-outcome applications in oncology^10–16; however, none have been published for mCRPC treatments to date.

The current study compared ENZA with AA using the NNT method with the following outcomes: (1) achieving one additional patient free of radiographic progression or death; (2) achieving one additional patient with chemotherapy delayed; and (3) achieving one additional patient with death avoided, with a 1-year time horizon. The corresponding associated incremental cost-effectiveness values (i.e. the incremental cost per additional outcome) among chemotherapy-naïve patients with mCRPC were also calculated from a US third-party payer perspective. The cost components considered in this study included medications, monitoring, adverse events, post-progression treatments, progression, and end-of-life care.

Methods

In the current study comparing ENZA with AA, NNT and incremental cost per additional patient with clinically meaningful outcome were estimated. The study outcomes were representative of an average patient with mCRPC treated with ENZA or AA and were evaluated over the 1-year time horizon following the initiation of ENZA or AA. A 1-year time horizon was selected after consideration of the annual formulary review process and budget time horizons of US payers.

Input data: efficacy in the base case

Data on the clinical efficacy of ENZA and AA were obtained from the PREVAIL and COU-AA-302 trials, respectively (Table 1)^3,6,7,17. Both trials included asymptomatic or mildly symptomatic patients with mCRPC without prior chemotherapy. Patient age, Gleason score, pain level, and laboratory measures including alkaline phosphatase and lactate dehydrogenase at baseline were similar between the two trials. In the PREVAIL trial, 11.2% of patients had visceral disease at screening, whereas the COU-AA-302 trial excluded patients with visceral disease. Given most patient characteristics were comparable at baseline except for visceral disease, the efficacy outcomes were directly used. The difference in the proportion of visceral disease may lead to a conservative efficacy estimate for ENZA in the current study. The outcomes evaluated in the study included radiographic progression-free survival (rPFS), time to initiation of chemotherapy, and overall survival. These three outcomes were considered clinically meaningful in the current study and were measured in both the PREVAIL and COU-AA-302 trials; specifically, rPFS and overall survival were the co-primary end-points. In addition, time to initiation of chemotherapy was evaluated and considered a valuable outcome for patients and caregivers¹⁸. The observed 1-year rates of rPFS, chemotherapy delay, and overall survival were obtained from the respective Kaplan-Meier curves using pixel analysis without adjustment.

Table 1. Input data for drug costs, monitoring, and efficacy.

Input	ENZA	AA
Drug cost
Daily dose	160 mg^6	1000 mg + prednisone 10 mg^7
WAC per mg, USD	$1.84^19	$0.29, $0.06 (prednisone)^19
Proportion of patients with concomitant prednisone use, %	27^6	100^7
Compliance rate (Astellas internal data), %	92	91
Mean treatment duration within 1 year, months*	9.6^6	9.0^7
Monitoring
Physician visit (CPT: 99213)	$72.94 × once monthly^20	$72.94 × once monthly^7^,^20
Basic metabolic panel test (CPT: 80048)	0^6	$15.55 × once monthly^7^,^21
Hepatic function panel test (CPT: 80076)	0^6	$15.02 × once monthly^7^,^21
1-year efficacy
rPFS rate, %	65.0^6	58.0^7
Mean rPFS duration, months^†	10.2^6	9.6^7
Chemotherapy-free rate, %	83.4^3	79.6^17
Mean chemotherapy-free duration, months^†	11.3^3	11.1^17
Overall survival rate, %	91.8^6	90.7^7
Mean overall survival duration, months^†	11.7^6	11.6^7

AA, abiraterone acetate plus prednisone; CPT, current procedural terminology; ENZA, enzalutamide; rPFS, radiographic progression-free survival; USD, United States Dollars; WAC, wholesale acquisition cost.

* Mean treatment durations were estimated using the area-under-the-curve technique based on the reported median treatment durations and constant hazard of discontinuation assumption.

^† Durations with clinically meaningful outcomes were calculated using the area-under-the-curve technique based on the respective Kaplan-Meier curves.

Input data: cost and healthcare resource use in the base case

Unit costs of ENZA, AA, and prednisone were obtained from Red Book^® Online (2016 wholesale acquisition cost [WAC], accessed on March 15, 2016; Table 1)¹⁹. Dosing schedules of ENZA, AA, and prednisone were based on product labels in the US^6,7. Furthermore, mean treatment duration within 1 year was estimated from the trial data^6,7, and assumptions for real-world compliance rates were obtained from the Astellas Clinformatics™ Data Mart database (OptumInsight, Eden Prairie, MN) by utilizing the medication possession ratio methodology. Concomitant use of prednisone is required as a labeled indication for patients using AA; similarly, the proportion of patients using ENZA with concomitant prednisone use was obtained from the product label^6,7.

Frequency of monitoring tests within 1 year after treatment initiation was based on each product label, and unit costs for monitoring tests were obtained from the Centers for Medicare & Medicaid Services (Table 1)^20,21. Grade 3/4 adverse events reported in ≥1% of patients from the PREVAIL or COU-AA-302 trials and any events noted in warnings in the product labels were considered in the model (Table 2)^3,6,7,22. The unit costs of treating each adverse event were obtained from the literature^20,23–30. The model also assumes that the adverse events, which may be etiologically related (e.g. increased alanine aminotransferase and increased aspartate aminotransferase, edema, and cardiac events/acute renal failure), are experienced by different patients, or at different times if incurred by the same patient.

Table 2. Input data for adverse events.

Adverse event	Unit cost, USD	1-year rates of Grade 3/4 adverse event*, %
		ENZA^3^,^6	AA^7^,^22
Acute renal failure^20^,^24^†	$9,958	1.0	–
Adrenocortical insufficiency^20^,^24^†	$9,533	–	0.4
Asthenia/fatigue^25^,^27	$8,419	2.3	1.9
ALT increased^20^,^24^†	$6,110	0.2	5.3
AST increased^20^,^24^†	$6,110	–	2.7
Atrial fibrillation^30	$17,098	0.2	1.6
Back pain^20^,^24^†	$7,026	1.7	–
Cardiac events^26	$20,241	1.7	6.6
Dyspnea^25^,^27	$10,863	0.4	2.1
Edema^20^,^24^†	$7,397	0.1	0.3
Fall^37	$10,752	1.1	–
Hematuria^20^,^24^†	$7,577	0.9	1.1
Hyperglycemia^23	$2,505	–	5.6
Hypertension^25	$2,284	5.0	3.4
Hypokalemia^20^,^24^†	$6,342	–	2.4
Joint swelling^20^,^24^,^†	$6,986	1.1	1.8
Lower respiratory tract infection^38	$5,211	1.0	–
Lymphopenia^25	$11,668	–	7.5
Non-pathological fracture^29	$4,229	1.4	–
PRES^20^,^24^†	$13,678	0.1	–
Seizure^28	$8,855	0.1	–

AA, abiraterone acetate plus prednisone; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ENZA, enzalutamide; PRES, posterior reversible encephalopathy syndrome; USD, United States Dollars.

* The adverse event rates for enzalutamide and abiraterone acetate plus prednisone were obtained by prorating the rates reported in the trials by the median treatment duration to a 1-year rate, except for PRES, which was based on assumption.

^† Due to lack of relevant information, it was assumed that costs for treating Grade 3/4 events would be equivalent to a hospital stay with relevant condition and a follow-up outpatient visit.

Upon disease progression, patients with mCRPC were assumed to incur one hospitalization and receive post-progression treatments. Unit costs for progression-related hospitalization were obtained from Healthcare Cost and Utilization Project data, based on admission costs of prostate cancer patients (Table 3)²⁴. Dosing schedules of post-progression treatments were obtained from product labels, and unit costs were obtained from Red Book^® Online^{®6,7,19,31–33}. The model assumed that only deceased patients within the study time horizon would incur end-of-life costs. Subsequently, to remain consistent with the model time horizon, only patients who progressed or were deceased within the first year following therapy initiation were considered to incur progression, end-of-life, and post-progression treatment costs. Aside from drug costs based on 2016 WAC values in US dollars, all other costs were inflated to 2015 US dollars using the medical component of the Consumer Price Index in the US³⁴.

Table 3. Input data for post-progression and end-of-life costs.

Inputs	Value
ENZA: proportion of patients receiving each post-progression treatment^3^*, %
Sipuleucel-T	2.2
Docetaxel	53.3
ENZA	0.0
AA plus prednisone	35.0
Cabazitaxel	9.5
AA: proportion of patients receiving each post-progression treatment^22^*, %
Sipuleucel-T	7.4
Docetaxel	50.8
ENZA	25.5
AA plus prednisone	0.0
Cabazitaxel	16.3
Costs for post-progression treatments, USD
Sipuleucel-T cost of one treatment course	$121,479^19^,^31
Docetaxel monthly cost	$1,317^19^,^32^,^39
ENZA monthly cost	$8,848^6^,^19
AA monthly cost	$8,646^7^,^19
Cabazitaxel monthly cost	$13,391^19^,^33
Hospitalization due to progression or death, USD
One-time hospitalization cost	$13,407^24
Hospice care
Proportion of patients receiving hospice care before death, %	53^40
Average length of stay in hospice, days	47^40
Cost per day in hospice, USD	$171^41

AA, abiraterone acetate plus prednisone; ENZA, enzalutamide; USD, United States Dollars.

* Proportions of patients receiving each post-progression treatment were derived from the proportions reported in the trials and reweighted to 100%, assuming that each patient would only receive one type of post-progression treatment, other than the original therapy, within the time horizon.

Calculation: costs

Total average costs included pre-progression, post-progression, and end-of-life costs. Pre-progression costs included drug costs, monitoring costs, and adverse-event-related costs, whereas post-progression costs included costs for post-progression treatments and hospitalizations. In addition, end-of-life care was comprised of costs for hospitalization and hospice service. For all cost components, only costs incurred within the 1-year period following the initiation of ENZA or AA were considered.

Drug costs for ENZA and AA within 1 year of therapy initiation were calculated, based on the dosing schedule, unit price, proportion of patients with concomitant prednisone use, compliance rate, and mean treatment duration within the year for each cohort. The mean treatment duration within 1 year was estimated using the area-under-the-curve method, based on the median treatment duration (17.5 months for ENZA and 13.8 months for AA) and a constant hazard rate of discontinuation assumption^6,7. Specifically, the probability of treatment discontinuation within a half-month cycle was derived for ENZA and AA, respectively, based on the median treatment durations and assuming a constant hazard rate for discontinuation. A time-to-discontinuation curve was constructed to model the probability of being on treatment for each therapy, and the mean treatment duration during the 1-year period after treatment initiation was estimated by the area-under-the-curve method based on the curve.

Monitoring cost within 1 year was estimated based on the monitoring schedule and unit costs. The AA product label specified that patients are monitored for blood pressure, serum potassium, symptoms of fluid retention, and hepatotoxicity at least once per month⁷. Therefore, patients incurred one physician visit, one test for basic metabolic panel, and one test for hepatic function panel per month. No specific monitoring schedule was required for ENZA, according to the product label⁶. Patients treated with ENZA were assumed to have the same frequency of physician visits as those treated with AA.

Adverse event-related cost within 1 year was calculated from event rates and associated unit costs. The 1-year adverse event rates were prorated from the reported rates in the trials to reflect gradual incidence of Grade 3/4 adverse events for chemotherapy-naïve patients with mCRPC in the PREVAIL trial^3,6,7,22. Furthermore, the 1-year adverse event rates were calculated as the overall event rates observed in the PREVAIL trial and the COU-AA-302 trial divided by the corresponding overall median treatment duration and multiplied by 12 months.

Post-progression costs within 1 year were calculated as the sum of post-progression treatment cost and hospitalization cost related to progression. The model assumed that only patients with disease progression would receive post-progression treatments and progression-related hospitalization. Subsequent treatments reported in the ENZA and AA trials were considered (including sipuleucel-T, docetaxel, and cabazitaxel)^3,22. Re-treatment with the initial product was not allowed in the model (e.g. patients initially treated with ENZA would not be re-treated with ENZA post-progression). Specifically, it was assumed that the proportion of patients re-treated with ENZA in the PREVAIL trial for ENZA would receive AA as the post-progression treatment in the real world, with ENZA for AA patients. The proportion of patients receiving each type of post-progression treatment was estimated based on the clinical trial reports. Sipuleucel-T was a one-time personalized therapy, and the full treatment course was considered among recipients. Costs of the other (recurring) post-progression treatments were calculated as the monthly drug cost multiplied by the eligible treatment duration. The eligible treatment duration was the difference of mean duration of overall survival and mean duration of rPFS within the 1-year period (i.e. the time patients were alive, but after progression during the 1-year period following initial treatment with ENZA or AA). Average post-progression treatment costs within 1 year were estimated as the weighted average cost of all post-progression treatments considered in the model.

End-of-life cost was calculated as the sum of end-of-life hospitalization cost and hospice cost; only deceased patients within 1 year incurred end-of-life costs. End-of-life hospitalization cost was calculated as the unit cost of hospitalization multiplied by the proportion of deceased patients. Additionally, hospice care cost was calculated as the proportion of deceased patients multiplied by the proportion of patients receiving hospice before death and multiplied by the average length of stay and daily cost.

Calculation: NNT and incremental cost per additional patient outcome

The NNT and cost per additional patient with clinically meaningful outcome comparing ENZA with AA were calculated for each outcome, respectively (Figure 1). NNT was calculated as the reciprocal of the event rate difference of clinically meaningful outcomes between ENZA and AA from published clinical trials of the same population. An incremental cost per additional patient with clinically meaningful outcome was calculated as the difference in total costs (ENZA vs AA) multiplied by the corresponding NNT. In situations where the difference in total cost is negative and the NNT is positive, the incremental cost per additional patient with clinically meaningful outcome was denoted as “dominant” to comply with the standard reporting approach for cost-effectiveness analyses³⁵.

Figure 1. Illustration of methodology. NNT, number needed to treat; rPFS, radiographic progression-free survival.

Sensitivity analysis

One-way sensitivity analyses were conducted to examine the effect of change in one key model input or assumption on model results, while others are held at the base-case value. The following was tested in the one-way sensitivity analyses: (1) varying the proportion of patients hospitalized upon disease progression from 100% in the base case to 50% and 0%, (2) changing post-progression treatment composition from values observed in the respective clinical trial in the base case to be the same between ENZA and AA, (3) varying rates of adverse events by ±25% of base-case values for both arms and by ±25% for ENZA only, (4) varying costs of adverse events by ±25% of base-case values, (5) varying monitoring costs by ±25% of base-case values, (6) adding monthly metabolic panel and hepatic function tests for ENZA, (7) adding prostate-specific antigen tests and bone scans every 3 months, and (8) varying end-of-life costs by ±25% of base-case values. The real-world concomitant use of corticosteroid with ENZA may also be different from the proportion observed in the clinical trial. However, due to the low cost of corticosteroid therapies, the impact on study results was expected to be minimal and, therefore, was not considered in the sensitivity analysis.

Results

Base-case results

Within a 1-year time horizon, the total estimated cost per treated mCRPC patient was $94,440 for ENZA and $97,105 for AA (Table 4). As a net impact, the 1-year total cost was $2666 lower for an average patient treated with ENZA compared with AA. While the acquisition costs for ENZA were considered higher than those for AA (difference of $6880, which is partly due to longer treatment duration), ENZA was associated with lower costs for monitoring (difference of −$236), adverse events (difference of −$2650), post-progression treatments (difference of −$6465), and end-of-life care (difference of −$194).

Table 4. Total cost per treated patient within a 1-year time horizon.

Cost component	ENZA, USD	AA, USD	Difference (ENZA–AA), USD
Drug cost	$77,791	$70,911	$6,880
Monitoring cost	$697	$933	−$236
Adverse-event-related cost	$1,361	$4,011	−$2,650
Post-progression cost	$13,142	$19,608	−$6,465
End-of-life cost	$1,449	$1,643	−$194
Total cost per treated patient	$94,440	$97,105	−$2,666

AA, abiraterone acetate plus prednisone; ENZA, enzalutamide; USD, United States Dollars.

The NNT for rPFS was 14 when comparing ENZA with AA; thus, treating 14 mCRPC patients with ENZA resulted in one additional patient with rPFS at 1 year vs treating with AA. Additionally, the NNT for chemotherapy delayed was 26 when comparing ENZA with AA; thus, treating 26 mCRPC patients with ENZA resulted in one additional patient with chemotherapy delayed at 1 year vs treating with AA. Lastly, the NNT for overall survival was 91 when comparing ENZA with AA; thus, treating 91 mCRPC patients with ENZA resulted in one additional patient alive at 1 year vs treating with AA (Table 5). ENZA was cost-effective compared to AA for all three outcomes; specifically, the modeled results suggest ENZA was associated with lower costs and favorable clinical outcomes (Table 5). Treatment of chemotherapy-naïve patients with mCRPC with ENZA vs AA would result in more mCRPC patients with potentially improved outcomes at a lower overall cost.

Table 5. Clinical outcomes and associated costs comparing ENZA with AA within a 1-year time horizon.

Outcome	NNT	Cost difference, USD	Incremental cost per additional patient with outcome
Free of progression or death  (based on rPFS)	14	−$2,666	Dominant
Chemotherapy delayed  (based on chemotherapy-free rate)	26		Dominant
Death avoided  (based on overall survival)	91		Dominant

AA, abiraterone acetate plus prednisone; ENZA, enzalutamide; NNT, number needed to treat; rPFS, radiographic progression-free survival; USD, United States Dollars.

One-way sensitivity analysis results

In the one-way sensitivity analyses (Figure 2), ENZA is consistently associated with cost savings compared with AA under all tested scenarios. The cost difference between ENZA and AA was most sensitive to post-progression treatment composition—when ENZA has the same treatment composition as AA, the cost savings were reduced to $398 from $2666 in the base case. In other sensitivity analyses, the cost savings comparing ENZA to AA ranged from $1727–$3328. In terms of the incremental cost per additional patients with outcome, ENZA was cost-effective compared with AA in all sensitivity analyses (specifically, potential for improved outcomes at a lower overall cost).

Figure 2. One-way sensitivity analysis.

Discussion

This study used clinical trial data supplemented with references in the literature to calculate NNT (and associated incremental costs) comparing two novel oral oncolytic agents (ENZA and AA) for treating chemotherapy-naïve patients with mCRPC. The results suggest ENZA compared with AA can potentially prolong survival, decrease the risk of radiographic progression, and delay chemotherapy in chemotherapy-naïve patients with mCRPC. Results from the incremental costs per additional patient with clinically meaningful outcome analysis conclude that the clinical benefits of ENZA were achieved at a lower cost compared with AA within a 1-year time horizon from a US payer perspective. Extensive sensitivity analyses have been conducted for the key inputs and assumptions used in the model, including post-progression treatment, adverse event rates and costs, disease monitoring, and end-of-life costs. The study results are robust in all sensitivity analyses.

NNT and incremental cost per additional patient with clinically meaningful outcome (overall survival, rPFS, and chemotherapy delayed) were selected as the evaluation approach in the current study because these outcomes are easily interpretable and relevant to payer decision-making. This approach allows the comparison of alternative treatments with respect to efficacy and costs. To the knowledge of the authors, no publication has applied this approach for evaluating treatments in mCRPC, although this approach has been used for evaluating other oncology treatments^11,14–16. For example, a published meta-analysis concluded that adding long-term hormone therapy resulted in an NNT of 14 to achieve one additional locally advanced prostate cancer patient with clinical progression-free survival compared with radiotherapy alone¹¹. However, there is no published threshold for NNT, and determining whether an NNT value is clinically meaningful depends on the disease and patient expectations. In the current study, the NNT to achieve one additional rPFS for ENZA vs AA was 14, and the clinical benefit was achieved with cost savings. Similarly, when chemotherapy delayed or death avoided was evaluated, treating mCRPC patients with ENZA was associated with more patients experiencing potentially improved outcomes at a lower cost compared to treating patients with AA.

A previous study compared ENZA and AA for the treatment of chemotherapy-naïve mCRPC in the US utilizing a different methodology; it concluded the cost per median overall survival month was lower for AA than for ENZA⁸. However, the study focused on drug acquisition costs and incorporated 2015 WAC values, without considering costs of adverse events and costs of downstream events associated with different clinical outcomes such as post-progression treatments and end-of-life care⁸. Similarly, another previous study used administrative claims data and demonstrated that AA was associated with lower monthly pharmacy costs, but similarly did not consider other costs associated with these treatments³⁶.

The current study was designed to compare ENZA and AA from a US third-party payer perspective and considered cost components beyond drug acquisition cost, specifically aligned with a 1-year budget decision cycle. Subsequently, although the estimated acquisition cost of ENZA was higher than AA, the clinical benefits of ENZA offset the higher acquisition cost. Additionally, treatment with AA requires monitoring of mineralocorticoid excess, adrenocortical insufficiency, and hepatotoxicity at least monthly; however, there is no requirement for regular monitoring with ENZA specified in the product label. The difference in toxicity profiles was also associated with a higher cost of adverse events for AA compared with ENZA.

Based on our assumptions, patients treated with ENZA experienced less disease progression and spent more time free of progression compared with those treated with AA; therefore, they were associated with lower post-progression treatment costs. In addition, the greater 1-year overall survival rate of patients treated with ENZA compared with those treated with AA resulted in lower end-of-life costs for ENZA. Lastly, the higher acquisition cost of ENZA in the model was partly due to the longer mean treatment duration of ENZA compared with AA (9.6 vs 9.0 months), which may be associated with improved clinical outcomes of ENZA (e.g. longer treatment duration and less discontinuation). When evaluating treatment options, it is important to consider aspects beyond the drug acquisition cost and examine the complete measure of overall costs in addition to the clinical benefits.

There are known limitations of the current study. First, the model used results reported in two randomized clinical trials to estimate the safety and efficacy profiles of ENZA and AA. Safety and efficacy outcomes from clinical trials may not be representative of the experience of the general patient population; therefore, the generalizability to real-world clinical practice may be limited. In addition, the current study did not test uncertainties associated with the efficacy of ENZA and AA. Future US real-world studies are needed to further differentiate these two treatment options in terms of benefits, risks, and associated costs. Second, the current economic model assumed that patients would have one hospitalization upon disease progression and would receive one of the following post-progression treatments: sipuleucel-T, docetaxel, ENZA, AA, or cabazitaxel, and the proportions of patients receiving each treatment are the same as reported in the respective clinical trials. Patients with mCRPC in the real world may receive treatments that are not considered in this study (e.g. radium-233) or may have different probabilities of receiving each treatment from those observed in the respective clinical trials, or even receive multiple sequential treatments, after progression. It is also possible that some patients with mCRPC will not be hospitalized upon disease progression. To evaluate the impact of these assumptions and inputs, different probabilities of hospitalization after progression and different composition of post-progression treatments were tested in these sensitivity analyses, and the results are robust. Third, while the time horizon is 1 year in the current study, some patients would incur additional costs beyond 1 year. Future studies would be beneficial to evaluate costs incurred by these patients beyond the time horizon evaluated in this study.

In conclusion, the current study suggests that ENZA is cost-effective compared with AA for treating chemotherapy-naïve patients with mCRPC. ENZA, compared with AA, potentially decreases the risk of radiographic progression and death and delays chemotherapy in chemotherapy-naïve patients with mCRPC based on results from randomized controlled trials with similar study populations. Additionally, ENZA is associated with a lower cost, compared with AA, due to reduced post-progression, end-of-life, monitoring, and adverse event costs.

Transparency

Declaration of funding

This research was sponsored by Astellas Pharma, Inc. and Medivation, Inc., the co-developers of enzalutamide.

Declaration of financial/other relationships

MM, CNB, BJP, BB, and SF are employees of Astellas. HY, JG, YS, and EW have served as consultants to Astellas. AB is an employee of Medivation. MB was an employee of Medivation at the time this work was initiated. Astellas Pharma, Inc. and Medivation, Inc. participated in the interpretation of data and the review and approval of the publication. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors would like to thank Jing Zhao from Analysis Group for a significant contribution toward analytical support and Ana Bozas from Analysis Group for medical writing assistance. Editorial assistance, funded by both sponsor companies, was provided by Charlene Rivera, PhD, Stephanie Rippon, and Lauren Smith from Complete HealthVizion.