Budget impact analysis of everolimus for the treatment of hormone receptor positive, human epidermal growth factor receptor-2 negative (HER2−) advanced breast cancer in Kazakhstan

Abstract

Objective:

The aim of this study was to determine the budget impact of everolimus (in combination with letrozole/anastrozole) as a second-line treatment for ER+ HER2− negative advanced and metastatic breast cancer in post-menopausal women.

Research design and methods:

A cumulative cohort model was developed to estimate the 5-year costs associated with introducing everolimus to the Kazakh healthcare system. Two alternative market share scenarios were compared: with everolimus and without everolimus. PFS and OS data were taken from the trial and extrapolated. The background costs of the pre-progressed and post-progressed health states, drug costs and costs associated with adverse events were included in the model.

Results:

The 5-year results from the budget impact analysis demonstrate that the introduction of everolimus leads to a 12% increase in drug costs, a 2% reduction in pre-progression health state costs, a 1% increase in post-progression health state costs, and a 2% reduction in adverse event costs. The net result is only a modest increase in total costs; a 2.69% increase of T201 million, from T7.5 billion to T7.7 billion over a period of 5 years.

Conclusions:

The analysis estimated that, if everolimus were to be introduced to the Kazakh healthcare market for the treatment of ER+ HER2− advanced breast cancer, there would be minimal impact upon overall healthcare expenditure. An increase in drug acquisitions costs was almost exactly offset by a reduction in other healthcare costs, due to improved management of the disease.

Keywords::

• Everolimus
• Breast cancer
• Budget impact
• Kazakhstan

Introduction

Breast cancer is the most common cancer in women across the world. The World Health Organisation (WHO) estimates that in 2011 breast cancer resulted in ∼508,000 deaths1. It is estimated that in 2012 over 1.6 million new cases of breast cancer were diagnosed and that the prevalence of breast cancer cases diagnosed in the previous 5 years was over 6 million2. The prevalence of breast cancer in Kazakhstan was estimated to be 27,137 in 20123. The number of newly diagnosed breast cancer patients in Kazakhstan in 2012 was 39513. Metastatic breast cancer is the presence of disease at distant sites. It is estimated that the total number of patients with breast cancer will be more than 27,000 in Kazakhstan in 2014, of which ∼8% will develop metastatic disease4.

The choice of treatment for women with locally advanced or metastatic disease depends on a number of variables, including previous treatment, site of metastases, receptor status of tumor cells, menopausal status, health, and informed patient choice. For women with estrogen receptor positive (ER+) advanced breast cancer, many authorities recommend offering endocrine therapy (e.g., a non-steroidal aromatase inhibitor (AI) such as letrozole and anastrozole) as first-line treatment for the majority of patients5. For post-menopausal women with locally advanced or metastatic ER+, human epidermal growth factor receptor 2-negative (HER2-negative) breast cancer, the National Institute for Health and Care Excellence (NICE) does not currently recommend biological therapies (e.g., trastuzumab; Herceptin)6. If disease progresses after treatment with AIs, patients with refractory disease typically move on to other treatments in second line. Everolimus (Afinitor), a new mTOR (mammalian target of rapamycin) inhibitor, is indicated for second-line treatment of ER+, HER2-negative advanced and metastatic breast cancer in post-menopausal women. Everolimus blocks the activation of S6K1 and 4E-BP1, thereby inhibiting cell growth, proliferation, and G1–S transition, and inducing apoptosis7. Everolimus has the potential to act on tumor cells both directly to inhibit growth and indirectly by inhibiting and displaying anti-vascular properties.

The effectiveness of everolimus (in combination with exemestane) has been demonstrated against exemestane alone in the BOLERO-2 trial. The BOLERO-2 trial8 was a randomized, double-blind, phase III study of everolimus plus exemestane vs exemestane plus placebo, which has been conducted in 189 centers in 24 countries. The trial participants were women with HER2-negative, ER+ advanced or metastatic breast cancer whose disease had recurred or progressed following prior treatment with non-steroidal aromatase inhibitors. BOLERO-2 is the only phase III registration trial assessing the use of everolimus in this indication and so is used as the basis for the current study.

Patients were randomized to either oral everolimus (10 mg/day) plus exemestane (25 mg/day) (n = 485) or Exemestane (25 mg/day) plus matching placebo (10 mg/day) (n = 239). Treatment continued until disease progression, the development of unacceptable toxicity, or withdrawal of consent. The protocol provided detailed guidelines for dose interruptions or reductions for everolimus and matched placebo for adverse events; two reductions in dose were permitted. The primary outcome measure was progression-free survival, whilst secondary outcomes included overall survival, overall response rate, clinical benefit rate, time to response, duration of response, safety (adverse events, biomarker analysis, vital signs, time to deterioration of ECOG performance status), pharmacokinetics, and quality-of-life. Further details on the BOLERO-2 trial have been published in Baselga et al.9 and Beaver and Park10.

The purpose of this study was to estimate the budget impact associated with the introduction of everolimus to the Kazakhstan healthcare system.

Methods

A cumulative cohort model was developed in Microsoft Excel to estimate the 5-year costs associated with introducing everolimus to the Kazakh healthcare system. In order to do so, two alternative scenarios were compared: (i) An existing state of affairs, with current market share being used to estimate the numbers of patients receiving different treatments, and (ii) an ‘intervention’ scenario, where everolimus is introduced to the market and assumed to replace a proportion of existing treatments (see Table 1). The budget impact was estimated for 2014–2018.

Table 1. Market share with and without everolimus.

	1st line	2nd line	3rd line
Current market share (2015–2018)
Everolimus + letrozole	0%	0%	0%
Letrozole/anastrozole	15%	30%	30%
Chemotherapy	70%	70%	70%
Tamoxifen	15%	0%	0%
Total	100%	100%	100%
Market share with everolimus (2014)
Everolimus + letrozole	1.0%	1.0%	0.0%
Letrozole/anastrozole	15.0%	30.0%	30.0%
Chemotherapy	69.1%	69.0%	70.0%
Tamoxifen	14.9%	0.0%	0.0%
Total	100%	100%	100%
Market share with everolimus (2015)
Everolimus + letrozole	2.0%	2.0%	0.0%
Letrozole/anastrozole	15.0%	30.0%	30.0%
Chemotherapy	68.3%	68.0%	70.0%
Tamoxifen	14.7%	0.0%	0.0%
Total	100%	100%	100%
Market share with everolimus (2015–2018)
Everolimus + letrozole	3.0%	3.0%	0.0%
Letrozole/anastrozole	15.0%	30.0%	30.0%
Chemotherapy	67.5%	67.0%	70.0%
Tamoxifen	14.6%	0.0%	0.0%
Total	100%	100%	100%

Everolimus is used in combination with letrozole/anastrozole in Kazakhstan, since exemestane is not registered there. As data on everolimus in combination with letrozole or anastrozole is not available, the data from the two arms of BOLERO-2, everolimus + exemestane and exemestane alone, were assumed in the current model to be representative of everolimus + letrozole/anastrozole and letrozole/anastrozole alone, respectively.

Epidemiology

A cumulative cohort approach to the modeling was taken. That is, in the first year of the model, the first year’s cohort enters the model and incurs the costs of their first year of treatment. In the second year of the model, any surviving members of the first year’s cohort are joined by a new incident population. Thus, in the second year, two separate cohorts are being modeled, the first cohort in their second year and the second cohort in their first year. In the third year of the model a new, third, incident cohort is introduced, and so forth until the model has completed 5 years. The model was limited to a 5-year horizon in order to avoid introducing additional uncertainty due to unknown long-term treatment outcomes, as well as unknown changes in market conditions. This is shown in Figure 1.

Figure 1. Cumulative cohort approach.

The Kazakh Scientific Research Institute of Oncology and Radiology has estimated that, in 2012, the total number of patients with breast cancer was 27,1373. This figure was applied as an estimate of the number of patients who would be entering at the start of the model in 2014. Of those, it is expected that 8.02%4 will progress to the advanced (metastatic) stage of disease. Therefore, 2176 patients per year will enter the advanced stage of the disease. Everolimus is indicated only for patients who are post-menopausal and have the ER+/HER2− form of the disease. It is estimated that 54% of advanced breast cancer patients are post-menopausal and that, of these, 66% are ER+/HER2−10. Therefore, the total number of patients eligible to be treated with everolimus is 776, and this is the total number of patients entering the model in 2014, the first year of the model. The incidence for years 2–5 was based on 3951 newly diagnosed patients per year, taken from 2012 official national statistics3. This was increased by an annual growth of 8.58%, which is the average of the growth in incidence from 2010–2011 and from 2011–2012, to estimate the incidence for 20143. This same growth was then applied in each subsequent year of the model to estimate the increasing incidence from years 2015–2018. Therefore, the estimated incident populations in years 2014–2018 were 4658, 5057, 5491, and 5952, respectively. The number of post-menopausal patients with advanced ER+/HER2− breast cancer was calculated using the same filters as used for the prevalent population in 2014, giving 133, 145, 157, and 170 newly diagnosed patients in 2015–2018, respectively, Of the newly diagnosed and eligible patients, it was assumed that 17.6% were receiving first-line treatment, 53.6% were receiving second-line treatment, and 28.8% were receiving third- or higher-line treatment (Novartis data on file).

The model uses current market share estimates to determine the total number of patients receiving each form of therapy in the base case model3. These are shown in Table 1. These market share figures are assumed to remain constant throughout the base case model (i.e., without the introduction of everolimus). To estimate the cost impact of introducing everolimus to the market, it was assumed that, for first- and second-line treatment, everolimus (plus letrozole or anastrozole) would gain 1% of the market share in the first year, 2% in the second year, and 3% in the third, fourth, and fifth years. For third-line therapy, everolimus was assumed to have zero uptake. For first-line therapy, the increase in the use of everolimus was assumed to be drawn proportionally from patients previously assigned to chemotherapy and tamoxifen. For second-line treatment, the increase was assumed to be drawn from patients assigned to chemotherapy.

Costs

To estimate the costs associated with each treatment, a number of different factors were considered. First, drug acquisition costs were accounted for based on the existing market share. Second, non-drug healthcare costs (e.g., monitoring, investigations, other treatments) were accounted for dependent upon the patient’s health status and severity of disease. Thus, the model accounts for different rates of progression-free survival (PFS) and overall survival (OS) rates associated with each treatment, drawn from published trials11. PFS and OS are used to calculate patients’ duration in each health state and the associated costs and are not used to calculate quality-adjusted life years (QALYs).

In order to allow extrapolation of PFS and OS beyond the observed trial period, parametric curves were statistically fitted to observed trial data. Extrapolation of the BOLERO-2-derived survival curve using parametric distributions (exponential, Weibull, log-normal, log-logistic, and Gompertz distributions) was performed and the goodness-of-fit examined.

Interim OS data were available up to 17 months from BOLERO-2 for everolimus + exemestane and exemestane alone, with a median follow-up of 16.2 months6^,12. Kaplan-Meier curves for this period were used to calculate parametric curve fits. Thereafter, the parametric survival curves were used to extrapolate the trial data to give the 5-year survival data required for the model’s time horizon. In the base case the parametric function was applied for the whole of the model time horizon, rather than using the observed data, as data points were not available on a monthly basis, making the survival curve very ‘stepped’. A Weibull function was chosen for the base case, guided by feedback from expert clinicians, which indicated that the Weibull function had the highest clinical plausibility. The R-squared values for the Weibull fits were 0.935 and 0.950 for everolimus and comparator arms, respectively. To estimate the probability of death in each monthly cycle of the model, the following Weibull formula was used: where t represents time and σ and λ represent the shape and scale parameters, respectively. This was used to calculate survival for each one1monthly cycle in the model.

Unlike the OS data the BOLERO-2 PFS data were available at monthly intervals. However, for consistency, the parametric curves were also applied throughout the model. As for OS, a Weibull function was applied in the base case, guided by feedback from expert clinical opinion. The extrapolations were based on PFS data up to 27 months for everolimus + exemestane and up to 22 months for exemestane alone. The R-squared values for the Weibull fits were 0.998 and 0.990 for everolimus and the comparator arm, respectively.

Data from the BOLERO-2 trial’s two arms (everolimus plus exemestane and exemestane alone) were assumed to be representative of the everolimus + letrozole/anastrozole and the letrozole/anastrozole groups in this model, respectively. PFS and OS data for the other comparators included in the model (i.e., chemotherapy and tamoxifen) were taken from published sources. Data for chemotherapy was based on docetaxel (19 month PFS data and 36 month OS data)12. Data for tamoxifen were taken from Novartis data on file from the TAMRAD trial (27 month PFS and 40 month OS data). For consistency, the same approach was taken as for the BOLERO-2 data outlined above: parametric Weibull curves were applied throughout the model. The Weibull PFS and OS parameters and estimated mean PFS and OS for all interventions in the model are shown in Table 2.

Table 2. PFS and OS parameters.

	Everolimus + letrozole/ anastrozole	Letrozole/anastrozole	Chemotherapy	Tamoxifen
Weibull PFS σ parameter	0.067	0.191	0.064	0.133
Weibull PFS λ parameter	1.118	1.006	1.348	1
Weibull OS σ parameter	0.013	0.014	0.047	0.009
Weibull OS λ parameter	1.105	1.172	1.11	1.304
Mean estimated PFS	0.90	0.44	0.59	0.63
Mean estimated OS	2.98	2.49	0.68	2.17

Probabilities for both progression and death are time-dependent. That is, a patient’s risk of mortality changes over time. Incorporation of time-dependence is standard for survival modeling in oncology and is accounted for within the parametric functions used to determine the transition probabilities within the model13. The PFS and OS curves assigned to each of the treatments are shown in Figures 2 and 3, respectively.

Figure 2. Progression-free survival curves.

Figure 3. Overall survival curves.

Drug costs were applied to patients for the entire period that they remained in the progression-free health state. The drug costs per month are shown in Table 3, whilst the costs for chemotherapy are shown in Table 4. Average body surface area was assumed to be 1.72 m² 14. Compliance was assumed to be 100% in all patients until disease progression. Costs associated with adverse events were also included within the model. Rates of grade 3 and 4 adverse events for each treatment were taken from published sources8^,12^,15, and were assigned an individual cost per event, as shown in Table 5.

Table 3. Drug costs.

	Dose (mg)	Unit size (mg per unit)	Pack size (no. units)	Cost per pack	Cost per unit	Cost per month
Letrozole	2.5	2.5	30	Т 9,309	Т 310	T 9,445
Anastrozole	1.0	1.0	28	Т 2,936	Т 105	T 3,191
Letrozole/Anastrozole*						T 6,318
Everolimus	10.0	10	30	Т 1,052,880	Т 35,096	T 1,068,235
Eve + Letrozole/Anastrozole						T 1,074,553
Tamoxifen	20.0	20	30	Т 690	Т 23	T 700

*The cost for letrozole/anastrozole is calculated as the mean cost of letrozole and anastrozole individually.

Source for cost of everolimus: {Order of Chairman of the Board of ‘SK-Pharmaceuticals’ LLP № 06/62 as of June 10, 2013}, Minister of Healthcare of the Republic of Kazakhstan Order No. 224 as of April, 15, 201317.

Source for costs of other drugs: based on average costs of drugs from http://pharmprice.kz/ database for the period June–September 2013 and the following official documents: Attachment to ‘Minister of Healthcare of the Republic of Kazakhstan Order No. 444 as of July, 1, 2011’18.

Table 4. Cost of chemotherapy regimes.

Chemotherapy regimens	Proportion of patients	Dose (mg)	Unit size (mg)	Pack size (no. units)	Cost per pack	Cost per unit	Units required per dose	Cost per dose	Doses per month	Cost per month
Docetaxel	60%	135	80	1.7	Т 151,298	Т 88,999	2	Т 177,998	1.45	Т 259,147
Doxorubicin	60%	108	50	2.2	Т 1,304	Т 593	3	Т 1,778	1.45	Т 3,734
Capecitabine	20%	63,000	500	112	Т 84,876	Т 758	126	Т 95,486	1.45	Т 139,554
Carboplatin	20%	600	150	4	Т 3,729	Т 932	4	Т 3,729	1.45	Т 6,561
Cyclophosphamide	60%	1,080	200	5.4	Т 97	Т 18	5	Т 97	1.45	Т 1,297
PEG-doxorubicin	3%	90	20	4.5	Т 77,798	Т 17,288	5	Т 77,798	1.45	Т 113,917
Epirubicin	20%	108	59	2.2	Т 1,870	Т 850	2	Т 1,700	1.45	Т 3,621
Fluorouracil	10%	1,080	250	4.3	Т 115	Т 27	5	Т 134	1.45	Т 1,351
Gemcitabine	5%	3,600	1000	3.6	Т 5,500	Т 1,528	4	Т 6,111	1.45	Т 10,014
Paclitaxel	30%	315	100	3.2	Т 12,100	Т 3,781	4	Т 15,125	1.45	Т 23,079
Vinorelbine	20%	90	10	9	Т 12,870	Т 1,430	9	Т 12,870	1.45	Т 19,810
Weighted* mean cost of all therapies (including T 2458 administration cost)										T 205,851

*The weighted mean cost is calculated using the proportion of patients receiving each drug.

Cost source: based on average costs of drugs from http://pharmprice.kz/ database for the period June–September 2013.

Table 5. Cost of adverse events.

	Eve + Letrozole/ Anastrozole	Letrozole/ Anastrozole	Chemotherapy	Tamoxifen	Costs	Assumptions
Stomatitis	8%	1%	9%	0%	Т 995	Micanazole + Mitronidazole
Rash	1%	0%	2%	0%	Т 1,000	Anhistamine (suprasin)
Fatigue	5%	1%	12%	6%	Т 2,000	Solamedrol
Diarrhoea	3%	1%	10%	0%	Т 1,444	Linex, Loperamide
Anorexia	1%	1%	0%	2%	Т 2,000	Solamedrol
Nausea	1%	1%	6%	0%	Т 1,950	Ondacentron + Cirucal
Cough	1%	0%	0%	0%	Т 280	Amburaxaan
Pneumonitis	0%	0%	0%	2%	Т 9,220	Cephalosporin + Amburaxan
Thromboembolic	0%	0%	0%	4%	Т 54,847	Klexan −30 days, Veraphalin
Pain	0%	0%	0%	10%	Т 15,300	Tramadol
Conjunctivitis	–	–	0%	–	Т 765	Eye drops
Infection	–	–	26%	–	Т 6,067	Cephalasporin + flucanozole
Fluid retention	–	–	3%	–	Т 6,350	Mannitol + Diacard
Peripheral neuropathy	–	–	5%	–	Т 10,320	Vitamines + TAD
Nail toxicity	–	–	5%	–	Т 467	Miconazole
Local phlebitis	–	–	1%	–	Т 980	Troxevasine or Heparin
Allergy	–	–	2%	–	Т 1,000	Anhistamine (suprasin)
Leukopenia	–	–	77%	–	Т 54,010	Filgrastim
Thrombocytopenia	–	–	3%	–	Т 256,200	Revolate
Anaemia	–	–	2%	–	Т 362,754	Epoetin
Weighted cost	Т257	Т72	Т59,477	Т4,068	–	–

Cost source: based on average costs of drugs from http://pharmprice.kz/ database for the period June–September 2013.

Non-drug costs were also included in the model. Costs in the stable disease state comprise acquisition costs for active treatment (see above), cost of disease management including hospitalisations and testing, and adverse event costs (see above). In the stable disease state, costs are applied for as long as the patient stays in the pre-progressed state. Costs in the pre-progression health state differed between chemotherapy and the other interventions. The total cost per month for chemotherapy was based on the cost of supportive palliative care and was Т114,338. The cost for the other interventions was T7354. The breakdown of these costs is shown in Table 6. Pre-progression health state costs were taken from the Minister of Healthcare of the Republic of Kazakhstan Order No. 44416.

Table 6. Progression-free health state costs.

		All interventions excluding chemotherapy		Chemotherapy
	Unit cost	Resource use per month	Cost per month	Resource use per month	Cost per month
Costs of outpatient services
Clinical Oncologist	Т 768	1.00	Т 768	1.33	Т 1,021
GP visit	Т 640	–	–	1.33	Т 851
Clinical nurse specialist	Т 548	1.00	Т 548	2.00	Т 1,096
Ambulance	Т 2,188	–	–	1.00	Т 2,188
Psychotherapist	Т 768	–	–	1.00	Т 768
Outpatient care	Т 548	–	–	4.00	Т 2,192
Clinical nurse home visit	Т 548	–	–	4.00	Т 2,192
Total cost of palliative outpatient care per month			T 1316		Т 5,155
Costs of in-patient services
Clinical Oncologist	Т 768	–	–	1.33	Т 1,021
GP practices	Т 973	–	–	1.33	Т 1,294
Radiotherapy (40% of patients)	Т 774	–	–	20.00	Т 6,190
Hospitalization days (40% of patients)	Т 5,000	–	–	28.00	Т 56,000
Total cost of palliative in-patient care per month			T 0		Т 64,505
Laboratory tests
Complete blood count	Т 1,272	1.00	Т 1272	4.00	T 5,088
Blood chemistry panel	Т 2,795	1.00	Т 2795	4.00	T 11,180
Liver function test	Т 500	1.00	Т 500	4.00	T 2,000
ECG	Т 501	1.00	Т 501	1.00	T 501
Creatinine (urine)	Т 415	1.00	Т 415	4.00	T 1,661
Haematuria (urine)	Т 555	1.00	Т 555	4.00	T 2,220
Total laboratory costs			T 6038		Т 22,650
Additional methods of examination
CT scan	Т 7,820	–	–	0.30	T 2,346
Ultrasound (general & for breast)	Т 1,940	–	–	0.30	T 582
Conventional x-ray (breast 2 projections)	Т 891	–	–	0.30	T 267
Additional therapy
Antibiotics (Cephalosporin)	Т 9,220	–	–	10%	T 922
Analgesics (Tramadol)	Т 15,300	–	–	70%	T 10,710
Dexamethasone	Т 1,200	–	–	50%	T 600
Diphosphonate (Zometa)	Т 11,000	–	–	60%	T 6,600
Grand total cost per month for health state			T 7354		T 114,338

Cost source: based on average costs of drugs from http://pharmprice.kz/ database for the period June–September 2013.

Post-progressed resource use was also taken from the Minister of Healthcare of the Republic of Kazakhstan Order No. 44416. The average monthly cost for the post-progressed state was based on the assumption that all patients would receive supportive palliative care and some patients would also receive therapy. The costs for supportive palliative care were the same for the post-progressed state as for those pre-progressed patients on chemotherapy. In addition, patients on chemotherapy incurred a cost of administration of T3562 per month. The different sequences of treatment and their weightings are shown in Table 7. Those patients receiving third line hormonal therapy in the post-progressed state were assumed to receive tamoxifen. Those patients receiving chemotherapy in the post-progressed state were also assigned a cost of T62,777 for additional palliative care (this cost was not applied to patients receiving chemotherapy in the pre-progressed state). It was assumed that a proportion of patients would receive active post-second-line chemotherapy, while a proportion would receive supportive palliative care only. Unit costs were taken from the Minister of Healthcare of the Republic of Kazakhstan Order No. 44416. The background costs for the progression-free and post-progressed states are shown in Table 6.

Table 7. Post-progression treatment sequences.

Post-progression sequences	% patients receiving sequences	Average total time (months)	Average total cost per treatment sequence
Third line hormonal therapy + supportive palliative care	7%	7.8	Т 897,296.25
Chemotherapy + supportive palliative care	30%	15.8	Т 6,068,305.60
Third line hormonal therapy + chemotherapy  + supportive palliative care	51%	18.6	Т 7,156,722.69
Supportive palliative care	12%	5	Т 114,338
Total (weighted averages)	100%	15.4	Т 5,546,952
Average monthly cost			Т 360,848

Terminal care costs included hospital and hospice resource use and at home community support. The total cost per month was T19,974.

Sensitivity analyses

A range of sensitivity analyses were conducted, in which the following factors were varied: Progression and mortality for both everolimus and its comparators; AE rates for both everolimus and its comparators; costs of comparator treatments; market share of everolimus; prevalence and incidence of breast cancer. Because each of the sensitivity analyses varied several parameters simultaneously (e.g., costs across a range of comparator treatments), all of these, with the exception of average growth rate of morbidity of all patients, used a multiplication coefficient to vary the whole range of inputs. Thus, a multiplication coefficient of 1.0 implies that the parameters were not changed, whilst a coefficient of 1.5 suggests that all relevant parameters were increased by 50%.

Results

The 5-year results from the budget impact analysis are shown in Table 8. These demonstrate that the introduction of everolimus leads to a 12% increase in drug costs, a 2% reduction in pre-progression health state costs, a 1% increase in post-progression health state costs and a 2% reduction in adverse event costs. The net result is only a modest increase in total costs; a 2.69% increase of T201 million, from T7.5 billion to T7.7 billion over a period of 5 years.

Table 8. Budget impact analysis results.

	2014	2015	2016	2017	2018	Total costs
Without everolimus
Drug costs	703,479,209	195,477,482	147,656,460	157,025,334	170,388,963	1,374,027,447
Pre-prog costs	400,093,960	111,220,480	84,030,545	89,331,806	96,927,330	781,604,120
Post-prog costs	977,149,347	1,411,775,407	1,115,840,850	918,853,301	840,971,947	5,264,590,854
AE costs	32,392,258	5,559,744	6,036,552	6,554,252	7,116,350	57,659,156
Total	2,113,114,773	1,724,033,113	1,353,564,407	1,171,764,693	1,115,404,590	7,477,881,577
With everolimus
Drug costs	741,488,831	221,657,242	177,187,186	189,717,645	206,101,774	1,536,152,678
Pre-prog costs	396,516,235	109,687,140	81,949,741	87,015,993	94,411,068	769,580,177
Post-prog costs	979,190,462	1,418,446,421	1,127,149,398	933,523,890	858,542,150	5,316,852,320
AE costs	32,073,459	5,451,431	5,860,150	6,362,721	6,908,393	56,656,154
Total	2,149,268,987	1,755,242,234	1,392,146,475	1,216,620,248	1,165,963,385	7,679,241,328
Incremental impact of introducing everolimus
Drug costs	38,009,622	26,179,760	29,530,726	32,692,311	35,712,812	162,125,231
Pre-prog costs	−3,577,725	−1,533,340	−2,080,804	−2,315,813	−2,516,262	−12,023,944
Post-prog costs	2,041,115	6,671,013	11,308,548	14,670,588	17,570,202	52,261,466
AE costs	−318,799	−108,313	−176,403	−191,531	−207,957	−1,003,002
Total	36,154,213	31,209,120	38,582,067	44,855,556	50,558,795	201,359,752

The results of the sensitivity analyses are shown in Figure 4. The tornado diagram shows that the results of the model are particularly sensitive to the prevalence and incidence of breast cancer, the progression-free survival and overall survival of everolimus, and the overall survival of the comparators. When the prevalence and incidence of breast cancer are increased and decreased by 20%, the total budget impact is increased by 44% and decreased by 32%, respectively, to T289 million and T136 million. When the market share of everolimus was increased and decreased by 20% this resulted in a corresponding increase and decrease in the budget impact of 20%, as the magnitude of the budget impact was directly impacted. The absolute budget impact ranged from T241.6 million to T161.0 million, respectively, When progression on everolimus is decreased by 20% the total budget impact is increased by 11%, to T223.7 million. This is because, as progression-free survival is prolonged, patients remain on treatment for longer and consequently incur more costs. Conversely, when progression on everolimus is increased by 20% there is an 8% decrease in the total budget impact, to T184.9 million When mortality for everolimus is increased and decreased by 20% the budget impact changes by −5% and 6%, to T191 million and T213 million, respectively. This is due to the fact that cancer patients incur high resource use and so extended survival results in increased healthcare expenditure, whereas decreased survival results in reduced healthcare expenditure. Similarly, when the risk of mortality for the comparator interventions is increased and decreased by 20%, the total budget impact of everolimus changes by −6% and 5%, to T188 million and T211 million, respectively. This is because poorer survival for the comparator interventions means a reduction in background healthcare costs and so a higher incremental cost resulting from the introduction of everolimus. In the sensitivity analyses conducted, the total budget impact of introducing everolimus to the Kazakh healthcare system varied from T136 millon to T223 million. In all of these scenarios everolimus + letrozole/anastrozole resulted in increased costs to the Kazakh healthcare system. When considering the estimated percentage budget impact change after the introduction of everolimus there was little variation throughout the sensitivity analyses. In the base case the total budget for post-menopausal patients with advanced ER+/HER2− breast cancer was estimated to increase by 2.69% following the introduction of everolimus. In the sensitivity analyses conducted this value ranged from 2.15–3.23%.

Figure 4. Sensitivity analyses.

Discussion

This analysis has estimated the financial impact of introducing a new therapy, everolimus, as an adjunct to current treatments. The analysis includes drug acquisition costs as well as impact on the other healthcare resources, and accounts for disease progression and patient mortality. The model also accounts for new incident patients entering the model in future years. The analysis demonstrates that, although uptake of everolimus results in an increase in drug acquisition costs, this is almost entirely offset by savings in other healthcare resource use, despite a greater survival rate for patients receiving everolimus resulting in greater longer-term care and healthcare resource use.

As with all economic models, the analysis uses statistical techniques to present a simplified version of reality, while aiming to retain the key nuances of the disease progression and treatment pathways. For example, it was not possible to build in a treatment effect associated with post-second-line treatment, as there are no data on the efficacy of such treatments. However, since the OS curves were based on actual trial data (in which patients were treated with post-second-line therapy), the effect of such treatments would already be included.

Because the model includes all healthcare costs, rather than simply the cost of treatment, the analysis accounts for any costs incurred as a result of increased survival. Because everolimus results in increased PFS and OS (with the incremental OS being greater than the incremental PFS), there are substantial non-drug costs associated with its increased effectiveness. As the per-cycle costs for progressed disease are higher than for progression-free disease, the increase in survival in the progressed state results in increased non-drug costs for everolimus. Because of this, an economic analysis of any effective treatment for metastatic breast cancer will, ultimately, penalise those therapies that result in increased survival of patients and, as such, the budget impact results generated by such a model are not necessarily reflective of the cost of drug treatment alone.

In order for the model to calculate costs and benefits over a 5-year time horizon, PFS and OS were extrapolated beyond the trial period. This brings additional uncertainty into the model inputs. The robustness of the model would be improved if trial data of longer follow-up were available. However, the impact of varying the PFS and OS for everolimus and for the comparator interventions has been explored in sensitivity analysis. Of these four factors (PFS with everolimus + letrozole/anastrozole; PFS with comparators; OS with everolimus + letrozole/anastrozole; OS with comparators) it was found that comparator PFS had the smallest impact on the absolute budget impact and everolimus + letrozole/anastrozole had the largest impact. Despite this it was found that the impact of the introduction of everolimus on the relative budget change was very consistent, with the increase in total budget following its introduction ranging from 2.47–2.99%.

In addition to the uncertainty due to extrapolation there is uncertainty around a number of the model inputs. For example, as the model is a projection over the next 5 years, by definition the market share for everolimus in these years is unknown and so assumptions from experts had to be used. It is likely that, if everolimus were to gain a higher market share than is estimated in the current model, the incremental budget impact would be higher. The distribution of patients over first-, second-, and third-line therapy was also unknown and so was based on Novartis data on file. To explore this uncertainty a number of sensitivity analyses were conducted. It was found that, when varying key inputs by ±20% the results of the model did not change greatly. The absolute budget impact of the model varied between T136 millon to T223 million. The relative impact of introducing everolimus to the market ranged from a 2.15% to a 3.23% increase on total costs.

Conclusion

This analysis has demonstrated that, if everolimus were to be introduced to the Kazakh healthcare market for the treatment of advanced breast cancer, there would be minimal impact upon overall healthcare expenditure. An increase in drug acquisition costs was almost exactly offset by a reduction in other healthcare due to improved management of the disease. Sensitivity analysis demonstrated the model is most sensitive to the probability of progression for patients on everolimus and the probability of mortality for the other interventions.

Transparency

Declaration of funding

LL & MT’s employer, York Health Economics Consortium, received funding for health economics consultancy from Novartis Pharma AG.

Declaration of financial/other relationships

York Health Economics Consortium receives funding for conducting health economics research and consultancy. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.