Cost-effectiveness analysis of secukinumab for the treatment of active psoriatic arthritis: a Canadian perspective

Abstract

Objective: The study evaluates the cost-effectiveness of secukinumab, a fully human monoclonal antibody that selectively neutralizes interleukin (IL)-17A, vs currently licensed biologic treatments in patients with active psoriatic arthritis (PsA) from a Canadian healthcare system perspective.

Methods: A decision analytic semi-Markov model evaluated the cost-effectiveness of secukinumab 150 mg and 300 mg compared to subcutaneous biologics adalimumab, certolizumab pegol, etanercept, golimumab, and ustekinumab, and intravenous biologics infliximab and infliximab biosimilar in biologic-naive and biologic-experienced patients over a lifetime horizon. The response to treatments was evaluated after 12 weeks by PsA Response Criteria (PsARC) response rates. Non-responders or patients discontinuing initial-line of biologic treatment were allowed to switch to subsequent-line biologics. Model input parameters (Psoriasis Area Severity Index [PASI], Health Assessment Questionnaire [HAQ], withdrawal rates, costs, and resource use) were collected from clinical trials, published literature, and other Canadian sources. Benefits were expressed as quality-adjusted life years (QALYs). An annual discount rate of 5% was applied to costs and benefits. The robustness of the study findings were evaluated via sensitivity analyses.

Results: Biologic-naive patients treated with secukinumab achieved the highest number of QALYs (8.54) at the lowest cost (CAD 925,387) over a lifetime horizon vs all comparators. Secukinumab dominated all treatments, except for infliximab and its biosimilar, which achieved minimally more QALYs (8.58). However, infliximab and its biosimilar incurred more costs than secukinumab (infliximab: CAD 1,015,437; infliximab biosimilar: CAD 941,004), resulting in higher cost-effectiveness estimates relative to secukinumab. In the biologic-experienced population, secukinumab dominated all treatments as it generated more QALYs (8.89) at lower costs (CAD 954,692). Deterministic sensitivity analyses indicated the results were most sensitive to variation in PsARC response rates, change in HAQ, and utility values in both populations.

Conclusions: Secukinumab is either dominant or cost-effective vs all licensed biologics for the treatment of active PsA in biologic-naive and biologic-experienced populations in Canada.

Keywords:

• Secukinumab
• psoriatic arthritis
• cost-effectiveness
• Canada
• biologics
• tumor necrosis factor inhibitors
• incremental cost effectiveness ratio (ICER)
• quality-adjusted life year (QALY)
• biosimilars

Introduction

Psoriatic Arthritis (PsA) is a chronic, systemic inflammatory disease, associated with psoriasis. PsA falls under the umbrella condition of Spondyloarthritis (SpA) and mainly affects peripheral joints and the axial skeleton^1–4. The global epidemiology estimates of PsA are not yet considered accurate due to the different diagnostic criteria that are currently implemented. The global prevalence of PsA in the general population has been reported to vary from 0.01–0.67%^5,6. The global annual incidence has been reported to range between 0.1 cases per 100,000 population in Japan to 23.1 cases per 100,000 population in Finland^7,8. In Canada, an annual incidence of 2.7 cases of PsA per 100 psoriasis patients was reported in 2016⁹.

The economic burden of PsA on the healthcare systems worldwide is significant. Based on the mean direct cost for PsA (US$3,638) extracted from claims data¹⁰, the direct annual healthcare costs for PsA were estimated to be as high as US$1.9 billion in the US in 2000¹¹. In a literature review on the economic burden of PsA, the mean direct costs were reported to range from $4,008 in Hungary to $5,646 in the US, whereas the mean indirect costs ranged from $4,359 in Hungary to $12,949 in Germany¹¹.

In Canada, an observational database study reported that the average annual drug costs (per patient in 2012 Canadian dollars) were $61.31 for non-steroidal anti-inflammatory drugs (NSAIDs), $386 for disease-modifying anti-rheumatic drugs (DMARDs), and $10,730 for biologics¹². In addition, patients spent an average of $257 per year on over-the-counter medications and $549 per year on complementary or alternative medicine. Their average annual healthcare costs in terms of clinic visits, diagnostic tests, hospital admissions, and surgeries were $15,667¹². In another study at a center in Toronto, Ontario, the mean annual direct cost of PsA was $15,802 per patient, of which $5,499 was non-pharmacologic costs, and $10,219 was pharmacologic costs¹³.

The primary goal of treating patients with PsA is to maximize health-related quality-of-life through control of symptoms, prevention of structural damage, normalization of function, and social participation, according to the recommendations by the European League against Rheumatism (EULAR)^14,15. Similar treatment goals have also been established by the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA)^16,17. Furthermore, the 2014 treatment recommendations of the Canadian Rheumatology Association/Spondyloarthritis Research Consortium of Canada (CRA/SPARCC) state that the therapeutic objective is “remission or minimal disease activity (MDA) using a treat-to-target approach”^18,19.

Conventional treatments for PsA include NSAIDs and synthetic DMARDs (e.g. methotrexate, sulfasalazine, leflunomide, cyclosporine), which are recommended as first-line agents by most of the mentioned treatment recommendations. The treatment paradigm of PsA has changed significantly since the introduction of biologic agents. Biologics (also known as bDMARDs) are currently recommended as second-line treatment for patients inadequately controlled by conventional treatments^4,20. These biologics include Tumor Necrosis Factor Inhibitors (TNFi) (adalimumab, certolizumab pegol, etanercept, golimumab, infliximab, or their biosimilars), and new emerging treatments like the interleukin (IL)-17A inhibitor secukinumab and the IL-12/IL-23 inhibitor ustekinumab.

The most recent approved biologic, secukinumab, is a high-affinity fully human monoclonal antibody with a novel mechanism of action by neutralizing IL-17A, which has been demonstrated to play a key role in the pathogenesis of PsA^4,21. Secukinumab is approved by Health Canada for the treatment of active PsA²². The recommended dose is 150 mg by subcutaneous injection, with initial dosing at Weeks 0, 1, 2 and 3, followed by monthly maintenance dosing starting at Week 4 for patients who had not been previously treated with biologics. For PsA patients with inadequate response to previous TNFi treatment or with concomitant moderate-to-severe plaque psoriasis, the recommended dose is 300 mg.

Two pivotal phase 3 trials (FUTURE 1 and 2) reported fast and sustained efficacy with a favorable safety profile of secukinumab in patients with PsA, up to 3 years^23–26. Biologic naive and biologic-experienced patients were included in these trials. Biologic-experienced patients represented approximately one third of the total trial populations. Those patients had an inadequate response, safety, or tolerability issues with up to three previous TNFi therapies, regardless of primary (initial non-response) or secondary (loss of response over time) lack of efficacy. In summary, the trial population of FUTURE 1 and FUTURE 2 represent a more difficult to treat patient population than any pivotal TNFi Randomized Clinical Trials (RCTs) have reported.

This analysis reports the results of a cost-effectiveness study on subcutaneously administered secukinumab in comparison with the currently licensed biologics (subcutaneous treatments adalimumab, certolizumab pegol, etanercept, golimumab, ustekinumab, and intravenous treatments infliximab, infliximab biosimilar) over a lifetime horizon (60 years of disease duration) in two patient populations: (i) patients who had not previously been treated with biologics (biologic-naive) and (ii) patients who had previously been treated and failed with biologics (biologic-experienced) in line with the marketing authorization of secukinumab in Canada²². The perspective is that of the Canadian Health Care System. This is the first cost-effectiveness analysis on secukinumab in Canada and the results are discussed in consideration of previous cost-effectiveness studies on different biologic agents used in the treatment of PsA in different countries^27–33.

Methods

Patient population and interventions

The analysis was based on a patient population of 18 years of age and older with active PsA, despite current or previous treatment with NSAIDs, conventional DMARDs, and/or TNFi. The study was conducted in two populations: (i) biologic-naive patients and (ii) biologic-experienced patients. Baseline characteristics of the FUTURE 2 study²⁴ due to its subcutaneous loading phase were considered for various population inputs and are presented in the Supplementary material (Table S1).

The model evaluated the cost-effectiveness of subcutaneous secukinumab compared with the currently licensed biologics (subcutaneous treatments adalimumab, certolizumab pegol, etanercept, golimumab, ustekinumab, and intravenous treatments infliximab, infliximab biosimilar) for the treatment of PsA from the Canadian Health Care System perspective. The dose and dosage frequencies of all biologics used in the analysis are available in the Supplementary material (Table S2). Also, 47% of patients were assumed to receive concomitant methotrexate, and the cost of methotrexate was included for these patients to calculate total costs associated with each treatment. In line with the marketing authorization of secukinumab, 150 mg was applied for the biologic-naive and 300 mg for the biologic-experienced population²².

Model structure

The model was based on a semi-Markov model structure which was developed in Microsoft Excel 2010 (Figure 1)³⁴. The overall structure of the model is similar to the York model^28,30, with few improvements by the inclusion of subsequent-line biological treatment (treatment sequencing) and potential impact on mortality due to severe adverse events (infection and malignancy). The York model has been used and accepted as the structural basis for several economic evaluations of TNFi treatments in PsA^{27,29,31–33}.

Figure 1. Structure of Markov model.

Patients entered the model at the start of treatment. The length of induction period was 3 months, after which response to the treatment was evaluated. Transitions occurred on a 3-month cycle. The primary response criterion in the model was defined by PsA Response Criteria (PsARC)^35,36. The York Model and several publications evaluating the cost-effectiveness of treatments in PsA have also utilized PsARC as a primary response criterion^{27,28,30–33}. The use of PsARC as a response criterion in PsA was also assessed positively by the National Institute of Health and Care Excellence (NICE)³⁷.

Transitions from the “Start Treatment” health state were based on the PsARC response rate, chance of serious infection, malignancy, dropout rate, and death. Possible transitions are represented by the decision tree in Figure 2. All of the grouped states (as indicated by the grey box in Figure 1) could transition to the Malignancy or Death health states. It was further assumed that patients experienced higher risk of mortality for an additional 5 years following an initial malignancy event. Once patients entered the Malignancy health state, they could transition to Death, or remain in the health state to account for the history of malignancy and to model this additional mortality risk.

Figure 2. Decision tree for start treatment and extended treatment health states.

Patients discontinuing the initial biologic treatment were allowed to switch to a subsequent-line biologic treatment. Subsequent-line biologic treatment was modeled using average values of the efficacy, costs, treatment withdrawal rates, and adverse event rates across all biologics in the model. The health states associated with subsequent biologic treatment were the same as those of the initial biologic treatment. Patients were assumed to continue on the subsequent biologic treatment for the remainder of the model time horizon or could drop-out to Standard of Care (SoC), which included methotrexate. Patients entering SoC treatment were assumed to be ineligible to switch back to biologics at any point. The annual discontinuation rate for the subsequent biologic treatment was assumed to be equal to the biologic with the lowest withdrawal rate among all biologics, which allowed patients to remain on biologics for long term.

Model inputs

Clinical inputs

The main clinical inputs for the model were the primary efficacy measure PsARC response at 3 months, the Psoriasis Area and Severity Index (PASI) response distribution at 3 months, and change in the Health Assessment Questionnaire (HAQ) at 3 months.

In the absence of head-to-head comparisons between treatments for clinical measures, comparative effectiveness data for treatments were obtained from a Bayesian network meta-analysis (NMA)³⁸ for the biologic-naive population. The NMA was conducted for the three efficacy outcomes (ACR, PsARC, and PASI). A total of 20 RCTs involving 6,021 adults with active PsA were identified through a systematic literature review to provide efficacy inputs for conducting the NMA. Out of these 20 trials, the evidence base for biologic-naive patients included 12 trials with 3,339 patients, with the PsARC outcome being reported by 6 trials. The evidence base for biologics-experienced patients was very limited. Hence, the clinical efficacy data for the biologic-experienced population were calculated by applying adjustment rates based on the difference in clinical data between biologic-naïve and biologic-experienced populations, as reported in some trials assessing both populations^24,39. Such adjustment rates were applied to account for the fact that biologics are known to have reduced efficacy in the biologic-experienced patients compared to biologic-naive patients⁴⁰. These adjustment rates were obtained from the FUTURE 2 trial for secukinumab²⁴ and from the Canadian ACCLAIM study on adalimumab representative for all TNFi³⁹.

For each modeled comparator, the comparative effectiveness data for the primary efficacy measure (PsARC response) at 3 months are reported in Table 1. The PASI response distribution was evaluated at a 3-month period, which represented the change in PASI from baseline at the time of assessment, and is reported in Table 2. The arthritis component of PsA was modeled as change in HAQ per a 3-month period, and patients experienced a different change in HAQ based on the presence or absence of PsARC response (Table 3). The rates of treatment withdrawal are reported in the Supplementary material (Table S3). For some of the comparators, few clinical input data were not available to include in the NMA calculations. For comparators with missing inputs, the clinical effectiveness data were assumed equal to the comparator with the lowest efficacy in that particular population, to enable comparisons vs all treatments.

Table 1. PsARC response inputs at 3 months.

Administration	Treatment	Biologic-naive	Biologic-experienced
Subcutaneous	Secukinumab 150	59.82%	N/A
	Secukinumab 300	N/A	82.81%
	Adalimumab	62.84%	51.72%
	Certolizumab pegol	51.18%*	67.89%
	Etanercept	70.91%	68.38%
	Golimumab	79.96%	59.90%
	Ustekinumab	51.18%*	46.04%
Intravenous	Infliximab	78.65%	67.73%
	Infliximab biosimilar	78.65%	67.73%

N/A, Not applicable; PsARC, Psoriatic Arthritis Response Criteria.

Data for Biologic-naive population were obtained from a fixed effects network meta-analysis (NMA)³⁸ unless indicated otherwise with *. Data for the biologic-experienced population were obtained via applying adjustment rates to account for the fact that biologics are known to have reduced efficacy in the biologic-experienced patients compared to biologic-naive patients.

* Data was not available. Assumed equivalent to the least responsive biologic for the biologic- naive population in NMA.

Table 2. PASI response distribution at 3 months.

		Biologic-naive				Biologic-experienced
Administration	Treatment	PASI <50	PASI 50–74	PASI 75–89	PASI 90–100	PASI <50	PASI 50–74	PASI 75–89	PASI 90–100
Subcutaneous	Secukinumab 150	25%	23%	23%	30%	N/A	N/A	N/A	N/A
	Secukinumab 300	N/A	N/A	N/A	N/A	20%	17%	23%	40%
	Adalimumab	30%	24%	22%	24%	59%	12%	14%	15%
	Certolizumab pegol	39%*	24%*	19%*	17%*	74%	12%	9%	5%
	Etanercept	39%*	24%*	19%*	17%*	74%	12%	9%	5%
	Golimumab	39%*	24%*	19%*	17%*	64%	13%	13%	11%
	Ustekinumab	39%*	24%*	19%*	17%*	58%	12%	14%	16%
Intravenous	Infliximab	13%	18%	23%	46%	49%	9%	14%	28%
	Infliximab biosimilar	13%	18%	23%	46%	49%	9%	14%	28%

N/A, Not applicable; PASI, Psoriasis Area Severity Index

Data for Biologic-naive population were obtained from a fixed effects network meta-analysis³⁸, unless indicated otherwise with *.

Data for biologic-experienced population were obtained via applying adjustment rates to account for the fact that biologics are known to have reduced efficacy in the biologic-experienced patients compared to biologic-naive patients.

* Data was not available. Assumed equivalent to the least responsive biologic for the biologic-naive population in NMA.

Table 3. Change in HAQ.

		Biologic-naive		Biologic-experienced
Administration	Treatment	PsARC response	No PsARC response	PsARC response	No PsARC response
Subcutaneous	Secukinumab 150	–0.54	–0.23	N/A	N/A
	Secukinumab 300	N/A	N/A	–0.54	–0.33
	Adalimumab	–0.49*	–0.14*	–0.49*	–0.14*
	Certolizumab pegol	–0.40**	0.00**	–0.40**	0.00**
	Etanercept	–0.64*	–0.20*	–0.64*	–0.20*
	Golimumab	–0.44*	–0.06*	–0.44*	–0.06*
	Ustekinumab	–0.40**	0.00**	–0.40**	0.00**
Intravenous	Infliximab	–0.66*	–0.20*	–0.66*	–0.20*
	Infliximab biosimilar	–0.66*	–0.20*	–0.66*	–0.20*

HAQ, Health Assessment Questionnaire; N/A, Not applicable; PsARC, Psoriatic Arthritis Response Criteria.

Data for secukinumab were obtained from the FUTURE 2 trial²⁴.

* Data were assumed equivalent to that of mixed population as reported in Cawson et al.³¹.

** Data were assumed equal to placebo, as reported in the FUTURE 2 trial²⁴.

Cost and resource use

Two main cost categories were incorporated in the model: direct costs, which include drug acquisition costs, disease-related costs, laboratory costs, adverse event costs; and indirect cost due to productivity losses. All costs were converted to Canadian dollars (CAD), if they were not available already in CAD, and inflated using the most recent values (2015) from the general Canadian Consumer Price Index⁴¹.

Drug acquisition costs

Drug acquisition costs for the treatments were obtained from the Ontario drug benefit formulary (Table 4)⁴².

Table 4. Drug acquisition costs.

Administration	Treatment	Dose/unit	Cost per unit
Subcutaneous	Secukinumab	150 mg pre-filled syringe	CAD 822.50
	Secukinumab	300 mg pre-filled syringe (2 × 150 mg pre-filled syringe)	CAD 1,645.00
	Adalimumab	40 mg pre-filled syringe	CAD 769.97
	Certolizumab pegol	200 mg pre-filled syringe	CAD 664.51
	Etanercept	50 mg pre-filled syringe	CAD 405.99
	Golimumab	50 mg pre-filled syringe	CAD 1,555.17
	Ustekinumab	45 mg pre-filled syringe	CAD 4,593.14
Intravenous	Infliximab	100 mg vial	CAD 987.56
	Infliximab biosimilar	100 mg vial	CAD 525.00

CAD, Canadian dollar.

Drugs acquisition costs were obtained from Ontario drug formulary⁴².

Disease-related costs

Disease-related costs consisted of both the arthritis and psoriasis components of PsA (Table 5). Arthritis-related costs were estimated via a linear regression based on HAQ scores, as per the following regression equation^28,43:

Table 5. Disease-related costs.

Input	Mean	Unit
Arthritis-related
Intercept	CAD 476	Per 3 months
Cost per HAQ change	CAD 925	Per 1-unit change per 3 months
Psoriasis-related
Health states
Uncontrolled psoriasis (PASI <75)	CAD 5,056	Per 3 months
Controlled psoriasis (PASI ≥75)	CAD 413	Per 3 months

CAD, Canadian dollar; HAQ, Health Assessment Questionnaire; PASI, Psoriasis Area Severity Index.

Arthritis-related cost from Rodgers et al.²⁸ and Michaud et al.⁴³; psoriasis-related cost from Levy et al.⁴⁴.

Psoriasis-related costs were modeled based on PASI response, incorporating the costs for uncontrolled psoriasis, defined as the proportion of patients not reaching PASI75 response, and the cost for controlled psoriasis (patients reaching PASI75)⁴⁴.

Laboratory costs

The laboratory tests included full blood count, erythrocyte sedimentation rate, liver function test, urea and electrolytes test, chest radiograph, tuberculosis Heaf test, anti-nuclear antibodies, and DNA double-stranded test (Supplementary material, Table S4).

Adverse event costs

The adverse events modeled included serious infection and malignancy. Serious infections included tuberculosis or other serious infections (e.g. septicemia, bronchopneumonia, kidney or unitary tract infection, lower respiratory disease, or bronchitis)²⁸. The distribution of serious infections was assumed to be 10% tuberculosis and 90% other serious infections. Costs associated with tuberculosis, other serious infections, and malignancy are reported in the Supplementary material (Table S5). Incidence of adverse events was taken from the manufacturer’s product monograph when available²², or published literature (Supplementary material, Table S6).

Productivity costs

The model also accounted for indirect costs in the form of productivity loss due to PsA. Data regarding loss of productivity in PsA patients were obtained from a real-world study of rheumatologists and dermatologists, and their PsA patients, across 11 countries⁴⁵. In this study patients were stratified by their Health Assessment Questionnaire Disability Index (HAQ-DI) score into six sub-groups, and respective unemployment rates for each sub-group were reported (HAQ-DI <0.6: 24% unemployment rate; 0.6 ≤ HAQ-DI <1.1: 41%; 1.1 ≤ HAQ-DI <1.6: 45%; 1.6 ≤ HAQ-DI <2.1: 44%; 2.1 ≤ HAQ-DI <2.6: 68%; 2.6 ≤ HAQ-DI <3: 87%). The gross domestic product (GDP) per capita of the full-time employed person (USD 43,248.5, converted to CAD) was obtained from the World Bank data for Canada⁴⁶. This value was multiplied by the unemployment rate to obtain productivity loss due to unemployment for each of the HAQ-DI sub-groups.

Other inputs

Utility weight inputs

For every cycle in the model, the utility values were produced on the basis of the estimated HAQ and PASI scores, using a regression equation. The coefficients applied in the regression equation were obtained from the FUTURE 2 trial data²⁴ for the base case analysis and from Rodgers et al.²⁸ for the alternative scenario (Supplementary material, Table S7).

Mortality inputs

Three types of mortality inputs were considered in the analysis: disease-specific mortality, adverse events-related mortality, and all-cause mortality. Patients with PsA have higher mortality rates compared to the general population. Hence, disease-specific mortality was incorporated as relative risks based on published international literature⁴⁷ (Supplementary material, Table S8). Relative risks of mortality due to adverse events were obtained from the literature⁴⁸ (Supplementary material, Table S8). All-cause mortality was estimated using Canadian sources⁴⁹ and are presented as annual rates (Supplementary material, Table S9).

Base case analysis

The analysis evaluated the cost-effectiveness of secukinumab 150 mg (biologic-naive population) and secukinumab 300 mg (biologic-experienced population) compared to other licensed biologics. The primary effectiveness outcome was quality-adjusted life years (QALYs).

To evaluate the cost-effectiveness, costs and QALYs of secukinumab relative to other modeled biologics were assessed for dominance first (i.e. secukinumab having higher QALYs at a lower cost relative to the comparator). When secukinumab did not dominate the comparator, the incremental cost-effectiveness ratio (ICER) was calculated.

The primary analysis was conducted over a lifetime horizon (60 years disease duration) which allowed for the extrapolation of response benefits due to the chronic nature of the disease. Annual discounting rates were applied in the model based on standard Canadian rates of 5% per annum for future costs and outcomes⁵⁰. The model design, input data, and analysis were assessed for credibility and relevance using the questionnaire from the ISPOR-AMCP-NCP task force report⁵¹.

Sensitivity analysis

Three types of sensitivity analyses were carried out: one-way sensitivity analysis, alternative scenario analysis, and probabilistic sensitivity analysis.

In the one-way sensitivity analysis, the model input parameters like PsARC response rate, HAQ change, laboratory costs, disease-related costs, adverse events costs, and utility weight inputs were varied (one parameter at a time) to identify parameters having the greatest effect on model results (list of all parameters varied and their ranges available in the Supplementary material, Tables S10 and S11).

Alternative scenarios were evaluated for three cases: HAQ rebound assumptions, utility weight inputs, and analysis time horizon. In the first alternative scenario, it was assumed that patients who withdraw from active treatment had HAQ rebound which equals the initial gains rather than the base case assumption that HAQ rebound equals to natural history. In the second alternative scenario, utility weights were obtained from the York model²⁸, whereas, in the base case analysis, they were obtained from the FUTURE 2 trial²⁴. In the third alternative scenario, a 10-year time horizon was considered instead of 60 years in the base case analysis. Further details regarding alternative scenarios can be found in the supplement under the “Alternative scenarios” section.

For the probabilistic sensitivity analysis, input parameters were randomly sampled from their assigned distribution during simulations (Supplementary material, Tables S10 and S11). The results of the probabilistic sensitivity analysis were summarized using cost-effectiveness acceptability curves calculated from the net monetary benefit statistic across a wide range of willingness-to-pay (WTP) thresholds for each treatment option for both populations.

Results

Base case results

Biologic-naive population

Secukinumab 150 mg was compared with the other biologics to assess its cost-effectiveness in the biologic-naive population. In this population, patients treated with secukinumab 150 mg achieved 8.54 QALYs at a cost of CAD 925,387 over a lifetime horizon (Table 6). Secukinumab dominated all other subcutaneous biologics, as it produced more QALYs at the lowest cost among these comparators. Although the intravenously administered biologic infliximab and its biosimilar generated a minimally higher number of QALYs (8.58; +0.04) in this population compared to secukinumab 150 mg, associated costs were higher (infliximab: CAD 1,015,437; infliximab biosimilar: CAD 941,004). This led to higher ICERs for infliximab (CAD 1,991,544/QALY) and its biosimilar (CAD 345,393/QALY) compared to secukinumab 150 mg. The costs and QALYs associated with all treatments are summarized in Table 6.

Table 6. Costs, QALYs, and ICER values for all analyzed populations.

Administration	Treatment	Total costs	QALYs	ICER (Secukinumab 150 mg vs comparator)
Biologic-naive
Subcutaneous	Secukinumab 150 mg	CAD 925,387	8.54
	Adalimumab	CAD 978,913	8.37	Secukinumab dominates adalimumab
	Certolizumab pegol	CAD 980,194	8.26	Secukinumab dominates certolizumab pegol
	Etanercept	CAD 985,925	8.44	Secukinumab dominates etanercept
	Golimumab	CAD 981,653	8.26	Secukinumab dominates golimumab
	Ustekinumab	CAD 985,776	8.30	Secukinumab dominates ustekinumab
Intravenous	Infliximab	CAD 1,015,437	8.58	CAD 1,991,544/QALY*
	Infliximab biosimilar	CAD 941,004	8.58	CAD 345,393/QALY**
Administration	Treatment	Total costs	QALYs	ICER (Secukinumab 300 mg vs comparator)
Biologic-experienced
Subcutaneous	Secukinumab 300 mg	CAD 954,692	8.89
	Adalimumab	CAD 1,003,259	8.38	Secukinumab dominates adalimumab
	Certolizumab pegol	CAD 1,006,976	8.36	Secukinumab dominates certolizumab pegol
	Etanercept	CAD 1,013,756	8.48	Secukinumab dominates etanercept
	Golimumab	CAD 1,002,222	8.29	Secukinumab dominates golimumab
	Ustekinumab	CAD 1,004,749	8.34	Secukinumab dominates ustekinumab
Intravenous	Infliximab	CAD 1,044,195	8.56	Secukinumab dominates infliximab
	Infliximab biosimilar	CAD 979,171	8.56	Secukinumab dominates infliximab biosimilar

* ICER value for Infliximab compared to Secukinumab 150 mg.

** ICER value for Infliximab biosimilar compared to Secukinumab 150 mg.

CAD, Canadian dollar; ICER, Incremental cost-effectiveness ratio; QALY, Quality-adjusted life years.

The total cost comprised of drug costs, disease related costs, productivity costs, and adverse events-related costs (Supplementary material, Table S12).

In summary, secukinumab 150 mg dominated all other subcutaneous biologics, and was cost-effective vs intravenous administered biologics, resulting in the most cost-effective treatment for biologic-naive patients with active PsA in Canada.

Biologic-experienced population

In the biologic-experienced population, patients treated with secukinumab 300 mg achieved the highest number of QALYs (8.89) at the lowest cost of CAD 954,692 compared to all other biologics over a lifetime horizon (Table 6). The costs and QALYs associated with all treatments are reported in Table 6, and the cost-break-up is shown in the Supplementary material (Table S12). As secukinumab 300 mg had the highest number of QALYs at the lowest cost, it dominated all other modeled treatments in the biologic-experienced population.

Sensitivity analysis

One-way sensitivity analysis results showed that the cost-effectiveness outcomes were mainly sensitive to variation in the PsARC response rates, change in HAQ, and utility regression equation coefficients in biologic-naive and biologic-experienced populations (Supplementary material, Figures S1 and S2).

For alternative scenarios, an alternative HAQ rebound assumption (HAQ rebound equals initial gain) demonstrated similar cost-effectiveness outcomes vs the base case analysis (HAQ rebound equals natural history). In this scenario, secukinumab dominated all other subcutaneous biologic treatments and was cost-effective compared to intravenous biologic treatments in the biologic-naive population. Secukinumab dominated all other modeled biologic treatments in the biologic-experienced population (Supplementary material, Table S13). For the other two alternative scenarios (utility weight inputs from York model and 10-years model time horizon), secukinumab was found out to be cost-effective compared to other biologic treatments in the biologic-naive and biologic-experienced populations (Supplementary material, Tables S14 and S15).

Probabilistic sensitivity analysis demonstrated that, at a variety of different WTP thresholds, the probability of being cost-effective was highest for secukinumab in both populations (Figure 3).

Figure 3. Probability of cost-effectiveness of PsA treatments at different willingness-to-pay thresholds for biologic-naive and biologic-experienced populations.

Discussion

This analysis investigated the cost-effectiveness of secukinumab compared to the currently licensed biologics (subcutaneous treatments adalimumab, certolizumab pegol, etanercept, golimumab, ustekinumab and intravenous treatments infliximab, infliximab biosimilar) in patients with active PsA over a lifetime horizon. The analysis was carried out from the perspective of the Canadian Health Care System. This is the first study which evaluated the cost-effectiveness of secukinumab in PsA in Canada.

The cost-effectiveness of secukinumab was evaluated in two different populations stratified by the previous failure of biologics (biologic-naive: secukinumab 150 mg, biologic-experienced: secukinumab 300 mg). These populations were analyzed separately for two reasons: (i) the clinical response to biologic treatments may vary significantly between the biologic-naive and biologic-experienced populations; and (ii) for secukinumab, the recommended dosing is different for the two populations.

In the biologic-naive population, secukinumab 150 mg dominated all modeled subcutaneous biologic treatments. The intravenous administered biologic infliximab and its biosimilar produced minimally higher QALYs (+0.04) compared to secukinumab. However, the costs were significantly higher and, therefore, the ICER values were very high compared to secukinumab (>CAD 300,000/QALY for both original and biosimilar). While no particular WTP threshold is mentioned by the Canadian Agency of Drug and Technologies in Health (CADTH) in Canada, a general acceptability threshold could be around CAD 50,000 to CAD 100,000 per QALY gained, based on which, infliximab and its biosimilar would not be considered cost-effective. Intravenously administered infliximab and its biosimilar also had additional administration costs compared to subcutaneous treatments.

In the biologic-experienced population, secukinumab 300 mg produced more QALYs at a lower cost relative to all biologics. Thus, secukinumab dominated all biologic treatments in patients with active PsA, who failed at least one previous biologic treatment.

The model has also been adapted to other countries. It has been submitted and reviewed by the National Institute for Health and Care Excellence (NICE) and the Swedish Dental and Pharmaceutical Benefits Agency (TLV). Both assessments resulted in positive decisions for secukinumab^37,52. The model structure was validated by a Canadian expert panel in September 2015. The robustness of the analysis was confirmed by one-way sensitivity analysis, scenarios analysis, and probabilistic analysis.

Several other cost-effectiveness studies, which were mainly conducted from a UK payer perspective, investigated the cost-effectiveness of biologics other than secukinumab in PsA (Supplementary material, Table S16). Most of them evaluated TNFi only. Rodgers et al.²⁸ evaluated the effectiveness, safety, tolerability, and cost-effectiveness of adalimumab, etanercept, and infliximab for the treatment of active and progressive PsA in patients with inadequate response to standard treatment. This analysis was based on a Markov structure model, adapted from the York model³⁰. The base-case results suggested that etanercept had the highest probability of being cost-effective at a threshold ICER between £20,000 and £30,000 per QALY gained²⁸. A few other studies also reported that etanercept is a cost-effective option for patients with active PsA. Bojke et al.²⁷ evaluated the cost-effectiveness of adalimumab, etanercept, and infliximab vs palliative care. The base-case analysis suggested that etanercept was the most cost-effective option, with an ICER of approximately £16,000 per QALY gained vs palliative care. Bravo Vergel et al.²⁹ and Woolacott et al.³⁰ suggested that etanercept had the highest probability of being cost-effective at threshold ICERs between £20,000 and £40,000.

Cawson et al.³¹ analyzed the cost-effectiveness of biologics licensed in the UK (adalimumab, etanercept, golimumab, and infliximab). The analysis results suggested that adalimumab, etanercept, and infliximab were cost-effective compared to palliative care, and in the comparison among TNFi, both golimumab and adalimumab were dominated by etanercept, and infliximab was not cost-effective compared to etanercept. A study by Cummins et al.³³ reported that ICERs for etanercept, adalimumab, and infliximab relative to palliative care lie within an accepted range of £20,000–£30,000 per QALY gained.

As none of these above-mentioned published studies have evaluated the cost-effectiveness of secukinumab, the present analysis is adding a significant new body of evidence. Especially as it is comparing a new mode of action (IL-17A inhibition, secukinumab) vs the previous standard (TNF-alpha inhibition). This study revealed that secukinumab (150 mg and 300 mg) is either dominant or the most cost-effective option for the treatment of active PsA compared to other biologics including the previously reported most cost-effective options etanercept, adalimumab, and infliximab in both biologic-naive and biologic-experienced populations from the perspective of the Canadian Healthcare System.

Limitations

The long-term efficacy of secukinumab in this analysis might be under-estimated because only short-term comparative effectiveness data (week 12/16), based on the NMA, were available for the model. However, secukinumab has shown significantly higher and sustained response levels compared to TNFi in Matching Adjusted Indirect Comparisons (MAIC) vs adalimumab, etanercept, and infliximab for up to 1 year^53–56. In addition, secukinumab has also confirmed high and sustained response levels in the long-term clinical trial data²⁶. In the future, cost-effectiveness analysis may be conducted using long-term comparative effectiveness data from those MAIC studies.

Furthermore, the clinical input data for the biologic-experienced population has some limitations. Due to the non-availability of robust RCT data in this population, best available evidence had to be utilized. Hence, adjustment rates were applied to the biologic-naive population clinical data which were obtained from the FUTURE 2 trial for secukinumab²⁴ and from the Canadian ACCLAIM study on adalimumab for all TNFi³⁹. However, the patient populations in the biologic-experienced population might still differ between FUTURE 2 and ACCLAIM with more difficult-to-treat patients in FUTURE 2. Patients in the biologic-experienced group of FUTURE 2 had to have an inadequate response, safety, or tolerability issue with up to three previous TNFi treatments, whereas in ACCLAIM the biologic-experienced population was defined by previous biologic exposure only, and a very limited number of patients reported a history of multiple previous TNFi treatments.

The use of PsARC as a primary response criterion could be considered a general weakness of all cost-effectiveness models in PsA. PsARC was developed initially due to the absence of a PsA-specific response criterion³⁵. However, it may not be used widely in daily clinical practice. In addition, there is an ongoing discussion around the best possible way to measure and define response or remission-like-levels for PsA treatments^57,58. As a response to this discussion, and due to the multifaceted nature of PsA, a variety of composite responder indices assessing a set of core domains of the disease such as skin, joints, pain, and entheseal involvement have been developed; e.g. Minimal Disease Activity (MDA) and Psoriatic Arthritis Disease Activity Score (PASDAS)⁵⁸.

Conclusions

This economic analysis conducted from the Canadian Healthcare system perspective showed that secukinumab dominated all other subcutaneous biologic treatments in the biologic-naive population. Secukinumab was also cost-effective compared to intravenous infliximab and its biosimilar. In the biologic-experienced population, secukinumab dominated all other subcutaneous and intravenous biologic treatments. This analysis suggests that secukinumab is the most cost-effective treatment option for active PsA in Canada for both biologic-naive and biologic-experienced patients, and is a significant advancement in the treatment of active PsA.

Transparency

Declaration of funding

This study was funded by Novartis Pharma AG, Basel, Switzerland.

Declaration of financial/other interests

RG is a consultant for Novartis. DG has received grant/research support from Abbvie, Amgen, BMS, Celgene, Eli Lilly, Janssen, Novartis, Pfizer, and UCB and is a consultant for Abbvie, Amgen, BMS, Celgene, Eli Lilly, Janssen, Novartis, Pfizer, and UCB. SC is an employee of Novartis Pharmaceuticals Canada Inc., Dorval, Canada. PG is an employee of Novartis Product Life Cycle Services-NBS, Novartis Healthcare Private Limited, Hyderabad, India. CNG and LM are employees of RTI Health Solutions, Research Triangle Park, USA. EN is an employee of Novartis Product Lifecycle Services–NBS, Novartis Global Service Center Dublin, Ireland. SMJ is an employee and shareholder of Novartis Pharma AG, Basel, Switzerland. Peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Acknowledgments

The authors thank Niraj Modi, Novartis Healthcare Private Limited, Hyderabad, India for editorial writing support and Doreen McBride and Tina Krieger from RTI Health Solutions for project management support during the model development phase.