Cost-effectiveness analysis of secukinumab in ankylosing spondylitis from the Canadian perspective

Abstract

Aim: To assess the cost-effectiveness of interleukin (IL)-17A inhibitor secukinumab vs the currently licensed biologic therapies in ankylosing spondylitis (AS) patients from a Canadian healthcare system perspective.

Methods: A decision analytic model (semi-Markov) evaluated the cost-effectiveness of secukinumab 150 mg compared to certolizumab pegol, adalimumab, golimumab, etanercept and etanercept biosimilar, and infliximab and infliximab biosimilar in a biologic-naïve population, over 60 years of time horizon (lifetime). The Bath Ankylosing Spondylitis Disease Activity Index (BASDAI 50) response rate was used to assess treatment response at week 12. Non-responders or patients discontinuing initial-line of biologic therapy were allowed to switch to subsequent-line biologics. Model input parameters (short-term and long-term changes in BASDAI and Bath Ankylosing Spondylitis Functional Index [BASFI], withdrawal rates, adverse events, costs, resource use, utilities, and disutilities) were obtained from clinical trials, published literature, and other Canadian sources. Benefits were expressed as quality-adjusted life years (QALYs). Cost and benefits were discounted with an annual discount rate of 1.5% for all treatments.

Results: In the biologic-naïve population, secukinumab 150 mg dominated all comparators, as patients treated with secukinumab 150 mg achieved the highest QALYs (16.46) at the lowest cost (CAD 533,010) over a lifetime horizon vs comparators. In the deterministic sensitivity analysis, results were most sensitive to changes in baseline BASFI non-responders, BASDAI 50 at 3 months and discount rates. Probabilistic sensitivity analysis showed that secukinumab 150 mg demonstrated higher probability of achieving maximum net monetary benefit vs all comparators at various cost thresholds.

Conclusions: This analysis demonstrates that secukinumab 150 mg is the most cost-effective treatment option for biologic-naïve AS patients compared to certolizumab pegol, adalimumab, golimumab, etanercept and etanercept biosimilar, and infliximab and infliximab biosimilar for a lifetime horizon in Canada. Treatment with secukinumab translates into substantial benefits for patients and the healthcare system.

Keywords:

• Secukinumab
• ankylosing spondylitis
• cost-effectiveness
• Canada
• biologics;quality-adjusted life year
• IL-17A
• anti-TNF

JEL classification codes::

• I10
• I19

Introduction

Ankylosing spondylitis (AS) is an axial spondyloarthritis (axSpA) that primarily affects the spine (debilitating inflammation of the sacroiliac joints). AS also affects peripheral joints, and entheses¹ and extra-articular manifestations are also observed^2–4. Inflammation of sacroiliac joints can lead to progressive worsening pain, fatigue, stiffness, poor health-related quality-of-life, discomfort, and eventual progression to fusion (ankylosis) of the inflamed spinal joints, reducing spinal mobility^5,6.

Globally the prevalence of AS ranges from 10 per 10,000 to 140 per 10,000^7,8, with a prevalence in Canada of 60 per 10,000 in 2009⁹. Clinical, humanistic, and economic burden of AS is significant¹⁰. In 2003, mean annual costs per AS patient was Canadian Dollar (CAD) 9,008 in Canada, with direct healthcare accounting for 29% of these costs¹¹.

As per the Assessment of SpondyloArthritis International Society (ASAS) and European League Against Rheumatism (EULAR), non-steroidal anti-inflammatory drugs (NSAIDs) are the first line treatment for AS, while biologic therapies, such as the anti-tumor necrosis factor (anti-TNFs), like etanercept, infliximab, adalimumab, certolizumab, and golimumab, and the IL-17A inhibitor secukinumab are recommended in patients with persistently high disease activity, despite conventional treatments. If the initial biologic therapy fails, it is recommended to switch to secukinumab (level of evidence 1b) or a second anti-TNF (level of evidence 2)¹². The 2016 updated recommendations also include a new overarching principle that the disease and its treatment are associated with high costs, and the cost of treatment should be considered while deciding on the treatment¹². The Canadian Rheumatology Association (CRA)/Spondyloarthritis Research Consortium of Canada (SPARCC) has similar recommendations for AS treatment¹³.

Secukinumab is the first fully human monoclonal antibody that selectively neutralizes interleukin (IL)-17A. Health Canada approved secukinumab in April 2016 for the treatment of adult patients with active AS who have responded inadequately to conventional therapy¹⁴. Secukinumab is a highly efficacious treatment for AS, demonstrating a rapid onset of action, sustained clinical response, and a good safety profile according to the results of several phase 3 clinical trials^15–18. Secukinumab showed higher improvement in signs & symptoms vs adalimumab in matching-adjusted indirect comparison (MAIC)¹⁹. A head-to-head superiority study (SURPASS) vs adalimumab with radiographic progression as the primary endpoint is underway²⁰.

We report the results of a cost-effectiveness study of secukinumab 150 mg in comparison with the currently licensed biologics (certolizumab pegol, adalimumab, golimumab, etanercept and etanercept biosimilar, infliximab and infliximab biosimilar) over 60 years of time horizon (lifetime) among the biologic-naïve population. The perspective is that of the Canadian healthcare system. This is the first cost-effectiveness analysis on secukinumab for AS in Canada.

Methods

Patient population and interventions

The target population is Canadian adults aged 18 years of age and older who fulfilled the modified New York criteria for AS and exhibited inadequate response to NSAIDs. The base case analysis was conducted in a biologic-naïve population (having no prior exposure to biologics). The population inputs shown in Supplementary Table S1 define the baseline characteristics for the selected model population. Population data are calculated as the weighted average across all patients from the MEASURE 1 and MEASURE 2 pooled trial data²¹.

Details regarding treatment regimens like dosing and frequencies are available in Supplementary Table S2. Biosimilars for etanercept and infliximab are included in the model with the assumption of equivalent efficacy to the originator and different drug acquisition costs.

Model structure

A semi-Markov model was developed using Microsoft Excel, and the model structure can be seen in Figure 1. This structure was chosen based on time-dependent probabilities associated with mortality. Patients start a given treatment, and the response to treatment was assessed using a 50% improvement of the initial Bath Ankylosing Spondylitis Disease Activity Index (BASDAI 50) response at 3 months.

Figure 1. Markov model structure with 3-month induction period.

In the model, patients transitioned to different health states dependent on the probabilities of BASDAI 50 response rate, drop-out rate, serious infection, malignancy, and death. Malignancy and adverse events (AEs) such as infections were modeled as separate health states to better track their associated costs and quality-of-life effects. Patients that experienced infections were allowed to continue biologic treatment or switch to conventional care (CC). It was assumed that all patients discontinued biologic treatment upon entering the “Malignancy” health state. Patients in the malignancy health state had a higher risk of mortality for an additional 2 years following initial malignancy event. Upon entering the malignancy health state, patients could move to death or stay in the health state to track the history of malignancy and to model time-dependent survival probabilities following malignancy.

Patients switched to a subsequent-line biologic treatment upon discontinuing the initial line of treatment. In the absence of relevant data for subsequent-line biologics, average values of the efficacy, costs, and AE incidence rates across all biologics were considered. No switching to biologics was allowed for patients receiving CC.

Model inputs

Clinical inputs

The main clinical input for the model was BASDAI 50 response at 3 months (Supplementary Table S3), which is consistent with the existing British Society for Rheumatology treatment recommendation²², National Institute for Health and Care Excellence (NICE) appraisals for AS^23,24, and the Canadian Agency for Drugs and Technologies in Health (CADTH) drug expert committee recommendation²⁵. BASDAI consists of a 1–10 scale, which is based on five major symptoms of AS including fatigue, spinal pain, joint pain/swelling, enthesitis, morning stiffness duration, and morning stiffness severity²⁶. BASDAI 50 response refers to patients having a 50% improvement in BASDAI score from baseline. This is used to classify patients as responders and non-responders. Additional clinical inputs include short-term changes in the BASDAI and Bath Ankylosing Spondylitis Functional Index (BASFI) responses (Supplementary Table S4), which measures change in BASDAI and BASFI scores at 3 months from baseline; long-term changes in BASFI response captured at 1 year through the modified Stoke Ankylosing Spondylitis Spinal Score (mSASSS) (Supplementary Table S5), and biologic withdrawal rates (Supplementary Table S6). Long term BASFI change over time is modeled to characterize the progression of AS, specifically to model the impact of different clinical processes on BASFI over time. Specifically, the independent effects on BASFI because of disease activity (BASDAI) and the extent and progression of radiographic disease (as measured by the mSASSS) for AS is measured. It allows for consideration of the impact on BASFI that might be achieved via symptomatic improvements (i.e. in terms of reductions in disease activity) and those which might be conferred by disease modification properties (i.e. the effect on the likelihood and/or rate of further radiographic progression).

The comparative effectiveness data for treatments including change in BASDAI and BASFI at 3 months were taken from a Bayesian network meta-analysis (NMA)²⁷. NMAs were conducted in accordance with health technology assessment guidance²⁸. A total of seven trials were identified for biologic-naïve patients involving 1,361 patients. Randomized clinical trial data for the respective drugs were used to inform the NMA²⁷. Wherever the data were missing for any biologic, an average of other biologics was taken.

Cost inputs

Four types of direct costs were incorporated into the model including drug acquisition costs, disease-related costs, medical support costs, and AE costs. The drug acquisition costs for secukinumab and comparators were obtained from the Ontario drug benefit formulary/comparative drug index (Table 1)²⁹.

Table 1. Drug acquisition costs for biologics for AS patients in Canada.

Administration	Drug	Dose (mg)	Cost (CAD)
Subcutaneous	Secukinumab	150	822.50
	Adalimumab	40	769.97
	Certolizumab	200	664.51
	Etanercept	50	405.99
	Etanercept biosimilar	50	255.00
	Golimumab	50	1,555.17
Intravenous	Infliximab	100	987.56
	Infliximab biosimilar	100	525.00

AS, Ankylosing spondylitis; CAD, Canadian dollar; mg, milligrams.

Drug acquisition costs were obtained from the Ontario drug benefit formulary/comparative drug index²⁹.

Disease-related costs included to account for the AS disease management costs are estimated based on an exponential BASFI regression model, where Cost (CAD) = CAD 2,421.51 × EXP (0.213 × BASFI)^30,31.

Medical support costs included costs related to medical visits, as well as laboratory test (Supplementary Table S7).

The model also included costs for treating and managing adverse events. The adverse events considered in the model are serious infection and malignancy as the product labels include caution for these events (Supplementary Table S8).

All costs were converted to CAD, if they were not already available in CAD. All cost inputs were inflated to 2018 using the general Canadian Consumer Price Index³². A sole Canadian healthcare system perspective was considered in this analysis. Hence, all indirect costs, such as productivity losses, were excluded.

Utility inputs

Utility weight inputs were used in the model to calculate quality-adjusted life years (QALYs) to reflect the improvement in health-related quality-of-life experienced by patients who achieved the various levels of BASDAI and BASFI responses. A linear mixed model was used to fit the EQ-5D utility score as a response variable and BASDAI and BASFI response scores, age and sex as predictors (Supplementary Table S9). The model also included disutilities associated with AEs of serious infection and malignancy (Supplementary Table S10).

Mortality inputs

Three mortality inputs were considered in the analysis: age-related all-cause mortality, disease-specific mortality, and AE-related mortality. Higher mortality rates associated with older age were incorporated in the form of all-cause mortality. All-cause mortality was estimated based on the National Life Tables for Canada and are presented in Supplementary Table S11³³. AS patients have a higher mortality rate compared to the general population. The disease-specific mortality risk was based on relative risk values to account for higher mortality in AS patients vs the general population (Supplementary Table S12). In addition, relative risks of mortality due to AEs, such as tuberculosis, other serious infection, or malignancy were obtained from the literature (Supplementary Table S12).

Base case analyses

The analyses assessed the cost-effectiveness of subcutaneous secukinumab 150 mg in a biologic-naïve population compared to other licensed biologics (subcutaneous treatments with certolizumab pegol, adalimumab, golimumab, etanercept and its biosimilar, and intravenous treatments with infliximab and its biosimilar). The primary effectiveness outcome was QALYs.

The primary analysis was conducted over a lifetime horizon due to the chronic nature of the disease (60 years of disease duration considered for the present analysis). Future costs and outcomes were discounted with a discounting rate of 1.5% per annum for all treatments³⁴.

Sensitivity analysis

Sensitivity analyses (deterministic one-way, probabilistic, and alternative scenario analyses) were carried out to assess the robustness of the findings.

In deterministic one-way sensitivity analysis, the model input parameters such as BASDAI 50 response rate, disease-related costs, laboratory costs, AE costs, and utility weight inputs were modified (one at a time) by ±10% to identify sensitive parameters with the largest effect on model results (see lower and upper bound assumptions in Supplementary Table S13).

In probabilistic sensitivity analysis, to estimate the uncertainty around the results, the model was run 1,000 times assuming a certain distribution for each input in the model and a specific cost and QALY was generated for each comparator. These were plotted as a scatter plot and, furthermore, results were presented using cost-effectiveness acceptability curves for each treatment calculated from the net monetary benefit statistic across a wide range of willingness-to-pay thresholds. The net monetary benefit statistic was selected because it allowed for the comparison of all treatment regimens simultaneously (distributions are shown in Supplementary Table S13).

Various alternative scenario analyses were also carried out in the model. This included analyses for a mixed population (a combination of biologic-naïve and biologic-experienced populations), using other published utility values³⁵, excluding disutilities, and assuming an initial gain on BASFI rebound assumption (see Table 2 for alternative assumptions).

Table 2. Summary of ICER results for secukinumab vs biologics by varying base-case assumptions.

Parameters	Base case	Alternative scenarios	Secukinumab 150 mg
Subpopulation	Biologic-naïve	Mixed (a combination of biologic-naïve and biologic-experienced) sub-population	Similar to base case^a
BASFI rebound assumption	Natural history	Initial gain (BASFI deteriorates by the amount of improvement that was achieved during response to therapy)	Similar to base case^a
Disutilities	Not included	Disutilities for adverse events included	Similar to base case^a
Utilities	Measure 1 and 2 trial data	Using McLeod et al.^35	Similar to base case^a

ICER, Incremental cost-effectiveness ratio; BASFI, Bath Ankylosing Spondylitis Functional Index.

^a Secukinumab dominates all biologics.

Results

Base case results

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 the highest number of 16.46 QALYs at a total cost of CAD 533,010. Golimumab achieved the second highest number of 16.13 QALYs at a total cost of CAD 588,053, followed by certolizumab pegol, which achieved 15.78 QALYs at a total cost of CAD 586,732. The total costs and QALYs for all comparators are reported in Table 3. As secukinumab had the highest number of QALYs at the lowest cost among all treatments, it dominated all comparators in this population.

Table 3. Total costs, QALYs for biologic-naïve AS patients over a lifetime horizon (60 years).

Administration	Drug	Total cost (CAD)	QALYs	Secukinumab vs comparator
Subcutaneous	Secukinumab	533,010	16.46
	Adalimumab	594,597	15.61	Secukinumab dominates adalimumab
	Certolizumab pegol	586,732	15.78	Secukinumab dominates certolizumab
	Etanercept	585,755	15.19	Secukinumab dominates etanercept
	Etanercept biosimilar	574,062	15.19	Secukinumab dominates etanercept biosimilar
	Golimumab	588,053	16.13	Secukinumab dominates golimumab
Intravenous	Infliximab	622,244	15.74	Secukinumab dominates infliximab
	Infliximab biosimilar	577,551	15.74	Secukinumab dominates infliximab biosimilar

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

Among all biologics, secukinumab had lower overall drug and other medical costs (consisting of disease related costs, monitoring, and adverse events related costs), leading to lower total costs (Table 4). Lower costs were seen due to lower dropouts beyond year 1 among those on secukinumab (withdrawal rates of 1.6% among those on secukinumab vs 6.2–25.1% among those on comparators beyond year 1) (Supplementary Table S6), which led to less patients on a more expensive subsequent line of treatment. Lower dropouts meant more patients tended to stay in the maintenance treatment phase leading to a higher BASFI score, thus contributing to higher QALYs. In addition, a lower incidence of adverse events like serious infections among patients on secukinumab (low serious infection rates of only 0.16%) (Supplementary Table S8) also ensured lower costs.

Table 4. Costs and QALYs breakdown.

	Secukinumab	Certolizumab pegol	Etanercept	Adalimumab	Infliximab	Golimumab
Costs (CAD)
Biologic drug costs	308,218	340,489	331,221	347,550	373,433	351,304
Disease related costs	213,054	232,497	242,684	233,914	235,363	224,575
Monitoring	4,760	4,597	4,471	4,558	4,522	4,652
Serious infection	6,092	8,119	6,272	6,984	7,414	6,377
Malignancy	886	1,030	1,107	1,592	1,512	1,146
Total	533,010	586,732	585,755	594,597	622,244	588,053
QALYs
Disease-related QALYs	16.462	15.788	15.199	15.614	15.749	16.137
Disutilities	–0.004	–0.005	–0.004	–0.004	–0.005	–0.004
Total	16.458	15.783	15.195	15.610	15.745	16.133

CAD, Canadian dollar; QALY, Quality-adjusted life years.

Sensitivity analysis

In the deterministic one-way sensitivity analysis, model results were most sensitive to BASDAI 50 response at 3 months, baseline BASFI non-responders, and discount rates (Supplementary Figure S1).

Probabilistic sensitivity analyses demonstrated that, at different willingness-to-pay (WTP) thresholds, the probability of being cost-effective was highest for secukinumab 150 mg in the biologic-naïve population among comparators (Figure 2). It was observed that, in nearly 89.4% of simulations, secukinumab exhibited lower costs and higher QALYs compared to other biologics. The scatter plot representing costs and QALYs for all treatment is available in Figure 3.

Figure 2. Probability of net monetary benefit of AS treatments at different willingness-to-pay thresholds. NMB, Net monetary benefit; SEC, secukinumab; ADA, adalimumab; CER P, certolizumab pegol; ETN, etanercept; GOL, golimumab; INF, infliximab.

Figure 3. Cost-effectiveness scatter plot. ADA, adalimumab; CAD, Canadian dollar; CE, cost-effectiveness; CER P, certolizumab pegol; ETN, etanercept; GOL, golimumab; INF, infliximab; SEC, secukinumab; QALYs, quality adjusted life years.

For the alternate scenario analyses, considering the mixed population (combination of biologic-naïve and biologic-experienced) perspective, results were similar to base-case analysis, with secukinumab 150 mg (14.98 QALYs at a total cost of CAD 1,141,452) dominating all other biologics (golimumab achieved the highest number of 14.74 QALYs at a total cost of CAD 1,230,255). Secukinumab 150 mg also dominated all comparators in all other alternative scenarios, reinforcing the robustness of our analysis. A summary of results for all alternate scenarios is presented in Table 2.

Discussion

The present analysis assessed the cost-effectiveness of secukinumab 150 mg vs currently licensed biologics (certolizumab pegol, adalimumab, golimumab, certolizumab pegol, etanercept and etanercept biosimilar, golimumab, infliximab and infliximab biosimilar) for the treatment of AS patients over a lifetime (60 years) in a Canadian healthcare system setting.

Secukinumab 150 mg achieved 16.46 QALYs at a cost of CAD 533,010 over a lifetime horizon in the biologic naïve population. Secukinumab 150 mg produced more QALYs at a lower cost relative to all other comparators. Thus, secukinumab 150mg dominated all comparators. The present model has been submitted and reviewed by NICE and resulted in a positive decision for secukinumab 150 mg in AS³⁶. Different sensitivity analyses confirmed the robustness of the results.

Several studies from different countries have reported the cost-effectiveness of secukinumab vs other biologics for the treatment of AS (Supplementary Table S14). These include studies from Turkey³⁷, Colombia³⁸, the UK^39,40, Bulgaria⁴¹, and Russia⁴². As these published studies were conducted for different countries, results are not directly comparable to our study. However, all these published studies also reported that secukinumab is a cost-effective treatment for AS.

The present analysis provides a new body of evidence, particularly as it assesses the cost-effectiveness of a treatment with a novel mechanism of action (IL-17A inhibition with secukinumab) vs the previous treatment standard (TNF inhibition). The model results suggest that, for physicians, payers, and patients, secukinumab should be the first biologic of choice, as more patients can be treated effectively and cost-efficiently.

Various factors add to the strength of this analysis. In the absence of head-to-head trials, the data for comparative efficacy of other biologics was obtained from a NMA which was conducted using Bayesian NMA techniques, which is scientifically robust²⁷. Inclusion of granular cost details (costs for drug acquisition, medical support, adverse events, as well as disease-related costs) provides a real-world picture of the economic burden on a healthcare system due to AS patients. Uncertainty analysis has been done using probabilistic and deterministic sensitivity analysis, as well as alternative scenario analyses, which strengthens the robustness of the results. Considering the chronic nature of diseases like AS, the base case analysis used a lifetime time horizon (60 years).

While the model reflects the merits of the approach used, the model also presents certain limitations. There might be an under-estimation of long-term efficacy of secukinumab, since the effectiveness data is derived from the NMA, which included data up to week 16²⁷. Secukinumab has shown sustained efficacy in signs, symptoms, physical function, and preventing radiographic damage in AS patients up to 4 years in long-term clinical trial data^43,44, and has also demonstrated higher improvement in signs and symptoms vs adalimumab in Matching Adjusted Indirect Comparisons (MAIC)⁴⁵. Other limitations of this analysis include lack of efficacy data from TNFi in the biologic-experienced population. In the case of missing data, it was assumed that patients are treated with an average biologic based on all other biologics modeled. The societal impact of AS is mainly related to indirect costs. However, these costs, along with costs borne by patients and their families, were not included in the analysis.

Conclusion

Secukinumab dominated (higher QALYs at lower total costs) all modeled biologic treatments (certolizumab pegol, adalimumab, golimumab, certolizumab pegol, etanercept and its biosimilar, golimumab, infliximab and its biosimilar) in the biologic-naïve population of patients with active AS from the Canadian healthcare system perspective. Treatment with secukinumab translates into substantial benefits for AS patients as well as the Canadian healthcare system.

Supplementary material

The supplementary material contains detailed information on population inputs, biologic drug dosing, clinical inputs, other model inputs and sensitivity analyses inputs and outputs.

Transparency

Declaration of financial/other interests

RG is a consultant for Novartis. SCR is an employee of Novartis Pharmaceuticals Canada Inc., Dorval, Canada. PG and MJ are employees of Novartis Product Life Cycle Services-NBS, Novartis Healthcare Private Limited, Hyderabad, India. SMJ is an employee and shareholder of Novartis Pharma AG, Basel, Switzerland. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentation

An abstract titled “Cost-effectiveness analysis of secukinumab in ankylosing spondylitis: a Canadian perspective” was accepted and presented as a poster at the ISPOR 20th Annual European Congress 2017⁴⁶.

Acknowledgments

The authors thank GS Ramakrishna, Novartis Healthcare Private Limited, Hyderabad, India for editorial writing support and RTI Health Solutions for project management support during the model development phase.

Additional information

Funding

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