Summary of the Sponsor’s Pharmacoeconomic Submission

The sponsor submitted a cost-utility analysis comparing tildrakizumab with the following biologic therapies reimbursed in Canada for moderate-to-severe plaque psoriasis: adalimumab, brodalumab, etanercept SEB, guselkumab, infliximab SEB, ixekizumab, risankizumab, secukinumab, and ustekinumab. The model was from the perspective of a Canadian publicly funded health care payer over a 10-year time horizon. An annual discount rate of 1.5% was applied to both costs and benefits. The target population was adult patients with moderate-to-severe psoriasis (a PASI of 8 and patches on 3% to 10% of body surface area according to the Canadian Dermatology Association guidelines). The model baseline characteristics were based on the reSURFACE clinical trials.³

The economic analysis was conducted using a Markov model where costs and benefits were assessed using 2-week cycles. The model was developed in Microsoft Excel and was an adaptation of the Markov York model originally developed to evaluate the cost-effectiveness of etanercept and infliximab for the treatment of psoriatic arthritis.⁵ The model had 2 time periods: the induction period (from treatment initiation up to the initial assessment of treatment response, i.e., 10 to 16 weeks) and the maintenance period (the period following primary response). The model included the following states defined by the PASI response categories: a PASI of less than 50, 50 to 74, 75 to 89, and 90 to 100. At the point of initial assessment (i.e., end of the induction period), patients were in 1 of the aforementioned response categories based on the probability of response to treatment (Table 10 in Appendix 4). Patients who achieved a PASI response score of less than PASI 75 (i.e., the primary outcome in the clinical trials) were switched to second-line treatment, which was assumed to consist of a mix of systemic therapies, including biologics, systemic immunosuppressants, and phototherapy. The sponsor assumed that 50% of the patients receiving second-line treatment were treated with biologics (assumed to represent a mix of all biologic treatments), while the remaining 50% were treated with a combination of methotrexate, cyclosporine, acitretin, and phototherapy, based on expert clinical advice. Patients on second-line treatment who did not achieve a PASI 75 response score after a 12-week induction period were switched to BSC, which comprises topical therapies. Those with a PASI score of 75 or greater could either continue in their existing health state, discontinue therapy, or die (due to all-cause mortality). Upon discontinuation, patients were assumed to receive the next-line therapy (either second-line treatment or third-line BSC). Patients moving to BSC were distributed across PASI health states based on placebo response from the tildrakizumab reSURFACE clinical trials. Patients would either remain in this state or die (due to all-cause mortality). The sponsor assumed that patients who respond to treatment will maintain their PASI score and remain in the same health state (either PASI 75 or PASI 90) until treatment discontinuation or death.

Treatment effectiveness for tildrakizumab was based on the reSURFACE phase III trials,³ whereas comparative efficacy was based on an NMA published by ICER, which assessed treatment response rates in terms of PASI 50, PASI 75, and PASI 90 for the rest of the biologic therapies.⁴ Second-line treatment was assumed to be 10% less effective than the average of first-line treatments. The sponsor implemented this assumption by increasing by 5% the probability of achieving a PASI response of less than 50 or a PASI response of 50 to 74, while decreasing by 5% the probability of achieving a PASI 75 or PASI 90 response.

Patients could discontinue treatment due to lack of efficacy. The model did not account for adverse events associated with treatments. Discontinuation rates during and after the first year of treatment were based on the 2018 ICER report.⁴ The probabilities of treatment discontinuation during the first year of therapy were based on a study that reported on US claims data from 2007 to 2012,⁶ while treatment discontinuation after the first year of therapy was based on results from the Biological Treatment in Danish Dermatology (DERMBIO) registry of patients receiving ustekinumab and secukinumab for the treatment of psoriasis⁷ (Table 11).

Health state utilities corresponding to PASI response scores were estimated using an additive approach in which an incremental value associated with each response category was added to a baseline utility value. The baseline utility was based on a systematic review of health utilities across conditions such as asthma, cancer, chronic disease, diabetes, and skin disease (including psoriasis).⁸ The systematic review identified 3 clinical trials that included Canadian patients and compared adalimumab with placebo for the treatment of moderate-to-severe plaque psoriasis. The sponsor estimated the baseline utility as the average of the baseline utilities of patients within the placebo arm of the 3 trials identified in the review. The incremental value associated with each PASI response was sourced from a cost-utility analysis based on the ustekinumab phase III clinical trial.⁹ The incremental values were added to the baseline utility in order to estimate a utility value for each PASI response health state.⁹

Mortality rates were based on all-cause Canadian mortality data.¹⁰ The cost of BSC was calculated as the average of the following topical therapies: fluticasone propionate, amcinonide, and mometasone furoate lotions, whereas the cost of second-line treatment was calculated as the weighted average of the lowest-cost biologic, the cost of phototherapy, and the average cost of immunosuppressants. Administration costs were assumed to be covered by the sponsor’s patient support programs, whereas monitoring costs were obtained from the Ontario Schedule of Benefits. Unit costs of drugs were obtained from the Ontario Drug Benefit Formulary.¹¹

Sponsor’s Base Case

In the base case, the sponsor reported that tildrakizumab was dominated by brodalumab (i.e., brodalumab was associated with lower total costs and higher QALYs). The sequential incremental cost-utility ratios (ICURs) were incorrectly calculated by the sponsor. This error was addressed by CADTH; the correct results are reported in Table 2. Adalimumab, tildrakizumab, secukinumab, and ustekinumab were dominated, whereas infliximab SEB, guselkumab, and ixekizumab were extendedly dominated.

Table 2

Summary of the Results of the Sponsor’s Base Case.

Etanercept SEB had the lowest costs and fewest QALYs followed by brodalumab and then risankizumab, all of which are on the CEF. The ICURs were estimated in the same order: the ICUR for brodalumab compared with etanercept SEB was $86,703, while the ICUR for risankizumab compared with brodalumab was $839,868 (Table 2). The sponsor reported that tildrakizumab was associated with a total cost of $116,234 and 7.589 QALYs over the 10-year time horizon. At a WTP threshold of $50,000 per QALY gained, tildrakizumab had a 0% probability of being cost-effective.

Summary of the Sponsor’s Sensitivity Analyses

The sponsor conducted a range of scenario analyses. Under each scenario, results in terms of costs and QALYs were estimated using probabilistic analyses.

The following scenarios were considered:

• Time horizon set to 1, 5, and 20 years and, additionally, with an induction phase only (16 weeks).
• Discount rates for both costs and benefits set to 0% and 5%.
• Proportion of patients in second-line treatment being treated with biologics set to 25% and 75%.
• Subgroup analysis of treatment-experienced patients.

The sequential ICURs were incorrectly calculated by the sponsor; CADTH calculated sequential ICURs for each scenario from the sponsor’s reported costs and QALYs. Etanercept SEB had the lowest costs and fewest QALYs in all scenarios except when the time horizon was set to the induction phase only (16 weeks).

The results of the sponsor’s scenario analysis led to findings that were similar to the base-case analysis. The CEF included etanercept SEB, brodalumab, and risankizumab in all scenarios except when the time horizon was set to the induction phase only, which resulted in brodalumab being the lowest-cost option and infliximab SEB being the only biologic on the CEF. In the treatment-experienced population, the sequential ICUR for brodalumab compared with etanercept was $84,205, and the ICUR for risankizumab compared with brodalumab was $801,306 (Table 13 of Appendix 4). In the treatment-experienced population, the ICURs for brodalumab and risankizumab ranged between $42,878 to $91,526 and between $752,159 to $1,823,250, respectively.

Limitations of Sponsor’s Submission

• Uncertainty with respect to treatment effectiveness and safety. Tildrakizumab has been compared head to head with placebo and etanercept; however, there are no head-to-head randomized studies comparing tildrakizumab with other biologics. Relative treatment efficacy was informed by the reSURFACE clinical trials³ and by a published NMA from ICER;⁴ however, these estimates may not be reliable, given the limitations identified by the CADTH clinical reviewers. In particular, the clinical reviewers noted that even though the trials reported efficacy outcomes up to week 52 and week 64, conclusions about the comparative efficacy of tildrakizumab could only be drawn from the induction period (12 weeks) due to the design of the studies and, as such, there is significant uncertainty around the long-term clinical effectiveness of tildrakizumab. Furthermore, efficacy outcomes reported after week 12 were reported descriptively based on observed case data, which could potentially inflate the effects of tildrakizumab. Additionally, the NMA report did not include an assessment of inconsistency or statistical heterogeneity (see CADTH Clinical Review Report for further details). The place in therapy of tildrakizumab is uncertain, as currently available biologics, especially the newer drugs (IL-17 and IL-23 inhibitors), provide good efficacy and durable response. Furthermore, according to the clinical expert, IL-17 inhibitors are anticipated to be favoured over IL-23 inhibitors such as tildrakizumab due to the limited data regarding the efficacy of IL-23 inhibitors in psoriatic arthritis.
  Evidence on the long-term comparative effectiveness of tildrakizumab was not available. As a result, the sponsor assumed that the difference in PASI scores between tildrakizumab and the rest of the biologics, and BSC at the end of the induction period, continues for patients remaining on treatment for the rest of the lifetime horizon, i.e., the model did not assess potential waning of the treatment effect of tildrakizumab or any other biologic. This was considered inappropriate by the clinical expert consulted by CADTH, as a reduction in the effectiveness of biologic treatments is expected over time.
  Additionally, the sponsor used a PASI response score of 75 (PASI 75) to measure treatment response during the trial period. However, the clinical expert consulted by CADTH advised that a PASI 75 response is not consistent with how treatment success is measured in clinical practice, as PASI 90 is now the preferred response score to measure treatment success. The use of the PASI 90 response may lead to different conclusions about both absolute and relative efficacy and result in different conclusions about the cost-effectiveness of tildrakizumab. As indicated in the CADTH Clinical Review Report, indirect evidence suggests that tildrakizumab may be less effective in inducing PASI 75 and PASI 90 responses compared with IL-17 inhibitors (ixekizumab, brodalumab, secukinumab), other IL-23 inhibitors (guselkumab, risankizumab), and infliximab, but may be more effective than etanercept (Table 10). However, given the structure of the model, it was not feasible to explore the cost-effectiveness of tildrakizumab using PASI 90 as a measure of response or using alternate assumptions about long-term treatment effects.
• Treatment pathway does not reflect clinical practice. The natural history of the condition was not captured in the model, as the sponsor only modelled PASI response to treatment but did not model disease progression over time. Additionally, the sponsor’s model assumed that patients who discontinue their second-line treatment switch to BSC. However, as per the clinical expert consulted by CADTH, in clinical practice, patients who discontinue or do not respond to treatment would likely receive a higher dose of the same drug or switch to another biologic treatment. Furthermore, in clinical practice, BSC is only used prior to patients being eligible for treatment with a biologic. The clinical expert consulted by CADTH also noted it is unlikely for BSC to be used as the last line of therapy. This assumption has been considered inappropriate in previous submissions to CADTH for psoriasis.¹² Therefore, the treatment pathway in the economic model does not reflect clinical practice.
  The sponsor included the second-line treatment health state; however, this inclusion did not fully address the treatment pathway limitation, as only 50% of the patients on second-line treatment were assumed to receive biologic therapy and the remaining patients received a combination of methotrexate, cyclosporine, acitretin, and phototherapy. CADTH addressed this limitation by assuming that 100% of the patients on second-line treatment would receive biologic therapy; however, CADTH was unable to address the limitation related to BSC as a third-line treatment because of the structural inflexibility of the model.
  Uncertainty in the treatment discontinuation rate. Treatment discontinuation rates in year 1 in the economic model were based on US claims data on patients receiving adalimumab, etanercept, and a very small number receiving ustekinumab.⁶ Discontinuation rates reported in the study used by the sponsor may also be due to coverage changes in the US (e.g., plans no longer supporting a particular psoriasis drug, or patients moving to different plans), as noted in the 2018 ICER report.⁴ There is a lack of evidence for the remaining biologics; therefore, the sponsor assumed a discontinuation rate for infliximab between those of etanercept and adalimumab, and assumed secukinumab, ixekizumab, and brodalumab, risankizumab, guselkumab, and tildrakizumab had the same discontinuation rate as ustekinumab. This is an important assumption, because patients who discontinue second-line treatment are assumed to receive BSC (instead of an active treatment), which is associated with a very low response rate. Since tildrakizumab has the lowest discontinuation rates in the model (16% during the first year and 5% after the first year), this approach favours tildrakizumab. CADTH also noted that previous submissions to CADTH for treatments for psoriasis have used constant discontinuation rates across all treatments (typically 20%).^13–19 The discontinuation rate used for tildrakizumab in the economic model (i.e., 16%) is also lower than the rates reported in the literature for other biologics (range between 19% and 31%). Additionally, the sponsor assumed discontinuation rates would decrease after the first year. Treatment discontinuation after the first year of therapy was based on results from the Danish DERMBIO registry of patients receiving ustekinumab and secukinumab for the treatment of psoriasis. Only evidence for ustekinumab was provided to support the assumption that discontinuation rates would decrease over time; however, ustekinumab discontinuation rates for the first year and the following years are based on different study designs and might not be comparable. No further evidence was provided to support this assumption, and no evidence was provided for the remaining biologics. In line with previous CADTH reviews, CADTH took the conservative approach of using an overall discontinuation rate of 20% for all treatments during and after the first year. However, since the assumption of differential discontinuation rates was considered plausible by the clinical expert, CADTH explored this assumption as a scenario analysis.
• Differential timing of initial assessment of treatment response. The sponsor assumed different time points for the initial assessment of treatment response for different comparators. For first-line treatments, the time to assessment was assumed to be either 10 weeks (infliximab), 12 weeks (tildrakizumab, etanercept, brodalumab, ixekizumab, guselkumab, secukinumab, and ustekinumab) or 16 weeks (adalimumab). At the time of assessment, the cohort was allocated a distribution of PASI scores based on the ICER NMA and the reSURFACE clinical trials; patients were then subject to treatment discontinuation. Thus, the differential timing would likely impact the results of the analysis, as it impacts the duration and benefit of the treatment. Furthermore, the sponsor used a 12-week induction period for tildrakizumab; however, response was evaluated at 28 weeks in the tildrakizumab clinical trials.
  The CADTH reanalysis adopted a consistent time point for the initial assessment of treatment response (16 weeks) for each biologic as per clinical expert advice, and explored a 28-week induction period for tildrakizumab and retained a 16-week period for the rest of the biologics as a scenario analysis.
• Cost of BSC. The cost of BSC used in the economic model is based on the average costs of cortical steroids. However, a Canadian costing study by Levy et al. is available.
  CADTH considered the Levy study to be a more appropriate source of BSC costs, as it estimated the direct costs of plaque psoriasis in a Canadian population taking into account health care provider visits, frequency of prescription and over-the-counter pharmacotherapy, laboratory tests and procedures, hospitalizations, and frequency of non-conventional treatment and management. Patients in the Levy et al.²⁰ study received a mix of phototherapy and pharmacotherapy, including 13% of patients who received a biologic therapy. As biologic therapy has significantly higher costs than topical therapy (average annual drug cost of biologics ranges between $16,023 and $39,080), CADTH excluded pharmacotherapy (topical, systemic, and biologic therapy) and phototherapy costs from the BSC arm in order to be consistent with the BSC efficacy based on placebo response assumption.
• Cost of biologics. Two biosimilars of etanercept became available in Canada,^21,22 but these are currently not approved for the treatment of psoriasis; the indication of these products is for ankylosing spondylitis and rheumatoid arthritis. The sponsor’s base case used the SEB cost for etanercept; however, the branded cost should have been used. Furthermore, the sponsor used outdated prices for several of the biologic treatments (infliximab SEB, ixekizumab, and secukinumab). The branded cost of etanercept as well as the correct price of the biologics (Table 12) were used in the CADTH base case.

CADTH Common Drug Review Reanalyses

The CADTH reanalysis could not address the following limitations: the lack of evidence on the long-term effectiveness of tildrakizumab beyond the clinical trial period, and the structural limitations of the model, which do not correctly reflect current clinical practice such as switching to BSC upon discontinuation of second-line therapy (instead of switching to a different biologic), and the use of PASI 75 as treatment response (instead of PASI 90). CADTH’s reanalysis included the following changes to the sponsor’s base case (see results in Table 3 and Table 14 in Appendix 4):

1. Assumed that 100% of the patients on second-line treatment receive biologic therapy (which represents a mix of all biologic treatments in terms of their efficacy; second-line treatment is assumed to be 10% less effective than the average of first-line treatments).
2. Used a 20% discontinuation rate for all biologics in all years.
3. Applied a consistent time point (16 weeks) for the initial assessment of treatment response for all biologics.
4. Used Levy et al. as a source for BSC costs, excluding pharmacotherapy and phototherapy costs from BSC.
5. Used the branded price for etanercept.
6. Used the correct prices of infliximab SEB, ixekizumab and secukinumab as per Table 4.
7. CADTH base case (1 + 2 + 3 + 4 + 5 + 6).

Table 3

CADTH Base Case.

The following scenario analyses were conducted based on the CADTH base case:

7a.

The subgroup of patients who are treatment experienced.

7b.

A 28-week induction period for tildrakizumab; 16-week induction period for the rest of the biologics.

7c.

The different discontinuation rates as per the sponsor’s base-case analysis.

7d.

An exploration of the alternative utilities from the ICER report, as per the sponsor’s scenario analysis.

7e.

A 20% discontinuation rate during the first year and a 5% discontinuation rate thereafter.

Based on the CADTH sequential reanalysis, adalimumab, tildrakizumab, guselkumab, secukinumab, ixekizumab, and ustekinumab were either dominated or extendedly dominated. Tildrakizumab was dominated by infliximab SEB, i.e., infliximab SEB was associated with lower total costs and higher QALYs. The following 4 treatments were on the CEF: etanercept, infliximab SEB, brodalumab, and risankizumab. Etanercept would be cost-effective if a decision-maker is willing to pay less than $5,459 per QALY, infliximab SEB would be cost-effective if a decision-maker was willing to pay at least $5,459 but less than $43,560 per QALY, brodalumab would be cost-effective if a decision-maker was willing to pay at least $43,560 but less than $879,094 per QALY, and risankizumab would be cost-effective if a decision-maker was willing to pay at least $879,094 per QALY (Table 3). At a WTP threshold of $50,000 per QALY gained, tildrakizumab had a 0% probability of being cost-effective. CADTH conducted a price-reduction analysis based on the CADTH base case. A price reduction of 20% was required for tildrakizumab to be cost-effective at a WTP threshold of $50,000 per QALY (Table 16).

A scenario analysis assuming a 28-week induction period for tildrakizumab resulted in ICURs of $86,745 for brodalumab compared with etanercept and $824,478 for risankizumab compared with brodalumab. Whereas using the ICER report utilities resulted in ICURs of $7,252 for infliximab compared with etanercept, $46,370 for brodalumab compared with infliximab, and $919,597 for risankizumab compared with brodalumab.

An additional scenario analysis on the treatment-experienced population produced ICURs of $7,252 for infliximab compared with etanercept, $46,370 for brodalumab compared with infliximab, and $919,597 for risankizumab compared with brodalumab. Full results of the sequential analysis can be found in Table 14 of Appendix 4. Tildrakizumab was consistently dominated by brodalumab across all scenarios.

Patient Input

Patient input was received from 2 patient groups: the Canadian Organization for Rare Disorders, and a joint submission from the Canadian Association of Psoriasis Patients, the Canadian Psoriasis Network, and the Canadian Skin Patient Alliance.

Patients reported that psoriasis resulted in itchiness and pain in the inflamed skin; additionally, the skin may crack and bleed. Patients also reported the significant impact of psoriasis on their quality of life when the condition is not being effectively treated. Patients experienced feelings of embarrassment, loss of sleep, problems with intimacy, lack of self-confidence, and feelings of depression. Quality of life was included in the economic model by using utility values for health states defined by PASI scores.

Patients described having used several treatments with different levels of response. Based on the survey and interview responses, many patients reported the treatments as ineffective in addressing their key concerns; additionally, respondents stated that treatment worked only for a period of time. However, no consideration was given to the waning of treatment effects in the sponsor’s submission, and the economic analysis did not evaluate active treatment sequences after initial treatment failure. Three patients had experience with tildrakizumab and were unanimous in their opinion that the drug was highly effective and that, although the long-term impact is still unknown, in the shorter term, tolerability was good.

Family members and caregivers of patients with psoriasis often experience challenges. However, this was not reflected in the sponsor’s submission as a societal perspective was not explored.

Issues for Consideration

• According to the clinical expert consulted as part of this CADTH review, there is uncertainty regarding the place in therapy for tildrakizumab in clinical practice. There are a number of comparators available in Canada; if approved, tildrakizumab will be the third IL-23 blocker for the treatment of plaque psoriasis in Canada. Even though there is some dosing advantage over guselkumab, there are no advantages over risankizumab.
• In 2017, 2 biosimilars of etanercept became available in Canada. ^21,22 These are currently not approved for the treatment of psoriasis. Additionally, a novel human tumour necrosis factor alpha (certolizumab pegol) received Health Canada approval for the treatment of moderate-to-severe plaque psoriasis. The potential introduction of these comparators could impact the findings of the economic analysis.

Conclusions

Based on CADTH’s reanalysis, tildrakizumab is not cost-effective at a WTP of $50,000 per QALY; CADTH’s findings are aligned with the sponsor’s results. In the CADTH base case: etanercept was the optimal therapy for moderate-to-severe psoriasis if the decision-maker’s WTP threshold is less than $5,459 per QALY; infliximab SEB was the optimal therapy if the WTP threshold is at least $5,459 but less than $43,560 per QALY; brodalumab was the optimal therapy if the WTP threshold is at least $43,560 but less than $879,094 per QALY; and, risankizumab was the optimal therapy at a WTP threshold of at least $879,094.

It should be noted that some biologic drugs provide better efficacy in terms of response at a lower total cost (e.g., adalimumab, brodalumab, and infliximab have better efficacy than tildrakizumab, at a lower total cost). At least a 20% reduction in the submitted price would be required for tildrakizumab to be cost-effective at a WTP threshold of $50,000 per QALY. It should be noted, however, that there is significant uncertainty around the clinical effectiveness of tildrakizumab. Additionally, the economic model did not allow CADTH to assess the impact of assumptions relating to the waning of treatment effect and the use of alternative treatment sequences in clinical practice. This adds to the uncertainty of the cost-effectiveness of tildrakizumab.