The model was designed to assess 52 weeks of treatment in a hospital setting (hospital pharmacy). It is based on a decision tree that incorporates each one of the drugs evaluated (adalimumab, etanercept, infliximab, and ustekinumab) and takes into account the possible escalation or de-escalation of the regimen and the need to switch drugs estimated on the basis of the probability of different particular responses at week 24. The probability in each case was calculated using data from a recently published meta-analysis12 (Fig. 1).

Figure 1.

Model Design. PASI indicates Psoriasis Area and Severity Index.

(0.27MB).

The reference document used to create the model was the European consensus document on treatment goals,13 which we adapted to the clinical practice of our country. In this model, patients are classified into 4 groups according to their response to treatment, which is assessed in terms of an improvement in the Psoriasis Area and Severity Index [PASI] of at least 50%, 75% or 90% (PASI 50, PASI 75, and PASI 90, respectively). In patients who achieve a PASI 90 response (excellent responders), the options are to continue the initial regimen or to reduce the dose or frequency of administration of the current biologic agent. Patients with a PASI 75 response who do not achieve a PASI 90 response (responders) continue on the initial regimen. In patients whose response is between PASI 50 and PASI 75 (partial responders), the choices are to continue on the initial regimen, to escalate the regimen (by increasing the dose or frequency of administration), or to switch to a different biologic agent. In patients who do not achieve a PASI 50 response, the options are to escalate the dosage regimen or to switch the patient to another biologic agent.

The distribution of patients according to response (Table 1) was determined on the basis of the absolute rates of efficacy for each biologic drug obtained from the rates of incremental efficacy over placebo published in the meta-analysis of reference12 (Table 2). Absolute efficacy rates were estimated on the basis of an average rate of response to placebo of 10.0%, 3.9%, and 1.1% for a PASI 50, PASI 75, and PASI 90 response, respectively (obtained by imputing the last observation before week 24 of all the patients treated with placebo in the meta-analysis12).

In this study, we analyzed the efficiency of treatment in a hospital setting on the basis of the direct cost (acquisition cost plus the cost of administration, if applicable) of each one of the biologic drugs at the manufacturer's price, as funded by the national health system as of April 2015,14 less the corresponding obligatory deduction15 plus 4% VAT. In calculating the cost of infliximab, we took into account the optimization of vial consumption (using the leftover portion of a vial in the treatment of the following patient) and the cost associated with intravenous infusion (€259.21).16 Annual costs were calculated by prorating the cost of the last dose received to the number of weeks remaining up to and including week 52.

The costs of treatment were calculated on the basis of the dosing regimens recommended in the Summary of Product Characteristics17–20: adalimumab, 80mg induction dose followed by 40 mg every other week starting 1 week after the initial dose; etanercept, 50 mg twice weekly for 12 weeks followed by 50 mg weekly; infliximab, 5 mg/kg administered at weeks 0, 2, and 6, followed by 5 mg/kg every 8 weeks; and ustekinumab, 45 mg or 90 mg at weeks 0 and 4, followed by a dose every 12 weeks. From week 24 onwards, we calculated the costs of treatment according to the dosing regimen that would be used depending on patient response to treatment (Fig. 1).

In the case of the agents that use dosing regimens adjusted to the patient's body weight (ustekinumab and infliximab), we calculated an average cost using the distribution of body weight in the Spanish population with psoriasis treated with biologics. This information was taken from the Spanish Registry of Systemic Therapy in Psoriasis (BIOBADADERM)21: < 60kg (13.2%); 61-80 kg (41.6%); 81-100kg (36.0%); 101-120kg (6.8%); and > 120kg (2.5%).

On the basis of the mean annual cost of treatment and the proportion of patients who continue on the initial biologic agent prescribed, we calculated the cost per successfully treated patient after one year for each of the 4 initial treatment strategies.

The mean annual cost per initial treatment strategy reflects the cost associated with the decision to start treatment with a particular drug and includes the cost of treatment with the biologic drug administered initially plus the cost of subsequent treatment based on the response observed at week 24.

Efficiency can also be evaluated by determining which treatment option offers the best health outcomes (the maximum number of patients treated successfully and the minimum number of patients treated unsuccessfully) for a specific budget. Using that approach, we determined how many patients would continue on the initial drug (successful treatment) and how many would be switched to a different biologic agent (unsuccessful treatment) at 1 year based on a fixed annual budget of €1000000 and using the data on mean annual cost taking into consideration the response at week 24.

Patients who are considered to be excellent responders (those who have a PASI 90 response, that is, an improvement of ≥ 90% over the baseline PASI score), either continue with the initial dosing regimen or change to a reduced regimen at a ratio of 1:1. Responders (those with a PASI 75-90 response, that is, a ≥ 75% but < 90% improvement over baseline PASI), continue on the initial dosing regimen. Partial responders (PASI 50-75 response, that is an improvement of ≥ 50% but < 75% over baseline PASI), continue on the initial treatment regimen, change to an escalated regimen, or switch to another biologic agent at a ratio of 1:1:1. In the case of nonresponders (those who do not achieve a PASI 50 response), the initial dosing regimen is escalated or the patient is switched to another biologic agent at a ratio of 1:1. The biologic agent selected for rescue therapy is assumed to be any of the other 3 available treatments at a ratio of 1:1:1. In the absence of real world data, these ratios are assumed to be representative of the average across different clinical practice settings.

The model assumes that patients on ustekinumab whose body weight is greater than 100kg (9.3%) are treated with 90 mg or 45mg in equal proportions (1:1) because both of these doses are effective for such patients and recommended in the Summary of Product Characteristics.20

De-escalation of treatment in excellent responders involves a reduction in either the dose or the frequency of administration. In this respect, the model is based on the assumptions enumerated below. In the case of adalimumab, frequency of administration is reduced to every 3 (33% reduction in the cost of acquisition) or every 4 weeks (50% reduction) at a ratio of 3:1. In the case of etanercept, the dose is reduced to 25 mg weekly or the frequency to 50 mg every other week (50% reduction in the acquisition cost) at a ratio of 3:1. In the case of etanercept, for which the Summary of Product Characteristics includes the option of intermittent therapy, responders (PASI 75-90 response) and excellent responders (at least a PASI 90 response) in whom therapy is not de-escalated by a reduction in dose or dosing frequency, receive continuous or intermittent treatment at a ratio of 3:1. Because infliximab is potentially immunogenic, many clinicians prefer not to reduce the frequency or dose of this drug. This preference is reflected in the model by the following assumptions: in at least 50% of optimum responders (at least PASI 90), neither the dose nor the interval between doses is modified; in 25%, the interval between doses is changed to 10 weeks (the equivalent of reducing the dose to 4 mg/kg/d if the dosing frequency were maintained); and in the remaining 25%, the interval is changed to 12 weeks (a 33% reduction). In the case of ustekinumab 45 mg, the dosing frequency is reduced to every 18 or every 24 weeks at a ratio of 3:1. With ustekinumab 90 mg, the dose is reduced to 45 mg or the dosing frequency is reduced at a ratio of 1:1; the patients in whom the frequency is reduced are changed to a regimen of every 18 weeks or every 24 weeks at a ratio of 3:1.

Escalation of the treatment regimen implies an increase in dose or frequency of administration. In the case of adalimumab, the model assumes that the prescribing physician either increases the dosing frequency to once a week or changes the dosage to 80 mg weekly at a ratio of 3:1. With etanercept, the frequency is increased to 50 mg twice weekly. With infliximab, the frequency is increased to every 6 or every 4 weeks at a ratio of 3:1. With ustekinumab 45 mg, the dose is increased to 90 mg or the frequency is increased at a ratio of 1:3; and the increase in frequency is achieved by administration every 10 weeks or every 8 weeks at a ratio of 3:1. With ustekinumab 90 mg, the frequency is increased to every 10 weeks or every 8 weeks at a ratio of 3:1.

The base case model assumes that there is no washout period when a patient is switched to a different biologic drug (when the initial treatment is ineffective), and that the first dose of the new drug is administered on the day the next dose of the initial treatment would have been administered.

To determine the consistency of the model, a number of deterministic sensitivity analyses were carried out on the most important parameters.

The first of these analyses presupposed a washout period of 4 half-lives when rescue therapy with any one of the other 3 biologic agents is implemented.

In the second sensitivity analysis, treatments with a dosing frequency in multiples of 4 weeks were changed to multiples of months9,22,23 (infliximab every 2 months and ustekinumab every 3 months, instead of every 8 and every 12 weeks, respectively).

In the third sensitivity analysis, we varied the proportion of patients with a body weight greater than 100 kg treated with ustekinumab 45 mg and 90 mg, applying 2 limit scenarios: all patients treated with 90 mg or all patients treated with 45 mg.

A fourth sensitivity analysis assumed the use of an induction dose (50 mg twice weekly for 12 weeks) in only 50% or in 0% of the patients treated with etanercept.

Since the base values used in the model for the percentages of escalation and de-escalation were arbitrary, several sensitivity analyses were conducted to explore the impact of variations in how the dosing regimen is modified according to initial patient response. In the case of excellent responders, 2 limit scenarios were analyzed: reduction in all cases and maintenance of the initial dosing regimen in all cases. The alternative scenario in partial responders was escalation of the dosing regimen in all cases, and in non-responders it was rescue therapy with another biologic agent in all cases.

To add more information and to prevent the model from becoming obsolete in the event of pricing changes, we incorporated two final scenarios, which contemplate a 5% and a 10% reduction in the prices of adalimumab and etanercept. Since the prices of both infliximab and ustekinumab have recently changed, no price variation was considered for those 2 treatments.

The mean cost for 1 year of treatment according to the model, was lowest for patients who started treatment with ustekinumab (€13873), followed by infliximab (€14234), adalimumab (€14262), and etanercept (€16286) (Fig. 2). However, the percentage differences with respect to the most cost-effective treatment were minimal in the case of infliximab (2.6%) and adalimumab (2.8%). The costs per initial treatment strategy in the proposed model differ from the mean theoretical cost calculated solely on the basis of the dosing regimen recommended in the Summary of Product Characteristics (which does not consider possible dose reductions in optimal responders or the need for escalated treatment or rescue therapy) and from per patient consumption derived from pharmacy records (which does not take into account that part of the consumption of each drug corresponds to rescue therapy for patients in whom treatment with another biological agent has failed) (Fig. 3).

Figure 2.

Average cost per patient at the end of the first year of treatment according to the biologic therapy initially prescribed. The colored portion of the bar represents the cost for patients who complete the year on the same therapy; the orange portion represents the cost of rescue therapy with a different biologic agent.

(0.08MB).

Figure 3.

Comparison of cost of treatment over a 1-year period using different methods. SPC indicates Summary of Product Characteristics.

(0.14MB).

In the present model, the mean cost per patient treated successfully at the end of 1 year is lowest in patients who start treatment with ustekinumab (€15209), followed by infliximab (€16136), adalimumab (€16820), and etanercept (€20178) (Fig. 4.)

Figure 4.

Comparison by initial treatment of the cost-effectiveness and efficiency of 1 year of biologic therapy. (With respect to efficiency, the colored bars represent patients treated successfully and the orange bars represent the patients in whom treatment was unsuccessful).

(0.16MB).

According to the present model, for a fixed budget of €1000000, the number of patients treated successfully (those who continue with the initial biologic agent) at the end of 1 year is greatest for the treatment regimen starting with ustekinumab (66), followed by infliximab (62), adalimumab (59), and etanercept (50). For the same budget, the number of patients treated without success (those who must be switched to rescue therapy with another biologic agent) after 1 year is lowest for the treatment regimen starting with ustekinumab (6), followed by infliximab (8), adalimumab (11), and etanercept (12) (Fig. 4).

The results of the sensitivity analyses (Table 3) show that ustekinumab presents the lowest mean annualized cost per initial treatment in all but the following cases: a 90 mg dose for all patients weighing over 100 kg (in that scenario, infliximab is the lowest cost option); no additional cost for induction therapy with etanercept (in which case etanercept would be the lowest cost option); rescue therapy with another biologic agent for all nonresponders (no escalation of the initial regimen); and a 5% or a 10% reduction in the price of adalimumab and etanercept. In the last 3 cases, adalimumab would become the dominant alternative, with the lowest annual cost of treatment.

Ustekinumab would also have the lowest mean annualized cost per patient treated successfully in all the sensitivity analysis scenarios except when induction with etanercept does not involve any additional cost, in which case etanercept would be the lowest cost option (Table 3). On the basis of a fixed budget, ustekinumab (followed by infliximab, adalimumab, and etanercept) would achieve successful treatment in the largest number of patients and would be associated with the lowest number of patients experiencing primary treatment failure in most of the scenarios evaluated (Table 4).

The authors declare that no experiments were performed on humans or animals during the course of this study.

The authors declare that no private patient data are disclosed in this article.

The authors declare that no private patient data are disclosed in this article.