2.1 Pharmacoeconomic Study Design: CCA

This pharmacoeconomic evaluation uses a cost-consequence approach to assess both the costs and the consequences (effects) of natalizumab therapy in patients with RRMS. The CCA separately estimates and evaluates the costs (expressed in €, year 2024 values) and the efficacy of compared treatments, allowing analysis of separately disaggregated cost components and efficacy outcomes. It also provides a comprehensive view, estimating costs and effects from different perspectives: national health service, patient, caregiver, and society.

Specifically, this analysis includes different types of outcomes and costs: direct healthcare and non-healthcare, indirect and intangible. The CCA was conducted by developing a cohort Markov model in Microsoft Excel for the Italian context. The model simulated disease progression in relation to DMTs, as DMTs can reduce disability worsening and relapse occurrence, which influences patient’s health state, quality of life, and disease management costs. In the base-case scenario of the CCA, we compared intravenous and subcutaneous natalizumab with ofatumumab or ocrelizumab, two high-efficacy therapeutic alternatives for patients with RRMS, with a focus on efficacy outcomes and costs. In this analysis, efficacy and the costs of subcutaneous natalizumab were assumed comparable to those with the intravenous formulation and so are not reported [11]. Further, we created an alternative scenario in which we compared intravenous and subcutaneous natalizumab with a focus on other issues related to the time and costs of drug administration, as other costs and efficacy outcomes were assumed comparable between the two formulations. The main aspects of the model are described in the following sections.

2.2 Intervention

The intervention considered in the economic evaluation, in both scenarios, was intravenous natalizumab 300 mg every 4 weeks in the first year of treatment and every 6 weeks from the second year (extended dose) [8, 13].

2.3 Comparators

In the base-case scenario, natalizumab was compared with the other two available high-efficacy RRMS treatments: (1) subcutaneous ofatumumab 20 mg at weeks 0, 1, and 2, followed by every 4 weeks (starting at week 4) and (2) intravenous ocrelizumab 300 mg, followed 2 weeks later by a second intravenous infusion of 300 mg and then a single intravenous infusion of 600 mg every 6 months.

In the alternative scenario, intravenous natalizumab was compared with subcutaneous natalizumab in terms of drug administration time and resource consumption from the perspective of the MS center, the patient, and society.

2.4 Model Description

The analysis was conducted using a cohort Markov model developed in Excel to compare natalizumab with other high-efficacy DMTs for the treatment of RRMS using disaggregated efficacy outcomes and costs in the Italian context. In the base-case scenario, the model simulated disease progression and estimated healthcare outcomes and costs associated with treatment and management of the disease, creating a simulated cohort for all the therapeutic options mentioned (natalizumab, ocrelizumab, and ofatumumab). In the model, the natural history of RRMS was regulated by three main events: disability progression, relapse occurrence, and mortality risk. RRMS is characterized by periods of relapse, which involve an exacerbation of MS symptoms. A relapse is defined as an exacerbation lasting at least 24 h and separated from the previous exacerbation by at least 30 days; most relapses last from a few days to several weeks [22]. In the model (Fig. 1), disease can progress through a series of disability states, which are based on the Expanded Disability Status Scale (EDSS). The EDSS quantifies disability in patients with MS, and it is useful for monitoring changes in the level of disability over time. The EDSS scale represents the level of disability and ranges from 0 to 10, with 10 being death due to MS [23]. At the beginning of the simulation, patients are assigned to a disability level (EDSS score) ranging from 0 to 6.5 according to the AFFIRM trial [24] and assigned to one of the highly effective DMTs included in the analysis (natalizumab, ocrelizumab, or ofatumumab). Over time, patients can progress to higher or lower disability states (according to the EDSS), maintain the same disability level, move to a second highly effective DMT, or experience relapses (as reported in detail in Sects. 2.6.2 and 2.6.3) or death. This model assumed that patients can discontinue treatment due to adverse events or loss of efficacy or when they reach an EDSS level ≥ 7. As such, the model has 18 health states, seven corresponding to EDSS levels (< 7) with a first highly effective DMT, seven with a second highly effective DMT, three EDSS levels (≥ 7) with no treatment, and one state related to death. Moreover, the model assumed that EDSS was the main determinant for evaluating costs and clinical outcomes.

Fig. 1

The structure of the cohort Markov model shows the Expanded Disability Status Scale (EDSS) [21] health states (0–9) through which patients with relapsing–remitting multiple sclerosis (RRMS) could transition, or remain within, during each cycle. EDSS transitions can involve one or more levels of worsening or improvement, based on the annual transition probabilities estimated by Palace et al. [25]. For simplicity, EDSS transitions exceeding one level are not displayed. Transition to death is possible from all health states. At any time, patients treated with a first highly effective disease-modifying therapy (DMT) could discontinue treatment for adverse events (AEs) or loss of efficacy and could move to second highly effective DMT. Treatment is discontinued once an EDSS health state of 7 is reached. EDSS 0 = normal neurological examination, no disability in any functional system; EDSS 1 = no dysfunction, minimal signs; EDSS 2 = minimal dysfunction; EDSS 3 = moderate dysfunction; EDSS 4 = relatively severe dysfunction not preventing ability to work or carry out normal activities of living, excluding sexual function; EDSS 5 = dysfunction severe enough to preclude working with maximal motor function, walking unaided up to several blocks; EDSS 6 = assistance required for walking; EDSS 7 = restricted to wheelchair; EDSS 8 = restricted to a bed but with effective use of arms; EDSS 9 = totally helpless bed patient; EDSS 10 = death. First highly effective DMT = natalizumab, ocrelizumab, or ofatumumab; second highly effective DMT = an hypothetical subsequent highly effective DMT with an efficacy computed as the mean of efficacy of the three alternatives compared

The model adopted a 5-year time horizon and 1-year cycles, and applied a half-cycle correction. The 5-year time horizon was defined in line with previous CCAs conducted in Italy, which adopted a period required from regional and national payers for this type of analysis [25]. In accordance with guidelines from the Italian Medicines Agency [26, 27], a discount rate of 3% was applied.

2.5 Outcomes and Costs

In the base-case scenario, the model estimated different efficacy outcomes to capture in detail the impact of MS and DMTs. The estimated outcomes were relapse frequency, divided into severe (requiring hospitalization) and non-severe, disease progression in terms of EDSS level distribution, percentage of patients with EDSS ≤ 3, EDSS ≥ 6, EDSS 7–9, and overall survival (life-years). The model also estimated intangible outcomes such as quality-adjusted life-years (QALYs), which combine length and quality of life into a single metric, and social outcomes such as days required for informal care (lost leisure time for assistance from relatives/friends), days of work lost, and the percentage of unemployed patients. In the alternative scenario, comparing methods of administering natalizumab, the model assessed the following outcomes: active working time for healthcare professionals (clinician, nurse), infusion chair occupation time, the patient’s time required for administration, the patient’s informal care, and productivity loss for patients, society, and caregivers during the administration day.

Costs assessed in the base-case scenario were divided into direct healthcare costs, direct non-healthcare costs, and indirect costs. Direct healthcare costs included expenses related to pharmacologic treatment purchase and administration and therapy monitoring and costs associated with relapses and disability management. Direct non-healthcare costs included all expenses related to services employed for disease management: community services and investments in equipment and devices to facilitate a patient’s mobility. Indirect costs are related to productivity losses and informal care.

In the alternative scenario, the model included the healthcare direct costs related to administration (cost for healthcare staff, chair occupation, and consumables), the non-healthcare direct costs (costs of transport and formal care or babysitting), and the indirect costs for administration (productivity loss for the patient, society, and the caregiver and loss of time for unpaid activities for patients).

2.6 Model Inputs2.6.1 Population

The simulated cohort had an initial mean age of 36.8 years, and 70% of the subjects were female [24]; subjects in the cohort were distributed across different EDSS levels, using data from the AFFIRM trial: 5% EDSS 0, 29% EDSS 1, 33% EDSS 2, 20% EDSS 3, 9% EDSS 4, 3% EDSS 5, 1% EDSS 6, and 0% each for EDSS 7, 8, and 9 [24]. The AFFIRM trial included 942 patients with RRMS aged 18–50 years selected from 99 centers in Europe, North America, Australia, and New Zealand. Patients must  have had at least one nuclear magnetic resonance image showing lesions consistent with the disease and at least one relapse in the year preceding but not within the last 50 days before the study began.

2.6.2 Natural History of the Disease

The natural history of the disease and its related progression was simulated using annual disability state transition probabilities in patients with RRMS derived from the British Columbia Multiple Sclerosis database [28]. Transition probabilities remained constant over time. The transition matrix with annual transition probabilities among EDSS states is shown in Table S1 in the electronic supplementary material (ESM). The patients could move to higher and lower disability levels with no restrictions. The British Columbia study [28] estimated transition probabilities for RRMS and secondary progressive MS (SPMS) altogether. So, the model assumed that patients moving to SPMS continued to be treated and were subject to the same transition probabilities as patients with RRMS and did not include specific natural history data related to SPMS. The assumption that patients with SPMS continued to be treated is consistent with study populations in the recent ASCLEPIOS I & II and OPERA [29] trials. Furthermore, the transition to SPMS is not always easy to diagnose because many different clinical definitions exist, and it is often diagnosed retrospectively, meaning that patients likely do not discontinue treatment at the time of progression to SPMS.

Natural history annual relapse rates in patients with RRMS and in relation to EDSS level were estimated using data from Patzold et al. [30] and the UK MS Survey [31] (Table 1). The model differentiated between relapse requiring and not requiring hospitalization. The proportion of relapses requiring hospitalization was estimated at 0.018 [31].

Table 1 Clinical parameters

Annual death probabilities by age were estimated using mortality rates of the Italian general population for the year 2021, stratified by age and sex, reported by the National Institute of Statistics [32]. Death probabilities at each age were computed as a weighted average of male and female probabilities; the proportion of females to males was assumed to be constant in each age group. Annual death probabilities were then adjusted by the relative risk of death in the MS population compared with that in the general population (relative risk 1.7), derived from Jick et al. [33].

2.6.3 DMTs Efficacy Data

DMTs modify the natural evolution of the disease, delaying the progression towards the highest EDSS levels and reducing the occurrence of relapses. The impact of DMTs on the natural course of disability and on the occurrence of relapses was modelled independently, as recognized in RRMS simulation models [34]. Clinical efficacy on annualized relapse rate and confirmed disability progression at 24 weeks for the analyzed DMTs were computed using data from a network meta-analysis (NMA) of clinical trials [35]. Specifically, for each DMT, we extrapolated a relapse rate ratio using placebo as a reference to determine relapse reduction associated with each treatment option and a hazard ratio (HR) for disease progression (Table 1). HRs were then applied to the baseline cumulative hazard derived from natural history disability progression probabilities and then transformed to DMT progression probabilities, using the formula by Briggs et al. [36]. In the model, patients with EDSS < 7 could discontinue treatment because of adverse events or loss of efficacy. In this case, a second line of treatment was introduced for patients with EDSS < 7 in the model (who discontinued treatment because of adverse events or loss of efficacy) assuming hypothetical subsequent DMT with efficacy computed as the mean of efficacy of the three alternatives evaluated in the base-case scenario (natalizumab, ocrelizumab, and ofatumumab). This method balanced the effect of all possible secondary therapy options. A sensitivity analysis employing the mean of DMT pairs not included as first-line treatment was conducted to assess the effect of model assumptions on results. Discontinuation probabilities for the three DMTs were populated using data from an NMA [35], whereas we assumed no discontinuation for the hypothetical subsequent DMT (Table 1). Moreover, patients discontinued treatment when they reached an EDSS level ≥ 7, and no subsequent treatments were introduced in this case.

Finally, the model indirectly included adverse events in the treatment discontinuation rate, as previous studies have shown that the impact of these events, including progressive multifocal leukoencephalopathy, constitutes a minimal part of the overall MS burden [37, 38].

2.6.4 Utilities Data

Utilities by EDSS in patients with RRMS were obtained from a study on MS burden [2] using specific data for Italy and applied in the base-case scenario (Table 1). Utilities applying UK tariffs [39] were multiplied by life-years in each disease state (EDSS level) to compute QALYs. Moreover, the model included a disutility equal to 0.18 for patients experiencing one or more relapses, regardless of EDSS level [31]. This value was calculated as the mean difference in utilities between patients without a relapse and those who had experienced a relapse in the past 3 months, after controlling for EDSS level.

2.6.5 Social and Services Use Outcomes

The base-case scenario analysis considered social outcomes, such as the number of days used for informal care, workdays lost, and the percentage of patients unemployed. The parameters (Table 2) for estimating these results were obtained from the Italian study by Battaglia et al. [2].

Table 2 Social parameters per the Expanded Disability Status Scale (EDSS)

In the alternative scenario analysis, when comparing costs and outcomes related to therapy administration between natalizumab variants, the model used data from the EASIER study [9] to evaluate the day time lost by a patient for treatment administration; the productivity loss for the patient, society, and the caregiver; the loss of time for unpaid activities for the patient; and the healthcare direct services: the time needed for healthcare staff (clinician and nurse) and use of healthcare services (chair occupation) during administration. Furthermore, direct healthcare costs relating to transport between home and the MS center (and return) and formal assistance (e.g. babysitters/caregivers) were also considered. The model parameters are reported in Table S2 in the ESM.

2.6.6 Costs

The following direct healthcare costs were considered in the base-case scenario analysis: disease management related to disability, acquisition of pharmacological treatment, drug treatment administration, patient monitoring, and relapses.

Costs for disease management by EDSS were obtained from an analysis conducted in Italian healthcare administrative databases combined with clinical data from MS disease registries [3]. Table 3 illustrates the healthcare cost data attributable to the disease that were entered into the model (base-case scenario).

Table 3 Direct and indirect costs for disease management per the Expanded Disability Status Scale (EDSS)

The model evaluated the costs related to acquiring, monitoring, and administering DMTs. Costs were calculated as an annual cost, based on the ex-factory prices of the individual packs [40] and on discounts for the Italian national health service. Values for each included DMT are reported in Table S3 in the ESM. Costs for natalizumab and ofatumumab are different for the first year of administration compared with subsequent years, depending on the different doses used for treatment.

DMT administration costs were assessed at €9.71 based on the tariff reported by the G.U. 2013 [41], whereas other costs related to treatment management were included as monitoring costs. For subcutaneous ofatumumab, administration costs were considered only for the first three, assuming they were administered under medical supervision to evaluate any systemic and local adverse events.

Annual monitoring costs of the three treatments were estimated with an ad hoc survey in three MS reference centers in Italy (Biogen, data on file). Specifically, the survey concerned clinical and instrumental tests that are usually performed at the centers to monitor the health status of patients in the first, second, third, and subsequent years of treatment and tests performed before starting therapy. The number of annual exams, estimated with the survey, was then multiplied by the unit cost according to the tariff of services reported in the G.U. 2013 [41]. The monitoring costs for treated patients used in the model are shown in Table S3 in the ESM. The costs of the first year also included tests and exams provided before starting the therapy.

Costs of treatment acquisition, administration, and monitoring of the subsequent hypothetical DMT were assumed to be the same as the mean cost of the three DMTs compared in this analysis: natalizumab, ofatumumab, and ocrelizumab. A sensitivity analysis to evaluate the impact of this assumption was conducted.

The cost of relapse (€432.73) was obtained from an analysis conducted in Italian healthcare administrative databases combined with clinical data in MS disease registries [3]. The cost of diagnosis-related group n.13 "Multiple sclerosis and cerebellar ataxia" (€1419.00) [41] was attributed to severe relapses requiring hospitalization.

The direct non-healthcare costs were also considered in the base-case scenario analysis. These are direct non-healthcare costs relating to services used for disease management: (1) community services and (2) investments in equipment and devices to facilitate patient mobility. The annual costs per patient used in the model varied according to the level of EDSS (Table 3) and were taken from the 2017 study by Battaglia et al. [2].

Indirect costs included in the analysis were also obtained from Battaglia et al. [2] and are shown in Table 3.

In the alternative scenario analysis, we compared intravenous and subcutaneous regimens. Healthcare direct costs per administration of intravenous or subcutaneous natalizumab were estimated using time and unit costs reported in the EASIER study [9]: €67/h for clinicians, €27/h for nurses, and €0.30/h for chair occupation. From the same study, we extrapolated and introduced into the model the cost per transport from home to MS center (and return) (€30.68 per administration) and formal care (babysitting/caregiver: €2.38 per intravenous infusion or €1.28 per subcutaneous administration) as direct non-healthcare costs.

Furthermore, data on indirect costs related to natalizumab administration and included only in the alternative scenario analysis were taken from the EASIER study [9]: €13.50/h of patient’s productivity loss, €15.92/h of patient and society productivity loss, €2.89/h of patient’s loss of time for unpaid activities, and €13.16/h of caregiver’s productivity loss.

Inflation adjustments were applied using the Italian inflation rates from the Italian National Institute of Statistics [42].

2.7 Uncertainty Estimation for Model Outcomes and Sensitivity Analysis

A probabilistic sensitivity analysis was implemented to compute confidence intervals (CIs) for outcome and cost estimates. After assigning a distribution and uncertainty range to main input parameters, a Monte Carlo simulation was conducted with repeated sampling sets of all inputs over 1000 simulation runs (Table S4 in the ESM). Mean values and the 2.5 and 97.5 percentiles are reported for each outcome and cost estimated by the model for each DMT.

A sensitivity analysis was conducted to evaluate possible bias related to model assumptions on efficacy and cost parameters of second-line high-efficacy DMTs. Specifically, the sensitivity analysis assumed a hypothetical subsequent DMT with efficacy and costs computed as the mean of efficacy or costs of the two DMT alternatives not used as first-line treatment.

2.8 Model Validation

The validity of the conceptual model was enhanced through consultation with clinical experts in MS. The model structure was reviewed to ensure it accurately represented the natural history of the disease and the clinical pathways. The model structure was also compared against other conceptual models reported in the literature for the evaluation of DMTs in MS.

Input data validation was conducted with a comprehensive literature review, and clinical experts validated the model inputs.

The computerized model was developed using Microsoft Excel and underwent testing by two modeling experts. The model was tested for extreme sets of input parameters, patient flow simulations, results, and uncertainty analysis calculations and for interface assessment.

Finally, model outcomes were defined and revised by clinical experts. The outcomes were also compared with published studies on DMTs for patients with MS. The results were consistent with the available literature, which included findings from both model simulations and empirical data. Sensitivity and scenario analyses were conducted to assess the impact of changing model inputs on the results.