Costs of treatment with biologiC and targeted synthetiC disease-modifying anti-rheumatiC drugs in rheumatiC diseases – data from the romanian registry of rheumatiC diseases

Given the limited resources of the health system, rheumatologists are interested in reducing the costs of modern treatments for rheumatoid arthritis, ankylosing spondylitis and psoriatic arthritis, given that the therapeutic targets are obtained and maintained. Data from the Romanian Registry of Rheumatic Diseases (RRBR, in Romanian) from 2016 to 2019 showed for all three diseases that: continuations on the same regimen with tapering experienced a marked increase; the yearly drug cost per patient steadily decreased towards 2019; adalimumab and etanercept originators present the most number of administrations per year and consequently the highest afferent costs in the entire observation period; new drugs (biologic biosimilars and targeted synthetic drugs) are gaining specific portions of the market; switches decrease costs of treatment since the hypothetical models in which switches would not have been performed and patients would have continued their previous treatment throughout the respective year showed 11% increases of costs. RRBR data have shown that reaching and maintaining therapeutic targets (including by switching strategies), reducing risks of adverse events by reducing exposure (tapering) and increasing the use of biosimilar biologics lead to significant cost reductions.


INTRODUCTION
The importance of maintaining costs at a sustainable level for the health system is due to the fact that innovative therapies have greater efficacy, but also much higher costs than traditional treatment, their impact on health systems, even the wealthy ones, being significant. For this reason, the authorities, as well as prescribing physicians, should be interested in controlling the costs for new therapies [1][2][3][4][5][6][7][8][9][10]. The advent of modern therapeutic molecules and availability of data derived from the Romanian Registry of Rheumatic Diseases (RRBR) allowed for cost estimation on a national level, which this study aims to report.
In Romania, patients with rheumatoid arthritis (RA), ankylosing spondylitis (AS) and psoriatic arthritis (PsA) are evaluated by senior attending rheumatologists in a clinical setting (hospital, day care or outpatient clinic) and, if treatment with biologic or targeted synthetic disease-modifying anti-rheumatic drugs (bDMARDs, tsDMARDs) is indicated and the patient fulfils severity criteria issued by the National Health Insurance House (NHIH), their data are uploaded into the RRBR database in the form of a visit, capturing the doctor's indication and posology of b/ tsDMARDs per visit, followed by issuing a reimbursed prescription which is filled by any community pharmacy. RRBR does not record actual data on issued and filled prescriptions, but data reported by each attending physician regarding decisions on bDMARDs indications. Each visit entered in the RRBR database is either an initiation visit (patients naïve to b/tsDMARDs who will receive a specific reimbursed b/tsDMARD), an initial monitoring visit (patients on a b/tsDMARD from non-reimbursed sources, such as completed clinical trials), a contin-Ref: Ro J Rheumatol. 2020;29(1) DOI: 10.37897/RJR.2020. 1.4 uation visit (patients who will continue their previous reimbursed b/tsDMARD) or a switch visit (patients with adverse events, primary or secondary non-responders who will change their previous reimbursed b/tsDMARD). More specifically, the analysis started from the number of patients in the RRBR for which visits lead to b/tsDMARD therapy indications, which were classified into the following cost groups: a) number of initiations (b/tsDMARD-naïve patients starting treatment within the reporting interval); b) number of initiations followed by switch (patients who, after initiation on a specific b/tsD-MARD, switched to another b/tsDMARD at the first re-evaluation because of adverse events or primary non-responder status); c) number of continuations on the same regimen (patients who continue treatment with the same bDMARD, without changing the frequency of administration or dose); d) number of continuations with increased dose or decreased interval of administration (patients who continue the treatment with the same b/tsDMARD, increasing the dose ordecreasing the administration interval); e) number of continuations with reduced dose or increased interval of administration (tapering: patients who continue treatment with the same b/tsDMARD, decreasing the dose or increasing the administration interval); f) number of simple switches (patients who switched their previously continued b/tsDMARD because of adverse events or secondary non-responder status); g) number of multiple switches (patients who, in the course of 12 months, had more than one simple switch).

COST CALCULATION ALGORITHM
For this analysis, initial monitoring visits were considered continuations with or without switch, taking into account that the patient started bDMARD therapy in the past, without being registered in the RRBR. Precautions have been taken to prevent multiple registrations so that a unique patient identifier can only be found in one of the above categories, as well as to avoid incomplete reporting, so that the total number of patients with reported visits during the period of interest to be found in the sum of the categories mentioned above. For cost calculation, the price used for each b/tsDMARD (trade name) was according to CANAMED (the national catalogue of prices for medicines for human use issued with medical prescription, authorized for marketing) in effect at that date and it was expressed in the national currency (lei).
The cost analysis was performed in August 2019 for the 2016-2018 period and in February 2020 for all of 2019, using a calculation model that is described below. The analysis considered all three indications for b/tsDMARDs therapy in rheumatology as indicated (RA, AS and PsA) and it was performed using data from the RRBR, for each calendar year, respectively for the period January 1 st 2016 -December 31 st 2019.
To allow the correct comparison of therapeutic costs of b/tsDMARDs, both for initiations and continuations, as well as between therapies, the cost of b/tsDMARDs was calculated taking into account strictly the period and the interval of administration introduced in the RRBR by the attending physician, for each b/tsDMARD. For the real cost calculation per patient, the cost is calculated taking into account each administration scheme and its duration of administration, stated in the RRBR by the attending physician. Each administration schedule has a calculated price which took into account the price per dose multiplied by the number of doses. Exceptions were intravenous tocilizumab (administrated in doses according to the bodyweight stated in the RRBR) and rituximab (the number of administrations was calculated taking into account the "Certify that, after this visit, the patient will receive rituximab in the next 4 weeks" field from the "Medication Page" of RRBR, where it exists, or taking into account the criteria for re-administration of rituximab according to the national guidelines issued by the NHIH).
The data are taken from the last valid visit of each year, as follows: a) from the "History of b/tsDMARD treatment" page, 1) if there is no price change during that year, -the calculation model for simple regimen only (maintenance doses only) is according to the "Administration regimen" page, namely the number of administrations from the respective year is calculated as the duration of administration in days from "History of b/tsDMARD treatment" page divided by the number of administration days related to the regimen, resulting in the cost for the entire administration period which is calculated as the number of administrations multiplied by the cost of an administration. -the calculation model for composite regimens (loading doses followed by maintenance doses) is according to the "Administration regimen" page, namely the number of loading administrations and the number of maintenance administrations for the reporting year. Each type of administration (loading and maintenance) has a calculated cost taking into account the number of administrations and the cost of an administration. The total cost for that administration period is the sum of the loading cost and the maintenance cost.
2) if there is a change in price during that year, the same above calculation algorithm is used, but the cost is calculated for each period of a b/tsDMARD price. b) from "Recommended b/tsDMARD treatment" page, the duration of administration is calculated on the same model as above, but the end date of administration period is set as December 31 st of the reporting year, and the start date of administration is set as either the date of consultation, if it is a continuation visit, or as the end date, if it is an initiation or switch visit.
For cost calculation, prices for every b/tsD-MARDs listed alphabetically are reported in the national currency (lei) in The above calculation produced the real costs of b/tsDMARDs for each year in the analysed period. Since there have been initiatives to alter the switch mechanism based on the hypothesis that switching b/tsDMARDs increases costs, this study also aimed to test this hypothesis using RRBR data. Therefore, a second hypothetical calculation aimed to estimate costs without switches, namely the overall costs assuming that patients who underwent switches, before the first switch, would have continued throughout the year with their previous b/tsDMARD. For the calculation of this "no-switch" hypothetical cost, the data from "History of b/tsDMARD treatment" RRBR pages were used for the first b/tsDMARD administered in the reference year. The last administration date is December 31 st of the reporting year and the first administration date is the existing date in the "History of b/tsDMARD treatment" RRBR page, or January 1 st if the first administration was performed in the previous year. For patients who started the year directly with a switch, the cost was considered identical to the real cost: in 2019 there were 241 such cases, in 2018, 91 cases, 40 cases in 2017 and78 cases in 2016. The no-switch model also assumed that each patient continued treatment using the last administration regimen of the previous year.

COSTS OF RA TREATMENT WITH b/tsDMARDs
In 2019, compared to the previous years, both the number of RA patients and treatment decisions (initiations, continuations and switches) increased significantly (Table 2)   Concordant with the rise in patient numbers, the costs of initiations and continuations (including tapered regimens) increased towards 2019, while switches maintained a comparable level throughout the analysed period (Table 3). Interestingly, taking into account all treatment decisions from 2016 to 2019, the cost per patient decreased: from a maximum of 36,067 lei/patient in 2017 (approximately 7,895 euros/patient at the average exchange rate in 2017) to a minimum of 33,612 lei/patient in 2019 (approximately 7,083 euros/patient at the average exchange rate in 2019; Figure 2).  (Table 4), Enbrel (etanercept originator) is the leading drug in terms of number of administrations and afferent costs in the entire observation period (Table  4): in 2019, 22.7% of the cohort's administrations were with Enbrel, accounting for 24.6% of total cost, with similar figures in 2018 (27.6% of administrations and 30.3% of total cost), 2017 (29.0% of administrations and 33.3% of total cost) and 2016 (31.6% of administrations and 37.5% of total cost).
The second position, with the same constancy over the 4 investigated years, is Humira (adalimumab originator) which accounted for 20.0% of the 2019 administrations and 20.3% of its total cost, with similar levels in the previous years (24.6% of administrations and 27.6% of total cost in 2018, 24.6% of administrations and 28.7% of total cost in 2017, 25.6% of administrations and 30.6% of total cost in 2016). Together, these two bDMARDs explained 45% of total costs in 2019 and more than half of total costs between 2016 and 2018.
The third position in terms of prevalence and cost was disputed between MabThera (rituximab originator, which accounted for 17.3% of administrations and 8.9% of total cost in 2016, respectively 13.7% of administrations and 6.9% of total cost in 2017) and RoActemra (tocilizumab) which gained weight progressively, reaching 11.4% of administrations and 12.6% of total cost in 2018 and respectively 12.7% of administrations and 16.9% of total cost in 2019. Similar to the rise of RoActemra, but with lower proportions, Simponi (golimumab) recorded an increase from a minimum of 2.7% of administrations and 2.5% of total costs in 2016 to a maximum of 4.3% of administrations and 5.5% of total costs in 2019. Aside from MabThera, there were other molecules which experienced a negative balance during the 4 investigated years, such as Orencia (abatacept, which decreased from a maximum of 5.9% of administrations and 6.5% of total costs in 2017 to a minimum of 3.0% of administrations and 4.1% of total costs in 2019) and Remicade (infliximab originator, which decreased from a maximum of 5.5% of administrations and 5.3% of total costs in 2016 to 2.5% of administrations and 1.4% of total costs in 2019).
While some molecules exhibited a relative stable balance over the 2016-2019 in terms of number of patients and costs (such as Cimzia, Table 4), it was expected that the new drugs (bDMARD biosimilars and tsDMARDs) to gain specific portions of the market.
Thus, Benepali (biosimilar etanercept), since it became available in Romania in 2017 as the only biosimilar etanercept, increased from 0.4% of administrations and 0.1% of total cost in its first year, to 5.6% of administrations and 4.8% of total cost in  In its first year with available biosimilars, already 3.8% of administrations and 1.8% of costs are accounted by adalimumab biosimilars (Hulio, Imraldi, Hyrimoz) compared to the total number of administrations on adalimumab (biosimilar and originator). Also, treatment with tsDMARDs (Olumiant and Xeljanz) increased almost 8-fold from 2018 when reimbursed tsDMARDs became available in Romania and when they accounted for only 1.1% of administrations and 0.4% of total costs, to 2019 when tsDMARDs accounted for 8.0% of administrations and 7.5% of total costs. The hypothetical model in which switches would not have been performed and patients would have continued their previous b/tsDMARD throughout the respective year, before the first switch was made, produces significant results which invalidates the assumption that switches increase costs of treatment: the "no-switch" model would have generated at least 11% more costs each year (Table 5).

COSTS OF AS TREATMENT WITH bDMARDs
In 2019, compared to the previous years, both the number of AS patients and treatment decisions (continuations and switches) increased significantly, while initiations fluctuated around the mean (Table  6). Of note, continuations on tapering regimes experienced a marked increase towards 2019: from 1.  The third position in terms of prevalence and cost was disputed between Remicade (infliximab originator, which accounted for 11.3% of administrations and 12.3% of total cost in 2016, respectively 13.5% of administrations and 17.7% of total cost in 2017) and Simponi (golimumab) which gained weight pro-   costs in 2017, 31.4% of administrations and 28.5% of costs in 2018 and respectively 36.0% of administratons and 34.7% of costs in 2019 ( Figure 6). The hypothetical model in which switches would not have been performed and patients would have continued their previous bDMARD throughout the respective year, before the first switch was made, produces significant results which invalidates the assumption that switches increase costs of treatment: the "no-switch" model would have generated at least 5% more costs for 2016-2018 (Table 9).

COSTS OF PSA TREATMENT WITH bDMARDs
In 2019, compared to the previous years, both the number of PsA patients and treatment decisions (in-itiations and switches) increased significantly, while continuations fluctuated around the mean (Table 10). Of note, continuations on tapering regimes experienced a marked increase towards 2019: from 0.4% of patients in 2016 to 0.5% in 2017, 1.9% in 2018 and 3.7% in 2019 (Figure 7).  of administrations and 13.8% of total cost in 2017) and Simponi (golimumab) which gained weight progressively, reaching 11.1% of administrations and 12.3% of total cost in 2018 and respectively 10.2% of administrations and 13.3% of total cost in 2019. Similar to the rise of Simponi, but with lower proportions, Cosentyx (secukinumab) recorded an increase from a minimum of 0.5% of administrations and 0.3% of total costs in 2017 to a maximum of 8.3% of administrations and 6.4% of total costs in 2019.
As expected, new drugs (bDMARD biosimilars) started to gain significant portions of the market of PsA treatment.
Thus, Benepali (biosimilar etanercept), since it became available in Romania in 2017 as the only biosimilar etanercept, increased from 0.2% of administrations and 0.1% of total cost in its first year, to 4.3% of administrations and 3.6% of total cost in 2019. From the total number of administrations and costs accounted by etanercept, biosimilar and originator, Benepali covered 0.5% of administrations and 0.2% of costs in 2017, increasing 5-fold in 2018 (5.5% of administrations and 3.4% of costs) and doubling in 2019 (11.0% of administrations and 9.1% of costs; Figure 9).
A similar pattern emerged for infliximab: from the total number of administrations and costs accounted by infliximab, biosimilar and originator, biosimilars (Inflectra and Remsima throughout 2016-2019) increased progressively from 2016 (7.7% of administrations and 6.1% of costs), reaching 8.6% of administrations and 7.4% of costs in 2017, 10.3% of administrations and 10.2% of costs in 2018 and respectively 14.3% of administrations and 15.5% of costs in 2019 (Figure 9).
The hypothetical model in which switches would not have been performed and patients would have continued their previous bDMARD throughout the respective year, before the first switch was made, produces significant results which invalidates the assumption that switches increase costs of treatment: the "no-switch" model would have generated at least 8% more costs for 2016-2018 (Table 13).

CONCLUSIONS
Given the limited resources of the health system, rheumatologists are interested in reducing the costs of modern treatments for chronic inflammatory rheumatic diseases, given that the therapeutic targets are obtained and maintained. Interest for therapeutic costs had led physicians to recommend reduction of therapeutic exposure (tapering) in selected cases, a strategy which corresponds to the medical need to reduce risks of adverse events, simultaneously reducing the costs. Switching of b/tsDMARDs for medical reasons (adverse events or failure to reach the therapeutic target with the previous b/tsDMARD) corresponds to the principle of higher priority of efficacy over efficiency, and it actually leads to savings, a fact confirmed by RRBR data, as it prevents wasting resources on medically inefficient drugs. Additionally, cost reduction can be facilitated by stimulating the use of bDMARD biosimilars, a fact confirmed by the presented RRBR data. If the real price difference in Romania of biosimilars compared to the originals would reflect the significant difference between these prices in other health systems (as in Nordic countries), the impact on cost reduction would be even higher. The emergence of new therapies and biosimilar bDMARDs also induced the phenomenon of price erosion of originator bD-MARDs, with contributes to maintaining expenses, lowering the cost per patient and treating more patients as long as the therapeutic target is reached and maintained.