Article Text
PDF
Abstract
This systematic literature review (SLR) regarding the efficacy, duration of use and safety of glucocorticoids (GCs), was performed to inform the 2022 update of the EULAR recommendations for the management of rheumatoid arthritis (RA). Studies on GC efficacy were identified from a separate search on the efficacy of disease-modifying antirheumatic drugs (DMARDs). A combined search was performed for the duration of use and safety of GCs in RA patients. Dose-defined and time-defined GC treatment of any dose and duration (excluding intra-articular GCs) prescribed in combination with other DMARDs were considered. Results are presented descriptively. Two included studies confirmed the efficacy of GC bridging as initial therapy, with equal efficacy after 2 years of initial doses of 30 mg/day compared with 60 mg/day prednisone. Based on a recently performed SLR, in clinical trials most patients starting initial GC bridging are able to stop GCs within 12 (22% patients continued on GCs) to 24 months (10% patients continued on GCs). The safety search included 12 RCTs and 21 observational studies. Well-known safety risks of GC use were confirmed, including an increased risk of osteoporotic fractures, serious infections, diabetes and mortality. Data on cardiovascular outcomes were Inconsistent. Overall, safety risks increased with increasing dose and/or duration, but evidence on which dose is safe was conflicting. In conclusion, this SLR has confirmed the efficacy of GCs in the treatment of RA. In clinical trials, most patients have shown to be able to stop GCs within 12–24 months. Well-known safety risks of GC use have been confirmed, but with heterogeneity between studies.
- Glucocorticoids
- Epidemiology
- Arthritis, Rheumatoid
Statistics from Altmetric.com
Introduction
To inform the task force responsible for the 2022 update of the EULAR recommendations for the management of rheumatoid arthritis (RA), three systematic literature searches (SLRs) were performed.1 2 These included an update of the 2019 SLRs on safety and efficacy of disease-modifying antirheumatic drug (DMARD) treatments in RA, and a third SLR on the efficacy, safety and for the first time also duration of use (ie, the inability to stop) of glucocorticoids (GCs).3 4
GCs have been successfully used for the treatment of RA for decades. Because of their disease-modifying properties and rapid mode of action, they are still frequently used in current treatment strategies of RA.5 6 However, it is also well known that the use of GCs is limited by a dose-dependent and duration-dependent risk of serious side effects.7 Therefore, GCs are mainly used currently as short-term ‘bridging’ therapy, with the aim of rapidly suppressing disease activity until slower-acting DMARDs become clinically effective.5 After this initial bridging phase, it has been recommended to taper and subsequently stop these GCs as rapidly as clinically feasible.6
In the SLRs in the context of the previous 2019 recommendations, GCs were included as part of the SLRs on safety and efficacy of DMARD treatment in RA.3 4 However, since then specific concerns have been raised about the duration of use of GCs, specifically about difficulties to stop GCs after their use as initial bridging therapy, which may lead to a higher risk of (serious) side effects.8 In the most recent update of the RA management guidelines of the American College of Rheumatology (ACR), the use of GCs—also as initial bridging therapy—is now specifically discouraged as it was argued that risks of GCs outweigh their benefits. This incentivised us to also investigate the duration of use (or the inability to stop) of GCs.8
Therefore, the aim of the current SLR is to update the evidence on the efficacy, safety and duration of use of GCs in RA, to inform the task force responsible for the update of the 2022 EULAR recommendations for the management of RA. For efficacy and safety, we focused on the effect newly started GCs (different routes of administration) as initial bridging therapy, rescue therapy and also the long-term effects of GCs in low doses. In addition, studies reporting the likelihood of long-term use of GCs after initial bridging (with or without protocolised tapering) compared with rescue therapy or reviewed. The results of two separate SLRs, on safety and on efficacy of treatment with DMARDs in RA, are published independently.1 2
Methods
The steering committee of the EULAR task force for the 2022 update of the EULAR recommendations for the management of RA developed an SLR protocol covering three GC-related topics: efficacy, duration of use and safety. For each topic, research questions were defined according to the PICO (Population, Intervention, Comparator, Outcomes) principle. The complete list of research questions and PICOs is available in online supplemental file 1.
Supplemental material
The population of interest was adult patients with a clinical diagnosis of RA. Different types of GCs and different routes of administration (of any dose and duration) were considered. GC treatment needed to be dose-defined and time-defined, and prescribed in combination with other DMARDs. Intra-articular GCs were not considered in this SLR.
Efficacy
We included studies investigating the efficacy of newly-started GCs in combination with DMARDs, as initial therapy or during follow-up. We also compared the efficacy of different routes of administration, protocolised versus non-protocolised tapering, and initial GC bridging versus rescue GC therapy in patients that did not start initial GC bridging. In addition, we studied the efficacy of long-term low dose GCs.
The prespecified comparator group varied per research question of interest, but in general included any other (combination of) DMARDs, or DMARD treatment in combination with a different type or dose of GCs. The efficacy outcomes of interest included core set measures, composite measures, physical functioning, other patient-reported outcomes, disease impact and structural damage (full list in online supplemental file 1). We included randomised controlled trials (RCTs) with >50 patients and ≥3 months follow-up. Studies on efficacy had to be published after 1 January 2019, as this study was an update of the previous EULAR SLR on the efficacy of pharmacological treatments in RA which had addressed GCs.3
Duration of use
We included studies investigating the likelihood of long-term continuation or restarting of GCs and the effects of withdrawal after initial GC bridging. Furthermore, we studied the likelihood of long-term use after initial GC bridging compared with rescue GCs, the influence of different routes of administration, and protocolised versus non-protocolised tapering. Recorded outcomes included GC use and dose during follow-up, disease activity, flare rate and DMARD changes after GC tapering. Randomised clinical trials with and without comparator group with >50 patients and ≥12 months of follow-up were included. Observational studies evaluating duration of use were not considered. A recent SLR (search performed on 9 February 2021) studied the likelihood of long-term continuation or restarting of GCs when used as initial bridging therapy in DMARD-naïve RA patients.9 This search was updated, and the studies previously excluded as patients were not DMARD-naïve (which was not required for the current SLR), were reviewed. For the other questions addressing duration of use, no date limit was defined.
Safety
Studies investigating the safety of newly started GCs, as well as studies on chronic low-dose GCs, were included. Furthermore, we included studies comparing the safety of different routes of administration, studies investigating the safety of GCs combined with biological DMARDs (bDMARDs) and studies comparing continued GC use with GCs that are tapered and stopped. The comparator group was adjusted to the research question of interest but in general included any other (combination of) DMARDs, or DMARD treatment in combination with a different GC routine. An extensive list of safety outcomes was defined, including (serious) adverse events ((S)AEs) of specific interest for GC treatment (online supplemental file 1). We included cohort studies and registries, and (long-term extensions of) phases 3 and 4 clinical trials including >30 patients. Studies needed to be published after 1 January 2012, as an update to the previous SLR to inform the task force of the EULAR recommendations on the management of medium to high-dose GC therapy.10
Systematic literature review
An experienced librarian (JWS) conducted a combined literature search for the duration of use and safety of GCs on 14 January 2022, in the databases MEDLINE, Embase, Web of Science and the Cochrane Library. In addition, the EULAR and ACR 2020 and 2021 abstract databases were searched. Preprints of included abstracts were also searched. A detailed search strategy for each database is provided in online supplemental file 2. Studies on GC efficacy were identified from the separately performed search on the efficacy of DMARD treatment in RA.1
Two reviewers (SAB and AK) independently performed the title screening and abstract screening of a random selection of 10% of the included abstracts. In case of a high level of agreement, one reviewer (SAB) screened the remaining abstracts, and performed the full text screening, risk of bias (RoB) assessment and data extraction. Data on the predefined outcomes were extracted using a standardised data extraction form. RoB assessment was performed using the ‘Hayden tool’ for observational studies and the Cochrane Collaboration’s tool for RCTs.11 12 In case of doubt regarding any of the described steps, disagreements were discussed among the reviewers until consensus was reached, and a senior methodologist was consulted whenever necessary (RBML).
Due to a large heterogeneity in included studies and reported outcomes, all results are presented descriptively.
Results
Efficacy
Article selection
Four studies were identified from the separate search on the efficacy of DMARD treatment and the GC search on duration of use and safety.1 Three of these had a high RoB (online supplemental file 4).13–15 One study was terminated early, and therefore, no data were extracted.16 One additional conference abstract was identified, which did not report sufficient data for data extraction.17 Extracted data are reported in online supplemental tables 5.1 and 5.2.
The efficacy of GC bridging as initial therapy
Two studies (both with high RoB) reported on the efficacy of GC bridging as initial therapy.14 15 The CareRA trial was a pragmatic open-label randomised trial with 2 years follow-up.14 Patients were stratified to high-risk or low-risk groups. For high-risk patients receiving COBRA classic (methotrexate (MTX)+sulfasalazine+GC initial dose 60 mg/day), COBRA slim (MTX+GC initial dose 30 mg/day) or COBRA Avant Garde (MTX+leflunomide+GC initial dose 30 mg/day) no significant differences were reported in ΔDAS28 (Disease Activity Score) (mean (SD) improvement −2.6 (1.2) to −2.7 (1.3)), DAS28 <2.6 (65% to 74%) or ΔHAQ (mean (SD) improvement −0.5 (0.7) to −0.7 (0.7)) at 24 months. Radiographic damage progression as measured by the Sharp/van der Heijde (SvdH) score was low in all treatment arms (mean progression between 0.5 (1.3) and 0.9 (1.7) at 2 years). In the second trial, patients started treatment with MTX+hydroxychloroquine, and either prednisone (initial dose 10 mg/day) or placebo. More improvement in ACR20 response, ΔDAS28 and ΔHAQ was shown for patients starting GC compared with placebo at 3 and 6 months, but not at 12 months.15
The efficacy of initial GC bridging compared with GCs used as rescue therapy
Low-risk patients included in the CareRA trial could be randomised to initial GC bridging (COBRA slim, MTX+GC initial dose 30 mg/day), or to MTX tight step-up treatment (no oral GCs allowed).14 At 24 months, the COBRA slim and tight step-up treatment arms reported similar improvements in ΔDAS28 (mean (SD) −2.4 (1.7) and −2.2 (1.9)), DAS28<2.6 (67% and 72%), ΔHAQ (mean (SD) −0.6 (0.8) and −0.5 (0.7)) and low radiographic damage progression (mean progression 0.3 (0.7) and 0.5 (1.3) at 2 years). During follow-up (ie, not only considering the 24-month visit), patients on COBRA slim had lower values of DAS28 (mean (95% CI) difference: 0.37 (0.00 to 0.7)) and were less likely to receive GC injections (19% vs 47%) than patients on tight step-up.
The efficacy of long-term low-dose GC
The efficacy of long-term low-dose GC was investigated in the double-blind, placebo-controlled GLORIA trial in patients with active RA aged ≥65 years during 2 years follow-up.13 Patients were randomised to prednisolone 5 mg/day or placebo, in addition to standard care antirheumatic treatment. Due to the pragmatic design of the trial, the study was assessed as having high RoB. At 3 months, ΔDAS28 improved more in patients on prednisolone compared with placebo (mean (SD) improvement −1.4 (1.1) and −0.7 (1.2), p<0.0001). The ΔHAQ improved in patients on prednisolone, but remained stable in patients on placebo (mean (SD) −0.2 (0.5) and 0.0 (0.4)). Also, patients on prednisolone more often achieved an ACR20 (36% vs 24%), 50 (20% vs 9%) or 70 response (8% vs 1%) and DAS28<2.6 (37% vs 14%), and slightly more often achieved Boolean remission (3.2% vs 0.7%). At 24 months, adjusted analyses showed that DAS28 was significantly lower for patients on prednisolone compared with placebo (2.97 vs 3.33), with similar HAQ (both groups 1.1) and significantly lower SvdH progression (0.3 vs 1.9 points).
One study was included on the efficacy of different routes of administration of GCs, but due to early termination of the study no data were extracted.16 No studies were included on the efficacy of GC bridging during the follow-up, or on the efficacy of GC bridging with a protocolised versus a non-protocolised tapering protocol.
Duration of use
Article selection
The combined search identified 2977 articles (online supplemental figure 3.1). Of these, 10% were double-screened by two readers (SAB and AK) based on title and abstract, resulting in 95% agreement. The remaining abstracts were screened by a single reader (SAB). Two hundred and twenty out of 2977 articles were selected for detailed full-text review. Of these, 14 fulfilled all inclusion criteria. However, 13 of these were included in a recently published SLR.9 Thus, the search identified only one new paper.18 No additional meeting abstracts were identified. RoB assessments were provided in the original paper, and are in online supplemental file 4.9 Baseline characteristics of included studies are provided in online supplemental table 5.1.
The likelihood of long-term continuation or restarting of GCs when used as initial bridging therapy
No additional articles were included on this topic after a recently published SLR.9 In this previous SLR, 137 papers were included from 10 unique clinical trials. One of these studies was assessed as having a low RoB, nine of these studies were assessed as having a high RoB, mostly because they were single-blinded (Cochrane RoB tool 2).9 Four of these studies (10 treatment arms) reported sufficient data to be included in a meta-analysis on GC use at 12 months follow-up, and 2 studies (4 treatment arms) on GC use at 24 months follow-up. In these four studies patients started with a dose of 30 or 60 mg/day of prednisone or prednisolone bridging, which was tapered to 5 or 7.5 mg/day and could be subsequently stopped between 20 and 32 weeks of follow-up in the different studies. The pooled proportions of patients taking GCs was 0.22 (95% CI 0.08 to 0.37) at 12 months and 0.10 (95% CI −0.01 to 0.22) at 24 months (figure 1).9 Insufficient data were reported to compare other outcome measures between studies.
Proportion of patients using GCs at 12 months (A) and at 24 months (B) after initial GC bridging in clinical trials. Reproduced with permission from van Ouwerkerk et al.9 CareRA COBRA ag, CareRA COBRA Avant Garde; CareRA COBRA slim (low), CareRA COBRA slim low-risk group, GC, glucocorticoids; IMPROVED early rem, IMPROVED early remission; PFT, proportions according to the Freeman Tukey transformation.
A follow-up study was performed in which individual patient data were combined from 7 out of 10 trials included in the original meta-analysis, to perform an individual patient data meta-analysis based on more detailed data.19–25 At the time of publication of this SLR these data were yet unpublished, but the results of this effort were presented to the task force responsible for the update of the 2022 EULAR recommendations for the management of RA and partly reproduced here with permission (table 1 and online supplemental tables 5.3 and 5.4). The meta-analysis on individual data from the seven included trials (table 1) shows that at 6 months after the end bridging 7% of patients used GCs. After excluding trials which started with parenteral GCs (i.m. or i.v. GC once), this percentage was slightly higher (12%). At 12–18 months after bridging, these numbers were very similar.
GC status at predefined time points after the initial bridging strategy, based on an individual patient data meta-analysis of seven studies
The effects of withdrawal of GCs after use as bridging therapy
The SEMIRA study (low RoB) was the only included study which assessed the effects of withdrawal of GCs.18 In this placebo-controlled study with 24 weeks follow-up, patients on tocilizumab and 5 mg/day prednisone with stable low disease activity were randomised to either continue or to taper prednisone to stop in 16 weeks. Patients who tapered prednisone had an increase in DAS28 of 0.54 (95% CI 0.35 to 0.73), whereas patients who continued prednisone had a decrease in DAS28 of −0.075 (95% CI −0.27 to 0.21) compared with baseline (p<0.001). The HR of being flare-free at 24 weeks was 2.77 (95% CI 1.47 to 5.25) for continuing versus stopping prednisone.
The likelihood of long-term use of GCs when used as bridging therapy, compared with GCs used as rescue therapy
Low-risk patients included in the CareRA study (high RoB, study described above) were randomised to COBRA slim, which consisted of MTX+prednisone bridging, or to MTX tight step-up, which consisted of MTX monotherapy without oral GC (i.m. GC allowed as rescue).14 The mean cumulative prednisone dose in the first year was higher for COBRA slim (1554 mg (SD 308)) than for tight step-up (36 mg (SD 50)). However, in the second year the cumulative prednisone dose was higher for the tight step-up arm compared with the COBRA slim arm (235 mg (SD 696) vs 151 mg (SD 346)), suggesting that GC use was higher in patients who did not initially receive GC bridging. This was mainly explained by a higher proportion of patients receiving GC injections in the tight-step up arm (COBRA slim: 19%, tight step-up: 47%). The proportion of patients with at least 3 months GC use outside of the induction scheme was similar between the two treatment arms (COBRA slim: 12%, tight step-up 11%), with a slightly higher median daily dose in the tight step-up arm (COBRA slim: 5.4 mg (IQR 2.9), tight step-up: 6.7 mg (IQR 3.3)).
No studies were included on the influence of the route of administration of GCs or on the effects of protocolised versus non-protocolised tapering and stopping of GC bridging.
Safety
Article selection
From the 220 out of 2977 articles that were identified for full-text review, 42 papers fulfilling all selection criteria were included. In addition, three meeting abstracts were identified (online supplemental figure 3.2). From these conference abstracts, one full text of a subsequent full manuscript was available as a preprint.13 The included studies reported on 12 unique RCTs and 21 observational studies. Two of these studies were not reported, 1 due to early termination,16 and another because it was reporting on the same cohort with the same outcomes as another included paper.26 RoB assessments are provided in online supplemental tables 4.1–4.3 (RCTs) (observational studies). Baseline characteristics of the included studies are provided in online supplemental tables 6.2 and 6.3 (RCTs) (observational studies).
Safety of therapy with GCs in RA
The number of included safety studies per outcome category is provided in online supplemental table 6.1. Observational studies were available for seven of the outcome categories (cardiovascular disease and hypertension, osteoporosis, infections, diabetes and hyperglycaemic, mortality, adverse pregnancy outcomes and glaucoma). These were considered the main study type. For outcome categories that are not reported in this table, no studies were included.
Cardiovascular disease and hypertension
Eleven observational studies reported on cardiovascular disease and/or hypertension (table 2, online supplemental tables 6.4 and 6.5).27–37 Outcome definitions varied per included study. Four studies reported on hypertension separately.27 29 34 35 Two studies, both at unclear RoB, reported an increased risk of hypertension.34 35 One of these studies only reported an increased risk for recent GC treatment with a dose ≥7.5 mg/day and for cumulative GC doses between 5000 and 10 000 mg.35 The other study reported an increased risk for all time-varying cumulative GC doses (adjusted HR, aHR 1.11, 1.39 (95% CI 1.03 to 1.48)), but not for different categories of daily GC doses, up to ≥7.5 mg/day.34 The two other studies (both high RoB) reported no increased risk of hypertension for any GC use and for different cumulative GC doses,27 or for daily versus alternate daily GC dosing.29
Cardiovascular disease and hypertension, results of observational studies
One study at high RoB reported no increased risk of a cerebrovascular accident after adjustment for potential confounders.36
The other observational studies reported on cardiovascular events in general.27 28 30–33 37 Two studies (one high, one unclear RoB) reported an increased risk of cardiovascular events for any reported daily or cumulative dose (aHRs 1.02, 4.98 (95% CI 1.00 to 6.03)).31 37 One study (high RoB) reported an increased risk for ever and current GC use and for most reported cumulative doses.27 Three studies (2 unclear, 1 high RoB) only reported an increased risk for higher daily or cumulative doses (aHRs 1.41, 2.05 (95% CI 1.03 to 2.79)).30 32 33 These were a daily dose ≥5 mg, cumulative doses >750 mg in the preceding 6 months or >1100 mg in the preceding year,33 ≥7.5 mg/day for≥3 months,32 or a cumulative dose ≥8421 mg over 10 years.30 One study at unclear RoB reported no increased risk of cardiovascular events for any reported duration or cumulative dose, after adjustment for potential confounders.28
The number of events in the 12 included RCTs was low, and was therefore not further described (online supplemental table 6.6).13 18 21 22 24 38–43
Osteoporosis and osteoporotic fractures
Five observational studies reported on osteoporotic fractures,26 29 30 32 44 and one on incident osteoporosis (table 3, online supplemental tables 6.7 and 6.8).27 Three studies (two high, one unclear RoB) showed an increased risk for osteoporotic fractures.26 32 44 This risk was only observed for higher daily or cumulative doses, although one study (unclear RoB) already reported an increased risk at 2.5 mg/day.26 A second study (high RoB) reported an increased risk for GC use ≥3 months, independent of whether the daily dose was below or above 7.5 mg/day.32 No difference for the risk of osteoporotic fractures was observed in another study, even with high cumulative doses (unclear RoB),30 and between daily versus alternate daily GC dosing (high RoB).29
Osteoporotic fractures and osteoporosis, results of observational studies
One study (high RoB) reported an increased risk of osteoporosis for current GC use, for all reported cumulative dose categories (adjusted ORs, aORs 1.31, 1.56 (95% CI 1.11 to 1.97)).27
Three additional RCTs reporting on osteoporotic fractures or osteoporosis are described in online supplemental table 6.9.40 41 45
Infections
Five observational studies reported on the risk of infections (table 4, online supplemental tables 6.10 and 6.11).27 29 30 46 47 Four of these reported on the risk of serious infections.27 29 30 46 Three studies (two high RoB, one unclear RoB) reported an increased risk of serious infections for (current) GC use.27 30 46 Two of these reported an increased risk with increasing dose.46 47 The two studies investigating cumulative doses showed conflicting results.27 30
Infections, results of observational studies
One study (high RoB) reported no difference in risk of serious infections for alternate daily GC dosing versus daily GC dosing. However, the risk of any infection was lower with alternate daily GC dosing (aOR 0.27 (95% CI 0.12 to 0.63)).29 Also, the risk of an infection occurring during an acute care hospitalisation was increased with increasing GC doses (one study at high RoB).47
Results of 11 additional RCTs reporting on infections, serious infections, tuberculosis or herpes zoster are described in online supplemental table 6.12.13 14 18 21–24 40 48–50
Diabetes and hyperglycaemia
Four observational studies reported on the risk of incident diabetes (table 5, online supplemental tables 6.13 and 6.14).27 29 51 52 Three studies (two unclear RoB, one high RoB) reported an increased risk for ever or current GC use, with an increased risk for higher daily and for higher cumulative doses.27 51 52 One of these studies (unclear RoB) reported no increased risk for daily doses <5 mg/d,51 whereas a second study (unclear RoB) did report an increased risk of incident diabetes for the lowest daily dose category (<5 mg/day).52
Diabetes, outcomes of observational studies
One study (high RoB) reported no difference in the risk of diabetes for daily versus alternate daily dosing.29
Results of five RCTs reporting on diabetes and three RCTs reporting on hyperglycaemia or blood glucose fluctuation are described in online supplemental table 6.15.13 18 22 24 40–42 50
Mortality
Four observational studies reported on the risk of all-cause mortality (table 6, online supplemental tables 6.16 and 6.17).27 30 51 53 One study (unclear RoB) reported no increased risk for ever use of GC or for different cumulative doses.30 Two studies (one unclear, one high RoB) reported an increased risk for current GC use.27 51 Two studies reported an increased risk with increasing daily dose, with one study (high RoB) reporting an increased risk starting at ≥8 mg/day and one study (unclear RoB) starting at ≥5 mg/day.51 53 Three studies (two high, one unclear RoB) reported an increased risk with increasing cumulative doses.27 51 53
Mortality, results of observational studies
Three additional RCTs reporting on mortality are described in online supplemental table 6.18.13 14 40
Adverse pregnancy outcomes
One observational study (unclear RoB) reported on preterm birth (online supplemental tables 6.19 and 6.20).54 An increased risk was shown for oral GC use ≥10 mg/day after gestational day 139 (aHR 2.47 (95% CI 1.32 to 4.63)).
Glaucoma
One observational study (high RoB) reported on the risk of glaucoma.27 No higher risk was reported for current or past GC use. For some of the cumulative dose categories a higher risk was reported, but this was not consistently increased with increasing doses.
Three additional RCTs reporting on glaucoma are described in online supplemental table 6.23.13 38 40
Other safety outcomes
The remaining safety outcomes (online supplemental tables 6.24–6.33) were only reported in RCTs.13 15 18 21–24 38–42 48 49 Due to a limited number of included studies per outcome and/or a limited number of events, the safety outcomes of these trials were not further discussed.
Safety outcomes of randomized trials investigating long-term GC treatment
Patients in the SEMIRA trial (low RoB) on tocilizumab with stable low disease activity were randomised to either taper and stop prednisone at 16 weeks, or to continue prednisone 5 mg/day.18 At 24 weeks follow-up, both the number of patients with treatment emergent AEs (80 vs 64) and the number of patients with treatment emergent SAEs (7 vs 4) were higher for patients that tapered and stopped versus the patients that continued prednisone (online supplemental table 6.30).
The safety of prednisolone 5 mg/day in addition to standard care in patients with active RA aged ≥65 years was investigated in the double-blind placebo-controlled GLORIA trial (high RoB) during 2 years follow-up.13 One of the main outcomes was the number of AEs of special interest. These included all SAEs, AEs leading to discontinuation and several GC-related AEs. More patients in the prednisolone (60%) than in the placebo (49%) group experienced this outcome (p=0.02, online supplemental table 6.30). The difference was mainly caused by a higher number of serious (26 vs 16) and non-serious (124 vs 91) infections in patients on prednisolone versus placebo (online supplemental table 6.12). The number of cardiovascular events was numerically higher for patients on prednisolone versus placebo (online supplemental table 6.6). The total number of patients with an SAE did not differ between the two groups (online supplemental table 6.30). Most GC-related AEs were rare (ie, newly occurring hypertension (online supplemental table 6.6), diabetes (online supplemental table 6.15), cataract (online supplemental table 6.32)), without relevant differences between groups.
No studies were included describing the safety of GC combined with bDMARDs compared with bDMARDs without GCs, or the safety of GCs with different routes of administration.
Discussion
In this SLR, an overview was provided of the efficacy, duration of use and safety of GCs, with the aim of informing the task force responsible for the 2022 update of the EULAR recommendations for the management of RA. The results of this SLR suggest that GCs are effective as initial bridging therapy, with equal improvements over 2 years in disease activity, physical functioning and radiographic progression for moderate (30 mg/day) compared with high (60 mg/day) initial prednisone doses.14 Studies about the efficacy of lower initial GC doses were not identified in the current SLR, but have been published previously and suggest benefits with doses ranging from 5 to 10 mg/day.55 Another recently published study that was not included in the current SLR is the NORD-STAR trial.56 In this RCT in patients with early RA, treatment with either abatacept, certolizumab or tocilizumab (all plus MTX) did not show superior efficacy compared with active conventional treatment consisting of conventional synthetic DMARDs+GCs. Since data on oral GCs and intra-articular GC use were not reported separately in the publication, this study could not be included.
For low-risk patients, after 2 years, efficacy outcomes were similar for patients starting initial GC bridging versus tight step-up treatment (without GCs).14 However, disease activity during follow-up was lower for patients starting with bridging GCs, which can have positive effects. In fact, early control of disease activity has been shown to improve long-term outcomes (eg, fatigue), and to reduce the use of non-steroidal anti-inflammatory drugs (NSAIDs) and analgesics.57–59 In addition, in the second year of follow-up cumulative GC doses were higher for patients in the tight step-up treatment group, indicating that also in terms of duration of use initial GC bridging may be advantageous in low-risk patients compared with ‘rescue’ treatment with GCs.14 Further studies are needed to compare the efficacy, duration of use and safety of initial GC treatment versus rescue GC treatment in high-risk patients, especially in those starting a b/tsDMARD, and in larger patient groups. Also, studies comparing different routes of administration of GCs, the efficacy of GC bridging during follow-up, or the efficacy of GC bridging with a protocolised versus a non-protocolised tapering protocol, are currently lacking.
Concerns have recently been expressed regarding the ability to stop GCs after initial GC bridging treatment.8 We included data from a recently published SLR and meta-analysis, which concluded that in clinical trials at 12–24 months of follow-up with prespecified tapering and stopping of GCs, only a limited proportion of patients still used GCs, or resumed GC after initial stopping.9
These data suggest that in clinical trials with protocolised tapering schemes, tapering and stopping of GCs is feasible for most patients. This observation is in line with data from a study in real-world setting (not included in this SLR), in which most patients on a step-down strategy with protocolised tapering of GC could stop treatment after 2 years.60 Even though protocolised tapering of GCs seems feasible, there might be challenges in implementing this strategy into clinical practice. Further research into possible barriers and facilitators of GC tapering in clinical practice will help in bridging this gap in the near future. Based on these RCTs, the recommendations on this point should clearly demand forced tapering and discontinuation of GCs in clinical practice. It should also be noted that although 12–24 months is longer than the maximum bridging period of 3 months that is advised in the 2019 EULAR recommendations, most trials included in this SLR only allowed stopping GCs after approximately 6 months, once the GC dose had first successfully been reduced to five or 7.5 mg/day within a timeframe of about 6 weeks.6 Therefore, current data do not allow us to draw conclusions on whether stopping after 3 months is also feasible. However, it must be assumed that the effect of parenteral GC bridging does not last for more than 3 months and from this perspective 3 months appears to be a realistic target time.
The effects of withdrawal of GCs in patients with established RA (mean disease duration of 9.6 years) were studied in the SEMIRA trial.18 Patients on tocilizumab with stable low disease activity who tapered prednisone had an increase in DAS28 and a lower likelihood of being flare-free, compared with patients who continued prednisone at 5 mg/day, suggesting that in advanced disease tapering might be more difficult.
The majority of studies included in this SLR reported on GC safety. Whereas we included both observational studies and (long-term extensions of) RCTs, we considered observational studies the main study type to evaluate safety in this SLR. Since RCTs are generally not powered to evaluate safety outcomes, often exclude patients with high risks of AEs and mostly have a relatively short follow-up, safety outcomes of RCTs can be difficult to interpret. Observational studies including large numbers of patients with longer follow-up allow a better interpretation of safety outcomes. However, due to non-random treatment allocation, a RoB such as confounding by indication may play a major role in the interpretation of the results. For this reason, all observational studies included in this SLR are evaluated to have an unclear or a high RoB.
Evidence from observational studies was available for 7 out of 21 outcomes reported in this SLR. In general, there is considerable heterogeneity in the definition of outcome measures and exposure categories across studies. Studies reporting on hypertension and cardiovascular events show conflicting results. In some studies a higher risk is only observed for higher cumulative or daily doses (≥5 to 7.5 mg/day prednisone equivalent dose (PEQ)).32 33 35 One study showed no increased risk of cardiovascular events at all after adjusting for disease activity.28 Of note, only one other study with cardiovascular events as an outcome directly adjusted for disease activity.33 This study only found an increased risk for prednisone/prednisolone daily doses >5 mg/day or cumulative doses >750 mg in the preceding 6 months. This stresses the potential influence of confounding by indication in studies investigating the side effects of GC where patients who taper and flare and thus appear to be unable to stop or lower GC are different from those who can discontinue or were never exposed to GC. In addition, Information bias can also occur when defining exposure to GCs, which might further challenge the interpretation of treatment effects.61
The risk of osteoporotic fractures, serious infections and risk of diabetes was increased for GC users across all included studies. In general, this risk seemed increased for higher daily doses and longer durations, but results on a safe daily dose and/or duration are conflicting.
Only one out of four studies, which only reported unadjusted results, showed no increased risk of mortality.30 The remaining studies showed an increased risk with increasing daily and/or cumulative doses, but again residual confounding (eg, by disease severity) cannot be ruled out.27 51 53 None of the studies reported an increased risk of mortality for the lowest dose categories (<5 to 7 mg/day PEQ).
In conclusion, the results of this SLR confirm the efficacy of GC bridging as initial therapy. Moreover, the majority of patients who start GC bridging in clinical trials are able to stop GCs within 12–24 months. Also, long-term low-dose GC use seems efficacious, but with an increased risk of infections. In general, this SLR confirms well-known safety risks of GC use such as an increased risk of osteoporotic fractures, serious infections, diabetes, cardiovascular events and mortality, but with large heterogeneity between studies in terms of definitions of outcome measures and exposure categories, and particularly methods to cope with bias. Most studies show an increased risk with increasing daily dose and/or duration, but the currently available evidence is too equivocal to indicate which dose and duration are safe for different safety outcomes. The results of this SLR were used by the task force responsible for developing the update of the EULAR recommendations for the management of RA.
Ethics approval
Not applicable.
Acknowledgments
KLH is supported by the NIHR Manchester Biomedical Research Centre.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor David S Pisetsky
Twitter @AlexSepriano, @Janetbirdope
Contributors SAB extracted the data for the SLR; JWS developed the search strategies; SAB wrote the first version of the manuscript and all authors revised the manuscript critically for important intellectual content and gave final approval of the version to be published.
Funding This work was supported by the European Alliance of Associations for Rheumatology.
Competing interests SAB: received an ASPIRE grant from Pfizer. AS: received speaker/consulting fees from UCB, Novartis and Lilly. AK: Speakers bureau, Consultancy: AbbVie, Amgen, Bristol-Myers Squibb, Eli Lilly, Gilead, Janssen, Merck Sharp and Dohme, Novartis, UCB und Pfizer. JWS: received grants from Abbvie, Astra-Zeneca, Janssen, Lilly, Novartis, Roche and honoraria from Abbvie, Amgen, Astra-Zeneca, Astro, BMS, Celgene, Celltrion, Chugai, Gilead, ILTOO, Janssen, Lilly, MSD, Novartis-Sandoz, Pfizer, Roche, Samsung, Sanofi, UCB. DvdH: received consulting fees from AbbVie, Amgen, Astellas, AstraZeneca, BMS, Boehringer Ingelheim, Celgene, Daiichi, Eli-Lilly, Galapagos, Gilead, Glaxo-Smith-Kline, Janssen, Merck, Novartis, Pfizer, Regeneron, Roche, Sanofi, Takeda, UCB Pharma and is Director of Imaging Rheumatology bv. RC received consulting fees from Abbvie, BMS, Amgen, Celltrion, Eli-Lilly, Fresenius-Kabi, Galapagos, Janssen, MSD, Novartis, Pfizer, UCB Pharma. CJE Has received fees from Abbvie, Astra Zeneca, BMS, Celgene, Celltrion, Fresenius Kabi, Gilead, Galapagos, GSK, Janssen, Eli Lilly, Pfizer, Roche, Sanofi for advisory boards, speakers bureau and research support. PV received consulting fees from Abbvie, BMS, Celltrion, Eli Lilly, Galapagos, Gilead, Nordic Pharma, Pfizer, Sidekick Health, UCB and speaker’s fees from Eli Lilly, Galapagos, MSD and Roularta. Holds the Pfizer Chair Early RA Management at KU Leuven. SdS: none. JEP received research grants from AbbVie, Frensenius Kabi, Mallinckrodt Pharmaceuticals, Pfizer, Seattle Genetics and consulting fees from AbbVie, Amgen, Astra Zeneca, BI, BMS, Celltrion, EMERALD, Frensenius Kabi, Galapagos, GSK, Janssen, Lilly, Mallinckrodt Pharmaceuticals, Medexus, Mitsubishi Tanabe Pharma, Novartis, Pfizer, Roche, Sandoz, Samsung, Sobi, Viatris. TT received grants from AbbVie GK, Astellas, Asahi Kasei, Chugai, Daiichi Sankyo, Eisai, Takeda, Mitsubishi Tanabe, Nippon Kayaku, and speakers/consulting fees from AbbVie GK, Astellas, Chugai, Daiichi Sankyo, Eli Lilly Japan, Eisai, Gilead, Mitsubishi Tanabe, Pfizer Japan. KLH received speaker’s fees from Abbvie and research grants from Pfizer and BMS. KLW received research grants from Pfizer, BMS, and GSK, and scientific consulting fees from UCB, Abbvie, BMS, Galapagos, Gilead, Lilly, GSK, Roche, and Novartis. DA has received grants and/or speaker/consulting fees from Abbvie, Amgen, Lilly, Janssen, Merck, Novartis, Pfizer, Roche, SoBi, Sanofi, Sandoz, and Roche. TAS has received grant/research support from AbbVie and Roche, has been a consultant for AbbVie and Sanofi Genzyme, and has been a paid speaker for AbbVie, Novartis, Roche, Sanofi, and Takeda. JSS: none. RBML: received consulting fees from AbbVie, BMS, Celgene, Eli-Lilly, Galapagos, Gilead, Glaxo-Smith-Kline, Janssen, Merck, Novartis, Pfizer, Roche, UCB and is Director of Rheumatology Consultancy bv. JSS, DvdH, KLH, KLW, DD and RBML are members of the Editorial Board of ARD.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.