Study flow diagram. GCA indicates giant cell arteritis; PMR, polymyalgia rheumatica.
Proposed therapeutic algorithm for the treatment of polymyalgia rheumatica (PMR). PDN indicates prednisone. *The 2008 International Consensus Classification Criteria47 are awaiting prospective validation. †This recommendation is not based on the data analyzed in the present review; rather, it relies on evidence that methotrexate administered at PMR onset can reduce glucocorticoid-related adverse effects, especially bone mass loss,39 and suggestions and reports from experienced investigators.2,38 Glucocorticoid-related adverse effects, including bone mass loss, should be managed according to local policies and guidelines. CRP indicates C-reactive protein; ESR, erythrocyte sedimentation rate.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Hernández-Rodríguez J, Cid MC, López-Soto A, Espigol-Frigolé G, Bosch X. Treatment of Polymyalgia Rheumatica: A Systematic Review. Arch Intern Med. 2009;169(20):1839–1850. doi:10.1001/archinternmed.2009.352
Polymyalgia rheumatica (PMR) treatment is based on low-dose glucocorticoids. Glucocorticoid-sparing agents have also been tested. Our objective was to systematically examine the peer-reviewed literature on PMR therapy, particularly the optimal glucocorticoid type, starting doses, and subsequent reduction regimens as well as glucocorticoid-sparing medications.
We searched Cochrane Databases and MEDLINE (1957 through December 2008) for English-language articles on PMR treatment (randomized trials, prospective cohorts, case-control trials, and case series) that included 20 or more patients. All data on study design, PMR definition criteria, medical therapy, and disease outcomes were collected using a standardized protocol.
Thirty studies (13 randomized trials and 17 observational studies) were analyzed. No meta-analyses or systematic reviews were found. The PMR definition criteria, treatment protocols, and outcome measures differed widely among the trials. Starting prednisone doses higher than 10 mg/d were associated with fewer relapses and shorter therapy than were lower doses; starting prednisone doses of 15 mg/d or lower were associated with lower cumulative glucocorticoid doses than were higher starting prednisone doses; and starting prednisone doses higher than 15 mg/d were associated with more glucocorticoid-related adverse effects. Slow prednisone dose tapering (<1 mg/mo) was associated with fewer relapses and more frequent glucocorticoid treatment cessation than faster tapering regimens. Initial addition of oral or intramuscular methotrexate provided efficacy at doses of 10 mg/wk or higher. Infliximab was ineffective as initial cotreatment.
The scarcity of randomized trials and the high level of heterogeneity of studies on PMR therapy do not allow firm conclusions to be drawn. However, PMR remission seems to be achieved with prednisone treatment at a dose of 15 mg/d in most patients, and reductions below 10 mg/d should preferably follow a tapering rate of less than 1 mg/mo. Methotrexate seems to exert glucocorticoid-sparing properties.
Polymyalgia rheumatica (PMR) is a syndrome characterized by aching and morning stiffness in the shoulder and pelvic girdles and neck in persons 50 years or older.1,2 Systemic manifestations such as low-grade fever, fatigue, and weight loss are frequently present, as are increased acute-phase reactants including high erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) levels, and anemia of chronic disease.1,2
Treatment with glucocorticoids is the preferred therapy for PMR.1,2 Before the glucocorticoid era, the occasional self-limiting nature of PMR was evidenced by spontaneous improvements in some patients,3,4 and musculoskeletal symptoms were treated with nonsteroidal anti-inflammatory drugs (NSAIDs).3,5 Today, prednisone and its principal active metabolite, prednisolone, considered to be equipotent at equivalent doses, are universally used in PMR. Other currently used glucocorticoids include methylprednisolone and deflazacort (not available in the United States).
An initial prednisone dosage of 10 to 20 mg/d is deemed appropriate for most patients who have PMR without associated giant cell arteritis (GCA).1,2,6 Symptoms usually resolve completely after a few days. Most patients require at least 2 years of treatment, but others have a more chronic, relapsing, or refractory course requiring steroid treatment for much longer.1,2 The adverse effects of long-term glucocorticoid therapy are common and sometimes deleterious in patients with PMR.1,2,7 To reduce the total cumulative dose of glucocorticoids and their adverse effects, some researchers have investigated the addition of cytotoxic drugs and, more recently, biologic agents with potential glucocorticoid-sparing effects to the PMR regimen.1
To our knowledge, no reports have summarized the evidence for glucocorticoid treatment or glucocorticoid-sparing therapies in PMR. The present review systematically analyzes the reported evidence on PMR therapy, especially the preferentially used glucocorticoid, its optimal initial and maintenance doses and tapering regimens, and glucocorticoid-sparing agents used.
We systematically searched the Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and MEDLINE/PubMed for English-language articles published between 1957 and December 2008, using the MeSH term polymyalgia rheumatica in combination with the terms treatment, glucocorticoids, prednisone, prednisolone, methylprednisolone, deflazacort, methotrexate, azathioprine, NSAIDs, and biological therapy. References of relevant articles retrieved were searched manually. Studies that included 20 patients or more were selected.
Two of us (J.H.-R. and X.B.) independently read titles and abstracts searching for articles on medical interventions in PMR. Articles considered to meet inclusion criteria, and those with inconclusive abstracts were fully reviewed to decide on their final inclusion. Three of us (J.H.-R., A.L.-S., and X.B.) recorded the types and initial doses of glucocorticoids and other therapies tested, subsequent tapering schedules, proportion of patients discontinuing glucocorticoid treatment, time to treatment cessation, and relapse rate during follow-up. Additional data recorded included primary end points, inclusion and exclusion criteria, number of patients enrolled, baseline demographics, misclassified patients (eg, patients initially diagnosed with GCA or later developing GCA or other inflammatory conditions), patients lost to follow-up, follow-up duration, and treatment-related adverse effects. All data were reviewed and confirmed by one of us (J.H.-R.).
Various proposed definition criteria for PMR (Table 1)8-11 and the authors' own criteria are noted when used. In studies including patients initially diagnosed with GCA, only those patients with isolated PMR were analyzed when possible. In studies with patients initially considered to have PMR alone who later developed symptoms suggestive of GCA, confirmed or not by temporal artery biopsy, the number of patients with isolated PMR at the end of the study was specified.
Methodologic quality was evaluated independently by 3 of us (J.H.-R., X.B., and G.E.-F.). Observational studies were evaluated according to the “Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.”12 Quality and susceptibility to bias in observational studies were appraised using the criteria recommended by Sanderson et al.13 The quality of randomized trials was assessed using the scale proposed by Jadad et al.14 Disagreements on data and the quality of selected studies were resolved by discussion among all authors.
According to the type of medication used to treat PMR, we analyzed glucocorticoids, glucocorticoid-sparing agents, and NSAIDs. Treatments for initial remission induction and maintenance phases were examined separately.
We identified 784 citations. After retrieving 163 articles, 133 were excluded. We finally analyzed 30 studies with ≥20 patients (13 randomized trials 17 and observational studies) (Figure 1). No meta-analyses or systematic reviews were found. Of the 2220 patients initially included, 153 (6.9%) were either initially diagnosed with GCA or developed GCA during follow-up, and 2161 (97.3%) were finally analyzed.
Prednisone or prednisolone alone was investigated in 14 observational studies15-28 and 2 randomized trials.29,30 Observational studies included 8 case series (5 retrospective15-19 and 3 prospective20-22), 3 retrospective case-control trials,23-25 and 3 prospective cohort studies26-28 (Table 2 and Table 3). Deflazacort was analyzed in 1 prospective case series31 and 3 randomized trials.32-34 Methylprednisolone35 and 6-methylprednisolone36 were investigated in 1 randomized trial each. Eight studies used glucocorticoid-sparing agents; 5 used methotrexate (3 randomized trials,37-39 1 retrospective case-control trial,40 and 1 prospective cohort study41); and 2 randomized studies (1 each) tested azathioprine42 and infliximab43 (Table 4 and Table 5). Three studies analyzed NSAIDs.19,23,44
All studies used different diagnostic PMR criteria (Table 1), outcome definitions (eg, relapse, recurrence, and disease remission) (Table 6), scoring systems, medications and routes of administration, initial dosages, tapering schedules, and length of follow-up. Most studies were observational, and only 2 randomized trials could be considered confirmatory studies with an appropriate sample size calculation.38,43 This heterogeneity did not allow a pooled estimator to be calculated or statistical heterogeneity to be tested. Study designs were therefore considered in the following order (listed from lowest to highest evidence quality): case series, case-control studies, cohort studies, and randomized trials.
Most studies evaluating the use of glucocorticoids alone for remission induction or maintenance in PMR were observational (Table 2 and Table 3).
In 1 study, patients with PMR who were treated with a dose of 10 mg/d of prednisolone required fewer dose increases than those taking less than 10 mg/d, although no patient treated with more than 10 mg/d required dose modification.27 When initial prednisolone doses of 15 mg/d or lower were compared with doses higher than 15 mg/d, the daily maintenance dose was higher in patients initially treated with more than 15 mg/d after the first and second year of treatment, although the initial dose did not influence relapses or treatment discontinuation.28 A case series using prednisolone at initial doses of 15 mg/d found that fewer than 1% of patients required doses higher than 15 mg/d to control symptoms.21 In the only randomized trial in which a prednisolone regimen was started at 10 mg/d or 20 mg/d for 4 weeks with rapid tapering in 2 months, patients initially receiving 20 mg/d had fewer relapses than those receiving 10 mg/d, and 30% of patients taking 10 mg/d had to increase to 15 to 20 mg/d to control symptoms.29
Based on these results, fewer than 1% of patients with PMR initially treated with prednisolone at 15 mg/d required higher doses to control their symptoms. Prednisolone doses of 10 mg/d or higher seemed to control initial PMR more efficiently than lower doses, and doses of 15 mg/d or lower appeared to be as effective as higher doses. Therefore, the best available evidence seems to indicate that 15 mg/d of prednisone as a starting dose could be effective in most patients with PMR.
Glucocorticoid regimens are usually tapered according to clinical and laboratory responses (generally ESR and CRP levels).16,31,34 González-Gay et al15 found that a tapering rate of less than 1 mg/mo was associated with fewer relapses than reductions greater than 1 mg/mo after initial prednisone doses of 10 to 20 mg/d. Similar results have been found by other authors.17,21,35 Two studies used prednisolone at 15 mg/d followed by gradual tapering until maintenance doses of 8 mg/d21 and 10 mg/d35 were reached, with subsequent reductions of 1 mg every 2 months until treatment discontinuation. Both of these studies showed optimal control of disease activity during the study period.21,35 Conversely, faster reduction regimens were associated with poorer results.27,30 These findings would indicate that once a stable prednisone daily dose of 10 mg is achieved after initial remission, further dose reductions should be smaller than 1 mg/mo (eg, 1 mg every 2 months).
For prednisone regimens initiated at 10 to 20 mg/d, discontinuation rates have been reported to be 41% to 50% after 2 years,16,24,25 70% after 3 years, 82% after 4 years,16 and 91% after 11 years.15 When regimens started at 10 mg/d or lower or 12.5 mg/d or lower, 70% of patients discontinued therapy after 4 years of follow-up.18,20 In a study of a starting prednisone dose of 20 mg/d, 33% of patients discontinued treatment in less than 1 year.26
Several studies have compared the effect of different initial glucocorticoid doses on treatment duration and/or relapse rate.23-25 When compared with prednisone doses lower than 15 mg/d, doses of 15 mg/d or higher did not show differences in any outcomes.25 Prednisone treatment discontinuation rates were similar when initial doses of 10 mg/d or lower were compared with initial doses of 15 to 20 mg/d23 and when initial doses of 10 mg/d (range, 7-12 mg/d) were compared with initial doses of 24 mg/d (range, 15-30 mg/d).24 However, patients taking 10 mg/d tended to have more relapses than patients taking higher initial doses.24 Starting doses of greater than 15 mg/d were associated with a higher risk of glucocorticoid-related adverse effects and no additional benefit.19,24
Relapses usually occurred in 23% to 29% of patients during the entire follow-up period15,16 and, depending on the study, in 33% of patients during the first year.21 A higher relapse rate (55%) was reported by a retrospective study using prednisone at wide dose ranges (1-100 mg/d; median dose, 15 mg/d).17
The only study assessing sex-related differences in newly diagnosed PMR cases found that women had more relapses, received higher cumulative doses, and had more glucocorticoid-related adverse effects than men.22
Despite the differing starting doses and tapering regimens, prednisone doses between 10 and 20 mg/d seemed to control disease activity at PMR onset and, overall, allowed glucocorticoid treatment cessation in about 50% of patients at 2 years.
Three randomized trials compared deflazacort with other glucocorticoids in terms of efficacy at controlling disease activity during a 12-month or shorter follow-up period.32-34 When used at the same doses, deflazacort was found to have a lower potency than prednisone,32 prednisolone,34 and 6-methylprednisolone.33 Although deflazacort treatment initiated at doses higher than 20 mg/d seemed to be effective in newly diagnosed PMR cases,31 these results are mainly based on a single observational study.
Two randomized, double-blind, placebo-controlled trials analyzed the efficacy and safety of intramuscular (IM) methylprednisolone treatment35 and methylprednisolone administered via local shoulder injections36 in the treatment of PMR. A 1-year study compared a depot preparation of IM methylprednisolone acetate (120 mg every 2 weeks for 12 weeks followed by monthly injections with dose reductions of 20 mg every 3 months) with an oral prednisolone regimen (15 mg/d gradually reduced to 10 mg/d) in newly diagnosed PMR cases. Prednisolone dose reductions below 10 mg/d were made at 1 mg every 8 weeks. Although both glucocorticoids induced and maintained disease remission, prednisolone tended to control symptoms more rapidly and consistently and also showed higher glucocorticoid treatment discontinuation rates than IM methylprednisolone. However, patients taking prednisolone received higher cumulative doses and had more glucocorticoid-related adverse effects.35
One study36 evaluated 6-methylprednisolone treatment given as bilateral shoulder injections every 4 weeks in newly diagnosed PMR cases limited to the shoulder girdle. Shoulder discomfort and systemic symptoms resolved initially in all patients, and this effect was sustained after 14 months in 50% of patients.
These limited results and the need for repeated invasive procedures suggest that routine methylprednisolone injections (IM or shoulder) do not represent a practical PMR treatment and should only be considered for patients at high risk of glucocorticoid-related adverse events (IM injections) and in cases of shoulder-limited PMR (shoulder injections).
Three randomized studies have investigated methotrexate regimens in newly diagnosed PMR cases (Table 4 and Table 5).37-39Oral methotrexate doses of 7.5 mg/wk plus 20 mg/d of prednisone offered no greater benefits than prednisone alone in all outcomes measured after 2 years of follow-up.37 However, these results may be misleading because (1) methotrexate doses of 7.5 mg/wk may be insufficient to exert glucocorticoid-sparing effects and (2) 15% of patients included had GCA, which usually requires higher prednisone doses to control disease activity. Oral38 and IM39 methotrexate at a dose of 10 mg/wk, when added to a prednisone regimen, showed glucocorticoid-sparing effects compared with a prednisone regimen alone regarding relapse rates, prednisone treatment discontinuation rates, duration of prednisone therapy, and cumulative prednisone dose.38,39 Intramuscular methotrexate treatment was discontinued at 18 months by all patients who had stopped prednisone therapy 6 months before.39 Overall glucocorticoid-related adverse effects and a significant decrease in bone mass density were observed only in patients receiving prednisone alone.39
A randomized, double-blind, placebo-controlled study tested the efficacy of infliximab as a glucocorticoid-sparing agent in newly diagnosed PMR cases43 (Table 4 and Table 5). No differences were observed between groups in (1) the proportion of patients without relapses through week 52 (primary end point), (2) the number of relapses, (3) the duration and cumulative dose of prednisone, or (4) prednisone treatment discontinuation rates. Thus, while infliximab cannot be considered a useful glucocorticoid-sparing agent in patients with newly diagnosed PMR. the addition of oral or IM methotrexate to the regimen at 10 mg/wk or higher seemed to reduce relapses, prednisone requirements, and prednisone-related adverse effects.
Oral methotrexate doses of 7.5 mg/wk (increased to 10.0-12.5 mg/wk, according to clinical response) used as cotreatment for remission maintenance in patients with PMR previously receiving prednisone (most requiring ≥20 mg/d) for ≥3 months did not show clinical or biochemical benefit after 9 months of follow-up.41 However, the inefficacy of methotrexate in this subset of patients requiring unusually high glucocorticoid doses may not be generalizable to the larger PMR population.
Long-term effects of oral methotrexate were retrospectively analyzed40 in a cohort that had earlier been studied in a randomized clinical trial.38 At 59 months after therapy initiation, a modest effect of methotrexate was maintained in that the number of flare-ups per patient was reduced, but no differences in other disease outcomes were found.40
The only study (randomized, double-blind, placebo-controlled) using azathioprine (150 mg/d) during the maintenance phase in PMR showed a high frequency of medication-related adverse effects, and 35% of patients withdrew (44% in the azathioprine group and 27% in the placebo group).42 After 52 weeks, patients receiving azathioprine required lower cumulative prednisolone doses than those taking placebo. The small number of completers and high proportion of included patients with GCA (29%) make the results difficult to interpret.
The addition of NSAIDs to glucocorticoid regimens for the treatment of patients with PMR has shown no advantage over glucocorticoids alone in duration of therapy or daily or cumulative prednisone doses, and it produced more adverse events.19 However, some patients with PMR may achieve sustained remission with NSAIDs.23 Anecdotally, the effects of tenidap, an unlicensed NSAID, were investigated during the maintenance phase in PMR, and a high toxic profile was found without glucocorticoid-sparing effects.44 Although these conclusions were based on low-quality evidence, it is safe to say that treatment with NSAIDs alone may relieve symptoms in a minority of patients with PMR, but it may also have undesirable adverse effects when administered long term with glucocorticoids.
Although studies evaluating treatment of PMR not associated with GCA have significant clinical and methodologic variations, and the quality of evidence does not allow specific therapeutic recommendations, the best available evidence suggests that prednisone or its equivalent, at a starting dose of 15 mg/d, may control disease activity in most patients. However, 0% to 13% of patients may still require higher initial doses to control symptoms.21,27-29
Initial prednisone dose reductions of 2.5 mg monthly or every 2 weeks until the dose of 10 mg/d is reached have been used.1,21,35 Subsequent reductions may be attempted at 1 mg/mo or less (eg, 1 mg every 6-8 weeks) until discontinuation.15,17,21,35 Since relapses usually occur when the prednisone dose is reduced to below 10 mg/d26,30 or 5 mg/d,15,16 near the time of,21,25 or within the first 3 months of,18 dose reduction, control visits every 3 months are reasonable, especially when the dose was reduced at the previous visit.
No studies have addressed the management of refractory or relapsing disease. Our own experience is based on maintaining the minimum prednisone dose that controls disease activity. Continuing treatment with glucocorticoids only or adding an agent with glucocorticoid-saving properties should be decided after considering the risks and benefits of long-term glucocorticoid therapy and contraindications to adjuvant therapy.
Of the glucocorticoid-sparing agents tested, oral38 or IM39 methotrexate at a dose of at least 10 mg/wk seems to be useful in new-onset PMR. However, relapsing cases or those treated with long-term prednisone doses of 10 mg/d or higher may require higher methotrexate doses. The initial addition of methotrexate to a prednisone regimen might benefit patients at high risk of glucocorticoid-related adverse effects: the combination treatment regimen has shown fewer adverse effects than prednisone alone.39 The increased benefit and lower adverse-event profile of IM and subcutaneous methotrexate treatment compared with oral methotrexate in the treatment of rheumatoid arthritis45,46 suggests that subcutaneous methotrexate might also be considered for the treatment of PMR. Methotrexate therapy discontinuation could be tentatively attempted 6 to 12 months after glucocorticoid treatment cessation.39
From the findings of the present review, we have designed an algorithm for treating PMR (Figure 2). Osteoporosis prophylaxis with bisphosphonates, oral calcium, and vitamin D supplementation are broadly recommended1 because glucocorticoid-related adverse effects are reported in widely divergent percentages of patients with PMR: one group of studies concludes the range to be 3.6% to 27%15,20,25,27,35 of patients with PMR, while another group reports it to range from 58% to 91%.19,21,39Adverse effects can be detected after 1 year of treatment19,21 and are more frequent in patients experiencing more disease relapses15 and those receiving higher glucocorticoid doses and for longer periods.15,19,24,25,27
It is challenging to identify patients with PMR who have more resistant disease and who may benefit from a tailored treatment strategy. Elevated ESR16,17,26,28,48,49 and CRP48 and interleukin 626,48 levels at the time of diagnosis correlate with an increased risk of relapse17,26,48,49 or higher glucocorticoid requirements,16,28 especially if abnormalities persist during treatment.26,48 High hemoglobin levels and low ESR values are associated with a better response to glucocorticoid therapy in PMR.28 Sex seems to influence the course of PMR: compared with men, women seem to have more resistant disease, more relapses,22 a need for greater cumulative amounts of glucocorticoids,22 more glucocorticoid-related adverse effects,19,22,25,35 and a need for longer-duration glucocorticoid treatment.16,25
The limitations of this review are mainly due to the lack of controlled intervention studies on PMR treatment. In addition, there was significant variation between studies, including different diagnostic PMR criteria, outcome definitions, initial dosages, tapering schedules, and follow-up periods. Specifically, the indistinct use of nonvalidated PMR classification criteria8-11 weakens the inclusion criteria. However, a recent consensus study by 27 international experts47 established 7 potential PMR classification criteria that are awaiting prospective validation. We have based the therapeutic algorithm on these classification criteria (Figure 2). Other multicenter studies have addressed PMR activity scoring systems for defining remission thresholds and developing response criteria for treatment monitoring.50,51
In conclusion, although with limited evidence-based information, the available data suggest a starting prednisone dose of 15 mg/d followed by a slow tapering regimen as appropriate treatment for most PMR cases. Although methotrexate has shown glucocorticoid-saving properties, the efficacy of all adjuvant medications included in this review and new biologic and nonbiologic glucocorticoid-sparing agents remains to be determined in larger randomized controlled trials, especially in patients with PMR who are glucocorticoid dependent.
Correspondence: Xavier Bosch, MD, PhD, Department of Internal Medicine, Hospital Clínic, Villarroel 170, 08036 Barcelona, Spain (firstname.lastname@example.org).
Accepted for Publication: July 22, 2009.
Author Contributions: All authors are responsible for the contents of this report. Study concept and design: Hernández-Rodríguez, López-Soto, and Bosch. Acquisition of data: Hernández-Rodríguez, López-Soto, Espigol-Frigolé, and Bosch. Analysis and interpretation of data: Hernández-Rodríguez, Cid, and Bosch. Drafting of the manuscript: Hernández-Rodríguez, Cid, Espigol-Frigolé, and Bosch. Critical revision of the manuscript for important intellectual content: Hernández-Rodríguez, Cid, López-Soto, and Bosch. Statistical analysis: Hernández-Rodríguez. Administrative, technical, and material support: Espigol-Frigolé and Bosch. Study supervision: Hernández-Rodríguez, Cid, López-Soto, and Bosch.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant SAF 08/04328 from the Ministerio de Educación y Ciencia and Fondo Europeo de Desarrollo Regional (FEDER), Madrid, Spain; and grant 06/0710 from Marató TV3, Barcelona, Spain. Drs Espigol-Frigolé and Cid are supported by the Instituto de Salud Carlos III, Madrid.
Additional Contributions: Ferran Torres, MD, PhD, of the Statistics and Methodology Support Unit, Department of Clinical Pharmacology–Support Assessment and Prevention Unit, Hospital Clinic, Barcelona, provided methodologic support.