Polymyalgia rheumatica (PMR) is the most common chronic inflammatory disease in the elderly 1 . Clinically, it is characterized by pain in the neck and lower back as well as proximal extremity complaints, including tender, aching, and stiff muscles. Patients feel fatigued and their level of physical activity is reduced 1 . These symptoms are accompanied by elevated erythrocyte sedimentation rate (ESR) and increased blood levels of C-reactive protein (CRP) 1 .

The knowledge of the etiology and pathophysiology of PMR is modest. The prevailing view is that PMR reflects inflammatory processes in synovial membranes in joints, bursae, and tendons 1 2 3 4 . Several studies have found elevated blood levels of various proinflammatory cytokines in PMR 5 . Recently, we showed that levels of proinflammatory cytokines, including IL-6 and TNF-α, which both potently induce the expression of acute phase reactants 6 and promote nociception 7 8 , are increased in the interstitium of affected muscles 9 .

The only effective treatment is medium-dose glucocorticoids (GC), which abolish symptoms within a few days 10 . However, because long-term treatment is necessary, serious adverse effects, including type 2 diabetes, hypertension, and osteoporosis, are frequent 11 .

In patients with rheumatoid arthritis (RA), another chronic inflammatory disease and an important differential diagnosis in PMR 10 12 13 , administration of TNF-α inhibitors has been a therapeutic success 14 . In PMR, however, no effect of the TNF-α antagonist infliximab on relapse frequency and use of prednisone was found in a recent randomized controlled trial (RCT) of newly diagnosed patients 15 . Still, it should be noted that seven clinics participated in that study 15 . It can be expected that a high number of clinics and doctors involved increases the variation associated with clinical evaluation and decisions concerning patients and, accordingly, decreases the ability to detect differences between treatment with TNF-α blockade and placebo. Furthermore, in the mentioned RCT, patients had successfully been treated with prednisone for some weeks before the start of infliximab therapy, which was applied in parallel with a fixed tapering of prednisone treatment 15 . If at all stages the scheduled prednisone dose per se would be sufficient to control the disease, this may have hampered the ability of the study to detect any potential beneficial effect of the added TNF-α blockade.

Finally, even if infliximab has no effect in patients with PMR, the TNF-α antagonist etanercept might still be effective, because the two TNF-α inhibitors act by different mechanisms, being an anti-TNF-α monoclonal antibody and a soluble recombinant Fc-coupled TNF-α receptor fusion protein, respectively. Correspondingly, infliximab and etanercept have different therapeutic potentials in other diseases, for example, only infliximab is effective in granulomatosis disorders such as Crohn's disease and Wegener's granulomatosis 16 . Also, small uncontrolled studies have pointed to a beneficial effect of etanercept in patients with PMR 17 18 . Moreover, in a RCT of patients with giant cell arteritis (GCA), which is intimately related to PMR, etanercept was shown to be an effective therapy 19 .

As there is a need for effective drugs other than GCs for PMR, and because existing evidence does not exclude a role of etanercept, in the present study we performed the first RCT of etanercept in patients with PMR. The study was a parallel group in a placebo-controlled, double-blinded, RCT with etanercept in a group of newly diagnosed, GC naïve PMR patients and non-PMR control subjects. The trial was carried out at a single center, and all patients were diagnosed and evaluated by the same chief rheumatologist. Furthermore, etanercept was the only anti-inflammatory drug administered. So, the study allows evaluation of the pure effect of etanercept treatment in patients with PMR without any possible blurring interference due to, for example GC treatment. The duration of the trial was 14 days, because for etanercept to be an attractive alternative to GCs in the treatment of patients with PMR, beneficial effects must occur rapidly. Rapid action of an effective anti-TNF-α therapy can also be expected, if TNF-α does in fact play a key role in the pathophysiology of PMR. In agreement with this view, in RA a clear effect of TNF-α antagonism can be expected within a week 20 .

Twenty-two patients with PMR were included in the trial; of these, two patients, both receiving placebo, withdrew from the study (Figure 1). Twenty-one control subjects without PMR were included; one control subject receiving etanercept withdrew (Figure 1). Baseline anthropometric and clinical findings did not differ (P > 0.05) between patients who received etanercept and patients who received placebo (Figures 2 and 3, and Table 1). Nor did these variables differ between control subjects receiving etanercept and placebo, respectively (Table 2, P > 0.05). Age, BMI, and blood pressure did not differ (P > 0.05) between patients and control subjects; before treatment all other measurements differed significantly (P < 0.001 to 0.01) between these groups (Tables 1 and 2, and Figure 3).

After the 14 days of trial participation, all patients were treated with 20 mg/day prednisolone. Within a week, prednisolone abolished any remaining PMR symptoms and normalized CRP and ESR in all patients, supporting the PMR diagnosis.

The present study is the first RCT of the effect of etanercept in patients with PMR. The major finding is that 14 days of etanercept monotherapy ameliorates disease activity in newly diagnosed, GC-naïve patients with PMR. However, the effect is modest indicating that TNF-α does not have a predominant role in the pathophysiology of PMR.

The positive effect of etanercept is evident from the 24% reduction of PMR-AS in patients treated with the drug, as well as the insignificant changes in PMR-AS observed in placebo-treated patients (Figure 2). The reduction of PMR-AS in patients treated with etanercept reflected decreases in all paraclinical as well as objective and subjective clinical components of PMR-AS. Among these, statistically significant improvements in shoulder mobility (E), physician's global assessment (VAS_ph) and CRP were seen, and these changes were paralleled by a significant decrease in ESR and plasma IL-6 concentration (Figures 3 and 4). The fact that both etanercept and placebo had no effect in non-PMR control subjects (Figures 2 to 4, and Table 2) shows that the significant effects of etanercept seen in patients with PMR were indeed disease specific.

However, the view that the effect of etanercept is modest in patients with PMR is suggested by the fact that the reductions in both MST and VAS_p were not statistically significant (Table 1). Furthermore, at the end of the 14 days of etanercept treatment, PMR-AS was still not only higher than in control subjects but also not significantly different from values in placebo-treated patients (Figure 2). Correspondingly, although the tramadol intake was lower in etanercept compared with placebo-treated patients, the difference was not statistically significant, and functional status as evaluated by HAQ did not improve in any of the groups.

In order to verify that the anti-TNF-α treatment was still effective at the final evaluation by day 15, that is, three days after the last etanercept injection, and that potential effects of the blockade accordingly were not underestimated, plasma TNF-α concentrations were determined. However, instead of being reduced, in etanercept-treated patients the TNF-α concentrations were higher than before treatment and also higher than in placebo-treated patients (Figure 5). The increased TNF-α values observed after etanercept treatment were probably due to the simultaneous detection of both free and accumulated etanercept-bound TNF-α 27 . In agreement with this view, during etanercept treatment of control subjects TNF-α concentrations were also markedly increased (Figure 5).

Furthermore, in patients treated with etanercept, TNF-α concentrations were reduced by 61% after IgG depletion of plasma. This process also removes etanercept (free etanercept as well as etanercept-TNF-α complexes), which contains the Fc fragment of IgG1. TNF-α concentrations were similar in IgG-depleted plasma from etanercept-treated compared with placebo-treated patients. As some etanercept-bound TNF-α will remain in IgG-depleted plasma from etanercept-treated patients, while only free TNF-α is present in IgG-depleted plasma from placebo-treated patients, the similar overall TNF-α concentrations in IgG-depleted plasma most likely reflect that free TNF-α concentrations were lower in plasma from etanercept-treated patients than in plasma from placebo-treated patients. So, these observations indicate that by the end of the trial, etanercept was still present in plasma and able to lower free TNF-α concentrations compared with values in placebo-treated patients 27 .

In the present study, TNF-α concentrations in plasma were higher in patients with PMR than in non-PMR controls subjects (Figure 2) before treatment in both treatment groups. This is in accordance with findings in a recent study, in which we showed that interstitial TNF-α concentrations in affected muscle are increased in untreated PMR and normalized in parallel with complete clinical remission during 14 days of prednisolone therapy 9 . Although these findings are suggestive, the fact that in the present study TNF-α blockade had only a modest limiting effect on disease activity indicates that TNF-α is not a major determinant of the pathophysiology in PMR. Interestingly, in a study of two patients with previously untreated GCA, a condition generally considered to be closely related to PMR, a dramatic improvement was observed within 14 days of monotherapy with the anti-TNF-α monoclonal antibody infliximab, and it was concluded that in this disease TNF-α plays a major role in mediating inflammation 28 .

In the present study, the same experienced rheumatology specialist evaluated all patients. It is interesting to note that although being blinded to treatment allocation, the specialist's evaluation of disease activity as reflected by shoulder mobility (E) and VAS_ph, was in better agreement with paraclinical measures (ESR, CRP, and IL-6) than the patients' own evaluations of disease activity as apparent from MST, VAS_p, and HAQ. Due to the monotherapeutic design of the present RCT, potential effects of etanercept were not obscured by simultaneous administration of other anti-inflammatory drugs such as GCs. As a consequence, however, the trial had to be brief. Thus, it is possible that a larger effect of etanercept would have been demonstrated had it been possible to extend the duration of the trial. However, in patients with RA the effect of etanercept occurs rapidly 20 . Furthermore, in patients with GCA a small uncontrolled study showed a dramatic effect of infliximab monotherapy treatment in the early phase 28 . Conversely, a RCT has shown that extension of infliximab therapy to 22 weeks does not diminish the need for simultaneous GC therapy 29 . Still, in some contrast to the latter study, a GC-sparing effect of etanercept has been demonstrated in a recent RCT when administered late in the course of GCA in patients with side effects from GC 19 . However, such patients may have a particularly strong inflammatory reaction making them relatively GC-resistant 6 30 and sensitive to TNF-α blockade 31 .

The disease ameliorating effect of etanercept seen in the present study contrasts with the lack of benefit of infliximab found in a RCT with GC-treated, newly diagnosed patients with PMR 15 . This discrepancy may reflect differences in study design or differences in mechanism of action of the two TNF-α antagonists. Arguing in favor of the latter possibility, in a small, uncontrolled study, it was demonstrated that etanercept was beneficial in the treatment of GC-treated patients with newly diagnosed PMR and decompensated diabetes 18 . However, arguing against the latter possibility, in line with our findings, infliximab monotherapy induced remission in a small, uncontrolled study with newly diagnosed patients with PMR 17 . Moreover, in another uncontrolled study of a small number of patients with longstanding, GC-resistant PMR, administration of infliximab was accompanied by diminished disease activity 32 .

Overall, current evidence indicates that the effect of TNF-α blockade in PMR and GCA is moderate. For that reason and considering the high cost, TNF-α therapy in PMR and GCA seems justified solely when these diseases can only be controlled by GC doses that are untenable due to serious adverse effects and comorbidities induced or worsened by steroids.

In conclusion, the present RCT has demonstrated that etanercept monotherapy ameliorates disease activity in newly diagnosed, GC-naïve patients with PMR. However, the effect is modest, indicating that TNF-α does not play a central role in the pathophysiology of PMR.

BMI: body mass index; CRP: C-reactive protein; CV: coefficient of variation; ESR: erythrocyte sedimentation rate; GC: glucocorticoid; GCA: giant cell arteritis; HAQ: health assessment questionaire; Ig: immunoglobulin; IL: interleukin; MST: morning stiffness; PMR: polymyalgia rheumatica; PMR-AS: PMR activity score; RA: rheumatoid arthritis; RCT: randomized controlled trial; TNF: tumor necrosis factor; VAS: visual analog score; VAS_p: patient's visual analog score; VAS_ph: physician's visual analog score.

The authors declare that they have no competing interests.

FK and HG contributed equally to the planning and conduct of the trial as well as to the analysis and interpretation of study outcomes. Both authors drafted and approved the manuscript.