Adult onset Still’s disease (ASD) is a rare condition, prevalence of which is estimated to be 1.5 cases per 100 000 to 1.5 cases per 1 000 000.^1–3 Its expression is very polymorphic and its progression highly variable.^4–8 Although some patients have a unique flare without any recurrence, most have recurrences, which progress to a chronic form of the disease.⁹ The literature distinguishes two main chronic forms: a systemic form with fever, asthenia, and other general symptoms, and an arthritic form with polyarthritis, which is sometimes erosive.^4–7,9–11

Because of its low prevalence and its heterogeneous expression, randomised controlled trials are difficult to perform in ASD patients. Thus, all available evidence concerning effective therapies has emerged from open label studies. Aspirin and other non-steroidal anti-inflammatory drugs have been found to stabilise the disease in only 10−30% of patients.^6,8,11 In fact, corticosteroids are the preferred treatment, resulting in a clinical response in 76−95% of patients.^4,8,10–13 However, dependence on or resistance to steroids is common, and other treatments are needed. By homology with rheumatoid arthritis, several disease modifying antirheumatic drugs (DMARDs), including gold salts, hydroxychloroquine, d-penicillamine, sulfasalazine, and cyclophosphamide, have been tested.^{4,9–12,14,15} More recently, two other treatments have been found to be effective in cases of steroid dependence or resistance: methotrexate^16–19 and polyvalent intravenous immunoglobulin (IV Ig).^20–22 As several immunological studies have previously shown that Th-1 cytokines, specifically TNFα, are involved in ASD pathogenesis, anti-TNFα agents, which are highly efficient in rheumatoid arthritis and ankylosing spondylitis,^23,24 have been also evaluated in ASD. Apart from reports of a few isolated cases, only the results of three open label studies have been published.^25–28 Two studies were retrospective and included five and six patients respectively, treated with either etanercept or infliximab.^26,27 The other study was a prospective open label trial of 12 patients with established ASD treated with etanercept.²⁸ Thus, complete and partial remission might be achieved with anti-TNFα therapy in some subsets of patients.

The Club Rhumatismes and Inflammation, a section of the French Society of Rheumatology dedicated to the study of inflammatory rheumatisms, organised a survey in French hospitals to retrospectively evaluate the efficacy and tolerance of anti-TNFα in the treatment of ASD.

MATERIALS AND METHODS

Patient identification

This retrospective study was based on a national postal survey of the departments of rheumatology and internal medicine in all French hospitals; general or teaching. Departments that had treated patients with established ASD, according to Yamaguchi classification criteria,²⁹ who were resistant to conventional DMARD therapy, for whom anti-TNFα therapy had been tested, were sent a standardised questionnaire asking about age and symptoms at ASD diagnosis, classification criteria, treatments tested before anti-TNFα therapy, and items related to the therapy.

Anti-TNFα therapy

At the time of the survey, only etanercept and infliximab were available in France. Patients receiving either drug were eligible, whatever the therapeutic scheme. The dose and mode of administration were those usually used in rheumatoid arthritis (3 mg/kg infusion of infliximab administered at weeks 0, 2, 6, 10, and every 8 weeks thereafter, and 25 mg etanercept administered subcutaneously twice a week). The study required no minimal duration of treatment, and all patients having received at least one injection or infusion were included.

Efficacy and tolerance of anti-TNFα therapy

The symptoms and circumstances having led to anti-TNFα therapy were recorded. The response to anti-TNFα was defined in three categories on the basis of the opinion of the physician in charge of the patient: remission was defined as a complete resolution of all clinical and biological ASD related symptoms, except joint erosion; failure was defined as an absence of significant improvement within 1−3 months following treatment start; and partial remission was defined as the persistence of one or several ASD related symptoms. In the latter case, the remaining symptoms were recorded. In the case of therapy discontinuation, the reason for discontinuation was recorded. All side effects, either suspected or certain, were also noted.

RESULTS

The survey identified 20 ASD patients in 12 different rheumatology or internal medicine departments. The mean age was 32.2 (19) years (median 28 years) at ASD diagnosis and 40.7 (17) years (median 34.5 years) at anti-TNFα therapy start (table 1). The ratio of males to females was 1 to 4 (5 men, 15 women). All patients satisfied Yamaguchi criteria for ASD.²⁹ Any other differential diagnoses had been ruled out. In four patients, the disease began during childhood, and the patients had active disease up to the adult age. At the time of anti-TNFα therapy, all patients were adult. Mean disease duration at therapy start was 8.5 (6) years (median 6.5 years, range 2–21 years). In five patients, the disease was predominantly systemic (mainly fever, arthralgia, and rashes); in 15, chronic polyarthritis was the dominant feature. All patients had had several treatments (table 1), especially prednisone and methotrexate, before anti-TNFα therapy.

All patients had refractory active disease at the start of anti-TNFα therapy (table 2). Ten received infliximab only and five etanercept only. In five cases, patients were successively treated by both drugs: four received etanercept first, then infliximab because of lost efficacy; one received the reverse sequence after experiencing a systemic anaphylactic reaction after the third infliximab infusion. All but two patients received anti-TNFα in association with oral prednisone. Six patients receiving etanercept also took another DMARD (four methotrexate, one azathioprine, and one IV Ig) and 14 receiving infliximab took another DMARD (12 methotrexate, two azathioprine). The mean treatment duration was 11 (9) months for etanercept and 9 (7) months for infliximab.

The mean follow up was 13 months for etanercept and 14 for infliximab (median 9 and 10 months, respectively) (table 3). Anti-TNFα efficacy was discernable within 2−6 weeks following treatment start. Complete remission was observed in only five patients (one treated with etanercept and four with infliximab). Most patients achieved a partial remission: 16 of 25 treatments (7/10 receiving etanercept and 9/15 infliximab). One patient had a partial and dissociated response to etanercept: all symptoms but fever improved. After a few months, a global flare occurred and etanercept was discontinued. Finally, failure to respond to anti-TNFα therapy was noted in four patients. Both forms of the disease seemed to have similar response patterns to therapy. Partial remission was achieved in 80% of patients (4/5 systemic; 12/15 articular ASD).

Five patients switched etanercept and infliximab treatment: four received first etanercept then infliximab, and one had the reverse combination (fig 1). In all cases, the switch was made because the first anti-TNFα was ineffective; one case had concomitant skin rash, which accelerated the discontinuation of infliximab. In all but two cases, the alternative agent produced an insufficient response, and after a mean follow up of 6 months, the second agent was discontinued. One patient had a skin rash that led to etanercept discontinuation. Another case had an accidental serious burn of the upper limbs that required infliximab discontinuation, despite a partial but significant response of ASD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Breakdown of the five patients with anti-TNFα switches.

At the last follow up, the agents had been discontinued in 17/25 treatment sequences (7/10 etanercept, 10/15 infliximab; table 4), the main reason being a lack of efficacy. Side effects were responsible for discontinuation in three patients. One patient had a skin rash and blurred vision, and received a diagnosis of optic neuritis. Antinuclear antibodies were present at a titre of 1/80, without any reactivity for double stranded DNA, soluble antigens, or histons, and the skin biopsy suggested an allergic origin (no lupus band on immunofluorescence analysis). Brain magnetic resonance imaging (MRI) and angio-MRI results were normal, as was visual evoked potential. With discontinuation of infliximab, all symptoms resolved within 6 months. A 74 year old woman died of cardiac failure 2 weeks after the second infliximab infusion. She had had one episode of pericarditis and arrhythmia when ASD was diagnosed 2 years before anti-TNFα therapy, but had had no other cardiac symptoms during the course of the disease. After the second infusion, she developed high fever (39°C) and features of refractory cardiac failure, the exact origin remaining unknown. The third patient discontinued both infliximab and etanercept because of an allergic skin reaction.

Mild side effects were observed in seven patients (table 5). Three patients had skin rashes, which responded to anti-histamine treatments. Serious side effects occurred in two patients. One had recurrent bronchitis and an episode of pneumonia requiring antibiotics. However, it was possible to maintain therapy with only minor changes in the infliximab infusion periodicity. Another case had a spontaneous thigh abscess. Antibiotics and etanercept discontinuation resolved the abscess without need for surgical intervention; anti-TNFα therapy was reintroduced within 2 months, without further infection.

DISCUSSION

Compared with previously published papers, the present series brings additional relevant data on the efficacy of anti-TNFα agents in refractory ASD, thanks to the number of patients, the presence of both systemic or articular forms of the disease, the use of either infliximab, etanercept, or both, and a rather long follow-up (13 months; up to 44 months for some patients). The retrospective design of our study is a limitation, but prospective randomised control trials are difficult to conduct for ASD. This limitation is shared by most of the studies published so far,^25–27 which included only a few patients, six at maximum. Only one series, of 12 cases, used a prospective, open label design and reported the efficacy of etanercept in patients with a polyarticular form of ASD.²⁸

As our study was retrospective and collected data from all over France, we tried to homogenise the information by defining three clinical responses based on physicians’ opinions. With use of this classification, we found that when anti-TNFα agents were efficacious, clinical improvement occurred rapidly, within the first month of treatment. Complete remission was possible under both agents but was rare, only 5/20 cases, which is consistent with data from the literature.^26,28 Partial response was observed in most patients (16/25 cases) in our series as well as in those of the literature. Only one patient experienced a remission of joint and skin symptoms without any substantial change in fever spikes. Such a dissociation had already been mentioned in ASD with use of other DMARDs¹⁶ and etanercept.²⁸ The patients with a systemic form of ASD seemed to respond the same as the patients with an articular form of the disease.

The present series offers information about the efficacy of switching from one anti-TNFα to the other. In opposition to what has already been suggested for rheumatoid arthritis,³⁰ such a switch did not seem to be efficacious in ASD. Two patients in whom therapy with one anti-TNFα failed showed partial response to the other; however, the response was transient and led to rapid discontinuation of the second agent. For the three other cases, a partial and transient response was observed with use of both drugs. The pathogenic basis of such occurrences is unknown. In general, the safety of anti-TNFα therapy was in accordance with what has been described for rheumatoid arthritis, psoriasitic arthritis, ankylosing spondylitis, or inflammatory enterocolitis.^31–33 We did not observe any cases of tuberculosis, which is probably explained by the large diffusion of warnings from pharmaceutical firms or governmental agencies at the time of the study. Other infectious side effects, noted in the literature,²⁷ were found in our series; no life threatening infection was recorded. Skin rashes were observed in our patients but were distinguishable from ASD specific skin lesions: they were pruritic urticarian lesions occurring soon after anti-TNFα injection or infusion, and had no vesperal timing and relation to fever spikes. The most serious side effect was an anti-TNFα induced autoimmune reaction in one patient, who had a skin rash and optic neuritis. Such events have been reported after anti-TNFα therapy in patients with other inflammatory conditions.^34–36 In general, therapy discontinuation resolved the symptoms.

In conclusion, anti-TNFα agents may be useful in the treatment of refractory ASD. However, this therapy does not seem to be as efficacious in ASD as in rheumatoid arthritis or spondylarthropathies.