Polymyositis (PM) and dermatomyositis (DM) are chronic inflammatory muscle disorders clinically characterised by progressive proximal muscle weakness. Treatment is based on glucocorticoids in high doses in combination with immunosuppressive drugs such as azathioprine or methotrexate.1 Many patients have only a partial response and some patients do not respond at all.2 3 Hence, there is a need for new therapies.

The molecular pathways driving the chronic inflammation in myositis have partially been clarified and proinflammatory cytokines have consistently been found in muscle tissue of patients with idiopathic inflammatory myopathies (IIM).4 5 A role for tumour necrosis factor (TNF) in the pathogenesis of PM and DM is suggested by the presence of TNF at the protein and mRNA level in muscle tissue as well as increased serum levels of soluble TNF receptors.6^–10 A few case reports have suggested beneficial effects of two different TNF blockers in PM or DM.11^–16 However, to date there have been no systematic studies on the effects of TNF blockade on clinical symptoms as well as on the underlying inflammatory process in muscle tissue in patients with IIM.

The aim of this study was to test the hypothesis that TNF is a key molecule in disease pathogenesis of IIM by targeting TNF, and measure the effects on clinical outcome, serological data and on the local inflammation in muscle tissue in patients with PM and DM that were refractory to conventional immunosuppressive treatment. We also included patients with inclusion body myositis (IBM), as the molecules produced by inflammatory cells in muscle tissue are similar to the ones seen in PM and DM including TNF and interleukin (IL)1.17

PATIENTS AND METHODS

Patients

A total of 13 patients (8 women and 5 men) with refractory myositis followed at the Rheumatology Unit, Karolinska University Hospital were included in the trial. Clinical characteristics of the patients are presented in table 1. The patients were diagnosed according to the Bohan and Peter criteria,18 19 five with definite PM, four with definite DM and four with IBM diagnosed according to Griggs criteria.20

Inclusion criteria were: persisting muscle weakness defined as ⩽80% of muscle strength as measured by functional index (FI) of myositis (21) and signs of disease activity and failure to respond to treatment with high doses of glucocorticoids for a minimum of 6 months in combination with azathioprine and/or methotrexate. Disease activity was defined as muscle oedema observed by MRI, or creatine kinase (CK) elevation, or inflammatory cell infiltrates in muscle biopsy. The trial was approved by the local ethics committee at Karolinska University Hospital and all patients signed a written informed consent prior to their participation in the study.

Study design

This was a 4-month, open label, uncontrolled trial. All clinical and laboratory assessments were carried out twice: before the first infusion and 2 weeks after the last infusion with infliximab (week 16). Infliximab was given as intravenous infusions at a dose of 5 mg/kg body weight four times (weeks 0, 2, 6 and 14). Concomitant treatment with glucocorticoids and methotrexate and/or azathioprine had to be stable for 2 months before study start and throughout the trial period (table 1).

Clinical and laboratory data

Clinical assessments included the disease activity core set proposed by The International Myositis Assessment and Clinical Studies Group (IMACS) including manual muscle test (MMT).22 23 Serum levels of CK and lactate dehydrogenase (LD) were performed as routine analyses at the Department of Clinical Chemistry, Karolinska University Hospital. Muscle impairment was also assessed by FI, which measures the number of repetitions in 11 muscle groups.21 The total score varies between 0–64, where 64 indicate full capacity. The same physical therapist (HA) performed all muscle function tests without knowledge of the laboratory or biopsy data. Clinical improvement was defined according to the proposal of IMACS as ⩾20% improvement in three or more of the core set parameters and no more than two worsened by ⩾25%, neither can be MMT.22^–24 Worsening was defined by ⩾30% reduction in any three of six variables of the IMACS core set disease activity measure.25

Serology

Autoantibody tests were performed as routine analyses at the Department of Clinical Immunology at Karolinska University Hospital and included ANA, analysed on HEp-2 cells (Immunoconcepts, Sacramento, California, USA), autoantibodies to Sjögren syndrome A (SSA), Sjögren syndrome B (SSB), Sm, ribonucleoprotein (RNP), Scl-70, Jo-1 and centromere analysed using antinuclear antibody (ANA)-profile ELISA (Pharmacia Diagnostics, Uppsala, Sweden), Innolia Immunoblot (Innogenetics, Ghent, Belgium) and Auto Immunodiffusion (Immunoconcepts, Sacramento, California, USA). Anti-double stranded DNA was assayed by ELISA dsDNA (Pharmacia Diagnostics, Uppsala, Sweden). Autoantibodies to cardiolipin (IgG and IgM) were analysed by ELISA (Orgentec, Mainz, Germany).

Magnetic resonance imaging

Examinations of pelvic and thigh muscles were performed on 1.5 T (Philips, Best, The Netherlands). Body coil was used. Axial spin echo T1 weighted images, short tau inversion recovery (STIR) images and spin echo T1 weighted images with fat suppression were acquired before and after contrast enhancement. All examinations were evaluated by a radiologist (HE) who was blinded to clinical data. Increased signals on the STIR images and increased contrast enhancement when compared to surrounding tissues were regarded as signs of acute inflammation. The changes were classified as minor, moderate, or extensive (arbitrary scale).

Muscle biopsies

Muscle biopsies were obtained from m. vastus lateralis (n = 6) or from m. tibialis anterior (n = 3) by a semi open technique (the referred biopsies were from patients that completed the study).26 27 The repeated biopsy was taken from the contralateral side. Muscle biopsies from one patient (table 1, patient 17) were excluded due to end stage histopathology. The muscle biopsy specimens were frozen in liquid isopentane, stored at −70°C. Conventional histopathological evaluation was performed on coded sections by an experienced neuropathologist (IN). Immunohistochemistry staining was used to identify presence of T lymphocytes, macrophages, expression of major histocompatibility complex (MHC) class I and class II on muscle fibres,28 cytokines,4 interferon inducible protein (myxovirus resistance protein A (MxA)),29 and membrane attack complex (MAC) as previously described.30 Antibodies used are listed in table 2. Positive tissue controls were lipopolysaccharide (LPS) L-6529 (Sigma Chemical Co., St Louis, Missouri, USA) stimulated human peripheral blood monocytes (PBMCs), synovial specimens and tonsil specimens.

The sections were assessed coded by conventional microscopic assessment for qualitative evaluation. The same microscope has been used in a previous publication.31 Conventional microscopic evaluation was also used for semiquantitative expression of MAC in capillaries: – = no positively staining; 1+ = 1–3, 2+ = 4–10, 3+ ⩾11 positively stained capillaries. For quantification the area of specific immunostaining on whole tissue sections was measured by computerised image analysis and is expressed as a percentage of the total tissue area. The following scoring system was arbitrary assigned for cells, cytokine expression and endothelial cell markers (CD3, CD4, CD8, Ki67, CD68, CD163, IL1α, IL1β, TNF(2C8), high mobility group box chromosomal protein 1 (HMGB-1), MxA and CD31), except: IL6, TNF (Mab 1+Mab 11), and interferon (IFN)γ due to no or few positively stained cells; – = 0%, 1+ = 0–0.5%, 2+ = 0.5–2.5%, 3+ = 2.5–5%, 4+ = 5–10% and 5+ ⩾10% positively stained area. The following scoring system was assigned for MHC class I and MHC class II expression; – = no positive staining in muscle fibres, 1+ = 1–20%, 2+ = 21–40%, 3+ = 41–60%, 4+ = 61–80% and 5+ = 81–100% positively stained fibres. For results see Supplementary material.

Type I IFN activity measurement in patient sera

The method to measure type I IFN activity and preparation of cDNA has been published previously.32 Primers for three genes that are highly induced by type I IFN signalling were used in quantitative real-time (RT)-PCR. Measurement of the first two interferon-induced genes, with tetratricopeptide repeats 1 (IFIT-1) and protein kinase R (PKR), has previously been described.32 The third gene used was myxovirus resistance 1 (MX-1) 5′-TACCAGGACTACGAGATTG-3′ (forward) and 5′-TGCCAGGAAGGTCTATTAG-3′ (reverse). The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) has also been used previously32 and was used to quantify the cDNA samples to control for background gene expression. The myositis samples were compared to the mean and SD of a pool of previously tested healthy donors33 and the sum of the number of SD above healthy donors for each of the three genes was calculated for each sample.

Statistical analysis

The Wilcoxon signed rank test was used to compare variables before and after infliximab treatment. Adjustment for multiple comparisons was made by the use of the Bonferroni correction method; p values ⩽0.05 were considered statistically significant.

RESULTS

Clinical data

The clinical results are summarised in table 3. Nine patients completed the trial. Three patients stopped prematurely due to adverse events and one due to a newly-diagnosed malignancy. Of the five patients with PM or DM who completed the trial, clinical improvement was recorded in two patients with PM, one remained unchanged and two worsened according to the IMACS definition.22 23 Improvement was observed mainly in the doctor’s global assessment, in Health Assessment Questionnaire (HAQ) score, and in the extramuscular domain of arthritis. MMT did not improve in any patient (table 3). FI improved >20% in one responder. Two patients with DM and one patient with PM stopped the trial prematurely due to disease flare with worsening of muscle fatigue and increased CK levels, severe erythaema, or severe cough, in one patient each (table 3). The patient who experienced disease flare improved after pulse doses of glucocorticoids and cyclophosphamide. The erythaema and the cough resolved within 2 weeks after infliximab was withdrawn. A fourth case (DM) was withdrawn from the trial when an ovarian malignancy was discovered after the first infliximab infusion. Adverse events among the patients who completed the trial were diffuse abdominal pain in five patients. These adverse events resolved spontaneously after the last infliximab infusions. Of the four patients with IBM, one improved according to the IMACS definition and three were unchanged. One had adverse events: a lichenoid skin rash and fatigue.

Extended treatment

The three responders (two PM and one IBM) continued treatment with infliximab. One patient with PM developed a severe flare of myositis 1 week after the last assessment in the trial, with myalgia, panniculitis, fatigue, fever, abdominal pain, increased CK levels and new signs of muscle inflammation on the repeated MRI (patient 2). A second responder (PM) had to stop 4 months later due to disease flare (patient 1). The third responder (IBM) stopped after 6 months due to frequent respiratory tract infections including pneumonia (patient 10).

Autoantibodies

One patient with PM who flared developed positive ANA and IgM anti-cardiolipin autoantibodies. Two non-responding patients with IBM developed IgM anti-cardiolipin autoantibodies, and one of these patients also developed ANA.

Magnetic resonance imaging

Before therapy, minor inflammatory changes were noted in two of the nine completers (one PM and one IBM). After treatment, inflammatory changes were noted in five of the nine completers (three PM and two IBM). Three of these did not have any inflammatory changes in the pretreatment MRI and two had persisting changes. Of the patients with new MRI changes indicating muscle inflammation, one had clinical signs of a flare, one was unchanged and one flared immediately after the trial (fig 1A,B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 MRI of thigh muscles in a patient with polymyositis before (A) and after (B) infliximab treatment. New changes were seen after treatment visualised as increased signals (bold red arrows). The sequences acquired were axial spin echo T1 weighted images, short tau inversion recovery (STIR) images and spin echo T1 weighted images with fat suppression. C–F. Muscle specimens from a representative patient with polymyositis after infliximab treatment stained with immunohistochemistry technique to visualise inflammatory markers. C. Positive staining for macrophages (brown staining) in an inflammatory cell infiltrate (black arrow) and no expression in the capillary (dotted arrow). D. Major histocompatibility complex (MHC) class I expression (brown staining) in inflammatory cells (thin black arrow) and muscle fibres and sarcolemma (bold black arrow). E. Tumour necrosis factor (TNF) (2C8) expression (brown staining) in inflammatory cells (black arrow) and no expression in the same capillary as in C (dotted arrow). F. Interleukin (IL)1α expression in inflammatory cells (black arrow) and endothelial cells (dotted arrow) (brown staining).

Cellular infiltrates

In the first biopsy inflammatory infiltrates were present in seven of eight available biopsies from the nine completers. In the post treatment biopsy seven patients still had infiltrates in their skeletal muscle. T Lymphocytes and macrophages were detected in all patients in the first and second biopsy in inflammatory cell infiltrates but also scattered over the entire sections. Some changes on individual level were seen in patients with PM and DM (Supplementary material); the two responding patients with PM (patients 1 and 2) had decreased expression of CD4+ and CD8+ T lymphocytes. Another patient with PM who worsened clinically (patient 3) had increased expression of CD3+ and CD4+ T cells and both macrophage markers (Supplementary material). Two patients with IBM (patients 11 and 12) had an increase of CD68 and CD163 positive macrophages (Supplementary material) (fig 1C).

MHC class I and class II expression

In all patients most muscle fibres expressed MHC class I, before and after treatment. MHC class II expression displayed the same staining pattern as MHC class I antigens on muscle fibres, inflammatory cells and endothelial cells, but to a lesser degree. No difference was detected in MHC class I or II expression by computerised image analysis before vs after infliximab treatment (fig 1D).

Cytokine expression and MAC deposits

In the whole group there was a tendency of increased MxA expression after infliximab treatment (p⩽0.05, without Bonferroni correction) (table 4). For the other cytokines there were some changes on individual level. In patients with PM and DM increased expression of TNF and IL1β was seen in three patients each and IL1β in one (fig 1E,F). Only one patient had a decreased in TNF expression (patient 1) (Supplementary material). The same patient was also clinically improved (table 3). All three patients with IBM had increased expression of TNF and IL1β. HMGB-1 was increased in two patients (patients 10 and 11) (Supplementary material).

TNF as visualised by the use of the neutralisation antibody (a mix of Mab 1+Mab 11), which means that if only stains cells producing TNF, was only expressed in a few cells in six of eight patients before and after treatment. IL6 was expressed in a few cells in two patients in both biopsies. IFNγ expression was not detected in any patient either before or after treatment. Unchanged MAC deposits were observed in scattered capillaries of all patients and on the surface of scattered non-necrotic fibres of two patients.

Type I IFN activity

Type I IFN activity in sera was analysed in 10 patients. There was a significant increase in the type I IFN activity after infliximab treatment compared to baseline (mean value 1.54 (2.41) before, 3.88 (4.03) after treatment) (p = 0.037) (table 4). Among the patients with PM and DM, one of the two who improved had decreased type I IFN activity and one was unchanged. All four patients with PM or DM who were worse or unchanged and on whom data were available showed an increase in type I IFN activity (table 4). Two of the four patients with IBM had detectable type I IFN activity before infliximab treatment. After treatment the activity increased in three and unchanged/decreased in one (table 4).

DISCUSSION

Infliximab treatment in this cohort of patients with treatment-resistant IIM had only limited and transient clinical effects in occasional patients. The improvement was mainly seen in non-muscular variables. The absence of clinical improvement in muscle strength was associated with persistent signs of muscle inflammation in muscle biopsies or in MRI of thigh muscles and with systemic increased type I IFN activity. As the subgroups of myositis are small comparisons in response pattern are not possible.

A limitation of our trial was the open uncontrolled design. Nonetheless, we believe that by using objective measures of muscle inflammation such as MRI and repeated muscle biopsies in addition to clinical outcome measures our results are reliable. Furthermore, an open study design usually carries a risk of false positive effects of study treatment rather than negative results, as seen in our study.

The clinical outcome measures that we used, a core set to assess disease activity, and definition of improvement, are outcome measures that have been proposed by IMACS and have partly been validated.23–25 34 The sensitivity to change has not been tested in any clinical trial so far. The recorded clinical improvement in three cases was mainly recorded in non-muscular variables such as doctor’s global assessment but also in the self-rated functional assessment, HAQ and in the extramuscular domain of arthritis whereas MMT was unchanged. Improvement was also noticed in the FI test in one of the responders and one non-responder. This could indicate that MMT, that measures muscle strength, is less sensitive than FI that measures muscle endurance. Although our study is limited it raises a need for further validation of this disease activity outcome tool, which was developed in consensus.

We included some patients with IBM in this study because the presence of TNF has been reported in muscle tissue of such patients. One of our patients with clinical response had IBM; however, no improvement was seen of inflammatory variables in muscle tissue.

The limited clinical improvement with infliximab treatment in our patients with PM or DM contradicts published case reports with good results.12^–16 35 One possible explanation could be the selection of patients, as one of the first reports of a positive effect of infliximab treatment was observed in two newly-diagnosed patients. The long-term follow-up report of those cases was less encouraging.13 A similar transient effect of infliximab treatment was seen in two of our patients who after an initial response developed a flare of the disease during continued treatment 1 week to 4 months after the trial. In a retrospective study in which patients with similar refractory PM and DM were treated with etanercept or infliximab, a favourable response was reported in six out of eight cases.35 Notably, the improvement was based on reduced serum CK levels as well as improved doctor’s global assessment. It cannot be excluded that etanercept, which was used in most of the patients that had improved, may have a more advantageous effect in patients with myositis than infliximab, but this still needs to be tested with validated outcome measures.

Worsening of muscle symptoms was seen in a few of our patients and one patient had to stop prematurely due to increased muscle symptoms. Furthermore, two of the responders worsened during the extended treatment period. Clinical worsening was accompanied by increased or persisting signals on MRI indicating inflammation of thigh muscles, by increased CK levels or by increased signs of inflammation in muscle biopsies. We cannot exclude that worsening occurred by chance. Interestingly, in a literature survey we found one case with RA who developed clinical myositis, anti-Jo-1 autoantibodies, interstitial lung disease and muscle biopsy changes typical of myositis during treatment with infliximab.36 The mechanism behind increased inflammation seen in some conditions treated with anti-TNF treatment is not known. One suggested explanation is that blocking TNF may cause activation of T lymphocytes.37 However, we could not determine T lymphocyte activation in muscle tissue of our patients. Another possibility is an increased activity of the type I IFN system that has been implicated to have a role in the pathogenesis in DM and PM.29 38 It has also recently been shown that type I IFN inducible gene is expressed in blood in patients with PM and DM and reflects disease activity.39 Although our numbers of patients are low, there was a trend towards lack of improvement in association with increased type I IFN activity among our patients with PM and DM. A similar increased type I IFN activity was recorded after TNF blockade by etanercept in patients with primary Sjögren syndrome, likewise without any clinical improvement.32

To study the effects of TNF blockade on inflammatory signs in muscle tissue we used repeat muscle biopsies. Inflammatory changes within affected muscles from patients with IIM are often focally distributed. To minimise the risk of sampling error we analysed two muscle biopsy samples from each patient at each biopsy site by routine histopathological assessment with similar results. In addition, the first and the last sections of a biopsy sample were compared for histopathological changes, without notable difference. The biopsies that were compared (before and after treatment) were stained at the same time to keep staining conditions standardised. For quantification of staining we used computerised digital image analysis. In general the inflammatory cell infiltrates as well as inflammatory molecules expressed in the inflammatory cells, activation markers in endothelial cells such as MAC and MHC-class I and II expression in muscle fibres were remarkably consistent in the two biopsies arguing against an immune modulating effect of infliximab on muscle inflammation. The only change we could determine in muscle biopsies in this cohort of patients was a tendency of increased MxA expression. This goes in line with the increased type I IFN activity in sera suggesting that some patients with IIM that are treated with infliximab could respond with activation of the type I IFN system.

In conclusion, anti-TNF treatment with infliximab was not effective in patients with myositis that was previously resistant to conventional immunosuppressive treatment. Arthritis improved but not muscle strength, suggesting that different molecular pathways drive different clinical symptoms. These data argue against TNF as a key molecule in chronic myositis. The clinical flares, signs of increased muscle inflammation on MRI scans, and the increase in type I IFN activity in circulation and local in muscle tissue suggest that infliximab could worsen muscle inflammation and that TNF blockade is not a drug to be used in patients with treatment-resistant myositis.