Patients

Diagnoses and demographic data of the investigated patient population have been described previously.¹⁸ In short, we reviewed the clinical data and muscle biopsy specimens at presentation of 268 patients (>16 years of age) with “myositis” or “possible myositis” diagnosed in the period 1977–98. In total, 103 patients were excluded because of suspected s-IBM, rhabdomyolysis or muscular dystrophy (n = 73), insufficient clinical data to determine the disease course (n = 18), absence of biopsy specimen (n = 4) or lack of muscle biopsy abnormalities (n = 8). The remaining 165 patients were classified according to the following predefined criteria:

1. Definite polymyositis (subacute onset, proximal weakness or muscle soreness without skin changes, inflammatory myopathy with mononuclear cells surrounding and preferably invading individual non-necrotic muscle fibres in the endomysium)¹⁹

2. Definite dermatomyositis (subacute onset, proximal weakness or muscle soreness with characteristic skin abnormalities of dermatomyositis or perifascicular muscle atrophy)

3. Unspecified myositis (clinical polymyositis: subacute onset, proximal weakness or muscle soreness without skin changes, histopathologically characterised by inflammatory myopathy with perimysial/perivascular localisation of mononuclear cells in the muscle biopsy specimen, without additional endomysially located cell infiltrate)

4. Possible myositis (clinical polymyositis: subacute onset, proximal weakness or muscle soreness without skin changes, serum creatine kinase levels raised more than double and necrotising myopathy).

Subclassification of each of these categories into isolated myositis, myositis associated with a connective tissue disorder (CTD; in the presence of a well-defined CTD at presentation^20–,²⁴) or myositis associated with malignancy (in the presence of a malignancy diagnosed <2 years before presentation of myositis²⁵) resulted in the following diagnoses: isolated polymyositis (polymyositis with endomysial cell infiltrates), n = 9 (5%); dermatomyositis, n = 59 (36%; 54 isolated, 3 with CTD, 2 with malignancy); unspecified myositis (clinical polymyositis with perivascular/perimysial cell infiltrates), n = 65 (39%; 38 isolated, 26 with CTD, 1 with malignancy); and possible myositis (clinical polymyositis with necrotising myopathy), n = 32 (19%; 29 isolated, 3 with CTD).

The medical ethics committees of all participating centres approved the study protocol.

Data extraction from clinical charts

The following data were extracted from the clinical files: age at presentation, sex; history of referral; disease duration (time span from start of symptoms to start of treatment) before initiation of treatment; severity of weakness at presentation; diagnosis of lung involvement; development of cancer (<2 years after onset of myositis) or of CTD (during the entire follow-up period); laboratory features at initial evaluation, type, dose and duration of treatment modalities; adverse effects of drugs; and cause of death. We also recorded the disease course (see below). Myositis-specific autoantibodies (MSAs; antibodies to Jo-1, other tRNA synthetases, Mi-2 and signal recognition particle (SRP)) were determined in all patients examined at follow-up.²⁶

Outcome assessment

Death was regarded as disease-related if caused by cancer diagnosed 2 years before or after the onset of myositis, pulmonary complications, cardiac complications in patients < 40 years without cardiac history, complications of prednisone or other immunosuppressive treatment, or complications specifically related to a CTD.

Two investigators (IMB and MFGvdM) personally assessed muscle strength, disability, quality of life and persistent use of drugs in patients who had a follow-up duration of at least 1 year. Muscle strength was tested manually and scored according to the Medical Research Council score (MRC). By adding up the scores of 12 proximal and distal upper and lower limb muscles, and head flexors and extensors, a sum score was calculated, resulting in a maximum score of 130 for normal muscle strength.²⁷ Disability was assessed by the modified Rankin scale (score 0, no disability; score 5, totally dependent), a short index of global disability, well validated for several neurological disorders and also applied in neuromuscular diseases.²⁸ Three of the subscales (body care and movement, walking, and mobility) of the Sickness Impact Profile (SIP) score, a well-validated quality-of-life test, are aggregated into a physical dimension.^29–,³¹ In a Dutch population of healthy people aged 41–50 years, the mean physical SIP score was 1.6%.³¹ Physical SIP scores >1.6% were defined as abnormal.

For the prognostic factor analysis, cut-off scores were determined. Significant muscle weakness was arbitrarily defined as an MRC sum score of <128, thus allowing for one muscle group to be rated bilaterally with an MRC grade 4. Considerable disability was assigned a modified Rankin score of ⩾3 (moderate handicap, symptoms that significantly restrict lifestyle and prevent totally independent existence). Finally, the use of prednisone at >10 mg/day or other immunosuppressive or immunomodulating drugs at follow-up was used as an outcome for the analysis of prognostic factors.

Assessment of prognostic factors

The following prognostic factors for death and unfavourable outcome were analysed: age at onset of myositis (> or <60 years), sex, classification of myositis at presentation, disease duration before initiation of treatment (> or <6 months) and presence of MSAs in survivors only. In all models we adjusted for duration of follow-up.

Assessment of clinical course

Disease course was defined as monocyclic, according to the criteria used by Huber et al³² in juvenile dermatomyositis, when the patient remained in remission (no detectable clinical or biochemical disease activity, and off all drugs) after 24 months since diagnosis. Disease course was designated as polycyclic when the patient had recurrence of disease activity (as determined by clinical or biochemical parameters) after remission, and as chronic continuous when there was persistent disease or continuation of drugs beyond 24 months after diagnosis. We determined whether disease course differed according to age at onset, diagnosis at onset, sex and MSAs.

Statistical analyses

Groups were compared using Student’s t test or Mann–Whitney U test for continuous variables and the χ² test for categorical variables. Log rank test was used to compare hazard rates. All tests were two sided, and a p value <0.05 was considered significant.

Cox proportional hazards regression analysis and logistic regression analyses were used to assess the prognostic factors predicting death and unfavourable outcome, respectively. The hazard ratio or odds ratio (HR/OR) and 95% confidence interval were used as measure of association. Subsequently, predictors that were univariately associated with death or unfavourable outcome (HR/OR with p values <0.25) were included in a multivariate regression model (full model) to evaluate their independent contribution. Model reduction was carried out by excluding predictors with p values >0.05, such that a final model was derived, including independent predictors of death and unfavourable outcome. To prevent overoptimism in future populations and to validate the model internally, the regression coefficients of each predictor in the final models needed to be subjected to shrinkage. For this, a heuristic shrinkage factor was calculated as follows: (full model χ²−p)/full model χ², where p is the number of predictors considered. This shrinkage factor was used as a multiplier for the regression coefficients of the selected predictors.³³ The prognostic ability of the models to discriminate between patients with and without poor outcome was shown by the area under the receiver operating characteristic curve.³⁴

All analyses were carried out using SPSS software V.10.

Follow-up time, treatment and development of associated disorders during follow-up

There were 120 women and 45 men, with a mean age of 45 years (standard deviation (SD) 17 years, range 16–80 years). Median follow-up period was 5 years (mean (SD) 6.0 (4.4) years, range 1–23 years). In all, 157 patients (95%) were treated with prednisone. Treatment was started within 3 months after onset of symptoms in 51% of patients and within 1 year in 87% of patients. In 125 patients (76%) prednisone was given in a dose of at least 1 mg/kg body weight/day for at least 4 weeks. In total, 94 patients (57%) were subsequently treated with one (n = 61) or more (n = 33) immunosuppressive or immunomodulating agents because of failure of the prednisone or impossibility of tapering off. Drugs used included azathioprine (n = 72), methotrexate (n = 41), ciclosporin (n = 15), cyclophosphamide (n = 6), intravenous immunoglobulins (n = 14) or plasmapheresis (n = 3). Adverse effects of the drugs were reported in 125 patients (76%), including Cushing appearance (n = 71), psychiatric symptoms (n = 35), osteoporosis (n = 29), infections (n = 29), peptic symptoms (n = 23), hyperglycaemia (n  = 18), hypertension (n = 16), acne (n = 10), glaucoma (n = 5), cataract (n = 5) and aseptic necrosis of the femur head (n = 1).

A malignancy was diagnosed within 2 years after onset of myositis in five patients with dermatomyositis, in three patients with unspecified myositis (clinical polymyositis) and in two patients with possible myositis (colon (n = 2), lung (n = 2), breast, stomach, renal cell, ovarian or oral squamous cell carcinoma, and Hodgkin’s lymphoma (all n = 1)). Development of a CTD during follow-up was diagnosed in one patient with dermatomyositis (systemic lupus erythematosus) and in 10 of 38 patients with isolated unspecified myositis (clinical polymyositis), including scleroderma (n = 4), systemic lupus erythematosus (n = 3), Sjögren’s syndrome, rheumatoid arthritis and mixed CTD (all n = 1). The cumulative risk of incident CTD in patients with unspecified myositis was 6% at 6 months, 17% at 1 year and 33% at 7 years. This risk was significantly different from the CTD risk in the rest of the patients, in whom it remained at 1% after 6 months (p<0.001, log rank test). Lung involvement occurred in 24 patients (15%) and included lung fibrosis (n = 9), restrictive pattern at respiratory function testing (n = 8), isolated alveolitis (n = 3), interstitial pneumonia (n = 3) and pleuritis (n = 1). Patients with lung involvement had Jo-1 autoantibodies significantly more often (44%) than patients without lung involvement (14%; p = 0.007).

Mortality

In all, 34 patients (21% of the 161 patients who could be traced) had died after a median follow-up of 4 years (range 0.1–20 years). The cause of death was unknown in eight patients. Myositis-related death appeared in 18 patients (11%) and occurred until 9 years after onset of myositis (median 2 years, range 0.1–9 years). Causes of death were associated cancer (n = 7), pulmonary complications (n = 4), lethal adverse effect of drugs (n = 4), CTD (n = 2) or cardiac complication of myositis (n = 1). Table 1⇓ shows factors predicting disease-related mortality.

Death was regarded as not disease-related in another eight patients, including one patient who died after a stroke, one because of suicide and two because of cancer, which occurred >5 years after myositis was diagnosed. In four elderly patients who died because of heart failure, we considered a relationship with myositis unlikely as the myositis was not active at the time of death.

Outcome in surviving patients

In all, 110 of 131 surviving patients (84%) were personally re-examined after a median follow-up of 5 years (mean (SD) 6.9 (4.4) years, range 1–23 years). There were no statistically significant differences with regard to age, sex, serum creatine kinase activity at onset, type of myositis, duration of clinical manifestations before treatment initiation, severity of weakness at presentation, or whether treatment was given in a second-line or third-line setting between the re-examined patients and the 21 patients who were lost to follow-up (n = 4, 3%) or refused to be re-examined (n = 17, 13%; data not shown).

Tables 2⇓ and 3⇓ present the results of outcome assessments in re-examined patients at follow-up, categorised by diagnostic group and myositis-specific autoantibody, respectively. At follow-up, 24% of the patients had considerable disability (Rankin score 3–5) and 25% of the patients had muscle weakness (MRC sum score <128). Abnormal physical SIP scores were found in 84% of the patients. In all, 41% of the patients were using prednisone (>10 mg) or immunosuppressive treatment at follow-up. The median total SIP score was 12.2% and the median physical SIP score 10.8%. All 37 patients without disability (Rankin score 0 or 1) had complete normal muscle strength (MRC 130): 15 (41%) of these patients were still on drugs and 23 (62%) had abnormal physical SIP scores. All patients with a normal physical SIP score (n = 17) had complete normal muscle strength and no (n = 14) or slight (n = 2) (Rankin score 2) disability.

The analyses of prognostic factors in surviving patients are presented in table 1⇑. Significant disability was predicted by male sex (OR 3.1; 95% CI 1.2 to 7.9) and muscle weakness by age >60 years (OR 3.6; 95% CI 1.3 to 10.3). The presence of Jo-1 autoantibodies predicted the persistent use of drugs (OR 4.4; 95% CI 1.3 to 15.0). Among the patients who still used drugs at the time of re-examination, the Jo-1-positive patients (n = 10) had significantly more lung involvement (50%) than the Jo-1-negative group (n = 28, 11%; p = 0.019). Duration of follow-up was not related to disability or muscle strength. An inverse relationship existed between duration of follow-up and persistent use of drugs (OR/year increase in follow-up 0.9; 95% CI 0.8 to 1.0). Outcome was not significantly different between the various myositis subtypes or groups defined by presence of various MSAs.

The receiver operatoring characteristic areas for the three final models predicting significant disability, muscle weakness and treatment dependency indicated that in 61%, 61% and 70% of the patients, respectively, outcome can be predicted correctly using these models (table 1⇑).

Clinical course of myositis in surviving patients

In the 104 re-examined patients with a follow-up time of at least 2 years, disease course was monocyclic in 21 patients (20%), polycyclic in 21 patients (20%) and chronic continuous in 62 patients (60%). Fifteen patients with a polycyclic course had one relapse, six patients had two relapses. Follow-up duration, age, sex, types of myositis and the presence of MSAs were not significantly associated with the type of disease course (table 4⇓).