Funding: Dr Vidya Limaye was the recipient of the Australian and State and Territory affiliates grant from Arthritis Australia.
Mortality and its predominant causes in a large cohort of patients with biopsy-determined inflammatory myositis
Article first published online: 22 FEB 2012
© 2010 The Authors. Internal Medicine Journal © 2010 Royal Australasian College of Physicians
Internal Medicine Journal
Volume 42, Issue 2, pages 191–198, February 2012
How to Cite
Limaye, V., Hakendorf, P., Woodman, R. J., Blumbergs, P. and Roberts-Thomson, P. (2012), Mortality and its predominant causes in a large cohort of patients with biopsy-determined inflammatory myositis. Internal Medicine Journal, 42: 191–198. doi: 10.1111/j.1445-5994.2010.02406.x
Conflict of interest: None.
- Issue published online: 22 FEB 2012
- Article first published online: 22 FEB 2012
- Accepted manuscript online: 1 DEC 2010 04:30AM EST
- Received 5 August 2010; accepted 18 October 2010.
- inclusion body myositis;
- Bohan and Peter criteria;
- standardised mortality ratio
Background: There is a paucity of literature on the patterns and predictors of mortality in idiopathic inflammatory myopathies (IIM).
Aims: To determine the patterns and predictors of mortality in a South Australian cohort of patients with biopsy-proven IIM.
Methods: The living/deceased status (and for deceased patients the causes of death) of patients with histologically determined IIM was determined from the Births, Deaths and Marriages Registry. Standardised mortality ratios (SMR) were generated compared with the age/gender matched South Australian population. The effect of presence/absence of the components of the Bohan and Peter criteria on risk ratios (RR) for mortality was determined. The effect of comorbidities and autoantibodies on mortality was investigated.
Results: The SMR for mortality in IIM was 1.75 and was significantly increased in all disease subgroups, being highest in patients with dermatomyositis (2.40). Dominant causes of death were cardiovascular disease (31%), infections (22%) and malignancy (11%). Risk factors for death were age at time of biopsy (hazard ratio 1.05), ischaemic heart disease (RR 2.97, P < 0.0001), proximal weakness at diagnosis (RR 1.8, P= 0.03), definite diagnosis of IIM per the Bohan and Peter criteria (RR 2.14, P < 0.0001), and the absence of autoantibodies (RR 1.9, P < 0.001).
Conclusions: Patients with IIM are at 75% increased risk for mortality, and cardiovascular diseases account for the commonest causes of death. This study suggests a thorough cardiovascular evaluation of these patients is indicated, and raises the possibility that targeted interventions such as the use of aspirin or statins may improve outcomes in IIM.
The idiopathic inflammatory myopathies (IIM) are a group of systemic autoimmune diseases with dominant manifestations on skeletal muscle. The three best recognised disease subsets are polymyositis (PM), dermatomyositis (DM) and inclusion body myositis (IBM).1 As for many autoimmune diseases, both genetic2 and environmental3 factors are considered important, and recently there has been considerable interest in the role of myositis-specific and myositis-associated autoantibodies (MSA and MAA) in pathogenesis.4
The classification criteria for IIM developed by Bohan and Peter in 1975 include the presence of symmetrical weakness, increased serum creatine kinase (CK), myopathic changes on electromyography (EMG) and a supportive muscle biopsy.5 According to these criteria, myositis is said to be definite, probable or possible if four, three or two criteria are present respectively, and the diagnosis of DM requires additionally the characteristic cutaneous features. Some of the major concerns with the use of these criteria are first and foremost the fact that they were established prior to the recognition of IBM as a distinct entity and thus do not distinguish IBM from PM. Furthermore, potential for misclassification of disease subtype is recognised to occur.6,7 Hence for the present muscle biopsy is considered the definitive diagnostic test and aids not only in diagnosis but also in accurate classification of disease.
Although survival of patients with IIM has improved since the widespread use of corticosteroids and immunosuppressive medications,8 patients with IIM continue to have increased mortality,9 with a reported 5-year survival rate of 60%.10,11 Accurate assessment of prognosis in IIM is hindered by the relative rarity of these conditions and by the lack of adequate follow-up of a homogeneous population of patients defined by the same classification criteria.12 The vast majority of studies have included patients based on the Bohan and Peter criteria and such studies have been criticised for the heterogeneity of selected patients.
The rationale for determining mortality and the causes of this chronic disease extends beyond a description of its epidemiology. An understanding of prognostic factors and the ability to delineate which patients are at risk for premature mortality will enable directed approaches and improve patient management.
The aim of the present study was to estimate mortality rates in patients diagnosed with IIM compared to those of the general population, and to examine the association of existing comorbidities and the presence of autoantibodies on the risk of death in a large homogeneous population of patients in South Australia (SA) with biopsy-proven IIM.
Materials and methods
We performed a retrospective cohort study using data from the South Australian Myositis Database (SAMD), a statewide register in which are recorded the details of patients who have a histologically confirmed diagnosis of inflammatory myositis subsequent to 1980. Information recorded includes demographic details, living/deceased status and comorbidities existing at the time of database entry. Also recorded were the presence/absence of MSA and MAA in a large number of patients. The presence or absence of individual components of the Bohan and Peter criteria had been recorded, as well as whether patients fulfilled definite, probable or possible diagnoses according to these criteria.5 The dates of muscle biopsy and where applicable, the date of death were recorded. The establishment of this database has been facilitated by all diagnostic adult muscle biopsies performed in SA being reported in a central diagnostic Neuropathology Laboratory, in the Institute of Medical and Veterinary Science. Histological diagnoses are made according to the criteria described by Hohlfeld,13 and biopsies are subjected to peer-review. The database has been approved by the Research Ethics Committees of all teaching hospitals in SA.
Data from the SAMD was combined with that from the Births Deaths and Marriages Registry, South Australia (BDMR), to determine current living status (alive or deceased), and for deceased patients the dates and causes of death. The date on which the BDMR was searched (2 October 2009) was used as the censoring date for the study. Patients were followed up from their date of muscle biopsy (and diagnosis) until their date of death, or 2 October 2009 for living patients.
The causes of death were categorised into myositis, cardiovascular (myocardial infarction, cardiac arrest, congestive cardiac failure, ischaemic heart disease (IHD), arrhythmia, acute coronary occlusion and cardiogenic shock), respiratory (respiratory failure, adult respiratory distress syndrome and apnoea), infections and malignancy. Patients whose listed cause of death did not fit into one of these five categories were assigned the category ‘other’.
Standardised mortality ratios (SMR) were determined for each of the subgroups of IIM, the comparator being the standard South Australian population stratified by sex and 5-year age groups. The increased risk of death for existing comorbidities, including IHD, diabetes mellitus and hypertension (HT), was assessed in univariate analysis using risk ratios (RR). RR for death by univariate analysis were also determined for the individual components of the Bohan and Peter criteria, as well as whether patients satisfied definite, probable or possible diagnoses according to these criteria. Kaplan–Meier curves were constructed for each IIM subgroup and survival between groups was compared using the log rank test. Cox regression was used to compare hazard rates between groups after adjustment for age and gender. SMR were determined for each of the myositis subgroups according to the presence or absence of MSA or MAA.
Four patients were excluded from analysis due to missing date of birth. Of the remaining 370 patients on the SAMD, 118 were ascertained as deceased according to the BDMR. An additional six were known to be deceased according to the SAMD (124/370 deceased, 33.5%); however, these patients were also excluded from analyses because the date of death was not given. The overall median follow-up time for the remaining 364 IIM patients was 4.78 years, 5.51 years for DM (390 patient-years), 6.23 years for PM (1577 patient-years) and 5.51 years for IBM (390 patient-years).
The number of deaths per year per 100 persons was 4.6 for IIM, and 3.8, 7.0 and 4.0 for PM, IBM and DM respectively. The SMR for IIM was 1.75 and was significantly increased compared to the general population in all subgroups, being highest in patients with DM (Table 1).
|Observed deaths||Expected deaths||SMR||95% CI||P-value for SMR versus 1.0|
The Kaplan–Meier survival estimates following muscle biopsy are shown in Figure 1. The overall estimated median survival time post-muscle biopsy for IIM was 13.7 years (95% CI 11.7–21.7) for 2598 patient-years of follow-up. Patients with PM, IBM and DM were followed up for a total of 1541, 601 and 325 patient-years, and the estimated 50% median survival times were 19.3 (12.8–25.2), 9.7 (8.4–12.0) and 28.7 (11.1.) years for each of the groups. Thus there was considerable variation in median survival time among the subgroups of IIM, and a log rank test for equality of survivor function confirmed that patients with IBM had lower survival compared with PM (P= 0.0024) and DM (P= 0.06). There was no difference in survival between patients with PM or DM (P= 0.85). Although the SMR was highest in patients with DM, the median survival time post-biopsy was also the highest for patients with DM (28.7 years), most likely explained by the younger age of these patients. Age-adjusted Kaplan–Meier survival curves (Fig. 1b) showed a trend that patients with PM have the lowest survival. In age and sex-adjusted Cox regression the hazard ratios (HR) for death were 1.2 and 1.18 for IBM and DM respectively. The HR for age at time of biopsy was 1.05 (1.04–1.07), indicating a 5% increase risk of death with every year increase in age at the time of muscle biopsy. No influence of gender was seen.
Causes of death
Cardiovascular diseases were by far the commonest cause of death. Of the 118 deceased patients 36 died from cardiovascular causes (30.5%), 26 from infections (22.0%), 16 from respiratory causes (13.6%), 13 from malignancy (11%), and for five patients myositis was listed as the primary cause of death (4.2%). A further 22 patients (18.6%) died from other non-classified causes (this included cerebrovascular accidents for seven patients and pulmonary embolism in two). Of the 36 patients who died from cardiovascular causes, 16 had acute myocardial infarction. Among the 26 patients with infection as the primary cause of death, 19 had pneumonia, one had a lung abscess, five had septic shock and one had urinary sepsis.
Effect of coexisting comorbidities on risk of death
We next determined the RR for death conferred by the presence of IHD, HT and diabetes mellitus in patients with IIM (Table 2). Patients with IIM who had IHD had a significantly increased RR for death (2.97, 1.65–5.28) compared with patients who did not have IHD. Although we have shown that IIM is associated with a high prevalence of HT and diabetes mellitus,14 the present study showed HT did not increase the risk for death in IIM. There was however a trend towards an increased risk of death in patients with diabetes mellitus (RR = 1.63, P= 0.11) which did not reach statistical significance. Together this suggests that IHD and diabetes mellitus are more strongly associated with the risk of death than HT in IIM.
|Alive||Deceased||Deceased/total with comorbidity||Deceased/total without comorbidity||Risk ratio for death||95% CI||P-value for ratio versus 1.0|
Effect of components of the Bohan and Peter diagnostic criteria on risk of death
Patients with inflammatory myositis may present with a combination of elevation of serum levels of CK, proximal weakness, myopathic triad on EMG, and it is noteworthy that all features are not necessarily present. The presence of proximal weakness at the time of presentation conferred a significantly increased risk for death, and elevation of CK levels was associated with a (non-significant) reduction in mortality (Table 3). Patients who satisfied ‘definite’ Bohan and Peter criteria had a significantly increased risk for death compared with patients with either ‘probable’ or ‘possible’ diagnoses (RR 2.14).
|Alive||Deceased||Deceased/total with variable||Deceased/total without variable||Risk ratio for death||95% CI||P-value for ratio versus 1.0|
Effect of autoantibodies on SMR
We next investigated the effect of autoantibodies (MSA or MAA) on SMR in IIM. Patients with IIM who had autoantibodies had a substantially reduced SMR compared to those without autoantibodies (0.618, 0.13–1.81), and the effect was strongest in patients with PM (Table 4). Autoantibodies in patients with IBM did not have a protective effect. Among patients without autoantibodies, the SMR was increased in all three subgroups of IIM, and the subgroup with DM had the highest SMR.
|Observed deceased||Expected deceased||SMR||95% CI||P-value|
Effect of autoantibodies on presence of existing comorbidities
We sought to determine whether the absence of autoantibodies is associated with an increased risk for IHD or diabetes or whether it is an independent risk factor for death in IIM. Among antibody-positive patients, IHD was present in 6/34 (17.6%) compared with 19/83 (22.9%) without antibodies; RR = 0.79 (0.37–1.69, P= 0.53). Diabetes was present in 8/34 (23.5%) of patients with autoantibodies and 23/87 (26.4%) patients without antibodies; RR = 0.89 (0.45–1.76, P= 0.74). HT was present in 20/37 (54.1%) with antibodies and 52/88 (59.1%) without antibodies; RR = 0.87 (0.50–1.49, P= 0.60). Although a significant difference in risk for IHD, diabetes and HT conferred by the presence of autoantibodies is difficult to detect because of our small numbers, it appears that the absence of autoantibodies does not significantly increase the risk for these comorbidities and therefore may be an independent risk factor for death.
In this study we have examined the mortality and causes of death associated with a large population of patients with biopsy-proven IIM. We found our population has a 75% increased mortality compared with the general population. The predominant risk factors for death in IIM we identified were age at time of biopsy (HR 1.05), presence of IHD (approximate threefold increased risk), proximal weakness at the time of diagnosis, definite diagnosis of IIM per the Bohan and Peter criteria, and the absence of autoantibodies.
The 75% increased mortality was seen in all disease subgroups and was the highest in patients with DM (Table 1). The SMR of 1.75 is considerably lower than that (2.92) observed in a Finnish study of PM/DM patients diagnosed between 1969 and 1985.15 Importantly, this discrepancy may reflect the fact that the present study included patients with IBM, whilst the other did not, but may also reflect improved management approaches over time. The excess mortality seen in IIM is comparable to that reported in rheumatoid arthritis (RA), in which the SMR is between 1.5 and 1.6,16,17 and ankylosing spondylitis, in which the SMR has ranged from 1.33–1.8,18–20 but overall somewhat lower than the SMR of 2.4 reported in an international systemic lupus erythematosus (SLE) cohort.21 In systemic sclerosis the SMR has been documented to be 1.5 (diffuse subset 2.92, overlap disease 2.4 and limited subset 1.3)22 with diffuse scleroderma having the highest SMR of the connective tissue disorders. The reported death rate in IIM has ranged from 22% in a French cohort of PM/DM followed for a median 4 years,23 to 26% in a Spanish population or PM/DM.24 Our higher death rate of 33.5% after 57 months of median follow-up is more comparable to the 33% death rate reported in a New Zealand population-based study of patients with 76 months median follow-up time, and this is likely to reflect the inclusion of patients with IBM in both these study populations.25 This discrepancy again reiterates the problems in comparing non-uniform patient populations.
Previous studies have shown that the strongest predictive factor for mortality in IIM is age at disease onset.15,23,24,26–28 Our observed HR for death (1.05) is similar to the 1.04 HR observed in a cross-sectional study of 107 patients with DM/PM based on the Bohan and Peter (BP) criteria.24
In terms of the muscular disease, patients who satisfied definite Bohan and Peter criteria had a higher risk of death compared with those with probable or possible diagnoses (Table 3). We further report that the presence of proximal weakness was also predictive for death. Data are conflicting as to whether CK levels have prognostic value in IIM. Lack of CK elevation in DM has been correlated with a poor prognosis29,30 and higher CK levels have been associated with greater increments in muscle power in PM/DM.31 However, CK levels were shown not to influence survival in a large Finnish study of 176 patients with DM and PM,15 and also in the French group of 77 DM/PM patients.23 In the present study, we observed a trend that patients with elevation of CK had a reduced mortality risk (RR for death = 0.80) (Table 3) which did not reach statistical significance.
Almost one-third of our patients with IIM died from cardiovascular diseases, and indeed cardiovascular problems have been ranked as the highest cause of mortality in IIM by several studies, the proportions of deaths attributed to cardiovascular causes ranging from 14.7% to 55%.15,27,32,33 This variation is largely accounted for by the selection criteria used. Notably the studies with the lower incidence of IHD had selected only patients with PM and DM,33 whilst in those including IBM, the incidence of IHD was in the order or 50–55%.27,32 The link between accelerated atherosclerosis and the systemic autoimmune rheumatic diseases RA and SLE is well established;34 however, the vascular risk profile of IIM patients seems underestimated. We have recently reported a previously unrecognised high prevalence of HT (62%) and diabetes mellitus (29%) in our cohort of patients with IIM (14), although in the present study, neither were found to be predictors of death. We had also observed that IHD was a commonly associated comorbidity in patients with IIM. In a Spanish cohort of 107 patients with DM/PM, left ventricular dysfunction conferred a strong risk for death, with a HR = 5.2 (95% CI 2.4–11.1, P < 0.0001).24
We observed a significantly higher death rate from infection (22%) compared with the 2–2.5% rate of infection reported in two studies of PM/DM diagnosed by the BP criteria.15,24 However, notably infectious complications have been reported to occur in up to 30% of patients27,35,36 and opportunistic infections in 11% of patients with PM and DM.35 Part of the problem in comparing these studies is not only the selection criteria, lack of inclusion of IBM with the BP criteria, but also the possible differences in classification of cause of death. For instance, pneumonia was classified as an infection in our study, but was classified as a respiratory cause of death in the French cohort of PM/DM,23 and as an extension of the musculoskeletal disease itself in the study by Airo et al.15 Indeed on closer analysis of their data, pneumonia was the main cause of death in 23/107 (21%) of the patients with PM and 5/42 (12%) of DM in the study of Airo et al. which is very similar to the proportions observed herein, and aspiration pneumonia resulting in death occurred in 55% of patients with PM/DM in the French study.23 Marie et al. also found that aspiration pneumonia occurred most frequently during the first 2 months after diagnosis of the IIM, and was frequently related to oesophageal motor involvement and compromised ventilatory capacity. Together these studies and ours suggest that infectious complications particularly pneumonia confer a significant risk for death in patients with IIM, and hence should be diagnosed and treated promptly.
Previous studies which have included patients with cancer-associated myositis have consistently reported malignancy as the commonest cause of death,10,23,27,28,37 accounting for up to 47% of deaths.23 The study by Airo et al. in which patients with cancer-associated myositis were not included showed a similar death rate from cancer (16%) as that observed in the present study (11%).
Autoantibody testing in IIM has clinical value as individual autoantibody specificities are associated with distinct clinical features.38,39 Autoantibody testing also helps predict the risk of malignancy as patients without MSA or MAA are at increased risk of cancer.40 We have shown for the first time that autoantibodies are protective for mortality in IIM, and of note this was an independent protective factor as autoantibodies were not observed with differential frequency in patients with IHD, diabetes or HT.
Many studies on mortality in myositis have selected patients from a single centre with inherent potential for selection bias towards differential severity of disease, and preference towards specific management strategies. Our study, however, included all patients diagnosed within SA, managed at both public and private hospitals and also included non-hospitalised patients. We acknowledge that in this study, the BDMR will not ascertain the living/deceased status of patients who were diagnosed by muscle biopsy with IIM in SA and subsequently moved interstate; however, it is likely this number is modest. Further there are some concerns with using the cause of death as that listed on the death certificate, which is more likely to list acute events as the cause of death rather than a sustained chronic rheumatic illness. Such limitations acknowledged, we have in this study confirmed the excess mortality associated with IIM and show for the first time that the presence of autoantibodies is protective from mortality. Our demonstration that IHD confers a significant increased RR for death in IIM enables the identification of patients at greatest risk, and raises the possibilities that targeted interventions such as the use of statins and aspirin may improve survival in patients with IIM.
The authors are grateful to Dr Sally Cox (Flinders Medical Centre, Adelaide) for her assistance in establishing the SAMD.
- 2Differences in idiopathic inflammatory myopathy phenotypes and genotypes between Mesoamerican Mestizos and North American Caucasians: ethnogeographic influences in the genetics and clinical expression of myositis. Arthritis Rheum 2002; 46: 1885–93., , , , , et al.
- 4Immunogenetic risk and protective factors for the idiopathic inflammatorymyopathies: distinct HLA-A, -B, -Cw, -DRB1 and –DQA1 allelic profiles distinguish European American patients with different myositis autoantibodies. Medicine (Baltimore) 2006; 85: 111–27., , , , , et al.
- 8Clinical features, classification and epidemiology of inflammatory muscle disease. In: Hochberg MC, Silman AJ, Smolen JS, Weinblatt ME, Weisman MH, eds. Rheumatology, 4th edn. Edinburgh: Mosby-Elsevier; 2008; 1433–48., .
- 13Polymyositis and dermatomyositis. In: Karpati G, ed. Structural and Molecular Basis of Skeletal Muscle Diseases. Basel, Switzerland: ISN Neuropath Press; 2002; 221–7..