To systematically identify and appraise the existing literature on the association between polymyalgia rheumatica (PMR) and vascular disease.
To systematically identify and appraise the existing literature on the association between polymyalgia rheumatica (PMR) and vascular disease.
The Medline, Embase, CINAHL, and Cochrane databases were searched from inception to September 2010. Search criteria included observational studies containing patients with isolated PMR reporting on a vascular outcome. Data were extracted and articles were assessed for quality.
The database search identified 545 articles. Eight articles reporting on 7 unique studies were included in the final review. Four studies reported on vascular mortality and 3 studies reported on other vascular outcomes. Four of the studies reported statistically significant positive associations between PMR and vascular disease and 3 studies reported no statistically significant results. The heterogeneity of the results was investigated; the strongest explanatory variable was that 2 of 3 studies that reported nonsignificant results were prospective, whereas all of those studies reporting significant results were retrospective.
There is some evidence to suggest that, like other inflammatory rheumatic disorders, PMR may be linked to an increased risk of vascular disease. Further studies are required to quantify the level of risk to determine whether patients with PMR should be actively screened for cardiovascular disease and its risk factors.
Polymyalgia rheumatica (PMR) is a common inflammatory rheumatologic condition of uncertain etiology affecting patients ages >50 years (1). The condition causes pain and stiffness in the extremity girdle muscles, with fever, weight loss, depression, and malaise as possible auxiliary symptoms (2).
PMR is associated with giant cell arteritis (GCA), which is caused by inflammation of large and medium-sized arteries typically in the head and neck. Although both conditions can be treated with steroids, studies suggest that more than 50% of patients remain on steroids at 2 years (3).
Around the world and particularly in Western countries, vascular diseases, such as cardiovascular disease, cerebrovascular events, and peripheral vascular disease, are major causes of mortality and morbidity. Research has emphasized the association of a number of inflammatory rheumatic disorders such as rheumatoid arthritis, systemic lupus erythematosus, gout, and psoriatic arthritis with accelerated atherosclerosis, resulting in an increased risk of vascular disease, especially cardiovascular and cerebrovascular disease (4–7). Recent European guidelines for the management of inflammatory disorders recommend that patients with these conditions receive screening for cardiovascular disease and aggressive management of cardiovascular risk factors to try to reduce this excess vascular risk (8).
PMR may be associated with vascular disease for a number of reasons: either the inflammatory nature of the disease itself, because of its association with GCA (which has a direct inflammatory effect on arteries and may result in either aneurysms or stenosis within affected segments) and subclinical arteritis, or via the potential side effects of corticosteroid treatment (including hypertension, hyperlipidemia, and diabetes mellitus) (9, 10). Paradoxically, however, corticosteroids in patients with PMR may actually decrease vascular risk by controlling inflammation (11).
As the populations of Western countries age, the economic and personal burdens of both PMR and vascular diseases will increase and any association between the 2 conditions will be magnified. The aim of this systematic review was to examine the association of PMR and vascular disease.
This is the first systematic attempt to synthesize the current literature about the cardiovascular comorbidity of polymyalgia rheumatica (PMR).
This study highlights the potential impact of PMR at both the personal and public health levels to influence potentially fatal or disabling vascular events and the need for further research, since PMR is a common condition in those already at high risk of vascular disease.
Current evidence is not homogeneous enough to recommend changes to the management of patients with PMR.
An investigation should be conducted in primary care to discriminate between PMR and giant cell arteritis.
With the aid of a health librarian, a search strategy was defined using subject heading terms, which were checked and found to contain a core set of disease terms and study design terms.
Search inclusion criteria included free text and heading terms for PMR combined with heading terms for vascular disease (covering but not limited to cardiovascular, cerebrovascular, and peripheral vascular disease) and heading terms related to observational study designs. Articles referring to GCA were searched in full text and were included in the review if they contained a subset of PMR patients without concurrent GCA.
Medline, Embase, and CINAHL were each searched separately from their inception to September 2010 (for full search strategies, see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). The Cochrane Library was searched manually using their drop-down heading system. The screening process was performed twice by the same reviewer (ATH), unclear candidate articles were screened in full text, and the reference lists of included articles were then screened using the same inclusion and exclusion criteria.
Data extracted included sample size, PMR definition, control source, mean age, sex proportions, proportion with concurrent GCA, study population location, followup length, loss to followup, steroid treatment information, vascular risk factor information, and outcomes. This was performed twice by the same reviewer (ATH).
Study quality was determined using a predesigned quality assessment tool adapted from the Newcastle-Ottawa Scale criteria supplemented with additional points from Altman and specific questions related to PMR and the study of vascular comorbidity (12, 13). This tool was used separately by 2 reviewers (ATH and either SLH or CDM), with disagreements being resolved by the third reviewer (SLH or CDM).
The studies were split into 2 groups to assess the possible causes of the heterogeneity of their results. Groups of studies were formed based on whether they found a positive association between PMR and vascular disease or whether they reported nonsignificant findings. The groups were then compared in terms of meeting or not meeting the 12 quality assessment criteria and also 4 other study design aspects. Statistical significance of this comparison was found using Fisher's exact test.
A total of 545 articles were identified by the initial search. Based on titles and abstracts, 531 articles were excluded, the primary reasons for which are shown in Table 1. Fourteen articles were reviewed in full text, 7 of which were subsequently excluded because they reported all-cause mortality with no specific information on vascular causes of death. Five further articles were identified from the reference lists of the included articles, 1 of which met the inclusion/exclusion criteria and was included in the final review, bringing the total to 8 articles. A summary of the search is shown in Figure 1.
|Excluded articles (n = 531), no. (%)|
|No isolated PMR subjects||112 (21.1)|
|No vascular outcome||383 (72.1)|
|Nonobservational design||34 (6.4)|
|No PMR/non-PMR comparison||12 (2.3)|
Two articles reported data from the same cohort at different lengths of followup and are therefore discussed in this review as a single study (14, 15). Four of the identified studies reported on vascular mortality and 3 reported on other vascular outcomes.
Two studies reported a statistically significant positive association between PMR (16) or PMR/GCA (17) with vascular mortality, whereas 2 studies did not demonstrate any association (14, 15, 18). The results of these studies are shown in Table 2.
|Author, year (ref.)||PMR group size||Reported outcome|
|Bengtsson and Malmvall, 1981 (14)||49||“total incidence of myocardial infarction and cerebrovascular disease … seems not to be higher than expected”|
|Andersson et al, 1986 (15)||49||25 vascular deaths vs. 26 expected|
|Schaufelberger et al, 1995 (16)||220||29 vascular deaths vs. 17 expected (P < 0.05)|
|Uddhammar et al, 2002 (17)||35|
|Female||SMR 1.49 (95% CI 1.18–1.89)|
|Male||SMR 1.58 (95% CI 1.12–2.24)|
|Myklebust et al, 2003 (18)||315||RR 0.78 (95% CI 0.52–1.18)|
Schaufelberger et al reported an increased all-cause mortality rate among a group of PMR patients (n = 220) compared to expected mortality rates from population statistics. They observed 41 deaths in their cohort (compared to 29 expected deaths). This overall increase in mortality was due to an excess of vascular deaths (29 deaths [70%] from vascular disease compared to population statistics, which estimated that 17 deaths [58%] would be from vascular causes), which was reported as being statistically significant (P < 0.05) (16).
Uddhammer et al also reported increased standardized mortality rates (SMRs) for vascular disease among patients with PMR (n = 35) and GCA. Death due to cardiovascular diseases in women and men was reported separately, with SMRs of 1.49 (95% confidence interval [95% CI] 1.18–1.89) for women and 1.58 (95% CI 1.12–2.24) for men. The small number of PMR patients without GCA was not compared separately to the control population; however, there was comparison of vascular mortality between GCA and PMR patients that was reported as being higher in those with GCA (P = 0.05) (17).
However, a small Swedish PMR cohort study found no difference between the rates of vascular mortality in patients with PMR (n = 49) or GCA compared to Swedish population rates at either 58 months (14) or 136 months (15). However, this was a small cohort of patients, the majority of whom (59.4%) also had concurrent GCA.
A larger study by Myklebust et al compared deaths from coronary heart disease or stroke and found no excess risk, with a relative risk of 0.78 (95% CI 0.52–1.18) for development of these outcomes in the PMR group (n = 315) (18).
The results of studies reporting on other vascular outcomes are shown in Table 3. Two studies that were based on similar cohorts of patients from Minnesota reported statistically significant associations between PMR and various vascular events (19, 20).
|Author, year (ref.)||PMR group size||Outcome||Result|
|Kremers et al, 2005 (19)||193||MI||OR 1.78 (95% CI 1.13–2.82)|
|CVE||OR 1.60 (95% CI 1.08–2.39)|
|PVD||OR 2.21 (95% CI 1.37–3.60)|
|Pfadenhauer et al, 2005 (21)||34||Vertebral narrowing||2.9% of PMR subjects vs. 3.0% of controls|
|Warrington et al, 2009 (20)||353||PVD||HR 2.50 (95% CI 1.53–4.08)|
Kremers et al explored the additional cost of health care usage by patients with PMR (n = 193) using a design that took into account vascular events both before and after diagnosis of PMR. The article reported statistically significant associations between PMR and myocardial infarction (odds ratio [OR] 1.78, 95% CI 1.13–2.82), peripheral vascular disease (OR 2.21, 95% CI 1.37–3.60), and cerebrovascular events (OR 1.60, 95% CI 1.08–2.39). However, it is not clear whether most of these events were pre- or postdiagnosis (19).
Warrington et al reported an increased risk for the development of peripheral vascular disease in PMR patients (n = 353) compared to controls (hazard ratio [HR] 2.50, 95% CI 1.53–4.08), which was adjusted for the presence of known risk factors such as diabetes mellitus, hypertension, and hyperlipidemia (20). In contrast, Pfadenhauer et al reported no statistically significant difference in the frequency of vertebral artery stenosis between PMR patients (n = 34) and controls (21).
The full results of the analysis of heterogeneity of the study findings, along with the adapted quality assessment tool, are shown in Supplementary Appendix B (available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658).
No individual quality point or specific aspect of study design was able to explain the observed heterogeneity. The design aspect closest to explaining the heterogeneity was the inclusion of increased specialist-led followup of patients; this was present in 2 of 3 studies reporting no significant association between PMR and vascular disease (14, 15, 18) and none of the 4 studies that reported a statistically significant association (P = 0.143).
The results of this systematic review appear to demonstrate an association between PMR and vascular disease, with 4 of 7 included studies reporting a positive association. However, caution is needed when interpreting these results, due to the degree of heterogeneity in the identified studies that prohibited a formal meta-analysis.
Point estimates of the strength of association ranged from an HR of 2.5 (95% CI 1.53–4.08) for the association of PMR and peripheral vascular disease (20) to a relative risk of 0.78 (95% CI 0.52–1.18) for vascular causes of death (18). It is likely that if PMR is associated with one form of vascular disease, since vascular diseases share common etiologies, then there would be an association with other forms and also an effect on mortality. As no single quality point, design feature, or outcome measure completely explained the heterogeneity of the results found by this review, it is likely that combinations of these factors are responsible for the differences in reported associations.
The estimates of the effect of PMR on vascular mortality reported by some studies in this review are comparable with the strength of association of other inflammatory conditions, for example, rheumatoid arthritis, which has been shown to have a meta-SMR of 1.50 (95% CI 1.39–1.61) (5). If PMR was confirmed to be associated with vascular mortality then it would represent a potentially important modifiable factor both to individuals with the condition and, since PMR is common, a potential impact on public health.
It is interesting that the aspect that came closest to explaining the heterogeneity was the inclusion of increased levels of specialist care in study design, which appears to make patients with PMR have no difference in vascular risk compared to the healthy population (14, 15). A similar effect to this was noted in a systematic review of the association of rheumatoid arthritis and vascular disease (5). This offers the intriguing possibility that even if PMR patients do have a higher vascular risk than the general population, it may be decreased or nullified by improved patient care. This may include increased screening or more aggressive management of known vascular disease risk factors, careful titration of corticosteroid therapy, or alteration of patient behavior (for instance, smoking cessation or use of aspirin). This should be explored further in future research.
A strength of this review is that our broad search strategy was able to identify studies whose primary purpose was to investigate the prognosis of GCA but that contained patients with pure PMR; however, some of these studies contained only small numbers of patients with PMR and did not separately compare patients with PMR or combined PMR and GCA to controls or population statistics, which makes it harder to draw conclusions from their results.
If only studies that reported on PMR separately were included in this review, then 3 studies would have shown PMR to have a positive association with vascular disease and 2 studies would have shown no association. However, it is likely that if PMR is associated with vascular disease, then GCA, which causes greater levels of inflammation, requires higher levels of corticosteroids, and also has direct effects on particular arteries, would share or in fact have a stronger association with vascular disease than PMR. Although there is a strong overlap between PMR and GCA, novel imaging such as positron emission tomography may enable future study designs to more accurately define PMR and GCA populations. Future studies should discriminate between PMR and GCA and should test the association between these conditions as a possible explanatory variable for any observed relationship.
Other strengths include the robust quality assessment process and the systematic assessment of heterogeneity. The method of assessing the cause of study heterogeneity in this review, while being systematic, may have been overly cautious. Since a limited number of studies were identified, to achieve statistical significance would have required all of the studies in one group to meet a quality criterion or design point and all of the studies in the other group not to meet the point. However, this technique may be useful for the systematic investigation of heterogeneity in reviews that identify large numbers of studies.
Weaknesses of this review include that screening was performed by a single reviewer; however, since it was done twice with no alteration of the results and only 1 additional study was identified by searching reference lists, it seems unlikely that additional studies were not identified. Publication bias may be limited in this review because the novelty of research into this study question would increase the likelihood of publication of studies showing nonsignificant results.The only study found by this review that was designed specifically to test whether PMR is associated with vascular disease was limited to the investigation of peripheral vascular disease (20). Applying similar methods, such as the exclusion of subjects with prediagnosis vascular events and adjusting the results for known vascular risk factors, to different populations using different outcomes such as myocardial infarction and stroke may help us to better understand the relationship between PMR and vascular disease.
In addition, it is disappointing that although PMR is primarily diagnosed and managed in a community-based setting, the majority of the research found in this review was conducted in secondary care, with the 2 Minnesota-based studies being the only ones to use community-based records (19, 20). Secondary care patients may have more severe or diagnostically ambiguous disease than those managed exclusively in the community, making results less applicable to the majority of PMR patients.
This review suggests that, similar to other inflammatory rheumatic disorders, there may be an increased risk of vascular disease in patients with PMR. Since the evidence found by this review is sparse and showed considerable heterogeneity, further studies are required to both examine this association and, if there is evidence of increased risk, examine whether this is due to atherosclerotic vascular events or other vascular events such as aneurysms.
Future studies should, where possible, discriminate between PMR and GCA and test the association between these conditions as a possible explanatory variable for any observed relationship. In addition, they should ideally adjust their analyses based on known vascular disease risk factors and potential modifiers such as use of statins or aspirin, while better representing PMR patients based in primary care.
Because PMR is a common disease in patients ages >50 years, and since this age group has a high risk of vascular events, illuminating the possible association between PMR and vascular diseases may provide evidence to change clinical practice, which may reduce the potential excess risk to patients with PMR and in turn may help to reduce the overall personal and economic burdens of vascular mortality and morbidity.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Mr. Hancock had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Hancock, Mallen, Belcher, Hider.
Acquisition of data. Hancock, Mallen.
Analysis and interpretation of data. Hancock, Mallen, Belcher, Hider.