Psoriatic arthritis (PsA) is a chronic inflammatory arthritis associated with psoriasis that occurs in 0.3–1% of the population.1 PsA is usually seronegative for rheumatoid factor and is classified among the spondyloarthropathies.2 3 The prototype of inflammatory arthritis, rheumatoid arthritis (RA), has been extensively researched, and multiple studies reiterate the increased cardiovascular morbidity and mortality rates, with standardised mortality ratios ranging from 0.87 to 3.0.3^–12 Patients with psoriasis are also at significantly higher risk for cardiovascular deaths.13^–15 We have previously demonstrated an increased mortality risk in patients with PsA and notably, the leading cause of death was cardiovascular disease (CVD).16 Others have recently noted the increased cardiovascular risk among patients with spondyloarthritis.17 Psoriasis was found to be an independent risk factor for myocardial infarction (MI).18 These findings indicate the possibility of increased risk of cardiovascular morbidities in PsA.

It has been suggested that the chronic inflammatory nature of RA and other inflammatory rheumatic diseases is very similar to that causing atherosclerosis.19^–21 Among risk factors for coronary heart disease are modifiable factors such as smoking, diabetes, hypertension, increased total low-density lipoprotein and decreased high-density lipoprotein cholesterol levels, obesity, sedentary lifestyle and fixed factors such as age and gender.19 However, additional risk factors, such as apoprotein B, Lp(a) lipoprotein, fibrinogen, C-reactive protein (CRP), soluble intercellular adhesion molecule-1, homocysteine, plasminogen activator inhibitor-1 and matrix metalloproteinase-3, may play a role in chronic inflammatory diseases.22 Treatments such as disease-modifying antirheumatic drugs, or some non-steroidal anti-inflammatory drugs (NSAIDs) such as cyclo-oxygenase-2 (COX-2) inhibitors may also exacerbate heart disease.21 While studies on the direct risk of CVD in PsA are few, the indirect evidence of a potential link is significant and suggests that further exploration of this topic would be beneficial.

The aim of this investigation was threefold: we sought to describe CVD in our PsA cohort, to determine the prevalence of CVD in comparison with the general population and to examine the risk factors of CVD relative to disease severity in PsA.

METHODS

Setting

The University of Toronto Psoriatic Arthritis Clinic in the Centre for Prognosis Studies in Rheumatic Diseases was founded in 1978 as a prospective longitudinal observational cohort of patients with PsA.23 The cohort consists of patients with PsA who are referred to the clinic by family doctors and other medical specialists. Patients followed up in the clinic may receive their primary rheumatological care at the clinic or may attend once or twice a year and continue to be followed up by their community rheumatologists. The clinic thus includes the spectrum of PsA, from mild to very severe disease.

Clinical assessments

Each assessment consisted of a medical history, physical examination and laboratory evaluation, including items related to PsA as well as comorbidities and known risk factors. A standard protocol, completed at each clinic visit at 6–12-month intervals, has been entered into a computerised database.

Patient selection

Patients included in this study were registered at the clinic between 1 January 1978 and 1 June 2004. The database was used to identify patients with cardiovascular morbidity. Cardiovascular disease comprised hypertension, angina, MI, cerebrovascular accident (CVA) and congestive heart failure (CHF). Hypertension was defined as a blood pressure >140/90 mm Hg on more than two consecutive occasions, or use of antihypertensive agents by the patients. MI was defined as one of the following: definite ECG abnormalities; typical symptoms with probable ECG abnormalities and abnormal enzymes (⩾2 upper limit of normal); typical symptoms and abnormal enzymes.

Angina was defined as severe pain or discomfort over the upper or lower sternum or anterior left chest and left arm, of short duration relieved by rest or vasodilators. A cerebrovascular event was defined as an abrupt onset of neurological dysfunction resulting in neurological damage. Additional information was sought from the patient charts to ensure accuracy and completeness of the database, since check-up visits may occur between protocol visits. Moreover, documentation of the clinical events was obtained from hospital records and primary care doctors. Dates of diagnosis for each type of event were obtained. Deaths attributable to cardiovascular causes were recorded. The procedures used to identify causes of death have been previously described.16

Predictive factors

Known risk factors included age, gender, daily alcohol use, smoking, raised triglycerides, raised cholesterol, hyperuricaemia and diabetes. Other risk factors relating to PsA disease severity at first clinic visit were age at onset (psoriasis and PsA), disease duration (psoriasis and PsA), actively inflamed joint count, clinically deformed joint count, Psoriasis Area and Severity Index (PASI) score and erythrocyte sedimentation rate (ESR).

For a comparison group, comparative data on CVD was obtained from the Canadian Community Health Survey (CCHS), a cross-sectional examination of health determinants, health status and health system use, carried out by Statistics Canada in 2000–1.24 This survey, conducted by telephone and computer-assisted personal interviews, took place at health region level and provincial level; 39 278 people from Ontario aged ⩾12 years were interviewed. The sample was weighted, and results extrapolated to the Ontario population were used for analysis. Available summary tables are subdivided by age group and gender, subject to sample sizes meeting release guideline criteria of Statistics Canada. Otherwise, data are pooled to meet these guidelines.

Statistical analysis

The prevalence of CVD morbidities in patients followed up in the PsA cohort was compared with CCHS data through standardised prevalence ratios (SPRs) calculated in the same fashion as the more familiar standardised mortality ratios. The prevalence of CVD morbidities in the PsA cohort was estimated over the same time period (ie, up to 2000–1) with matched gender and age groups as was done for the cross-sectional CCHS. Specifically, all deaths before 1 January 2000 were excluded and CVD events that occurred before 31 December 2001 were included for the SPR calculation. Additionally, where gender-specific information was not available for particular age categories in the CCHS for particular morbidities, only the corresponding age categories in the PsA cohort were collapsed over gender when calculating the SPRs for those particular morbidities.

Owing to long follow-up period in the PsA cohort, we also investigated the possible differential effects of risk factors over calendar time. Interaction terms between calendar time and risk factors were examined individually. In addition to medication variables defined as ever-used only, we also checked the medication variables defined as whether used since last visit, when possible. For hypertension, age at onset of psoriasis and steroids since last visit were found to have possible interactions with calendar time that disappeared after multiplicity adjustment. Similarly, ESR and gender for MI, as well as age at onset of psoriasis, steroids since last visit and PASI for first CVD event were inconclusive evidence for differential effects over calendar time.

Risk factor analyses were undertaken using Cox relative risk regression analysis with time-dependent explanatory variables and age as the chosen time scale.25 Left truncation was incorporated into the Cox model. Data analyses included patients’ records when information of the relevant potential risk factors was available from the previous clinic visit. However, if laboratory tests measurements were not available at previous visit, the last recorded test measurements were used and the appropriate patients’ records included. Because PASI was recorded for patients only since 1991, missing values for this variable were handled by introducing an unknown PASI classification. This allowed more patients’ records to be included in analyses, but the qualitative findings were not sensitive to this decision. Additionally, to alleviate some of the problems that would result from missing information on medication information since the last visit, in particular NSAIDs, ever-used medication variables were constructed. The logarithmic transformation of a laboratory measurement was chosen to assess that covariate’s effect on risk of having a CVD event if it either lessened the impact that a few extreme/outlying observations of the covariate may have on the results, thus making them more robust, or made the effect of the covariate approximately linear on the log hazard scale.

RESULTS

Between 1 January 1978 and 16 June 2004, 648 patients were registered in the University of Toronto PsA clinic database; 364 (56.2%) were male and 284 (43.8%) were female. Mean age at first visit was 43.5 years and at last visit was 51.8 years. Average duration of follow-up was 8.3 years. At first visit, durations of psoriasis and PsA were 14.6 years and 7.4 years; mean initial active joint count was 10 and mean initial PASI score was 5.5 for those enrolled since 1991 (table 1).

The number of cardiovascular morbidities observed at any time (including events before entry) in patients from the PsA cohort was as follows: hypertension 206, MI 50, angina 33, CVA 8 and CHF 12. Overall, 227 people had at least one of these conditions, corresponding to a prevalence of 35.0%. Increased prevalences were observed for hypertension, MI and angina with overall gender and age-adjusted SPRs of 1.9, 2.6 and 2.0, respectively (table 2), where the corresponding numbers of observed hypertension, MI and angina up to 2001 were 122, 25 and 18, respectively.

Unfortunately, for the youngest age group (ie, for MI and angina: age 12–44; and for CHF: age 12–54), gender-specific information on MI, angina and CHF was not available in the CCHS data. Therefore gender-specific SPRs could be calculated only for hypertension and CVA. For male and female subjects, the SPRs for CVA were 1.0 (95% CI 0.3 to 4.1) and 0.8 (95% CI 0.2 to 3.3) respectively, whilst the gender-specific SPRs for hypertension were 2.1 (95% CI 1.7 to 2.6) and 1.7 (95% CI 1.3 to 2.2) for male and female subjects, respectively.

We performed sensitivity analyses to examine the results for SPRs. For hypertension, after redefining the 39 cases that were solely identified by consecutive high blood pressure measurements in the PsA cohort after the first visit to be non-hypertensive, the estimated SPR was reduced to 1.29 (95% CI 1.04 to 1.60). These 39 excluded patients in the PsA clinic represent those patients who might be identified as having hypertension through the more regular follow-up that takes place in the Toronto PsA clinic. For MI and angina, less likely to be over-reported owing to regular follow-up, we examined the potential effect of reclassification on reported significance tests. The observation of just 16 or more of the 25 MI cases and 15 or more of the 18 angina cases would provide statistical evidence for elevated prevalences in 0.05 significance level tests.

After clinic entry, 122 clinic patients with PsA developed hypertension, 38 had an MI and 5, 21 and 11 had CVA, angina and CHF, respectively. A total of 155 patients had at least one of these conditions observed after clinic entry.

Table 3 presents single risk factor time to event analyses for hypertension, MI and any CVD conditions. Small numbers of events precluded informative analyses of the other specific cardiovascular conditions. Effects for diabetes, triglycerides and PASI are evident for the hypertension and first CVD event analyses. With smaller numbers the only risk factor found for MI were smoking and triglycerides.

Table 4 presents multivariate time to event analyses. For hypertension and first CVD event, the interactions for age at onset of psoriasis and steroids since last visit with calendar time were not evident. However, for first CVD event, the interaction between PASI and calendar time is borderline significant (p = 0.05), thus suggesting a possible changing effect of PASI over calendar time. An increased effect of PASI on hypertension was seen for a PASI score greater than a threshold of 20. It can also be concluded that diabetes and triglycerides are significantly related to hypertension and the occurrence of any CVD event. If the interaction of steroid use since last visit and calendar time is excluded, then the main effects of steroid use since last visit are found not to be significant in the models for hypertension and first CVD event (p = 0.11 and 0.18, respectively).

For MI, the differential effect of gender and ESR over calendar time is borderline statistically significant (p = 0.04). Smoking and triglycerides were both found to be positively related to the risk of experiencing MI.

DISCUSSION

Much evidence has been accumulating surrounding the link between rheumatic diseases and CVD. Studies have found increased rates of cardiovascular death in RA, systemic lupus erythematosus, ankylosing spondylitis, psoriasis and PsA.4 14 16 20 26 CVD is among the leading causes of death among patients with PsA. Recent evidence suggests that patients with PsA are at an increased risk for developing CVD.27^–29 A study from the integrated outcomes database identified 3066 patients with PsA who were matched 1:4 with subjects in the database on the basis of age, sex, location and length in the plan. The prevalence ratios of peripheral CVD (1.6), CHF (1.5), atherosclerosis (1.4), ischaemic heart disease (1.3), cerebrovascular disease (1.3) and hypertension (1.3) were higher in patients with PsA than controls. They also found that risk factors for coronary artery disease such as hypertension, diabetes and hyperlipidaemia produced higher prevalence ratios in patients with PsA (1.3, 1.5, 1.2, respectively) than in controls.28

We have now found an indication of such effects in patients with PsA followed up prospectively. For hypertension, MI and angina, prevalences were significantly higher in the PsA cohort than data from the CCHS, with standardised prevalence ratios of 1.9, 2.57 and 1.97, respectively. Higher prevalences of risk factors for coronary artery disease were also reported for patients with psoriasis compared with the general population, although it is not clear how many of those patients had PsA.30 Similar observations were previously reported from our PsA cohort.31 Also, two recent studies have shown an increased frequency of subclinical atherosclerosis, expressed by endothelial dysfunction and increased carotid artery intima-media thickness, in patients with PsA without clinically evident CVD or classic cardiovascular risk factors.32 33 Interestingly, in keeping with our results, these authors did not find any association between the severity or the pattern of joint involvement and the development of subclinical atherosclerosis. No information about the severity of psoriasis skin involvement was given in these two studies. However, we did observe a role for high PASI scores representing more severe psoriasis. Severe psoriasis was found to be an independent risk factor for MI in a population study of patients with psoriasis.18 Possibly, the additive effect of PsA by itself as a chronic inflammatory disease together with a higher frequency of traditional cardiovascular risk factors in these patients may account for the increased incidence of cardiovascular events seen in patients with PsA.

We have previously reported that ESR and radiological damage are associated with increased relative risk values for deaths related to the circulatory system.34 However, no substantial evidence for a role for ESR emerged in our current analysis. Traditional Framingham factors, high triglyceride levels and presence of diabetes, were found to be prognostic for cardiovascular morbidity.

Chronic inflammation with perpetually raised levels of proinflammatory cytokines and immune cells can lead to endothelial dysfunction and formation of atherosclerotic plaque.4 19 Interleukin 6 and tumour necrosis factor can also induce hepatic synthesis of CRP, a serum marker of inflammation associated with heart disease.4 Anti-inflammatory drugs can also contribute to the development of cardiovascular disease.35 Glucocorticoids, although anti-inflammatory, can lead to hypertension, hyperglycaemia and a negative lipid profile, and NSAIDs, similarly, raise blood pressure and also may promote thrombosis through COX-2 inhibition.35 Elderly patients receiving rofecoxib were at increased risk of congestive heart failure.36 Methotrexate affects homocysteine levels, which is a risk factor for atherosclerosis.4 37 In patients with arthritis limited movement and sedentary lifestyle may contribute, since regular exercise is a protective factor against heart disease.4 In psoriasis, abnormalities of coagulation and fibrinolysis and raised homocysteine levels have been documented.38 39

Many of these measures might be useful prognostic factors in predicting heart disease, but our data were primarily rheumatological and not cardiological. As a result, we did not have data on body mass index or CRP. In addition, the true morbidity rate is difficult to measure, even with data from regular clinic visits, since it was found that up to half of ischaemic heart disease in patients with RA is clinically silent.40 Events in our study were recorded by patient self-reporting at clinic visits, protocol physical examinations and possible referrals for consultations on symptoms presenting at clinic visits. An appropriate control group is also important, which was the reason for using data from the CCHS in our study. The community cohort in the CCHS is comparable to our outpatient PsA clinic and they were compared over the same time period. Collection of data in the CCHS was systematic, and data specific to the province of Ontario was readily available. However, it should be noted that while the patients with PsA were followed up prospectively, the information on the controls was collected cross-sectionally.

Possibly, significance in the other outcomes was not detected owing to small numbers of events (ie, lack of power), although this group of patients with PsA is the largest group followed up prospectively to date. However, unlike RA, there is no national patient registry of patients with PsA.

Our clinic serves as a primary, secondary and tertiary referral centre and includes the whole spectrum of PsA, from mild to severe disease, thus it is representative of patients with PsA who see rheumatologists. Moreover, our cohort is similar in demography and disease characteristics to others reported in the literature.1 As far as we know, this is the first study to examine cardiovascular morbidity among patients with PsA followed up prospectively. Our findings show significantly increased risk of hypertension, MI and angina in PsA. However, our analyses were exploratory, and multiple testing may inflate type I error rates. Even so, our results are in accord with findings from studies in other rheumatic diseases, suggesting that the inflammatory process in PsA also predisposes these patients to these comorbidities.5 6 10 25 These findings need to be confirmed with other large cohorts of patients with PsA followed up prospectively. Moreover, attempts should be made to record and deal with the risk factors identified in this study in all patients with PsA.