Methods

We first searched the EUSTAR database which has previously been described in detail.8 LVEF was measured echocardiographically at each participating centre using Simpson’s method and was classified as depressed if <55% or <50%. In a second phase, a case-control study of a patient subset was performed to further identify independent factors associated with LV dysfunction by simple and multiple regression including parameters not registered in the EUSTAR database. Details of the methods and statistical analysis used are given in file 1 in the online supplement.

Results

By April 2008, 7283 patients had been enrolled. Since neither LVEF nor the presence of LV dysfunction was recorded in 209, this analysis includes data collected on 7073 patients (mean (SD) age 56 (14) years, 981 (13.9%) men). A total of 383 patients had a reduced LVEF, corresponding to a prevalence of 5.4%. The main characteristics of patients with and without LV dysfunction are shown in table 1.

Table 2 shows the ORs (95% CI) of factors associated with LV dysfunction.

It is noteworthy that men presenting with diffuse cutaneous SSc and active or past digital ulcerations, representing 203 patients (2.8%) in this study, were at a particularly high risk of LV dysfunction (OR 3.2; 95% CI 2.1 to 4.9).

Case-control analysis

The second phase of the study included 385 patients (129 with LVEF <55%) and 256 controls. Diffuse cutaneous SSc was present in 145 patients, pulmonary fibrosis in 176, past or active digital ulcer in 204 and a history of renal crisis in 6 patients. Echocardiography revealed the presence of a systolic pulmonary artery pressure >40 mm Hg in 114 patients, of whom 34 had precapillary pulmonary arterial hypertension (PAH) confirmed by cardiac catheterisation. Associated factors with reduced LVEF according to univariate and multivariate analyses are shown in table 3.

We conducted similar analyses for LVEF <45% (n = 36) which confirmed the results shown in table 3 (see file 1 in online supplement).

Discussion

The main observations made in this analysis were (1) a 5.4% prevalence of LV dysfunction and (2) the identification of age, male gender, myositis, digital ulcers, lung involvement and absence of previous treatment with calcium channel blockers as associated factors with reduced LVEF.

Few large studies have evaluated LV dysfunction. In a multicentre study of PAH, a post hoc analysis identified only 8 of 570 patients with SSc (1.4%) with LVEF <45%.2 The EUSTAR database offers a unique opportunity to study the complications of SSc. The 5.4% prevalence of reduced LVEF measured in the first phase of our study is concordant with the 7.2% prevalence in patients with diffuse SSc and 5% in patients with limited SSc observed in the first report from EUSTAR.8 A depressed LVEF is therefore one of the four main manifestations of major organ involvement in SSc, together with PAH,9 10 11 renal crisis12 and interstitial lung disease.

The second objective was the identification of factors associated with a reduced LVEF. We used a two-step strategy that included (1) an analysis of the entire EUSTAR database, allowing the inclusion of robust factors unequivocally pertinent considering the large number of patients included; and (2) a nested prospective case-control study of a subgroup that included more data such as atherosclerosis risk factors, drug regimen and the presence of precapillary PAH. In order to limit the impact of disease duration—a typical contributor to organ dysfunction—we matched the two study groups for disease duration. Both analyses confirmed that male gender, age, muscle involvement and digital ulceration were independent associated factors with LV dysfunction. The association between digital ulceration and cardiac dysfunction may be a manifestation of the diffuse microvascular lesions which characterise the disease.1 The significant association between systolic pulmonary artery pressure and reduced LVEF observed in the first part of the study should be interpreted with caution since systolic pulmonary artery pressure is influenced by LV function, and this is emphasised by the lack of an independent association with precapillary PAH found in the second part of our study. In the nested study, except for male gender, typical cardiovascular risk factors were not associated with a reduced LVEF. This observation is concordant with a previous report of a predominant contribution of microangiopathy—compared with the controversial results regarding atherosclerosis—in the development of primary myocardial involvement in SSc.1 Another important finding of the nested study was the markedly lower proportion of patients with reduced LVEF who had previously been treated with calcium channel blockers. This is in agreement with previous short-term studies.1 In addition, Steen et al reported that patients presenting with PAH were significantly less often treated with calcium channel blockers than patients without PAH13 and, in another study, the development of digital ulcers was also delayed by vasodilator therapy.14 Altogether, these observations suggest that calcium channel blockers may protect against microvascular complications. Pending the results of a long-term prospective study, the broad use of calcium channel blockers in patients with SSc, unless contraindicated, should be strongly considered.

Our study is limited by its observational design and, since we studied patients with an LVEF <55% which has not been associated with an increased mortality in patients with SSc, our results do not apply to mortality. We expect the prospective follow-up that is ongoing to enable the identification of predictors of LV dysfunction and outcome in patients with SSc with mildly decreased LVEF.