Background Systemic sclerosis (SSc) remains a devastating condition associated with a high risk of mortality. However, the main causes of death and risk factors for mortality are only partially known.
Objectives To investigate causes and independent predictors of death in a large European population of SSc-patients.
Methods We performed a longitudinal observational study using the latest 2014 data extract from the EUSTAR cohort. We looked at (i) deaths and causes of death (SSc, non-SSc) and (ii) effect of demographics, disease characteristics and comorbidities on mortality. Investigators from each center were asked to indicate the single primary cause of death according to standardized definitions. Associated factors with mortality were analyzed first in univariate analysis using t-test, chi2 or fisher exact test as necessary, followed by multivariable Cox modeling. Based on cox multivariate analysis, a simple score was generated (SScore) by adding the integer of each hazard ratio of variables independently associated with mortality. The ability of the SScore was compared to the previous modified 5-factor score to predict overall and 5-years mortality, and SSc-related mortality during follow-up using C-statistics.
Results A total of 11,194 SSc-patients were included: 86% of women, 31% of diffuse cutaneous forms, and mean disease duration: 5.9 [2.4–11.8] years. During mean follow-up of 27±33 months, 1072 (9.6%) patients died. Of these, 617 patients (58%) died from SSc-related cause, 270 from non-SSc cause (25%), whereas the cause of death remains unknown in 185 (17%). Main causes of death were lung fibrosis (14.7%) followed by PAH (12.9%), primary heart involvement (10.5%), cancer (11.5%), infections (8%) and atherosclerosis (4.8%).
The SScore comprised the following predictors: male gender (2 points), upper gastrointestinal involvement (2), proteinuria (5), LVEF<50% (3), DLCO<60% of predicted (2). Each increase of 1 point of the SScore was associated with a 35.9% [31.7–40.3] increase in mortality at 5 years. Areas under the ROC of the SScore to predict total mortality and SSc-related mortality were 0.70 (0.68–0.82) and 0.74 (0.72–0.76).
The accuracy of the previous 5-factor score was lower than the SScore to predict total mortality (p=0.005) and 5 years-mortality (p=0.003).
Further analyses looked at various subsets of SSc: by example, DcSSc patients had reduced survival as compared to LcSSc and causes of death were different: in DcSSc, the main causes of death were pulmonary fibrosis (18.6%), primary heart involvement (13.3%) and cancer (10.0%), whereas PAH (18.1%), cancer (12.8%) and pulmonary fibrosis (11.6%) were the leading causes in LcSSc (Figure).
Conclusions Using the largest worldwide available cohort of SSc patients in which 1072 deaths have been recorded and characterized, we provide insight into the causes of deaths and emphasis on the contribution of the cardiopulmonary system.
In addition, we have established a new score to estimate the 5-year survival of SSc patients.
Disclosure of Interest None declared