Mortality risk factors in primary Sjögren syndrome: a real-world, retrospective, cohort study

Abstract

Background: What baseline predictors would be involved in mortality in people with primary Sjögren syndrome (SjS) remains uncertain. This study aimed to investigate the baseline characteristics collected at the time of diagnosis of SjS associated with mortality and to identify mortality risk factors for all-cause death and deaths related to systemic SjS activity measured by the ESSDAI score.

Methods: In this international, real-world, retrospective, cohort study, we retrospectively collected data from 27 countries on mortality and causes of death from the Big Data Sjögren Registry. Inclusion criteria consisted of fulfilling 2002/2016 SjS classification criteria, and exclusion criteria included chronic HCV/HIV infections and associated systemic autoimmune diseases. A statistical approach based on a directed acyclic graph was used, with all-cause and Sjögren-related mortality as primary endpoints. The key determinants that defined the disease phenotype at diagnosis (glandular, systemic, and immunological) were analysed as independent variables.

Findings: Between January 1st, 2014 and December 31, 2023, data from 11,372 patients with primary SjS (93.5% women, 78.4% classified as White, mean age at diagnosis of 51.1 years) included in the Registry were analysed. 876 (7.7%) deaths were recorded after a mean follow-up of 8.6 years (SD 7.12). Univariate analysis of prognostic factors for all-cause death identified eight Sjögren-related variables (ocular and oral tests, salivary biopsy, ESSDAI, ANA, anti-Ro, anti-La, and cryoglobulins). The multivariate CPH model adjusted for these variables and the epidemiological features showed that DAS-ESSDAI (high vs no high: HR = 1.68; 95% CI, 1.27-2.22) and cryoglobulins (positive vs negative: HR = 1.72; 95% CI, 1.22-2.42) were independent predictors of all-cause death. Of the 640 deaths with available information detailing the specific cause of death, 14% were due to systemic SjS. Univariate analysis of prognostic factors for Sjögren-cause death identified five Sjögren-related variables (oral tests, clinESSDAI, DAS-ESSDAI, ANA, and cryoglobulins). The multivariate competing risks CPH model adjusted for these variables and the epidemiological features showed that oral tests (abnormal vs normal results: HR = 1.38; 95% CI, 1.01-1.87), DAS-ESSDAI (high vs no high: HR = 1.55; 95% CI, 1.22-1.96) and cryoglobulins (positive vs negative: HR = 1.52; 95% CI, 1.16-2) were independent predictors of SjS-related death.

Interpretation: The key mortality risk factors at the time of SjS diagnosis were positive cryoglobulins and a high systemic activity scored using the ESSDAI, conferring a 2-times increased risk of all-cause and SjS-related death. ESSDAI measurement and cryoglobulin testing should be considered mandatory when an individual is diagnosed with SjS.

Funding: Novartis.

Keywords: Cardiovascular; Infection; Lymphoma; Mortality; Sjögren syndrome; Systemic disease.

Fig. 1

Design of the statistical approach of the study based on a directed acyclic graph including all-cause and SjS-related mortality as dependent variables, and the following variables classified as: • Independent variables (those that define the disease phenotype at the time of diagnosis, grouped into three categories: glandular involvement, immunological variables, and systemic involvement). They are the variables that characterize the disease at the time of its diagnosis and that have been related to the prognosis of the disease according to previous studies. • Confounding variables: the epidemiological variables that have been related to both the independent variables (phenotype of the disease) and the dependent variable to be studied (mortality) according to previous studies. • Competing exposure variables (socioeconomic status, non-communicable diseases): variables that can influence the dependent variable to study but for which no published evidence supports a significant influence in the disease phenotype (independent variables).

Fig. 2

Forest plot of HRs and 95% confidence interval for SjS-related death yielded by the CPH regression model treating cases of death unrelated to Sjögren as competing events (right-censored observations).

Fig. 3

Kaplan–Meier analysis and Log-rank test comparing the survival curves for SjS related-deaths (treating other causes of death as competing events) during the first 20 years of follow-up of patients: 3a) males vs females; 3b) abnormal vs normal results in oral tests at baseline; 3c) high vs non-high DAS-ESSDAI at baseline; 3d) positive vs negative cryoglobulins at baseline.

Fig. 3

Kaplan–Meier analysis and Log-rank test comparing the survival curves for SjS related-deaths (treating other causes of death as competing events) during the first 20 years of follow-up of patients: 3a) males vs females; 3b) abnormal vs normal results in oral tests at baseline; 3c) high vs non-high DAS-ESSDAI at baseline; 3d) positive vs negative cryoglobulins at baseline.

Fig. 4

Cumulative survival for all-cause (4a) and SjS-related (4b) death according to the disease activity states (DAS) scored at the time of SjS diagnosis (no activity, low activity, moderate activity, high activity).

Fig. 5

Cumulative survival for all-cause (5a) and SjS-related (5b) death according to the number of clinESSDAI domains classified as high activity at diagnosis (0, 1, 2 or more).