Pos1331 development and validation of a clinical score for mortality prediction in systemic sclerosis: the sscore study

Background Systemic sclerosis (SSc) is one of the most lethal systemic autoimmune diseases. Considering recent advances in drug therapy and safety concerns, predicting mortality may contribute to a personalized treatment decision [1,2] . Objectives Our goal was to develop and validate a clinical score for mortality prediction in SSc patients (SScore) Methods For SScore development, a systematic literature review (SLR) in PubMed and EMBASE (up to July 2022) was performed by two independent reviewers and hazard ratios (HR) from clinically relevant and statistically significant 5-year SSc mortality predictors were meta-analyzed. Considering heterogeneity (I2 test) and publication bias (Egger’s test and funnel plots), we selected the most representative HR of each predictor. The weighted risk of death was estimated by Poisson’s regression model, and the final SScore was the sum of each predictor score, varying from 0 to 1. For SScore validation, 91 SSc patients (ACR-EULAR 2013 or LeRoy and Medsger classification criteria) were consecutively included in a prospective cohort and followed-up between 2019 and 2022 in the outpatient clinics of two tertiary hospitals. Death was confirmed by medical records and phone calls. Results Out of 6,893 studies (2,539 in PubMed; 4,354 in EMBASE), 33 were included in the meta-analysis [3,449 deaths in 15,403 SSc patients over 8.6 years (mortality 22.4%)]. Six predictors were included in SScore: age, dyspnea, malignancy, renal impairment, anti-Scl70 and scleroderma renal crisis (Table 1). In the validation cohort, there were 13 deaths (mortality 14.3%) among the 91 SSc patients over 2.6 years [age 60.5 (±10.3) years-old; women 92.3%; 81.2% white; SSc duration 13.0 (±8.1) years; diffuse cutaneous SSc 22.0%]. The best SScore cut-off was 0.18 (sensitivity 61.5%, specificity 87.2%, positive and negative predictive values 44.4% and 93.2%, respectively, and accuracy 79.1%). The area under the receiver operating characteristic curve (Figure 1) was 0.78 (95%CI 0.64-0.92). SScore was higher among those who died (p<0.001). Table 1 : Clinical score for mortality prediction in systemic sclerosis. Predictor Relative risk (95%CI) Score Age ≥ 60 anos 1.85 (0.61-5.56) 0.10 NYHA functional classes III and IV 2.18 (0.60-7.96) 0.15 Malignancy 5.33 (2.02-14.1) 0.30 Renal impairment 3.60 (1.28-10.1) 0.20 Anti-Scl70 1.89 (0.50-7.06) 0.10 Scleroderma renal crisis 2.44 (0.45-13.2) 0.15 NYHA (New York Heart Association) functional classes III and IV: dyspnea, palpitation or exhaustion with mild physical activity/at rest; renal impairment: creatinine elevation/proteinuria. Figure 1 : Receiver operating characteristic curve for death prediction by SScore. Conclusion An accessible score for SSc mortality prediction (SScore) was developed from a SLR and validated in a real-life scenario with high specificity and negative predictive value and acceptable accuracy. Larger cohort studies are needed to clarify SScore usefulness in clinical practice, including treatment individualization. References [1]Khanna D, et al. Systemic Sclerosis-Associated Interstitial Lung Disease: How to Incorporate Two Food and Drug Administration-Approved Therapies in Clinical Practice. Arthritis Rheumatol. 2022 Jan;74(1):13-27. doi: 10.1002/art.41933. [2]Bukiri H, et al. Current advances in the treatment of systemic sclerosis. Curr Opin Pharmacol. 2022 Jun;64:102211. doi: 10.1016/j.coph.2022.102211. Acknowledgements: NIL. Disclosure of Interests None Declared.