Beyond screening for pulmonary arterial hypertension: The DETECT score is a potential promising prediction tool for all-cause mortality in systemic sclerosis: Analysis from the EUSTAR database

Abstract

BACKGROUND: Systemic sclerosis (SSc) is characterized by an increased mortality. Various mortality risk factors are included in the DETECT algorithm, a screening tool for SSc-associated pulmonary arterial hypertension. We tested the DETECT score as a predictor of all-cause mortality in SSc. METHODS: SSc patients from the European Scleroderma Trial And Research (EUSTAR) cohort, with available data for calculating the DETECT and the SCOpE (currently proposed risk algorithm) scores and follow-up were included. Patients from the University Hospital Zurich served as derivation cohort, the remaining EUSTAR patients formed the validation cohort. Uni- and multivariable Cox regression tested the DETECT score as a predictor of mortality. A time-dependent ROC curve analysis was used to assess predictive accuracy (at 1, 3, and 5 years), and to derive and validate optimal cutoffs. RESULTS: The derivation cohort (n = 605) showed less cardio-pulmonary and diffuse cutaneous involvements, but longer follow-up and higher mortality than the validation cohort (n = 1017). The DETECT score independently predicted mortality in both cohorts, even after excluding pulmonary hypertension patients. Time-dependent ROC analysis showed excellent predictive accuracy for mortality (AUC>0.85) in the derivation cohort, non-inferior to the SCOpE score. In the validation cohort, a moderate-to-good performance for 1-year mortality was retained. A DETECT score>40 demonstrated strong performance (sensitivity≥0.68; specificity≥0.83) in the derivation, and performed moderately in the validation cohort (sensitivity = 0.54; specificity = 0.71). CONCLUSION: The DETECT score robustly predicts all-cause mortality in SSc across phenotypically different cohorts. A DETECT score>40 may refine risk stratification, guiding tighter monitoring and management. Further validation over 1-year outcomes is warranted.