[2009] [FRI0319] CLINICAL PREDICTION OF 5-YEAR SURVIVAL IN EARLY SCLERODERMA: VALIDATION OF A SIMPLE PROGNOSTIC MODEL IN A MULTI-CENTRE STUDY

J. Fransen¹, D. Diaconu¹, R. Hesselstrand², P. Carreira³, G. Valentini⁴, L. Beretta⁵, P. Airò⁶, M. Inanc⁷, S. Ullman⁸, A. Balbir-Gurman⁹, S. Sierakowski¹⁰, L. Czirjak¹¹, V. Riccieri¹², R. Giacomelli¹³, A. Gabrielli¹⁴, G. Riemekasten¹⁵, M. Matucci-Cerinic¹⁶, D. Farge¹⁷, N. Hunzelmann¹⁸, F.H.J. Van den Hoogen¹⁹, M.C. Vonk^{1 1}Rheumatology, Radboud University Nijmegen Medical Centre, Nijmegen, Netherlands; ²Rheumatology, Lund, Sweden; ³Rheumatology, Hospital de 12 octobre, Madrid, Spain; ⁴Rheumatology, Napoli; ⁵Rheumatology, Milano; ⁶Rheumatology, Brescia, Italy; ⁷Rheumatology, Istanbul, Turkey; ⁸Dermatology, Copenhagen, Denmark; ⁹Rheumatology, Haifa, Israel; ¹⁰Rheumatology, Bialystok, Poland; ¹¹Rheumatology, Pecs, Hungary; ¹²Rheumatology, Roma; ¹³Rheumatology, L'Aquila; ¹⁴Rheumatology, Ancona, Italy; ¹⁵Rheumatology, Berlin, Germany; ¹⁶Rheumatology, Florence, Italy; ¹⁷Internal medicine, Paris, France; ¹⁸Rheumatology, Cologne, Germany; ¹⁹Rheumatology, Nijmegen, Netherlands

Background: Systemic sclerosis (SSc) is associated with a significant reduction in life expectancy. When patients with a high risk of SSc related death could be identified already at diagnosis, this opens an opportunity for early and intensive treatment. A simple prognostic model to predict 5-year survival has already been developed in a British tertiary referral centre in 1999 in 280 patients, but it was not validated in other samples. Usually, the predictions of a prognostic model are less accurate when the model is applied to new patients, especially from other centres or countries.
Objectives: To validate the prognostic model to predict 5-year survival in SSc provided by Bryan et al. (1999) in other SSc centres throughout Europe.
Methods: A European multi-centre cohort of SSc patients diagnosed before 2002 was established. Data were collected in daily clinical practice and stored in patient charts or clinic-owned databases. Patients with Scleroderma according to the preliminary ACR criteria were eligible for this study when they were followed for at least five-years or shorter if they died. Patient-observation time started at the moment the diagnosis of SSc was secured (baseline) and ended at 5-years follow-up or at the moment of death, whichever came first. The primary outcome was 5-year survival after diagnosis of SSc. For the prognostic logistic regression model, the following baseline variables were used: age, gender, urine protein presence, ESR, and carbon monoxide diffusing capacity (DLCO).
Results: Complete data were available for N=618 patients, 59 (9.5%) of them died within 5 years after diagnosis. Of the patients there were 85% female, mean (SD) age at diagnosis was 50 (14), and 33% were classified as having diffuse SSc. At diagnosis, the median (P25-P75) ESR was 16 (9-32), elevated ESR (>25) was present in 203 (33%) patients, urine protein was present in 39 (6.3%) patients, the median (P25-P75) DLCO in '% expected' was 77 (63-90), low DLCO (<70%) was present in 224 (36%) patients. The prognostic model with age (OR 1.04), gender (OR 1.92), urine protein (OR 2.43), elevated ESR (1.79) and low DLCO (OR 2.64) had an area under the ROC curve of 0.76. Of the patients with no risk factors, 7/277 (2.5%) died, with 1 risk factor 29/230 (13%) died, with 2 risk factors 20/97 (21%) died, and with 3 risk factors 3/14 (21%) died.
Conclusion: A simple prognostic model using 3 disease factors to predict 5 year survival at diagnosis in SSc showed reasonable discriminatory performance upon validation in a European multi-centre study. Update of the formula may increase discrimination and precision of the predictions.
References:

1. Bryan C, Knight C, Black CM, Silman AJ: Prediction of five-year survival following presentation with scleroderma - Development of a simple model using three disease factors at first visit. Arthritis and Rheumatism 1999;42(12):2660-5.