Improving outcome prediction of systemic sclerosis from isolated Raynaud's phenomenon: role of autoantibodies and nail-fold capillaroscopy

Abstract

Objective: A simple weighted prognostic algorithm, based on capillaroscopy and autoantibodies, is developed to classify patients at different risk of transition from isolated RP to SSc within 5 years from the screening visit.

Methods: Two hundred and eighty-eight of 768 patients with isolated RP who underwent capillaroscopy were recruited. The prognostic contributions of capillaroscopic findings (giant loops, haemorrhages and the number of capillaries) and SSc-associated autoantibodies (ACAs, anti-topo I and ANAs) were assessed in a semi-parametric regression models suitable for competing risks. A prognostic index was built by a bagging technique. A structured tree approach was used to extract simple classificatory rules that can be directly interpreted.

Results: Thirty-four transitions from isolated RP to SSc and 42 to other CTDs were observed. All of the chosen variables had a substantial prognostic impact. A complex non-linear prognostic pattern was observed for capillaries, with the risk of developing SSc increasing as the number of loops decreased. The presence of ANAs had a strong impact on prognosis [hazard ratio (HR) = 9.70], which was increased by the presence of ACA (HR = 3.94; P < 0.001). A weighted prognostic classification for the development of SSc was constructed using capillary number, giant loops and ANAs. The prognostic discrimination was satisfactory (Harrell's C-index = 0.86).

Conclusion: Our prognostic model is based on easy-to-obtain features (i.e. the number of capillaries, giant loops and ANAs) and could be used to facilitate clinical decision making in the screening phase, and may also have important implications for stratifying patients into risk groups for future clinical assessment.