S100B protein detected in the serum of patients with cutaneous melanoma has been long reported as a prognostic biomarker. However, no consensus exists on its implementation in the routine clinical setting. This study aimed to comprehensively and quantitatively summarize the evidence on the suitability of serum S100B to predict patients' survival. Twenty-two series enrolling 3393 patients with TNM stage I to IV cutaneous melanoma were reviewed. Standard meta-analysis methods were applied to evaluate the overall relationship between S100B serum levels and patients' survival (meta-risk). Serum S100B positivity was associated with significantly poorer survival (hazard ratio [HR] = 2.23, 95% CI: 1.92-2.58, p < 0.0001). Between-study heterogeneity was significant, which appeared to be related mainly to dissemination bias and the inclusion of patients with stage IV disease. Considering stage I to III melanoma (n = 1594), the meta-risk remained highly significant (HR = 2.28, 95% CI: 1.8-2.89; p < 0.0001) and studies' estimates were homogeneous. Subgroup analysis of series reporting multivariate survival analysis supported S100B as a prognostic factor independent of the TNM staging system. Our findings suggest that serum S100B detection has a clinically valuable independent prognostic value in patients with melanoma, with particular regard to stage I-III disease. Further investigation focusing on this subset of patients is justified and warranted before S100B can be implemented in the routine clinical management of melanoma.
(c) 2008 Wiley-Liss, Inc.