Distilling Cancer Biomarkers From the Serum Peptidome: High Technology Reading of Tea Leaves or an Insight to Clinical Systems Biology?

摘要

In this issue, Mian et al use high-resolution mass spec-trometry (MS) to define the serum peptidome of patients with melanoma. They trained machine learning algorithms to recognize serum peptidomic patterns that clustered with specific disease states. This strategy enabled them to discriminate stage I melanoma patients from stage IV patients. Furthermore, the peptidomic patterns were able to discriminate stage III melanoma patients who developed tumor recurrences within the first year after diagnosis from stage III patients who remained disease free. An artificial neural network analysis of the peptide patterns was able to correctly identify 80% of patients who eventually experienced recurrence, compared with a 21 % prediction rate when these investigators used the conventional tumor marker S100-beta.