Abstract
In search for novel markers for breast cancer, we aimed to identify and validate novel serum protein profiles specific for breast cancer, and assess the influence of clinical (subjects age) and pre-analytical (sample storage duration) variables on the constructed classifiers. To this end, sera of breast cancer patients (n=152) and healthy controls (n=129), randomly divided into a training and test set, were analysed by surface-enhanced laser desorption/ionisation time-of-flight mass spectrometry (SELDI-TOF MS). In the training set, 14 peak clusters were found to differ significantly in expression between cases and controls. None of the peak clusters were influenced by subjects age and sample storage duration. Ten peak clusters were also found significantly discriminative in the test set. Peak clusters were structurally identified as C3a des-arginine anaphylatoxin, (tentative) inter-alpha-trypsin inhibitor heavy chain 4 fragments and a fibrinogen fragment. Logistic regression analyses on the training set yielded a classification model with a moderate performance on the test set, corresponding to those reported in previously performed validation studies. Most likely originating from the highly heterogeneous nature of breast cancer, selection of breast cancer subgroups for comparison with healthy controls is expected to improve results of future diagnostic SELDI-TOF MS studies.