Evaluation of auto-antibody serum biomarkers for breast cancer screening and in silico analysis of sero-reactive proteins

Aberrantly expressed proteins in tumours evoke an immunological response. These immunogenic proteins can serve as potential biomarkers for the early diagnosis of cancers. In this study, we performed a candidate marker screen on macroarrays containing 38,016 human proteins, derived from a human fetal-brain expression library, with the pools of sera from breast cancer patients (1 pool of benign samples, 3 pools of ductal carcinoma and 2 pools of lobular carcinoma) and 1 pool of sera from healthy women. A panel of 642 sero-reactive clones were deduced from these macroarray experiments which include 284 in-frame clones. Over-representation analyses of the sero-reactive in-frame clones enabled the identification of the sets of genes over-expressed in various pathways of the functional categories (KEGG, Transpath, Pfam and GO). Protein microarrays, generated using the His-tag proteins derived from the macroarray experiments, were used to evaluate the sera from breast cancer patients (24 malignant, 16 benign) and 20 control individuals. Using the PAM algorithm we elucidated a panel of 50 clones which enabled the correct classification prediction of 93% of the breast-nodule positive group (benign & malignant) sera from healthy individuals’ sera with 100% sensitivity and 85% specificity. This was followed by over-representation analysis of the significant clones derived from the class prediction