INTRODUCTION Multi photon detection (MPD) of 125-I -streptavidin increases the sensitivity of immunassays by 500-1000x. (∼1 fg/ml). This allows accurate measurement of low-abundance cytokines in peripheral blood. These techniques allow data on whole patient cohorts to be analysed. MATERIALS and METHODS. Three different groups of cytokines (angiogenic -vEGF and IL-8, cancer-specific - PSA and immune system markers IL-6 and TNFα ) were measured in normal serum (n=95)and in patients with untreated breast cancer (n=264). Non-Gaussian distribution of correlates requires novel statistical methods. RESULTS. Serum concentrations of individual cytokines ranged over 6 logs but using MPD and a panel of 5 biomarkers 252/264 (95.4%) women with breast cancer were correctly identified. 5/264 fell in the uncertain range and 7/264 (2.7%) were wrongly classified. 89/95 healthy women were correctly classified with 2/95 being misclassified. DISCUSSION. Parallel measurement of multiple markers in patients allows accurate discrimination between normal and malignant sera. There is an age-dependant change in cytokine expression and these results are strongest in younger women. For many of the potential biomarkers for which MPD assays have been established, overlap of serum concentrations between healthy and breast cancer patients has been observed. Analysis of correlations between multiple biomarkers provides much more predictive power than analysis of single biomarkers for detection of breast cancer or other epithelial cancers (ovarian, prostate, melanoma). Diagnostic power is potentiated by the use of panels of biomarkers that include not only tissue markers, but also systemic markers such as inflammatory and/or angiogenic factors. Many of the most informative proteins, e.g. PSA, are very low abundance proteins in serum. Since the coverage of each biomarker multiplicatively reduces cohort coverage the use of correlations between different proteins requires high coverage of patient cohorts for each biomarker. This suggests that very high sensitivity coverage of complete patient cohorts, accurate quantitation over several orders of magnitude of the abundance of proteins and the ability to quantitatively measure multiple biomarkers concurrently will all be essential elements of successful methodologies for serum-based diagnostic proteomics.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]