This discrepancy could be due to the use of different disease types versus ductal carcinoma in situ, different sample materials, and different technical platforms. More importantly, we have conducted a pre–validation of the high throughput screening discovered protein biomarkers using an independent sample set. Interestingly, CA6, which was validated in our study, was also discovered in this previous proteomic profiling study using saliva samples from noninvasive breast cancer patients , indicating the potential of this biomarker for the early detection of breast cancer. In order to obtain a more realistic estimate of the clinical utility of the validated biomarkers, and avoid the consequences of potential data overfitting, we employed leave-one-out crossvalidation. The cross validation rate reflects a more accurate estimate of the true prediction accuracy of the biomarker. Except CA6, all comparisons have cross validation rates of #0.333, indicating that the validated biomarkers in general have high prediction accuracy. Despite our moderate sample size, we appear to have identified biomarkers that significantly correlate with the presence of breast cancer. Although the underlying relationships among systemic diseases and the saliva biomarkers are unclear, our recent study using mouse models has indicated that upon systemic disease development, cancer-specific changes occur in the salivary transcriptomic profiles. Stimulation of the salivary glands by mediators released from remote tumors plays an important role in regulating the salivary surrogate biomarker profiles. There may be extracellular communication between the ductal tissues of the breast and those of the salivary glands, since the histophysiology is very similar between these two distant tissues. Interestingly, all validated biomarkers were previously implicated in breast cancer or other cancers. Further investigation into the mechanism of salivary biomarkers for systemic cancers is warranted. In summary, our study has identified transcriptomic and proteomic biomarkers in saliva that have the potential to impact current diagnostic triage for breast cancer. The salivary biomarkers’ discriminatory power paves the way for a PRoBE-designed definitive validation study. The critical feature of PRoBE design involves prospective clinical sample collection, before outcome ascertainment, from a study cohort that is relevant to the clinical application. Any biomarker test intended for FDA approval and clinical use should incorporate the PRoBE principles as early as possible, as these principles eliminate potential biases commonly seen at the discovery stage. The study of real life networks, such as the world-wide web, internet, power-grids and math co-authorship, has put forth properties that distinguish them from classical Erdo¨s-Re´nyi random networks. The variety of degree distributions and other statistical measures that emerge has heightened the interest in complex networks.