Vermillion is applying a multi-biomarker approach to identify associations between genes, proteins and diseases that translate into meaningful information about disease states. We plan to use this information to create and commercialize high-value diagnostic tests with sufficient sensitivity and specificity to aid physicians in predicting disease risk, selecting appropriate therapies, monitoring disease progression, and improving outcomes for patients.

Our strategy is to create a diagnostics paradigm that is based on risk stratification, multiple-marker testing and information integration. This strategy is based on the belief that any specific disease is heterogeneous; therefore, relying on a single disease marker to provide a simple “yes-no” answer is likely to fail.

Diseases, such as cancer and heart disease, can be traced to multiple causes and can develop within an individual at different rates and with varying symptoms. Additionally, individuals with the same disease may respond differently to treatments. Consequently, diagnosis, disease monitoring and treatment decisions can be challenging. This heterogeneity of disease and difference in human response to disease and/or treatment underlies the shortcomings of single biomarkers to predict and identify many diseases. A better understanding of heterogeneity of disease and human response is necessary for improved diagnosis and treatment of many diseases.

When designing clinical studies, Vermillion begins with the clinical question, since this drives the downstream clinical utility of the biomarkers. Recognizing that validation is the point at which most biomarkers fail, we have adopted a strategy focused on reducing the attrition rate between discovery and clinical implementation by building validation into the discovery process. With this as a starting point, we design our studies so that they include the appropriate cases and control groups. We place an emphasis on multi- institutional studies, inclusion of clinically relevant controls, and partitioning of training from validation data. For example, in the 2004 Cancer Research paper, which describes the first three markers in the ovarian cancer panel, more than 600 samples taken from five hospitals were analyzed. These samples were divided into individual training sets, followed by a first round of validation samples and then a second round of independent validation samples. To date, Vermillion has analyzed more than 2,500 samples from five additional medical centers.

An important characteristic of proteins is that their functional activity is often modulated by changes in their structure. Conventional approaches to assay proteins have variable ability to detect these changes, and may depend on the specificity of the antibody to the original or altered forms of the proteins. Additionally, a conventional assay may inadvertently measure only one form of a protein while many exist. Vermillion has developed programs for biomarkers in which mass spectrometry provides an advantage over traditional assays in characterizing and quantifying disease markers.