Cancer diagnostics: decision criteria for marker utilization in the clinic

A new diagnostic tool must pass three major tests before it is adopted for routine clinical use. First, the tool must be robust and reproducible; second, the clinical value of the tool must be proven, i.e. the tool should reliably trigger a clinical decision that results in patient benefit; and, third, the clinical community has to be convinced of the need for this tool and the benefits it affords. Another factor that can influence the adoption of new tools relates to the cost and the vagaries of insurance reimbursement. The Cancer Diagnosis Program (CDP) of the US National Cancer Institute (NCI) launched the Program for the Assessment of Clinical Cancer Tests (PACCT) in 2000 to develop a process for moving the results of new technologies and new understanding of cancer biology more efficiently and effectively into clinical practice. PACCT has developed an algorithm that incorporates the iterative nature of assay development into an evaluation process that includes developers and end users. The effective introduction of new tests into clinical practice has been hampered by a series of common problems that are best described using examples of successes and failures. The successful application of the PACCT algorithm is described in the discussion of the recent development of the OncotypeDX assay and plan for a prospective trial of this assay by the NCI-supported Clinical Trials Cooperative Groups. The assay uses reverse transcription (RT)-PCR evaluation of a set of 16 genes that were shown to strongly associate with the risk of recurrence of breast cancer in women who presented with early stage disease (hormone responsive, and no involvement of the auxiliary lymph nodes). The test is highly reproducible. It provides information to aid the physician and patient in making important clinical decisions, including the aggressiveness of the therapy that should be recommended. A trial is planned to test whether OncotypeDX can be used as a standalone trigger for specific treatment decisions. The problems that have been encountered and have delayed the development of other diagnostic tools are exemplified in the development of tests for human epidermal growth factor receptor (HER2) overexpression, for predictors of response to epidermal growth factor receptor inhibitors, and for the detection of residual disease following chemotherapy.