21109 Background: Standard treatment for most patients with CUP involves empiric chemotherapy. Since specific treatment now exists for most types of advanced carcinoma, precise identification of the primary site could lead to improved therapy. Veridex developed an optimized set of 10 gene markers, for a qRTPCR assay to identify tissue of origin of metastatic carcinoma in formalin-fixed, paraffin-embedded (FFPE) tissue samples (J Mol Diagn 8:320, 2006). The assay includes markers for 6 primary sites: lung, pancreas, colon, breast, ovary, and prostate. In this retrospective study, we evaluated the Veridex assay in patients with CUP. Methods: We obtained FFPE tissue from diagnostic biopsies on 69 CUP patients previously enrolled in empiric chemotherapy studies. The Veridex assay was performed as previously described. Assay results were correlated with clinical features, pathologic features, and response to treatment. Results: The Veridex assay yielded provisional diagnoses in 42 of 69 patients (61%): lung (15), pancreas (11), colon (12 ), ovary (4), breast (0), and prostate, (0 ). Most patients with diagnoses of lung and pancreas cancer had clinical and pathologic features compatible with these diagnoses; response rates to empiric chemotherapy (usually taxane/platinum-based) in patients with these diagnoses were 29% and 9%, respectively. The 12 patients with colon cancer diagnoses had predominantly intra-abdominal metastases (liver, peritoneum); response rate to therapy (usually taxane/platinum- based) was low (8%). The 4 patients with ovarian cancer had atypical clinical and pathologic features, and only 1 of 4 had PR to first-line taxane/platinum therapy. Conclusions: In this retrospective study, the Veridex 10-gene molecular assay was feasible and provided provisional diagnoses in a majority of patients with CUP. The diagnoses made using this assay (except ovarian cancer) were compatible with clinicopathologic features. The efficacy of cancer-specific treatment in patients diagnosed by this assay will be evaluated in prospective studies. No significant financial relationships to disclose.