Prospective study of a 10-gene molecular assay to predict tissue of origin in patients with carcinoma of unknown primary (CUP)

Abstract

21096 Background: Carcinoma of unknown primary (CUP) where metastatic disease presents without an identifiable primary represents ∼ 3–5% of all cancers. Identifying the origin of the primary tumor in CUP pts can facilitate rational choice of therapeutic regimens. Veridex developed an optimized set of 10 gene markers for a quantitative reverse transcriptase polymerase chain reaction (qRTPCR) assay, and demonstrated high accuracy in predicting the tissue of origin with formalin-fixed, paraffin-embedded (FFPE) metastatic carcinoma samples (J Mol Diagn 2006, 8: 320–9). In this study, the 10-gene assay was prospectively evaluated in CUP pts. Methods: We collected FFPE biopsy tissue specimens from consenting CUP pts at MD Anderson. Eligibile pts met our definition for CUP with adenocarcinoma or poorly differentiated carcinoma. Samples were obtained prior to treatment. 51 pts have been enrolled so far and 11 were ineligible [insufficient samples].Of the 40 pts, qRTPCR assay has been performed on 33 pts. Data on 27/33 is available. A statistical model was used to determine the probability that the metastatic carcinoma tissue assayed originated from 1 of the following 7 categories: lung, pancreas, colon, breast, prostate, ovarian, and other. Subsequently, prediction of the primary by qRTPCR was independently compared with metastatic pattern spread, tumor pathological features, and results of clinical and pathology diagnostic workups. Results: Assay results on 27 prospectively collected CUP patient biopsy specimens are available. In total, CUP tissue of origin prediction by the assay correlated with clinical and pathological assessment in 21 out of 27 evaluated pts (78 %). The most common cancer type predicted by the assay was colon cancer, which correlated with predominantly intra-abdominal metastatic spread in this pt cohort. Conclusions: This prospective study demonstrated the feasibility of conducting gene analysis to predict metastatic carcinoma tissue of origin in FFPE tissue specimens derived from CUP patients. Overall distribution of various primary cancer types as predicted by the assay was consistent with the historical distribution reported for CUP. Assay prediction was concordant with clinical and pathological assessment in 78 % CUP pts. No significant financial relationships to disclose.