Do collision tumors of the gastroesophageal junction exist? A molecular analysis

Collision tumors are thought to arise from the accidental meeting of two independent tumors. Here we present five gastroesophageal junction tumors consisting of two collision tumors and three composite tumors (characterized by two divergent lineages originating from the same neoplastic clonal proliferation), as diagnosed on histology. In an attempt to prove this distinction at a genetic level, we performed TP53 sequence analysis and p53 immunohistochemistry. In addition, loss of heterozygosity (LOH) analysis using 10 microsatellite markers was carried out. An identical TP53 mutation and a similar pattern of retention and LOH were found in both neoplastic components of the presumed collision tumors, suggesting that both components are derived from a single precursor cell that undergoes divergent differentiation in the evolution of the tumor. In the composite group, 1 case had a genetic basis for the possible diagnosis of a collision tumor, with a TP53 mutation in the adenocarcinoma component only, and a different pattern of retention and loss of heterozygosity. These findings imply that it is not possible to recognize true collision tumors from immunohistologic appearance alone and suggest that the long-standing histologic criteria for the diagnosis of these neoplasms have no molecular basis