45 Background: New animal models of esophageal cancer are required to accelerate research into novel treatment strategies. Mouse xenografts used in pre-clinical drug testing are typically derived from cell lines, but several esophageal cancer cell lines have been contaminated (PMID 20075370). Xenografts implanted from surgically resected human tumors are a potentially viable model, and may better recapitulate the characteristics of the original tumor. We describe the feasibility of such an approach. Methods: 2-10 mm esophageal carcinoma fragments obtained from resected patient tumors were implanted directly into NOD/SCID mice. Tumors that grew beyond 1.5 cm in diameter were passaged serially. Pathologic properties were assessed from passage to passage. Potential clinicopathological predictors of engraftment were evaluated. Results: Of the 63 patients with tumors used for implantation, 67% were male; median age 66 (range 31-89); 31% node negative disease; 69% node positive disease; 12% metastatic; 55% received neoadjuvant therapy; 12% had prior Barrett's esophagus; 46% had pre-existing heartburn; 88% were distal esophageal/GE junction cancers. Engraftment occurred in 14/43 (33%) adeno-, 1/14 (7%) squamous, and 1/1 adenosquamous carcinomas (p=0.05; 5 others were rare tumors e.g. granular or small cell). Engraftment occurred in 6/31 well/moderately vs 7/18 poorly differentiated tumors (p=0.08). Because we amended our implantation procedures, we compared engraftment rates pre- (7/39; 18%) and post-amendment (8/23; 35%; p=0.13). The primary architecture of the xenografted tissue remained similar pathologically for up to 5+ passages. None of the following variables predicted for engraftment (p>0.20 each comparison): cancer location, age, gender, history of heartburn, pre-operative therapy, presence of Barrett's esophagus, disease stage, or recurrence of primary cancer. Conclusions: Creation of esophageal cancer xenografts from primary human resected specimens is feasible. Adenocarcinomas, higher tumor grade, and newer implantation strategies were associated with improved engraftment. We are treating these models with chemotherapy to discover new pharmacogenomic predictive biomarkers. No significant financial relationships to disclose.
Predictive biomarkers for Barrett's esophagus: so near and yet so far