Efficacy of Pre-Treatment Biopsies in Predicting Final Histopathology of Endoscopically Resected Early Malignancies in the Upper Gastrointestinal Tract

Abstract

Background: Endoscopic resection (ER) is an efficient and valuable treatment option for early mucosal well differentiated neoplastic lesions in the upper gastrointestinal tract (Pech 2008). Treatment algorithms are very often based on pathological findings from endoscopically obtained biopsies, however the diagnostic reproducibility of these are not as good as for ER specimens (Mino-Kenudson 2007). The current study aimed to correlate prospectively pre-ER pathological findings to the final pathological diagnosis obtained from the ER specimen. Methods: ERs from the upper GI tract were prospectively registered in a database between 2006 and November 2008. ERs were performed using the cap technique or a multiband mucosectomy device in the stomach and the esophagus, or by the lift and snare technique in the duodenum. The worst pathology known from biopsies from neoplastic lesions before ER was compared to the histology of the ER specimen. All biopsies and ER specimens were reviewed by at least two pathologists with specialized expertise in gastrointestinal pathology. Results: We studied the histology of 100 consecutive ERs in the upper GI tract. ERs were performed in squamous esophagus (4), in Barrett's esophagus (63), in the cardia (6), in the stomach (19) and in the duodenum (8). The final diagnosis included: 4 squamous cancer of the esophagus, 1 columnar lined esophagus, 6 intestinal metaplasia (Barrett), 13 low grade and 13 high grade Barrett's dysplasia, 30 Barrett's adenocarcinomas, 4 adenocarcinomas in the cardia, 4 gastric adenocarcinomas, 12 gastric tubular adenomas with high(6) or low grade dysplasia(5) and 8 tubular adenomas in the duodenum (1 high grade). The overall accuracy of pre-ER biopsies in predicting final histology was 61%. 21% of the lesions were upgraded to a worse pathology. 16 of these 21 lesions were upgraded from low or high grade dysplasia to mucosal cancer or even submucosal cancer. The majority of these lesions (63%) were clearly visible as slightly elevated or depressed lesions: 7 type IIa, 1 type IIc, 2 type IIa-c and 1 type Is lesions. Remarkably, all pre-ER biopsies in Barrett's that were classified as high grade dysplasia with suspicion of carcinoma could be formally classified to either dysplasia (3) or carcinoma (2) after staging ER. Finally, 18% of the lesions were downstaged to a more benign final pathology. Conclusion: Endoscopically prelevated biopsies only moderately predict the final diagnosis after endoscopic resection. Reassuring histology in the presence of small visible early lesions should therefore be considered as an indication for staging ER to obtain a final histological diagnosis.