Limitations of ampullectomy in the treatment of nonfamilial ampullary neoplasms

Background: Pancreaticoduodenectomy (PD) is the standard surgical management of invasive ampullary neoplasms. A rational plan to use ampullectomy (AMP) for lesions at this location requires careful analysis of preoperative clinical information (comorbidity, lesion size, and histopathology) and intraoperative data (frozen section pathology and clinical impression) to properly select patients for this treatment. Methods: We identified 140 consecutive cases of nonfamilial ampullary neoplasms from our prospective institutional database over a 7-year period (1996-2003). Preoperative and intraoperative factors were analyzed and related to outcomes. Results: AMP was planned for 37 patients with small lesions (median, 1.86 cm [range, 0-3 cm] vs. 2.6 cm [range, 0-8 cm] in PD). AMP was converted to PD because of the extent of disease in three and an intraoperative diagnosis of invasive cancer in five patients. Preoperative biopsy had a diagnostic accuracy of 79% (97 of 123) but missed 23 cancers. Intraoperative frozen section had a diagnostic accuracy of 84%; two cases of high-grade dysplasia and invasive cancer were missed. Patients with invasive cancer treated by AMP had a decreased recurrence-free and disease-specific survival compared with those treated by PD. Lymphatic spread of disease was associated with diminished long-term survival. Although both vascular invasion and tumor stage independently predicted lymphatic metastases, both were limited by their sensitivity. Conclusions: The reduced morbidity and mortality of AMP makes this the preferred treatment for benign lesions of the ampulla. Conversion to PD should be considered when intraoperative or final pathology identifies invasive adenocarcinoma. Refinement of clinicopathologic factors may reduce the occasional PD for benign disease and AMP for malignancy.