Curative potential of multimodality therapy for locally recurrent rectal cancer

OBJECTIVE: To assess the results of multimodality therapy for patients with recurrent rectal cancer and to analyze factors predictive of curative resection and prognostic for overall survival. SUMMARY BACKGROUND DATA: Locally recurrent rectal cancer is a difficult clinical problem, and radical treatment options with curative intent are not generally accepted. METHODS: A total of 394 patients underwent surgical exploration for recurrent rectal cancer. Ninety were found to have unresectable local or extrapelvic disease and 304 underwent resection of the recurrence. The latter patients were prospectively followed to determine long-term survival and factors influencing survival. RESULTS: Overall 5-year survival was 25%. Curative, negative resection margins were obtained in 45% of patients; in these patients a 5-year survival of 37% was achieved, compared to 16% (P <.001) in patients with either microscopic or gross residual disease. In a logistic regression analysis, initial surgery with end-colostomy and symptomatic pain (both univariate) and increasing number of sites of the recurrent tumor fixation in the pelvis (multivariate) were associated with palliative surgery. Overall survival was significantly decreased for symptomatic pain (P <.001) and more than one fixation (P =.029). Survival following extended resection of adjacent organs was not different from limited resection (28% vs. 21%, P =.11). Patient demographics and factors related to the initial rectal cancer did not affect outcome. Perioperative mortality was only 0.3%, but significant morbidity occurred in 26% of patients, with pelvic abscess being the most common complication. CONCLUSIONS: This study demonstrates that many patients with locally recurrent rectal cancer can be resected with negative margins. Long-term survival can be achieved, especially for patients with no symptoms and minimal fixation of the recurrence in the pelvis, provided no gross residual disease remains.