Colorectal cancer pelvic recurrences: Determinants of resectability

PURPOSE: This study was designed to identify preoperative and intraoperative features of locally recurrent colorectal cancer that predict RO resection in patients scheduled for attempted complete resection followed by intraoperative radiation therapy. METHODS: Review of a prospective data base identified 119 patients brought to the intraoperative radiation therapy suite for planned complete resection of locally recurrent rectal (n = 101) and colon (n = 18) cancer between January 1994 and November 2000. R0 resection was achieved in 61 patients. This group was compared with patients in which an R1 (n = 38), R2 (n = 7), or palliative procedure (n = 13) was performed. Variables evaluated included: tumor location, features of the primary tumor, and preoperative findings on computed tomography, magnetic resonance imaging, and history/physical. Tumor location was established by review of operative/pathologic reports and classified as axial (anastomotic/perineal), anterior (bladder/genitourinary organs), posterior (presacral), or lateral (pelvic sidewall). RESULTS: When recurrence was confined to the axial location only, or axial and anterior locations, RO resection was achieved significantly more often than when other locations were involved (P < 0.001, P = 0.003, respectively). When a lateral component was present, R0 resection was achieved significantly less often than when there was no lateral component (P = 0.002). For patients with available preoperative computed tomography and/or magnetic resonance imaging results (n = 70), the finding of lateral tumor involvement was associated with R0 resection significantly less often than when lateral disease was not identified (P = 0.004). CONCLUSIONS: Pelvic recurrences confined to the axial location, or axial and anterior locations, are more likely to be completely resectable (R0) than those involving the pelvic sidewall. Efforts to enhance preoperative identification and imaging of these patients are clearly justified.