Sentinel lymph node biopsy in rectal cancer - Not yet ready for routine clinical use

Background. The value of sentinel node biopsy in visceral cancers is uncertain. We evaluated the feasibility and utility of radiocolloid lymphatic mapping and selective lymph node sampling in patients with rectal cancer. Methods. Forty-eight patients with rectal cancer were investigated. Thirty-seven patients had already undergone preoperative radiochemotherapy for locally advanced tumors. Eleven patients underwent primary surgery. An endoscopic injection of 1 mL technetium 99m-sulfur-colloid into the peritumoral submucosa was performed 15 to 17 hours before surgery. Ex vivo identification of the nuclide-enriched "sentinel lymph nodes" (SLNs) was performed using a hand-held gamma-probe. The selected SLNs were then carefully and systematically examined using serial sections and immunohistochemistry. Results. One or more SLNs were found in 46 of the 48 patients. The SLN detection rate was 96%. Sixteen of the 48 patients had lymph node metastases (35%). In 7 of the 16 patients, the SLNs correctly represented the nodal status. In 9 of the 16 patients, the SLN was tumor-free whereas non-SLN harbored metastases. This result represents a sensitivity of only 44%, and a false-negative rate of 56%. Further analysis showed that the method correctly predicted the nodal status only in the small subgroup of 5 patients with early cancer without preoperative radiation. In 4 patients, juxtaregional lymph nodes were excised on the basis of intraoperative radiocolloid detection, leading to upward staging in 1 patient. Conclusions. Sentinel lymph node biopsy using the radiocolloid technique with ex vivo lymph node identification shows a relatively high detection rate; however, the sensitivity in patients with locally advanced/irradiated rectal cancer is low. Nevertheless, the detection of juxtaregional metastases can improve staging in some patients. Further studies should focus on patients with early rectal cancers where the data were more promising.