PURPOSE: Nodal metastasis is the best predictor of survival for patients wi
th colon cancer. Statistical models based on random distribution of positiv
e lymph nodes suggest that to correctly classify nodal status with 95 perce
nt confidence, 20 nodes are needed for T1 lesions, 17 nodes for T2, and 15
nodes for T3. The mean number of nodes identified in American patients is 8
, suggesting that they might not be accurately staged. Patients in our tumo
r registry staged as "node-negative" had a short survival when they had les
s than or equal to 10 lymph nodes evaluated when compared with patients wit
h > 10 lymph nodes evaluated (P < 0.01). We hypothesized that the use of se
ntinel lymph node may assist in the staging of colon cancer. METHODS: Thirt
y-eight consecutive patients with colon lesions were prospectively enrolled
into this trial between February 1998 and November 1993. Thirty-one patien
ts met criteria for analysis. During surgery, Lymphazurin blue dye was inje
cted subserosally into the area around the tumor. Routine nodal evaluation,
with extra cuts of all sentinel nodes, was undertaken. RESULTS: At least o
ne sentinel lymph node was found in 18 of 31 patients (58 percent). Sensiti
vity of 67 percent, specificity and positive predictive value of 100 percen
t, and negative predictive value of 94 percent were found when sentinel lym
ph nodes were identified in 2 of these 18 patients, the sentinel lymph node
was the only positive lymph node found. CONCLUSIONS: Application of the se
ntinel lymph node technique to colon cancer may make it easier to identify
lymph nodes most likely to contain metastatic disease, potentially "down-st
aging" more patients. This may have implications in postoperative care.