PURPOSE: Although local excision can be curative in patients with early-sta
ge rectal cancer, approximately 20 per cent of patients will develop local
recurrence, many as a result of unrecognized and unresected regional lymph
node metastases. Our objective was to determine if standard pathologic fact
ors can predict lymph node metastases in small intramural rectal cancers an
d provide a basis for patient selection for nonradical surgery. METHODS: Be
tween June 1986 and September 1996, 318 patients with T1 or T2 rectal cance
rs underwent radical resection at our institution. Of these, 159 patients (
48 T1 and 111 T2) were potentially eligible for curative local excision (le
ss than or equal to 4 cm in size, less than or equal to 10 cm from the anal
verge, no synchronous metastases), and the prevalence of lymph node metast
ases based on T stage and other pathologic factors was analyzed in this gro
up. RESULTS: The overall frequency of lymph node metastasis was 15 percent
(24/159 patients). T stage (T1, 10 percent; T2, 17 percent), differentiatio
n (well-differentiated or moderately differentiated, 14 percent and poorly
differentiated, 30 percent), and lymphatic vessel invasion (lymphatic vesse
l invasion-negative, 14 percent and lymphatic vessel invasion-positive, 33
percent) influenced the risk of lymph node metastasis. However, only blood
vessel invasion (blood vessel invasion-negative, 13 percent and blood vesse
l invasion-positive, 33 percent) reached statistical significance as a sing
le predictive factor (P = 0.04). Tumors with no adverse pathologic features
(low-risk group) had a lower overall frequency of lymph node metastasis (1
1 percent) compared with the remaining tumors (high-risk group, 31 percent;
P = 0.008). However, even in the most favorable group (T1 cancers with no
adverse pathologic features) lymph node metastases were present in 7 percen
t of patients. CONCLUSION: In rectal cancer patients potentially eligible f
or local excision, the overall risk of undetected and untreated lymph node
metastases is considerable (15 percent). The use of pathologic factors alon
e after local excision does not reliably assure the absence of lymph node m
etastases.