Between January 1, 1990 and December 31, 2007, a total of 811 patients diagnosed with stage III adenocarcinoma of the rectum within 12 cm from the anal verge after undergoing curative resection followed by selective adjuvant chemoradiotherapy were extracted from our prospectively collected database. Of the 811 patients, due to patient refusal, poor performance status, or other reasons, only 525 patients completed the scheduled adjuvant chemoradiotherapy. Among the 525 patients, 76 in whom CRM was not measured were also excluded. The remaining 449 patients were eligible for this study.

Patients were staged via either a combination of appropriate imaging studies such as chest radiography, transrectal ultrasonography, pelvic magnetic resonance imaging (MRI), abdominopelvic computed tomography (CT), or 18-FDG positron emission tomography (PET) before the operation. Patients received mechanical bowel preparation with polyethylene glycol solution the day before surgery. Prophylactic antibiotics were administered at the induction of anesthesia. The inferior mesenteric artery was doubly ligated from the aorta at its root. During surgery, care was taken to preserve the hypogastric nerve. The inferior mesenteric vein was ligated at the lower border of the pancreas. A sharp pelvic dissection was performed from the entrance to the pelvic cavity down to the pelvic floor, or the anal hiatus within the levator ani muscles along the visceral mesorectal fascia.15 Tumor-specific mesorectal excisions were performed according to the tumor distance from the anal verge. For cases of upper rectal cancer, the mesorectum was excised 4 cm from the lower edge of the tumor. Total mesorectal excision was performed for cases of mid and low rectal cancer at the level of the levator ani muscle.16 A potentially curative procedure was defined as one in which the surgeon believed that the entire tumor had been removed at the time of surgery. All of the patients registered in the prospectively collected colorectal database in our hospital made postoperative follow-up visits every 2-3 months for 3 years. After 3 years, follow-up visits were reduced to every 6 months until 5 years, and annually thereafter. A physical examination, serum carcinoembryonic antigen (CEA) level measurement, and chest radiography were performed at each follow-up visit. Abdominopelvic CT scans and bone scintigraphies were performed annually. Chest CT, pelvic MRI, or 18-FDG PET scan was performed as indicated according to the clinician's decision. Patient follow-up lasted until the cutoff date (October 31, 2010), or when the patient died. The median follow-up period for all the patients was 56.6 months (range, 4.9-230.9 months).

Lymph nodes were retrieved after gross examination and manual palpation. All dissected and retrieved lymph nodes were stained with hematoxylin and eosin (H&E), and examined by light microscopy for the presence of tumor metastasis. After confirming their proper orientation, surgical specimens were thoroughly inspected during gross examination. CRM was then identified as described by Quirke, et al.9 In some cases, the CRMs of the freshly received specimens were stained using the Davidson Marking System (Bradley Product, Inc., Bloomington, MN, USA) and subsequently fixed for 48 h. Dissection consisted of slicing 5 mm serial sections of the whole tumor and the surrounding mesorectum in the transverse plane. Using a pathology slide (H&E stain), the shortest distance was measured from the microscopically lateral portion of the tumor to the stained CRM.11 CRM involvement was defined as a distance from the circumference margin of ≤1 mm.9 Patients were divided into two groups: "CRM-positive group" (CRM ≤1 mm) and "CRM-negative group" (CRM >1 mm) according to the shortest distance between the CRM and the tumor (Fig. 1).

In this study, 5-FU (450 mg/m² for 5 days) and leucovorin (20 mg/m² for 5 days) was given intravenously each month with six cycles. A total dose of 5040 cGY of external beam radiation therapy was delivered after the second round of chemotherapy (as supplied by a 6 MV/10 MV dual photon linear accelerator, CLINAC 2100 C, Varian, Palo Alto, CA, USA) in 30 fractions of 180 cGY each, five times weekly to the pelvic cavity with individually shaped portals and using a four-field box technique.17

LR was defined as any recurrent tumor growth within the pelvic cavity or perineal area confirmed by clinical, radiological or pathological evidence. For histological confirmation, an imaging-guided biopsy was performed, when possible. Overall local recurrence rates were calculated as the sum of isolated LR and LR with concomitant systemic metastases.8 Systemic recurrence was defined as recurrence outside the pelvis.

Lymph node ratio (LNR) was defined as the ratio of metastatic lymph nodes to total retrieved lymph nodes. Patients were categorized into three groups based on their LNR: Group 1=0<LNR≤0.099 (LNRG1); Group 2=0.01≤LNR ≤0.249 (LNRG2); or Group 3=0.25≤LNR≤1.0 (LNRG3). This cut-off value was chosen based on the tertile number of LNR in this study group.

All statistical analyses were performed using SPSS software, version 18.0 (SPSS, Chicago, IL, USA). Categorical variables were analyzed using the chi-square test or Fisher's exact test, and continuous variables were analyzed using Student's t-test. Local recurrence-free survival was defined from the date of surgery to the date of detection of local recurrence. Systemic recurrence-free survival was defined as the date of surgery to the date of detection of systemic recurrence. Differences in survival and recurrences between the groups were compared using the Kaplan-Meier method and tested with the log-rank test. Factors associated with local-recurrence free survival and systemic-recurrence free survival were analyzed by a Cox proportional hazards model done by a backward stepwise selection of variables. p-values <0.05 were considered statistically significant.

Of the 449 patients, 79 (17.6%) were classified as CRM-positive group and 370 patients (82.4%) as CRM-negative group. The clinicopathological characteristics of the patients are summarized in Table 1. There were significant differences in CRM status according to the degree of lymphovascular invasion, tumor size, pathologic T stage, and pathologic N stage.

Overall, 203 patients (45.2%) exhibited recurrence during the study period. Of these patients, 42 patients (20.7%) were CRM positive and the remaining 161 patients (79.3%) were CRM negative. Isolated SR rate was higher in the CRM-positive group than that of the CRM-negative group, although the p-value thereof was only marginally significant (40.5% vs. 29.7%, p=0.062) (Table 2).

Among various patterns of SR, liver metastasis and carcinomatosis were significantly higher in the CRM-positive group than in the CRM-negative group (25.3% vs. 10.8%; p=0.001, and 6.3% vs. 1.9%; p=0.043, respectively) (Table 3).

The 5-year overall survival rate was 54.2% in the CRM-positive group and 66.1% in the CRM-negative group. On univariate analysis, LNR group and pathologic N stage were shown to be significant predictors of LR, while lymphovascular invasion, circumferential resection margin involvement, LNR group, pathologic T stage, and pathologic N stage were shown to be significant predictors of SR (Table 4). Cox proportional hazards analysis showed that LNR group was an independently predictor of LR. Additionally, preoperative CEA level, CRM involvement, pathologic N stage, and LNR group were shown to be independent predictors of SR on multivariate analysis (Table 5).

The cumulative incidences of LR and SR according to CRM status are shown in Fig. 2. The overall local recurrence rate (the sum of isolated LR and LR with concomitant distant metastases) was not different regardless of whether the patient was CRM positive or not (5-year LR rate in the CRM-positive group 13.0% vs. CRM-negative group 13.5%; p=0.677).