Schematic drawing of sampling sites for genetic examination. Approximately 4 × 3 × 3-mm cubic specimens were dissected from the tumor (T1 and T2), the surrounding liver parenchyma (N1 and higher), and Glisson capsule (G, dashed rectangle) when it appeared on the cut surface. T1 and T2 were randomly sampled from viable parts of the tumor. Best sampling sites for N1 and higher were selected in a systematic manner as shown in this figure.
Distribution of sampling sites in hepatic parenchyma for genetic examination. One hundred ninety-nine samples were harvested from documented noncancerous hepatic parenchyma surrounding liver tumors from 39 patients whose tumors had mutations in the K-ras or p53 genes or in both. Shaded bars indicate samples containing tumor cells with the same genetic alterations as were detected in the metastatic lesions in the liver.
Distribution of sampling sites in Glisson pedicle for genetic examination. Twenty-one samples were harvested from Glisson pedicle that appeared on the cut surface from 21 patients with genetic mutations in metastatic tumors. Shaded bars indicate samples containing tumor cells with the same genetic alterations as were detected in the metastatic lesions.
Histopathological findings of micrometastases. A tiny metastatic lesion was noted 3 mm from the main tumor.
Distribution of micrometastases detected by conventional histopathological examination. Micrometastases were found in 15 patients (24.2%). All micrometastases were found within 5 mm of the tumor border. Shaded bars indicate patients whose micrometastases were confirmed by genetic examination.
Kaplan-Meier estimates of overall survival after hepatic resection for metastatic colorectal cancer according to the surgical margin (SM). There was no significant difference in patient survival among groups 1 through 4.
Kaplan-Meier estimates of disease-free survival after hepatic resection according to the surgical margin (SM). There were marginally significant differences in disease-free survival (P = .054, group 4 vs groups 1-3).
Kokudo N, Miki Y, Sugai S, Yanagisawa A, Kato Y, Sakamoto Y, Yamamoto J, Yamaguchi T, Muto T, Makuuchi M. Genetic and Histological Assessment of Surgical Margins in Resected Liver Metastases From Colorectal CarcinomaMinimum Surgical Margins for Successful Resection. Arch Surg. 2002;137(7):833-840. doi:10.1001/archsurg.137.7.833
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
There have been few reports on the minimum surgical margins (SMs) required for successful liver resection in patients with colorectal metastases. This minimum requirement may be narrower than the previously recommended margin of 10 mm.
To identify the minimum margins by assessing the presence of micrometastases around the tumor using genetic and histological techniques, and to investigate whether SMs are associated with patterns of tumor recurrence or patient survival.
Prospective and retrospective studies.
Tertiary referral cancer center.
Patients and Methods
Fifty-eight patients who underwent 62 liver resections for hepatic metastasis from colorectal cancer between December 1, 1996, and November 30, 2000, were included in the study. Tissue samples taken from the tumor, surrounding liver parenchyma, and Glisson pedicle near the tumor were tested for K-ras and p53 mutations using the mutant allele–specific amplification method. For the retrospective study on patient outcomes, 194 patients who had undergone liver resections between 1980 and 2000 were analyzed according to their SMs.
Of the 62 sets of samples from liver metastases, 39 were positive for K-ras and p53 gene mutations or both. Micrometastases in the liver parenchyma surrounding colorectal metastases were present in 2.0% (4/199) of tested samples and were located within 4 mm of the tumor border. Micrometastases via Glisson pedicle were more common (14.3% [3/21]), but these were also confined to a short distance from the tumor edge (≤5 mm). Of the 5 micrometastases detected by genetic analysis, only 2 were confirmed by histopathological examination. The analysis of patient outcomes demonstrated that the incidence of cut-end recurrence (relapse in the bed of resection) decreased from 20.0% to a range of 5.6% to 7.5% when the SM is 2 mm or more. The incidence of definite cut-end recurrence in patients with SMs less than 2 m, 2 to 4 mm, and 5 mm or wider was 13.3% (6/45), 2.8% (1/36), and 0% (0/102), respectively. The SM was not a significant prognostic factor in patient survival.
Micrometastases around liver tumors are not common, and most are confined to the immediate vicinity of the tumor border. We propose an SM of 2 mm as a clinically acceptable minimum requirement, which carries approximately a 6% risk of margin-related recurrence. Because liver resection provides the only chance of cure, complete removal of the tumor with a minimum margin is justified when technically unavoidable because of the size, location, number of tumors, or successive resections.
ALTHOUGH SURGICAL margins (SMs) wider than 10 mm have been recommended for hepatic resection for colorectal metastases,1- 3 this is not always achievable when multiple lesions are present or when they are located deep inside the liver and close to major vessels. Furthermore, it is not uncommon for patients who undergo liver resection with minimum margins to survive without recurrence. There have been few reports4,5 on the minimum SMs required for successful liver resection. This may have limited the application of surgical techniques for colorectal liver metastases that cannot yet be cured by other treatment modalities.6
Using a sensitive method for detecting K-ras and p53 mutations in DNA samples, investigators have recently identified micrometastases in tissue samples that appeared to be tumor-free by routine histopathological examination.7,8 The mutant allele–specific amplification method is capable of detecting a single tumor cell containing genetic changes in a sample containing thousands of normal cells. We hypothesized that this powerful genetic analysis might detect tumor infiltration or micrometastases around liver metastases and provide useful information regarding minimum requirements for SMs.
We conducted a prospective study to assess the presence of micrometastases around the tumors using genetic and histological techniques. We also retrospectively investigated whether SMs are associated with patterns of tumor recurrence or patient survival in a single-institution series.
Between December 1, 1996, and November 30, 2000, 63 patients underwent 72 liver resections for colorectal metastases with curative intent at Cancer Institute Hospital, Tokyo, Japan. Of these, 58 patients (62 hepatectomies) were prospectively assigned to the present study after informed consent had been obtained. Ten resections were excluded because of positive SMs (9 patients) or multiple tumors (1 patient, >10 tumors). There were 36 men and 22 women, aged 35 to 78 years (mean, 60.1 years). The primary site was the colon in 42 patients and the rectum in 16. The mean ± SEM number of tumors, maximum diameter of tumors, and macroscopic SMs were 2.2 ± 0.2 (range, 1-9), 4.3 ± 0.4 cm (range, 0.7-18.0 cm), and 5.8 ± 0.6 mm (range, 0-25 mm), respectively. Written informed consent was obtained from all patients for genetic analysis of resected specimens.
For the retrospective study on patient outcomes, 215 patients who had undergone liver resection for colorectal metastases between January 1, 1980, and December 31, 2000, were selected from our database. Of these, 194 patients with operative extrahepatic metastases had undergone surgery with curative intent. The remaining patients were excluded because of residual tumor of the liver in 6, concomitant lung metastases in 7, para-aortic lymph node metastases in 4, peritoneal dissemination in 3, and inguinal lymph node metastases in 1. There were 108 men and 86 women, aged 35 to 82 years (mean, 59.0 years). The primary site was the colon in 137 patients and the rectum in 57. The SM was not clearly documented in 11 patients, who were excluded from the study. The remaining patients were divided into 4 groups according to the macroscopic width of the SM: group 1, less than 2 mm (45 patients); group 2, 2 to 4 mm (36 patients); group 3, 5 to 9 mm (53 patients); and group 4, 10 mm or more (49 patients). Of these 183 patients, 3 (1.6%) had macroscopic tumor thrombus in the portal branch, 5 (2.7%) had metastases in the intrahepatic bile duct, and 1 patient (0.5%) had disease in both locations. In such cases, the SM was measured between the proximal edge of the thrombus and the bed of resection. Patient survival, disease-free survival, and site of recurrence after liver resection were evaluated.
Resected specimens were cut in the plane that included the maximum cross section of the tumor. The specimens were divided into 2 parts, one of which was fixed in 17% formaldehyde and submitted for histopathological examination. From the remaining halves, approximately 4 × 3 × 3-mm cubic specimens were dissected from the tumor (T1 and T2), surrounding liver parenchyma (N1 and higher), and Glisson pedicle (G) when it appeared on the cut surface. T1 and T2 were randomly sampled from viable parts of the tumor. Best sampling sites for N1 and higher were selected in a systematic manner as shown in Figure 1. All of the dissected specimens were immediately frozen in liquid nitrogen and stored at −80°C. DNA was extracted from the frozen tissue as described elsewhere.9 In patients with multiple metastatic lesions, the evaluation was performed in the cross section of the largest tumor.
Codons 12 and 13 of the K-ras gene were examined by the mutant allele–specific amplification method as described previously.10 The 3′-ends of 20–base pair oligonucleotides used as polymerase chain reaction primers corresponded to variants of the first or second nucleotides of K-ras codons 12 and 13. DNA sequences corresponding to exons 5, 6, 7, and 8 of p53 were amplified by polymerase chain reaction, and the products were sequenced as described by Miyoshi et al.11
To find corresponding genetic alterations in the liver parenchyma or Glisson pedicle where mutations had been identified in metastatic tumors, we synthesized mutant allele–specific amplification primers with 3′-ends corresponding to each variant and amplified the respective sample DNA (N1 and higher and G) by polymerase chain reaction using these primers.7
Histological examination was done in the same cross section as was genetic analysis, using the other half of the divided specimens. Tumors and surrounding liver parenchyma were examined by conventional histopathological examination after hematoxylin-eosin staining. Micrometastasis was histologically defined as microscopically detectable tumor nests without direct connection to a tumor edge of the intrahepatic metastases.
All values represent the mean±SEM. The Mann-Whitney test or Kruskal-Wallis test was used to compare group data. The χ2 test was used for categorical data. The level of significance was established at P<.05. Survival time was calculated from the date of hepatic resection until death. Survival curves were generated using the Kaplan-Meier method and were then compared using the Breslow-Gehan-Wilcoxon test. A multivariate stepwise Cox proportional hazards regression analysis (backward elimination method)12 was performed to identify factors that were independently associated with survival and disease-free survival after liver resection. Commercially available software (SPSS, version 6.0; SPSS Inc, Chicago, Ill) was used for statistical analysis.
Of the 62 sets of samples from liver metastases, 15 (24.2%) were positive for K-ras mutation (codon 12, 14 cases and codon 13, 1 case) and 31 (50.0%) were positive for p53 mutation (exon 5, 8 cases; exon 6, 7 cases; exon 7, 9 cases; and exon 8, 7 cases). Overall, 39 sets of samples (62.9%) were positive for K-ras or p53 mutation or both, and these were analyzed for the remainder of the study.
One hundred ninety-nine samples were harvested from documented noncancerous hepatic parenchyma surrounding liver tumors from 39 patients whose tumors had mutations in the K-ras or p53 genes or in both (N1 and higher, Figure 1). Sampling sites (distance from the edge of the tumor) were distributed as shown in Figure 2. Four samples (2.0%) from 2 patients contained tumor cells that had the same genetic alterations as were detected in the metastatic lesions in the liver. The positive samples were harvested within 4 mm of the tumor border. One of these was confirmed by conventional histological examination.
Twenty-one samples were harvested from Glisson pedicle, which appeared on the cut surface (G, Figure 1) in 21 patients with positive genetic mutations in metastatic tumors. There was no macroscopically identifiable Glisson pedicle on the cut surface in the remaining 18 patients. Sampling sites were distributed between 2 and 25 mm from the tumor border (Figure 3). Three samples (14.3%) were genetically positive for cancer and were harvested within 5 mm of the edge of the tumor. One of these was confirmed by histological examination.
Histological micrometastases were found in 15 patients (24.2%) (Figure 4). All of the micrometastases were found within 5 mm of the tumor border. Of these, 2 were confirmed by genetic analysis (Figure 5). In the remaining patients, samples for genetic analysis were not taken from the corresponding area, primarily because of the close proximity to the edge of the tumor.
Patient characteristics in groups 1 through 4 are summarized in Table 1. There was no significant difference among the 4 groups in patient age, sex, primary site (colon vs rectum), temporal relationship (synchronous vs metachronous), or surgical procedure (limited resection, single segmentectomy of Healey and Schroy,13 or bisegmentectomy or more). There were significant differences among the 4 groups in lobar involvement (unilobar vs bilobar) and number of tumors. Bilobar distribution and a greater number of tumors were associated with significantly narrower SMs. Differences in tumor size and presence of extrahepatic disease were marginally not significant among groups 1 through 4.
After a median follow-up of 29.1 months, 107 patients (58.5%) developed recurrence inside or outside the liver. The overall recurrence rate and recurrence rate in the liver were not statistically different among the 4 groups, although group 4 (margins ≥10 mm) tended to have better results.
Cut-end recurrence was defined as a relapse in the bed of resection. The location of relapsing tumors and the bed of previous resection were identified by computed tomography or intraoperative inspection. Cut-end recurrence was considered definite when the 2 locations matched. Cut-end recurrence was suspected when the 2 locations were within the same Couinaud segment14 but not in the same location.
Cut-end recurrence, including suspected cases, developed in 20.0% of patients with SMs narrower than 2 mm, whereas the incidence of cut-end recurrence was 5.6% to 7.5% in patients with margins from 2 to 9 mm. There was no cut-end recurrence in patients with SMs 10 mm or wider. The incidence of definite cut-end recurrence in patients with SMs of less than 2 mm, 2 to 4 mm, and 5 mm or wider was 13.3% (6/45), 2.8% (1/36), and 0% (0/102), respectively. Overall, there were 15 patients (8.2%) in whom cut-end recurrence was the only treatment failure after liver resection (Table 2). Of these, 9 patients underwent a second liver resection, with a 3-year survival of 57.1%. The 5-year survival after the first hepatic resection was 53.6%.
The overall 5-year and disease-free survival rates were 41.9% and 29.7%, respectively. There were no significant differences in patient survival among groups 1 through 4 (P = .90) (Figure 6). However, there were marginally significant differences in disease-free survival (P = .054, group 4 vs groups 1 through 3) (Figure 7).
Among the 8 possible prognostic factors studied, univariate analysis showed a significant difference in patient survival based on the primary site (colon vs rectum), tumor size, and presence of extrahepatic disease (Table 3). Multivariate analysis identified 6 independently significant variables: tumor size, primary site, temporal relationship (synchronous vs metachronous), number of tumors, presence of extrahepatic disease, and tumor distribution (unilobar vs bilobar) (Table 4). The SM was not a significant prognostic factor in patient survival.
Univariate analysis of disease-free survival showed a significant difference in primary site, temporal relationship, tumor size, and presence of extrahepatic disease. The SM was marginally not significant (P = .10) (Table 3). Multivariate analysis identified 4 independently significant variables: presence of extrahepatic disease, temporal relationship, primary site, and tumor size (Table 5).
In the prospective study, the survival and disease-free survival rates in 15 patients with histological micrometastases were comparable or slightly better than those in 43 patients without micrometastases: 3-year survival, 100.0% vs 70.1%; and 2-year disease-free survival, 53.3% vs 32.3%, respectively.
Using a sensitive genetic analysis for detecting K-ras and p53 mutations, we found micrometastases in the liver parenchyma surrounding colorectal metastases. However, the incidence was as low as 2.0%, and they were located only in the immediate vicinity of the tumor (within 4 mm of the tumor border). Micrometastases via Glisson pedicle were more common (14.3%), but they were also confined to a short distance from the tumor edge (≤5 mm). Of the 5 micrometastases detected by genetic diagnosis, only 2 were confirmed by conventional histopathological examination.
These results are consistent with the relationship between SMs and patterns of recurrence. An analysis of patient outcomes demonstrated that the incidence of cut-end recurrence decreased from 20.0% to the range of 5.6% to 7.5% when the SM is 2 mm or more. The incidence of definite cut-end recurrence in patients with SMs of 2 to 4 mm and 5 mm or wider was 2.8% and 0%, respectively.
Because micrometastases detected in the present study were confined to the immediate vicinity of the main tumors, they may have originated from the closest main tumor, rather than from distant metastatic lesions or primary sites. Intrahepatic distant metastases are considered to be rare in colorectal metastases.15,16
Although many surgeons may try to achieve SMs as wide as possible during hepatic resection for colorectal metastases, there has been no concrete evidence regarding how much margin is necessary for successful hepatectomy. Since the report of Hughes and coworkers,17 other authors1- 5,17- 24 have noted the prognostic significance of the SM after liver resection for metastatic colorectal carcinoma (Table 6). Among these 14 reports, including the present study, only 4 stated that an SM 10 mm or wider was a favorable prognostic factor.1- 3,23 Furthermore, SMs 10 mm or wider were achieved in a mean percentage of only 34.4% (range, 15.9%-54.9%) of patients.
For positive margins (SM, 0 mm), 7 of the 14 reports indicated that a positive margin was associated with a significant or marginally significant unfavorable prognosis. Although patients with a positive margin may tend to have a recurrence at the cut end or may have other undesirable prognostic factors,23 the treatment failure that actually leads to patient death is not clear. Patients with a positive margin should not be considered inoperable, as the group 1 patients in our study had a chance of cure, with a 5-year survival of 34.7% (Figure 6).
We hypothesized that the minimum SM for successful liver resection without cut-end recurrence may lie somewhere between 0 mm and 10 mm. Based on the results summarized in Table 7, an SM of 2 mm appears to be a clinically acceptable minimum requirement, carrying an approximately 6% risk of margin-related recurrence. Gayowski4 and Ohlsson5 and their colleagues reported that patients with SMs 1 mm or wider, but less than 10 mm, had outcomes comparable to those of patients with SMs wider than 10 mm. However, considering the measurement error associated with a macroscopic SM, a margin of 1 mm may be too narrow for clinical use.
If the SM is an independent significant prognostic factor, cut-end recurrence should be common in patients with a narrow margin, and this type of recurrence should lead to a poor outcome. Hughes et al18 reported that recurrence in the remnant liver was more common in patients with a positive margin (68% [25/37]) than in patients with a negative margin (38% [217/570]). Cady et al3 reported similar findings. However, these reports did not distinguish cut-end recurrences from other types of recurrence in the remnant liver. Bozetti et al25 reported that cut-end recurrence accounted for only 25% (4/16) of all recurrence in the remnant liver. Harned et al26 also reported a low incidence of cut-end recurrence, accounting for only 8% of all recurrences in the remnant liver. This rate was 21.7% (15/69) in the present study.
Consistent with previous reports,1,25,26 the SM did not affect the overall recurrence rate or the recurrence rate in the liver (Table 3). However, cut-end recurrence developed in 7.5% to 20.0% of patients with SMs narrower than 10 mm, whereas no recurrence was observed when SMs were wider than 10 mm. Cut-end recurrence was definite in 7 patients, and 5 of these had positive margins. Margins narrower than 10 mm may lead to cut-end recurrence, but the incidence in this study was only 8.2% (15 patients). Furthermore, 9 of these underwent a second hepatectomy, with a 3-year survival of 57.1%. Although cut-end recurrence is an undesirable treatment failure, its effect on patient survival is small. There have been no data on whether cut-end recurrence causes other types of remnant liver recurrence.
The higher prevalence of micrometastases via The Glisson pedicle may require further evaluation. Yamamoto et al15 reported 9 cases of colorectal metastases that had gross extension into The Glisson pedicle. Macroscopic invasion extended up to 23 mm from the tumor edge. Histopathological examination of multiple slices may be needed to evaluate microscopic tumor extension into The Glisson pedicle.
In conclusion, micrometastases around liver metastases from colorectal carcinoma are not common, and most are confined to the immediate vicinity of the tumor border. Cut-end recurrence after liver resection occurs in 5.6% to 7.5% of patients whose SM is 2 mm or more, but the incidence of definite cut-end recurrence was less than 3% with a margin of this width. We propose an SM of 2 mm as a clinically acceptable minimum requirement, carrying approximately a 6% risk of margin-related recurrence. However, the SM itself is not an independent significant prognostic factor after liver resection, but rather is probably in concert with other factors, including the number of tumors, tumor distribution, and extrahepatic disease. Because liver resection provides the only chance of cure, complete removal of the tumor with a minimum margin is justified when technically unavoidable based on the size, location, or number of tumors. Hepatic resections with SMs narrower than 2 mm are acceptable, especially for patients with multiple liver metastases.
Corresponding author and reprints: Norihiro Kokudo, MD, Hepato-Biliary-Pancreatic Surgery Division, Department of Surgery, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, Japan 113-8655 (e-mail: email@example.com).