Liver resection for metastatic colorectal cancer: assessing the risk of occult irresectable disease

Liver Resection for Metastatic Colorectal Cancer: Assessing the Risk of Occult Irresectable Disease William R Jarnagin, MD, Yuman Fong, MD, FACS, Alex Ky, MD, Lawrence H Schwartz, MD, Philip B Paty, MD, Alfred M Cohen, MD, FACS, and Leslie H Blumgart, MD, FRCS, FACS

disease. Resectability ranged from 95% in patients with a score of 0 (solitary, unilobar) to 62% in those with a score of 3 (multiple, bilobar; p 5 0.0001). The predictive value of this scoring system was then validated by applying it prospectively to an additional group of 118 patients taken to surgery for resection; the results were similar.

Background: Liver resection is standard therapy for selected patients with metastatic colorectal cancer. Extrahepatic metastases and inability to remove all hepatic disease usually preclude curative resection and are the most common contraindications. This study analyzes irresectability in patients considered to have resectable disease taken to operation for potentially curative hepatic resection. We describe preoperative factors associated with irresectability and propose a preoperative scoring system that identifies patients at particularly high risk for occult irresectable disease.

Conclusions: Standard preoperative investigations predicted resectability in 79% of patients with hepatic colorectal metastases. Unresectable disease limited to the liver and extrahepatic disease were seen with nearly equal frequency. The majority of patients with extrahepatic metastases had resectable hepatic disease (31 of 43, 72%). A preoperative scoring system is proposed that identifies patients at high risk for unrecognized irresectable disease and may help focus the use of additional diagnostic modalities such as laparoscopy and positron emission tomography (PET). (J Am Coll Surg 1999; 188:33–42. © 1999 by the American College of Surgeons)

Study Design: Patients considered to have resectable hepatic colorectal metastases were identified from a prospective database. Intraoperative findings that precluded liver resection were recorded. Demographic data, characteristics of the primary tumor, and characteristics of the hepatic metastases were recorded and analyzed. Results: From April 1992 through July 1997, 416 patients were explored with the intention of performing a potentially curative liver resection; 329 (79%) were resected. Eighty-seven patients (21%) had apparently resectable tumors on preoperative imaging but irresectable disease at laparotomy. Forty-four patients (51%) had irresectable disease limited to the liver; 32 had extensive bilobar disease not appreciated before surgery, and 12 were not resected for technical or other reasons unrelated to disease extent. Forty-three patients (49%) had extrahepatic disease, 31 of whom had resectable hepatic tumors. Of the several preoperative factors analyzed, only the estimated number of hepatic tumors was an independent predictor of irresectable findings at operation. This held true for patients with extrahepatic metastases and those with extensive hepatic disease. From these data, we devised a preoperative scoring system that estimates the probability of finding occult irresectable

Liver resection is potentially curative therapy for selected patients with metastatic colorectal cancer.1-3 Candidates for partial hepatectomy are generally those in whom liver metastases represent the only site of distant disease, although a select few with resectable extrahepatic recurrence may also benefit from resection.4-6 In the United States, approximately 7,000 to 9,000 patients fall into this category annually.7 Effort has been directed toward identifying patients who would most benefit from liver resection. Several variables have been identified that are associated with an increased risk of early recurrence.1,2,5 Most surgeons agree, however, that the only absolute contraindications to liver resection in otherwise fit patients are: the presence of unresectable extrahepatic disease or the inability to resect completely all hepatic disease.1,5 Preoperative imaging plays a critical role in patient selection. Improvements in radiologic techniques have enhanced our ability to identify patients

Received June 3, 1998; Revised September 18, 1998; Accepted October 5, 1998. From the Departments of Surgery (Jarnagin, Fong, Ky, Paty, Cohen, Blumgart) and Radiology (Schwartz), Memorial Sloan-Kettering Cancer Center, New York, NY. Correspondence address: William R Jarnagin, MD, Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021. © 1999 by the American College of Surgeons Published by Elsevier Science Inc.

33

ISSN 1072-7515/99/$19.00 PII S1072-7515(98)00272-5

34

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

with irresectable disease before surgery. Despite these improvements, unrecognized irresectable disease remains a persistent problem. Irresectable findings at laparotomy have been reported in 40% to 70% of patients.8-10 Extrahepatic disease, in particular, remains difficult to detect preoperatively.10,11 To avoid the morbidity, cost, and potential treatment delay associated with unnecessary exploration, several authors advocate routine use of additional studies, specifically laparoscopy with laparoscopic ultrasound12-15 or 18F-fluorodeoxyglucose positron emission tomography (18FDG PET).16-18 Several reports suggest that laparoscopy and 18FDG PET change the management in a subset of patients by identifying additional disease not seen on conventional imaging.12-15,17,18 A major unanswered question is whether either or both of these modalities should be applied routinely or reserved for patients at high risk for additional disease. The aim of this study is to identify patients at high risk for occult irresectable disease. We evaluated the incidence and pattern of intraoperative findings that precluded liver resection in patients considered to have resectable tumors. Comparison to a group of resected patients revealed several preoperative factors associated with occult irresectable disease at operation. From these results, we created a scoring system that stratifies patients from low to high risk, and validated its predictive value by applying it prospectively to a separate group of patients. It is hoped that such analysis will identify patients likely to benefit from more extensive preoperative investigation. METHODS Patients with metastatic colorectal cancer to the liver seen at Memorial Sloan-Kettering Cancer Center between April 1992 and July 1997 were identified from a prospective hepatobiliary database. Additional data were retrieved from medical records as necessary. Only those patients subjected to laparotomy for potentially curative liver resection were included. The objective of the procedure was determined from the surgeon’s preoperative note. Exclusion criteria were as follows: 1) exploration performed for any reason other than curative liver resection (ie, placement of hepatic artery infusion pump, palliation, etc.); 2) previous liver resection for any reason; 3) known extrahepatic metastases, even if concomitantly resected with liver disease. Patients with synchronous hepatic metastases, considered for simultaneous partial colectomy and hepatic resection, were included.

J Am Coll Surg

Curative resection was defined as any resection that the surgeon believed had removed all hepatic disease and might reasonably be expected to result in cure. Palliative resections and resections that clearly left tumor behind were excluded. The incidence of positive microscopic margins was less than 5%, and these were included. The majority of resections were performed by 2 surgeons (YF and LHB); however, all surgeons adhered to the same criteria of resectability. The absolute number of tumors was not used as a criterion of resectability. No patient was denied a resection based on the number of tumors found at operation, provided that all hepatic disease could be removed and there was no extrahepatic disease. The vast majority of patients were referred after having some or most of their workup completed elsewhere. The initial evaluation included a history, physical exam, and assessment of fitness for major hepatic resection. Outside radiographic studies were reviewed at a biweekly Hepatobiliary Disease Management conference attended by surgeons, radiologists, oncologists, and gastroenterologists. Studies were repeated or additional studies obtained as necessary, depending on the quality of the previous imaging; this judgment was made by the surgeon in consultation with the radiologist. The extent of hepatic disease was evaluated by CT, CT arterial portography (CTAP), MRI, or a combination of these. Duplex ultrasound (US) was frequently used to supplement these studies. CT scan of the chest and pelvis was performed in all patients. Colonoscopy was repeated in the absence of a normal examination within the previous year. At laparotomy, all patients underwent thorough exploration before hepatic resection. The pelvis, peritoneal cavity, retroperitoneum and portal, celiac, and retropancreatic lymph node basins were examined for evidence of extrahepatic disease. The extent of hepatic disease was assessed by bimanual palpation and intraoperative ultrasound. Findings suspicious for extrahepatic disease or unanticipated hepatic disease that would prevent resection were biopsied. Any palpable lymph nodes were also biopsied. Demographic information, date of initial surgery, stage of the primary tumor, and history of postoperative adjuvant therapy were determined from the database and review of the medical record. All tumors were staged according to the American Joint Committee on Cancer TNM classification.19 Disease-free interval, defined as the time from resection of the colorectal primary to the initial radiographic appearance of liver metastases, was deter-

Vol. 188, No. 1, January 1999

Jarnagin et al

mined before hepatic resection in all patients; this was 0 in patients who presented with synchronous hepatic metastases. It should be noted that diseasefree interval does not refer to recurrence after hepatic resection. The time from discovery of hepatic metastases to surgery and preoperative CEA levels, drawn within 4 weeks of exploration, was recorded. The extent of hepatic disease and segmental involvement was based on the Couinaud nomenclature.20 Patients were considered to have bilateral disease if any segment of both right and left hemilivers was involved with cancer. The number of hepatic metastases, number of segments involved, and size of the largest tumor were estimated from review of the radiographic reports and surgeon’s notes. Tumors were scored only if the radiographic interpretation was unequivocal and were not counted if too small to characterize or otherwise unclear. Findings suspicious for extrahepatic disease on the preoperative imaging were recorded for all patients. Operative findings, findings that precluded resection, and operation performed were determined from the surgeon’s operative note. Variables analyzed included age, gender, history of postcolectomy adjuvant therapy, time from discovery of hepatic metastases to surgery, preoperative CEA level, disease-free interval, stage and location of the primary tumor, number of hepatic metastases and hepatic segments involved, size of the largest tumor, and bilobar versus unilobar disease. Statistical analyses were performed using SPSS for Windows, version 7.0 (Statistical Package for Social Science, SPSS, Inc., Chicago, IL). Continuous variables were compared using Student’s t-test and categorical variables were compared using a chi-square test. Variables with p values # 0.1 on univariate analysis were further analyzed using a linear logistic regression model. Two-sided p values , 0.05 were considered statistically significant. Numeric data are expressed as the mean 6 standard deviation, unless otherwise indicated. RESULTS From April 1992 through July 1997, 416 patients with hepatic colorectal metastases were explored with curative intent. Of these, 329 (79%) were completely resected; 87 (21%) were found to have irresectable disease. Eighteen of the resected patients were excluded for insufficient data, leaving 311 for comparison.

Liver Resection in Metastatic Colorectal Cancer

35

Table 1. Intraoperative Findings that Precluded Resection in 87 Patients Finding Disease limited to the liver Extensive bilobar disease Vascular proximity/limited hepatic disease Abnormal liver Adhesions/vascular control impossible Extrahepatic disease Lymph node metastases Portal Retropancreatic Celiac Aorto-caval Carcinomatosis Pelvis Spleen Kidney

n 44 32 6 4 2 43 22 13 4 3 2 14 5 1 1

Patients unresectable because of vascular proximity had 1 or 2 tumors that involved either all 3 hepatic veins or all of the portal inflow.

Findings that precluded resection Forty-four patients had disease limited to the liver, 32 of whom had extensive bilobar liver metastases (Table 1). Six patients had limited hepatic disease that was irresectable because of unsuspected proximity to major vascular structures. These patients had 1 or 2 tumors that involved either all 3 hepatic veins or all of the portal inflow. Four patients had abnormal livers, either diffuse fatty infiltration or cirrhosis that was unsuspected before surgery. Resection was aborted because the risk of postoperative hepatic failure was considered excessive. Two patients were not resected because adhesions prevented extrahepatic access to the porta hepatis and hepatic veins. Forty-three patients were not resected because of extrahepatic disease. However, 31 of these (72%) had resectable hepatic tumors, based on the surgeon’s intraoperative assessment. Lymph node metastases (22 patients) and carcinomatosis (14 patients) were the most common findings. Five patients had recurrence in the pelvis and 1 each had metastatic disease involving the spleen and kidney. Factors associated with irresectability Table 2 summarizes the demographic data and characteristics of the primary tumor and liver metastases in the resected and irresectable groups. In the latter, only those patients not resected because of disease extent were included. The 12 patients not resected for reasons unrelated to extent of disease (vascular proximity, abnormal liver, adhesions) were excluded from this analysis. Irresectable patients were analyzed as a group and as separate ‘extrahepatic disease’

36

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

J Am Coll Surg

Table 2. Univariate Analysis of Preoperative Variables in Resected and Irresectable Patients Resectable (n 5 311)

Irresectable (all) (n 5 75)

Irresectable (HDZ) (n 5 32)

p value

p value

p value

Preoperative variable Demographics Demographics Age (y) 62 6 11 Age . 50 y 82% Male:Female 179:132 Discovery of liver disease to surgery (weeks) 16 6 28 Chemotherapy after colectomy 60% Characteristics of the primary tumor Synchronous hepatic metastases 38% No. of Positive nodes (colectomy) 2.1 6 2.7 Positive nodes 5 0 (colectomy) 38% Positive nodes . 3 (colectomy) 23% Disease-free interval (weeks) 58 6 80 Disease-free interval . 1 y 42% CEA (ng/dL) 176 6 1121 CEA . 20 (ng/dL) 41% Characteristics of the hepatic metastases No. of hepatic metastases 2.2 6 1.7 Bilobar metastases 52% No. of hepatic segments involved 3.1 6 1.4 Size of largest hepatic tumor 5.2 6 3.2

Irresectable (EHDZ) (n 5 43)

58 6 12 67% 49:38 18 6 24 56%

.01 .005 .8 .5 .6

58 6 12 67% 22:21 17 6 21 60%

.06 .03 .4 .7 .9

57 6 11 66% 20:12 19 6 27 47%

.05 .03 .6 .6 .2

55% 2.2 6 3.2 34% 26% 40 6 69 28% 167 6 534 57%

.01 .7 .7 .6 .08 .03 .9 .01

45% 2.6 6 3.4 26% 28% 51 6 77 33% 108 6 377 49%

.4 .3 .2 .5 .6 .3 .7 .4

68% 1.7 6 2.9 46% 23% 27 6 54 22% 245 6 690 64%

.001 .4 .3 .9 .03 .03 .8 .02

3.6 6 1.9 74% 3.7 6 1.5 4.9 6 2.9

.0001 .001 .001 .5

3.4 6 1.6 72% 3.5 6 1.6 4.6 6 2.5

.0001 .02 .2 .3

3.9 6 1.6 77% 4.1 6 1.3 5.3 6 3.4

.0001 .01 .0001 .8

Univariate p values are derived from statistical comparisons of irresectable patients to resectable patients. Irresectable patients were further subdivided into those with extrahepatic disease (EHDZ) and extensive hepatic disease (HDZ). These irresectable subgroups were again compared to the resected group. Twelve patients not resected for reasons unrelated to extent of disease (vascular proximity, abnormal liver, adhesions) were excluded from this analysis.

(EHDZ) and ‘extensive hepatic disease’ (HDZ) subgroups. Statistic comparisons were made to the resected patients. Demographics. Irresectable patients were significantly younger than those in the resected group (58 6 12 versus 62 6 11 years, p 5 0.01). Eightytwo percent of the resected patients were older than 50 years compared with 67% of the irresectable patients (p 5 0.005). These differences persisted when patients with extrahepatic disease and extensive hepatic disease were analyzed separately. There was a slight male predominance that was similar in all groups. Likewise, the interval from radiographic discovery of the liver metastases to surgery and the proportion of patients receiving adjuvant chemotherapy after resection of the primary tumor were similar. Characteristics of the primary tumor. The location of the colorectal primary was evenly distributed in the resected and irresectable groups (Figure 1A). The proportion of patients with primary rectal cancers was identical in each group. The disease stage at initial presentation (Stage), extent of bowel wall invasion (T Stage), and number of cancer-bearing lymph nodes in the resected colorectal specimen (N Stage) are illustrated in Figure 1B. Stage 4 disease at

initial presentation (synchronous hepatic metastases) was significantly more common in irresectable patients (55% versus 38%, p 5 0.01) (Table 2). The T stage and number of involved lymph nodes were similar (Figure 1B, Table 2). Moreover, the proportion of patients with uninvolved lymph nodes in the primary tumor was similar in the resected and irresectable groups. The disease-free interval was not significantly greater in resected patients compared with the irresectable group (58 6 80 versus 40 6 69 weeks, p 5 0.08). However, the percentage of resected patients with disease-free interval longer than 52 weeks was higher (42% versus 28%, p 5 0.03). Preoperative CEA levels were not significantly different. However, CEA levels . 20 ng/dL were noted in a significantly greater proportion of irresectable patients (57% versus 41%, p 5 0.01) (Table 2). None of the characteristics of the primary tumor was significantly different in patients with extrahepatic disease when compared with the resected group. The differences in primary tumor characteristics between resected and all irresectable patients were attributable to patients with extensive hepatic disease (Table 2). There was no correlation between

Vol. 188, No. 1, January 1999

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

37

Table 3A. Summary of Resectability Scoring System Parameter No. of hepatic tumors 1 2 .2 Extent of hepatic disease Unilobar Bilobar

Points 0 1 2 0 1

sive hepatic disease but was not significant in those with extrahepatic disease. Size of the largest hepatic tumor was similar (Table 2). Multivariate analysis. All variables with p # 0.1 on univariate analysis were included in a multivariate logistic regression analysis. Number of hepatic tumors was the only independent predictor of irresectability (p 5 0.0004, odds ratio 0.7, 95% confidence interval 0.6–0.9). Moreover, when the irresectable group was further analyzed, number of hepatic tumors remained an independent predictor of irresectability secondary to extrahepatic disease (p 5 0.01, odds ratio 0.8, 95% confidence interval 0.7–1.0) and extensive hepatic disease (p 5 0.02, odds ratio 0.8, 95% confidence interval 0.6–1.0).

Figure 1. (A) Distribution of primary tumor location in all patients. (B) Extent of disease at initial presentation for all patients. Stage refers to overall disease stage according to the American Joint Commission on Cancer. T stage refers to the extent of penetration through the bowel wall, and N stage refers extent of nodal involvement.

the T stage or N stage of the primary tumor and irresectability from carcinomatosis or lymph node metastases. Characteristics of the hepatic metastases. Resected patients had significantly fewer hepatic metastases than irresectable patients (2.2 6 1.7 versus 3.6 6 1.0, p 5 0.0001). Likewise, the percentage of patients with bilobar disease was lower in the resected group (52% versus 74%, p 5 0.001). These differences remained significant when the irresectable patients were analyzed separately (Table 2). The number of hepatic segments involved with tumor was also greater in irresectable patients (3.7 6 1.5 versus 3.1 6 1.4, p 5 0.001). This difference was more pronounced in patients with exten-

Resectability score Based on the above analysis, the number of hepatic metastases and extent of hepatic disease (bilobar versus unilobar) were used to calculate a resectability score (Table 3A). Points were assigned based on the number of hepatic tumors (1 tumor 5 0 points; 2 tumors 5 1 point; .2 tumors 5 2 points) and whether or not the disease was unilateral (0 points) or bilateral (1 point). The sum was the final resectability score. The 12 patients not resected for reasons unrelated to disease extent were excluded, as were 4 patients with incomplete data (1 resected, 3 irresectable). Predictive value (retrospective). Resectability scores were tabulated for all patients retrospectively. There was a progressive decline in resectability with increasing score, from 95% in patients with a score of 0% to 62% in those with a score of 3 (Table 3B). On logistic regression analysis, the likelihood of resection decreased significantly with increasing score (p 5 0.00001, Table 4). The percentage of resected patients with a score of 0 was significantly higher than those with scores of 2 (p 5 0.008) or 3 (p 5 0.0001). Likewise, the percentage of resected patients with scores of 1 or 2 were significantly greater than those with a score of 3

38

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

J Am Coll Surg

Table 3B. Results of Resectability Scoring System Applied Retrospectively to Original Cohort of Patients Resected†

Irresectable†

Score

n*

n

%

n

%

EHDZ/HDZ‡

0 1 2 3

113 65 90 114

107 57 75 71

95 88 83 62

6 8 15 43

5 12 17 38

6/0 4/4 8/7 24/19

*Total number of patients with each score. †Number of patients in each category with the indicated score. ‡Numer of irresectable patients with these findings at laparotomy. Patients not resected for technical reasons (12 in B and 1 in C) were excluded from this analysis. EHDZ, extrahepatic disease; HDZ, extensive hepatic disease.

(p 5 0.0001 and p 5 0.001, respectively). When compared to the resected group (1.4 6 1.2), mean resectability score was significantly higher in all irresectable patients (2.3 6 1, p 5 0.0001), patients with extrahepatic disease (2.2 6 1.1, p 5 0.0001) and patients with extensive hepatic disease (2.5 6 0.7, p 5 0.0001). Resectability score could not be used to determine the reason for irresectability (extrahepatic metastases versus extensive hepatic disease, p 5 0.2). Predictive value (prospective). To validate these results, the scoring system was applied prospectively to a separate group of patients taken to surgery for potentially curative liver resection. Scores were determined as described above, using the same eligibility criteria. From August 1997 through August 1998, 118 patients were explored for possible curative liver resection. Eighty-seven were resected (74%). Thirtyone were irresectable (26%), 17 because of extrahepatic disease, 13 because of extensive bilobar disease, and 1 for technical reasons (solitary tumor involving all hepatic veins). The breakdown of resected and irresectable patients by score was similar to that observed in the previous group (Table 3C). On logistic regression analysis, the likelihood of resection decreased significantly with increasing score (p 5 0.002, Table 4). The average score was significantly lower in the re-

sected group (1.4 6 1.1 versus 2.2 1.0, p 5 0.001). Likewise, resectability in patients with a score of 0 was significantly greater than resectability in patients with a score of 2 (93% versus 73%, p 5 0.05) or 3 (93% versus 57%, p 5 0.001). Preoperative imaging Figure 2A shows the breakdown of the preoperative imaging studies in irresectable patients; the results were similar for resected patients. This tabulation does not include studies obtained as part of routine followup after resection of the colorectal primary. More than half the patients underwent abdominal CT or duplex US (57% and 51%, respectively). CTAP was performed in 47% and MRI in 22%. A total of 152 preoperative studies was performed, nearly 2 per patient. One-third of patients underwent 1 preoperative imaging study, most commonly CTAP. Two or more studies were performed in 70% of patients (Figure 2B). It is not possible to analyze the predictive value of the various imaging studies, since many were done elsewhere and the techniques used were not uniform. One or more preoperative imaging study identified enlarged portal, celiac, retropancreatic, or aortocaval lymph nodes in 19 patients in the irresectable group. Of these, 11 (58%) were negative at exploration. Only 8 of 22 patients (36%) with lymph nodal

Table 3C. Results of Resectability Scoring System Applied Prospectively to an Additional Group of Patients Resected†

Irresectable†

Score

n*

n

%

n

%

EDHZ/HDZ‡

0 1 2 3

29 27 26 35

27 21 19 20

93 78 73 57

2 6 7 15

7 22 28 43

2/0 5/1 2/5 8/7

*Total number of patients with each score. †Number of patients in each category with the indicated score. ‡Numer of irresectable patients with these findings at laparotomy. Patients not resected for technical reasons (12 in B and 1 in C) were excluded from this analysis. EHDZ, extrahepatic disease; HDZ, extensive hepatic disease.

Vol. 188, No. 1, January 1999

Jarnagin et al

39

Liver Resection in Metastatic Colorectal Cancer

Table 4. Predictive Value of Resectability Score

Score

p Value

Coefficient (SE)

Resectability score (retrospective) Resectability score (prospective)

0.00001 0.002

0.14 0.21

Odds Ratio

95% Confidence Interval

0.45 0.51

0.35–0.6 0.34–0.78

Logistic regression analysis of resectability score was performed separately for the retrospective group (Table 3B) and the prospective group (Table 3C).

involvement had suspicious findings on any preoperative study. The sensitivity and specificity of any preoperative study correctly identifying extrahepatic lymph nodal metastases were 36% and 79%, respectively. Lymphadenopathy was reported in 23 of the resected patients, all of which were negative at oper-

ation. Carcinomatosis was not suspected in any of the 14 patients with this finding at exploration. Procedures and complications Table 5 summarizes the procedures performed in the irresectable patients. The majority had exploration and biopsy only. Twenty-seven patients, explored with curative intent, had placement of hepatic artery pump after irresectable findings were discovered. Four patients with synchronous disease, initially scheduled to undergo colectomy and hepatectomy, had colectomy only. There were 9 complications (10%) and 0 deaths. Mean length of stay was 7 6 3 days for all patients and 6 6 3 days for those undergoing exploration and biopsy only. DISCUSSION The curative potential of hepatic resection for colorectal metastases is well established.1-3 While subsets of patients are at greater risk for early recurrence, longterm survival has been documented despite the presence of poor prognostic factors.1 Extrahepatic metastases and inability to remove all hepatic disease Table 5. Summary of Procedures Performed, Complications and Hospital Stay in Irresectable Patients Procedures

Figure 2. Breakdown of preoperative imaging studies by type (A) and number (B) in irresectable patients.

Exploration/biopsy only Hepatic artery infusion pump/cholecystectomy Colectomy Colostomy reversal Cholecystectomy Ventral hernia repair Thoracotomy Complications Wound infection Pulmonary embolus Pneumonia Thrombosed hepatic artery Supraventricular tachycardia Otitis media Total Deaths Hospital stay Overall Exploration and biopsy only

n 51 27 4 2 1 1 1 4 1 1 1 1 1 9 (10%) 0 763d 663d

40

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

Table 6. Resectability Rates in Four Published Series, Including the Present Study Resected Author Fortner et al Steele et al10 Gibbs et al9 Present study 8

Year

n

n

%

1984 1991 1998 —

265 150 159 534

75 69 97 416

30 46 61 78

predict poor outcomes and remain the only absolute contraindications to resection in otherwise healthy patients. Given the lack of effective alternative therapy, we believe that resection should be considered in all patients who meet these criteria. The importance of adequate preoperative imaging is obvious. Identifying irresectable disease before laparotomy can be difficult, however. Previous reports suggest that unsuspected irresectable disease is found in 40% to 70% of patients subjected to laparotomy.8-10 The rationale for this study was to identify preoperatively patients at high risk for occult irresectable disease. Such patients would then be candidates for more intensive preoperative investigation that might demonstrate irresectability and spare them needless laparotomy. By contrast, patients at low risk might be spared the time and expense of additional studies which would have a low yield. In this study, 78% of patients (416 of 534) considered to have resectable liver metastases were resected. This rate of resection, among the highest reported in the literature (Table 6), is principally a reflection of improved imaging combined with an aggressive surgical approach. Preoperative evaluation was performed using standard imaging studies (CT, CTAP, US, MRI). Extrahepatic metastases and unanticipated hepatic disease accounted for the vast majority of irresectable cases and were seen in nearly equal proportions. Technical and other factors, unrelated to extent of disease, accounted for the remainder. Interestingly, nearly three-fourths of patients with extrahepatic metastases had resectable hepatic disease. Although our resectability rate was high, there is a clear need for better patient selection. First, while the operative morbidity in the irresectable patients was low, the average length of stay was 1 week, and alternative therapy (chemotherapy) was usually further delayed. Second, this study confirms the relative insensitivity of current imaging modalities in predicting extrahepatic nodal disease and carcinomatosis. Because patients denied exploration based on ra-

J Am Coll Surg

diographic findings are not included, the true predictive value of such findings cannot be estimated. Nevertheless, the data suggest that radiographic findings of enlarged perihepatic lymph nodes are frequently not an indicator of metastatic disease. In our analysis of preoperative factors, the estimated number of hepatic tumors was the only independent predictor of overall irresectability. Surprisingly, this variable was the only independent predictor of both extrahepatic metastases and occult disease within the liver. None of the characteristics of the primary tumor (number of positive nodes, stage, disease-free interval, etc.) was significant on multivariate analysis. To assess an individual patient’s risk of irresectability, we devised a scoring system based on the extent of disease within the liver: number of hepatic metastases and bilobar versus unilobar disease. These parameters emerged as the most potent predictors of resectability after analysis of all preoperative variables and were chosen for several reasons. First, as shown in the initial analysis, the number of hepatic metastases was the only independent predictor of irresectability. Second, except for patient age, these were the only factors that were significantly different in both subgroups of irresectable patients. Third, these variables provided the greatest contrast between the resected and irresectable groups. The predictive value of the scoring system was not improved by adding some or all of the variables that were significantly different on univariate analysis. When applied retrospectively to the original cohort of patients, this scoring system stratified the risk of irresectable disease. Resectability decreased progressively and significantly with increasing score, from 95% in those with a score of 0 (single, unilobar tumor) to 62% in those with a score of 3 (3 or more tumors, bilobar involvement). Importantly, the ability of this system to predict resectability was confirmed prospectively in an entirely different group of patients, with similar results. Clearly, this scoring system is dependent on the quality of preoperative imaging and therefore subject to some error. The number of hepatic tumors was probably underestimated in some patients in the resected and irresectable groups. However, this error was likely similar in both groups and probably not a major factor, a belief supported by the similar results obtained on prospective analysis. The aim of this study was to analyze the risk of finding additional metastatic disease that would ren-

Vol. 188, No. 1, January 1999

Jarnagin et al

der a patient irresectable. While surgeons may have different views as to what constitutes irresectable disease, all surgeons at this institution adhere to the same criteria: presence of extrahepatic disease or inability to resect all hepatic disease. Whether resection should be offered to some patients is the subject of some debate. Several previous studies have identified features of the primary tumor and other factors that portend poor outcomes after hepatic resection.1,2,5,21 However, we and others believe that the presence of poor prognostic factors should not necessarily preclude resection, because some of these patients will benefit.1,5 The scoring system we propose is meant to identify patients at high risk for irresectable disease at the time of laparotomy and is not intended to predict early recurrence or survival. It is apparent from our data that many of the variables associated with early recurrence are not critical for assessment of resectability, once the decision to resect has been made. This study shows that the risk of occult irresectable disease can be estimated preoperatively by assessing the extent of hepatic disease. Patients with solitary, unilobar tumors rarely have unrecognized irresectable disease and are nearly always resected. On the other hand, patients with multiple, bilobar tumors are at higher risk of occult hepatic and extrahepatic disease that is below the level of detection using current imaging techniques. These patients are therefore likely to benefit from more intensive preoperative investigation. The best method of further investigating high risk patients remains to be determined. The results of this study show that improvements in identifying extrahepatic metastases and occult hepatic disease are necessary. PET scan and laparoscopy with laparoscopic ultrasound have emerged as the 2 most likely candidates. PET scanning is a relatively new modality that may distinguish benign from malignant tissue based on differences in glucose metabolism.16 Previous reports have suggested greater sensitivity of PET over conventional imaging in detecting occult extrahepatic disease and disease within the liver.17,18 Although promising, PET remains investigational and is not routinely available. Moreover, its ability to identify small tumors (, 1 cm) is questionable18; its role in patients with metastatic colorectal cancer remains to be defined. Laparoscopy has had a major impact on the preoperative staging of patients with pancreatic, gastric and other abdominal malignancies.13,22,23 Laparoscopy with laparoscopic ultrasound allow thorough evaluation of the liver, peritoneal cavity, and perihe-

Liver Resection in Metastatic Colorectal Cancer

41

patic lymph nodal stations in most patients, even those previously explored.15 In patients with metastatic colorectal cancer, laparoscopy with laparoscopic ultrasound has identified irresectable findings in 31% to 48% of patients, significantly enhancing the resectability rate.12,14,15 In summary, using standard imaging techniques, we resected 78% of patients taken to surgery for curative resection. Analysis of several preoperative variables demonstrated that the number of hepatic tumors was the only independent predictor of irresectable findings at laparotomy. A scoring system based on the extent of hepatic disease stratifies the risk of irresectable findings at laparotomy and identifies patients at high risk. Patients with solitary, unilobar tumors are rarely unresectable and would probably not benefit from further investigations. Patients with 3 or more tumors and bilobar involvement have occult irresectable disease in approximately 40% of cases. These patients should routinely undergo laparoscopy with laparoscopic ultrasound to avoid unnecessary laparotomy. References 1. D’Angelica M, Brennan MF, Fortner J, et al. Ninety-six five-year survivors after liver resection for metastatic colorectal cancer. J Am Coll Surg 1997;185:554–559. 2. Kavolius J, Fong Y, Blumgart LH. Surgical resection of metastatic liver tumors. Surg Oncol Clin N Am 1996;5(2):337–352. 3. Scheele J, Stang R, Altendorf-Hofmann A, Paul M. Resection of colorectal liver metastases. World J Surg 1995;19:59–71. 4. Adam R, Bismuth H, Castaing D, et al. Repeat hepatectomy for colorectal liver metastases. Ann Surg 1997;225:51–62. 5. Bismuth H, Adam R, Navarro F, et al. Re-resection for colorectal liver metastases. Surg Oncol Clin N Am 1996;5:353–364. 6. Nakamura S, Sakaguchi S, Nishiyama R, et al. Aggressive repeat liver resection for hepatic metastases of colorectal carcinoma. Surg Today 1992;22(3):260–264. 7. Steele G, Ravikumar TS. Resection of hepatic metastases from colorectal cancer. Ann Surg 1989;210:127–138. 8. Fortner J, Silva JS, Golbey RB, et al. Multivariate analysis of a personal series of 247 consecutive patients with liver metastases from colorectal cancer: I. Treatment by hepatic resection. Ann Surg 1984;199:306–316. 9. Gibbs JF, Weber TK, Rodriguez-Bigas MA, et al. Intraoperative determinants of unresectability for patients with colorectal hepatic metastases. Cancer 1998;82:1244–1249. 10. Steele G, Bleday R, Mayer RJ. A prospective evaluation of hepatic resection for colorectal carcinoma metastases to the liver: Gastrointestinal tumor study group protocol 6584. J Clin Oncol 1991;9:1105–1112. 11. Sugarbaker PH. Surgical decision making for large bowel cancer metastatic to the liver. Radiology 1990;174:621–626. 12. Babineau T, Lewis D, Jenkins R, et al. Role of staging laparoscopy in the treatment of hepatic malignancy. Am J Surg 1994;167:151– 155. 13. Callery MP, Strasberg SM, Doherty GM, et al. Staging laparoscopy with laparoscopic ultrasonography: optimizing resectability in hepatobiliary and pancreatic malignancy. J Am Coll Surg 1997;185:33–39. 14. Field RI, Liu JB, Nazarian L, et al. Laparoscopic liver sonography: preliminary experience in liver metastases compared with CT portography. J Ultrasound Med 1996;15:289–295.

42

Jarnagin et al

Liver Resection in Metastatic Colorectal Cancer

15. John TG, Greig JD, Crosbie JL, et al. Superior staging of liver tumors with laparoscopy and laparoscopic ultrasound. Ann Surg 1994;220:711–719. 16. Delbeke D, Martin WH, Sandler MP, et al. Evaluation of benign versus malignant hepatic lesions with positron emission tomography. Arch Surg 1998;133:510–516. 17. Lai DTM, Fulham M, Stephen M, et al. The role of whole-body positron emission tomography with 18F-fluorodeoxyglucose in identifying operable colorectal cancer metastases to the liver. Arch Surg 1996;131:703–707. 18. Vitola JV, Delbeke D, Sandler MP, et al. Positron emission tomography to stage suspected metastatic colorectal carcinoma to the liver. Am J Surg 1996;171:21–26. 19. American Joint Committee on Cancer. Fleming ID, Cooper JS,

20. 21. 22. 23.

J Am Coll Surg

Henson DE, et al. AJCC Cancer Staging Manual. 5th ed. Philadelphia: Lippincott-Raven, 1997; 12, Colon and Rectum: 83–88. Couinaud C. Etude anatomiques et chirugicales. Paris: Masson, 1957 400–409. Hugh TJ, Kinsella AR, Poston G. Management strategies for colorectal liver metastases—Part II. Surg Oncol 1997;6(1):31–48. Burke EC, Karpeh MS, Conlon KC, Brennan MF. Laparoscopy in the management of gastric adenocarcinoma. Ann Surg 1997;225: 262–267. Conlon KC, Dougherty E, Klimstra DS, et al. The value of minimal access surgery in the staging of patients with potentially resectable peripancreatic malignancy. Ann Surg 1996;223: 134–140.