Liver Resection for Multiple Colorectal Metastases: Title and subTitle BreakInfluence of Parenchymal Involvement and Total Tumor Volume, vs Number or Location, on Long-term Survival FREE

CARCINOMA OF the colon is the third most common malignancy in Western countries, and its incidence is increasing.¹ One third of patients with colorectal cancer develop liver metastases during the course of the disease.² The natural course of unresected colorectal hepatic metastases carries a median survival time of less than 1 year, and there are almost no survivors 3 years after diagnosis.^{3 - 5} Liver resection is the only effective therapeutic approach that offers a chance of long-term survival of 25% up to 39% after 5 years.^{5 - 8}

Several prognostic factors have been analyzed and widely discussed in the scientific literature, but most of them failed to find universal acceptance. Furthermore, the role of liver resection remains controversial regarding the indication for multiple and/or bilateral hepatic metastases. Disappointing results have been reported in the treatment of multiple and bilateral nodules,^{8 - 13} and some authors did not recommend a surgical approach under these circumstances. Alternative treatments, such as systemic or locoregional intra-arterial chemotherapy and cryoablation, have been accepted as palliative measures in patients with unresectable disease^{8 ,13 - 15} ; however, poor results have been reported with the use of these procedures.^{8 ,14 - 15}

We reviewed our experience with 245 liver resections for colorectal metastases. Several prognostic factors that might affect long-term survival were analyzed, including total tumor volume (TTV). We then evaluated operative features, treatment-related complications, and long-term results in patients who underwent resection with multiple and/or bilobar metastases compared with patients with single metastases to define whether surgical resection should be advocated in the former cases.

From November 1, 1984, to January 31, 2000, 257 consecutive liver resections for colorectal hepatic metastases were performed at the Department of Surgery and Transplantation, University of Bologna, Bologna, Italy. Among them, 245 patients who underwent curative liver resections were enrolled in this analysis; the other 12, who underwent palliative procedures, were not. Selective diagnostic workup for patients with liver metastases included the evaluation of liver function, measurement of carcinoembryonic antigen level, ultrasonography, chest x-ray film, abdominal spiral computed tomographic scan, colonoscopy, and, in a few patients, liver biopsy to confirm the presence of metastatic disease. Follow-up ended April 30, 2000. There were 144 men (58.8%) and 101 women (41.2%). The mean ± SD age was 59.5 ± 9.9 years (range, 32-81 years). The mean ± SD number of lesions was 1.7 ± 1.1 (range, 1-10). The mean ± SD diameter of the largest metastasis was 4.9 ± 2.5 cm (range, 0.4-20 cm).

The type of procedure was defined according to the classification of liver segments by Coinaud.¹⁶ Resection of 3 or more segments was defined as a major hepatectomy. Liver resection was defined as "curative" when there was no extrahepatic disease and, macroscopically, all the hepatic tumor was completely removed with a sufficient surgical margin, which was confirmed as being free from cancer infiltration by microscopic examination. Surgical technique of liver resections and different methods of vascular control to reduce intraoperative bleeding have been described elsewhere.^{17 - 18}

During the study period, an adjuvant systemic postoperative chemotherapy based on fluorouracil and folinic acid was administered to improve long-term survival in 101 patients (41.2%) without any randomization. Among them, 32 (13.1%) patients also received regional chemotherapy by an infusion catheter placed in the hepatic artery through the gastroduodenal artery after the hepatectomy.

Several prognostic factors, which are detailed in Table 1, were analyzed to define which patients were expected to be long-term survivors. Among them, percentage of liver involvement was calculated on the basis of the preoperative radiologic findings, as reported by Gennari et al,¹⁹ and only more recently by computed tomographic volumetry.

To evaluate the influence on survival of the number and location of liver metastases, the 245 patients were divided into 3 groups: 146 patients (59.6%) with a single lesion (group M1), 60 (24.4%) with multiple unilobar lesions (group M2), and 39 (16.0%) with multiple bilobar lesions (group M3). The presence of at least 2 lesions, 1 on the right of the middle hepatic vein and 1 on its left, was defined as bilobar metastases.

Demographic features, tumor characteristics, operative variables, early postoperative outcome, and adjuvant chemotherapy were evaluated in the 3 groups (Table 2). Since most of the recent technical innovations were applied since 1992,¹⁸ we also evaluated survival rate in the 3 groups after 1992.

After liver resection, patients were followed up for recurrence every 3 months with carcinoembryonic antigen level and ultrasonography for the first year and then every 6 months, with chest x-ray film every 6 months, and with colonoscopy every year. A computed tomographic scan was performed whenever a local or distant recurrence was suspected from the other examinations.

Death occurring within 30 days after the surgical procedure was defined as operative mortality. Death occurring after surgery and before discharge was defined as hospital mortality. Survival was considered from the day of surgery to the day of death or the most recent follow-up visit.

The results were expressed as mean ± SD. The χ² test was used for categorical variables. The t test was used to compare continuous variables. The Spearman test was used to demonstrate a correlation between 2 different variables. Survival curves were estimated by means of the Kaplan-Meier method. Differences in survival curves between the groups were compared by the log-rank test. A multivariate analysis using Cox proportional hazards model was used between the prognostic factors analyzed at the univariate analysis. A P value of less than .05 was defined as significant. Statistical analysis was carried out with the SPSS test.²⁰

Among the 245 curative surgical procedures, 40 wedge resections (16.3%), 103 segmentectomies (42.1%), and 102 major hepatectomies (41.6%) were performed. Forty-six postoperative complications appeared in 37 patients (15.1%); the most frequent was pleural effusion in 13 cases (5.3%); in 13 patients (5.3%) a second laparotomy was required. Mean hospital stay was 12.9 ± 1.7 days (range, 6 to 21 days).

The overall operative mortality was 0.8% (2/245 patients); there was no additional in-hospital mortality.

The median length of follow-up was 30 months (range, 3 months to 14 years). The global actuarial survival rate at 3 and 5 years was 53% and 34%, respectively (Figure 1). On univariate analysis, only tumor diameter (P = .04), presence of 1 or 2 metastases vs 3 or more (P = .03), and percentage of liver involvement (P<.001) significantly affected long-term survival.

Multivariate regression analysis demonstrated that only the percentage of hepatic involvement by the tumor significantly affected long-term outcome (P = .001), and 25% was a useful cut-off in stratifying patients with different prognoses (Figure 2). No other feature contributed to survival; in particular, the system rejected tumor diameter and number of nodules as independent predictors of survival.

The rejection of the 2 already mentioned variables from the multivariate analysis led to the persistence of doubts about the effectiveness of our analysis. In particular, the acceptance of percentage of liver involvement by the tumor suggested that the total amount of tumor in the liver should play a fundamental role.

We therefore decided to calculate the TTV, assuming that all the metastases were sphere shaped. This could represent a kind of approximation, but, in consideration of the sample size, we assumed that this newly obtained variable would not represent a bias of the study.

The volume (V) of the nodule was calculated with the following expression: V = (4/3)πr³. We already had the diameter of all the lesions and their number, which were stored in our electronic database. In patients with multiple lesions, TTV was calculated as the sum of the volume of each metastasis.

Mean TTV was 128 ± 313 cm³ (range, 1.5-4199 cm³). Consequently, patients were arbitrarily divided into 3 groups: patients with TTV of less than 125 cm³ (T1), patients with TTV from 125 to 380 cm³ (T2), and patients with TTV of more than 380 cm³ (T3).

The 3- and 5-year survival in the 3 groups was 59.7% and 40.3%, 44.8% and 22.8%, and 23.2% and 15.4%, respectively (Figure 3). This difference was statistically significant (P<.01). Among the 181 patients in group T1, 68 (37.6%) had multiple and/or bilateral metastases, and a maximum of 10 nodules were resected per patient.

The multivariate analysis using the Cox model was performed again including the already mentioned prognostic factors analyzed at univariate analysis and TTV, thus excluding the percentage of liver involvement. As a result, the only variable that independently affected long-term outcome was TTV (P<.01). Analysis by Spearman rank test showed a significant correlation between the percentage of hepatic involvement by the tumor and the TTV (r = 0.41; P<.001).

An additional step was to divide patients with single lesions and patients with multiple and/or bilateral metastases among each of the 3 groups obtained according to TTV. These divisions are summarized in Table 3. As shown in Table 4, among each group of patients selected on the basis of TTV, there was no difference in survival between patients with single lesions and patients with multiple lesions. No difference was found between patients with multiple lesions in group T1 and patients with single lesions in group T2. However, patients with multiple metastases in group T1 had a significantly better prognosis than patients with single lesions in group T3 (P = .005).

The results of the statistical analysis are summarized in Table 2. The 3 groups were comparable regarding age, sex (except for a slight prevalence of male patients in group M2), diameter of the largest lesion, postoperative complications, postoperative stay, operative mortality, and adjuvant chemotherapy. In group M3 there was no operative mortality.

In group M2, the mean number of lesions was 2.6 ± 1.3 (range, 2-10) and 8 patients (13.3%) carried more than 3 lesions; in group M3, the mean number of lesions was 2.9 ± 1.3 (range, 2-6) and 11 patients (28.2%) had more than 3 lesions.

Fewer major hepatectomies (only 46 [31.5% of cases]) were performed in group M1, resulting in significantly less operative time compared with groups M2 and M3.

The 3- and 5-year survival rate was 57% and 41% in group M1, 49% and 17% in group M2, and 43% and 34% in group M3, respectively (Figure 4A). From 1992, the 3- and 5-year survival rate in the 3 groups was 63% and 45%, 54% and 24%, and 46% and 46%, respectively (Figure 4B). A tendency to a better prognosis was shown in group M1 compared with M2 (P = .05); the difference did not, however, reach statistical significance between groups M1 and M3, or between groups M2 and M3.

Long-term life expectancy was compared between patients with single metastases (group M1) and those with multiple metastases (groups M2 + M3). The actuarial survival rate at 5 years was 41% and 23%, respectively (not significant) (Figure 5A). Similar results were obtained when patients with unilobar metastases (groups M1 + M2) and patients with bilobar metastases (group M3) were compared; the 5-year survival rate was 35% and 32%, respectively (not significant) (Figure 5B).

Thirty-three patients survived for more than 5 years after resection. Seven of them (21.2%) had multiple metastases: multiple unilobar in 3 cases and bilobar in the other 4. In 28 cases (84.8%), the TTV was less than 125 cm³, while only 5 (15.1%) were in the T2 group. No patients survived for 5 years in the T3 group.

A review of 245 curative hepatic resections for colorectal metastases at a single tertiary center for liver surgery was carried out with the aim of defining which variables can predict long-term outcome, and whether surgical therapy should be advocated for patients with multiple and/or bilobar metastases. Among patients who underwent resection with 2 or more lesions, wherever located, a significant benefit in long-term survival was shown in the present series, with a 5-year survival of 23%. For this group of patients, alternative treatments such as cryoablation, laser hyperthermia, and intra-arterial or systemic chemotherapy have been applied with dismal results^{15 ,21 - 22} : about 20% survival after only 2 years.²² A recent review reported that 5-year survival for untreated but potentially resectable colorectal liver metastases was less than 2%.⁵

Because of the improvement in preoperative workup, anesthesiology management, and perioperative care, the mortality rate after hepatic resections for metastases has been reported as being less than 5% in recent series^{7 - 8 ,23} and even less than 1% in the present one.

Several prognostic factors have been evaluated to define which of them affect long-term survival. Minagawa and coworkers²⁴ stressed that most variables are considered significant in 8% to 86% of the published articles. Only the presence of lymph node metastases at the hepatic hilum has been consistently associated with a poor prognosis.²⁴ In the present study, which is one of the largest single-center series, the presence of 3 or more metastases, a tumor diameter greater than 6 cm, and a percentage of liver involvement higher than 50% were significantly related to a shorter life expectancy. However, on multivariate analysis, only the percentage of hepatic invasion by the tumor showed a significant influence on long-term outcome. This variable was taken from the original article by Gennari et al,¹⁹ which is still today one of the most widely accepted classifications of liver metastases. Nevertheless, this variable remains poorly described and suffers from a lack of objective measures, since the extent of the tumor depends on a subjective evaluation by the operator based on the diagnostic imaging. Because of this consideration and to stress the role of the total amount of the tumor in prognosis, we decided to calculate the TTV, which replaced the percentage of liver involvement in the analysis. It was remarkable that long-term survival was significantly related to the TTV (Figure 3). By substituting TTV for the percentage of liver involvement, on multivariate analysis the TTV was the only independent variable that affected survival. A statistical correlation using the Spearman test was found between the percentage of liver involvement by the tumor and the TTV, to confirm our assumption.

From our results, sex, age, preoperative carcinoembryonic antigen level, time of diagnosis, Dukes stage of primary tumor, type of resection, and adjuvant systemic and/or regional intra-arterial chemotherapy (even if this is not a controlled study) did not affect long-term survival.

To date, liver resection can achieve a 5-year survival rate of up to 40%, and it still represents the treatment of choice for colorectal liver metastases.^{7 - 8} Application of locoregional or medical therapies as the primary treatment for patients with single resectable metastases does not therefore seem justified.^{5 ,8} On the other hand, the role of liver resection remains unclear in patients with multiple and/or bilateral metastases.^{9 - 13} In the present series, there was a clear tendency to a better outcome in patients with small single lesions; however, no statistically significant difference was found between patients with single lesions and patients with multiple lesions, nor between patients with monolobar metastases and patients with bilobar metastases. Furthermore, even in patients with 3 or more metastases, a 5-year survival close to 20% was achieved.

Dividing the patients who underwent resection into 3 groups on the basis of the TTV, we were able to demonstrate that there was no difference in survival depending on the number of metastases. It was interesting that, in this group of patients with TTV less than 125 cm³ (group T1), the long-term outcome of 68 patients with multiple and/or bilateral metastases was significantly better than in patients with single metastases and TTV greater than 380 cm³ (group T3). Most important, no difference in survival was found between patients with multiple metastases in group T1 and patients with single metastases in group T2. To the best of our knowledge, this is the first report to advocate the TTV of colorectal liver metastases, and not number or location, as a reliable predicting factor of prognosis.

We have also reported that, among the 33 patients who survived more than 5 years, 7 (21.2%) had resection for multiple and/or bilateral metastases. This finding is remarkable compared with the dismal results in patients who did not undergo resection or who were treated by alternative therapies.⁵ Considering TTV, there were 28 and 5 long-term survivors in groups T1 and T2, respectively, while there were no 5-year survivors in group T3. The TTV, therefore, showed a stronger influence on survival than did number and location of metastases.

More than 75% of patients with advanced disease (multiple and/or bilobar metastases) underwent surgery after 1992. The improvement in surgical technique and vascular control of the hepatic pedicle^{17 - 18} were the key points that allowed us to extend the indication for liver metastases in recent years. Since 1992, thanks to some technical innovations like the routine use of anatomic segmentectomies, the restrictive use of blood transfusion, and strict respect of the surgical margin, we found an improvement in global survival and achieved a 5-year survival rate from 25% up to 40% even in patients with multiple bilateral metastases (Figure 4B).

The greater number of intraoperative blood transfusions and amount of operative time in groups M2 and M3 compared with the group M1 result from the complexity of the surgical technique when multiple lesions have to be resected. Significantly more major hepatectomies were required in groups M2 and M3. However, no differences in mortality, postoperative complications, or hospital stay were found between the 3 groups. We continue to stress that, in a tertiary center for liver surgery, the mortality rate is less than 1%, even for far advanced cases.²⁴

This mortality rate and the absence of other proven treatments support the concept of an aggressive surgical approach even for patients with multiple colorectal metastases. The other systemic or locoregional treatments represent palliative procedures, which may have a role as adjuvant treatments. Recently, encouraging results of aggressive surgical therapy have been reported even in cases of synchronous hepatic and pulmonary colorectal metastases.²⁵

In conclusion, although several prognostic factors have been reported to influence long-term survival and to predict poor outcome, we believe that patients with multiple colorectal liver metastases, irrespective of number and location, should always be considered by an experienced surgeon, and the only reasonable contraindications to surgery should be the presence of extrahepatic metastases, except for resectable lung metastases, and the impossibility of radically resecting the tumor while maintaining sufficient liver parenchyma. Further adjuvant therapies must be evaluated to prolong survival, especially in patients with a large total volume of metastases, which strongly affects prognosis and seems to be the only independent predictor of survival.

Corresponding author and reprints: Gian Luca Grazi, MD, Department of Surgery and Transplantation, S. Orsola-Malpighi Hospital, University of Bologna, Via Massarenti 9, 40138 Bologna, Italy (e-mail: glgrazi@unibo.it).