scout
|Articles|September 1, 2006

Oncology

  • ONCOLOGY Vol 20 No 10
  • Volume 20
  • Issue 10

Management of Liver Metastases From Colorectal Cancer

The liver is a frequent site of metastatic colorectal disease. Over the past 20 years, improvements in systemic chemotherapy and surgical techniques have improved the survival of patients with hepatic metastases. For 4 decades, fluorouracil and leucovorin were the only drugs available to treat metastatic colorectal cancer, but several new drugs and a variety of novel regimens are now available. Further improvements in results have been seen with the delivery of chemotherapy via the hepatic artery. Surgical resection of liver metastases has been encouraged when possible, and recent advances in surgery such as portal vein embolization, have made liver resection a possibility for more patients. This review considers the timing and sequence of chemotherapy and surgery in this setting, as well as the roles of cryoablation, radiofrequency ablation, and radiation therapy.

The liver is a frequent site of metastatic colorectal disease. Over the past 20 years, improvements in systemic chemotherapy and surgical techniques have improved the survival of patients with hepatic metastases. For 4 decades, fluorouracil and leucovorin were the only drugs available to treat metastatic colorectal cancer, but several new drugs and a variety of novel regimens are now available. Further improvements in results have been seen with the delivery of chemotherapy via the hepatic artery. Surgical resection of liver metastases has been encouraged when possible, and recent advances in surgery such as portal vein embolization, have made liver resection a possibility for more patients. This review considers the timing and sequence of chemotherapy and surgery in this setting, as well as the roles of cryoablation, radiofrequency ablation, and radiation therapy.

The liver is a frequent site of metastatic disease, especially for cancers of the gastrointestinal tract. Since venous drainage from the colon and rectum flows via the portal vein to the liver,[1] it is not surprising that patients with colorectal cancer frequently develop liver metastases. Approximately 15% of patients will have liver metastases at the time of diagnosis, and another 60% of patients who develop metastatic disease will have metastases to the liver. For many years, the approach to patients with hepatic metastases was nihilistic. In the past 2 decades, improvements in systemic chemotherapy, in modalities to detect liver metastases, and in surgical techniques for hepatic resection, have improved the survival of these patients.

Systemic Chemotherapy

TABLE 1

Randomized Trials of First-Line Chemotherapy for Metastatic Colorectal Cancer

For 4 decades, fluorouracil (5-FU) plus leucovorin (LV) were the only drugs available to treat metastatic colorectal cancer, yielding a response rate of 20% to 30% and a median survival of 11 to 12 months. A number of new agents are now available, including irinotecan (Camptosar), a topo-isomerase inhibitor, and oxaliplatin (Eloxatin), a platinum compound with in vivo and in vitro activity against colon cancer cell lines and the ability to synergize with 5-FU.[2,3]

Randomized trials using irinotecan with 5-FU/LV vs 5-FU/LV alone[4,5] produced an increase in response rate and survival (Table 1). When irinotecan/5-FU/LV (IFL) was compared to oxaliplatin plus 5-FU/LV (FOLFOX), the response rate was increased from 35% to 45%, and survival was increased from 15 to 19.5 months for the IFL and FOLFOX groups, respectively.[6] That said, 5-FU given by continuous infusion is more effective than bolus 5-FU; therefore, when 5-FU administration was changed to infusion (ie, FOLFIRI rather than IFL), the regimen of irinotecan/5-FU/LV became more effective and produced results similar to those seen with FOLFOX.[7]

TABLE 2

Second-Line Therapy: Systemic Therapy in Previously Treated Colorectal Cancer Patients

In the past few years, targeted agents have become available: bevacizumab (Avastin), a monoclonal antibody to vascular endothelial growth factor, and cetuximab (Erbitux), an antibody to epidermal growth factor receptor. The addition of bevacizumab to IFL[8] increased response rates and survival (35% to 45%, and 15.6 to 19.5 months, for IFL vs bevacizumab/IFL, respectively). For almost 4 decades, the 2-year survival for metastatic colorectal patients treated with 5-FU or 5-FU/LV was 25%; with these new agents, 2-year survival rates have increased to 30%-39%, with a marked improvement in overall survival (Table 1).

In the second-line setting, even with the new agents, results are less impressive. Irinotecan is associated with a response rate of 5% to 14% and a median survival of 9.9 months.[9] FOLFOX administered to irinotecan-refractory patients produces a 9.9% response rate, with a median survival of 9.8 months (Table 2).[10] After progression on irinotecan, cetuximab alone can produce a 10% response rate, which increases to 23% when this agent is combined with irinotecan.[11] Second-line bevacizumab combined with FOLFOX increased survival to 12 months from 10 months for FOLFOX alone (Table 2).[12]

Hepatic Arterial Infusional Chemotherapy

To further improve on results, adding direct liver perfusion with chemotherapy might be useful. The first trials compared hepatic arterial infusion (HAI) alone to systemic chemotherapy. The rationale for HAI is based on the following facts: (1) liver metastases are perfused almost exclusively by the hepatic artery, whereas the normal liver is perfused by the portal vein,[13] (2) certain drugs are largely extracted by the liver during the first pass, allowing for minimal systemic toxicity,[14] and (3) the liver is often the first and only site of metastatic disease.[15] Therefore, aggressive treatment of metastases confined to the liver by resection or hepatic infusion may yield prolonged survival for some patients.

Regional therapy can be delivered using a hepatic arterial port or a totally implantable pump. Early studies with catheters produced clotting and bleeding that did not allow for long-term hepatic infusion with reliable patency.[16] Studies in Europe still use the catheters rather than pumps, perhaps explaining the inferior results in the European trials.

Randomized Trials

TABLE 3

Randomized Trials of Hepatic Arterial Infusion vs Systemic Chemotherapy for Unresectable Liver Metastases

Ten randomized studies have compared HAI to systemic chemotherapy (Table 3).[17-26] Almost all the studies showed an increase in response rate. Only the English trial that used HAI 5-FU failed to increase response rate.

Why did the superior response rates seen with HAI not translate into improved survival in the earlier trials? Most of these trials were too small, and a number of them allowed crossover to HAI at the time of progression on systemic chemotherapy, potentially diluting statistical results that might have demonstrated a survival benefit. Two earlier European studies demonstrated an increase in survival with HAI, but appropriate systemic therapy was not always used.

Two recent European trials did not show an increase in survival-the first being the Medical Research Council and European Organization for Research and Treatment of Cancer (EORTC) study,[26] which randomized patients to HAI 5-FU/LV via a port rather than a pump (using 5-FU rather than floxuridine [FUDR]) vs systemic 5-FU/LV. In this study, 37% of the patients assigned to the HAI arm did not receive treatment and 29% had to stop treatment. No differences were seen in response rate at 12 weeks (22% vs 19%, for HAI and systemic therapy, respectively), and no differences were seen in toxicity, progression-free survival (PFS), or survival. The study was not analyzed to look at the patients who actually received treatment.

A German cooperative group[17] randomized patients to HAI FUDR, HAI 5-FU/LV, or systemic 5-FU/LV. Tumor response rates were 43.2%, 45%, and 19.7%, and development of extrahepatic disease was 40.5%, 12.5%, and 18.3% for the HAI FUDR, HAI 5-FU/LV, and systemic 5-FU/LV groups, respectively. Toxicity data indicated that 5-FU/LV therapy was much more toxic than FUDR. A port was used, rather than a pump, and the FUDR regimen was different from what was used in the American studies in that the dosage was reduced from 0.2 to 0.15 mg/kg/d after three cycles rather than adjusting for patient toxicity. The median survival was 12.7, 18.7, and 17.6 months for the HAI FUDR, HAI 5-FU/LV, and systemic 5-FU/LV groups, respectively. Only 66% of patients randomized to HAI FUDR were treated, but all were included in the survival analysis. Eight patients in the HAI FUDR group died before ever receiving treatment, perhaps explaining the very low survival with HAI FUDR.

The Cancer and Leukemia Group B (CALGB)[18] trial differs from the other HAI studies in that it included the use of dexamethasone in the HAI arm.[27] HAI FUDR, dexamethasone plus LV was compared to systemic bolus 5-FU/LV. No crossover was allowed. The HAI group had a significant increase in survival (24.4 months, vs 20 months in the systemic group, P = .0034). The time to hepatic progression was better in the HAI arm (9.8 vs 7.3 months, P = .034) but the time to extrahepatic progression was better in the systemic arm (14.8 vs 7.7 months in the HAI group, P < .029). The toxicities were as expected, with a significant increase in diarrhea and neutropenia for the systemic arm and a significant increase in biliary toxicity for the HAI arm (18% vs 0%). A quality-of-life assessment was performed as part of the study, and the HAI group experienced improvement in this parameter, as measured at 3 and 6 months.

Differences between the CALGB study[18] and the European studies might explain differences in the outcomes. The CALGB study used pumps instead of ports, and HAI therapy included FUDR with dexamethasone to decrease toxicity.[27] Survival was based on intent to treat in all three studies, and the actual number of patients treated was much lower in the European studies-66% in the German study and 63% in the English study-while it was 86% in the CALGB study. The CALGB study did demonstrate that regional therapy alone can improve survival over systemic 5-FU/LV with a survival similar to that seen with newer agents. Randomized studies of HAI therapy vs the new therapies have not been conducted. In the future, studies comparing HAI or HAI plus new agents vs the new agents alone would be appropriate.

Second-Line Therapy

TABLE 4

Hepatic Arterial Infusion in Previously Treated Patients: Second-Line Therapy

HAI-based therapy in patients refractory to systemic chemotherapy has produced much higher response rates in small single-institution studies (Table 4). A trial using HAI FUDR/LV/dexamethasone[28] in both chemotherapy-naive and previously treated patients produced response rates of 72% and 52%, and median survivals of 23 and 13.5 months for the chemotherapy-naive and previously treated patients, respectively. HAI FUDR/dexamethasone plus mitomycin administered through the pump sideport, produced a 70% response rate in previously treated patients, with a median survival of 19 months from the start of HAI therapy after progression on systemic 5-FU/LV.[29]

A phase I study of HAI FUDR combined with systemic irinotecan in previously treated patients (45% had previously received irinotecan) reported a response rate of 74%, a time to disease progression of 8.1 months, and a median survival of 20 months.[30] A total of 13 of the 16 patients with prior irinotecan exposure responded to this regimen. Systemic oxaliplatin plus 5-FU/LV or oxaliplatin plus irinotecan with concurrent HAI FUDR/dexamethasone in 36 previously treated patients (74% had received prior irinotecan) produced response rates of 86%, with a median survival of 36 months, and a 1-year survival of 80%.[31] These and other small studies (Table 4) suggest that a benefit may be derived from HAI and systemic therapy as second-line therapy, and randomized studies exploring the use of HAI plus systemic therapy vs new agents alone in the second-line setting should be performed.

Toxicity of Hepatic Arterial FUDR Infusion

The most common problems with HAI therapy are hepatic toxicity and gastric ulcerations. Myelosuppression, nausea, vomiting, and diarrhea do not occur with HAI FUDR. If diarrhea does occur, shunting to the bowel should be suspected.[32] Clinically, biliary toxicity is manifested as elevations of aspartate transaminase, alkaline phosphatase, and bilirubin.[28] In early stages of the disease, hepatic enzyme elevations will return to normal when the drug is withdrawn and the patient is given a rest period, whereas in more advanced cases, it does not resolve. Therefore, careful monitoring of liver function tests is necessary to avoid this toxicity. The bile ducts derive their blood supply almost exclusively from the hepatic artery,[33] and thus are also perfused with high doses of chemotherapy during HAI treatment. HAI 5-FU causes fewer biliary problems and more arteritis.[34]

In patients who develop jaundice, an endoscopic retrograde cholangiopancreatogram (ERCP) may demonstrate lesions resembling idiopathic sclerosing cholangitis in 5% to 29% of patients treated. The strictures may be focal and present at the hepatic duct bifurcation, and therefore drainage procedures either by ERCP or by transhepatic cholangiogram may be helpful. Duct obstruction from metastases or bile duct strictures from surgery can also be causes of elevated bilirubin and must be ruled out before concluding that the elevation in liver function tests is due to HAI therapy.

Liver Resection

Since autopsy series show that in one-third of patients dying from colorectal cancer, the liver is the only site of metastatic disease, surgical resection of liver metastases has been encouraged when possible. More than 20 large series show that liver resections are possible with reasonable mortality and morbidity, producing an average 5-year survival of about 30%.[35-41]

No randomized studies of liver resection vs systemic chemotherapy have been performed. Historically, however, the median survival for patients with liver metastases is about 6 to 12 months. Even with the newer therapies, the 2-year survival rate is generally less than 40%, and this drops rapidly at 3 and 4 years. In contrast, patients with liver metastases who have undergone resection generally have a 2-year survival of around 60% to 70% and a 5-year survival of 25% to 30%.

Surgical Advances

Recent advances in surgery have made liver resection a possibility for more patients. One technique, portal vein embolization,[42] removes the blood supply to the affected liver and thereby induces hypertrophy of the nondiseased portion that will remain after liver resection. If small remnant parts of the liver can undergo hypertrophy, larger resections of diseased areas can be done. The use of intraoperative ultrasonography allows the surgeon to find lesions missed by other modalities and also helps define the relationship of the tumor to hepatic vasculature.[43] Improved vascular clamping techniques,[44] controlled anatomic resection,[45] and other new strategies are outlined in a review by Khatri et al.[46] The use of radiofrequency ablation to small lesions in the remaining liver also extends liver resection options to more patients. Some reports have stated that even patients with positive nodes or small lung metastases can undergo liver resection with better outcomes than those seen with chemotherapy alone,[46] but these data are not widely accepted.

Patient Selection

There is no consensus for determining which patients are suitable for resection. A computer program (the OncoSurge model) was developed to aid in selecting these patients.[47] A panel of 16 experts from oncology, radiology, and liver surgery developed hypothetical patient profiles to decide when resection should be done first, chemotherapy should be done first followed by possible resection, or resection should be absolutely contraindicated. This consensus panel decided that absolute contraindications were unresectable extrahepatic disease, more than 70% liver involvement, liver failure, and being surgically unfit. Factors that did not influence treatment strategy were age, primary tumor stage, timing of metastasis detection, past blood transfusion, liver resection type, preresection carcinoembryonic antigen (CEA), or previous hepatectomy. They also suggested that patients with bilobar disease or more than four metastases receive resection only after tumors were shrunk with chemotherapy.

Prognostic Indicators

A number of published studies have attempted to identify prognostic factors associated with poorer outcomes, and two large series have developed a scoring system.[48,49] At Memorial Sloan-Kettering Cancer Center (MSKCC), for the 1,001 patients undergoing liver resection, the most important factors adversely affecting survival were size of tumor (> 5 cm), disease-free interval (< 12 months), number of tumors (> 1), lymph node-positive primary, and preoperative CEA (> 200 ng/mL). Giving 1 point for each of these factors, a score was obtained that correlated with survival. For scores of 0, 1-2, 3, and 4-5, the predicted 5-year survival rates were 60%, 42%, 20%, and 18%, respectively.[48]

Nordlinger et al[49] reviewed data on 1,568 patients who underwent liver resection for colorectal metastases. The 2- and 5-year survival rates of 64% and 28% were affected by age, size of the largest metastases, CEA level, stage of the primary, disease-free interval, number of liver nodules, and resection margin. Giving 1 point to each of these risk factors, the investigators calculated scores that divided the patients into three risk groups: 0-2, 3-4, and 5-7 with corresponding 2-year survival rates of 79%, 60%, and 43%, respectively.

The Liver Met Survey is an international Internet-based registry that includes 2,122 patients. In a recent report on this registry, the overall 5- and 10-year survival rates were 42% and 26%. The 5-year survival rates for three or fewer vs more than three nodules were 48% vs 24% (P = .0001). Another factor that affected survival was whether the tumor was unilateral or bilateral. In this review, preoperative chemotherapy did not benefit patients with solitary metastases (5-year survival rate: 45% vs 58%), whereas for patients with more than five metastases, the 5-year survival rate was 22% and 12% for patients with and without prior chemotherapy.[50]

Is Neoadjuvant Chemotherapy Useful?

TABLE 5

Resectability Rate After Systemic Chemotherapy: Retrospective Studies

The use of new effective systemic chemotherapy has increased the possibility of obtaining a response and thus making liver metastases resectable. Several systemic chemotherapy studies have retrospectively evaluated resectability after chemotherapy (Table 5). The first retrospective reviews were written by Bismuth et al[51] and Giacchetti et al.[52] Adam and colleagues,[53] from the same group, reviewed the records of 1,104 patients with unresectable disease who received mainly FOLFOX as neoadjuvant therapy; 12.5% became resectable , and the 5-year survival rate of these patients was 34%-similar to what could be obtained with patients who were initially resectable.

TABLE 6

Prospective Trials Evaluating Neoadjuvant Systemic Chemotherapy for Unresectable Disease: Resectability Rate

Recently, several trials were designed to address resectability after chemotherapy (Table 6), but variations in patient selection makes it difficult to compare these studies. The Mayo Clinic trial considered patients to be unresectable if they had (1) involvement of all three major hepatic veins, portal vein bifurcation, or the hepatic vena cava; (2) involvement of the main right or main left portal vein and the main hepatic vein of the opposite lobe; (3) disease requiring more than a right or left trisegmentectomy; or (4) six or more metastatic lesions distributed diffusely in both lobes of the liver. In the presence of any of the above, patients were treated with preoperative oxaliplatin/5-FU/LV. Of 44 patients, 17 (38%) became suitable for resection, but after resection, 73% had a recurrence in the liver. Median survival was 26 months.[54] A retrospective review of these data demonstrated that 10% of patients were actually resectable prior to neoadjuvant therapy.

Pozzo et al treated 40 patients who had unresectable disease in a nonrandomized trial using neoadjuvant FOLFIRI.[55] Their criteria for unresectability were (1) six metastases or three per lobe, (2) size > 5 cm for one lesion if six metastases were present, or (3) contiguity with two hepatic veins, inferior vena cava, or liver hilum. Overall, 32% of patients were able to under-go liver resection after chemotherapy. With a median follow-up of 30 months, the disease-free survival of resected patients is presently 28 months, and overall survival has not been reached. Other trials evaluating neoadjuvant chemotherapy are listed in Table 6.

Improving Outcomes

TABLE 7

Initially Unresectable Disease: Resectability After HAI With or Without Systemic Therapy

How can we improve outcomes for unresectable patients? The use of targeted agents such as cetuximab or bevacizumab may be helpful. Rougier et al treated 23 patients with cetuximab and FOLFIRI, and 7 (30%) became resectable.[56] Would adding HAI therapy improve results? In a phase I study of oxaliplatin/irinotecan plus HAI chemotherapy in 44 clearly unresectable patients, 34% became resectable (70% of patients were receiving this treatment as second- or third-line therapy).[31] The median survival of the entire group is 36 months, and median survival for the resected group has not been reached (Table 7).

In a Japanese study of 51 patients with unresectable disease with HAI 5-FU and systemic UFT (tegafur/uracil), 31 patients were judged to be resectable after preoperative chemotherapy, but only 24 agreed to surgery. The 3- and 5-year survivals were 58% and 42% for the resected group, but for those patients who did not have resection, the 3- and 5-year survivals were 25% and 0%.[57] Using all three active agents may increase resectability. The Gruppo Oncologico Nord Ovest (GONO) randomized 244 patients to FOLFIRI or FOLFIRI plus oxaliplatin (FOLFOXIRI). In patients with only liver metastases, the resection rate was 12% and 36% for the FOLFIRI and FOLFOXIRI regimens, respectively.[58]

Predicting Poor Outcomes

Adam,[53] in a retrospective review of 131 patients who underwent liver resection after preoperative -systemic chemotherapy, found in a multivariate analysis that tumor progression on chemotherapy, elevated CA 19-9, and an increase in the number of metastases were all predictors of poor outcome.

Even if metastases disappear on CT, they may remain viable. In a review by Nordlinger et al on 586 patients[59] treated at their institution, 38 patients had disappearance of at least one lesion on CT. Pathologic examinations of sites that were considered to have a complete response showed that in 12 of 15 (80%) of these lesions, there were still viable tumor cells on pathologic examination. Areas where there was a complete response (although they could not be resected at the time of surgical exploration) were closely followed. Of these 31 sites, 23 (74%) developed a reccurrence. Clearly, this analysis shows that further postresection therapy needs to be given to these patients if sites are left intact at the time of liver resection.

Toxicity of Neoadjuvant Chemotherapy

As use of preoperative chemotherapy increases, reports are emerging about liver toxicity associated with this strategy. Investigators have found an increased incidence of steatosis, sinusoidal abnormalities, veno-occlusive disease, and steatohepatitis in this setting. In the Rubbia-Brandt[60] report on 153 patients undergoing liver resection, 51% of the 87 patients who received chemotherapy prior to resection had sinusoidal dilation, while 66 patients treated with surgery alone did not. Of 43 patients who had received prior oxaliplatin, 34 (78%) showed striking sinusoidal alterations, and 48% had veno-occlusive fibrosis.

Kooby reported on 325 patients with fatty livers undergoing resection at MSKCC. They compared this group to 160 patients with normal livers. Men and those with higher body mass index were more likely to have steatosis. Patients treated with preoperative chemotherapy were more likely to have steatosis (66%), with marked steatosis being an independent predictor of complications following hepatic resection.[61]

M. D. Anderson researchers reported no increase in preoperative deaths from neoadjuvant irinotecan in 2003.[62] However, they recently reported a 20% incidence of steatohepatitis in patients receiving preoperative irinotecan vs 4.4% in those who received no prior chemotherapy, and an 18.9% incidence of sinusoidal dilation in those receiving preoperative oxaliplatin vs 1.9% in those receiving no prior chemotherapy. Moreover, patients with steatohepatitis had an increased 90-day mortality (14.7% vs 1.6%, P = .001).[63]

Timing of Surgery and Chemotherapy

Internal server error