First-Line Chemotherapy vs Bowel Tumor Resection Plus Chemotherapy for Patients With Unresectable Synchronous Colorectal Hepatic Metastases FREE

Colorectal cancer (CRC) is the third most prevalent cancer in the world, with more than 940 000 cases and nearly 500 000 deaths occurring annually worldwide.¹ Almost 20% to 25% of patients with CRC (hereinafter referred to as CRC patients) present with distant metastases at the time of diagnosis.^2- 3 The liver is involved in 80% to 90% of the cases, and in almost 50% it is the only site of metastasis.⁴ From 10% to 30% of CRC patients with newly diagnosed disease and liver-only deposits have resectable metastases at presentation. In such instances, resection of the primary tumor with immediate or delayed hepatic resection is the only potentially curative option with a low mortality rate and 5-year survival rates of 18% to 46%.^5- 6 However, an estimated 70% to 90% of these patients have unresectable hepatic metastases.⁷ In such cases, surgery is a necessary indication in symptomatic patients with tumor complications such as intestinal obstruction, perforation, or hemorrhage, whereas therapeutic options for asymptomatic or relatively asymptomatic patients (nearly 80% of all cases) are still debated.^8- 10 Some authors recommend resection of the bowel tumor followed by systemic chemotherapy to perform correct abdominal tumor staging, prevent tumor complications, and potentially improve patients' status and efficacy of chemotherapy through tumor load reduction.^3,11- 13 In recent studies, other authors^14- 17 have argued that bowel tumor resection may result in high mortality and morbidity rates and delay systemic chemotherapy and proposed immediate administration of systemic chemotherapy, with surgery reserved for treatment of tumor complications or, if tumor shrinkage occurs, for a potentially curative resection. Thus, the best treatment strategy for asymptomatic CRC patients with unresectable liver-only metastases still remains unclear, warranting additional investigations.^7,18

In this study, we report on our experience on treatment of asymptomatic CRCs with unresectable metastases confined to the liver. Specifically, we retrospectively compared clinical outcome in patients undergoing bowel tumor resection plus chemotherapy (resection group) with that in patients treated with chemotherapy first (chemotherapy group), in an effort to identify selection criteria for the most beneficial treatment of these patients.

Clinical records from 53 consecutive patients with stage IV bowel cancer and unresectable liver-only metastases were retrospectively selected from the database of the Division of Surgical Oncology from January 4, 1995, to December 23, 2005. These patients had hepatic metastases judged unresectable because of invasion of major liver pedicles (metastatic disease adjacent to or involving all 3 hepatic veins, and/or the portal vein bifurcation, and/or the retrohepatic vena cava, and/or the vascular structures of the opposite lobe) or intrahepatic dissemination (bilobar disease) requiring a liver resection potentially capable of jeopardizing postoperative liver function. Eight of these 53 patients (15%) were symptomatic and needed immediate surgical exploration to manage tumor complications. Among the remaining 45 asymptomatic or relatively symptom-free patients (ie, complaining only of constitutional symptom such as fatigue or anorexia), 3 received only supportive care because of very poor general conditions, and the remainder underwent resection of the bowel tumor. The 42 remaining patients represented the resection group of this study.

The chemotherapy group consisted of 23 asymptomatic or relatively symptom-free CRC patients with unresectable liver-only metastases who had refused surgery and were thus treated with chemotherapy first. These patients were treated by the medical oncology team in our department during the study period. Their clinical records were retrieved from the medical oncology database.

All tumors were staged on evaluation of findings of physical examination, routine laboratory tests, and diagnostic imaging (including complete colonoscopy with biopsy, abdominal ultrasonography, chest and abdomen computed tomography scintigraphic bone scan, and, more recently, whole-body magnetic resonance imaging and fludeoxyglucose F18–labeled positron emission tomography).

Liver metastases were staged according to the degree of hepatic parenchymal replacement by tumor (ie, H1, <25%; H2, 25%-50%; and H3, >50% of whole liver volume) by means of computer-aided analysis of computed tomography with a dedicated software. In the resection group, all patients underwent resection of the bowel cancer; no explorative laparotomy or colon bypass was performed in these patients. First-line systemic chemotherapy based on intravenous fluorouracil plus folinic acid or, after 2001, fluorouracil, folinic acid, and oxaliplatin or fluorouracil, folinic acid, and irinotecan hydrochloride was started immediately after diagnosis in the chemotherapy group and at least 3 weeks after surgery in the resection group. The response to systemic therapy was evaluated every 3 months and metastases were restaged with regard to their resectability. A switch to second-line chemotherapy, with the eventual addition of new biological therapies such as bevacizumab and cetuximab, was decided in cases of partial response (metastatic reduction <25%) or disease progression. All of the patients were available for follow-up and the statistical analysis was completed by July 31, 2006.

The main characteristics of the patient population are summarized in Table 1. In the resection group, extent of the primary tumor and nodal status (T and N variables of the TNM staging system),¹⁹ degree of histologic differentiation (well, moderate, or poor), tumor growth pattern (expanding or infiltrating),²⁰ number of resected nodes, tumor size, curative or noncurative resection of the bowel tumor (defined as macroscopic and microscopic removal of the tumor, free resection margins, absence of peritoneal diffusion, and lymphadenectomy extended beyond involved nodes at the postoperative pathological examination), and postoperative complications were also recorded. Finally, in the chemotherapy group, the complication rate related to unresected bowel tumors was recorded.

We used the BMDP statistical package (BMDP Statistical Software Inc, Los Angeles, California) to perform statistical analysis. In all analyses, the significance level was specified as P < .05. The equality of group means and comparisons between proportions were analyzed by using the unpaired t test and χ² test with the Yates correction, respectively. Univariate statistical analysis related to survival was determined by the log-rank test (Mantel-Cox). Curves were plotted using the product-limit method (Kaplan-Meier) and analyzed using the generalized Savage test or the Mantel-Cox test (BMDP1L; BMDP Statistical Software, Inc). The independent significance of prognostic variables related to overall survival was determined by multivariate analysis, using the Cox proportional hazards model. The level of significance was obtained by means of a score test (BMDP2L; BMDP Statistical Software, Inc). Finally, a stepwise multivariate analysis was performed to generate a model of the best linear combination of variables able to predict overall survival.

The 2 groups were well matched with regard to clinical characteristics (Table 1). In the resection group, 4 patients (9.5%) underwent noncurative resection of the bowel tumor owing to diffuse nodal and/or peritoneal spreading. No postoperative mortality was recorded. Nine patients (21.4%) experienced slight postoperative complications. All patients could be discharged from hospital and were able to undergo systemic chemotherapy as scheduled.

The mean (SD) follow-up time was 21 (15) months (range, 5-61 months; median, 16 months). None of the patients was readmitted because of bowel obstruction or other late complications of surgery. During follow-up, 28 patients (67%) died of disease within 5 to 47 months after tumor resection, whereas 14 patients are still alive. Of interest, 5 of the surviving patients (12% of the resection group) underwent curative hepatic resection after metastasis downstaging by means of chemotherapy, and they are disease free at 12, 17, 37, 43, and 61 months after bowel resection (2 patients underwent a successful hepatic second resection for liver recurrence). The remaining 9 patients are still being administered systemic chemotherapy. The mean and median overall survival times were 26 and 15 months, respectively; the 1- to 4-year actuarial survival rates were 72%, 38%, 35%, and 15% (Figure 1).

In the chemotherapy group, the mean (SD) follow-up time was 16 (12) months (range, 5-45 months; median, 12 months). The mean (SD) interval from the diagnosis to the start of chemotherapy was 8 (2) days (range, 6-12 days; median, 8 days), which was significantly shorter than that observed in the resection group, as expected. Seven patients (30%) had severe complications related to unresected bowel tumor. Two colon perforations, 1 intestinal hemorrhage, and 1 bowel obstruction required emergency surgery. Three cases of intestinal obstructions were initially managed with endoscopic stent placement, but 2 patients required surgery owing to bowel perforation or stent dislocation. In these 7 patients, complications occurred 3, 7, 9, 10, 14, 16, and 22 months (mean, 11 months; median, 10 months) after the initial diagnosis. In 3 of these patients, a reduction in the size of metastatic liver lesions was obtained by chemotherapy, although they still remained unresectable. Surgical mortality and morbidity rates were 14% (1 patient died of septic shock after bowel perforation), and 43% (3 patients showed prolonged intestinal paralysis, pleural effusion, and wound infection), respectively. All but 3 patients were able to continue systemic chemotherapy. Twenty-one patients (91%) died of disease, whereas 2 patients were still alive 37 and 43 months after the start of chemotherapy. The first is alive with unresectable disease in the liver and stented bowel tumor; the second, after shrinkage of liver metastases, underwent successful synchronous curative resection of the bowel tumor and hepatic deposits. The mean and median overall survival time were 17 and 12 months, respectively; the 1- to 4-year actuarial survival rates were 61%, 17%, 13%, and 0%, respectively (Figure 1).

Overall, a curative hepatic resection after shrinkage of initially unresectable liver metastases was possible in 6 patients and it appeared to be dependent on the type of treatment (5 in the resection group and 1 in the chemotherapy group), extent of liver involvement, and performance status (Table 2), although the small sample size did not allow reaching a significant difference.

In the whole series, age, sex, site of tumor bowel, basal serum levels of carcinoembryonic antigen and carbohydrate antigen 19-9, degree of histologic differentiation, tumor growth pattern, and different treatment schedules (before and after 2001) were not significantly associated with overall survival.

Likewise, in the resection group, tumor size, T and N staging of bowel tumor, number of resected nodes, and postoperative complications were not shown to correlate with the long-term outcome. Although patients administered novel chemotherapeutic agents, such as oxaliplatin or irinotecan, displayed no better survival rates than patients administered fluorouracil-based chemotherapy only, it has to be emphasized that the 4-year actuarial survival rate was 40% in the former and 3% in the latter (46% and 7%, respectively, in the resection group and 22% and 0%, respectively, in the chemotherapy group). On the contrary, performance status, basal serum levels of lactic dehydrogenase and alkaline phosphatase, percentage of liver involvement, potentially curative resection of the bowel tumor, and type of treatment (resection vs chemotherapy) were demonstrated to be the only variables significantly correlated with long-term survival (Table 3). On multivariate analysis, performance status, liver involvement, and modality of primary treatment were shown to be the only covariates independently associated with survival rate. Patients in the chemotherapy group had a 2-fold increased risk of cancer-related death than patients undergoing resection of bowel tumor. Metastatic seeding involving more than 50% of the whole liver was associated with a 5-fold increased risk of cancer death than more limited hepatic spread. After backward elimination, performance status was removed from the model; stepwise regression selected liver involvement and type of treatment as the best combination of variables capable of predicting long-term survival (Table 4). On the basis of the model generated by the stepwise process, a 4-year survival rate was plotted stratifying for liver involvement and treatment strategy. The mean and median survival times and the 4-year survival rate for 27 patients with H1 to H2 liver involvement undergoing potentially curative bowel tumor resection were 34.5 and 25.0 months and 21.5%, respectively. These figures were significantly better than those observed in 16 patients in the chemotherapy group who had H1 and H2 liver involvement (mean and median survival times, 20.6 and 13 months, respectively; 4-year survival rate, 0%) (Figure 2).

Treatment options for asymptomatic or relatively symptom-free CRC patients with unresectable liver-only metastases are still debated.^9- 10 For several decades tumor resection followed by chemotherapy has been considered the treatment of choice because it allows precise definition of the tumor stage (peritoneal carcinomatosis or other misdiagnosed metastases), treatment of associated symptoms, prevention of local complications, provision of psychological benefit for the patients who believe that the cancer has been completely removed, and, finally, performance of a reduction of neoplastic masses.^8,12 In particular, tumor debulking was considered very important because it could result in more effective postoperative chemotherapy, as previously demonstrated for advanced renal cell and ovarian cancers.^21- 22 Major drawbacks of this approach are high postoperative mortality and morbidity rates, poor control of symptoms when these are related to metastases, and, in particular, a remarkable delay in starting chemotherapy.^17,23 To avoid delayed chemotherapy, Sarela et al²³ first suggested modifying the general approach taken with these patients by immediately starting chemotherapy so as to treat systemic disease and to identify responders to chemotherapy who would have a better prognosis. Major drawbacks of this nonoperative treatment are the low percentage of responders, lack of long-term survivors, and risk of intestinal complications caused by unresected CRCs requiring, in turn, emergency surgery with high mortality and morbidity rates owing to chemotherapy-induced immunosuppression.⁷

Thus far, only 6 trials have compared the following 2 modalities of treatment: tumor resection followed by systemic chemotherapy, or systemic chemotherapy first with surgery only in case of complications or metastases that become amenable to curative resection. In 4 series, high postoperative morbidity rates, low rates of emergency surgery for tumor complications, and equivalent median survival times suggested that asymptomatic CRC patients with unresectable metastases could be safely treated without tumor resection.^14- 17 On the contrary, 2 other reports showed low postoperative morbidity rates, high frequency of tumor complications requiring emergency surgery in patients who did not undergo resection, and significantly different 2-year survival rates between patients undergoing tumor resection and patients treated with chemotherapy.^13,18 However, excluding a case-matched study,¹⁷ the remaining trials compared nonhomogeneous patient populations, thus complicating interpretation of results. Therefore, it is still unclear how to best treat these patients, considering that even the most recent strategy proposed by Mentha et al²⁴ requires further validation.^9- 10

To overcome the limits intrinsic to noncomparative series, patients included in this study were all treated at the same institution and had matched clinical and tumor characteristics when subgrouped according to treatment option. In the resection group, surgery allowed more precise staging of abdominal disease, as in 4 cases, for instance, where it turned out to be noncurative owing to preoperative undiagnosed diffuse cancer spreading. The postoperative complication rate was low and chemotherapy could be administered to all patients after 3 weeks. Mean and median survival times and the actuarial survival rate were significantly better than those recorded in the chemotherapy group and in previous series dealing with patients undergoing systemic chemotherapy first.^14- 17 Although these latter patients could benefit from a shorter interval from the diagnosis to the start of chemotherapy, the complication rate related to unresected tumor was not negligible, frequently requiring emergency surgery with high postoperative morbidity. Along with good performance status and limited liver involvement, tumor resection was shown to be an independent variable associated with survival rate. Of interest, the best model to predict long-term survival included curative tumor resection along with a hepatic replacement volume of 50% or less of the whole liver. To our knowledge, this is the first time that tumor resection has been demonstrated to be an independent prognostic factor in stage IV CRC patients.

However, shrinkage of liver deposits to allow curative hepatic resection is crucial to obtain long disease-free survival in patients with unresectable liver-only metastases at presentation.^10,25 In previous reports, potentially curative treatment of downstaged liver metastases ranged from 1.9% to 9% in patients treated with chemotherapy first^15,17,23; conversely, these percentages were higher (15.6%-29%) in patients undergoing curative bowel resection.^17,26- 27 In the present series, the curative hepatic resection rate was almost 3-fold higher in the resection group than in the chemotherapy group; patients with limited hepatic involvement who underwent resection had a significantly better survival rate than those with comparable liver involvement who did not, suggesting that debulking of primary tumor may improve outcome of postoperative chemotherapy.¹³

Recently, an increased hepatic resection rate (33.3%) has been reported with an aggressive approach that includes curative bowel resection followed by hepatic arterial infusion and systemic chemotherapy²⁸; this percentage is expected to exceed 40% by extending surgical frontiers of distant metastases and using more effective novel immunochemotherapeutic agents such as bevacizumab and cetuximab.^27,29 Thus, to obtain a long disease-free survival, aggressive chemotherapy that eventually includes hepatic arterial infusion, major hepatectomies, and resections of other involved organs such as lung and brain must be anticipated.^25,28,30 In this scenario, immediate bowel tumor resection will allow shifting the focus of treatment toward distant metastatic sites that will eventually require aggressive surgery.

Novel surgical and chemotherapeutic treatment options could change our attitude toward asymptomatic CRC patients with unresectable liver-only metastases by giving us the tools to pursue prolonged survival, improved quality of life, and even complete cure. Bowel tumor resection allows precise disease staging, avoids complications related to unresected colon cancer, does not prevent postoperative therapy, appears to promote metastatic shrinkage and improved antitumor therapy, and allows aggressive surgery toward metastatic sites. According to our experience and those of others,^2,11,13 such a treatment option is feasible for patients showing good performance status, metastatic liver involvement of less than 50% of the whole hepatic volume, and conditions for potentially curative resection of bowel tumor. In the remaining patients, chemotherapy first is preferable as palliative treatment while waiting to switch to curative surgery of distant metastases.

Correspondence: Gennaro Galizia, MD, Second University of Naples School of Medicine, F. Magrassi–A. Lanzara Department of Clinical and Experimental Medicine and Surgery, c/o II Policlinico, Edificio 17, Via Pansini, 5, 80131 Naples, Italy (gennaro.galizia@unina2.it).

Accepted for Publication: December 7, 2006.

Author Contributions:Study concept and design: Galizia, Lieto, and Orditura. Acquisition of data: Galizia, Lieto, Castellano, Imperatore, Pinto, and Zamboli. Analysis and interpretation of data: Galizia, Lieto, Imperatore, Pinto, and Zamboli. Drafting of the manuscript: Galizia, Lieto, and Orditura. Critical revision of the manuscript for important intellectual content: Galizia, Lieto, Orditura, Castellano, Imperatore, Pinto, and Zamboli. Statistical analysis: Galizia and Imperatore. Administrative, technical, and material support: Galizia, Lieto, Orditura, Castellano, and Pinto. Study supervision: Castellano and Zamboli.

Financial Disclosure: None reported.

Additional Contributions: Fortunato Ciardiello, MD, PhD, and Ferdinando De Vita, MD, PhD (Division of Medical Oncology), Filippo Beneduce, MD, PhD, Ilario de Sio, MD, PhD, and Luigi Vitale, MD, PhD (Division of Gastroenterology), and Ciro Romano, MD, PhD (Division of Internal Medicine, Allergy and Clinical Immunology) of Second University of Naples School of Medicine provided helpful suggestions and excellent collaboration. Carlo Pignatelli, MD, PhD (Second Division of General Surgery, Second University of Naples School of Medicine), provided invaluable collaboration and very constructive discussions and criticisms.