Fax: (713) 745-1921
Is resection of colorectal liver metastases after a second-line chemotherapy regimen justified?†
Version of Record online: 28 MAR 2011
Copyright © 2011 American Cancer Society
Volume 117, Issue 19, pages 4484–4492, 1 October 2011
How to Cite
Brouquet, A., Overman, M. J., Kopetz, S., Maru, D. M., Loyer, E. M., Andreou, A., Cooper, A., Curley, S. A., Garrett, C. R., Abdalla, E. K. and Vauthey, J.-N. (2011), Is resection of colorectal liver metastases after a second-line chemotherapy regimen justified?. Cancer, 117: 4484–4492. doi: 10.1002/cncr.26036
We thank Stephanie P. Deming and Ruth J. Haynes for editing the manuscript.
- Issue online: 16 SEP 2011
- Version of Record online: 28 MAR 2011
- Manuscript Accepted: 3 JAN 2011
- Manuscript Revised: 22 NOV 2010
- Manuscript Received: 13 SEP 2010
- National Institutes of Health
- Anderson Cancer Center. Grant Number: CA016672
- colorectal liver metastases;
- liver resection;
- second-line chemotherapy;
Patient outcomes following resection of colorectal liver metastases (CLM) after second-line chemotherapy regimen is unknown.
From August 1998 to June 2009, data from 1099 patients with CLM were collected prospectively. We retrospectively analyzed outcomes of patients who underwent resection of CLM after second-line (2 or more) chemotherapy regimens.
Sixty patients underwent resection of CLM after 2 or more chemotherapy regimens. Patients had advanced CLM (mean number of CLM ± standard deviation, 4 ± 3.5; mean maximum size of CLM, 5 ± 3.2 cm) and had received 17 ± 8 cycles of preoperative chemotherapy. In 54 (90%) patients, the switch from the first regimen to another regimen was motivated by tumor progression or suboptimal radiographic response. All patients received irinotecan or oxaliplatin, and the majority (42/60 [70%]) received a monoclonal antibody (bevacizumab or cetuximab) as part of the last preoperative regimen. Postoperative morbidity and mortality rates were 33% and 3%, respectively. At a median follow-up of 32 months, 1-year, 3-year, and 5-year overall survival rates were 83%, 41%, and 22%, respectively. Median chemotherapy-free survival after resection or completion of additional chemotherapy administered after resection was 9 months (95% confidence interval, 4-14 months). Synchronous (vs metachronous) CLM and minor (vs major) pathologic response were independently associated with worse survival.
Resection of CLM after a second-line chemotherapy regimen was found to be safe and was associated with a modest hope for definitive cure. This approach represents a viable option in patients with advanced CLM. Cancer 2011;. © 2011 American Cancer Society.
Colorectal cancer is one of the leading causes of cancer death worldwide and was responsible for an estimated 49,920 deaths in 2009 in the United States.1 In most cases, death occurs at the end of the evolution of metastatic disease, and in the majority of patients, the liver is the first site of metastasis. In patients with colorectal liver metastases (CLM), liver resection is the only treatment associated with long-term survival. In patients with unresectable CLM, chemotherapy is the only therapy with a proven survival benefit, and resection may be possible in case of good response to chemotherapy.2 In patients with resectable CLM, preoperative systemic chemotherapy has been used as part of the treatment strategy to decrease the risk of recurrence after resection.3 Disease progression during preoperative chemotherapy predicts poor outcome after resection.4 For patients with advanced disease and progression during preoperative chemotherapy, a change to an alternate chemotherapy regimen is usually recommended, and liver resection may be reconsidered if there is disease response or stabilization in response to the alternative chemotherapy.5 However, objective response rates to second-line preoperative chemotherapy are only 4%-28%, hence surgery is rarely an option after failure of first-line chemotherapy.6-10 Furthermore, the safety, efficacy, and outcome of hepatic surgery in patients who received multiple lines of chemotherapy have been evaluated only in small series of patients who received cetuximab or intra-arterial hepatic artery chemotherapy infusion as a “rescue” regimen.11-13
In this study, we analyzed the feasibility and outcomes of resection of CLM after a second-line chemotherapy regimen. We aimed to determine the role of liver surgery in patients with advanced metastatic disease who received multiple lines of chemotherapy because of suboptimal response or toxicity after first-line chemotherapy.
MATERIALS AND METHODS
Patient Population and Definitions
Between August 13, 1998, and June, 17, 2009, data from 1099 consecutive patients who underwent surgical resection for CLM at 1 institution were collected prospectively. Inclusion criteria for this study were resection of CLM following second-line (2 or more) chemotherapy regimens and at least 1 year of follow-up after surgery. The study was approved by the institutional review board of The University of Texas MD Anderson Cancer Center (RCR01-116).
At The University of Texas MD Anderson Cancer Center, the institutional guidelines for patients with CLM include preoperative chemotherapy in almost all patients.5 Exceptions to this practice are patients who develop CLM less than 1 year after adjuvant chemotherapy for the primary tumor and patients with small volume CLM at risk of disappearing during chemotherapy and that cannot have their position radiographically labeled by a fiducial marker.14, 15 In patients with initially resectable CLM, preoperative short-course chemotherapy (8-12 weeks) is used. Patients with advanced CLM initially unsuitable for resection are seen by the surgeon prior to initiation of chemotherapy to determine whether the patient may be a surgical candidate with a chemosensitive tumor (radiographic response to chemotherapy). In patients with initially unresectable disease, surgical resection is considered as soon as the lesions become resectable.5 During treatment, restaging is performed every 3 to 4 cycles on the basis of clinical assessment and imaging. For this study, imaging studies were retrospectively reviewed for each patient to measure response according to the Response Evaluation Criteria in Solid Tumors (RECIST)16 (response defined as complete response, partial response, stable, or progressive disease) and morphologic criteria17 (response defined as none, incomplete, or optimal response). Second-line chemotherapy was defined as any change in chemotherapy regimen administered for CLM except for the discontinuation of bevacizumab 6-8 weeks before surgery. Adjuvant chemotherapy administered for the primary tumor was not considered first-line chemotherapy for CLM except in patients with CLM detected at the same time as the primary tumor (synchronous CLM) and patients who developed CLM during adjuvant chemotherapy for the primary tumor. Reasons for receiving second-line chemotherapy were classified into the following categories: (1) progression during first-line chemotherapy, (2) advanced metastases with suboptimal response to the first-line chemotherapy regimen, or (3) intolerable toxicity of the first-line chemotherapy regimen. A suboptimal response was defined as a response insufficient to allow safe resection based on volumetric assessment.
Surgical Procedure, Postoperative Complications, and Pathologic Findings
Liver resections were performed only with curative intent—that is, only if it was believed that a complete tumor resection could be achieved. During laparotomy, the peritoneal cavity was inspected to rule out extrahepatic spread of disease. Hepatic palpation and intraoperative ultrasonography were performed in all cases to better define the location of the CLM in the liver and their relationship to portal pedicles and hepatic veins. Major hepatectomy was defined as hepatic resection including 3 or more contiguous liver segments. Radiofrequency ablation was used in combination with liver resection and only for small tumors in selected patients when a complete resection with staged resections and portal vein embolization were not otherwise possible. Data on postoperative morbidity and 90-day mortality were collected prospectively. The severity of postoperative complications was graded using the classification of Dindo et al.18 Complete pathologic response was defined as 0% of tumor cells viable, major pathologic response as 1%-49% of tumor cells viable, and minor pathologic response as 50% or more of tumor cells viable.19 Changes of the nontumorous liver parenchyma, including nonalcoholic fatty liver disease (steatosis and steatohepatitis) and sinusoidal dilation, were assessed semiquantitatively as described previously.20-22 Hepatic injury was defined as steatosis in more than 30% of the hepatocytes and/or the presence of steatohepatitis (Kleiner score >4) and/or moderate or severe sinusoidal dilation.
The decision whether to deliver additional chemotherapy after resection was made at the time of the follow-up postoperative visit on an individualized basis, based on the presence of possible residual disease, response to preoperative chemotherapy on imaging,16, 17 tolerance and number of cycles of preoperative chemotherapy, and pathologic response. The policy was to avoid chemotherapy after resection when possible, because most patients who undergo resection of CLM have already received prolonged chemotherapy. Patients were reassessed via physical examination, carcinoembryonic antigen serum measurement, and multiphase (liver protocol) computed tomography examination every 3-4 months after resection, and decisions about further treatment were made according to the findings on follow-up workup.
Quantitative and qualitative variables were expressed as mean ± standard deviation, median (range), and frequency. Overall and disease-free survival were calculated using the Kaplan-Meier method from the date of resection of CLM and compared using a log-rank test. Chemotherapy-free survival was calculated from the date of resection of CLM in patients who did not receive postoperative chemotherapy and from the date of last dose of postoperative chemotherapy in those who did receive postoperative chemotherapy.
For the detection of factors associated with survival after resection of CLM following second-line chemotherapy regimen, univariate analysis was used to examine the relationship between overall survival and the following clinicopathologic variables: rectal (vs other) location of the primary tumor, presence of regional lymph node metastases (v absence), synchronous (vs metachronous) CLM, multiple (vs single) CLM (greater than 2), size of CLM measured at the time of diagnosis (greater or less than 5 cm), presence of extrahepatic disease, preoperative serum CEA level >5 ng/dL, progression of disease during first-line chemotherapy, intolerable toxic effects during first-line chemotherapy, total number of cycles of chemotherapy, number of cycles of the last chemotherapy regimen, partial or stable disease after last chemotherapy regimen according to RECIST criteria, morphologic response after last chemotherapy regimen, occurrence of major postoperative complications, pathologic response (complete or major vs minor), and postoperative chemotherapy. All variables associated with survival with P < .2 in univariate proportional hazards model were subsequently entered into a Cox multivariate regression model with backward elimination. P < .05 was considered statistically significant. Comparisons between groups were analyzed via chi-squared or Fisher's exact test for proportions, Mann-Whitney U test for medians, and Student t test for means, as appropriate. Statistical analysis was performed using SPSS version 17.2 (SPSS, Chicago, IL).
Among the 1099 patients who underwent resection of CLM during the study period, 230 did not receive any preoperative chemotherapy, and 809 received only 1 line of preoperative chemotherapy before surgery. The remaining 60 (5%) patients received 2 or more lines of preoperative chemotherapy and were the subjects of our study. The number of patients who underwent resection of CLM after a second-line chemotherapy increased over time (Figure 1).
The patient characteristics are summarized in Table 1. The majority (38/60 [63%]) of patients had synchronous and multiple CLM. Twelve (20%) patients had at least 5 CLM, and 25 (42%) patients had CLM measuring at least 5 cm in diameter. The 13 patients who had metachronous CLM had previously received adjuvant chemotherapy for node-positive primary tumors (5-fluorouracil and levamisole in 8 patients, 5-fluorouracil and oxaliplatin in 5 patients). The median time between the last cycle of adjuvant chemotherapy and the detection of CLM in these 13 patients was 28 months (range, 4-90 months). Fourteen (23%) patients had extrahepatic disease, including 8 patients with resectable lung metastases, 5 patients with portal node involvement, and 1 patient with pelvic local recurrence of rectal cancer.
|Age, y, mean±SD||59 ± 11|
|Rectal primary tumor||17 (28)|
|Positive lymph nodes||43 (72)|
|Synchronous liver metastases||47 (78)|
|No. of CLM, mean±SD||4 ± 3.5|
|Size of CLM at diagnosis, cm, mean±SD||5 ± 3.2|
|Bilateral liver metastases||24 (40)|
|Extrahepatic metastases||14 (23)|
|Preoperative plasma CEA level, ng/dL, median (range)||7 (1-1192)|
|Preoperative portal vein embolization||6 (10)|
|Two-stage liver resection||5 (8)|
|Type of liver resection|
|Extended hepatectomy||15 (25)|
|Major hepatectomy||25 (42)|
|Minor liver resection||20 (33)|
|Resection combined with radiofrequency ablation||15 (25)|
|Estimated blood loss, mL, median (range)||337 (25-3100)|
|Transfusion of packed red blood cells||10 (17)|
|Positive surgical margins||12 (20)|
|Major or complete pathologic response||19 (32)|
|Death within 90 days after surgery||2 (3)|
|Postoperative morbidity||20 (33)|
|Major postoperative complication||10 (17)|
Characteristics of preoperative chemotherapy are detailed in Table 2. Most patients received oxaliplatin- or irinotecan-based chemotherapy as first-line treatment. In most cases, the reason for a switch to a different chemotherapy regimen was tumor progression or advanced disease with insufficient response. In all patients, the last chemotherapy regimen before resection included irinotecan or oxaliplatin, and in the majority, it was associated with either cetuximab or bevacizumab. Most patients (39/60 [60%]) had prolonged chemotherapy (more than 6 months or 12 cycles). Twenty-two (37%) patients had objective response (complete or partial tumor response) with the last preoperative chemotherapy regimen according to the RECIST criteria, and 16 (27%) patients had an optimal morphologic response to the last preoperative chemotherapy regimen.
|First chemotherapy regimen|
|5-FU- and oxaliplatin-based||11 (18)|
|5-FU- and irinotecan-based||37 (62)|
|First chemotherapy regimen included bevacizumab||26 (43)|
|Indication for switch to second chemotherapy regimen|
|Progression during first regimen||30 (50)|
|Advanced metastases with insufficient response with or without toxicity||24 (40)|
|Intolerable toxicity||4 (6)|
|Empiric or unknown||2 (4)|
|No. of chemotherapy regimens received|
|Last chemotherapy regimen|
|5- FU + irinotecan||14 (23)|
|5- FU + irinotecan + bevacizumab||17 (28)|
|5- FU + irinotecan + cetuximab||10 (17)|
|5- FU + oxaliplatin||4 (7)|
|5- FU + oxaliplatin + bevacizumab||12 (20)|
|5- FU + oxaliplatin + cetuximab||3 (5)|
|No. of cycles of last chemotherapy regimen, mean±SD||7.5 ± 4.6|
|Total no. of cycles of preoperative chemotherapy, mean±SD||17 ± 7.8|
|Radiographic response to last chemotherapy regimen according to RECIST|
|Complete response||1 (2)|
|Partial response||21 (35)|
|Stable disease||22 (37)|
|Progressive disease||16 (27)|
|Optimal morphologic response to last chemotherapy regimen||16 (27)|
Feasibility of Resection of CLM After Second-Line Chemotherapy Regimen
Operative details and postoperative morbidity and mortality are summarized in Table 1. The majority (40/60 [67%]) of patients underwent major liver resection. Two (3%) patients died within 90 days postoperatively. One patient developed postoperative bleeding after right hepatectomy that necessitated reoperation; this patient subsequently developed irreversible hepatic failure. The other patient had a bile leakage with multiple infectious complications after segmental resection and lymphadenectomy. Ten (17%) patients developed major postoperative complications necessitating a surgical, endoscopic, or radiologic procedure. On pathologic examination, 48 (80%) patients had a complete resection, and 19 (32%) patients had major or complete pathologic response to systemic therapy. Upon histopathologic review of the nontumorous liver, 15 (25%) patients had substantial hepatic injury. Four patients had moderate or severe sinusoidal injuries, 12 patients had steatosis >30%, and 4 patients had steatohepatitis. Five (8%) patients had overlapping injuries (steatosis >30% associated with steatohepatitis in 4 patients, steatosis >30% associated with moderate sinusoidal dilation in 1 patient).
Long-Term Outcomes and Postoperative Chemotherapy
After a median follow-up time of 32 months, 1-year, 3-year, and 5-year overall survival rates were 83%, 41%, and 22%, respectively, and 1-year, 3-year, and 5-year disease-free survival were 37%, 11%, and 11%, respectively (Figure 2).
Twenty-six (43%) patients received postoperative chemotherapy. Forty-nine (81%) patients experienced recurrence, and palliative chemotherapy was restarted in all of them. Among these patients, 8 underwent surgery (resection of lung metastases in 5 patients; repeat liver resection in 2 patients; resection of local recurrence of rectal cancer in 1 patient). Median chemotherapy-free survival time was 9.2 months (95% confidence interval, 4-14 months) after surgery or after the end of postoperative chemotherapy (Figure 3). In the 34 patients who did not receive postoperative chemotherapy, the median chemotherapy-free survival time was 12 months (95% confidence interval, 6-18 months). Eight patients were alive and free of disease 12 months after resection (Figure 4).
Predictors of Survival After Resection of CLM After Second-Line Chemotherapy Regimen
Results of univariate and multivariate analysis for factors associated with survival after resection of CLM after second-line chemotherapy regimen are summarized in Table 3. Univariate analysis showed that synchronous CLM, multiple CLM, absence of radiographic response or stability of CLM by RECIST, and minor pathologic response were associated with worse survival. On multivariate analysis, only synchronous CLM and minor pathologic response were independent factors of worse survival (Figure 5). The 5-year survival rate was 10% in patients with both factors, compared with 32% in patients with neither factor (P < .001).
|Predictor||Univariate Analysis||Multivariate Analysisa|
|P||P||HR (95% CI)|
|Rectal primary tumor||.26|
|Positive lymph nodes||.98|
|Synchronous liver metastases||.02||.003||4.5 (1.7-12)|
|Maximum size of CLM ≥5 cm||.6|
|CEA level >5 ng/dL||.47|
|Progression during first chemotherapy regimen||.28|
|Toxic effects during first chemotherapy regimen||.82|
|Six or more cycles of second-line chemotherapy regimen||.64|
|Sixteen or more cycles of total chemotherapy||.16||.06|
|Radiographic response or stability after last chemotherapy regimen||.01||.4|
|Major postoperative complication||.08||.1|
|Positive surgical margins||.29|
|Minor pathologic response||.05||.006||3.4 (1.4-8.2)|
Our study shows that hepatectomy for CLM after a second-line chemotherapy regimen is feasible and associated with a modest survival benefit in patients who present with advanced CLM and have a suboptimal response to systemic therapy. Although oncologic outcomes seen in this series are not as good as previously reported after resection of CLM after first-line chemotherapy, resection of CLM after second-line chemotherapy regimen can be associated with prolonged survival and a chemotherapy-free interval and therefore represents a reasonable alternative in patients with advanced CLM.
To our knowledge, this is the largest series evaluating outcome of patients undergoing resection of CLM after second-line chemotherapy regimen. We found 1-year and 3-year overall survival rates of 83% and 41%, respectively, and 1-year and 3-year disease-free survival rates of 37% and 11%, respectively. The 1-year overall survival rate reported in this study compares favorably to the 1-year survival rates of 28%-40% reported previously in small series of patients treated with cetuximab or intra-arterial hepatic artery chemotherapy infusion as second-line therapy.11, 13, 23 This approach was associated with a median chemotherapy-free interval of 9 months after liver resection or after completion of additional chemotherapy administered after resection. We believe this represents an alternative to maintenance chemotherapy in patients with advanced disease in addition to modest possible survival benefit.
Over the past decade, the oncosurgical strategy of combining effective preoperative chemotherapy with an aggressive surgical approach has received wide acceptance for the treatment of CLM.2, 3, 19 In addition to lowering recurrence rates after resection,3 preoperative chemotherapy guides the treatment strategy, because tumor response to preoperative chemotherapy represents a powerful prognostic factor in patients undergoing resection.17, 19 Tumor progression during chemotherapy is considered a contraindication to resection of CLM by most oncologic teams.4 In patients who experience disease progression or develop severe toxic effects during first-line chemotherapy, a second-line regimen is usually introduced. Although this approach has been shown to improve survival, reported radiographic response rates after second-line chemotherapy are only 4%-28%.6-10 Our finding of a 37% objective radiographic response rate is consistent with rates reported by others, illustrating the difficulty in obtaining a tumor response with second-line chemotherapy. The extent of the disease probably contributes to the 20% positive margins after resection in our population. However, this rate remains acceptable with regard to previously reported data in patients with advanced colorectal liver metastases that can be up to 46%.24 Recently, yttrium 90 microsphere hepatic artery radioembolization has been proposed as an alternative to second-line chemotherapy as it has been shown to improve survival compared to intravenous fluorouracil in patients with unresectable CLM refractory to therapy.25 However, with yttrium 90 radioembolization, the radiographic tumor response rate rarely exceeds 10%, and the median survival rarely exceeds 1 year.
The majority of patients in our series had disease progression or a suboptimal response following first-line chemotherapy. Interestingly, patients with progression during first-line systemic therapy did not appear to have a worse outcome than patients who were switched to a second regimen because of inadequate response or toxic effects. We identified synchronous metastases and pathologic response as prognostic factors in our patients and found that the combination of these 2 factors adversely affected survival (10% 5-year overall survival rate when the 2 factors were present). The low rate of major or complete pathologic response (32%) in this study compared with data reported previously using modern first-line chemotherapy (45%-58%) probably reflects unfavorable tumor biology. Although the type of regimen, the duration of chemotherapy, and the use of a monoclonal antibody in first-line or last chemotherapy regimen was not associated with prognosis, we believe that the administration of a monoclonal antibody (e.g., bevacizumab or cetuximab) as second-line therapy when it was not given as first-line therapy may be an option, because it is associated with improved radiographic response rate or survival in patients with advanced metastatic colorectal cancer.11, 19, 26, 27 Other recommendations to optimize treatment strategy in these patients receiving second-line chemotherapy for CLM might focus on the timing of surgery that should be proposed as soon as the disease is resectable.
This study has some limitations. Study inclusion was not based on the extent of the disease, because the definition of resectability is often controversial,27 but on treatment with second-line preoperative chemotherapy regimens and subsequent liver resection with curative intention. Not all patients received the same chemotherapy regimens before surgery, and we did not analyze patients with advanced CLM who received second-line chemotherapy regimen and did not undergo resection. It was a retrospective review of data acquired prospectively and thus suffers from the limitations of retrospective studies, including potential selection bias. However, we believe that a phase 2 or 3 trial would be impractical to address this question. The comparison with a nonresected group was not performed, because resectable CLM remain a surgical indication, and progression during chemotherapy a relative contraindication to resection. It was not possible to select a population of patients receiving chemotherapy only with comparable disease extent.
In conclusion, this study showed that resection of CLM following second-line chemotherapy regimen is feasible and may be associated with a modest survival benefit. Given the lack of highly effective treatment options for patients with liver only metastatic disease with suboptimal response to initial chemotherapy, we consider liver resection for selected patients following second-line therapy an appropriate alternative. Not all patients receiving second-line chemotherapy for advanced CLM can benefit from resection, however; in the future, refinements in the assessment of tumor response should help to select surgical candidates in this challenging therapeutic setting.17, 28, 29
Supported in part by the National Institutes of Health through MD Anderson Cancer Center Support Grant CA016672.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 16New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst. 2000; 92: 205-216., , , et al.
- 25Phase III trial comparing protracted intravenous fluorouracil infusion alone or with yttrium-90 resin microspheres radioembolization for liver-limited metastatic colorectal cancer refractory to standard chemotherapy. J Clin Oncol. 2010; 28: 3687-3694., , , et al.