It is known that 60–70% of recurrent colorectal cancer involves the liver and that the liver is the only involved organ in 35% of cases. Colorectal liver metastases used to be thought of as systemic disease that involves many organs and systems until recently when local liver therapy in the form of liver resection has been reported to results in 5-year survival of 27–39% and even longer survival and cure [
Over the last 20 years liver resection for colorectal metastases has seen many refinements; the improvement in anaesthesia and postoperative care have reduced the morbidity and mortality with subsequent more aggressive surgical approach. Strategies that have widened the indications for liver resection includes portal vein embolisation, staged liver resection, neoadjuvant chemotherapy, ablative procedures, and locoregional chemotherapy [
Many studies have developed and validated scoring systems to predict prognosis and recurrence of colorectal metastases based on clinical and pathological data of large number of hepatic resections for colorectal metastases. These scores are based on variety of factors that include stage of the primary disease, time interval between diagnosis of the primary lesion and occurrence of liver metastases, level of CEA immediately prior to liver resection, size and number of resected liver lesions, surgical resection margin, blood transfusion, and bilateral distribution of liver disease [
Various factors were found to have influence on disease recurrence and overall survival; resection margins and lymph nodes involvement are common predictors of recurrence. Other controversial factors are number and size of lesions, blood transfusion, and disease-free interval. Other factors are extrahepatic disease and portal nodes metastases [
This systematic review is conducted to evaluate the risk factors influence of overall long-term survival following hepatic resection for colorectal metastases.
Electronic search was performed and relevant reports were identified using electronic databases (MEDLINE 1950–2010 and EMBASE 1980–2010), the search was restricted to human adults and English language literature. The search terms used were colorectal neoplasm, overall survival, and disease-free survival. All the references used in the published original and review studies were searched to identify more studies.
To be included, studies had to meet the following criteria. Design: prospective or retrospective cohort studies. Population: patients with liver metastases from colorectal cancer who had liver resection as a curative treatment. Exposure: surgical liver resection for metastases whether anatomic resection or segmental nonanatomic resection regardless of whether they had or did not have adjuvant chemotherapy. That included studies that evaluated patients’ survival following re-resection following primary liver resection. Outcome: overall ten-year survival following liver resection for colorectal liver metastases.
Duplicate publications were excluded and wherever publications that evaluated the same population group were encountered, the report with the most relevant and comprehensive data was selected.
Articles that met all the inclusion criteria were retrieved as full text articles. Two independent reviewers using standard data collection form extracted all relevant data from the full text articles. Inconsistencies were resolved by discussion to reach a reasonable consensus. Whenever missing data were encountered, the authors were contacted to request the data required to be included in the meta-analysis. One study in non-English language was encountered and was excluded.
Methodological quality of the studies was evaluated independently by two reviewers using the Newcastle-Ottawa Scale [
Odds ratio (OR) of overall survival was used as the primary effect estimate in this meta-analysis. From the eligible studies that met the inclusion criteria, estimates of the OR and its associated 95% confidence interval (CI) were calculated using the Review Manager software (Version 5 for Windows, Copenhagen, Denmark; The Nordic Cochrane Centre, The Cochrane Collaboration, 2008). Data that could not be extracted directly were reconstructed indirectly by two reviewers when required.
Prespecified factors that was thought to affect the overall survival after liver resection for colorectal metastases were analysed. Those included resection margin, tumour size, number of metastases, bilateral versus unilateral disease, T stage of the primary, lymph nodes positive primary versus lymph nodes negative, disease-free interval, CEA level, and blood transfusion. Sensitivity analysis on the included studies was conducted on the Review Manager.
Heterogeneity between the included studies was appraised using the
The initial search revealed 164 titles, abstracts for those articles (Figure
Characteristics of the included studies.
Study | Study design | Total number of patients | 5-year survival | 10-year survival |
---|---|---|---|---|
Giuliante et al. 2009 [ | Retrospective | 251 | 38.9% | 24.2% |
Hamady et al. 2006 [ | Retrospective | 293 | 44% | 36% |
Jamison et al. 1997 [ | Retrospective | 280 | 27% | 20% |
Minagawa et al. 2000 [ | Retrospective | 235 | 38% | 26% |
Scheele et al. 1995 [ | Retrospective | 469 | 39.3% | 23.6% |
Shimizu et al. 2007 [ | Retrospective | 164 | 51.8% | 36.6% |
Tomlinson et al. 2007 [ | Retrospective | 612 | 21% | 17% |
Wanebo et al. 1996 [ | Retrospective | 74 | 24% | 12% |
Quality assessment of included studies (Newcastle-Ottawa Scale).
Study | Selection | Comparability | Outcome |
---|---|---|---|
Giuliante et al. 2009 [ | 4 | 1 | 3 |
Hamady et al. 2006 [ | 4 | 2 | 3 |
Jamison et al. 1997 [ | 4 | 2 | 3 |
Minagawa et al. 2000 [ | 4 | 1 | 3 |
Scheele et al. 1995 [ | 4 | 2 | 3 |
Shimizu et al. 2007 [ | 4 | 1 | 3 |
Tomlinson et al. 2007 [ | 4 | 2 | 3 |
Wanebo et al. 1996 [ | 4 | 2 | 3 |
Flow diagram of study selection process.
There was no statistical evidence of publication bias among the included studies based on the funnel plot used in Review Manager.
The eleven studies reviewed 3442 patients who had liver resection for colorectal metastases, the eight included studies had a total number of 2387 patients, all studies reported five- and ten-year survival [
In the study of Tomlinson et al. [
Factors that influence disease-free survival and may predict that has been reported only in the study by Minagawa et al. [
Four studies reported adequate data to determine the relationship between positive and negative resection margin and long-term survival. The average overall survival for positive margin was 29%, which is significantly better compared with 20% of negative resection margin (
Positive and negative margins.
Four studies reported the influence of wider negative margin of more than 1cm compared with negative margin of 0–10 mm [
Resection margin less than 1 cm or more than 1 cm.
Six studies reported the effect of tumour size on overall survival [
Tumour size.
Four studies reported data on the effect of time interval between the diagnoses of the primary colorectal cancer and the occurrence of liver metastases on the overall patients’ survival. Pooled estimate of the survival time after liver resection for colorectal metastases showed no significant prognostic relationship (OR = 1.22; 95% CI: 0.75–1.99;
Disease-free interval.
Six studies reported data to determine the relationship between lymph nodes metastasis status of the primary colorectal cancer and survival after liver resection for colorectal metastases [
Nodal disease of the primary cancer.
Three studies reported data for determination of the relationship between types of liver resection (segmental or anatomic) [
Type of resection.
Four studies provided data for determining the relationship of CEA levels, prior to resection of colorectal liver metastases, and the overall patients’ survival [
CEA level.
Five studies provided data to determine the effect of bilateral distribution of resected colorectal metastases on the ten-year survival [
Distribution of liver lesions.
Seven studies reported data to determine the effect of the number of the resected liver lesions on long-term survival [
Number of liver lesions.
Six studies provided data to determine the relationship of timing of liver metastases; whether it was found at the time of diagnosis of the primary colorectal cancer or afterward [
Synchronous versus metachronous metastases.
Four studies provided data for determining the relationship of blood transfusion, after liver resection, and long-term survival [
Blood transfusion.
Three studies provided data that compared the difference in survival between patients who had a single lesion resected and those who had more than a single lesion [
Single versus multiple lesions.
Three studies provided data that compared survival of patients who had satellite lesions along with larger lesion or lesions compared with those who had no satellite lesions [
Satellite lesions.
This systematic review and meta-analysis showed that factors that affect long-term survival following hepatic resection, for colorectal cancer metastases, include clear resection margins, advanced primary colorectal cancer with nodal metastases, CEA levels, distribution of liver lesions, timing of diagnosis of liver metastases (synchronous or metachronous), quantity of blood transfusion, single lesion compared with multiple lesions, and presence or absence of satellite nodules close to the main lesion. Patients who had clear resection margins had significantly better long-term survival than those with positive resection margins. Patients who had early stage colorectal cancer with no lymph nodes metastases had better survival than those with lymph nodes metastases. It also showed that metachronous presentation of liver metastases is a good prognostic factor compared with synchronous presentation. A single liver lesion particularly in the absence of satellite nodules has better outcome than multiple lesions with or without satellite nodules; however, when comparison was made between patients with more than four nodules and those with four or less lesions, there was no difference in overall long-term survival. CEA levels less than 50 ng/mL, unilateral liver disease, and two units or less of perioperative blood transfusion were found to be favourable prognostic factors. There was no single prognostic factor of sufficient power to predict long-term survival and cure.
Other factors that were analysed in this meta-analysis and found not to have significant influence on long-term survival included a the width of resection margin, it was found that if the resection margin is clear there is no survival benefit from a wider resection margin more than 1 cm. Whether the resected largest lesion was more or less than 5 cm did not have effect on survival. The interval between the diagnosis of the primary and liver recurrence less or more than 12 months did not seem to affect long term survival. Whether the lesion removed in anatomic resection technique or segmental resection, number of resected lesions, and synchronous versus metachronous metastases, all had no effect on long-term survival.
During the past two decades, liver resection for colorectal liver metastases has been increasing the standard of care whenever the disease is limited to the liver and is technically possible by leaving adequate liver remnants. There is overwhelming evidence to support the survival benefit with reports of actuarial 5-year survival of 25–40% compared with patients who are treated only with chemotherapy who rarely survive up to five years [
Tomlinson et al. had investigated risk factors for 10-year survival and redesigned the original score devised by the same investigators. The original score had five components with one point for each component that includes tumour number more than four, size more than 5 cm, CEA level more than 200 ng/mL, and disease-free interval less than 12 months and positive resection margin with one point for each. After analysing 10-year survival, patients fell in two groups; the low risk group (0–2 points) who had 10-year survival of 21% and high-risk group (3–5 points) with 10-year survival of 10% (
It was not possible in this review to analyse the effect of adjuvant chemotherapy on survival after hepatic resection for colorectal metastases. However, most of the study patients had been treated in the era when the standard chemotherapy was 5-flourouracil, which has limited effect, compared with the modern chemotherapeutic agents like irinotecan, oxaliplatin, and bevacizumab [
The improvement of outcome in liver resection has been attributed to a combination of factors such as aggressive surgical treatment, improved chemotherapy, and improvement of preoperative imaging and patient selection [
Repeat hepatectomy for recurrent hepatic disease following initial liver resection is being increasingly used, also staged liver resection and resection of isolated extrahepatic disease is being more and more utilised with encouraging results that lead to improvement in survival [
Improvement of preoperative staging by liberal use of helical CT scans and MRI and the introduction of PET scan and PET CT have allowed for better patients selection by early detection of extrahepatic and bilobar disease [
Many authors have reported using radiofrequency ablation in conjunction with surgery for nonresectable liver metastases either intraoperative or postoperative with variable results [
The emergence of new chemotherapeutic agents such as oxaliplatin, irinotecan, and bevacizumab has increased the treatment option available for clinicians to deal with metastatic colorectal disease. These new agents have been used in conjunction with liver resection either as neoadjuvant or adjuvant manner in many studies but the effect of this approach on survival has not been test in a randomised controlled trial [
In conclusion this review defines 10-year survival and cure to be between 12% and 28%; we described the factors that affect survival in this meta-analysis. There is no single factor that was of sufficient power to rule out cure with the possible exception of positive resection margin. This leads to the fact that patients’ selection for resection with disease limited to the liver or liver disease with resectable extrahepatic metastases remains a matter of trial and error particularly for patients who have marginal suitability for resection. This review also indicates that we need newer prognostic factors perhaps based on tumour biology that may discriminate between curable and noncurable metastatic colorectal cancer. An important limitation of this study that reflects the quality of the available data is the fact that raw data was not available to all the studies’ patients and the presence of heterogeneity among the included studies and the fact that those studies are generally retrospective reviews. Also patients who had 10-year survival are likely to have been treated prior to the era of PET scan routine use and likely have been treated with old and less effective chemotherapeutic agents.