Randomized clinical trial of prehabilitation before planned liver resection.

BACKGROUND: Patients with low fitness as assessed by cardiopulmonary exercise testing (CPET) have higher mortality and morbidity after surgery. Preoperative exercise intervention, or prehabilitation, has been suggested as a method to improve CPET values and outcomes. This trial sought to assess the capacity of a 4-week supervised exercise programme to improve fitness before liver resection for colorectal liver metastasis. METHODS: This was a randomized clinical trial assessing the effect of a 4-week (12 sessions) high-intensity cycle, interval training programme in patients undergoing elective liver resection for colorectal liver metastases. The primary endpoint was oxygen uptake at the anaerobic threshold. Secondary endpoints included other CPET values and preoperative quality of life (QoL) assessed using the SF-36®. RESULTS: Thirty-eight patients were randomized (20 to prehabilitation, 18 to standard care), and 35 (25 men and 10 women) completed both preoperative assessments and were analysed. The median age was 62 (i.q.r. 54-69) years, and there were no differences in baseline characteristics between the two groups. Prehabilitation led to improvements in preoperative oxygen uptake at anaerobic threshold (+1·5 (95 per cent c.i. 0·2 to 2·9) ml per kg per min) and peak exercise (+2·0 (0·0 to 4·0) ml per kg per min). The oxygen pulse (oxygen uptake per heart beat) at the anaerobic threshold improved (+0·9 (0·0 to 1·8) ml/beat), and a higher peak work rate (+13 (4 to 22) W) was achieved. This was associated with improved preoperative QoL, with the overall SF-36® score increasing by 11 (95 per cent c.i. 1 to 21) (P = 0·028) and the overall SF-36® mental health score by 11 (1 to 22) (P = 0·037). CONCLUSION: A 4-week prehabilitation programme can deliver improvements in CPET scores and QoL before liver resection. This may impact on perioperative outcome. REGISTRATION NUMBER: NCT01523353 (https://clinicaltrials.gov).

Evaluation of a novel tissue stabilization gel to facilitate clinical sampling for translational research in surgical trials.

BACKGROUND: The aim was to establish the feasibility of using a tissue stabilization gel (Allprotect™) as an alternative to liquid nitrogen to facilitate collection of clinical samples for translational research. METHODS: Tumour samples from patients undergoing surgery for primary or metastatic colorectal cancer were either snap-frozen in liquid nitrogen or stored in Allprotect™ under a number of different conditions. Sample integrity was compared across different storage conditions by assessing biomolecule stability and function. DNA quality was assessed spectrophotometrically and by KRas genotyping by pyrosequencing. Total RNA retrieval was determined by nanodrop indices/RNA integrity numbers, and quality assessed by reverse transcription-PCR for two representative genes (high-mobility group box 1, HMGB1; carboxylesterase 1, CES1) and two microRNAs (miR122 and let7d). Western blot analysis of HMGB1 and CES1 was used to confirm protein expression, and the metabolic conversion of irinotecan to its active metabolite, SN-38, was used to assess function. RESULTS: Under short-term storage conditions (up to 1 week) there was no apparent difference in quality between samples stored in Allprotect™ and those snap-frozen in liquid nitrogen. Some RNA degradation became apparent in tissue archived in Allprotect™ after 1 week, and protein degradation after 2 weeks. CONCLUSION: In hospitals that do not have access to liquid nitrogen and -80°C freezers, Allprotect™ provides a suitable alternative for the acquisition and stabilization of clinical samples. Storage proved satisfactory for up to 1 week, allowing transfer of samples without the need for specialized facilities. Surgical relevance Access to clinical material is a fundamental component of translational research that requires significant infrastructure (research personnel, liquid nitrogen, specialized storage facilities). The aim was to evaluate a new-to-market tissue stabilization gel (Allprotect™), which offers a simple solution to tissue preservation without the need for complex infrastructure. Allprotect™ offers comparable DNA, RNA and protein stabilization to tissue snap-frozen in liquid nitrogen for up to 1 week. Degradation of biomolecules beyond this highlights its role as a short-term tissue preservative. Allprotect™ has the potential to increase surgeon participation in translational research and surgical trials requiring tissue collection.

The Liverpool uveal melanoma liver metastases pathway: outcome following liver resection.

AIM: To determine the outcome of patients that underwent liver resection for metastases from uveal melanoma. METHODS: Over a 9-year period, patients referred with uveal melanoma metastases were included. Following treatment of primary uveal melanoma, high-risk patients were offered to be enrolled into a 6-monthly non-contrast liver magnetic resonance imaging (MRI) surveillance. Following detection of liver metastases, patients were staged with a contrast-enhanced (Primovist(®)) liver MRI, computer tomography (CT) of the thorax and staging laparoscopy. RESULTS: 155 patients were referred with uveal melanoma liver metastases, of which 17 (11.0%) patients had liver resection and one patient was treated with percutaneous radio-frequency ablation. The majority of patients undergoing liver resection were treated with multiple metastectomies (n = 8) and three patients had major liver resections. The overall median survival for patients treated with surgery/ablation was 27 (14-90) months, and this was significantly better compared to patients treated palliatively [median = 8(1-30) months, P < 0.001]. Following surgery, 11 patients had recurrent disease [median = 13(6-36) months]. Patients who had undergone a major liver resection had a significantly poorer disease-free survival (P = 0.037). CONCLUSIONS: Patients who can undergo surgical resection for metastatic uveal melanoma have a more favorable survival compared to those who do not.

Disappearing colorectal liver metastases in the era of state-of-the-art triple-modality diagnostic imaging.

INTRODUCTION: Systemic therapy can result in disappearance of colorectal liver metastases in up to 40% of patients. This might be an overestimation caused by suboptimal imaging modalities. The aim of this study was to investigate the use of imaging modalities and the incidence, management and outcome of patients with disappearing liver metastases (DLMs). METHODS: This was a retrospective study of consecutive patients treated for colorectal liver metastases at a high volume hepatobiliary centre between January 2013 and January 2015 after receiving induction or neoadjuvant systemic therapy. Main outcomes were use of imaging modalities, incidence, management and longterm outcome of patients with DLMs. RESULTS: Of 158 patients included, 32 (20%) had 110 DLMs. Most patients (88%) had initial diagnostic imaging with contrast enhanced-CT, primovist-MR and FDG-PET and 94% of patients with DLMs were restaged using primovist-MR. Patients with DLMs had significantly smaller metastases and the median initial size of DLMs was 10 mm (range 5-61). In the per lesion analysis, recurrence after "watch & wait" for DLMs occurred in 36%, while in 19 of 20 resected DLMs no viable tumour cells were found. Median overall (51 vs. 28 months, p < 0.05) and progression free survival (10 vs. 3 months, p = 0.003) were significantly longer for patients with DLMs. CONCLUSION: Even state-of-the-art imaging and restaging cannot solve problems associated with DLMs. Regrowth of these lesions occurs in approximately a third of the lesions. Patients with DLMs have better survival.

Systematic review and meta-analysis of follow-up after hepatectomy for colorectal liver metastases.

BACKGROUND: The evidence surrounding optimal follow-up after liver resection for colorectal metastases remains unclear. A significant proportion of recurrences occur in the early postoperative period, and some groups advocate more intensive review at this time. METHODS: A systematic review of literature published between January 2003 and May 2010 was performed. Studies that described potentially curative primary resection of colorectal liver metastases that involved a defined follow-up protocol and long-term survival data were included. For meta-analysis, studies were grouped into intensive (more frequent review in the first 5 years after resection) and uniform (same throughout) follow-up. RESULTS: Thirty-five studies were identified that met the inclusion criteria, involving 7330 patients. Only five specifically addressed follow-up. Patients undergoing intensive early follow-up had a median survival of 39·8 (95 per cent confidence interval 34·3 to 45·3) months with a 5-year overall survival rate of 41·9 (34·4 to 49·4) per cent. Patients undergoing routine follow-up had a median survival of 40·2 (33·4 to 47·0) months, with a 5-year overall survival rate of 38·4 (32·6 to 44·3) months. CONCLUSION: Evidence regarding follow-up after liver resection is poor. Meta-analysis failed to identify a survival advantage for intensive early follow-up.

Cholecystectomy-related bile duct and vasculobiliary injuries.

BACKGROUND: Combined vasculobiliary injury is a serious complication of cholecystectomy. This study examined medium- to long-term outcomes after such injury. METHODS: Patients referred to this institution with Strasberg type E bile duct injuries were identified from a prospectively maintained database (1990-2010). Long-term outcomes were evaluated by chart review. RESULTS: Sixty-three patients were referred with bile duct injury alone (45 patients) or vasculobiliary injury (18). Thirty patients (48 per cent) had septic complications before transfer. Twenty-six patients (41 per cent) had long-term biliary complications over a median follow-up of 96 (range 12-245) months. Nine patients (3 with bile duct injury, 6 with vasculobiliary injury) required further interventions after a median of 22 (8-38) months; five required biliary surgical revision and four percutaneous dilatation of biliary strictures. Vasculobiliary injury and injury-related sepsis were independent risk factors for treatment failure: hazard ratio 7·79 (95 per cent confidence interval 2·80 to 21·70; P < 0·001) and 4·82 (1·69 to 13·68; P = 0·003) respectively. CONCLUSION: Outcome following bile duct injury repair was worse in patients with concomitant vasculobiliary injury and/or sepsis.

Effect of specialist decision-making on treatment strategies for colorectal liver metastases.

BACKGROUND: One hundred and ten patients were treated with palliative chemotherapy, of whom 53 had liver-only disease and had not been reviewed by a specialist liver surgeon. One scan was excluded as all reviewers felt it to be of insufficient quality to assess. Improved surgical technique and better chemotherapeutic manipulation of metastatic disease has increased the number of patients eligible for potentially curative resection of colorectal liver metastases. The rapid evolution in this field suggests that non-specialist decision-making may lead to inappropriate management. This study aimed to assess the management of colorectal liver metastases by non-liver surgeons. METHODS: All patients who underwent chemotherapy with palliative intent for metastatic colorectal cancer at a regional oncology centre between 1 January and 31 December 2009 were identified from a prospectively maintained local database. Six resectional liver surgeons blinded to patient management and outcome reviewed pretreatment imaging and assigned each scan a score based on their own management choice. A consensus decision was reached on the appropriateness of palliative chemotherapy. RESULTS: One hundred and ten patients were treated with palliative chemotherapy, of whom 53 had liver-only disease and had not been reviewed by a specialist liver surgeon. One scan was excluded as all reviewers felt it to be of insufficient quality to assess [corrected]. Tumours in 33 patients (63 per cent) were considered potentially resectable, with a high level of interobserver agreement (κ = 0 · 577). When individual approach to management was considered, interobserver agreement was less marked (κ = 0 · 378). CONCLUSION: Management of patients with colorectal liver metastases without the involvement of a specialist liver multidisciplinary team can lead to patients being denied potentially curative treatments. Management of these patients must involve a specialist liver surgeon to ensure appropriate management.

Outcomes of intensive surveillance after resection of hepatic colorectal metastases.

BACKGROUND: The impact of computed tomography (CT)-based follow-up for the detection of resectable disease recurrence following surgery for colorectal liver metastases (CRLM) was evaluated. METHODS: Some 705 patients undergoing resection of CRLM between January 1993 and March 2007 were included. Surveillance comprised 3-monthly CT (thorax, abdomen and pelvis) in the first 2 years after surgery, 6 monthly for 3 years and annually from years 6 to 10. Survival differences following recurrence between patients managed surgically and palliatively were determined, and the cost was calculated. RESULTS: Five-year disease-free and overall survival rates were 28.3 and 32.3 per cent respectively. Of 402 patients who developed recurrence within 2 years, 88 were treated with liver resection alone and 36 with lung and/or liver resection. Their 5-year overall survival rates were 31 and 30 per cent respectively, compared with 3.9 per cent in 278 patients managed palliatively (P < 0.001). For each 3-month interval during the first year of follow-up, patients with recurrence treated surgically had better overall survival than those treated palliatively. The cost of surveillance that identified 124 patients amenable to further resection was 12,338 pounds per operated recurrence. Assuming that patients with recurrence gained 5 years' survival, the mean survival gain was 4.28 years per resection and the cost per life-year gained was 2883 pounds. CONCLUSION: Intensive 3-monthly CT surveillance after liver resection for CRLM detects recurrence that is amenable to further resection in a considerable number of patients. These patients have significantly better survival with a reasonable cost per life-year gained.

Multicentre validation of a clinical prognostic score integrating the systemic inflammatory response to the host for patients treated with curative-intent for colorectal liver metastases: The Liverpool score.

BACKGROUND: This study aimed to create a new prognostic score integrating the systemic inflammatory response to predict survival in patients treated with curative intent for colorectal liver metastases (CLM). METHODS: We identified independent prognostic factors in patients who underwent liver surgery for CLM in a tertiary centre in the United Kingdom (UK) between 2010 and 2015. A pre- and a postoperative score (Liverpool score) were created by combining these factors to stratify patients into different risk groups. These new scores were validated in an international cohort of 219 patients from China and France. RESULTS: Multivariate cox regression analysis of the 364 patients of the UK cohort identified 6 preoperative and 1 postoperative prognostic factors for overall survival (OS): American society of anaesthesiologists (ASA) score, location and node status of the primary tumour, number and size of CLM, neutrophil-to-lymphocyte ratio (NLR) and resection margin. Both pre- and postoperative scores can be calculated with an online calculator at https://jscalc.io/calc/PXatrmjfrEFpYy2t. Using the pre-operative model on the UK cohort, median OS was 61.22 (50.23, not reached) months in the low-risk group (n = 162) and 30.36 (23.68, 35.95) months in the high-risk group (n = 162, p < 0.0001). The same difference was observed in the validation cohort. The Liverpool score outperformed previously published scoring system with a c-index of 0.619 pre-operatively and of 0.637 post-operatively. CONCLUSION: We developed a new prognostic score based on clinicopathologic characteristics including the site of the primary tumour location and on measurement of the systemic inflammatory response which could help to tailor patients' management.

Defining enhanced recovery after resection of peri-hilar cholangiocarcinoma.

INTRODUCTION: Enhanced recovery after surgery (ERAS) for peri-hilar cholangiocarcinoma (pCCA) has not been described in the literature. This study examined patients undergoing pCCA resection within a standard post hepatectomy ERAS pathway to define achievable targets suitable for these patients. METHODS: Patients undergoing pCCA resection at University Hospital Aintree (January 2009-October 2017) were identified. Achievement of key ERAS outcomes was assessed. Patients were stratified on incidence of major complications and pre-operative cardiopulmonary exercise testing. Chi Square and Mann Whitney analyses were undertaken as appropriate. Achievable ERAS targets were derived from patients who did not develop a major complication. RESULTS: 46 patients underwent resection with enhanced recovery. Median age 65 (24 male: 22 female). Key ERAS outcomes in patients who did not experience major complications are described as medians (interquartile range): length of stay 8 days (6-13), duration critical care 2 days (2-4), inotropes 6 h (0-24), epidural 3 days (3-4), early mobilization day 1 (1-2), full mobilization day 3 (3-4), urinary catheter removal day 4 (3-5), NGT removal day 1 (1-2) and restoration oral nutrition day 2 (2-4). Patients deemed high risk pre-operatively or those who developed major complications post-operatively required significantly longer critical care (p = 0.008 and p = 0.002 respectively). Other ERAS targets remained achievable in similar timeframes. CONCLUSIONS: ERAS for pCCA is achievable. Applicable ERAS standards are defined which take into account minor complications. High risk patients and those with major complications can be appropriately managed in an ERAS pathway, though there is increased need for critical care support.

European Society of Surgical Oncology's strategy for clinical research: Paving the way for a culture of research in cancer surgery.

As part of its mission to promote the best surgical care for cancer patients, the European Society of Surgical Oncology (ESSO) has been developing multiple programmes for clinical research along with its educational portfolio. This position paper describes the different research activities of the Society over the past decade and an action plan for the upcoming five years to lead innovative and high quality surgical oncology research. ESSO proposes to consider pragmatic research methodologies as a complement to randomised clinical trials (RCT), advocates for increased funding and operational support in conducting research and aims to enable young surgeons to be active in research and establish partnerships for translational research activities.

Intermittent Pringle manoeuvre is not associated with adverse long-term prognosis after resection for colorectal liver metastases.

BACKGROUND: Intermittent clamping of the porta hepatis, or the intermittent Pringle manoeuvre (IPM), is often used to control inflow during parenchymal liver transection. The aim of this study was to determine whether IPM is associated with an adverse long-term outcome after liver resection for colorectal liver metastasis (CRLM). METHODS: All patients undergoing resection for CRLM in 1993-2006, for whom data on IPM were recorded, were included in the study. A total of 563 patients was available for analysis. RESULTS: IPM was performed in 289 (51.3 per cent) of the patients. The duration of IPM ranged from 2 to 104 (median 22) min. There were no differences in clinicopathological features or postoperative morbidity between patients who had an IPM and those who did not. The median survival of patients undergoing IPM was 55.7 months compared with 48.9 months in those not having an IPM (P = 0.406). There was no difference in median disease-free survival between the two groups (22.1 versus 19.9 months respectively; P = 0.199). CONCLUSION: IPM is not associated with an adverse long-term prognosis in patients undergoing liver resection for CRLM.

Exome sequencing of synchronously resected primary colorectal tumours and colorectal liver metastases to inform oncosurgical management.

BACKGROUND: Next generation sequencing technology has facilitated mapping of the colorectal cancer genotype and furthered our understanding of metastogenesis. The aim of this study was to investigate for conserved and different mutations in the exomes of synchronously resected primary colorectal tumour and liver metastases. This information could potentially be utilised to guide the treatment of advanced disease with the help of biological information from the primary tumour. METHODS: We performed exome sequencing of synchronously resected primary colorectal cancer and colorectal liver metastases as well as normal colonic mucosa and liver parenchyma, from four patients who had received neo-adjuvant chemotherapy, at a depth of 50X using the Ion Proton platform. Raw data was mapped to the reference genome prior to variant calling, annotation and downstream analysis. RESULTS: Exome sequencing identified 585 non-synonymous missense single nucleotide variants (SNVs), of which 215 (36.8%) were unique to the primary tumour, 226 (38.6%) unique to the metastasis and 81 (13.8%) present in patient matched pairs. SNVs identified in the ErbB pathway appear to be concordant between primary and metastatic tumours. CONCLUSION: Only 13.8% of the metastatic exome can be predicted by the genotype of the primary tumour. We have demonstrated concordance of a number of SNVs in the ErbB pathway, which may inform selection of therapeutic agents in advanced colorectal cancer.

A critical appraisal of the role of neoadjuvant chemotherapy for colorectal liver metastases: a case-controlled study.

BACKGROUND: The aim of this study was to analyze the outcome of patients that received neoadjuvant chemotherapy prior to resection for colorectal liver metastases (CRLM) and compare them with a matched cohort of patients that underwent resection followed by adjuvant chemotherapy. METHODS: 687 patients have undergone curative resection between January 1993 and January 2006. In this period, 84 patients received neo-adjuvant chemotherapy and 71 of this group went on to resection. A control group was chosen, matched with these patients, made up of patients who underwent resection followed by adjuvant chemotherapy. RESULTS: There was no difference in clinico-pathological features between the neoadjuvant and the control group. However patients in the control group had more-extended resections and longer hospital stays than those in the neoadjuvant group (p = 0.015). Patients in the control group had an increased incidence of early recurrences (p < 0.001). Despite this, there was no significant difference in either the cancer-specific or the disease-free survival between the two groups of patients. CONCLUSION: Neoadjuvant chemotherapy has a role in the management of patients with disease that is considered initially unresectable as a down-sizing technique. In patients with resectable disease, the test-of-time approach that neoadjuvant therapy offers is yet to be proven.

Steatosis predicts postoperative morbidity following hepatic resection for colorectal metastasis.

BACKGROUND: Few studies are available on the effect of steatosis on perioperative outcome following hepatic resection for colorectal liver metastasis (CRLM). METHODS: Patients undergoing resection for CRLM from January 2000 to September 2005 were identified from a hepatobiliary database. Data analysed included laboratory measurements, extent of hepatic resection, blood transfusion requirements and steatosis. RESULTS: There were 386 patients with a median age of 66 (range 32-87) years, of whom 201 had at least one co-morbid condition and 194 had an American Society of Anesthesiologists grade of I. Anatomical resection was performed in 279 patients and non-anatomical resection in 107; 165 had additional procedures. Steatosis in 194 patients was classified as mild in 122, moderate in 60 and severe in 12. The overall morbidity rate was 36 per cent (139 patients) and the mortality rate was 1.8 per cent (seven patients). Admission to the intensive care unit, morbidity, infective complications and biochemical profile changes were associated with greater severity of steatosis. Independent predictors of morbidity were steatosis, extent of hepatic resection and blood transfusion. CONCLUSION: Steatosis is associated with increased morbidity following hepatic resection. Other predictors of outcome were extent of hepatic resection and blood transfusion.

Effect of type of resection on outcome of hepatic resection for colorectal metastases.

BACKGROUND: Non-anatomical liver resections have become more common in the management of colorectal liver metastases. This study examined survival and patterns of recurrence following surgery for colorectal liver metastases. METHODS: Data were collected prospectively on all patients who had hepatic surgery for colorectal liver metastases at St James' University Hospital, Leeds between 1993 and May 2003, and analysed with respect to type of resection. RESULTS: A total of 96 patients underwent non-anatomical liver resection, 280 patients had an anatomical resection, and 108 patients had a combined procedure. There was no significant difference in overall survival between the anatomical and non-anatomical groups (hazard ratio 1.14 (95 per cent confidence interval 0.60 to 2.17); P = 0.691). Intrahepatic recurrence was significantly less common in the anatomical group, whereas morbidity and mortality rates were lower in the non-anatomical group. On multivariable analysis, multiple metastases and poorer primary T stage predicted poorer overall survival and a positive resection margin predicted poorer disease-free survival. CONCLUSION: Non-anatomical resection can be performed with lower rates of surgical morbidity and mortality than anatomical resection, and does not disadvantage the patient in terms of overall survival.

Prognostic influence of multiple hepatic metastases from colorectal cancer.

AIMS: The aim of this study was to report the results of surgery for multiple colorectal liver metastases on patient outcome. METHODS: This was a review of 484 consecutive patients who underwent liver resection for colorectal liver metastases between 1993 and 2003. The cohort was divided into 2 groups, those with 1-3 metastases and those with "multiple" metastases, namely 4 or more lesions. The later group was subdivided into those with less than 8 ("several") or 8 or more ("numerous") separate lesions. MAIN OUTCOME MEASURES: the post-operative hospital stay was calculated and morbidity and mortality were assessed. RESULTS: On multivariate analysis the presence of multiple metastases was the only predictor for both poorer overall survival (p=0.007) and disease-free survival (p=0.031). However, when patients with multiple metastases are analysed in detail this survival disadvantage appears to be only present in patients with numerous (8 or more) lesions. CONCLUSION: Although patients with multiple metastases appear to have a poorer outcome, significant number of patients with multiple metastases survive to 5 years or more and should not be denied surgery. Patients with numerous (8 or more) metastases showed a poorer survival disadvantage. These patients need alternative treatment speculatives.

Predictors of early disease recurrence following hepatic resection for colorectal cancer metastasis.

BACKGROUND: With the broadening indications of hepatic resection for colorectal liver metastases (CRLM), the exact group of patients who would benefit from surgery is still debatable. The aim of this study was to identify predictors for early recurrence, defined as recurrence within 6 months of CRLM resection, in order to identify those patients who may require further pre-operative radiological staging of the disease prior to surgery. METHODS: Prospectively collected dataset of patients undergoing curative resection for CRLM during the 10-year period (January 1993-May 2003) were analyzed. Patients who received neo-adjuvant chemotherapy and patients who underwent repeat hepatic resections whose primary resection was not performed during the study period were excluded. RESULTS: Four hundred and thirty patients (89%) were included in the analysis. Eighty-six (20%) patients developed early recurrence. Early recurrence was associated with poorer outcome when compared to late recurrences (p<0.001). The predictor of early recurrence on multivariable analysis was the presence of eight or more metastases (p=0.036). CONCLUSION: We have identified a group of patients with multiple metastases who recur early following resection of CRLM. We suggest that these patients should be considered for additional pre-operative radiological workup in the form of PET scanning to identify those patients who would be deemed suitable for resection.

Outcome following repeat liver resection for colorectal liver metastases.

AIM: Our aim was to determine independent predictors of survival after second liver resection and to confirm whether the type of first resection influences survival after repeat resection. METHODS: Fifty-four patients who underwent a second liver resection for colorectal liver metastases were analyzed. To find independent predictors of survival, possible prognostic factors regarding the primary tumor, and the first and second resections were used in the Cox regression analysis. RESULTS: There were three postoperative deaths within 90 days of surgery. The 3- and 5-year overall survival rates were 53% and 46%, respectively. The size of the tumor (>50mm) (p=0.005), serum carcinoembryonic antigen level (>30microg/L) (p=0.002), and the presence of a positive surgical margin at the second resection (p=0.006) were independent predictors of poor survival following the second resection. The type of first resection was not associated with survival but was associated with the ability to achieve a histological negative surgical margin at the second liver resection (p=0.01). CONCLUSION: Three independent predictors of survival were identified. Major initial liver resection was associated with a reduced ability to achieve surgical clearance at the second resection. For colorectal liver metastases, major resection should only be performed if a negative margin cannot be achieved by minor resection.