All patients undergoing a liver resection for CRLM at the University Hospital of Wales (Cardiff, UK) across a 6-year period, 1^st January 2013 until 1^st January 2019, were included. Patients were excluded if post-operative histology identified an alternative pathology. Operative procedures were performed with the intention of curative liver resection after discussion at a regional specialist multidisciplinary team meeting (MDT). There was no rigid protocol for the use of primary first surgery or neo-adjuvant therapy, but rather this was assessed on a case by case basis. Operative strategies utilised open and laparoscopic procedures. If the future liver remnant was considered inadequate either Portal vein embolization (PVE) or partial ALPPS (Associating Liver Partition and Portal vein ligation in staged hepatectomy) was performed.⁶ Patients were routinely managed for the first post-operative night in the post anaesthetic care unit, a level 2 care environment, before returning to a ward.

Clinicopathological and operative data were obtained from both a prospectively maintained database, and retrospectively from electronic patient databases. Data was anonymised and included baseline characteristics (American Society of Anaesthesiologists class (ASA), gender, age, chemotherapy exposure, extrahepatic disease) operative variables (extent of liver resection, intraoperative blood loss), tumour variables (margin status, bile duct involvement, vascular involvement, number of metastases, size of largest metastases), post-operative complications, disease recurrence and date of death. The tumour burden score (TBS) was developed by Sasaki et al.⁷ as a prognostic marker. This was calculated where TBS²=(maximum tumour diameter)²× (number of liver lesions)².

Comparisons between the study groups were assessed with the student t-test for continuous variables, and the chi-square test for categorical variables. The Kaplan-Meier method was used to calculate survival estimates and comparisons made with the log rank test. Univariate and Multivariate regression analysis was performed using the Cox Proportional Hazards model. All statistical calculations were undertaken using SPSS Version 26 (IBM Corp. Armonk, NY). A p-value less than 0.05 was considered statistically significant.

The CRLM was classified as synchronous disease if identified at, or before, the time of surgery for the primary colorectal cancer. Liver metastases presenting within 12 months or after 12 months after the colorectal surgery were considered early and late metachronous disease respectively.

Neoadjuvant and adjuvant chemotherapy refers to treatment before and after the liver surgery, not the primary colorectal surgery. Patients were classed as receiving chemotherapy if they started treatment. Resection margin positivity (R1) was defined as microscopic tumour within 1 mm of the margin on any specimen. Recurrent hepatic disease that occurred simultaneously with extra-hepatic disease was coded as a hepatic recurrence for analysis purposes.

Patients were followed-up for 10 years or until a clinician determined further treatments would be inappropriate. Patients were seen annually for a clinical review, serum tumour markers and a CT thorax, abdomen and pelvis to identify treatable recurrent disease.

Across the study period, 339 operative liver procedures were performed on 289 consecutive patients for CRLM. The median age was 66 years (range 20-93), the median ASA was 2 and 66.1% of patients were males. The majority of patients had synchronous CRLM (54.9%), compared with 14.1% early and 30.0% late metachronous disease. The median length of stay for all procedures was 7 days with a 30-day and 90-day mortality of 1.4% (4/289) and 2.4% (7/289) respectively. Four deaths were for liver related complications and 3 patients were discharged home well but died of a cardiac event within 90 days of surgery.

The most common procedure was a liver metastasectomy (44.2%). Sixteen patients underwent a partial ALPPS to treat CRLM and of these, 4 patients did not return for a second stage due to disease progression and were excluded form survival estimate analysis. A further 8 patients underwent their first liver resection for CRLM prior to 2013 and therefore were also excluded. Fig. 1 summarises the patient procedures during the study period and those included in survival analysis. The 1, 3 and 5-year OS of 277 patients was 90.2%, 70.5% and 52.3% respectively, with a median survival of 70 months.

A total of 67 operations (19.8%) were performed on patients aged 75 years and older. The OS for those under 75 years was not statistically different from those over 75 years old (p=0.772, median 70 vs. 60 months) and is displayed in Fig. 2A. The 1, 3 and 5-year survival rates were 90.6%, 70.4%, 53.4% vs. 88.8%, 70.9%, 47.6% respectively. The patient characteristics of the two age groups are compared in Table 1. There was a lower rate of R1 resection in the older age group (20.4% vs. 4.5%, p=0.02) as well as a lower adjuvant chemotherapy rate (60.3 vs. 35.8%, p<0.001).

The median total number of tumours resected per patient was 2 in both age cohorts but ranged from 1 to 20 in those aged <75 years and 1 to 9 in those ≥75 years (p=0.383).

In patients who underwent curative liver resection in the absence of extra-hepatic disease, the disease-free survival was not different in those <75 years vs. those greater than or equal to 75 years old (1, 3 and 5-year survival rates 68.7%, 43.7% and 41.2% vs. 80.8%, 43.3% and 34%, p=0.521).

Univariate analysis of patient and tumour characteristics showed age >75 years did not influence OS (p=0.773). Table 2 indicates the prognostic ability of other clinicopathological factors. Operation type (p=0.010), greater TBS (p=0.006) and the presence of synchronous extra-hepatic disease (p=0.027) were associated with decreased OS. In multivariate analysis, both operation type and TBS remained predictive of OS (Table 3).