Two-stage hepatectomy in resection of colorectal liver metastases – a single-institution experience with case-control matching and review of the literature

A retrospective review of a prospectively obtained database of 632 consecutive liver procedures for colorectal liver metastases at the Clinical Department of Abdominal and General Surgery of the University Medical Centre Maribor in Slovenia was performed. This department is a specialised referral centre for hepato-pancreato-biliary surgery. The study period was from 1 January 2000 until 31 December 2020.

Before the surgery, patients consented to their anonymous data being used for research. Therefore, their records were anonymised and deidentified before analysis. Ethical approval for this study was obtained from the National Medical Ethics Committee of the Republic of Slovenia (0120-455/2020/3). All methods were performed following the relevant guidelines and regulations.

Routinely available clinical characteristics were analysed, including patient demographics, performance status defined according to the American Society of Anaesthesiologists Classification (ASA classification)⁶, application of neoadjuvant chemotherapy, preoperative carcinoembryonic antigen (CEA) level, and presence of extrahepatic disease. Primary colorectal tumour variables included the tumour location and nodal invasion. Liver metastasis variables included synchronous/metachronous metastases and the number and size of metastases.

Patients were presented at the multidisciplinary team meeting.² Bilateral colorectal liver metastases were resected in a single procedure when both the volume and function of the future liver remnant were considered sufficient. The parenchyma-sparing principle of liver surgery for colorectal liver metastases was applied.⁷ The types of liver resections were classified according to the Brisbane terminology.⁸ Major liver resections involved three or more adjacent liver segments, including conventional major resections (left/extended left hepatectomies, right/extended right hepatectomies, central hepatectomies).⁸ The analysis of future liver remnant consisted of computed tomography (CT) volumetry, laboratory liver tests (prothrombin time and albumins), and the indocyanine green clearance test.²

Specimens were analysed by a gastrointestinal histopathologist who assessed the resection margin. The histological surgical margins for malignant lesions were defined as microscopically negative (R0) or positive (<1 mm, R1).⁷ In addition, the Clinical Risk Score devised by Fong et al. was applied.⁹

Portal vein embolization, intraoperative selective portal vein ligation, or the ALPPS procedure were performed when the analysis suggested an insufficient future liver remnant.² Portal vein embolization was followed by atrophy of the embolized hemiliver and hypertrophy of the other hemiliver.² TSH with portal vein ligation was performed when the intraoperative findings were unfavourable.⁴ In the first stage, the metastasectomy of one hemiliver was performed along with portal vein ligation for the other hemiliver.⁴ The effect was similar to that of preoperative portal vein embolization.¹⁰ The second stage followed a few weeks later and consisted of a major hepatectomy.^2,4 ALPPS was performed with the same rationale; the difference was the addition of liver parenchyma transection in the first stage.¹¹

FIGURE 1.

RFA has been applied where radical liver resection has been infeasible due to the proximity of large vessels.² Therefore, RFA has been applied intraoperatively as an independent procedure or adjunct to liver resection.² RFA has also been used as a percutaneous procedure. However, these patients were excluded from the analyses.

Patients were followed-up at the outpatient clinic at periodic intervals. The follow-up protocol consisted of a CEA level, a chest radiograph or CT, an abdominal ultrasound, CT, or magnetic resonance imaging every three months for the first two years and every six months afterwards.²

IBM SPSS for Windows Version 28.0 (IBM Corp., Armonk, NY, USA) was used for the statistical analysis. Percentages are reported to one decimal place. A P value ≤ 0.05 was considered statistically significant.

Categorical variables are displayed as numbers with percentages. The differences between categorical variables were tested using the chi-square or Fisher-Freeman Halton test when more than two categories were present. Continuous variables were expressed as medians (minimum-maximum, interquartile range) and analysed with the Mann-Whitney test since the distribution analysis showed the non-normal distribution of data.

Survival data for median OS and RFS are presented as Kaplan-Meier curves, and groups were compared by a log-rank test. The results are expressed in months as the median (95% confidence interval (95% CI)). Survival tables were used for 3-and 5-year OS and RFS, given in percentages.

Case-control matching was performed for 15 patients from the TSH group.¹⁶ Patients from the OSH group (controls) were selected based on the variables that were statistically significant in a bivariate analysis. The sampling was performed without replacement and with maximising execution performance modality. Matched patients were assigned to the case-control matching-OSH group. The statistical analysis of continuous variables was performed with the Wilcoxon signed ranks test. The analysis of categorical variables was performed with the McNemar test.¹⁶ Survival was estimated as described previously.

This study included 166 patients: 151 in the OSH group and 15 in the TSH group. Their clinical characteristics and perioperative outcomes are summarised in Table 1.

To reduce the bias and equilibrate the number of group members, case-control matching was conducted. Patients from the OSH group (controls) were selected based on the predictors that were statistically significant in bivariate analysis (Table 1): number of liver metastases, neoadjuvant chemotherapy, and extent of liver resection.

Case-control matching returned 14 controls among the OSH group, annotated as case-control matching-OSH. All three variables were statistically significant in the case-control matching model (P < 0.001). The Wilcoxon signed ranks test for the median of differences between before and after matching was insignificant (P = 0.317).

After case-control matching, the TSH and OSH groups were compared (Table 2).

Perioperative morbidity and 90-day mortality rates are provided in Table 1 and Table 2.

FIGURE 2.

In the OSH group, seven (4.6%) patients died postoperatively. The causes of death were sepsis (n = 1), cardiorespiratory failure (n = 1), multiorgan failure (n = 3), and posthepatectomy liver failure (n = 2). In the TSH group, two (13.3%) patients died postoperatively. One patient suffered from colonic perforation, was reoperated on several times, and died of multiorgan failure. The second patient died of cardiorespiratory failure after acute myocardial infarction. In the case-control matching-OSH group, the patient (7.1%) died of posthepatectomy liver failure.

Patients were followed until their death or until 31 December 2022. The median follow-up was 174 (95% CI 113–235) months. A summary of the survival analysis is provided in Table 3 and Figure 2.

The RFS, median OS, and 3- and 5-year survival rates were 5 months, 21 months, 33% and 13% in the TSH group; 11 months, 35 months, 49% and 27% in the OSH group; and 8 months, 23 months, 36% and 21% in the case-control matching-OSH group, respectively.