Effect of the learning curve on survival after laparoscopic liver resection for colorectal metastases

This paper aimed to evaluate the effect of the learning curve of laparoscopic liver resection on the long-term survival of patients with colorectal liver metastasis. In this study of 286 patients, RA-CUSUM of the conversion rate and log regression analysis of blood loss estimated a learning curve of 68 cases to attain proficiency in laparoscopic liver resection. When this was applied to patients with colorectal liver metastasis, the overall and recurrence-free survival rates were not compromised by the learning curve. Background Laparoscopic liver resection (LLR) is a highly demanding procedure with great variability. Previously published randomized trials have proven oncological safety of laparoscopic liver resection (LLR) as compared to open surgery. However, these were started after the learning curve (LC) was established. This leaves the question of whether the LC of LLR in the early laparoscopic era has affected the survival of patients with colorectal liver metastasis (CRLM). Methods All consecutive LLRs performed by a single surgeon between 2000 and 2019 were retrospectively analysed. A risk-adjusted cumulative sum (RA-CUSUM) chart for conversion rate and the log regression analysis of the blood loss identified two phases in the LC. This was then applied to patients with CRLM, and the two subgroups were compared for recurrence-free (RFS) and overall survival (OS). The analysis was repeated with propensity score-matched (PSM) groups Results A total of 286 patients were included in the LC analysis, which identified two distinct phases, the early (EP; 68 patients) and the late (LP; 218 patients) phases. The LC was applied to 192 patients with colorectal liver metastasis (EPc, 45 patients; LPc, 147 patients). For patients with CRLM, R0 resection was achieved in 93 per cent: 100 per cent in the EPc group and 90 per cent in the LPc group (P = 0.026). Median OS and RFS were 60 and 16 months, respectively. The 5-year OS and RFS were 51 per cent and 32.7 per cent, respectively. OS (hazard ratio (h.r.) 0.78, 95 per cent confidence interval (c.i.) 0.51 to 1.2; P = 0.286) and RFS (h.r. 0.94, 95 per cent c.i. 0.64 to 1.37; P = 0.760) were not compromised by the learning curve. The results were replicated after PSM. Conclusion In our experience, the development of a laparoscopic liver resection programme can be achieved without adverse effects on the long-term survival of patients with CRLM.