Introduction The deleterious effect of complications on long term survival following resection of colorectal liver metastases (CRLM) remains controversial. Is it truly the effect of complications per seor simply a reflection of the burden of disease and tumour biology?

Method From our prospective database of patients resected with CRLM between January 2000 and January 2014 (n = 1457), we used Kaplan-Meier analysis to calculate cancer-specific survival (CSS) and disease-free survival (DFS) for patients with no complications (Group A n = 1084), any complication (Clavien-Dindo I-IV) (Group B n = 373) and major complications (Clavien-Dindo III/IV) (Group C n = 95). In addition, the Basingstoke Predictive Index (BPI), a validated predictor of CSS comprised of primary tumour factors and metastatic disease burden, was calculated for each group. From the BPI, we created case-matched pairs for patients with no complications (Group D n = 456), any complications (Group E n = 154) and serious complications (Group F n = 33) and recalculated CSS and DFS in those groups.

Results Patients without complications (Group A) had significantly better CSS and DFS compared with patients with complications (Group B) (CSS p = 0.001; DFS p = 0.02) or serious complications (Group C) (CSS p = 0.009, DFS p = 0.045). The median preoperative and postoperative BPI scores were significantly higher for Group B (6 and 7) and Group C (6 and 7) compared to Group A (5 and 6) (p < 0.05). Analysis of case-matched survival showed no difference between patients without complications (Group D) compared to those with complications (Group E) (CSS p = 0.098, DFS p = 0.089) or serious complications (Group F) (CSS p = 0.129, and DFS p = 0.982).

Conclusion These data show that when the influence of primary tumour biology and metastatic disease burden are corrected for, complications per sedo not influence survival after resection of CRLM.

Disclosure of interest None Declared.