Portal vein embolization does not affect the long-term survival and risk of cancer recurrence among colorectal liver metastases patients: A prospective cohort study.

BACKGROUND: Previous studies comparing the survival outcomes of liver resections with and without preoperative portal vein embolization (PVE) for colorectal liver metastases (CLM) have linked PVE to higher rate of tumor progression, lower overall survival (OS) and lower disease-free survival (DFS). The lack of adjusted models to compare these outcomes is a limitation of these studies since patients requiring PVE may differ significantly from the ones receiving upfront surgery. MATERIALS AND METHODS: Prospective cohort study of 128 patients undergoing CLM resection. The OS analysis followed an intent-to-treat (ITT) approach. The adjusted impact of PVE on OS and DFS was evaluated using multivariate Cox regression models. RESULTS: Seventy-one patients underwent PVE before attempting a liver resection while 57 received upfront surgery (NoPVE). All NoPVE patients were resected while 14 PVE participants (19.7%) were not operated (tumor progression = 9/14). PVE patients had a significantly higher preoperative lesions count (3 [1.75-4] vs 1 [1-2.5]; p < 0.001), a higher prevalence of bilateral metastases (23.5% vs 8.8, p = 0.028) and a higher count of neo-adjuvant chemotherapy cycles compared to NoPVE patients. The OS of PVE patients was similar to NoPVE participants (44.7 months [26.9-69.5] vs 49.0 [24.9-64.8], p = 0.761). The DFS of resected PVE patients was higher than NoPVE patients (33.2 months [10.7-54.6] vs 23.4 months [14.1-58.1], p = 0.991). In the adjusted models, preoperative lesions count was the only significant predictor of overall mortality (HR+IC95 = 1.06 (1.02-1.11) p = 0.005) and cancer recurrence (HR+IC95 = 1.14 (1.03-1.27) p = 0.012). CONCLUSION: In the context of CLM, patients requiring PVE differ significantly from patients receiving upfront surgery. This confirms the need for adjusted models when comparing the clinical outcomes of both groups. Our adjusted analysis suggests that PVE is not a significant predictor of a lower OS or DFS. PVE allowed the resection of 80% of participants with initially unresectable CLM. INSTITUTIONAL PROTOCOL NUMBER: 12.106 STUDY REGISTRATION NUMBER: NCT03168230.