Reappraisal of the risks and benefits of major liver resection in patients with initially unresectable colorectal liver metastases.

3609 Background: Improvements in both surgical technique and efficacy of chemotherapy have increased the rate of resection for patients with initially unresectable colorectal liver metastases (IU-CRLM). We aimed to evaluate the short and long-term outcomes of major hepatectomy for such patients. Methods: From 2000 to 2011, 257 patients underwent major hepatectomy for CRLM. Seventy-eight (30%) of these patients were considered IU and required portal vein occlusion and/or ≥12 cycles or change in induction chemotherapy regimen to achieve resectability. Results: IU patients had respectively more lesions (5.6 vs.3.6, p=0.001), more frequently bilobar (70% vs.50% p=0.008) and synchronous (83.3% vs.70%, p=0.027) than initially resectable (IR) patients. Post-operative mortality (12.8% vs.1.7%, p=0.001) and major complications (46.2% vs.22.3%, p=0.0001) were higher in IU patients. An associated metabolic syndrome (HR 5.2, CI 1.2-21.9, p=0.025), high grade sinusoidal lesions (HR 2.4, CI 1.1-5.8, p=0.044) and the need for vascular reconstruction (HR 6.3, CI 1.2-34.4, p=0.032) were significant risk factors for major morbidity in IU patients. Significantly fewer IU patients received adjuvant chemotherapy in case of major postoperative complications compared to IR patients (47% vs. 83%, p=0.001). Overall 5-year survival was significantly lower in IU than in IR patients (26% vs.55%, p=0.032) and all IU patients had tumor recurrence within 3 years. The absence of adjuvant chemotherapy and tumor size ≥5 cm were the only factors associated with poor survival in multivariate analysis for IU patients. Conclusions: IU-CRLM patients requiring major liver resection displayed higher morbidity and mortality rates than IR ones, therefore compromising both short and long-term outcomes. Multimodal strategy should be reassessed in the presence of metabolic syndrome, sinusoidal lesions or major vascular involvement.