Background: The best measure of the technical success of radiofrequency ablation (RFA) is local recurrence (LR). The aim of this prospective study is to identify factors that predict LR.
Methods: Three hundred thirty-five patients with 1032 unresectable liver tumors underwent laparoscopic RFA between November 1999 and August 2005. All lesions were assessed prospectively regarding tumor type, size, liver segment, blood vessel proximity, and central or peripheral location in the operating room and size of ablation zone at 1-week computed tomographic (CT) scans. Lesions that recurred in follow-up CT scans were identified prospectively. LR was categorized as contiguous or adjacent. Univariate Kaplan-Meier and Cox proportional hazard models were used for statistical analysis.
Results: LR was identified 21.7% of tumors on CT scans with a mean follow-up of 17 months (median, 12 months; range, 3-68 months). This was contiguous in 70% and adjacent in 30%. LR rate per tumor was highest for colorectal metastasis (34%), followed by noncolorectal, nonneuroendocrine metastasis (22%), hepatocellular carcinoma (18%), and neuroendocrine metastasis (6%). By univariate analysis, tumor type and size, ablation margin, liver segmental location, blood vessel proximity, and type of ablation (first time vs. repeat) were found to affect LR. The Cox proportional hazard model identified tumor type, tumor size, ablation margin, and blood vessel proximity to be independent predictors of LR.
Conclusion: LR after RFA is predicted by certain tumor characteristics and technical factors. This information can be used intraoperatively to identify those tumors at a higher risk for failure.