Radiology. 2020 Nov 10:200109. doi: 10.1148/radiol.2020200109. Online ahead of print.
Background Local tumor progression (LTP) is associated with poorer survival in patients undergoing radiofrequency ablation (RFA) for colorectal liver metastasis (CLM). An algorithmic strategy to predict LTP may help in selection of patients who would benefit most from RFA for CLM. Purpose To estimate local tumor progression-free survival (LTPFS) following RFA of CLM and develop an algorithmic strategy based on clinical variables. Materials and Methods In this retrospective study, between March
2000 and December 2014, patients who underwent percutaneous RFA for CLM were randomly split into development (60%) and internal validation (40%) data sets. Kaplan-Meier method was used to estimate LTPFS and overall survival (OS) rates. Independent factors affecting LTPFS in the development data set were investigated by using multivariable Cox proportional hazard regression analysis. Risk scores were assigned to the risk factors and applied to the validation data set. Results A total of 365 patients (mean age, 60 years ± 11 [standard deviation]; 259 men) with 512 CLMs were evaluated. LTPFS and OS rates were 85% and 92% at 1 year, 73% and 41% at 5 years, 72% and 30% at 10 years, and 72% and 28% at 15 years, respectively. Independent risk factors for LTP included tumor size of 2 cm or greater (hazard ratio [HR], 3.8; 95% CI: 2.3, 6.2; P < .001), subcapsular tumor location (HR, 1.9; 95% CI: 1.1, 3.1; P = .02), and minimal ablative margin of 5 mm or less (HR, 11.7; 95% CI: 4.7, 29.2; P < .001). A prediction model that used the risk factors had areas under the curve of 0.89, 0.92, and 0.90 at 1, 5, and 10 years, respectively, and it showed significantly better areas under the curve when compared with the model using the minimal ablative margin of 5 mm or less alone. Conclusion Radiofrequency ablation provided long-term control of colorectal liver metastases. Although minimal ablative margin of 5 mm or less was the most dominant factor, the multifactorial approach including tumor size and subcapsular location better predicted local tumor progression-free survival. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Soulen and Sofocleous in this issue.