医学
肝细胞癌
置信区间
射频消融术
倾向得分匹配
接收机工作特性
逻辑回归
磁共振成像
烧蚀
内科学
放射科
胃肠病学
肿瘤科
作者
Sunyoung Lee,Tae Wook Kang,Kyoung Doo Song,Min Woo Lee,Hyunchul Rhim,Hyo Keun Lim,So Yeon Kim,Dong Hyun Sinn,Jong Man Kim,Kyunga Kim,Sang Yun Ha
出处
期刊:Annals of Surgery
[Ovid Technologies (Wolters Kluwer)]
日期:2019-05-01
卷期号:273 (3): 564-571
被引量:244
标识
DOI:10.1097/sla.0000000000003268
摘要
Objective: We compared surgical resection (SR) and radiofrequency ablation (RFA) as first-line treatment in patients with hepatocellular carcinoma (HCC) based on the risk of microvascular invasion (MVI). Background: The best curative treatment modality between SR and RFA in patients with HCC with MVI remains unclear. Methods: Data from 2 academic cancer center-based cohorts of patients with a single, small (≤3 cm) HCC who underwent SR were used to derive (n = 276) and validate (n = 101) prediction models for MVI using clinical and imaging variables. The MVI prediction model was developed using multivariable logistic regression analysis and externally validated. Early recurrence (<2 years) based on risk stratification between SR (n = 276) and RFA (n = 240) was evaluated via propensity score matching. Results: In the multivariable analysis, alpha-fetoprotein (≥15 ng/mL), protein induced by vitamin K absence-II (≥48 mAU/mL), arterial peritumoral enhancement, and hepatobiliary peritumoral hypointensity on magnetic resonance imaging were associated with MVI. Incorporating these factors, the area under the receiver operating characteristic curve of the predictive model was 0.87 (95% confidence interval: 0.82–0.92) and 0.82 (95% confidence interval: 0.74–0.90) in the derivation and validation cohorts, respectively. SR was associated with a lower rate of early recurrence than RFA based on the risk of MVI after propensity score matching ( P < 0.05). Conclusions: Our model predicted the risk of MVI in patients with a small (≤ 3 cm) HCC with high accuracy. Patients with MVI who had undergone RFA were more vulnerable to recurrence than those who had undergone SR.
科研通智能强力驱动
Strongly Powered by AbleSci AI