使用增强磁共振成像（MRI）和临床因素开发和验证危险评分系统，以预测单一肝细胞癌（HCC）复发无病生存率（RFS）。从两个中心回顾性纳入经过治疗单一HCC的连续295名患者，并建立Cox比例风险模型来开发危险评分系统，使用外部数据验证其鉴别能力，并与巴塞罗那肝癌诊所（BCLC）或美国联合委员会癌症（AJCC）分期系统进行比较，使用Harrell的C指数。独立变量为肿瘤大小（每厘米；风险比[HR]，1.07；95％置信区间[CI]：1.02-1.13；p = 0.005），靶样外观（HR，1.74；95％CI：1.07-2.83；p = 0.025），可影像化的静脉内肿瘤或肿瘤血管侵犯（HR，2.59；95％CI：1.69-3.97；p＜0.001），肝胆相位中有非高血供低信号结节的存在（HR，4.65；95％CI：3.03-7.14；p＜0.001），以及病理大血管侵犯（HR，2.60；95％CI：1.51-4.48；p = 0.001）-与肿瘤标志物（AFP ≥ 206 ng / mL或PIVKA-II ≥ 419 mAU / mL）预测术前和术后危险评分系统。验证集中的风险评分显示相当好的鉴别能力（C指数为0.75-0.82），并且在BCLC（C指数为0.61）和AJCC分期系统（C指数为0.58）之上表现出色（ps＜0.05）。术前评分系统将患者分类为无复发风险、中度风险和高风险组，其2年复发率分别为3.3％、31.8％和85.7％。开发和验证的术前和术后危险评分系统可以估计单一HCC手术后RFS。• 风险评分系统比BCLC和AJCC分期系统更好地预测了RFS（C指数为0.75-0.82，而0.58-0.61；ps＜0.05）。• 与肿瘤标志物相结合，五个变量 - 肿瘤大小，靶样外观，可影像化的静脉内肿瘤或肿瘤血管侵犯，肝胆相位中有非高血供低信号结节的存在以及病理大血管侵犯 - 预测了单一HCC的手术后RFS。• 使用术前可获得的因素的风险评分系统将患者分为三个不同的风险组，在验证集中，低风险组、中度风险组和高风险组的2年复发率分别为3.3％、31.8％和85.7％。© 2023年。作者（s），独家许可欧洲放射学会。

To develop and validate risk scoring systems using gadoxetic acid-enhanced liver MRI features and clinical factors that predict recurrence-free survival (RFS) of a single hepatocellular carcinoma (HCC).Consecutive 295 patients with treatment-naïve single HCC who underwent curative surgery were retrospectively enrolled from two centers. Cox proportional hazard models developed risk scoring systems whose discriminatory powers were validated using external data and compared to the Barcelona Clinic Liver Cancer (BCLC) or American Joint Committee on Cancer (AJCC) staging systems using Harrell's C-index.Independent variables-tumor size (per cm; hazard ratio [HR], 1.07; 95% confidence interval [CI]: 1.02-1.13; p = 0.005), targetoid appearance (HR, 1.74; 95% CI: 1.07-2.83; p = 0.025), radiologic tumor in vein or tumor vascular invasion (HR, 2.59; 95% CI: 1.69-3.97; p < 0.001), the presence of a nonhypervascular hypointense nodule on the hepatobiliary phase (HR, 4.65; 95% CI: 3.03-7.14; p < 0.001), and pathologic macrovascular invasion (HR, 2.60; 95% CI: 1.51-4.48; p = 0.001)-with tumor markers (AFP ≥ 206 ng/mL or PIVKA-II ≥ 419 mAU/mL) derived pre- and postoperative risk scoring systems. The risk scores showed comparably good discriminatory powers in the validation set (C-index, 0.75-0.82) and outperformed the BCLC (C-index, 0.61) and AJCC staging systems (C-index, 0.58; ps < 0.05). The preoperative scoring system stratified the patients into low-, intermediate-, and high-risk for recurrence, whose 2-year recurrence rate was 3.3%, 31.8%, and 85.7%, respectively.The developed and validated pre- and postoperative risk scoring systems can estimate RFS after surgery for a single HCC.• The risk scoring systems predicted RFS better than the BCLC and AJCC staging systems (C-index, 0.75-0.82 vs. 0.58-0.61; ps < 0.05). • Five variables-tumor size, targetoid appearance, radiologic tumor in vein or vascular invasion, the presence of a nonhypervascular hypointense nodule on the hepatobiliary phase, and pathologic macrovascular invasion-combined with tumor markers derived risk scoring systems predicting postsurgical RFS for a single HCC. • In the risk scoring system using preoperatively-available factors, patients were classified into three distinct risk groups, with 2-year recurrence rates in the low-, intermediate-, and high-risk groups being 3.3%, 31.8%, and 85.7% in the validation set.© 2023. The Author(s), under exclusive licence to European Society of Radiology.