Zhang Wei, Wu Di, Wang Xinping, Zhang Hua, Yu Ming
Department of Digestive Surgery, Xi'an People's Hospital (Xi'an Fourth Hospital), Xi'an, China.
Department of Hematology, The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, China.
Front Oncol. 2025 Aug 21;15:1524765. doi: 10.3389/fonc.2025.1524765. eCollection 2025.
This study aims to develop a prediction model for invasive metastasis of primary liver cancer based on serum extracellular matrix metalloproteinase-inducing factor (CD147) and interleukin-6 (IL-6).
Between July 2022 and August 2024, 170 surgically treated primary hepatocellular carcinoma patients at our hospital were recruited. They were divided into a training group ( = 120) and a validation group ( = 50) at a 7:3 ratio. Univariate and multivariate logistic regression analyses were applied in the training group to identify factors related to invasive metastasis. A risk factor-based bar chart prediction model was then constructed and internally tested. Its goodness of fit was evaluated, and the model's diagnostic efficacy was assessed using the ROC curve. Finally, decision curve analysis (DCA) was performed to evaluate the model's clinical value.
In the training group, compared with the noninvasive metastasis group, patients in the invasive metastasis group had a significantly lower percentage of intact envelope and tumor size ≥5 cm, and significantly higher serum alpha-fetoprotein (AFP), alkaline phosphatase (ALP), C-reactive protein/albumin ratio (CAR), oncoglobulin (CEA), CD147, and IL-6 levels (all < 0.05). After logistic multifactorial analysis, intact envelope, tumor > 5 cm, AFP, CAR, CD147, and IL-6 were identified as independent influencing factors for invasive metastasis of primary hepatocellular carcinoma (all < 0.05). A column chart model was constructed. The C-index of the training and validation groups was 0.884 (95% confidence interval [CI]: 0.738-0.932) and 0.841 (95% CI: 0.741-0.939), respectively. The calibration curves showed good agreement between the predicted probability and the actual probability in both the training and validation groups, without significant deviation. The area under the curve (AUC) of the ROC analysis was 0.852 (95% CI: 0.824-0.979) and 0.839 (95% CI: 0.791-0.912), respectively. DCA indicated that the model had clinical application value within a certain range of threshold probabilities.
The prediction model based on serum CD147, IL-6, and other risk factors for the invasion and metastasis of primary hepatocellular carcinoma demonstrates high diagnostic value.
本研究旨在基于血清细胞外基质金属蛋白酶诱导因子（CD147）和白细胞介素-6（IL-6）建立原发性肝癌侵袭转移的预测模型。
2022年7月至2024年8月，招募我院170例接受手术治疗的原发性肝细胞癌患者。按照7:3的比例将他们分为训练组（n = 120）和验证组（n = 50）。在训练组中应用单因素和多因素逻辑回归分析来确定与侵袭转移相关的因素。然后构建基于危险因素的柱状图预测模型并进行内部测试。评估其拟合优度，并使用ROC曲线评估模型的诊断效能。最后，进行决策曲线分析（DCA）以评估模型的临床价值。
在训练组中，与非侵袭转移组相比，侵袭转移组患者的完整包膜比例和肿瘤大小≥5 cm的比例显著更低，而血清甲胎蛋白（AFP）、碱性磷酸酶（ALP）、C反应蛋白/白蛋白比值（CAR）、癌胚球蛋白（CEA）、CD147和IL-6水平显著更高（均P < 0.05）。经过逻辑多因素分析，完整包膜、肿瘤>5 cm、AFP、CAR、CD147和IL-6被确定为原发性肝细胞癌侵袭转移的独立影响因素（均P < 0.05）。构建了柱状图模型。训练组和验证组的C指数分别为0.884（95%置信区间[CI]：0.738 - 0.932）和0.841（95%CI：0.741 - 0.939）。校准曲线显示训练组和验证组的预测概率与实际概率之间具有良好的一致性，无显著偏差。ROC分析的曲线下面积（AUC）分别为0.852（95%CI：0.824 - 0.979）和0.839（95%CI：0.791 - 0.912）。DCA表明该模型在一定阈值概率范围内具有临床应用价值。
基于血清CD147、IL-6等危险因素的原发性肝细胞癌侵袭转移预测模型具有较高的诊断价值。
