背景:晚期胃癌(AGC)的预后相对较差,长期生存取决于及时的干预。目前,预测生存率仍然是一个热门话题。影像组学和免疫组织化学相关技术在癌症研究中的应用日益广泛。然而,它们在预测AGC患者长期生存率方面的整合尚未得到充分探索.
方法:收集2015-2019年大连大学附属中山医院诊断为AGC的患者150例。遵循严格的纳入和排除标准,90例患者纳入分析。我们收集了入选患者的术后病理标本,使用免疫组织化学技术分析MAOA的表达水平,并将这些水平量化为MAOAHScore。获得患者腹部普通CT图像,描绘了L3椎体水平的感兴趣区域,并提取了影像组学特征。LassoCox回归用于选择重要特征以建立放射学风险评分,将其转换为名为风险的分类变量,并使用Cox回归来识别独立的预测因子以构建临床预测模型。ROC,DCA,和校准曲线验证了模型的性能。
结果:入选患者的平均年龄为65.71岁,包括70名男性和20名女性。多因素Cox回归分析显示风险(P=0.001,HR=3.303),MAOAHScore(P=0.043,HR=2.055),和TNM分期(P=0.047,HR=2.273)是3年总生存期(OS)的独立预后危险因素,在3年疾病特异性生存期(DSS)的分析中也发现了类似的结果。开发的列线图可以预测3年的OS和DSS率,ROC曲线下面积(AUC)分别为0.81和0.797。联合校准和决策曲线分析(DCA)证实了列线图的良好预测性能和临床实用性。
结论:结合免疫组织化学和肌肉脂肪特征可以更准确地预测胃癌患者的长期生存。本研究为更深入地理解AGC中的生存预测提供了新的视角和方法。
BACKGROUND: The prognosis of advanced gastric cancer (AGC) is relatively poor, and long-term survival depends on timely intervention. Currently, predicting survival rates remains a hot topic. The application of radiomics and immunohistochemistry-related techniques in cancer research is increasingly widespread. However, their integration for predicting long-term survival in AGC patients has not been fully explored.
METHODS: We Collected 150 patients diagnosed with AGC at the Affiliated Zhongshan Hospital of Dalian University who underwent radical surgery between 2015 and 2019. Following strict inclusion and exclusion criteria, 90 patients were included in the analysis. We Collected postoperative pathological specimens from enrolled patients, analyzed the expression levels of MAOA using immunohistochemical techniques, and quantified these levels as the MAOAHScore. Obtained plain abdominal CT images from patients, delineated the region of interest at the L3 vertebral body level, and extracted radiomics features. Lasso Cox regression was used to select significant features to establish a radionics risk score, convert it into a categorical variable named risk, and use Cox regression to identify independent predictive factors for constructing a clinical prediction model. ROC, DCA, and calibration curves validated the model\'s performance.
RESULTS: The enrolled patients had an average age of 65.71 years, including 70 males and 20 females. Multivariate Cox regression analysis revealed that risk (P = 0.001, HR = 3.303), MAOAHScore (P = 0.043, HR = 2.055), and TNM stage (P = 0.047, HR = 2.273) emerged as independent prognostic risk factors for 3-year overall survival (OS) and The Similar results were found in the analysis of 3-year disease-specific survival (DSS). The nomogram developed could predict 3-year OS and DSS rates, with areas under the ROC curve (AUCs) of 0.81 and 0.797, respectively. Joint calibration and decision curve analyses (DCA) confirmed the nomogram\'s good predictive performance and clinical utility.
CONCLUSIONS: Integrating immunohistochemistry and muscle fat features provides a more accurate prediction of long-term survival in gastric cancer patients. This study offers new perspectives and methods for a deeper understanding of survival prediction in AGC.