基于超声特征指标的原发性甲状腺淋巴瘤鉴别诊断预测模型构建

Prediction model construction of differential diagnosis of primary thyroid lymphoma based on ultrasonic characteristics

摘要:

目的 探讨原发性甲状腺淋巴瘤鉴别诊断预测因子并构建基于超声特征指标的预测模型，旨在为临床早期准确鉴别诊断及指导后续治疗方案制定提供更多借鉴。方法 回顾性纳入2014年1月至2023年1月于河南省濮阳市人民医院诊治原发性甲状腺淋巴瘤患者47例和甲状腺未分化癌患者50例；比较两组临床及超声特征，采用单因素和多因素法评估原发性甲状腺淋巴瘤鉴别诊断预测因子，原发性甲状腺淋巴瘤鉴别诊断预测模型构建及效能分析。结果 两组合并桥本氏甲状腺炎比例、病灶类型、内部回声类型、钙化比例、囊性变比例及后方回声增强比例比较差异有统计学意义（P<0.05）。多因素分析结果显示是否合并桥本甲状腺炎、病灶类型、内部回声类型、有无后方回声增强、钙化及囊性变情况均可作为原发性甲状腺淋巴瘤鉴别诊断独立预测因子（P<0.05）。利用回归模型的独立影响因素以及P值预测概率对患者原发性甲状腺淋巴瘤风险进行预测，约登指数分别为53.70%、37.84%、79.90%、77.74%、35.49%、82.70%、92.33%。结论 是否合并桥本甲状腺炎、病灶类型、内部回声类型、有无后方回声增强、钙化及囊性变均可指导原发性甲状腺淋巴瘤诊断；而基于超声特征指标构建预测模型用于原发性甲状腺淋巴瘤与甲状腺未分化癌鉴别诊断具有良好效能。

Abstract:

Objective To investigate the prediction index of differential diagnosis of primary thyroid lymphoma and construct prediction model based on ultrasonic characteristics to provide more reference for the early differential diagnosis and treatment of invasive lung adenocarcinoma.Methods Forty-seven patients with primary thyroid lymphoma and 50 patients with undifferentiated thyroid cancer were retrospectively chosen in the period from January 2014 to January 2023. The clinicopathological and ultrasonic features of the two groups were compared, and the prognostic factors for differential diagnosis of primary thyroid lymphoma were evaluated by univariate and multivariate analysis method. Construction and efficacy analysis of differential diagnosis prediction model for primary thyroid lymphoma were done.Results There were significant differences in the proportion of combined Hashimoto's thyroiditis, lesion type, internal echo type, calcification ratio, cystic change ratio and posterior echo enhancement ratio between the two groups (P<0.05). The results of multivariate analysis showed that whether Hashimoto's thyroiditis was complicated, lesion type, internal echo type, posterior echo enhancement, calcification and cystic change could be used as independent predictors for differential diagnosis of primary thyroid lymphoma (P<0.05). The independent influencing factors of the regression model and the prediction probability of P-value were used to predict the risk of primary thyroid lymphoma of the patients, and the Youden index was 53.70%, 37.84%, 79.90%, 77.74%, 35.49%, 82.70%, 92.33%, respectively.Conclusion The presence or absence of Hashimoto thyroiditis, lesion type, internal echo type, posterior echo enhancement, calcification and cystic change can guide the diagnosis of primary thyroid lymphoma. The prediction model based on ultrasonic characteristics has satisfactory efficacy in the differential diagnosis of primary thyroid lymphoma and undifferentiated thyroid carcinoma.