基于SEER数据库对5634例唾液腺腺样囊性癌患者生存预测模型的构建  

Establishment of prognostic prediction model in 5634 patients with salivary adenoid cystic carcinoma based on SEER database

作　　者：刘英[1] 李晨曦[1] 方昌 刘慧 龚忠诚[1] LIU Ying;LI Chenxi;FANG Chang;LIU Hui;GONG Zhongcheng(Department of Oral and Maxillofacial Oncology&Surgery,the First Affiliated Hospital of Xinjiang Medical University,School/Hospital of Stomatology Xinjiang Medical University,Stomatological Research Institute of Xinjiang Uygur Autonomous Region,Urumqi 830054)

机构地区：[1]新疆医科大学第一附属医院(附属口腔医院)口腔颌面肿瘤外科,新疆维吾尔自治区口腔医学研究所,乌鲁木齐830054 [2]上海市口腔医院口腔颌面外科,上海市颅颌面发育与疾病重点实验室

出　　处：《现代口腔医学杂志》2023年第3期166-171,共6页Journal of Modern Stomatology

基　　金：口腔颌面发育与再生湖北省重点实验室开放课题(2022kqhm008);新疆护理协会2022年科研基金(2022XH19)。

摘　　要：目的 本研究回顾性分析癌症统计数据库(Surveillance, Epidemiology, and End Results, SEER)中唾液腺腺样囊性癌(adenoid cystic carcinoma, ACC)患者的资料并构建相关列线图模型。方法 利用SEER数据库,提取2010~2017年诊断为ACC的所有患者信息。将数据集分为训练队列和验证队列,并对训练队列和验证队列的平衡进行研究。对训练队列进行单因素logistic回归分析,将有意义的变量纳入多因素模型。采用后向归纳法并利用绩效评价模型,以受试者工作特征(receiver operator characteristic, ROC)曲线和校准曲线确定最终预测模型。结果 预测模型结果显示,≥80岁患者发生淋巴结转移的风险较低(OR=0.610, 95%CI:0.424~0.877),临床Ⅱ期ACC患者淋巴结转移风险低于Ⅰ期(OR=0.422, 95%CI:0.237~0.750)。在模型预测能力方面,训练队列曲线下面积(area under curve,AUC)为0.851(95%CI:0.829~0.873),验证队列曲线下面积为0.849(95%CI:0.813-0.885),表明模型具有良好的判别能力。模型的准确度、灵敏度和特异度分别为0.791(95%CI:0.779~0.804)、0.803(95%CI:0.759~0.846)和0.790(95%CI:0.777~0.804)。结论 本研究SEER分析表明,年龄、原发肿瘤部位和接受手术与否是ACC患者生存的影响因素。多元logistic回归模型对ACC患者的总体预测具有良好的效果。有高危险因素的患者在诊断后应给予更多关注。这些发现可能为ACC患者的治疗和预后提供参考。Objective The nomogram model of the patients with salivary adenoid cystic carcinoma(ACC)was constructed based on the authoritative cancer statistical database(Surveillance,Epidemiology,and End Results,SEER).Methods The SEER tumor registration database was used to extract the information of all patients diagnosed with ACC from 2010 to 2017.The data set is divided into training queue and verification queue,and the balance of training queue and verification queue is studied.Single factor logistic regression analysis was carried out on the training cohort,and meaningful variables were included in the multifactor model.Finally,the final prediction model is determined by the receiver operator characteristic(ROC)curve and calibration curve using the backward induction method and performance evaluation model.Results The prediction model showed that the risk of lymph node metastasis in patients≥80 years old was lower(OR=0.610,95%CI:0.424~0.877),and the risk of lymph node metastasis in patients with clinical stage II ACC was lower than that in stage I(OR=0.422,95%CI:0.237~0.750).In terms of the prediction ability of the model,the area under the curve(AUC)of the training queue is 0.851(95%CI:0.829-0.873),and the area under the curve of the validation queue is 0.849(95%CI:0.813~0.885),indicating that the model has good discrimination ability.The accuracy,sensitivity and specificity of the model were 0.791(95%CI:0.779~0.804),0.803(95%CI:0.759~0.846)and 0.790(95%CI:0.777~0.804),respectively.Conclusion Our SEER analysis showed that age,location of primary tumor and whether to undergo surgery were the factors affecting the survival of ACC patients.Multivariate logistic regression model has a good effect on the overall prediction of ACC patients.Patients with high risk factors should be given more attention after diagnosis.These findings may provide reference for the treatment and prognosis of ACC patients.

关 键 词：唾液腺腺样囊性癌 预后分析 SEER开放获取数据库 

分 类 号：R739.8[医药卫生—肿瘤]