建立结核感染T淋巴细胞斑点试验阳性的纵隔淋巴结结核与Ⅰ/Ⅱ期结节病诊断模型初探  

Preliminary study on establishing the diagnostic model for smear negative tuberculosis and stageⅠ/Ⅱsarcoidosis with hilar/mediastinal lymphadenopathy and positive T-SPOT.TB results

作　　者：陈凤芳 马俊[2] 黄劲 尹洪云[2] 沙巍[2] 杨光红 冯永红[2] CHEN Feng-fang;MA Jun;HUANG Jin;YIN Hong-yun;SHA Wei;YANG Guang-hong;FENG Yong-hong(School of Public Health,Guizhou Medical University,Key Laboratory of Environmental Pollution and Disease Control,Ministry of Education,Guiyang 550025,China)

机构地区：[1]贵州医科大学公共卫生学院环境污染与疾病监控教育部重点实验室,贵阳550025 [2]同济大学附属上海市肺科医院结核病临床研究中心上海市结核病(肺)重点实验室

出　　处：《中国防痨杂志》2020年第8期832-837,共6页Chinese Journal of Antituberculosis

基　　金：国家自然科学基金(81471563,81771692,81760578,81971558);贵州省教育厅开放项目(黔教合KY字〔2018〕482)。

摘　　要：目的建立结核感染T淋巴细胞斑点试验(T-SPOT.TB)检测为阳性而痰涂片为阴性的纵隔淋巴结结核与Ⅰ/Ⅱ期结节病患者的诊断模型,以提高临床对两种疾病鉴别诊断的敏感度和特异度。方法回顾性分析2013年1月至2018年12月上海市肺科医院收治的T-SPOT.TB检查阳性,且未接受过抗结核药品及糖皮质激素治疗的经胸部CT检查和活检组织病理学检查确诊的100例纵隔淋巴结结核患者(纵隔淋巴结结核组)和50例结节病患者(结节病组;Ⅰ期结节病8例,Ⅱ期结节病42例)的临床实验室检查指标,建立两种疾病鉴别诊断的logistic回归模型,通过ROC曲线下面积(AUC)、敏感度和特异度等指标评价模型的诊断效能。结果对选取的纵隔淋巴结结核患者与Ⅰ/Ⅱ期结节病患者临床检查指标的多因素非条件logistic回归分析显示,血清血管紧张素转化酶(SACE)、血红细胞沉降率(ESR)和淋巴细胞计数(Lym)是纵隔淋巴结结核与Ⅰ/Ⅱ期结节病鉴别的相关独立因素[Waldχ^2=5.826,P=0.016,OR(95%CI)=0.968(0.942~0.994);Waldχ^2=5.583,P=0.018,OR(95%CI)=1.036(1.006~1.068);Waldχ^2=4.074,P=0.044,OR(95%CI)=4.487(1.044~19.279)]。建立的logistic回归模型为:logit(P)=-0.418-0.033×SACE+0.036×ESR+1.501×Lym,ROC曲线分析模型的诊断效能为AUC=0.850(95%CI:0.773~0.927),最佳临界值为1.307,敏感度为82.4%,特异度为80.9%。结论本研究建立的诊断模型具有较高的敏感度和特异度,可以作为临床医师鉴别诊断T-SPOT.TB检测为阳性的纵隔淋巴结结核与Ⅰ/Ⅱ期结节病的辅助参考工具。Objective To establish diagnostic models of mediastinal lymph node tuberculosis patients and stageⅠ/Ⅱsarcoidosis patients,who were both positive in tuberculosis infection T lymphocyte spot(T-SPOT.TB)test and with negative sputum smear,to improve the clinical sensitivity and specificity of the differential diagnosis of the two diseases.Methods Clinical and laboratory examination data of 100 mediastinal lymph node tuberculosis and 50 stageⅠ/Ⅱsarcoidosis patients diagnosed by chest CT examination and biopsy histopathology(nstageⅠ=8,nstageⅡ=42)from Shanghai Pulmonary Hospital between January 2013 and December 2018 were retrospectively analyzed,their T-SPOT.TB test were positive and were not be treated with glucocorticoid or anti-tuberculosis treatment,in a logistic regression model for differential diagnosis of the two diseases was established.The diagnostic efficiency of the model was evaluated by the area under the ROC curve(AUC),sensitivity and specificity.Results Multivariate unconditional logistic regression analysis showed that serum angiotensinconverting enzyme(SACE)(Waldχ^2=5.826,P=0.016,OR(95%CI)=0.968(0.942-0.994)),erythrocyte sedimentation rate(ESR)(Waldχ^2=5.583,P=0.018,OR(95%CI)=1.036(1.006-1.068))and lymphocyte count(Waldχ^2=4.074,P=0.044,OR(95%CI)=4.487(1.044-19.279))were independent factors related for the identification of mediastinal lymph node tuberculosis and stageⅠ/Ⅱsarcoidosis.The established logistic regression model was logit(P)=-0.418-0.033×SACE+0.036×ESR+1.501×Lym.The AUC of diagnostic efficiency of the model were 0.850(95%CI:0.773-0.927),the optimal cut-off value was 1.307,the sensitivity was 82.4%,and specificity was 80.9%,respectively.Conclusion The diagnostic model established in this study has high sensitivity and specificity,and is auxiliary for the differentially diagnose of mediastinal lymph node tuberculosis and stageⅠ/Ⅱsarcoidosis with positive T-SPOT.TB test.

关 键 词：纵隔疾病 结核 淋巴结 结节病 诊断 鉴别 LOGISTIC模型 ROC曲线 评价研究 

分 类 号：R522[医药卫生—内科学] R446.6[医药卫生—临床医学]