痰液细胞图像分析法筛检锡矿接尘工人肺癌的方法学评价

A test of screening to predict lung cancer among dust-exposed tin miners with sputum imaging cytometry

目的评价痰液细胞图像分析对锡矿接尘工人肺癌的筛检试验，分析影响锡矿肺癌高危人群发病的危险因素。方法1998年用单纯随机抽样的方法选取广西某锡矿队列中45岁以上的男性接尘矿工345人为研究对象，采用全自动细胞图像分析法用经验指标（痰液）对研究对象进行肺癌筛检，其后追访该人群至2006年。以临床确诊为金标准，用筛检试验方法评价痰液细胞图像分析对锡矿接尘工人肺癌的筛检效能及最佳工作点。结果追访期间，共确诊11例肺癌病例；除年龄和职业接触外，吸烟是锡矿高龄接尘工人肺癌发病的危险因素（P=0．0384），调整混杂因素后，高吸烟组的相对危险度为18．21（2．15～154．39）。细胞图像分析法经验筛检指标的灵敏度、特异度和正确指数分别为27．3％、83．5％、10．8％；经拟合受试者工作特征（ROC）曲线，C2．5（DNA指数介于1．25～2．50之间所占百分比）以1．70％为最佳工作点，痰液细胞图像系统检测的灵敏度、特异度、一致率、阳性预测值、阴性预测值和正确指数分别为72．7％、62．3％、62．6％、6．0％、98．6％、35．0％。结论吸烟是高龄接尘工人肺癌的危险因素；使用痰液细胞图像分析法，将指标C2．5的筛检值调整为1．70％，有助于早期发现锡矿高危人群中的可疑肺癌患者。

Objective To evaluate efficacy of sputum imaging cytometry in early diagnosis on lung cancer among tin miners exposed to dust and analyze possible risk factors related to lung cancer among tin miners. Methods From a cohort of tin miners in Guangxi Province, a total of 345 male tin miners higher than 45 years old and with high exposure to crystalline silica dust were randomly selected as the objects. Imaging cytometry was used to implement the sputum analysis for the screening on lung cancer according to the experi- ence diagnostic standard. All objects were then followed up to the end of 2006. Clinical diagnosis of lung cancer was used as the golden standard to evaluate the efficacy of screening. Results From 1998 to 2006, 11 new cases were diagnosed as clinical lung cancer. Except of age and exposure to occupational hazards,smoking status （P=-0.0384） and mean smoking dose （P=-0.0078） were significantly associated with lung cancer, and the adjusted odds ratio of high level to the low was 18.21（2.15-154.39）. The sensitivity, specificity and Youden＇s index of the sputum imaging cytometry for the experience diagnosis were 27.3%, 83.9% and 11.2% respectively. According to the ROC curve analysis, area under ROC Curve （AUC） of C2.5 （the percentage when the DNA index ranged from 1.25 to 2.50） was 0.647 （0.525-0.768）,with the optimal operating point （OOP） of 1.70%. Sensitivity,specificity,agreement rate,positive predictive val-ue,negative predictive value and Youden＇s index for predicting lung cancers in highexposure tin miners were found to be 72.7% ,62.3%, 62.6% ,6.0% ,98.6% and 35.0% respectively. Conclusion Smoking is confirmed as an important risk factor of lung cancer in tin miners. The diagnostic efficiency can be improved if the diag-nostic point of C2.5 is adjusted to 1.70%.