摘要：
Objective To evaluate the reliability of automated cell and differential counts for different sources of body fluid (BF) specimens.Methods 150 BF specimens were collected from outpatients and inpatients of the Peking Union Medical College Hospital from 2009 May to 2013 March,including 75 serous cavity effusion,24 cerebrospinal fluid (CSF),44 benign bebronchoalveolar lavage fluid (BALF),and 7 others.Cell and differential counts were measured by the BF mode of XE-5000 hematology analyzer,and compared to the microscopy analysis,Wright' s-Giemsa and Hematoxylin-Eosin (HE).Data analysis were performed by Intraclass Correlation Coefficient (ICC),Spearman Correlation,Passing-Bablok Regression,and ROC curve.Results The consistency for serous cavity effusion and CSF was superior to that of BALF (ICC ＞ 0.9518 and ＜ 0.8986).The consistency of differential count for serous cavity effusion (ICCMN ＞0.8413,ICCPMN ＞ 0.8950,ICCEOS ＞ 0.9048,ICCHF ＞ 0.7715) were superior to that for CSF and BALF (ICCMN ＜ 0.7276,ICCPMN ＜ 0.7525,ICCEOS ＜ 0.8421 及 ICCHF ＜ 0.1615).The poor precision (CV ＞ 20％) was obtained from 14 specimens with RBC ≤490 × 106/L,including 8 CSF,5 serous cavity effusion and 1 drainage.The area under ROC curve of HF％ for predicting large cells and tumor cells were 0.806 and 0.654 respectively.The sensitivity and specificity for predicting large cells of HF％ were 0.83％ and 0.71％,respectively.The median (min,max) of HF％ for malignant specimens [19.4％ (0％-100.1％)],were lower than that for benign BALF[31.1％ (2.3％-260.7％)],the differences of them showed statistical significance (Z =-2.836,P ＜ 0.05).With the exception of benign BALF,the median (min,max) of HF％ for other benign specimens was 1.5％ (0％-85％),which was significantly lower than that for malignant specimens 19.4％ (0％-100.1％),the differences of them showed statistical significance (Z =-2.276,P ＜ 0.05).Conclusions Cell and differential counts for serous cavity effusion are better than that of CSF and BALF,and the HF for predicting the large cells is superior to tumor cells by the BF mode of XE-5000 analyzer.%目的 评估自动化体液模式对不同类型的体液标本进行细胞计数与分类的可靠性.方法 收集2009年5月至2013年3月北京协和医院门诊及住院患者的150份体液标本,包括75份浆膜腔积液、24份脑脊髓液(CSF)、44份良性支气管肺泡灌洗液(BALF)和7份引流液及穿刺液,通过XE-5000全自动血液分析仪体液模式进行细胞计数和分类,结合人工镜检法、瑞氏-吉姆萨及苏木素-伊红染色法评估仪器法的可靠性,通过组内相关系数(ICC)、Spearman相关、Passing-Bablok回归及ROC曲线进行数据分析.结果 浆膜腔积液和CSF的自动化红细胞和有核细胞计数优于BALF(ICC ＞0.9518与ICC＜0.8986),浆膜腔积液的自动化细胞分类计数(ICCMN＞0.8413,ICCPMN＞0.8950,ICCEOS ＞0.9048,ICCHF ＞0.7715)优于CSF和BALF(ICCMN＜0.7276,ICCPMN＜0.7525,ICCEOS＜0.8421,ICCHF ＜0.1615).细胞数≤490×106/L的14份标本(包括8份CSF、5份浆膜腔积液以及1份引流液),其精密度较差(CV＞20％).高荧光强度细胞用于预测体液中大细胞(巨噬细胞、间皮细胞、肿瘤细胞)和肿瘤细胞的曲线下面积分别为0.806和0.654,HF预测大细胞的敏感度和特异度分别为82％和76％.恶性标本HF水平为19.4％(0％ ～100.1％),低于良性BALF标本的31.1％(2.3％ ～260.7％),差异有统计学意义(Z=-2.836,P＜0.05);除外良性BALF标本,其他良性标本HF为1.6％(0％～85％),明显低于恶性标本的19.4％ (0％ ～ 100.1％),差异有统计学意义(Z=-2.276,P＜0.05).结论 XE-5000体液模式用于浆膜腔积液的细胞计数与分类优于CSF和BALF,HF用于大细胞的预测优于肿瘤细胞.
