CA19-9测定在4个免疫检测系统的差异及互认能力的评估

摘要

目的 研究4个免疫检测系统对CA19-9检测的差异和一致性,探讨实现CA19-9检测结果互认的基础.方法 检测系统比对研究.汇集整理CA19-9室间质量评价活动批号的数据(卫生部200811 ～ 201215,浙江省080309 ～ 120211)；搜集120份不同CA19-9浓度的新鲜血样本；搜集2010年9月至2012年3月不同检测系统非肿瘤体检人群实测CA19-9数据.选择4个稳定的检测系统Abbott Architect i2000、Roche E170、Beckman Dxi800和Siemens CentaurXP作为研究对象,统计分析在卫生部和浙江省室间质评(冻干粉和液体血清)中的组间差异；参照CLSI EP9-A2进行120份新鲜血样本的比对分析；计算各检测系统实测CA19-9的95％置信区间并分析.采用散点图、加权回归方程进行统计比较.结果 室间质评与新鲜血比对出现颠倒的结果.室间质评均表现出检测系统组间较大的偏倚,斜率bw在1.340到4.683之间,相关系数r较好；新鲜血比对中r不理想,但斜率bw接近于1.新鲜血比对中在推荐浓度27 U/ml时,检测系统两两间偏移分别为Abbott-Roche(-6.41％),Beckman-Roche(-5.07％),Siemens-Roche(13.15％),Beckman-Abbott(2.46％),Siemens-Abbott(22.52％),Siemens-Beckman(39.66％).部分检测系统间实测95％置信区间差异有统计学意义.结论 在研究检测系统差异时,在较低浓度CA19-9的结果在4个检测系统间具有互认的基础,但在高浓度时,检测系统间的差异明显增大,风险较高,尚无法实现互认.%Objective To investigate the differences and harmonization of immunoassay systems in detecting CA19-9 and to assess the possibility of mutual recognition in different laboratories.Methods Data were collected and analyzed from External Quality Assessments (EQA) of NCCL(Lots:200811-201215) and ZJCCL(Lots:080309-120211).120 fresh serum with different concentrations of CA19-9 were collected.The CA19-9 results of healthy people were also collected from September 2010 to March 2012.Four kinds of stable immunoassay systems were involved in our research,including Abbott Architect i2000,Beckman UniCel DxI 800,Roche E170 and Siemens ADVIA Centaur XP.The differences among four system groups were calculated with the EQA data.The fresh serum comparisons were also performed following the guideline of CLSI EP9-A2 The 95％ confidence interval of each immunoassay system was calculated.Comparisons were made by scatter diagrams and weighted regression.Results Both EQA of NCCL and ZJCCL showed better correlation coefficients and larger bias (bw ranged from 1.340 to 4.683) than in fresh serum comparisons.Although the correlation coefficients were all unsatisfactory,the bw were all close to 1 in fresh serum comparisons.When the recommended serum concentration of 27 U/ml was used,the biases were Abbott-Roche-6.41％,Beckman-Roche-5.07％,Siemens-Roche 13.15％,Beckman-Abbott 2.46％,Siemens-Abbott 22.52％ and Siemens-Beckman 39.66％,respectively.Differences of 95％ confidence intervals were statistically significant among parts of the immunoassay systems.Conclusions Only in the lower concentration can CA19-9 results be mutual recognized among four different immunoassay systems,there will be larger differences and risks in the higher concentration.