四种血清α-淀粉酶厂家系统测量结果的正确性评价

The evaluation of trueness of 4 a-amylase measurement systems

目的寻找满足ISO15189要求、测量结果正确的血清仅一淀粉酶（AMY）常规检测方法。方法收集2008年3至12月AMY活性为（40．0～750．0）U／L的50份不同浓度的人血清组成患者冷冻血清样本组，优选瑞士Roche公司、日本和光公司、深圳迈瑞公司、四川迈克公司4种AMY厂家系统，分别采用国际临床化学和实验室医学联和会（IFCC）参考方法和基于7170全自动生化分析仪的4种AMY厂家系统同时测量浓度为（70．1±3．7）U／L和（418．3±22．1）U／L的具有标准物质特性的冷冻血清和患者冷冻血清样本组。用Excel软件进行各厂家系统与IFCC参考方法测量结果的直线回归分析。采用MVS1．380软件的EP9-A2文件方法、Bland—Altman图形分析法和改良Bland-Altman图形分析法分别进行各厂家系统与IFCC参考方法测量结果的等价性判定、一致性评价和各厂家系统测量结果的正确性评定。结果以IFCC参考方法测量结果做评价标准，Roche、和光、迈瑞、迈克公司厂家系统测量结果存在明显差异：（1）浓度为（70．1±3．7）U／L和（418．3±22．1）U／L的具有标准物质特性的冷冻血清各厂家系统测量结果分别为66．4、70．6、69．4、49．2U／L和394．0、456．4、406．7、302．4U／L。迈瑞公司系统结果与IFCC参考方法一致。（2）患者冷冻血清样本组的各厂家系统与IFCC参考方法测量结果的直线回归方程斜率分别为：0．934、1．070、0．930和0．731，截距分别为：0．886、6．249、5．388和3．574。按EP9-A2文件方法评价，罗氏和光、迈瑞公司系统测量结果与IFCC参考方法等价。按Bland-Altman图形分析法评价，Roche和迈瑞公司系统测量结果与IFCC参考方法测量结果一致。各厂家系统平均偏倚分别为-6．11％（“平均偏倚±2s”分别为-2．81％和-9．40％）、1．99％（“平均偏倚4-2s”分别为10．35％和-6．36％）、-2．70％（“平均偏倚4-2s”分别为2．37％和-7．77％）和-34．72％（“平均偏倚4-2s”分别为-24．20％和-45．24％）。按改良Bland—Ahman图形分析法评价迈瑞公司系统测量结果正确。各厂家系统平均偏倚分别为-5．92％（“平均偏倚4-2s”分别为-2．81％和-9．03％）、2．10％（“平均偏倚±2s”分别为10．74％和-6．53％）、-2．64％（“平均偏倚4-2s”分别为2．24％和-7．51％）和-29．51％（“平均偏倚±2s”分别为-21．82％和-37．21％）。结论（1）不同厂家系统测量结果不一定都具有正确性，尽管各厂家系统测量结果都已声称具有溯源性。（2）当采用不同方法评价测量系统正确性时，同一厂家系统测量结果评价结论可能不一致。实验室应选择并建立常规方法测量结果正确性评价方法，选择并使用符合ISO15189正确性要求的常规方法。

Objective To investigate the routine methods of or-amylase （AMY） test in serum which meets the requirements of ISO 15189. Methods Fifty human serum samples with different concentrations of AMY （40 -750 U/L） were collected from March to December in 2008, to form the patients＇ frozen serum group. Four AMY measurement systems including Roche, Wako, MINDRAY and MAKER were used. The frozen standard materials with concentrations of （70. 1 ± 3.7 ） U/L and （418.3 ± 22. 1 ） U/L and the patients＇ frozen serum group were measured simultaneously by using IFCC reference method and 4 AMY measurement systems based on 7170A automatic biochemistry analyzer. The linear regression analysis was made between the measurement results of each system and IFCC reference method. The equivalence, agreement and trueness were also evaluated by using the file EP9-A2 method , Bland-Altman Graphical- Analysis and the improved Bland-Altman Graphical-Analysis of MVS1.80 software. Results Judging by the standards of IFCC reference method, the measurement results of 4 measurement systems were obviously different. （1） When measuring standard materials the results were 66.4,70. 6, 69.4 and 49.2 U/L respectively and 394. 0,456. 4,406. 7,302.4 U/L respectively. The measurement results of MINDRAY were in agreement with that of IFCC reference method. （2） When measuring the patients＇ serum group by 4 measurement systems and IFCC reference method, the slopes of the linear regression equations were 0. 934,1. 070,0. 930 and 0. 731, respectively. And the intercepts were 0. 886, 6. 249, 5. 388 and 3. 574, respectively. According to the EP9-A2 method, the measurement results of Roche Wako, MINDRAY were equivalent to that of IFCC reference method. According to Bland-Altman Graphical-Analysis, the measurement results of Roche and MINDRAY were in agreement with that of IFCC reference method. The average biases of each measurement system were - 6. 11% （ ＂ Average bias ± 2s＂ were 2. 81% and - 9.40% ）, 1.99% （ ＂Average bias ± 2s＂ were 10. 35% and - 6. 36% ）, - 2. 70% （ ＂Average bias ± 2s＂ were 2. 37% and -7.77% ） and -34. 72% （＂Average bias ±2s＂ were -24. 20% and -45.24% ） , respectively. According to the improved Bland-Ahman Graphical-Analysis, the measurement results of MINDRAY are correct. The average biases of each measurement system were - 5.92% （＂ Average bias _± 2s＂ were -2.81% and -9.03%）, 2. 10% （＂Average bias ±2s＂ were 10.74% and -6.53%）, - 2. 64% （ ＂Average bias _± 2s＂ were2. 24% and - 7.51% ） and - 29. 51% （ ＂Average bias ± 2s＂ were -21.82% and - 37.21% ）, respectively. Conclusions （ 1 ） The measurement results of different measurement systems do not necessarily have crreet results though they have claimed to have traceability. （2） The trueness of measurement results using the same system may not come to the same conclusion when evaluated by different methods. So laboratories should select and establish a procedure to evaluate trueness of routine methods and adopt those meeting the trueness requirements of ISO 15189.