| ObjectiveWe design the study to evaluate the trueness of the routine systems (methods) to provide the scientific evidence for the effective use of the measurement results and for standardization, and to evaluate the commutability of processed materials and to select candidate reference materials for EQA scheme and standardization programs.MethodInductively coupled plasma mass spectrometry internal standard method was applied as comparative method,14 (or 12) the routine systems, composed of kits, its supporting calibrators and Hitachi 7180 automatic analyzer were chosen as test systems. 48 (or 45) single patients serum were collected as the samples of the comparison experiment, according to the EP9-A2. The 25 processed materials evaluated for commutability were:(1) Fresh frozen mixed human serum, prepared from leftover patient serum and then processed (2) the certified reference materials with serum matrix(3)animal serum(4) the conventional external quality assessment materials(5) aqueous reference materials, candidate certified reference materials from our laboratory(6)calibrators of routine methods.The results from the 48(or 45) patient samples were used to generate a scatter plot. The results from the comparative method were designated as the x-axis, and the results of each of the routine systems were designated as the y-axis, and used the least squares linear regression (OLR). The bias at the three medical decision levels were calculated, according to the regression equation. The three criterions used to evaluate the bias were the specifications derived from biological variation, the health industry standard of analytical quality specifications for routine analytes in clinical biochemistry (WS/T403-2012), and the half of allowable total error of CLI’A88.The 95% prediction intervals of each regression line were calculated, and the commutability of processed materials were evaluated by comparing the plots with the limits of the intervals. The results beyond the range showed matrix effect, within the range showed no matrix effect.ResultsFor calcium, the precision of all systems were well (total CV<2.26%) and the correlation coefficient was high(r>0.99). The 2013 EQA samples show negative matrix effect in some systems, and 2014 EQA samples show positive matrix effect in some systems. The human serum pools were well commutable and only show matrix effect in two assays. The calibrators showed matrix effect in some assays. The certified reference materials were commutable in most of the assays, only GBW09152 show matrix effect in one assays. The animal serum show matrix effect in some assays. The aqueous reference materials show matrix effect in most of the assays and can only use in the reference methods.For calcium, the range of mean bias for the 48 fresh frozen serum were -0.129 mmol/L--0.001 mmol/L (-5.61%～-0.01%). The range of expected bias at three medical decision levels(1.75mmol/L,2.74mmol/L,3.37mmol/L) were -0.108 mmol/L 0.053mmol/L (-6.16%～3.02%),-0.165 mmol/L ～- 0.001 mmol/L (-6.02%～-0.04%),-0.215mmol/L～0.067mmol/L (-6.38%～1.99%), respectively. Most routine systems can met the standard of the half allowable total error of CLI’A 88 (0.125mmol/L) at the three medical decision levels, nearly half of the systems met the standard of WS/T 403-2012(2%), only a few of the assays met can meet the minimum bias specifications derived from biological variation (1.3%).For Roche, Beijing Strong, and Mindray systems, the bias were small, the trueness of these system were well, and other systems still existed calibration bias.For magnesium, the precision of all systems were well (total CV<2.01%) and the correlation coefficient was high(r>0.99). The 2013 and 2014 EQA samples show positive matrix effect in some systems. The human serum pools were commutable in most of the systems, and only showed matrix effect in one system. For the three certified reference materials, only SRM956c2 were commutable in all the system, GBW09152 and BCR304 both showed matrix effect in two systems. The calibrators showed matrix effect in some assays. The animal serum showed matrix effect in some assays. The aqueous reference materials show matrix effect in most of the assays, and can only use in the reference methods.For all assays, the range of mean bias was-0.040 mmol/L-0.063 mmol/L (-4.52%～7.20%). The range of expected bias at three medical decision levels (0.6mmol/L,1.00 mmol/L,2.50 mmol/L) were -0.026 mmol/L～0.059mmol/L (-4.27%～9.30%),-0.048 mmol/L～0.063 mmol/L (-4.79%～6.31%),-0.341-0.101 mmol/L(-13.66%～4.05%), respectively. Most assays can meet the limits of WS/T 403-2012 (5.5%) at three medical decision levels, nearly half of the systems can meet the minimum performance standard based biologic variation (2.80%), only few assays can meet the appropriate performance standard based on biologic variation (1.80%). For Diasys, Wako and Mindray systems, the bias at three medical levels all can meet the limit of 2.80%, the bias were small, the trueness of these system were well, and other systems still existed calibration bias.ConclusionFor the measurement of serum calcium and magnesium, most of the system still existed calibration bias, the manufacturers should pay attention to the traceability of the assigned value of calibrators. Some processed materials showed matrix effect in routine system, we should the commutability of these materials, when used them.We describe a method for determination of serum calcium using isotope dilution inductively coupled plasma mass spectrometry (ID/ICP-MS), and confirm the reliability of the method. Then we compare the isotope dilution mass spectrometry with the internal standard method to evaluate whether the difference is existed between the two methods.Isotope dilution inductively coupled plasma mass spectrometry (ID/ICP-MS) method is a definitive approach with high accuracy and precision. The measurement of isotopic ratio was 44Ca/42Ca, with 42Ca as the enriched isotope. The sample preparation was simple, sample and 42Ca spike solution were weighed accurately, then digested with nitric acid. The octopole collision cell instrument utilized H2 as cell gas to remove polyatomic ion interferences, and the isotope ratio measurement repeatability of 44Ca/42Ca was better than 0.10%(n=10). The concentration of the 42Ca enriched isotope was calibrated by reverse isotope dilution method, with SRM3109a as the standard. The mass discrimination correction factor can be achieved by measurement of the calibration standard (SRM3019a) consisting of natural calcium. For each analytical measurement, samples and spike calibration solution were measured alternately, with spike calibration solution of known isotope ratio as the standard, the mass discrimination factor was obtained by analyzing the spike calibration solution which was measured before and after the sample in a bracketing procedure. Using this method, the uncertainty in the correction of mass bias and the instrument drift were minimized. The relative expanded uncertainty of the sample was 0.40%, k=2.The instrument blank was 0.01%～0.07% and the sample processed blank was 0.02%-0.12%, so the correction of blank was not needed.The precision of the method was well, the range of repeatability (within-run precision), the intermediate precision (between-run precision) and intra-laboratory precision were 0.12%～0.19%,0.07%～0.09%,0.16%-0.17%, respectively, for the determination of two concentration samples of 2014,2015 electrolyte trueness verification. SRM909b Ⅰ, SRM909bⅡ, SRM909c and GBW09152 were measured to verify the trueness of methods, all the results were within the certified values range. For the four reference materials, the mean relatively bias were 0.29%,-0.02%,0.10%,-0.19%, respectively; and the intra-laboratory CV were 0.18%,0.18%,0.11%,0.16%, respectively. The range of the recoveries was 99.87%-100.37%.According to EP9-A2, the 46 fresh serum samples were collected and detected by the isotope dilution mass spectrometry and the internal standard method. The results of isotope dilution method were as reference values, and the relative bias of the two methods were small. The range of relative bias was -0.45%～0.45%, and the positive and negative bias were homogeneous distribution, and the means bias of all samples were 0.01%, so the system error were not existed. The linear regression equation of the methods was Y=0.99885X+0.112255, and the comparability of the methods was very high. |
PDF Full Text Request