同位素稀释质谱法与高效液相色谱法测定胆固醇的方法学比较：JCTLM列表中两个国际公认参考方法的比对研究

Comparison and bias estimation of isotope dilution mass spectrometry and HPLC for cholesterol： methodological comparison of two internationally accepted reference methods listed in JCTLM

目的对国际检验医学溯源联合委员会（JCTLM）收录的两种胆固醇参考方法进行方法学比对研究，评价两法测定结果的一致性和可比性。方法方法学比对研究。用同位素稀释液相色谱串联质谱法（LC/MS/MS法）和高效液相色谱法（HPLC法）同时测定52份新鲜冰冻人血清样本中的胆固醇，浓度范围为3 ～ 10 mmol/L。计算两法的精密度和准确度，对测定结果绘制散点图和偏差图，进行离群值检验，并选取最佳拟合模型进行回归分析，计算回归方程及各项参数。将两法测定结果进行统计学分析，计算胆固醇医学决定水平处的预期偏差及95%的可信范围。结果HPLC法仪器测定的CV为0.22%（0～0.49%），测定样本的总CV为0.36%（0.02%～0.83%），分析有证参考物质的平均偏倚－0.37%（－1.31%～0.14%）。LC/MS/MS法仪器测定的CV为0.50%（0～1.51%），测定样本的总CV为0.55%（0.03%～1.17%），分析有证参考物质的平均偏倚－0.24%（－0.53%～0.14%）。经统计分析两法测定结果未出现离群点，可采用线性回归方法进行分析。两法测定新鲜冰冻人血清样品的线性回归参数如下：斜率b＝0.989 5，斜率标准误Sb＝0.003 4，截距a＝0.063 4 mmol/L，截距标准误Sa＝0.019 9 mmol/L，Y估计值的标准误Sy/x＝0.034 8 mmol/L，相关系数r＝0.999 7，n＝52。同LC/MS/MS法相比，HPLC法测定新鲜血清样品的绝对偏差为0.000 mmol/L（－0.083～0.076 mmol/L），相对偏差为0.07%（－1.22%～1.24%）。t检验结果显示两个方法的测定结果没有统计学差异。在胆固醇医学决定水平处的预期偏差均在正负95%的可信区间范围内，且预期偏差远远小于允许误差。结论现有的HPLC法同权威的测量原理LC/MS/MS法测定血清总胆固醇的结果具有较好的一致性和可比性，且HPLC法比LC/MS/MS法价格低廉、技术简单、仪器精密度更高，故HPLC法可望在胆固醇的溯源过程中发挥重要作用。

ObjectiveTo evaluate the consistency and comparability of two kinds of cholesterol reference methods listed in the Joint Committee for Traceability in Laboratory Medicine （JCTLM）.Methods52 fresh frozen sera with cholesterol concentrations among 3-10 mmol/L were tested by isotope dilution liquid chromatography tandem mass spectrometry （ID-LC/MS/MS） and high performance liquid chromatography （HPLC）. The precision and accuracy of the two methods were calculated, and then a series of analysis were conducted including plotting scatter plots and deviation graphs, testing outliers, selecting the best fitting regression models and calculating the regression equations and parameters, and so on. The results of the two methods were statistically analyzed to estimate the expected deviation at the level of medical decision of cholesterol and the 95% confidence range.ResultsFor HPLC method, the CV of instrument measurement was 0.22% （0-0.49%）, the total CV of samples measurement was 0.36% （0.02%-0.83%）, and the average bias of the reference materials was 0.37% （-1.31%-0.14%）. For ID-LC/MS/MS method, the CV of instrument measurement was 0.50% （0-1.51%）, the total CV of samples measurement was 0.55% （0.03%-1.17%）, and the average bias of the reference materials was 0.24% （-0.53%-0.14%）. No outliers were found from the scatter plots and the statistical analysis and the linear regression were fitted to analyze the results of the two methods. The linear regression parameters of two methods for 52 fresh frozen human sera were as follows： the slope was 0.989 5, the standard error of slope was 0.003 4, the intercept was 0.063 4 mmol/L, the standard error of intercept was 0.019 9 mmol/L, the standard error of Y-estimate was 0.034 8 mmol/L, and the correlation coefficient was 0.999 7. Compared with the ID-LC/MS/MS method, the absolute deviation of fresh sera by HPLC method was 0.000 mmol/L （-0.083-0.076 mmol/L）, with a relative deviation of 0.07% （-1.22-1.24%）. T-test results showed no statistically significant difference between the two methods. The expected deviations at the cholesterol medicine decision level were within the range 95% confidence range, and the expected deviations were far less than the allowable error.ConclusionsThe HPLC method of cholesterol has good consistency and comparability with ID/MS method using the primary measurement principle. Because of more advantages of HPLC method such as less cost, more simple, requirement, and better precision, HPLC method is expected to play an important role in the process of standardization and traceability of serum cholesterol.