Introduction: It has been recognised in EQA schemes in Europe and America that the reproducibility between labs using LCMS for testosterone analysis is not optimal for this technique. We decided to conduct a calibration exercise to investigate the variability seen between labs.
Methods: Aqueous and matrix matched serum samples were sent to labs participating in the NEQAS testosterone scheme. The labs were asked to measure these samples blind using their routine assays and their own calibration material. We then re-calculated the results for these samples using assigned values from our routine assay which has been shown to align closely to a reference method. Most labs used liquidliquid extraction to prepare samples but a variety of different LC columns and calibration matrices were used.
Results: Re-calculating the results from serum samples using matched serum calibrators improved the inter laboratory precision (CV) from 10% down to 5%, likewise re calculating the results from aqueous samples using aqueous calibrators improved the inter laboratory precision from 7% down to 3%. Trying to re calculate the serum results using aqueous calibrators actually made the inter laboratory variability worse. This suggests that the matrix in which the calibrator is made will affect the result significantly.
Conclusion: All laboratories gave clinically acceptable results but the accuracy of the results and hence the variability between laboratories was improved if common calibration material was used. If labs are to use a variety of different LC columns it is important that they thoroughly evaluate the effects of ion suppression, caused by sample matrix, which can vary profoundly with LC columns from different suppliers. The choice of column will also have a bearing on the type of calibrator matrix that is suitable for use in the assay.