Sensitivity and specificity of detection methods for erythropoietin doping in cyclists / Jules A.A.C. Heuberger, Peter van Eenoo, Joris I. Rotmans, Pim Gal, Frederik E. Stuurman, Titiaan E. Post, Johannes M.A. Daniels, Herman Ram, Olivier de Hon, Jacobus Burggraaf, Adam F. Cohen. - (Drug Testing and Analysis 11 (2019) 9 (September); p. 1290-1301).
- DOI: 10.1002/dta.2665.
- PMCID: PMC6790592.
- PMID: 31232530
Abstract
Recombinant human erythropoietin (rHuEPO) is used as doping a substance. Anti‐doping efforts include urine and blood testing and monitoring the athlete biological passport (ABP). As data on the performance of these methods are incomplete, this study aimed to evaluate the performance of two common urine assays and the ABP. In a randomized, double‐blinded, placebo‐controlled trial, 48 trained cyclists received a mean dose of 6000 IU rHuEPO (epoetin β) or placebo by weekly injection for eight weeks. Seven timed urine and blood samples were collected per subject. Urine samples were analyzed by sarcosyl‐PAGE and isoelectric focusing methods in the accredited DoCoLab in Ghent. A selection of samples, including any with false presumptive findings, underwent a second sarcosyl‐PAGE confirmation analysis. Hematological parameters were used to construct a module similar to the ABP and analyzed by two evaluators from an Athlete Passport Management Unit. Sensitivity of the sarcosyl‐PAGE and isoelectric focusing assays for the detection of erythropoietin abuse were 63.8% and 58.6%, respectively, with a false presumptive finding rate of 4.3% and 6%. None of the false presumptive findings tested positive in the confirmation analysis. Sensitivity was highest between 2 and 6 days after dosing, and dropped rapidly outside this window. Sensitivity of the ABP was 91.3%. Specificity of the urine assays was high; however, the detection window of rHuEPO was narrow, leading to questionable sensitivity. The ABP, integrating longitudinal data, is more sensitive, but there are still subjects that evade detection. Combining these methods might improve performance, but will not resolve all observed shortcomings.
