Background Hormone measurements using automated immunoassays (IAs) are notorious for being affected by some factors that are present in serum (e.g. protein binding globulins, fatty acids). Different concentrations of these factors or a different composition of the serum is called an altered serum matrix. Liquid chromatography tandem-mass spectrometry (LC-MS/MS) is not influenced by matrix effects and can thus be used as reference method for hormone measurements. In clinical laboratories cortisol and testosterone are frequently measured using IAs. Therefore, the goal of this study is to unravel the extent of IA accuracy difficulties of cortisol and testosterone measurements in samples of hemodialysis patients, known to have a difficult matrix.

Methods Thirty samples from hemodialysis (HD) patients and 30 samples from healthy controls (HC) were used to measure cortisol and testosterone by an isotope dilution (ID)-LC-MS/MS method and by 4 commercially available automated immunoassays (Alinity (Abbott), Atellica (Siemens), Cobas (Roche) and Lumipulse (Fujirebio)). Method comparisons (Bland Altman plots and Correlation coefficients) were performed.

Results The cortisol immunoassays in HC deviated on average -8 to 4% from the LC-MS/MS method and 4% up to 49% in HD patients. The testosterone immunoassays in HC deviated on average -2 to 16% from the LC-MS/MS method and -6,7% up to 42% in HD patients. All correlation coefficients calculated based on the method comparison were significantly lower in the HD patients than in the HC group.

Discussion and conclusion Several automated immunoassays have great difficulty measuring cortisol and testosterone concentrations accurately in the altered serum matrix of hemodialysis patients. Physicians and laboratory specialists should be aware of these pitfalls in this specific patient group. It should be noted that these pitfalls might account for other patient groups characterized by an altered serum matrix as well.