[Recommendations for the standardization of growth hormone assays].

The diagnosis of growth hormone (GH) deficiency is based on the GH biological response to pharmacological stimulation tests. The cut-off value defining normality is the same whatever the GH assay used. In a group of the French Society for Clinical Biology (SFBC), we have evaluated whether differences between the GH concentrations obtained with the 9 commercial GH assays available in France exist or not. The study samples consisted of 72 serum pools and serial dilutions of the recombinant GH 22 kDa international standard, IS 98/574. These dilutions were performed by using 3 different diluents: the specific diluent provided by the manufacturers and thus different from one assay to another, serum without GH and heparin plasma without GH. Despite being calibrated against the same international standard, the different assays proposed variable conversion factors between microg and mIU, and we decided to express the results in mIU. The GH concentrations obtained for the 72 serum pools with the 9 assays were highly correlated, but absolute concentrations were significantly different from one assay to another. In particular, the ratio between the concentrations measured with both assays giving the lowest and highest concentration in the same sample respectively was about 50%. In the recovery test executed by adding the international standard, the slope of the regression curve describing the relationship between expected and measured concentrations was different of 1 in all but one assay. Furthermore, for a given assay and a given expected concentration, the measured values were sometimes different by up to 30% depending on the diluent used. These results led us to advise the manufacturers to calibrate their assays against the recombinant GH international standard, IS 98/574, to take into account the matrix effect detected in our study and to use the official conversion factor of 3 mIU/microg. Waiting for this new calibration, it is recommended that the results should be expressed in mIU/L and that serum samples should be used for the measurement of GH instead of plasma samples.