Influence of matrix on concentrations of somatotropin measured in serum with commercial immunoradiometric assays.

Using immunoradiometric assays (IRMAs) from Hybritech Inc. (H) and Nichols Institute Diagnostics (ND), we measured somatotropin (human growth hormone, hGH) in serum samples obtained every 20 min for 24 h from 10 prepubertal subjects with short stature. Results obtained with the ND reagents were 2.74 times greater than those obtained with the H reagents (P = 0.00001, r = 0.94, SEE = 3.9, n = 720). We therefore compared the IRMAs with the standard hGH RIA from the National Institutes of Health (NIH) National Hormone and Pituitary Program, using the genetically engineered hGH preparations (from Genentech Inc.) 22-kDa hGH and methionated 20-kDa hGH. We also assayed human pituitary hGH (NIH, lot no. AFP-4793B). Each hGH preparation was diluted in three diluent buffer systems: horse serum from H and from ND, and human serum. The RIA and H-IRMA gave superimposable standard curves for all hGH preparations in each diluent. The methionated 20-kDa hGH was not detected in the H-assay. Use of human serum matrix in the ND-IRMA shifted the standard curve as compared with the horse-serum matrix, giving equivalent binding at lower concentrations; i.e., serum hGH was overestimated in samples assayed against standards diluted in horse serum. Quality-control materials (Ciba-Corning) yielded disparate results in all three assays, yet human serum pools containing hGH gave similar results in the H and the NIH assays, and higher values in ND. When a human serum standard was used in the ND assay, both IRMAs gave similar results to the RIA assay for human serum samples. Reference intervals for hGH should be determined by each analytical laboratory, to prevent misdiagnosis of patients. Furthermore, quality-control material should be of human origin, because commercially supplied quality-control material does not react the same as human serum in some hGH assays.