Lack of specificity of cortisol determinations in human urine.

The specificity of 7 different binding proteins (4 antibodies and 3 transins) was investigated in human urine. Pregnant and nonpregnant urine samples were extracted and chromatographed, and values were compared before and after chromatography. Without chromatography, all methods grossly overestimated the amount of cortisol present. Four methods gave higher values than the other 3 even after chromatography, possibly due to cross-reactivity with 20-dihydrocortisone which coelutes in part with cortisol. Interference also occurred in both the less polar and the more polar regions of the chromatograms with all assay systems. Values for a series of 20 urines were closely similar for the RTAs but widely divergent for the RIAs. Close correlations were found for all of the RTAs with each other and with the RIAs if simple methylene chloride extraction was used. A high degree of correlation was also found between extracted and unextracted urine values in the 4 systems studied. Cortisol values by RTA (dog transcortin) after chromatography on Sephadex LH-20 gave mean values of 20.5 +/- 9.7 microgram/day in men, 14.0 +/- 5.7 micrograms/day in cycling women, and 38.0 +/- 24.5 micrograms/day in women in late pregnancy (n = 6 in each group). It is concluded that there is no simple practical method currently available for true cortisol in urine, but that the measurement of adrenal corticoids in urine can afford an accurate reflection of adrenocortical function provided there is no gross metabolic abnormality present and that the normal range is carefully established for each particular method used.