Direct enzyme-linked immunosorbent assay and a radioimmunoassay for melatonin compared.

A novel commercially available ELISA for direct measurement of melatonin concentration in serum was evaluated and compared with an RIA routinely used in our laboratory. The direct ELISA is technically simpler, requires a smaller sample volume (0.1 mL), and compares well with RIA in terms of stability of the calibration curve and intra- and interassay CVs. Correlation with RIA measurements is, however, suboptimal (y = 0.39x + 56; r = 0.65, P < 0.001; n = 138), which may be due to a serum effect, as evidenced by dilution studies. Furthermore, the detection range of the ELISA does not cover the physiological daytime melatonin concentrations in humans. Adding an extraction and 10-fold concentration step shifted the detection range of the ELISA to include low physiological concentrations as well. Correlation with RIA measurements also improved significantly (y = 0.97x-23; r = 0.95, P < 0.001; n = 105), probably due to removal of the serum effect. Although extraction increases the required sample volume (1.5 mL), work load, and procedure time, this step is necessary for the ELISA to compete successfully with RIA.