Plasma leptin measurements in epidemiological investigation: comparison of two commonly used assays and estimate of regression dilution bias.

As leptin is the object of intensive clinical research, we compared the radio-immunological assay (RIA) and enzyme-linked immunosorbent assay (ELISA) commercially available for measuring its plasma concentration in humans (Study 1), and sought to determine the power of a single plasma leptin measurement to characterise adequately a subject within a population on the basis of its intra- and inter-individual variations (Study 2).Study 1--Plasma leptin concentrations were determined by means of RIA and ELISA in a sample of 80 males. The measurements obtained using the two methods were closely correlated (r = 0.942), but the bias of the means was 21.1 +/- 73.5\% (M +/- SD, p < 0.001) and indicated that the two assays were not in agreement with each other. As expected, there were strong statistical associations between plasma leptin and a number of anthropometric indices, but the slopes of the regression of leptin concentration was significantly steeper when measured by ELISA. Study 2--ELISA was used to measure plasma leptin concentrations in three different samples obtained from 12 males and 12 females at two-week intervals. The inter-individual variation in plasma leptin was much greater than its intra-individual variation (the ratio of intra-to inter-individual variance = 0.05 and 0.04 in males and females, respectively), thus suggesting that a single fasting measurement is sufficient to characterise an individual's plasma leptin level within a population.ELISA is at least as effective as RIA in measuring plasma leptin, and is fully suitable for epidemiological investigations. A single measurement made in the morning and under fasting conditions is sufficient to characterise an individual within a population.