Evaluation of the comparability of commercial ghrelin assays.

To the Editor: Several assays are available to measure human ghrelin (h-GHS), but the comparability of results among assays is unknown. As an addition to previous investigations concerning the stability of h-GHS (1 ), it was our objective to compare the two most commonly used commercial h-GHS RIAs, focusing on side-by-side determinations, recoveries, and assay imprecision. The assays were purchased from Phoenix Pharmaceuticals (cat. no. RK-031-30) and Linco Research (cat. no. GHRT-89HK). Samples were stored at 80 °C and assayed as recommended by the manufacturers. For side-by-side comparisons, we used anonymized surplus sera (n 151) from clinical studies of lean and obese individuals [mean body mass index (BMI), 26 kg/m; range, 17–39 kg/m]. Separate side-by-side comparisons were performed with 68 samples from two individuals (34/ profile; BMI, 19.5 kg/m) taken at 20-min intervals to evaluate the suitability of the assays for determining intraindividual changes. All participants gave informed consent. For recovery and imprecision studies, serum was collected from four males (27–33 years of age; no drug use; normal blood pressure; BMI, 22–24 kg/m). Assay calibrators were added to achieve added concentrations of 100, 200, 400, or 800 ng/L. Each was measured 20 times. Recovery was calculated by subtracting the basal value of the sample. The CV was determined for each sample as well as for a Lyphochek control serum (cat. no. 40141; BioRad), which was also measured 20 times. The interassay CV was calculated from repeated measurements of the Lyphochek control (n 12) and pooled serum controls (n 12) by the Phoenix method over a longer period. No data for the interassay CV of the Linco assay are available because it had not been used previously in our laboratories. EVAPAK 3.0 evaluation software (Boehringer-Mannheim) was used for linear regression analysis (Passing–Bablok) (2 ). Each assay uses 100 L of sample. Linco provides ready-to-use calibrators; Phoenix provides a lyophilized preparation that must first be dissolved in assay buffer and then diluted repeatedly. Linco, in contrast to Phoenix, provides controls (two) per assay. Both assays have a 100% crossreactivity with h-des-octanoyl-GHS. Moreover, the Linco assay can be used for rodent and canine sera, whereas the specificity of the Phoenix antibody surprisingly changed between January 2003 (lot 500117) and August 2002 (lot 419523) to 1% cross-reactivity with non-human GHS. Linco gives the concentrations in ng/L, whereas Phoenix gives concentrations in pg/tube, presented in a general protocol for peptide RIAs. The Phoenix calibration curve includes concentrations of 10–1280 ng/L. The calibrators of the Linco assay range from 100 to 10 000 ng/L. In each assay, a better curve fit was obtained when the two lowest calibrators were excluded. The side-by-side comparison yielded linear regressions of r 0.976 (n 151) and r 0.890 (intraindividual profiles). The measured values obtained with the Phoenix RIA ranged from 10 to 1240 ng/L (median, 340 ng/L), whereas the measured values obtained with the Linco assay were 10-fold higher (131–11 666 ng/L; median, 3282 ng/ L). The regression is shown in Fig. 1. The 10-fold difference was observed neither when the calibrator of the Linco assay was replaced by the calibrator of the Phoenix assay nor vice versa. Recovery of h-GHS was 54–104% [mean (SD), 75 (22)%] with the Linco assay and 67–123% with the Phoenix assay, with the higher recoveries at higher concentrations of added hGHS. Intraassay CV for the Phoenix method were 8–12% at the lower concentrations and 7–8% at higher concentrations. The CV of the Linco assay were 7–11% in the lower range and 5–9% in the upper concentration range. We consider intraassay CV 10%