Anthropometric factors have significant influence on the outcome of the GHRH–arginine test: establishment of normative data for an automated immunoassay specifically measuring 22 kDa human growth hormone

Abstract

Context Adult growth hormone (GH) deficiency (GHD) is diagnosed by provocative testing of GH secretion. Objective To improve the diagnostic accuracy of GH-releasing hormone (GHRH) plus arginine (GARG) testing, we evaluated the influence of age, BMI and sex and established normative data for an automatic immunoassay specifically measuring 22 kDa human GH. Design/setting Prospective multicenter study. Participants Eighty-seven patients with hypothalamic–pituitary disease and 200 healthy controls. Patients were classified according to the number of pituitary hormone deficiencies (PHD). GHD was assumed when ≥2 PHD (in addition to GH) were present (n = 51); 36 patients with <2 PHD were considered GH sufficient (GHS). ROC analysis identified cutoffs with ≥95% specificity for GHD. Controls were prospectively stratified for sex, age and BMI. Interventions All participants received GHRH and l-arginine. Main outcome measures GH was measured by immunoassay (iSYS, IDS). Results In controls, multiple stepwise regression analysis showed that BMI (21%, P < 0.0001), sex (20%, P < 0.0001) and age (5%, P < 0.001), accounted for 46% of GH peak level variability during GARG. Comparison of peak GH during GARG (GHD vs GHS + controls) revealed an overall cutoff of 3.9 ng/mL (sensitivity 86%, specificity 95%). After adjustment for BMI and sex, optimal cutoffs (male vs female) were 6.5 vs 9.7 ng/mL in lean, 3.5 vs 8.5 ng/mL in overweight and 2.2 vs 4.4 ng/mL in obese subjects respectively. Conclusion BMI and sex account for most of the variability of peak GH levels during GARG. Consequently, diagnostic accuracy of the GARG test is significantly improved by use of adjusted cutoffs.