VASOPRESSIN TEST: DIAGNOSTIC INACCURACY IN EVALUATION OF HYPOTHALAMIC-PITUITARY-ADRENOCORTICAL AXIS

ABSTRACT An im vasopressin test was given to 141 children and adolescents, 52 with normal HPA axis and 89 with evident or suspected defect of the axis, and repeated in 36 cases, to establish criteria of a normal response, and to examine the accuracy and precision of the responses. Comparisons were made with the responses to the 2-h ACTH, insulin and 3-h metyrapone test. The distributions of plasma cortisol levels and increments were positively skew, and a log transformation was made for appropriate statistical analysis. Maximal plasma cortisol level was positively and maximal increment negatively correlated with the basal level. In precision, the maximal level was superior to the maximal increment. Hence, a normal result was best defined by an area around the regression of maximal level on basal level in the normal series. The best single index of the response was the maximal level. A useful new method was introduced for quantitative comparison of plasma cortisol responses to different tests. The vasopressin test result was frequently normal in patients who, according to repeated insulin tests were ACTH-deficient. Furthermore, the 10 patients with organic expansive hypothalamic lesions had a mean vasopressin response, that was greater relative to the insulin response than that of the reference series. However, 3 of 21 patients with organic non-expansive hypothalamic disease gave a subnormal vasopressin response but a normal insulin response. Moreover, in isolated GH deficiency and after prednisone medication the mean vasopressin response was lower relative to the insulin response than in the reference series. Thus, this test is not reliable in screening for, or in anatomical diagnoses of ACTH deficiency.