The choice of the optimal secondary adrenal insufficiency screening method

Background. The insulin tolerance test (ITT) is the “gold standard” of the secondary adrenal insufficiency (SAI) diagnosis but it is rather difficult to carry out, has some contraindications and requires patient’s hospitalization. The availability of a reliable screening method could reduce the necessity of using ITT.

Aim: to compare different methods of screening with ITT and to work out an optimal diagnostic algorithm of SAI.

Methods. 40 patients (20 women) after craniospinal (CSI) irradiation in a doze 35 Gy were examined. The average age at the time of the observation was 19,5±3 years, at the time of treatment 12,5±3,5 years. Patient’s blood samples were collected for basal cortisol (BC), DHEA-S. ITT was performed for all patients, glucagon stimulation test (GST) was for 27 persons. Patients were divided into groups: SAI and without SAI (W-SAI) after ITT. ROC- analysis was conducted to identify the thresholds for BC, DHEA-S and GST. Cut-off points for BC and DHEA-S levels corresponding to 100% sensitivity (Se) for SAI group and 100% specificity (Sp) for W-SAI patients were estimated to select a group of patients which do not require stimulative tests. Linear regression was used to construct a predictive model (PM) of SAI occurrence after CSI.

Results. 22/40 subjects failed ITT, 13/27 passed GST. 3 patients failed ITT but passed GST. Their level of neutrophils and monocytes was higher than the other patients'. SAI-patients had BC and DHEA-S lower than W-SAI (321±102 vs 516± 183; p=0,003 and 2,6±1,4 vs 5,1±2,1, p=0,003). ROC-analysis showed area under curve (AUC) for GST=0,91 with optimal cut-off for cortisol=489 which corresponds to 100% Sp and 62% Se. AUC was 0,83 for BC and 0,84 for DHEA-S.In 70% patients' BC was in a “grey zone” (32% of them passed ITT), 8% had BC lower than 200 (which corresponds to 100% Se) and 22% more than 499 (100% Sp). 50% of patients were in a “grey zone” for DHEA-S (50% of them had SAI), 18% had DHEA-S level below 2,0 and 32% above 4,7. A combination of BC and DHEA-S in the PM of SAI (0,592+0,001*BC+0,11*DHEA-S) had AUC 94%. This PM didn’t give the prognosis of SAI for 32% (CI: 18-49%) of patients. The addition of maximal cortisol (MC) level during GST (0,53+0,01*BC+0,066*DHEA-S+0,001*MC) increased AUC to 99% and didn’t allow to predict SAI in 8% (3,2% - 24,9%) of patients only.

Conclusions. When the screening methods were used separately, they showed comparable accuracy and it was not high. The PM may be used as optimal screening method for SAI and may allow to use ITT more rarely. But further studies are required to validate the PM proposed in this study.