RESUMO
OBJECTIVE: Because over 90% of serum cortisol is bound to albumin and corticosteroid-binding globulin (CBG), changes in these proteins can affect measures of serum total cortisol levels in cirrhotics without altering serum-free and salivary cortisol concentrations. METHODS: We assessed basal (T0) and post-synacthen (T60) serum total cortisol, serum-free and salivary cortisol in 125 consecutive cirrhotics (95 non-septic and 30 septic patients with a Child>8). RESULTS: Serum total cortisol levels significantly decreased from the Child A-C non-septic group, as did albumin and CBG levels, with a non-significant rise in serum-free cortisol concentrations. Non-septic patients with low albumin (≤25 g/L) or CBG levels (≤35 mg/L) had lower T0 serum total cortisol levels than patients with near-normal albumin (303.4 vs. 382.6 nmol/L; P=0.0035) or with normal CBG levels (289.9 vs. 441.4 nmol/L; P<0.0001), respectively, despite similar serum-free cortisol or salivary cortisol concentrations. Subnormal T60 serum total cortisol concentrations (<510.4 nmol/L) were measured in 7.2% of all patients (Child C: 14.5% vs. Child A and B: 0%; P=0.0013) but no patients exhibited symptoms suggesting adrenal insufficiency. Patients with or without subnormal T60 total cortisol had similar T0 salivary cortisol and serum-free cortisol concentrations. A trend was observed towards high serum-free cortisol concentrations and mortality in multivariate analysis. CONCLUSIONS: Serum total cortisol levels overestimated the prevalence of adrenal dysfunction in cirrhotics with end-stage liver disease. Since serum-free cortisol cannot be measured routinely, salivary cortisol testing could represent a useful approach but needs to be standardized.