BackgroundIncreased cortisol levels and genetic polymorphisms have been related to both major depressive disorder and antidepressant treatment outcome. The aim of this study is to evaluate the relationship between circadian salivary cortisol levels, cortisol suppression by dexamethasone and genetic polymorphisms in some HPA axis-related genes to the response to placebo and fluoxetine in depressed patients.MethodsThe diagnosis and severity of depression were performed using the Mini International Neuropsychiatric Interview (M.I.N.I.) and Hamilton depression scale (HAM-D17), respectively. Euthyroid patients were treated with placebo (one week) followed by fluoxetine (20 mg) (two months). Severity of depression was re-evaluated after placebo, three weeks and two months of fluoxetine treatments. Placebo response was defined as HAM-D17 score reductions of at least 25% and to < 15. Early response and response were reductions of at least 50% after three weeks and two months, and remission with ≤ 7 after two months. Plasma TSH, free-T4, circadian salivary cortisol levels and cortisol suppression by dexamethasone were evaluated. Seven genetic polymorphisms located in the Corticotrophin-releasing-hormone-receptor-1 (rs242939, rs242941, rs1876828), Corticotrophin-releasing-hormone-receptor-2 (rs2270007), Glucocorticoid-receptor (rs41423247), FK506-binding-protein-5 (rs1360780), and Arginine-vasopressin (rs3729965) genes were determined. Association analyses between response to placebo/fluoxetine and polymorphism were performed by chi-square or Fisher exact test. Cortisol levels were compared by t-test, ANOVA and the general linear model for repeated measures.Results208 depressed patients were recruited, 187 of whom were euthyroid. Placebo responders, fluoxetine responders and remitters exhibited significantly lower circadian cortisol levels than those who did not respond (p-values of 0.014, 0.008 and 0.021 respectively). Patients who abandoned treatment before the third week also exhibited a trend to low cortisol levels (p = 0.057). The polymorphisms rs242939 (CRHR1) and rs2270007 (CRHR2) were not in Hardy-Weinberg equilibrium. Only the rs242939 polymorphism (CRHR1) exhibited association with early response (three weeks) to fluoxetine (p-value = 0.043). No other association between outcomes and polymorphisms was observed.ConclusionsThese results support the clinical relevance of low salivary cortisol levels as a predictor of antidepressant response, either to placebo or to fluoxetine. Only one polymorphism in the CRHR1 gene was associated with the early response. Other factors may be involved in antidepressant response, although further studies are needed to identify them.Electronic supplementary materialThe online version of this article (doi:10.1186/s12888-014-0220-0) contains supplementary material, which is available to authorized users.