Citalopram, a new bicyclic antidepressant, is the most selective serotonin reuptake inhibitor. In a number of double-blind controlled studies, citalopram was compared to placebo and to known tricyclic antidepressants. These studies have shown their efficacy and good safety. The inefficacy of a psychotropic treatment in at least 20% of depressives has led a number of authors to propose original drug combinations and associations, like antidepressant/lithium (Li), antidepressant/sleep deprivation (agrypnia), antidepressant/ECT, or antidepressant/LT3. The aim of this investigation is to evaluate the clinical effectiveness and safety of a combined citalopram/lithium treatment in therapy-resistant patients, taking account of serotonergic functions, as tested by the fenfluramine/prolactin test, and of drug pharmacokinetics and pharmacogenetics of metabolism. DESIGN OF THE STUDY: A washout period of 3 days before initiating the treatment is included. After an open treatment phase of 28 days (D) with citalopram (20 mg D1-D3; 40 mg D4-D14; 40 or 60 mg D15-D28; concomitant medication allowed: chloral, chlorazepate), the nonresponding patients [less than 50% improvement in the total score on the 21 item-Hamilton Depression Rating Scale (HDRS)] are selected and treated with or without Li (randomized in double-blind conditions: citalopram/Li or citalopram/placebo) during the treatment (D29-D35). Thereafter, all patients included in the double-blind phase subsequently receive an open treatment with citalopram/Li for 7 days (D36-D42). The hypothesis of a relationship between serotoninergic functions in patients using the fenfluramine/prolactin test (D1) and the clinical response to citalopram (and Li) is assessed. Moreover, it is evaluated whether the pharmacogenetic status of the patients, as determined by the mephenytoin/dextromethorphan test (D0-D28), is related to the metabolism of fenfluramine and citalopram, and also to the clinical response. CLINICAL ASSESSMENT: Patients with a diagnosis of major depressive disorders according to DSM III are submitted to a clinical assessment of D1, D7, D14, D28, D35, D42: HDRS, CGI (clinical global impression), VAS (visual analog scales for self-rating of depression), HDRS (Hamilton depression rating scale, 21 items), UKU (side effects scale), and to clinical laboratory examens, as well as ECG, control of weight, pulse, blood pressure at D1, D28, D35. Fenfluramine/prolactin test: A butterfly needle is inserted in a forearm vein at 7 h 45 and is kept patent with liquemine. Samples for plasma prolactin, and d- and l-fenfluramine determinations are drawn at 8 h 15 (base line). Patients are given 60 mg fenfluramine (as a racemate) at 8 h 30. Kinetic points are determined at 9 h 30, 10 h 30, 11 h 30, 12 h 30, 13 h 30. Plasma levels of d- and l-fenfluramine are determined by gas chromatography and prolactin by IRNA. Mephenytoin/dextromethorphan test: Patients empty their bladders before the test; they are then given 25 mg dextropethorphan and 100 mg mephenytoin (as a racemate) at 8 h 00. They collect all urines during the following 8 hours. The metabolic ratio is determined by gas chromatography (metabolic ratio dextromethorphan/dextrorphan greater than 0.3 = PM (poor metabolizer); mephenytoin/4-OH-mephenytoin greater than 5.6, or mephenytoin S/R greater than 0.8 = PM). Citalopram plasma levels: Plasma levels of citalopram, desmethylcitalopram and didesmethylcitalopram are determined by gas chromatography--mass spectrometry. RESULTS OF THE PILOT STUDY. The investigation has been preceded by a pilot study including 14 patients, using the abovementioned protocol, except that all nonresponders were medicated with citalopram/Li on D28 to D42. The mean total score (n = 14) on the 21 item Hamilton scale was significantly reduced after the treatment, ie from 26.93 +/- 5.80 on D1 to 8.57 +/- 6.90 on D35 (p less than 0.001). A similar patCitalopram, a new bicyclic antidepressant, is the most selective serotonin reu