Tolerance in Antidepressant Treatment

Despite compelling evidence of rapid relapse into depression within 1–3 months after discontinuing long-term antidepressant treatment, and a strong relationship of this risk to prior illness history, the place for maintenance antidepressant treatment in recurrent major depression and other disorders remains uncertain [1, 2]. Most long-term therapeutic studies in major depression involve tricyclic antidepressants; they average 17 months in duration [1, 2]; only one small trial of imipramine reached 5 years [3]. The therapeutic model involved derives from evidence of protective effects of long-term lithium treatment in unipolar as well as bipolar major affective disorders [4, 5]. Evidence of long-term effectiveness and safety for antidepressant treatment in bipolar disorders is lacking [2, 6]. There is also growing evidence that a loss of benefit can emerge with long-term antidepressant treatment or on retreatment after discontinuation [7, 8, 9, 10], perhaps even more commonly among bipolar than unipolar depressed patients [11].In this issue, Schmidt et al. [12]and Fava et al. [13]provide relevant observations from their long-term collaborative study of outpatients with moderately severe, recurrent unipolar depression who relapsed during long-term treatment with (±)-fluoxetine. They extend previous observations that increasing fluoxetine doses from 20 to 40 mg/day can rescue some late failures [14], as did increasing the dose of enteric-coated fluoxetine tablets from 90 to 180 mg/week [12](table 1). Underlying mechanisms and implications of this phenomenon remain unclear, but invite comment.After responding to an open trial of fluoxetine at 20 mg/day, 31.4% (119/379) of depressives relapsed within an average of 107 days of blinded fluoxetine maintenance treatment, in which daily dosing was somewhat more effective than weekly (25.9 vs. 36.8% relapse rate; χ2 = 5.24, p = 0.022) [15]. Among 56 subjects relapsing at 20 mg/day, 57.1% responded to a blindly determined doubled dose of fluoxetine for another 6–24 weeks, as did 71.9% of 64 patients who relapsed on 90 mg/week of enteric-coated drug. Therefore, 35.0% (42/120) of patients relapsing during long-term fluoxetine treatment were not rescued by doubling the dose. These treatment-resistant subjects represent 11.0% (0.314 × 0.350) of all subjects starting long-term fluoxetine treatment.As expected, more patients relapsed during blinded maintenance with a placebo (61/122 = 50.0%) than with fluoxetine [119/379 = 31.4%; χ2 (1 d.f.) = 13.9, p = 0.0002], and with a shorter average latency (2.87 vs. 3.57 months) [12]. The larger relapse risk with placebo suggests that a loss of placebo response did not account for relapses during fluoxetine maintenance. Of those maintained on placebo after blinded withdrawal from 13 weeks of open treatment with fluoxetine, 62.3% (33/53) regained and maintained response to 20 mg/day on retreatment [13].Response during secondary trials of fluoxetine was unexpectedly greater than in primary trials [431/932 = 46.2% vs. 42/120 = 35.0% of subjects; χ2 (1 d.f.) = 5.43, p = 0.020]. This finding does not indicate any loss of response to retreatment. However, sampling may be biased by including all eligible subjects initially, but retreating only initially fluoxetine-responsive patients. Relative risk of unresponsiveness to retreatment with other antidepressants remains inadequately studied [7, 8, 9, 10, 11], but preliminary findings suggest variance among specific agents [11]. Similarly, responses to retreatment after failing lithium among bipolar disorder patients have been inconsistent [16, 17].It is reassuring that increasing the dose of fluoxetine led to sustained secondary responses in a majority (78/120 = 65.0%) of relapsed depressive patients, though only with limited follow-up (6–24 weeks) [12]. More concerning, however, is that 35% of patients relapsing during maintenance fluoxetine treatment did not respond to an increased dose, and that 11% of those started on long-term fluoxetine treatment became treatment resistant within only about 3.6 months of maintenance treatment [12, 15].Loss of previous antidepressant benefit over time, with incomplete return of response even after doubling the drug dose, are consistent with pharmacological tolerance. Underlying mechanisms are unknown, but induced metabolic clearance of antidepressants is readily tested with assays of blood drug concentrations over time. It also remains to be proved, as we suspect [11], that antidepressants vary in risk of therapeutic tolerance, and whether longer antidepressant treatment, or higher maintenance doses might further increase the risk of tolerance.That dosing may be significant is suggested by the finding [12]that response to increased dosing was nearly significantly greater among those maintained with the lesser dose of 90 mg/week (equivalent to 12.9 mg/day) vs. 20 mg/day of fluoxetine [respective secondary failure rates: 18/64 = 28.1% vs. 24/56 = 42.9%; χ2 (1 d.f.) = 2.85, p = 0.091]. Based on serum concentrations of fluoxetine, 90 mg/week (12.9 mg/day) delivers less dose-by-time than 20 mg/day – the equivalent of about 15 mg/day [12, 15]. Maintenance on placebo led to an intermediate failure rate on retreatment at 20 mg/day (20/53 = 37.3%), perhaps reflecting confounding effects of prior (zero) and current (relatively low) doses.An additional concern is that, among 28 long-term antidepressant trials including the fluoxetine study [1, 15], risk of postdiscontinuation relapse was nearly significantly greater after longer treatment following recovery from an index episode of major depression (Spearman rs = +0.367, p = 0.052). This relationship suggests that pharmacodynamic dependence may develop as treatment is continued, with an associated risk of withdrawal-emergent depression, including after reassignment to placebo. Excess risk of early relapse, especially after rapid drug-withdrawal, is better documented with benzodiazepines, antipsychotics, and lithium [1, 18, 19]. This phenomenon raises clinical and ethical concerns, and can confound interpretation of research findings. This possibility for antidepressants could be clarified by directly comparing the impact of discontinuing long-term antidepressant treatment at controlled rates.The reports by Schmidt et al. [12]and Fava et al. [13]make an important contribution by addressing unresolved questions about the impact of long-term antidepressant treatment. As additional information is gathered about the value and risks of long-term antidepressant treatment in various disorders, some tentative clinical recommendations can be offered. First, reemergence of illness in treated depressed patients calls for reconsideration of differential diagnoses, particularly bipolar disorder [2, 20]. Second, clinical prudence encourages use of the smallest amount of medicine that seems both effective and well-tolerated for long-term use in patients with recurrent or chronic forms of antidepressant-responsive illness. Third, changes in any psychotropic medication should be made gradually, and rapid discontinuation avoided, whenever possible.