Treatment-Resistant Unipolar Depression: Antidepressant Treatment Options
If a depressed patient does not respond to an antidepressant after an optimized medication trial, the physician can choose one of four treatment strategies: augment the failed antidepressant, use a higher dose of the antidepressant, combine two classes of antidepressants, or switch to another antidepressant. The choice between augmentation (the addition to ongoing antidepressant treatment of one or more pharmacological agents that alone do not have established antidepressant efficacy but that enhance antidepressant response or short response latency), higher dose, combination (the concomitant use of more than one antidepressant agent), and switching strategies is determined by the nature of the patient’s response to the failed antidepressant, willingness to take more than one medication, and ability to withstand potential adverse drug reactions that might arise from those strategies. If the initial drug trial has resulted in a partial response (defined as 25%-49% reduction in baseline symptom severity) without adverse drug reactions, then either augmentation or higher dose is recommended. However, if response with or without residual symptoms is accompanied by extremely bothersome side effects, then switching would be a better strategy. If the failed drug has caused no improvement whatsoever (<25% reduction in baseline symptom severity), switching might be reasonable, but no data are available to show that switching is superior to augmentation or combination in this situation.

In the following subsections we review the evidence for augmentation strategies, higher-dose antidepressants, combination strategies, and switching.

Augmentation Strategies
Lithium augmentation  Lithium augmentation is a commonly used strategy to manage treatment-resistant depression. Psychiatrists with experience prescribing lithium for bipolar disorder since the early 1970s are familiar with the possible adverse effects, blood levels, and possible long-term effects, such as hypothyroidism and nephrotoxicity. Although isolated case reports on the efficacy of lithium augmentation of antidepressants for nonbipolar depression were published in the 1960s and 1970s, wider clinical use followed a report by de Montigny and colleagues. Eight of eight depressed antidepressant nonresponders who had undergone at least 3 weeks of treatment with imipramine, amitriptyline, or doxepin responded dramatically within 48 hours of the addition of lithium. Several uncontrolled case series followed that supported the efficacy of lithium augmentation, but the quality and time course of the response were more variable than in the original report. Twelve double-blind controlled trials have since been published. A recent meta-analysis of nine placebo-controlled studies of lithium augmentation found that the pooled odds ratio of response during lithium augmentation compared with the response during placebo treatment was 3.31 (95% confidence interval = 1.46-7.53). The probability that a group of patients will respond to lithium augmentation is predicted at 30% for complete remission and another 25%-35% for substantial improvement but without a full return to euthymia.

How much lithium should be given, when, and for how long?  In contrast to the use of lithium for bipolar disorder, few studies have explored the relationship between lithium dose (or blood levels) and response to lithium augmentation. In a comparison of placebo with 250 mg lithium and 750 mg lithium daily in a double-blind study of 34 depressed patients, 250 mg lithium daily was not significantly better than placebo, but 750 mg lithium was superior to both 250 mg lithium and placebo. At least a 50% improvement on the Hamilton Rating Scale for Depression was observed in 18% (6 of 34) of patients taking 250 mg lithium, and 44% (15 of 34) of patients taking 750 mg lithium daily. In contrast, Thase et al. (1989a) found that no relationship existed between lithium dose (or blood levels) and response to lithium augmentation in 20 patients, and similar results were obtained by Dinan (1993) in a smaller sample of patients. The relevance of lithium dose and blood levels for response to lithium augmentation merits further study, but a target plasma level above 0.4 mEq/L is recommended.

Clinicians must decide when to add lithium to the ongoing but failed antidepressant trial. Some investigators have reported adding lithium to an antidepressant after 12 weeks of antidepressant nonresponse. In actual clinical practice, however, most clinicians would choose to intervene as early as during the fourth week of treatment for patients who have no sign of improvement. In this case, clinical judgment may be a better guide than the available literature for choosing the timing of adding lithium. The reports of latency to response for lithium augmentation of an antidepressant range from 2 days to 6 weeks. Although some patients will respond within 48-72 hours, many patients will experience a greater delay of onset; thus, the best course for patients who require lithium augmentation would be a trial of lithium for between 3 and 6 weeks before abandoning this strategy due to nonresponse.

Little is known about the long-term benefits of successful lithium augmentation (e.g., how many patients maintain the response or experience relapses or recurrence over time). Patients who respond to lithium augmentation tend to stop the lithium within 6 months to 8 years during naturalistic follow-up of responders from controlled and open studies. In one of the few studies available, good outcome with lithium augmentation was reported in 38 (72%) of 53 patients followed up naturalistically for 4-8 years. Good outcome was associated with a less endogenous nature of depression and an absence of hospitalizations.

Another study randomized 12 geriatric patients (10 women and 2 men; mean age 76.2 years) with unipolar depression who were receiving lithium augmentation to receive it or switch to placebo. Over the 2-year observation period, the placebo group reported a decrease in the composite side-effect score; two patients in each group had a recurrence of depression. This small study precludes definitive recommendations, but results suggest that it may be reasonable and safe to try to discontinue successful lithium augmentation after several months of complete remission.

Does efficacy vary with specific antidepressants and/or with patients’ characteristics?  Lithium has been studied as an augmenting agent most extensively with the classic TCAs. Only a few, mostly uncontrolled reports have focused on lithium augmentation of the newer generation of antidepressants (e.g., SSRIs, venlafaxine, or bupropion). Although the efficacy of lithium augmentation for those with nonresponse to TCAs as well as for those with response but who have residual symptoms is well established, the efficacy in SSRI nonresponders is not established. Lithium administration increases the function of serotonin neurons (de Montigny 1994). One may argue that such an effect may be somewhat redundant, perhaps increasing the risk of serotonergic side effects in the context of adequate SSRI dosing. In fact, Hawley et al. (1994) reported 14 cases of depression resistant to multiple treatments in which lithium augmentation of fluoxetine was poorly tolerated. One group found that although lithium augmentation appeared to be equally effective when added to either desipramine or fluoxetine, 33% of lithium augmentation responders who had not responded to fluoxetine alone had a relapse within 14 weeks, whereas none of the patients who had been taking desipramine relapsed (Ontiveros et al. 1991). More research is warranted on the course of response to lithium augmentation of different types of antidepressants. Joffe and colleagues (1993b) found lithium augmentation nonresponders to be more severely depressed, with more insomnia and weight loss, than responders. In addition, it appears that the rates of response to lithium augmentation in adolescent (Strober et al. 1992) and elderly (Flint and Rifat 1994) patients may be somewhat lower than those reported in other age groups. No other correlates of response have been identified to the best of our knowledge.

Conclusion  Although many patients can benefit from lithium augmentation, there is no clear dose-response relationship. However, if lithium augmentation is attempted, it seems reasonable to begin the lithium at 600 mg/day and expect a 4- to 6-week duration of the lithium trial before assuming lack of efficacy. The efficacy of this strategy with the SSRIs and the atypical antidepressants remains to be established.