Antidepressants, in the PL view, should not be used in bipolar illness. The only studies in which they have been compared to mood stabilizers find that mood stabilizers alone are as effective as antidepressants for the acute major depressive episode in bipolar disorder (reviewed here). Further, repeated studies show that antidepressants are ineffective in prevention of future depression (as meta-analyzed here). Thus, if after using one or two mood stabilizers, an antidepressant is added for an acute major depressive episode in bipolar disordre, it should be tapered off after improvement from the acute depression, in the view of PL, not longer than two months later. Indefinite maintenance of antidepressants in bipolar illness should be avoided.
Antidepressants Inefficacy and Harms
To summarize the above reviews, antidepressants are not effective in long-term prevention of bipolar disorder, and only marginally beneficial for short-term acute treatment. PL concludes that antidepressants should not be used long-term in most people with bipolar illness; even in the studies which support their use, only about 20% of patients were reported to have benefit at one year (as in one of the few long-term RCTs of modern antidepressants in bipolar illness). Thus 80% of patients should not receive antidepressants long-term.
Besides this lack of efficacy, antidepressants have two risks: causing acute mania, and causing long-term rapid cycling or mood destabilization. In terms of acute mania, this risk is minimized with concomitant mood stabilizer or antipyschotic treatment, and it is lower with certain antidepressants, as discussed below. The overall risk is abouot 10-50% depending on which antidepressant is used in bipolar disorder type I. The risk is somewhat lower in type II, and <1% in unipolar depression. Of the available agents, only paroxetine, sertraline, and bupropion have been shown to have a relatively low risk of causing acute mania, and therefore PL prefers to use those agents. Citalopram also appears to have a low risk. Venlafaxine (Effexor) has over twice the risk of inducing mania as those other agents, and should be avoided. TCAs are also high risk. It is reasonable to infer, based on the high mania switch rates wtih venlafaxine and TCAs, that noradrenergic (including Cymbalta and Strattera) or dopaminergic agents (including amphetamine stimulants) should not be used in bipolar illness. Mania induction is also very likely to be dose related, so the lower the dose the better: Pl recommends that about half the dose of antidepressants be used, if they are used at all, in bipolar depression compared with unipolar depression.
Antidepressants are mood destabilizers
A second risk with antidepressants is long-term mood destabilization, with more and more mood episodes (of both depression and mania) over time. This claim is based on two randomized studies of the topic. In the first study, with tricyclic antidepressants, about one-third of patients treated with antidepressants develop rapid-cycling bipolar disorder, based on two randomized studies of the topic. In the second study, with SRIs and bupropion mainly, continuation of antidepressants led to about two times more depressive episodes than stopping antidepressants. Both studies show that antidepressant discontinuation was effective in slowing down rapid cycling. In fact, this is the only proven intervention for rapid cycling.
It is worth noting that opponents of this view often cite non-randomized studies to claim that antidepressants do not cause or worsen rapid-cycling. Supporters of this view fail to make a basic scientific distinction: Randomized data are more valid than non-randomized data – a basic concept of evidence-based medicine. For that reason, the two randomized studies cited are the most valid evidence on this topic, and are not refuted by non-randomized reports to the contrary.
Antidepressants are contraindicated in rapid-cycling bipolar illness
Regarding rapid-cycling bipolar illness, then a few important points remain to be clarified:
Lithium is proven ineffective in multiple RCTs, as is lamictal in two RCTs where it was equivalent to placebo for rapid cycling. And, contrary to popular belief, anticonvulsants have never been shown to be more effective than lithium for rapid cycling. Valproate (Depakote) is only slightly more effective than lithium for rapid cycling according to the only randomized comparison. Rapid cycling is a severe form of bipolar disorder, with less treatment response than non-rapid cycling illness. No single agent is effective, and the only proven helpful intervention is to stop antidepressants. Beyond that, PL recommends that clinicians combine the use of multiple mood stabilizers, often with dopamine blockers (antipsychotics), in rapid cycling – but the key issue is to use these combinations always without antidepressants. This is the one area of treatment where PL feels a clear statement can be made, based on the available randomized data, that antidepressants are absolutely contraindicated.
Clinical Pearl: Therapeutic trials in bipolar illness have to occur without antidepressants
In sum, antidepressants can act as mood destabilizers, counteracting the benefits of mood stabilizers. Thus, no patient has received a therapeutic trial of a mood stabilizer, unless it occurs in the absence of antidepressants. Frequently, past trials of mood stabilizers “failed” with concomitant antidepressant use, but later the same mood stabilizers will have efficacy in the absence of antidepressant use.