This section explores the literature on each type of antidepressant, mostly with regards to the risk of causing manic switch and efficacy of the antidepressant for bipolar depression.
We will be revisiting some of the literature mentioned above in our previous discussion of tricyclics. In one of the earlier retrospective chart reviews, switch rates were determined as: 22% of patients receiving ECT, 15% of patients receiving tricyclics, and 34% of patients who received no treatment at all on admission. This study concluded that there was no risk of switch with tricyclics. The general consensus throughout the ensuing 20 years of research on switch rates is that these data were skewed, not in the least because there was no randomized treatment, and no way to protect against selection bias.
Early studies compare the use of tricyclics as monotherapy in the treatment of bipolar depression (or as maintenance treatment). As discussed earlier, they often are studies of both bipolar and unipolar patients. A 2-year prospective, placebo-controlled, randomized trial comparing imipramine and lithium by Prien et al. in 1973 showed that lithium was superior to imipramine and placebo in preventing affective episodes. Though statistical analysis was not done in this study, manic episodes occurred in 12% of patients on lithium, 33% of patients on placebo, and 67% of patients on imipramine. These early placebo-controlled trials are good indicators of rates of mania in untreated patients.
A study of bipolar patients comparing monotherapy with lithium or imipramine or the combination found that over the 2-year maintenance phase, lithium was the superior treatment in preventing depressive or manic episodes. Fifty-three percent of patients in the imipramine monotherapy group had a manic episode, vs. 28% and 26% of patients in the combination group and lithium monotherapy group, respectively. Another prospective, randomized study found no statistical difference between depressive and manic relapses in patients given lithium versus patients given lithium and imipramine. This same group did a placebo-controlled analysis using a lithium/imipramine combination compared with lithium and placebo. There was no difference in the incidence of depressive relapses. Female patients on the lithium/imipramine combination were at a significantly higher risk of becoming manic.
Just as the recent study by Nemeroff and colleagues showed that there was no additional efficacy to adding antidepressants to lithium at lower serum lithium levels, an earlier chart review showed that at lower serum lithium levels, there is a significantly higher incidence of tricyclic-induced switch to mania.
Monoamine Oxidase Inhibitors
The use of monoamine oxidase inhibitors remains a viable treatment alternative in psychiatric practice for those patients who are treatment-refractory and can follow the dietary restrictions. There are few studies looking at the efficacy and side effects of these medications in bipolar patients. One group performed studies on MAOI’s ability to treat atypical, or anergic, depression in bipolar patients (motor retardation, fatigue, volitional inhibition, and reversed neurovegetative features). Both of these studies looked at antidepressants as monotherapy (again, limiting the results’ application in current clinical practice). The first study done by this group was a small study looking at both bipolar and unipolar patients. Tranylcypromine was compared with placebo, and was shown to be efficacious in treating depression. No switches were observed in this 6-week trial. A second study, looking solely at bipolar patients with anergic bipolar depression, compared tranylcypromine with imipramine (no placebo). There was a significantly greater response in the patients treated with tranylcypromine, and there were equivalent switch rates.
The only placebo-controlled, randomized, double-blind study looking at a reversible MAOI (moclobemide) compared with imipramine in bipolar depressed patients also has limitations. In the total of 156 patients, there was no statistical difference in efficacy between the two medications. More patients on imipramine became manic than patients on moclobemide, but the difference was not statistically significant. The shortcoming of this study is that some, but not all, of these patients were on mood stabilizers. These differences were not worked into the statistical analysis, and obviously would give more credence and applicability to the utility of these findings in today’s practice.
Despite the paucity of evidence of the utility and risks of MAOIs in the treatment of bipolar disorder, these medications remain a credible alternative to the first-line treatments of bipolar depression. The fact that these medications are still recommended, usually before the use of tricyclic antidepressants, in bipolar patients, should remind us that much of psychiatry’s clinical practice in treating bipolar patients is still not based on evidence.
Selective Serotonin Reuptake Inhibitors
Again, a relatively small amount of evidence exists on SSRIs in bipolar depression, particularly concerning the use of SSRIs in addition to mood stabilizer therapy. Some case reports of either bipolar II depression or non-bipolar depression show that a dosage decrease of an SSRI resolves the treatment-emergent hypomania/ mania. One case report of fluoxtine-induced mania at 10 mg/day showed that the mania resolved when the fluoxetine was decreased to 10 mg twice weekly.
A large meta-analysis of pooled data from pharmaceutical industry-sponsored studies looked at switch rates in both bipolar and unipolar patients. In the unipolar group, there were very low switch rates and no differences between SSRIs and tricyclic antidepressants. In the bipolar patient data, the rate of manic switch of patients on sertraline and paroxetine was no greater than placebo (3.7% vs. 4.2%), and was significantly lower than the switch rate on tricyclics (11.2%). This meta-analysis included no information on mood stabilizers, nor were there strict diagnostic criteria used for mania or hypomania.
A naturalistic study of 44 bipolar depressed patients attempted to look at risk factors for antidepressant-induced mania, and attempted to differentiate between patients on or off mood stabilizer therapy. One-quarter of the patients on SSRIs had switches to mania or hypomania, and no differences were found among types of SSRIs. There was a significantly lower incidence of mood switching on patients with lithium in comparison to patients receiving anticonvulsant mood stabilizer therapy or no mood stabilizer therapy.
A placebo-controlled, randomized study of unipolar vs. bipolar II patients looked at monotherapy of fluoxetine for various lengths of time (the longest interval being 1 year). Relapse rates to depression were similar across all groups. In the short-term treatment group, there was a 3.8% vs. 0.3% manic switch of bipolar vs. unipolar patients. In the long-term treatment group, there was a 3.6% vs. 0.8% manic switch. The conclusion of this study was that fluoxetine was safe and effective with a low rate of switch to mania; however, this study was limited to bipolar II patients.
In a class of its own, venlafaxine warrants its own discussion, and its own evidence. A number of case reports exist on venlafaxine-induced mania, and in a few cases, mania occurred on venlafaxine while it had not occurred while the patient was on an SSRI. One small non-placebo-controlled trial studied bipolar and unipolar patients being treated with venlafaxine. There was a significant reduction in depression in both groups and no switches to mania occurred over the 6-week trial. In an RCT comparing venlafaxine and paroxetine in bipolar I and II patients, no difference in efficacy was found, but a slightly higher switch rate was found in the venlafaxine group. Venlafaxine has the highest switch rate in the Stanley Network Foundation’s double-blind, randomized, placebo-controlled trial comparing sertraline, buproprion, and venlafaxine (R. Post, personal communication.
Bupropion is considered by some, including the guidelines discussed above, to be the first-line antidepressant for bipolar depression. This recommendation is supported by a number of case reports and studies, but careful investigation of the literature shows that bupropion is itself not without risks in a bipolar population. One report of three cases showed successful use of bupropion in “difficult-to-treat” bipolar depressed patients — it did not cause mania in these three patients, and their depression was put into remission. It is also potentially helpful for bipolar II rapid cycling patients. However, there do exist some case reports of bupropion-related mania, specifically at high doses (>450 mg/day).
A year-long prospective, naturalistic life-charting study looked at 69 patients who received 113 trials of antidepressants. Using the criteria for antidepressant-induced mania discussed above and defined by Altshuler and colleagues, the researchers found no difference in switch rates, nor cycle acceleration, in patients receiving SSRIs versus bupropion. This was a large trial, but was potentially subject to clinician bias as it was not blinded. Similarly, a small prospective trial consisting of the acute treatment of 11 depressed bipolar patients on baseline mood stabilizer therapy showed that 7 out of 11 patients responded to the bupropion within 6 weeks. However, 6 out of 11 of the patients overall (four of the non-responders) experienced hypomania or manic symptoms necessitating the discontinuation of the medication. The lack of stringency in both of these trials calls the results into question, but each of these trials do question bupropion’s superiority to other antidepressants in the treatment of bipolar depression.
Some literature exists on the use of bupropion as maintenance treatment in bipolar disorder. It should be noted that “maintenance” treatment usually refers to studying patients for 1 year on a medication. One author published three case reports of bipolar patients maintained on bupropion for 1 year without affective episode: these patients relapsed to depression or mania within 8 weeks of bupropion discontinuation. As part of an open-label study of bupropion for the treatment of depression, 11 schizoaffective/bipolar patients were presented. These 11 patients were maintained on the medication for 1 year and had a significant decrease in affective episodes (some patients were taking concomitant treatment).
Though soon to be accompanied by more current data, the only double-blind trial to date comparing bupropion with tricyclic therapy was done by Sachs and colleagues. Bupropion and desipramine were randomly added to lithium or anti-convulsant treatment for an 8-week acute trial followed by a 1-year crossover trial. Similar efficacy was observed in both groups; however, there were more manic episodes in the TCA acute group (30% mania vs. 11%) and in the maintenance group (two TCA patients experienced manic episodes, vs. no bupropion patients). This study, though rigorous by design, is limited by its small numbers.
Of note, we are currently awaiting the unmasking of the results of the Stanley Foundation Bipolar Network double-blind randomized placebo-controlled comparison of buproprion, sertraline, and venlafaxine. There was an acute phase of 10 weeks of medication, and apparent and possible responders were offered the opportunity for 1 year of continued treatment — the subsequent section in this site discuses new evidence recommending more lengthy treatment. In the acute phase, 12.6% of the patients became hypomanic, and 12.6% of the patients became manic. In the 1-year continuation phase, 17.9% of the patients became hypomanic and 20.9% of the patients became manic. However, 42% of these patients reported a history of rapid cycling, so this switch rate may not be more than expected.
Combination of Antidepressants
There have been no studies on the risk of using combined antidepressant treatments in causing a manic switch. However, to date at least two case reports exist. The first, by Gabbay et al., is a report of a patient who switched to mania 3 days after paroxetine 20 mg was added to buproprion 400 mg po qd. It is postulated that cytochrome 2D6 interaction caused an increased level of the paroxetine. A second report was in a bipolar patient who became manic during a switch of antidepressants: buproprion was begun 4 days prior to the switch (200 mg/day) and fluoxetine was discontinued 2 days prior to the switch. This case of mania may have been caused by the combination of the buproprion and fluoxetine (because of fluoxetine’s long t1/2 ) or abrupt withdrawal of the fluoxetine.
Less studied and understood is whether withdrawal of antidepressants causes a switch to mania. In a study of imipramine in bipolar II and unipolar depression, there was a 2.6% overall incidence of hypomania (no difference between the groups). Sixty-six percent of these affective switches occurred upon withdrawal of the antidepressant. Other case reports discuss mania and hypomania upon tricyclic withdrawal — observed between 24 hours and 7 days of discontinuation. Cases have also been evaluated wherein the patients were on adequate mood stabilizer therapy, and cases have been observed upon discontinuation of SSRIs as well.
Henry poses an interesting hypothesis about why withdrawal of antidepressants might cause mania. The author postulates that the discontinuation syndrome (often seen upon abrupt cessation of antidepressants) could cause or exacerbate mania. If such were true, then a slow taper of the antidepressant would be indicated even if hypomanic symptoms were beginning to manifest. One of the studies discussed above specifically ruled out discontinuation syndrome symptoms in the patients. Inquiring about symptoms such as dizziness, light-headedness, nausea, headache, and sensory disturbance can help differentiate between discontinuation syndrome and hypomania.