Susan L. McElroy, M.D. Alex Israel, M.D. Antidepressant medications have long been viewed with suspicion as a treatment for bipolar disorder (Suppes 2010; Vieta and Garriga 2016). In an International Society for Bipolar Disorders (ISBD) consensus statement from a collaboration of 70 experts in which 173 studies were reviewed, it was concluded that antidepressants had a questionable risk-benefit ratio in the treatment of bipolar disorder patients and should not be used as monotherapy in bipolar I disorder (BD I) patients (Pacchiarotti et al. 2013). Indeed, meta-analyses are inconsistent as to whether antidepressants are efficacious in bipolar depression and whether they induce switch into hypomania and/or mania (hypo/mania) (Gijsman et al. 2004; Liu et al. 2017; McGirr et al. 2016; Vázquez et al. 2013; Zhang et al. 2013). However, the ISBD consensus statement acknowledged that the data for antidepressants in the treatment of bipolar II disorder (BD II) were more mixed than for BD I, and that antidepressant treatment, including as monotherapy, might be appropriate for some BD II patients. In this chapter we review randomized controlled trials (RCTs) of antidepressants in the treatment of BD II—either alone or in combination with BD I or major depressive disorder (MDD) patients. We also evaluate pertinent naturalistic studies of antidepressant discontinuation in bipolar disorder patients, studies comparing antidepressant response rates between BD II and BD I patients, and studies of antidepressant-induced hypomanic and/or manic (hypo/manic) switch episodes and of antidepressant-associated suicidality in BD II versus BD I. Finally, we discuss when antidepressants might be appropriate for BD II patients and how these medications might be optimally used in such patients. In particular, we attempt to answer the following questions. Do antidepressants have therapeutic effects in some patients with BD II? How safe are antidepressants in patients with BD II? Can we predict response (therapeutic and adverse) of BD II patients to antidepressants? Are some antidepressant classes more effective or safer for BD II patients? If the decision to use an antidepressant in a BD II patient has been made, which specific antidepressant should be used and how long should it be continued? RCTs of antidepressants in BD II patients can be divided into those assessing just BD II patients and those assessing BD II patients as well as MDD or BD I patients. Below, we review studies in both categories that included at least 10 patients with BD II. We found five published RCTs of antidepressants in patients only with BD II that included at least 10 patients. The first study was a two-phased, placebo substitution study of fluoxetine in patients with BD II or bipolar disorder not otherwise specified experiencing a major depressive episode (Amsterdam and Shults 2005). In the first phase, 37 patients received open-label fluoxetine 20 mg/day for up to 8 weeks; 12 (32.4%) were considered responders. In the second phase, the 12 responders were randomly assigned to either continue fluoxetine 20 mg/day or switch to placebo for up to 6 months. During this phase, 43% of fluoxetine recipients relapsed as compared with 100% of placebo recipients (P=0.08 by Fisher’s exact test). Manic symptoms were slightly but significantly higher in fluoxetine-treated patients than in patients receiving placebo. However, this difference was not considered clinically meaningful, and no hypo/manic switch episodes were observed in either group. The second RCT was a 9-month, proof-of-concept cross-over study exploring escitalopram monotherapy (10 mg/day) as a mood-stabilizing agent in 10 patients with BD II with depressive or hypomanic episodes occurring at least monthly (Parker et al. 2006). The study had three phases: 1) a 3-month no-treatment baseline to ensure that criteria for episode frequency were met; 2) a 3-month period of treatment with escitalopram or placebo; and 3) a 3-month period of receiving the opposite treatment. When compared with placebo treatment, escitalopram treatment produced significant reductions in depressive symptom severity, percentage of days with depressed or elevated mood (as assessed with patient-completed mood charts), and percentage of days impaired. There was no evidence that escitalopram caused a worsening of the course of illness. The authors concluded that escitalopram had antidepressant, and possibly antihypomanic, properties in BD II. In the third study, a randomized, double-blind placebo substitution trial, 83 BD II depressed patients who recovered with open-label treatment with fluoxetine (49.7% of the initial sample of 167 patients) were randomly assigned to receive placebo, lithium (mean average level=0.69 mmol/L), or continued fluoxetine (mean average maximum daily dose=34.3 mg) for 50 weeks (Amsterdam and Shults 2010). The mean time to depressive relapse or recurrence (the primary outcome) was significantly longer with fluoxetine (249.9 days) than with placebo (186.9 days) or lithium (156.4 days) (P=0.03). There were no significant differences in hypomanic symptoms over time among the groups. There were also no differences between rapid and non-rapid cyclers in depressive relapse or hypo/manic switch (Amsterdam et al. 2013). The fourth study had two phases: the first was a randomized, double-blind, parallel-group 12-week study comparing venlafaxine (n=65) and lithium (n=64) in BD II outpatients experiencing major depressive episodes (Amsterdam et al. 2016a). On the primary outcome, venlafaxine produced a greater response rate (67.7%) than lithium (34.4%; P<0.0001). On secondary outcomes, compared with lithium, venlafaxine produced a greater remission rate, a greater reduction in depressive symptoms, a greater reduction in global severity scores, and greater global improvement. There were no statistically significant differences in hypomanic symptoms between the two groups. In a post hoc analysis from this trial, there was a significant association between the number of prior antidepressant trials and likelihood of response (Amsterdam et al. 2016b). Specifically, a 25% reduction in response to venlafaxine was observed with each increase in the number of prior antidepressant trials, suggesting there is a reduced chance of response with an increase in the number of antidepressant trials. In the second phase of this study (Amsterdam et al. 2015), responders (n=44 for venlafaxine and n=22 for lithium) continued to receive the same monotherapy for another 6 months. There was no difference in relapse rate or time to relapse between venlafaxine and lithium in this study period. Also, there were no differences in manic symptom ratings or in syndromal or subsyndromal hypomanic episodes between the two groups. The venlafaxine group, however, had significantly fewer subsyndromal depressive episodes than the lithium group. In a post hoc analysis of both phases of this study, rapid-cycling status did not affect frequency of depressive response or of depressive relapse (Lorenzo-Luaces et al. 2016). Rapid cyclers had a significantly higher frequency of hypomanic symptoms than non–rapid cyclers during continuation monotherapy (P=0.005), but rates did not differ between venlafaxine recipients (17.6%) and lithium recipients (42.9%). In the fifth and largest study of antidepressants in BD II only, 142 patients with BD II depression were randomly assigned to receive sertraline plus placebo (n=45), lithium plus placebo (n=49), or sertraline plus lithium (n=48) for 16 weeks (Altshuler et al. 2017). The three treatment groups did not differ on the primary outcome measure, which was switch into hypo/mania: switch rates were 19.9% for sertraline, 19.4% for lithium, and 13.4% for the combination. Also, there were no differences in response rates across the three groups, which were 73.3% for sertraline, 67.4% for lithium, and 47.9% for the combination. However, the combination of sertraline and lithium was associated with a significantly higher dropout rate (70.8%) than the rates seen with either monotherapy (42.2% for sertraline and 55.1% for lithium). There was no evidence of a group×rapid-cycling status interaction for switch. Also, although response rates were similar across treatment groups among rapid cyclers, response rates were superior in the monotherapy groups compared with the combination group for nonrapid cyclers. Finally, lithium recipients who experienced a switch had significantly lower lithium levels than those who did not experience a switch. To our knowledge, nine RCTs involving antidepressants have included at least 10 patients with BD II as well as patients with BD I or patients with MDD. In the first of these studies, 22 BD II and 27 MDD patients in remission for 6 months were randomly assigned to receive lithium plus imipramine placebo, imipramine plus lithium placebo, lithium plus imipramine, or lithium placebo plus imipramine placebo for at least 2 years (average time in the study was 11 months) (Kane et al. 1982). Of the 49 patients, 43% had a depressive relapse, 4% had a manic relapse, and 2% had a hypomanic relapse. Depressive relapse rates for the four treatment groups across the entire sample were 27% for lithium, 83% for imipramine, 14% for lithium plus imipramine, and 67% for placebo; for the BD II patients, they were 25%, 67%, 17%, and 50%, respectively. Manic or hypomanic relapses were more likely to occur among patients not receiving lithium than among those receiving lithium. The authors concluded that lithium helped prevent depressive relapse in BD II and MDD patients, while imipramine had no prophylactic effect in either patient group. The second study was a 6-week double-blind comparison of tranylcypromine versus placebo in 59 patients with depressive episodes, 19 with BD II, 10 with BD I, and 30 with MDD (Himmelhoch et al. 1982). Of 28 bipolar disorder patients, 20 (71.4%) responded to tranylcypromine, while 4 (12.9%) of 31 patients responded to placebo. Switch into hypo/mania was not reported. In the third study, 56 outpatients with anergic bipolar depression, 32 of whom had BD II disorder, were randomly assigned to receive tranylcypromine or imipramine for 6 weeks (Himmelhoch et al. 1991). Tranylcypromine was superior to imipramine regarding greater symptomatic improvement, higher global response, and lower attrition rate without increased risk of hypo/manic switch (21% for tranylcypromine and 25% for imipramine). Of note, although hypo/manic switch rates were comparable, switches tended to occur sooner and be more severe with imipramine than with tranylcypromine. BD I and BD II patients had comparable outcomes, except that in the former, there were more treatment-emergent mood swings. The fourth study was a post hoc analysis of a randomized, placebo-controlled relapse prevention trial that included 839 patients with acute major depressive episodes (Amsterdam et al. 1998). After receiving fluoxetine 20 mg/day for up to 12 weeks, patients who had achieved remission were randomly assigned to receive fluoxetine 20 mg/day for 52 weeks; fluoxetine 20 mg/day for 38 weeks followed by placebo for 14 weeks; fluoxetine 20 mg/day for 14 weeks followed by placebo for 38 weeks; or placebo for 52 weeks. The 89 patients with BD II were compared with 89 age- and sex-matched MDD patients and 661 unmatched MDD patients. Both acute antidepressant response and long-term relapse rates were similar in BD II and matched and unmatched MDD patients. Over the short term, 3.8% of BD II patients had a hypo/manic switch compared with 0 matched and 0.3% unmatched MDD patients; over the long term, 2% of BD II patients had a hypo/manic switch compared with 0 matched and 1% unmatched MDD patients. The authors concluded that fluoxetine might be just as effective and safe for the treatment of BD II depression as MDD, both acutely and over the long term. The fifth study was a 6-week trial in which 27 patients with BD depression receiving lithium or valproate, 16 of whom had BD II disorder, were randomly assigned to the addition of paroxetine or the other mood stabilizer (Young et al. 2000). Both groups showed similar improvement in depressive symptoms with no differences in mania scores. Response was not compared between BD I and BD II patients. The sixth study was a post hoc analysis of an RCT comparing imipramine, phenelzine, and placebo in 310 outpatients with major depressive episodes, 62 of whom had BD II disorder and 248 of whom had MDD (Agosti and Stewart 2007). In an intent-to-treat analysis, there were no significant differences between BD II patients and MDD patients in rates of response to imipramine, phenelzine, or placebo. No patient developed manic symptoms that required antidepressant discontinuation or addition of a mood stabilizer. The seventh study was a comparison of mood switch among bupropion, sertraline, and venlafaxine in 184 patients with bipolar depression who were receiving at least one mood stabilizer, 46 of whom had BD II (Post et al. 2006). Although all patients were randomly assigned to receive one of these antidepressants, the last 156 patients enrolled in the study received the study drug under double-blind conditions. Hypo/manic switch rate was significantly greater for venlafaxine (29%) than for bupropion (10%) or sertraline (9%) (P=0.002). There was no difference in antidepressant response rates, which were 37% for venlafaxine, 33% for bupropion, and 41% for sertraline. In a related article exploring 228 acute antidepressant trials and 87 antidepressant continuation trials in 159 of these patients, the rate of hypo/manic switch was again higher for venlafaxine than for bupropion and sertraline—both acutely and with continuation treatment (Leverich et al. 2006). In the eighth study, 366 patients with bipolar depression receiving mood stabilizer therapy, 114 of whom had BD II, were randomly assigned to receive antidepressant therapy (with paroxetine or bupropion) or placebo for 26 weeks (Sachs et al. 2007). There were no significant differences between the two groups on the primary outcome, percentage of patients who had a durable recovery, defined as 8 consecutive weeks of euthymia (23.5% for the antidepressant group and 27.3% for the placebo group), or the secondary outcome, rate of hypo/manic switch (10.1% for the antidepressant group and 10.7% for the placebo group). Antidepressant durable recovery rates were similar in patients with BD II and those with BD I, although there was a nonsignificant trend for more improvement in BD II patients receiving a mood stabilizer plus placebo than a mood stabilizer plus an antidepressant. In a post hoc analysis of these data, improvement in self-esteem or loss of energy after 2 weeks of treatment was significantly associated with subsequent durable recovery, whereas improvement in psychomotor retardation was associated with subsequent hypo/manic switch (Mizushima et al. 2017). However, results comparing BD II and BD I patients were not reported. In the ninth study, 740 patients with bipolar depression, 262 with BD II, were randomly assigned to receive quetiapine 300 mg/day, quetiapine 600 mg/day, paroxetine 20 mg/day, or placebo for 8 weeks (McElroy et al. 2010). Both quetiapine doses produced significantly greater improvement in depressive symptoms than placebo in the entire bipolar disorder group, in BD I patients, and in BD II patients, while paroxetine was not superior to placebo in any of these three patient groups. Paroxetine, however, significantly reduced anxiety scores in the combined bipolar disorder group, as did both quetiapine doses, but findings were not reported for the individual BD I or BD II groups. The incidence of treatment-emergent hypo/mania was numerically lower with either quetiapine dose (2.1% with 300 mg/day and 4.1% with 600 mg/day) than with paroxetine (10.7%) and placebo (8.9%), but these results were not provided for the individual BD I and BD II groups. The authors concluded that quetiapine (at 300 or 600 mg/day), but not paroxetine (20 mg/day), was efficacious in the treatment of acute BD I and II depressive episodes. Naturalistic antidepressant discontinuation studies in BD patients suggest some patients who respond well to these drugs acutely do better over the long term regarding depressive relapse when antidepressants are continued than when they are discontinued (Altshuler et al. 2003). These studies included both BD I and BD II patients, and one found that bipolar disorder subtype did not affect relapse rates for depression or mania (Joffe et al. 2005). In a prospective comparison of BD I and BD II patients after discontinuation of successful antidepressant treatment, 70 bipolar depressed patients, 49 with BD II, who became euthymic for 2 months after the addition of an antidepressant to a mood stabilizer were openly randomly assigned to antidepressant continuation versus discontinuation for 1–3 years (Ghaemi et al. 2010). Antidepressant continuation was associated with numerically less severe depressive symptoms (mean difference in depression criteria=–1.84; 95% CI, –0.08–3.77) and a mildly delayed depressive episode relapse (HR=2.13; 95% CI, 1.00–1.20), but the differences were not significant. There was no significant increase in manic symptoms. BD II was not associated with a different response to antidepressant continuation as compared with BD I. However, in a post hoc analysis of these data, more improvement in depressive episode frequency was seen in BD I patients (mean decrease in depressive episodes per year of 0.35) than in BD II patients (mean decrease in depressive episodes per year of 0.26) (Vöhringer et al. 2015). The authors concluded that their findings were not consistent with the hypothesis that BD II has better outcomes with antidepressants than BD I. A number of studies have enabled comparison of antidepressant response in BD II versus BD I patients. In a meta-analysis of 15 trials, comprising 3,407 patients, that compared an antidepressant with either a placebo or an active comparator, Sidor and Macqueen (2011) concluded that antidepressants were not statistically significantly superior to placebo or other current standard treatments for bipolar depression. In a sensitivity analysis, bipolar disorder subtype (I or II) did not change the overall profile of estimated treatment effects. In an observational study of 221 bipolar depressed patients (n= 71 with BD II disorder) receiving an antidepressant conducted after this meta-analysis, BD II and BD I patients were found to respond at comparable rates (Pacchiarotti et al. 2011). Importantly, bipolar disorder patients who responded to antidepressants were more likely to have a past history of antidepressant response. The few studies allowing comparison of antidepressant response between patients with BD II depression and those with MDD found that acute antidepressant response and long-term relapse rates were similar in BD II and MDD patients (Amsterdam et al. 1998; Kane et al. 1982). We found no RCTs exploring the relationship between antidepressant dose and antidepressant response in BD II versus BD I or MDD. However, in a post hoc analysis of an RCT of paroxetine or bupropion in BD I and II patients receiving mood stabilizers, patients were divided into three groups: those receiving high doses of antidepressant, those receiving low doses of antidepressant, or those receiving placebo (Tada et al. 2015). Patients receiving high doses were significantly more likely to respond than those receiving low doses. Placebo recipients, however, were also more likely to respond than low-dose recipients. Randomized, controlled dose-finding studies of antidepressants are needed in BD II depression. Bond and colleagues (2008) conducted an important systematic review and meta-analysis of 13 prospective studies (7 RCTs, 2 single-blind clinical trials, 1 open-label trial, and 3 naturalistic treatment reports involving 777 bipolar disorder patients) that provided antidepressant-associated switch rates in BD II versus BD I, as well as 5 prospective studies (4 RCTs and 1 involving 185 MDD patients) that provided antidepressant-associated switch rates in BD II versus MDD. Estimated mean switch rates in studies of BD II and BD I were 7.1% and 14.2%, respectively, in acute trials and 13.9% and 23.4%, respectively, in maintenance trials. Mean switch rates in BD II versus MDD studies were 8.1% and 1.5%, respectively, in acute trials and 16.6% and 6.0%, respectively, in maintenance trials. The relative risk (RR) of mood switches was higher in BD I than in BD II (RR=1.78, 95% CI=1.24–2.58, P=0.002), and higher in BD II than in MDD (RR=2.77, 95% CI=1.26–6.09, P=0.01). The authors concluded that rates of antidepressant-associated hypo/manic switch in BD II were intermediate between those for BD I and MDD. They also observed that BD II patients, like MDD patients, switched primarily into hypomania, whereas BD I patients switched with similar frequencies into hypomania and mania. Indeed, in a recent evaluation of 591 bipolar disorder patients (465 with BD I) receiving antidepressants while acutely depressed, 205 with and 386 without a history of antidepressant-induced hypomania, the clinical risk factors associated with antidepressant-associated switch were female sex, current rapid cycling with anxiety disorder comorbidity, and a diagnosis of BD I (Frye et al. 2015). In a subset restricted to patients just receiving selective serotonin reuptake inhibitor (SSRI) or serotonin-norepinephrine reuptake inhibitor (SNRI) antidepressants, the only clinical risk factor associated with hypo/manic switch was BD I diagnosis. It has been observed that antidepressant-induced hypo/manic switch might be more common or more severe with tricyclics (Himmelhoch et al. 1991; Sachs et al. 1994) and venlafaxine (Leverich et al. 2006; Post et al. 2006; Vieta et al. 2002) than with SSRIs, bupropion, or monoamine oxidase inhibitors (MAOIs). These studies included mixed samples of BD I and II patients, and we are unaware of any RCTs comparing switch rates with different antidepressant classes in BD II patients only. It has also been observed that antidepressant-induced hypo/manic switch might be dose related, with higher doses associated with greater likelihood of switch (Ramasubbu 2001). However, we found no prospective RCTs of antidepressant dose and hypo/manic switch in BD II (or BD I) patients, although in the RCT of sertraline plus placebo, lithium plus placebo, and sertraline plus lithium in BD II depressed patients mentioned earlier (Altshuler et al. 2017), there was no difference in the proportion of sertraline recipients who reached 200 mg/day and switch status. We also found no studies exploring antidepressant dose and switch between patients with BD II and those with BD I or MDD. It is therefore unknown if antidepressant-associated switch into hypo/mania in BD II, or in BD I or MDD, is a dose-related phenomenon. Some studies suggest some BD I and BD II patients, like some MDD patients, may develop suicidality with antidepressant treatment, even when they are receiving concurrent mood stabilizers (Marangell et al. 2008; Yerevanian et al. 2007). Such suicidality has been hypothesized to represent antidepressant-induced switches into mixed states as well as an antidepressant side effect unrelated to switch. However, a meta-analysis of antidepressant trials in bipolar disorder patients found no association between antidepressants and suicidality (Zhang et al. 2013). Indeed, a 27-year longitudinal observational study in 139 BD II patients, 206 BD I patients, and 361 MDD patients found that antidepressant exposure was associated with reduced risk of suicidal behavior in both types of bipolar disorder patients but not those with MDD (Leon et al. 2014). To our knowledge, there have been no studies examining antidepressant dose and antidepressant-induced suicidality in BD II or BD I patients. More research into the possibility of antidepressant-induced suicidality is greatly needed in BD II (and in BD I). Case Vignette Lucia is a 26-year-old woman who initially presented to the psychiatric emergency department after her roommate found a suicide note on her bedside table. Over the past 2 months she reported worsening depressed mood. She stayed up all night ruminating about being a burden on the people around her and had been unable to motivate herself to go to work. She worried nearly constantly that her friends were saying negative things about her and often felt that people did not like her because she smelled bad. She denied substance use and basic lab work was unremarkable. Lucia was admitted to the inpatient psychiatric unit, diagnosed with a major depressive episode with psychotic features. She reported having several months–long major depressive episodes a year since she was 15. Her primary care provider had initiated a trial of sertraline (up to 200 mg/day) for presumed MDD in the past. She had found this to be somewhat helpful, but she had never stayed on it for more than a few months. A thorough history revealed at least two sustained periods of increased energy, decreased need for sleep, increased goal-directed activity (knitting scarves every night until 3 A.M.), impulsivity (e.g., spending thousands of dollars over a few days on clothing), and feeling “really great,” each lasting 10 days. Lucia was surprised to learn that these episodes of hypomania meant that her diagnosis was BD II instead of MDD. In the hospital, Lucia was started on quetiapine, titrated to 300 mg/day. Her psychosis remitted, and her sleep and mood improved. She was discharged to an intensive outpatient program that included an evidence-based psychotherapy for bipolar disorder, interpersonal and social rhythm therapy. However, she continued to experience residual depressive symptoms and was unable to return to work. Medication options were discussed with Lucia, including adding lithium, adding lamotrigine, switching atypical antipsychotics, and adding a selective serotonin reuptake inhibitor. Lucia chose to add a selective serotonin reuptake inhibitor to the quetiapine because of her previous partial response to sertraline. The treatment team discussed risks and benefits, including risk of switch to hypomania and mania. Early warning signs of switching, including disruption of sleep cycle, were discussed. Sertraline was started, and the dosage was increased to 200 mg/day. After 4 weeks, her residual depressive symptoms began to remit, and she was able to return to work. She, and the treatment team, continued to monitor carefully for signs of hypomania. Lucia remained symptom-free for the next 6 months.
9
Antidepressant Medications in Bipolar II Disorder
Randomized Controlled Trials of Antidepressants in Patients With Bipolar II Disorder
RCTs of Antidepressants in Patients Only With BD II
RCTs of Antidepressants in Patients With BD II and BD I and/or MDD
Naturalistic Antidepressant Discontinuation Studies in Patients With Bipolar II Disorder
Antidepressant Response of Bipolar II Disorder Depression Versus Bipolar I Disorder Depression or Major Depressive Disorder
Antidepressant-Associated Switch in Bipolar II Disorder Versus Bipolar I Disorder and Major Depressive Disorder
Antidepressant-Associated Suicidality in Bipolar II Disorder Versus Bipolar I Disorder
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
