#1 MedServ WebAward

The Psychopharmacology of Bipolar Disorder

Peter M. Brigham, MD

Note: I have not updated this site in quite a long time, and I am aware that it needs a substantial revision. Most pressing is the status of the atypical neuroleptics, which are now commonly accepted as effective antimanic agents and two of which have been approved for the treatment of bipolar depression. Please stay tuned for more up-to-date information. In the meantime, I encourage you to make use of the links labeled "search" that appear after each medication, which will take you to a MedLine search for recent articles on that medication in the treatment of bipolar disorder.

A somewhat impassioned note on drug names

(click to go to the relevant section)

DSM-4 diagnostic criteria
Differential diagnosis and comorbid factors
Mood charting
Mood stabilizers
Accepted mood stabilizers
Investigational mood stabilizers
Adrenergic blockers
Ca-channel blockers
Anticholinesterase inhibitors
Adjunctive treatments
Other agents
Omega-3 fatty acids
Special cases
Rapid cycling and mixed-state/dysphoric mania
Bipolar depression
Pregnancy & breastfeeding


Bipolar disorder has a prevalence of 1-3%, although some think it may be higher. Goodwin & Jamison (1990) estimate that approximately 1/3 of bipolar disorder is ever diagnosed, and only 1/3 of those diagnosed are in treatment, and only a small proportion of those in treatment are receiving optimal treatment. The peak ages of onset are 15-19, a fact that is well documented but under-appreciated. Going by the statistics, the average untreated bipolar patient will have the first episode of mood disruption at age 16 and 10 episodes by age 26! (Sachs) Considering the importance of this decade of development in establishing autonomy, vocational independence, and primary relationships outside the family, it is obvious that early diagnosis and treatment of bipolar disorder can have a profound effect on the course of a patient's life. In addition, the lifetime incidence of completed suicide in bipolar patients is on the order of 20% - this is an illness with a high degree of lethality, so our efforts to treat it should be vigorous and well informed.

This page is designed to be a resource for psychopharmacologists and other MH professionals, though obviously everyone with internet access can view it, and my intention is to keep it updated as new developments arise in the field. I owe much to the contributors to the Psychopharmacology internet mailing list moderated by Ivan Goldberg, MD, from whom I have learned an enormous amount on this and many other subjects over many years. I have also benefited from the conferences on bipolar disorder given by Gary Sachs and his colleagues at MGH; anyone who wants to increase their knowledge on this topic should consider attending one of these. An early version of this article was written in 1998-99 while I was at Harvard Vanguard Medical Associates, and the text was reviewed at that time by fellow-members of the Psychopharmacology and Therapeutics Committee, to whom I am likewise indebted. The first version was (and still is) posted on the Harvard Vanguard intranet site for reference by HVMA clinicians. What you see here is several revisions beyond the original. Thanks also to Ron Pies, MD, who kindly suggested a number of corrections and additions along the way.

I highly recommend the Expert Consensus Guidelines on Bipolar Disorder (2000), chaired by Gary Sachs; the American Psychiatric Association's Bipolar Practice Guidelines are another very useful resource. The main difference between those documents and this one is that the others focus on treatments that have solid research documenting their efficacy, whereas I also include here treatment approaches that have smaller studies, case reports, or anecdotal evidence to support their use. My feeling is that there is value, too, in having access to knowledge of more speculative treatments, since some patients will not respond to a "classic" approach. I also provide more practical, hands-on information (eg about dosing, drug interactions, etc) than is offered in the more general guidelines.

Nothing here is to be regarded as a substitute for your own clinical judgment. This document is by its nature provisional and subject to revision. You should always determine for yourself whether anything you read here is accurate, current, or pertinent to the clinical situations you face. Non-clinicians should take this material as general information only, and not as specific clinical advice — any questions or concerns you have as a result of reading this document should be taken up with your own mental health treatment provider. I make no warrantee that the material presented here is without error or omission.

You are encouraged to read through the whole document or to browse sections of immediate interest. Any underlined colored text is a hyperlink to another section of this document, to another related document, or to a page on the web.

Click on the [search] link following any medication to perform a current Medline search for articles pertaining to that med and bipolar disorder treatment.

(to table of contents)


DSM-4 diagnostic criteria

DSM-4 distinguishes between mood episodes and diagnosis.

A manic (hypomanic) episode is a distinct period of abnormally euphoric, expansive or irritable mood lasting at least one week accompanied by at least 3 of the following (4 if mood is only irritable):

For a manic episode, there must be "marked impairment" in social or occupational functioning and/or hospitalization must be required for patient's protection. If this criterion is not met then the episode is hypomanic.

If the history includes one or more manic episodes, then the diagnosis is bipolar I. If the history includes only episodes of hypomania, then the diagnosis is bipolar II.

Cyclothymia is characterized by nearly continuous mild episodes of hypomania and depression that does not meet criteria for major depression, with no more than two months at a time of euthymia during an interval of two years.

(to table of contents)

Differential diagnosis and comorbid factors

When diagnosing bipolar disorder attention should be paid to ruling out external factors, such as hyperthyroidism, acute drug intoxication, delerium or brain injury, etc. The fact that the age of onset of bipolar disorder is usually in the first three decades, and that about 90% of all first episodes occur before age 40 suggests that anyone presenting with new-onset mania after age 40 with no prior depression should be strongly suspected of having an underlying medical condition or substance abuse as the prime causative factor.

Unipolar depression

Mania is not "the opposite" of depression

Akiskal and others have described a population of patients who fail to meet strict DSM-4 criteria for hypomania and propose a "bipolar spectrum" model. Some patients in the "bipolar spectrum" have predominantly dysphoric/depressed mood but do not respond to multiple trials of ADs; they may benefit from the addition of a mood stabilizer. McElroy, among others, suggests that a "dimensional" view is more useful; the degree or severity of depressive symptoms should be assessed independently of the degree and severity of manic symptoms. Although these reputable and respected researchers are voicing a commonly held viewpoint, this is an admittedly "gray" area nosologically.

Consider bipolar spectrum diagnosis if patient has current depression with:

Always ask about bipolar symptoms when assessing depression.

DSM-4 diagnostic criteria mandate that major depression is excluded if a patient "has ever had an episode of mania or hypomania." The implication of this is that, strictly speaking, major depression is a diagnosis that can never be made with 100% confidence, since a significant proportion of those eventually diagnosed with bipolar disorder have an initial episode (or episodes) of depression. This is particularly true with early onset depression; one study (Geller et al 1994) found that over 30% of preadolescent children diagnosed with major depression went on to develop manic episodes. Early onset of depression (especially with comorbid anxiety disorder), family history of bipolar disorder, or multigenerational history of any mood disorder should heighten suspicion for bipolar rather than unipolar disorder. For instance, a teenager presenting with major depressive symptoms who has a first-degree relative with bipolar disorder should be strongly suspected of having bipolar not unipolar depression.

Inquire about all the criteria for manic episode listed above. DSM-4 criteria are actually very useful! Note: history of hypomania is often not reported by the patient but reported clearly by close relatives or friends. If you have a suspicion, get permission to talk to those who know the patient well. Tertiary experts on bipolar disorder uniformly insist on talking to family members as part of the evaluation process.

If "soft bipolar" symptoms are confirmed and the patient has had poor or equivocal responses to standard ADs, a trial of a mood stabilizer may be worth considering. Some of these patients, if not treated early, will develop more clear-cut hypomanic episodes as time goes on, clarifying the earlier suspicions regarding diagnosis.

A note about antidepressant-induced mania: The DSM-4 subcommittee reversed DSM-3 by excluding from the diagnosis of bipolar disorder those patients whose only symptoms of mania occur upon antidepressant treatment. There is now broad consensus that this was an error; approximately 80% of such patients will go on to have spontaneous episodes of mania or hypomania within 2 years (Sachs, Akiskal). Despite DSM-4, such patients should probably be treated as bipolar (I or II).

(to table of contents)


Historically, bipolar disorder has been confused with schizophrenia. Patients with severe mixed affective states can show prominent psychotic symptoms with no apparent "classic" euphoria, pressure of speech, etc., and can present as disorganized and thought-disordered, sometimes with florid paranoia. Adolescents especially sometimes present with mood-incongruent hallucinations, paranoia, and marked thought disorder, and turn out to have mixed-state episodes. Although affective symptoms usually are evident on exploration of history and mental status exam, they may be overshadowed by the psychotic picture. It is probably a good heuristic rule that any first-break psychotic patient (particularly if the index episode is in the mid-teens) should be tried sooner rather than later on a mood stabilizer (for instance, valproate), on the chance that a lifetime of neuroleptics might be avoidable.

Some question whether schizoaffective disorder is a distinct pathologic entity, but it continues to be useful as a diagnosis of exclusion at present.

(to table of contents)

Substance abuse

Approximately 60-75% (Sachs, Goodwin & Jamison) of the bipolar disorder clinical population have significant substance abuse (SA). Estimates vary widely (3% - 98% overall !) depending on whether rating scales or formal diagnostic criteria are used and whether the distinction is made between abuse and dependence. For females, bipolar II is associated with a higher incidence of alcohol abuse than bipolar I; for males the incidence is about equal. Cocaine abuse is associated with a several-fold increase in incidence of bipolar disorder; the rate of bipolar disorder in cocaine abusers may be higher than in any other category of substance abuse. Some increases in comorbidity exist in opiate abusing populations as well. Careful history can sometimes elucidate whether the SA is primary or secondary, but this is often obscure. SA is high both in depressed and especially in manic and mixed phases; some bipolar patients may cease drinking when depressed. The comorbidity may be due to genetic factors, attempts to "self-medicate," and/or a result of overall impairment of judgment. SA (especially cocaine or hallucinogens) can precipitate affective episodes and can alter the course of bipolar disorder by inducing switching. Bottom line: all bipolar disorder patients should receive a careful assessment of SA patterns and history.

With primary alcoholism and secondary affective disorder, affective symptoms usually remit after 2-4 weeks of sobriety (Akiskal). Persistence of affective symptoms longer than a month should heighten the suspicion of underlying affective disorder. Very early onset of SA suggests primary affective disorder - Famularo et al found 7 out of 10 patients with onset of alcohol abuse before age 13 had bipolar disorder. The prognosis for treatment of bipolar disorder is poorer for those patients who are also substance abusers. There is data suggesting that valproate is the preferred mood stabilizer in this population and that lithium is less effective.

Patients need to be educated about the relationship between the two disorders. AA or NA should be encouraged for those with substance dependence, but the patient should be "inoculated" against the attitude sometimes encountered in AA regarding abstaining from all "mind altering drugs" - more than one bipolar disorder patient has become manic or depressed after discontinuing mood stabilizers on advice from well-intentioned AA members. There is an AA pamphlet entitled The AA Member — Medications and other drugs, which clearly distinguishes between necessary and important prescription medications and self-administered drugs. Patients can be encouraged to obtain this pamphlet and refer to it if fellow AA members start pressuring them about their medications.

When both SA and bipolar disorder coexist - and they often do - each needs treatment in its own right. Successful treatment of bipolar disorder usually requires successful treatment of substance abuse.

(to table of contents)


There is considerable overlap between symptoms of ADHD and those of bipolar disorder. The hyperactivity and distractibility of ADHD can be mistaken for the racing thoughts and increased motor activity of bipolar disorder. Comorbidity is high. In fact, Wozniak et al found 90% of bipolar children had comorbid ADHD and 19% of ADHD children had comorbid bipolar disorder. The differential is often obscure, but the following characteristics may be helpful. While bipolar disorder may have childhood onset, ADHD always does. Bipolar disorder usually shows a cyclic or episodic pattern (at least after age eight), while ADHD is relatively constant in its effects on attention and impulsivity. Bipolar disorder hyperactivity is usually more goal-directed compared to the scattered hyperactivity of ADHD. Sleep patterns in bipolar disorder show episodic alterations, particularly phase reversal, whereas ADHD patients tend to have more consistent difficulty, often with onset insomnia. When stimulants worsen the clinical picture bipolar disorder may be suspected, though the converse is not necessarily true, as sometimes bipolar patients improve with the addition of stimulants (see below).

(to table of contents)

Borderline personality disorder

There are some (Akiskal, Gunderson & Elliott) who have proposed that borderline personality disorder (BPD) should be seen as part of the affective spectrum, but others (e.g., Pope) who maintain that it is independent, though frequently coexisting with affective disorder. Although much of the overlap may be on the basis of the depressive features of BPD, Akiskal and his colleagues have emphasized the close linkage to bipolar disorder: in a series of 100 DSM-4 borderlines, 25% met criteria for bipolar II or cyclothymia on followup. Bipolar depression and the depressive mood of BPD patients can sometimes be distinguished by the latter's relative high day-to-day or even hour-to-hour variability (as opposed to distinct phases of depression with a beginning and an end), high reactivity, and lack of changes in sleep and appetite. An occasional atypical mixed or rapid cycling bipolar patient, however, will look very "borderline," with quite volatile and reactive mood. All too often, bipolar features can be missed in the "sturm und drang" of BPD, especially in adolescents, and care should be taken to rule out bipolar disorder before prescribing antidepressants. When racing thoughts are a clear symptom, a bipolar component should be suspected.

In sum, distinguishing the two disorders is problematic, and there is probably a subgroup of borderlines with "bipolaroid" (predominantly mixed state) features, for whom mood stabilizers are an important treatment modality. Lack of response to several antidepressants may be a clue here.

(to table of contents)


Mood charting

The usefulness of mood charting for complex or treatment-resistant cases cannot be overemphasized. Some clinicians are mildly uncomfortable with using structured tools like this, feeling they tend to be a Procrustean bed onto which the treatment must be painfully fitted. While no checklist can capture a clinical picture perfectly, the advantages of using mood charting far outweigh its limitations if a patient is not responding to first-line treatments. Mood charting makes it possible to follow in great detail the patient's mood and relate it to a variety of variables in a way that would otherwise be impossible.

A sample mood chart is available here, adapted from one used by Sachs' MGH Bipolar Clinic. Charting assists in the tracking of medication use, mood, sleep, menstrual cycle, and other symptoms, as an invaluable aid for the psychopharmacologist's prescribing. It also becomes an ongoing reminder for the patient of the existence of his/her illness and the importance of monitoring and managing it with the team. Generally, patients rapidly come to appreciate its usefulness and become committed to filling out the charts faithfully. Failing to fill out mood charts is sometimes an early sign of trouble in the treatment alliance, though it can also result from anergic depression or disorganization.

A note about sleep. Sleep deprivation can precipitate mania. (It is a proven - if temporary - cure for depression.) Patients and their families should be educated about the high risk for mood disruption that is incurred when sleep hygiene is not maintained. A regular sleep/wake schedule should be followed, and alterations of sleep should be noted and reported. Mood charting is very useful for monitoring this.

(to table of contents)

Mood stabilizers

There is no agreed-upon definition of the term "mood stabilizer!" We all use the term, but it is nowhere officially defined. Sachs proposes: an agent that has efficacy in at least one of the primary treatment objectives (acute mania, acute depression, prophylaxis) that does not worsen an acute episode and does not increase affective switching.

In many cases more than one mood stabilizer will be necessary for full control of mood episodes. Serial trials of agents one after another is certainly the recommended way to begin treatment, but often adding a small dose of another agent can add considerably to therapeutic effect. Robert Post at NIMH has been an advocate of this approach, observing that at times using concurrent medications in lower doses can have synergistic therapeutic effects while avoiding side effects from any single med. Some patients do not fully respond until three or even four mood stabilizers are used concurrently, with full therapeutic doses of each.

Because polypharmacy is often necessary in bipolar treatment, an organized approach to the psychopharmacology of these patients is crucial. Except in very acute (usually inpatient) situations, one should not change or add more than one drug at a time, as this will obscure the evaluation of both response and side effects. Careful attention should be paid to drug interactions that affect dosing (e.g., CBZ lowers VPA levels, VPA raises LTG levels). Doses should be pushed to the maximum suggested or tolerated before concluding there is no benefit. Sufficient time should be given for any clinical improvement; pressure from the patient to move faster should be resisted as much as possible, since almost invariably this will complicate the picture with needless polypharmacy and/or premature conclusions of inefficacy. Keep in mind the rule of thumb that approximately five half-lives is required for a drug to achieve steady state — this becomes important for drugs with long half-lives, like zonisamide, which will not equilibrate until nearly two weeks after a dose change. Usually in mood disorders a minimum of 4 weeks after equilibration at maximal doses is necessary to have any confidence of full clinical effect.

An inadequate trial of a medication — either insufficient dose or too brief a course — is worse than no trial at all, since at best it is a waste of time and at worst it may permanently remove from consideration a potentially useful agent. "I already tried that, Doc -- it didn't help at all."

Mood charting is by far the best way of assessing response. Often patients report feeling "no better" globally when mood charting reveals that cycling frequency or amplitude is improving, which occurs typically well before any return to euthymia. In fact, months may be required for final stabilization of mood on the correct regimen.

A question that comes up often from patients is how long to stay on a mood stabilizing regimen. Current guidelines recommend 6-12 months after euthymia for bipolar I with 1-2 mild to moderate manic episodes (though some clinicians would be more ready to recommend longer term treatment even in this case), and indefinitely for bipolar I with >2 manic episodes or one manic episode if severe or with a strong family history of bipolar disorder. When stopping mood stabilizers, the taper should be done over 1-3 months. For bipolar II disorder, the Expert Consensus Guidelines recommend indefinite treatment after 3 episodes of hypomania or antidepressant-induced mania. The prevailing data suggests that there is a positive correlation between number of previous affective episodes and the development of treatment-resistance, so the decision to stop medications must be recognized as incurring significant long-term risk.

A note about mood cycling: Cycling does not necessarily imply phases of (hypo)mania followed by phases of depression. The cycling may consist solely of episodes of depression; with a past history of mania or hypomania, such patients are still bipolar, and the same criteria apply — the frequency of episodes is significant and should be monitored, and more than four episodes per year qualifies as rapid cycling. In addition, the pattern of cycling may be significant, since there is some evidence that with patients with either predominant depression or an "MDE" pattern (mania followed by depression followed by euthymia) the course and response to treatment may be different compared to those with "DME" pattern (see below).

A note about "compliance": Often a patient will present with hypomania announcing s/he stopped meds "because I didn't think I needed them any more." Don't mistake the cart for the horse in this situation - probe for the possibility that mild breakthrough hypomania resulted in some grandiosity that led to discontinuation of meds, rather than the other way around. If so, instead of simply restarting medications that were not fully effective, one should perhaps add or change mood stabilizers.

(to table of contents)

Accepted Mood Stabilizers

Lithium [search]

Initial studies in the 1960's showed (Sachs 1998):

Antimanic efficacy: four double-blind studies, n=116, 78% improved. Ten other controlled studies, n=413, 81% improved. Antidepressant efficacy: 11 controlled studies, n=269; 8 with lithium superior to placebo, 3 failed to show benefit (one was only a 10 day trial). Prophylactic efficacy: 10 controlled studies, n=514, all 10 show lithium > placebo, the overall rate of recurrence was lithium 37%, placebo 79%.

Naturalistic studies show a very different picture!

More recent open retrospective studies fail to detect the benefit of lithium maintenance (eg, Mander 1989). Lithium monotherapy was found to be inadequate for most patients: an NIMH collaborative study showed 33% episode-free at 18-24 month follow-up; MGH followed up 100 unselected bipolar patients and found 4% (!) episode free at 1 year on lithium alone!

Baldessarini et al (2000) surveyed the literature and found that "[n]either reported recurrence rates nor average proportions of time ill nor patient improvement of 50% or more during lithium maintenance therapy in a stable clinic setting has changed significantly since the 1970s." It is hard to reconcile this study with the above data, however. One could speculate that over the years there has been a gradual recognition of less typical bipolar patients with variants of the disorder (rapid cycling, atypical bipolar II, etc) that are less amenable to lithium treatment. Further, the growth of antidepressant treatment may have contributed to this shift by promoting more rapid cycling in susceptible patients, and certainly some proportion of previously lithium-responsive patients have discontinued their lithium and become lithium non-responsive. Discontinuation of lithium for a patient previously responsive to it incurs a 50% incidence of relapse within 9 months (!) (e.g., Viguera et al); the recurrences may be more treatment-resistant than prior episodes and require switching or adding other mood stabilizers. In sum, there appears to be at least a subpopulation of bipolar patients now for whom lithium is not the treatment of choice. Nonetheless, lithium should not be underestimated as an effective first-line mood stabilizer. In addition, there is evidence that lithium-responsiveness is a genetic trait, so a family history of response to lithium should suggest this drug as a primary treatment.


Attention should be paid to the possibility over time of subclinical hypothyroidism. The usual TSH screening picks up hypothyroidism once it has progressed to an endocrinologically significant point, but well before that a subtle functionally hypothyroid state can begin to compromise lithium responsiveness (see below under thyroid). If breakthrough hypomania begins to occur, it is probably worthwhile to consider low-dose thyroid augmentation, especially if the TSH is > 2.0.

Lithium shows poor response in:

(to table of contents)

Valproate (eg, Depakote, Depakene) [search]

VPA may now be the treatment of choice for acute uncomplicated mania and its continuation treatment. VPA is the current treatment of choice for mixed state and rapid cycling, and probably as well for psychotic mania and bipolar disorder with comorbid substance abuse.

Note: the paper by Bowden et al in Archives of General Psychiatry raises some controversial issues about VPA vs lithium in maintenance treatment. In their study, valproate did not differ from placebo in delaying recurrence of mood episodes, although it did show benefit over placebo in lower rates of discontinuation for a recurrent mood episode, and was superior to lithium in longer duration of successful prophylaxis. The study was criticized for the low relapse rate in its placebo group, suggesting that selection had been biased towards milder illness, partly because sicker patients may have been reluctant to risk getting placebo for a year. In addition, the study did not address the phenomenon of polypharmacy; in the real clinical world, optimal treatment may involve utilization of more than one mood stabilizer, and only maintenance monotherapy was examined, with severely limited "rescue" interventions. The long-term usefulness of VPA as part of a polypharmaceutical regimen has not been studied.

In acute mania, oral loading of VPA can achieve rapid control of symptoms (McElroy et al): day 1: single dose of 20 mg/kg (rule of thumb is weight in pounds x 10 = mg of VPA); days 2-4 same dose but split bid; day 4 get labs (VPA level, platelets, LFTs) then titrate dose to get level > 80 ug/ml or best clinical response. Some patients need > 100ug/ml.

For cyclothymia, VPA seems to be effective at surprisingly low doses, in one study 125-500 mg/day.


The accepted position has for years been that VPA has poor efficacy for bipolar depression. It turns out that this was based in large part on a Finnish study. Close examination of the study suggests the evidence for this is shaky; the inclusion criteria allowed a huge percentage of patients with unipolar depression, a fact that was lost in the English translation of the report. There are more recent claims that VPA may have some antidepressant effect for bipolar (as opposed to unipolar) patients. Five small open trials of divalproex for bipolar depression showed a 30% response rate — not impressive, but perhaps not negligible. Whether this might be due to its effect on treating mood cycling is unclear.

(to table of contents)

Olanzapine (Zyprexa) [search]

Olanzapine is now approved by the FDA for the treatment of mania. Although there have been reports of olanzapine inducing mania on occasion, it is proving to be a very effective mood stabilizer. Its major disadvantages are weight gain and sedation. Although the formal FDA approval is only for acute mania, there are now longer term studies demonstrating its usefulness for maintenance mood stabilization, and some indication (Sanger et al 2003) that it is useful for rapid cycling bipolar disorder. Olanzapine is proving to be an excellent mood stabilizer, but there is growing recognition of its role in causing the "metabolic syndrome" of obesity, elevated serum lipids, and insulin resistance, which must be addressed especially with long-term use.


(to table of contents)

Carbamazepine (eg, Tegretol) [search]

Evidence for CBZ's antimanic efficacy is not as robust as that for VPA, though it has been in use even longer than VPA for this purpose. It seems to have some antidepressant effects. It may be even more effective as a second mood stabilizer, though its tendency to induce CYP enzymes requires that blood levels of all agents need to be followed. Dosage should be titrated to serum levels, though the correlation between serum level and therapeutic effect is near zero in some studies. Note that this may mean that different patients may require differing serum levels for efficacy, not necessarily that serum levels in any given patient have no meaning at all. Apparently the new Tegretol XR form of carbamazapine will be submitted for approval by the FDA for acute mania.


(to table of contents)

Lamotrigine (Lamictal) [search]

Lamotrigine has just been approved by the FDA for "the maintenance treatment of adults with Bipolar I Disorder to delay the time to occurrence of mood episodes (depression, mania, hypomania, mixed episodes) in patients treated for acute mood episodes with standard therapy." Although the approval refers to "mood episodes," the data are only convincing for prevention of depression — it does not do well at preventing recurrences of mania. Although there is data supporting its use in acute mania, it seems not to be as reliable for this as valproate, lithium, or olanzapine, and there have even been case reports of it causing mania. It may be useful for treating acute bipolar depression. It is generally less sedating than the other anticonvulsant mood stabilizers. Its major drawback is the risk of serious rash, requiring slow titration of initial dose, and its interactions with CBZ and VPA (see below). An FAQ is available online.


(to table of contents)

ECT [search]

ECT seems to be equally effective for acute depressive and acute manic episodes, and is more effective statistically than any drug for depression, whether unipolar or bipolar. Bilateral ECT seems to be more effective for bipolar patients, depressed or manic, though more recent experience suggests greatly suprathreshold unilateral ECT may be used. Generally, anticonvulsant mood stabilizers and benzodiazepines will be discontinued for ECT, though they may just be held the morning of treatment with bilateral ECT; lithium and/or neuroleptics (including clozapine) need not be, though lithium dose may also be held acutely to let levels drop because of concerns about increased neurotoxicity. Maintenance ECT for long-term control of bipolar disorder appears to be a viable treatment — see Vaidya et al.

(to table of contents)

Investigational mood stabilizers

These are rational options with less data as yet to back them up. Some of them are mainly of theoretical interest, others have some clinical experience or uncontrolled studies to support their use. None of them can be recommended as first-line agents, though lamotrigine, gabapentin, and topiramate are approaching that status. I include here even agents for whom the evidence of efficacy is very sparse; when a patient does not respond to first-line treatments one must try less accepted ones, since occasionally a less orthodox treatment can be quite successful.


(to table of contents)

Oxcarbazepine (Trileptal) [search]

Oxcarbazepine is a keto-analogue of carbamazapine. The modification to the molecular structure prevents it from being metabolized by oxidation to CBZ-10,11-epoxide, which is the metabolite that is thought to be responsible for much of the toxicity of CBZ. OCB is metabolized by conjugation rather than by oxidation, and largely excreted by the kidney. See e.g. Lloyd et al. It has been on the market for over 10 years in various countries. The neurologists that I have spoken to seem to regard it as exactly equivalent to CBZ for seizure control ("It's Tegretol without the toxicity"). The clinical efficacy for seizure control of OCB compares favorably with CBZ in clinical trials, and there are case reports of its efficacy for bipolar disorder, and a very small placebo-controlled 1983 study (interestingly, this study was one of the early favorable reports for the use of VPA in mania). More recently, other studies have compared OCB (favorably) to lithium and to haloperidol, and at the 2001 APA meeting Reinstein presented a poster reporting equal effectiveness compared to valproate for acute mania. See also the recent open label study by Gaemi et al 2003. There have been no instances of marrow suppression in 7000 cases according to one citation. Compared with CBZ, P450 enzyme induction is greatly reduced; there is only a moderate induction of CYP3A4 and other isoenzymes seem not to be affected. It seems to be less prone to causing cognitive impairment than CBZ (see Grant & Faulds 1992). When switching from CBZ to OCB, one should be aware that the previous induction of CYP enzymes will be considerably decreased, and as a result the serum levels of other drugs may rise. The only medical issues of note is its occasional side effect of hyponatremia.


(to table of contents)

Gabapentin (Neurontin) [search]

GBP is an anticonvulsant drug particularly notable for its relative safety and lack of interactions with other drugs. Its use in bipolar disorder is based on clinical impressions of efficacy, usually as an adjunctive agent; it has not at this point been adequately established as a primary mood stabilizer. In fact, Parke-Davis sponsored a study preparatory to an FDA application for use as a mood stabilizer and found no efficacy above placebo (the FDA application has not been pursued, but a related compound is being looked at which may have greater efficacy). For many patients it seems to have significant benefit in combination with other agents, e.g., in reducing depression — but not mania — in mixed states (see Schaffer & Schaffer 1999). It cannot be recommended in most cases as a sole treatment. The prevailing clinical impression is that GBP appears to have fairly definite antianxiety effect. It also works well in lower doses in some patients for sleep, as an alternative to trazodone. An FAQ is available online.


(to table of contents)

Topiramate (Topamax) [search]

Topiramate is yet another anticonvulsant with mood stabilizing effects for some patients (see, eg, Marcotte 1998, McElroy et al 2000, Guile & Sachs 2002). It blocks sodium channels, enhances GABA, inhibits non-NMDA (kainate/AMPA) receptors, and inhibits carbonic anhydrase; which of these effects is most important for anti-seizure or mood stabilizing effect is not clear. It may also have some efficacy in anxiety, bulimia, and PTSD. It has been reported to cause depression, but the incidence of this is undetermined. Some clinicians — notably Alan Schatzberg (personal communication) — are now trying low doses (25-75mg/d) of TPM as an adjuctive treatment to counteract the tendency toward weight gain from VPA, with occasional success. The latest data on its antimanic effect is disappointing; nonetheless, it could be tried as a second-tier mood stabilizer when more reliable ones fail. One study finds it quite effective in preventing relapse in alcohol dependence. An FAQ is available online.


(to table of contents)

Tiagabine (Gabatril) [search]

Tiagabine is yet one more anticonvulsant with possible mood stabilizing properties. It is a selective GABA reuptake inhibitor. Although a small open trial failed to find any therapeutic effect in acute mania, there are some reports (eg, Kaufman 1998, Suppes et al 2002) of it being effective in augmentation of mood stabilization. Dosage seems to be 4-8mg/day; in doses of 12-16mg/day it tends to cause cognitive impairment.

(to table of contents)

Zonisamide (Zonegran) [search]

Zonisamide has been in used in Japan as an anticonvulsant (Excegran) since 1989. Its precise mechanism of action remains obscure, although it has been noted to have calcium channel blocking (CCB) activity. ZNS is long-lived, with a elimination half-life of 63-68 hours. Its renal/hepatic metabolism is roughly 85%/15%. It does not exhibit inducing or inhibitory effects on CYP enzymes in the mouse and does not autoinduce its own metabolism, which is at least partially through CYP3A4, for which it is a substrate. It is a sulfonamide derivative, so patients allergic to sulfonamides should probably avoid it. There are small uncontrolled series in which ZNS was found to have definite anti-manic effect. Recommended initial adult dosage in Japan is 100-200 mg/d, increased if necessary to 200- 400 mg/d, up to a maximum of 600 mg/d. See the preliminary package insert for more details, and also this article from 1994.

(to table of contents)

Levetiracetam (Keppra) [search]

Levetiracetam is an anticonvulsant approved in the US in late 1999. There are now anecdotal reports of its action as a mood stabilizer, though no published accounts or studies are available in humans (one animal study suggests promise for bipolar disorder). It is a novel anticonvulsant, unrelated chemically to other drugs in the class.  It has no known interactions with any other drugs, and low protein binding, and its mechanism of action is unknown but doesn't seem to involve GABA pathways or any common neurotransmitter kinetics, nor does it affect benzodiazepine receptor sites. It is largely excreted by the kidney, and undergoes little if any hepatic metabolism; it does not affect CYP enzymes. Dosage for anticonvulsant purposes is 1000-3000mg/day in divided doses. See this webpage for some more details. It is being tried in both adults and children for treatment-resistant bipolar disorder, and it seems to have some efficacy in selected patients. Stay tuned for more as it develops.

(to table of contents)

Adrenergic blockers

Clonidine, guanfacine, propranolol [search]

Several small studies have suggested that high dose propranolol has antimanic effect. Similarly, clonidine (actually an alpha-2 agonist which functionally decreases NE tone) in doses of 0.45-0.75mg/day was found to have acute antimanic effect, with less sedation than neuroleptics. One study used clonidine for maintenance treatment with good effect over three years in four out of seven cases. Guanfacine has been reported to actually induce mania, especially in children with family histories of bipolar disorder; it has also been found to raise valproate levels.

(to table of contents)

Ca-channel blockers

Verapamil, nifedipine, nimodipine [search]

Nimodipine is a dihydropyridine Ca-channel blocker. Post (1998) reports "clinically important responses in 10 of the first 30 patients exposed to nimodipine monotherapy, and in many instances this responsivity was confirmed in B-A-B-A (on-off-on-off) double-blind clinical trials. However, in the majority of instances, the response was less than complete, and required augmentation with carbamazapine (4 of 13 patients showed a good response to this augmentation regimen) and other agents as well." This response faded in three patients when they were shifted to the less expensive verapamil; isradipine and amlodipine, both dihydropyridine-type Ca-channel blockers, seemed to preserve the response. Research on nifedipine has shown inconsistent effects on mood stabilization. Post suggests that dihydropyridines may have specific utility for bipolar disorder. Nimodipine does not cause weight gain or tremor, and incidence of GI side effects is low. It does not seem to cause cognitive dulling, and may even sharpen cognition in some patients; it increases CSF somatostatin levels (permanently reduced in patients with Alzheimer's and transiently reduced in patients with depression and MS). Dosage is 30 mg BID initially, then 30 mg QID; one case report used 270mg/day. Needless to say optimal dosing remains to be established.

(to table of contents)


Acetazolamide (Diamox) [search]

Acetazolamide is a carbonic anhydrase inhibitor. One paper (Hayes 1994) reported using acetazolamide in treatment refractory bipolar patients, and it has been used since with occasional success. In the 1994 study, seven of 16 patients had positive results, some maintained over 2 years. "[A]ll of the responders were either in a depressive phase or in a rapid-cycling phase of a bipolar illness and... all had experienced partial positive response to at least one other anticonvulsant and were being maintained on anti-convulsant therapy when the response occurred." Dosage used for glaucoma and altitude sickness is 375-1000 mg/day divided into two doses, dosage for use in bipolar disorder is not yet well established.

(to table of contents)

Cholinesterase inhibitors [search]

Burt et al (1999) reported on a small open trial of donepezil (Aricept) as an adjunctive treatment for acute manic and mixed states. Six of 11 had "marked" improvement and 3 more had "slight" improvement. Five of the six marked responders experienced improvement within two weeks or less. Although one patient did not tolerate the medication, side effects were generally mild. Presumably, controlled studies will follow. Meanwhile, this might be worth trying in some non-responders. Note that in one study looking at its efficacy in psychotropic-induced memory loss (it was helpful!), two bipolar patients became manic, so be alert for switching.

(to table of contents)

Mexiletine (Mexitil) [search]

One study (Schaffer et al) found mexiletine, an anti-arrhythmic drug with anticonvulsant activity, to be helpful in rapid-cycling bipolar disorder. To quote from the results section of the medline abstract:

"Thirteen subjects (10 female, 3 male), mean age 41 years (S.D.=7.6), and mean duration of illness 20 years (S.D.=7.7) completed the study. The dose range of mexiletine was 200-1200 mg/day. Full response (>/=50% reduction in BMS) was seen in 46% of the subjects, and a partial response (25-49% reduction in BMS) in 15%. Of note, 5/5 subjects with a mixed or manic state demonstrated a full or partial response."

This was a small open trial, with the attendant limitations, but may point to another mood stabilizing agent to try in treatment-resistant cases.

(to table of contents)

Adjunctive treatments

Typical Neuroleptics [search]

Naturalistically, 30-40% of bipolar patients receive chronic neuroleptic treatment. Why? 20 studies since 1970 show high incidence of psychotic symptoms in mania: delusions 48%, hallucinations 15%, Schneiderian first-rank symptoms 18%, prior history of psychosis 58%. The earlier the onset of first episode the greater the incidence of psychotic symptoms (Rosen et al 1983). Most bipolar patients will respond to lowish doses (e.g., haloperidol 2-10 mg/day) especially if combined with benzodiazepines.

Note that if psychotic symptoms occur during hypomania, DSM-4 classifies the episode as severe and requires changing the diagnosis to mania even if other symptoms are mild. Implication: usually if a bipolar patient ends up on neuroleptics, the diagnosis will be bipolar I not bipolar II. If chronic neuroleptics are needed, the question of schizoaffective disorder arises. It is still not clear if schizoaffective disorder is really a separate illness in any sense - for instance, what if some "schizoaffective" patients end up responding in a few years to some new mood stabilizer currently unavailable and turn out not to need neuroleptics? Will we then conclude they were bipolar after all? Or that the "mood stabilizer" has neuroleptic properties? This is a murky area. Given the risks of long-term neuroleptic use, any bipolar patient who seems to need chronic neuroleptics probably should be considered for a more vigorous mood stabilizing regimen, since there is a reasonable chance that the psychotic symptoms represent inadequately treated affective disorder.

Acute phase indications for neuroleptics:

Maintenance phase indications for neuroleptics: Neuroleptic downside:

Adverse effects: acute EPS, akathisia, tardive dyskinesia (incidence may be higher in affective disorders and elderly women), seizure, NMS. Mukherjee et al (1984) report a high incidence of abnormal neurological signs in bipolar patients with high neuroleptic exposure, and suggest a causal relationship. (Note that this was an era when bipolar patients were universally treated with lithium, which has been found to heighten neurotoxicity in combination with neuroleptics; other mood stabilizers may not do this.) Negative impact on course of illness: possible increase in cycle frequency, increased incidence of depression (both post-manic and during maintenance).

Sachs suggests using neuroleptics as follows:

(to table of contents)

Atypical neuroleptics

There is less controlled data on atypical neuroleptics, but all of them now have data supporting their use in acute mania. Studies suggest they have a superior side effect profile and efficacy equal to or better than standard neuroleptics. Some of them, however, can occasionally induce or worsen mania. Note that all of them (except ziprasidone and aripiprazole) can cause weight gain, hyperglycemia (not always related to weight gain), and elevated triglycerides (ibid). Olanzapine, now approved for the treatment of acute mania, has been covered above.

Clozapine [search]

There is quite good evidence that clozapine is effective both in refractory depression and mania in bipolar disorder, with or without psychosis. The major drawback is the obvious one - the need for regular WBC monitoring and the risk of bone marrow suppression. It should probably be avoided in combination with carbamazapine.

Risperidone [search]

Ghaemi (1997) says "In an unpublished review (Ghaemi and Goodwin, in preparation), in all reports including three or more bipolar patients (nine studies, n=113), the risperidone response rate was 70% and the mania induction rate was 4.4% (five out of 113)...." The reports of inducing mania seem to be associated with doses above 6 mg per day, and this may indeed be a phenomenon that is more common if RSP is given unopposed by other mood stabilizers. Vieta et al is one of the more recent studies demonstrating the efficacy of RSP as an adjunctive treatment in bipolar disorder. More recent studies suggest its usefulness as a primary treatment for acute mania (eg, Segal, et al).

Other atypicals

Quetiapine (Seroquel) [search] is beginning to have some research support for use in bipolar disorder — see Vieta et al 2002, Brown et al 2003 , Altamura et al 2003. Recent data presented by Jones et al at the 2003 APA found it comparable to lithium for acute mania.
Ziprasidone (Geodon) [search] is still an unknown; we'll also have to wait for more clinical experience and data to see how useful this will be for bipolar disorder — early anecdotal evidence suggested it has an activating effect, and questions have been raised about its potential to induce mania. However, one recent study (Keck et al 2003) suggests it is useful for acute mania. Some people think that the activation is much more prominent at lower doses, and rarely an issue at 80mg bid.
Aripiprazole (Abilify) [search] has little published data yet for its use in bipolar disorder, though Keck et al have an article showing it significantly more effective than placebo in acute mania. Rumor has it that the manufacturer will apply for FDA approval for this indication. It is being called a dopamine stabilizer because of its partial agonist activity; it blocks the overstimulation of mesolimbic dopamine receptors, while increasing stimulation in underactive mesocortical DA receptors, overall restoring relatively normal DA levels in the brain. It is not yet clear whether its blocking effects at 5-HT1A and 2A receptors (thus boosting serotonin) forebode a risk of inducing mania.

Trimipramine [search]

Trimipramine may be an exception to the rule of avoiding tricyclics in bipolar patients. According to Post (1998), TMI's mechanism of effect does not appear to be reuptake inhibition of serotonin, dopamine, or norepinephrine, and it appears to have dopamine blocking efficacy perhaps similar to some atypical neuroleptics. It has been reported to be effective in some patients with rapid cycling bipolar disorder. However, an open trial in four acutely manic patients found no benefit at all, so this may be a case of the pharmacologic profile being of poor predictive value.

(to table of contents)

Other agents

Omega-3 fatty acids [search]

According to Stoll, omega-3 fatty acids may inhibit neuronal signal transduction pathways in a manner similar to that of lithium carbonate and valproate. His studies in bipolar patients have used 9-10g/day of omega-3 fatty acids, in the form of fish oil, and found significant mood-stabilizing effects. A good summary article on this work appeared in in Psychiatric Times. He has used flaxseed oil for the same purpose, but his controlled study used fish oil. To quote from the handout sheet he gives to patients:

"The 2 omega-3 fatty acids studied for mood stabilization and elevation are EPA (eicosapentenoic acid) and DHA (docosahexenoic acid), both from fish oil. Flaxseed is a terrestrial source of a different, untested omega-3 fatty acid (alpha-linolenic acid)." It may be that some people do not metabolize a-linolenic acid to EPA reliably. Stoll also says that a high EPA to DHA ratio may be important for mood effects, citing patients who have switched to a high DHA fish oil and lost the mood stabilization only to then regain it on switching back to a high EPA fish oil.

He goes on to say in his handout, "I consulted on the the development of an omega-3 supplement called OmegaBrite. This product has been developed specifically for mood elevation and stabilization and has the qualities which I believe make it the ideal omega-3 fatty acid supplement. The characteristics of OmegaBrite include over 92% omega-3 per capsule, over 70% EPA and 10% DHA (plus 12% other omega-3s), tasteless with no, repeat, no cholesterol, and pharmaceutical grade purity, with complete manufacture and encapsulation under nitrogen to prevent oxidation and preserve potency." [Note: I am told that Dr. Stoll has a financial stake in the OmegaBrite product. As an alternative, check ConsumerLab.com's website for a report on independent testing of various omega-3 products for purity and potency. Subscription is necessary for the full report, but three brands that passed muster are viewable for non-subscribers.]

(to table of contents)

Estrogen/progesterone [search]

It is a common observation that the hormonal changes of the menstrual cycle, pregnancy, and menopause can have the effect of either stabilizing or destabilizing mood, but unfortunately the direction of effect is quite unpredictable. As a result, it is hard to generalize about the utility of exogenous hormones in the treatment of bipolar disorder. Selected patients, particularly those with a clear history suggesting worsening course during (e.g.) menopause, may benefit from hormone replacement therapy if this is otherwise not medically contraindicated. Conversely, leuprolide (a GnRH blocker) may worsen the course of bipolar disorder.

(to table of contents)

Tryptophan [search]

This serotonin precursor was used with some success as an augmentor for tricyclics and mood stabilizers many years ago before a trace contaminant in one manufacturer's product resulted in eosinophilia-myalgia syndrome in a number of patients and the FDA banned its sale in the US. It is now legal again but unfortunately only available, as far as I have been able to determine, through compounding pharmacies (requires special licensing). Dosages of 2-8 grams/day are cited in the old literature, and, interestingly, bipolar patients were studied (e.g., TP as augmentor for lithium), without indication of increases in affective switching. TP should not be used with SSRIs, as there have been cases of serotonin syndrome reported with this combination.

(to table of contents)

5-Hydroxytryptophan [search]

More recently touted as a tryptophan alternative, 5-HTP is also a serotonin precursor, and is available in health food stores. Dosage is 150mg/day. Antidepressant effect is claimed, and there is anecdotal confirmation of this clinically. There is, however, no word yet on whether it might promote switching in bipolar patients. Also, there is a recent report of finding in some batches of 5-HTP low levels of the same trace contaminant previously suspected of causing eosinophilia-myalgia syndrome with tryptophan; no reports (yet?) of actual symptoms or adverse reactions. Theoretically, there is a risk of serotonin syndrome if 5-HTP is used in combination with SSRIs, but they have been used together with occasional success according to anecdotal reports.

(to table of contents)

Choline [search]

Stoll reported that 5 out of 6 rapid cycling patients on lithium showed clear reduction in manic symptoms when choline was added to their regimen. It may be that choline promotes lithium transport or utilization in the neuron, and that therefore choline is specifically useful as an augmentor of lithium.

(to table of contents)

Inositol [search]

Inositol (a second-messenger precursor) in doses of 12 grams/day is reported to be helpful for bipolar depression. There were previous reports of efficacy in panic disorder, unipolar depression, and OCD (see Levine 1997). Major drawbacks seem to be GI upset and expense, and there are reports of inositol inducing mania. Sachs says nevertheless it is "impressive" at times.

(to table of contents)

Thyroxine [search]

The relationship between thyroid status and mood disorders is truly an obscure subject. Many studies document a higher than expected incidence of various thyroid abnormalities in depressed and bipolar patients, and the empirical fact is that thyroid augmentation can improve response to antidepressants in unipolar depression and to mood stabilizers in bipolar disorder, perhaps especially in rapid-cycling bipolar disorder, and especially in women. Evaluating thyroid status is a necessary step in successful treatment of any treatment-resistant depressed or bipolar patient. It is especially important when lithium is (or has been) part of the treatment regimen, because of lithium's potential for cumulative toxicity on the thyroid.

However, the mechanisms for thyroid hormone's effects on mood remain largely unstudied. The thyroid axis is a complicated hierarchy of regulatory loops involving multiple levels of the brain and peripheral tissues. The hypothalamus responds to changing T3 and T4 levels by altering TRH secretion, which changes the pituitary's TSH output, which alters the levels of thyroid hormone (TH) secreted by the thyroid. It is generally accepted that, while the thyroid puts out predominantly levothyroxine (T4), it is liothyronine (T3) that is the more active form of the hormone, and that T4 is converted to T3 peripherally. Both depressed and bipolar patients tend to have blunted pituitary TSH response to TRH challenge, but other disruptions of the thyroid axis are possible, and no one knows how common they might be, or how important. In particular, mood-disordered patients may have impaired hypothalamic response to fluctuations in circulating TH, impaired peripheral T4 to T3 conversion, receptor resistance to T3, antibody blockade to TRH or TSH, iodide uptake failure, transthyretin deficiency in transferring T4 into CSF, and/or under-responsiveness of target tissues (specifically limbic regions of the brain) to TH (Klein). The net result can be functionally hypoactive thyroid status, even in the presence of "normal" TSH. Usually there is no hint of the existence of any problem in a routine evaluation, much less where any underlying pathology might be, and the sophisticated testing that might shed light on these matters is impractical.

There are several ways of approaching this Gordian knot. One is simply to augment mood stabilizers with TH empirically, without reference to any laboratory testing. Most clinicians feel more comfortable, however, with some strategy that tries to make sense of the lab tests available. The simplest heuristic approach is to consider augmentation if the TSH is > 2.0, and to titrate the dose of TH to get the TSH below 1.0. Some psychopharmacologists prefer to essentially ignore TSH and measure free T3 and free T4, augmenting if they are not both high normal, and titrating doses so that both are high normal, or even beyond (several studies have looked at augmenting until the free T4 is at 150% of upper limit of normal). See also Bauer et al 2002. Neither approach has any firm basis in the research literature, but both have been discussed in various clinical forums and reports and are defensible current practice.

Another area of controversy is whether to augment with T3, T4, or both. Most of the research on rapid cycling bipolar disorder has used augmentation with T4, while T3 has been more widely studied for augmentation in depression. However, there are many cases in which mood-disordered patients with preexisting hypothyroidism already on T4 replacement have responded to addition of exogenous T3 with improved mood stabilization or resolution of depression (see, eg, Cooke et al 1992 or Bunevicius et al 1999); one would hypothesize that some of these patients might have impaired T4 to T3 conversion. It is probably safer to use either T3 alone or T3 + T4, to avoid this possibility.

It has recently been suggested that relative selenium and possibly zinc deficiency might be implicated in some such cases. Selenium is a trace metal found in T4 deiodinase, which converts T4 to T3; selenium-deficient diets are common in certain areas of the US, and presumably with certain patients in all areas. There are some reports also that adding zinc to the diet of patients with relatively low free T3 normalizes T3 levels and TRH responsiveness. Selenium and zinc supplementation (Se 50-100 mcg/d and Zn 50-100 mg/d) might be considered for those patients who do not respond to T4 augmentation. Both are commonly available in drug stores and are not terribly expensive. Note: multivitamins that contain selenium and zinc often do not contain adequate doses for this purpose (e.g., Centrum Silver contains 15 mg Zn and 25 mcg Se).

A little-known fact with important clinical consequences is that dietary iron interferes significantly with the absorption of exogenous thyroid (see, eg, Shakir et al). Patients should be cautioned not to take any multivitamin containing iron within 4 hours of taking their thyroid. I have seen one case in which starting a multivitamin with iron inadvertently resulted in a TSH of over 20 and a recurrence of dense depression in a previously stable hypothyroid bipolar patient on replacement T4. This phenomenon deserves to be much more widely known and appreciated — even experienced psychopharmacologists are often not aware of it, much less primary care clinicians who do the bulk of thyroid hormone precribing. Tell everyone you know.

The subject of thyroid augmentation is sometimes a bone of contention between psychopharmacologists and endocrinologists or internists, who often object to a patient being on thyroid hormone when there is "no evidence of thyroid dysfunction." The concern is often about the risk of osteoporosis. The literature here is actually not definitive; the studies that show any decrease in bone density in patients on low-dose thyroid are pretty much counterbalanced by the studies that show no decrease at all (see, eg, Bauer, et al for a report on lack of problems even with supraphysiological doses). Don Klein (the "eminence grise" of NIMH) and others on the psychopharmacology mailing list feel that this risk is overblown, and certainly in most cases does not outweigh the benefit of successful control of mood.

(to table of contents)

Benzodiazepines [search]

Both clonazepam and lorazepam have been shown to have antimanic properties and are useful for acute control of mania, often in combination with neuroleptics. They are useful also for the maintenance phase, when they may be used prn for early signs of mania. Benzodiazepines generally, perhaps especially clonazepam, have been reported to promote depression. If a bipolar patient remains depressed while on more than prn benzodiazepines, discontinuing all benzos should be considered. On the other hand, clonazepam was found in one study to have antidepressant effect for bipolar depression, perhaps because of its effect on the anxiety component of dysphoria. There is one study that found alprazolam to have antidepressant effect in 60% of bipolar patients in whom it was tried (though other studies have failed to replicate this), but it may also induce mania, so alprazolam may not be the best choice in a bipolar patient.

(to table of contents)

Melatonin [search]

One study done in Italy suggested that 3mg qhs of melatonin can treat insomnia associated with mania that does not respond to usual hypnotic therapies (benzodiazepines). Interestingly, residual manic symptoms also improved over the course of the month of treatment. Since this pilot study was uncontrolled, delayed response to concurrent mood stabilizers cannot be ruled out, however. Also keep in mind that melatonin has been reported anecdotally to promote depression.

(to table of contents)

Stimulants [search]

For years, the dogma was that stimulants should be avoided for bipolar patients because they will cause mania. Although mood charting sometimes reveals that stimulants will promote cycling, they may occasionally be quite effective in bipolar depression as an adjunct to a mood stabilizer regimen. Some patients, perhaps especially those with comorbid ADD, will in fact have a mood stabilizing effect from stimulants; there are cases in the literature of acute mania responding to stimulants extremely rapidly (see. e.g., Bschor et al). More often, a bipolar II patient with prominent depression will benefit from the addition of a stimulant with diminished anergy; of course, monitoring for a destabilizing effect is necessary, and mood charting here will be useful. Careful and close follow-up of response to stimulants is clearly warranted, but they should not be absolutely shunned, either in depressed or manic phases.

Atomoxetine (Strattera) [search]

Atomoxetine, the newest ADD drug, may not be quite so benign in bipolar patients. It is a selective norepinephrine reuptake inhibitor, with a mechanism of action that is analogous to many antidepressants. In fact, it may have significant antidepressant activity (although it is not appproved for this indication), and early reports suggest a tendency to induce mania.

(to table of contents)

Special cases

Rapid cycling and mixed-state/dysphoric mania

Rapid cycling is defined in DSM-4 as four or more episodes of abnormal mood (depression or mania) within 12 months, demarcated either by partial or full remission for at least two months or a switch to an episode of opposite polarity. Mixed-state or "dysphoric" mania describes those patients with simultaneous symptoms of depression and mania; according to DSM-4, the full criteria for both must be met. A great many more patients will have subsyndromal mixed states, failing to meet criteria for either mania or depression, or both. Many now feel (e.g., McElroy) that the DSM criteria are too restrictive, and that these subsyndromal patients should be treated as having mixed bipolar disorder. Mood in these patients is typically irritable and/or depressed, with other signs of activation (agitation, a feeling of internal pressure, racing thoughts, restless energy).

Mood charting is extremely revealing with regard to cycling patterns, and many dysphoric/mixed-state patients show cycling patterns when moods are charted. If a mixed-state patient is closely questioned, s/he may reveal ultradian (within one 24 hour period) patterns of mood cycling. Many patients who cycle never get into the hypomanic range; if the cycles of waxing and waning depression are more frequent than four times a year, such patients should be considered rapid cycling bipolar patients when they have a history of (hypo)mania.

The single most effective treatment for rapid cycling is to discontinue antidepressants!!!

The best predictor for cycle acceleration is a history of antidepressant-induced mania. All antidepressants can accelerate cycling; bupropion is said to be least likely to do so (actual studies are meager), with MAOIs a close second, but even these agents can be counterproductive. By and large, antidepressants should be used sparingly if at all. Stopping an antidepressant will sometimes have quite rapid effects on settling down mood cycling. Care must be taken, however, to taper (over ten days to two weeks) rather than suddenly stop ADs, since rapid discontinuation also promotes mania. As mentioned above, the trend of the cycling pattern should usually be the basis of decision-making, since decreased cycling will typically precede euthymia. The best early sign of response is not improvement of mood per se but decrease in cycle amplitude and frequency. It may take several months on what will prove ultimately to be the correct regimen for cycling to disappear entirely. Patience in this regard is essential.

Thyroid status should be assessed carefully in rapid cycling patients, as the evidence for the utility of thyroid augmentation is stronger in this subcategory. Hyperthyroid augmentation has been tried in a handful of studies, and could be considered for very treatment-resistant patients. Thyroid doses (T4, with or without T3) can be pushed to the point of tachycardia, obviously with informed consent, and careful assessment of the risk/benefit ratio.

Valproate is the treatment of choice for rapid cycling and mixed or dysphoric manic states. One study suggested that combining valproate with lithium may be useful. Carbamazapine is generally accepted as the second choice agent. (It remains to be seen if oxcarbazepine will prove to be equivalent to CBZ here.) Both gabapentin and lamotrigine have been used increasingly for this as well, with some success (see Calabrese et al 2000), though lamotrigine has been reported to backfire and promote mania. Nonetheless, the Expert Consensus Guidelines now recommend lamotrigine as a first-line option for current depression in the context of rapid cycling. Stoll et al have a study suggesting that lithium plus choline may be effective for rapid cyclers. A calcium channel blocker could be considered. Combination treatment with several mood stabilizers with or without an atypical neuroleptic is recommended for treatment-resistant cases. ECT is sometimes effective.

(to table of contents)

Bipolar depression

The phenomenology of bipolar depression can be exactly the same as that of unipolar depression, though bipolar depression shows a trend towards symptoms of anergia, hyperphagia, and hypersomnia. Abrupt onset and/or termination of depression is quite suggestive of bipolar disease, as is early response (<10 days) to more than one antidepressant. Seasonal pattern of depression is somewhat more associated with bipolar than unipolar depression. The differential diagnosis is usually made by a careful history, including collateral contacts, looking for prior episodes of hypomania; a strong family history of (hypo)mania is suggestive.

The "switch rate" (switching from depression to mania) in bipolar patients on TCA or MAOI alone is 31-70% (various studies). Standard antidepressants tend to promote switching, and often accelerate cycling. Bupropion and MAOIs, and possibly SSRIs, may have a lower rate of switching than TCAs, though when patients switch into mania on SSRIs one report suggests the mania is very severe and recalcitrant. Reports vary as to which SSRIs might be more or less prone to induce switching; paroxetine may be the least likely SSRI to do this, and venlafaxine is a viable choice. However, beware — even bupropion can induce switching on occasion. TCAs have the strongest track record for inducing switching or cycle acceleration, and should almost always be avoided; MAOIs also have a high switch rate. Unopposed antidepressants of any class generally should not be used in bipolar I patients. With bipolar II disorder, the picture is less clear. Several studies suggest that these patients may do well on an antidepressant alone. Care must be taken, however, to monitor for emergence of hypomania, particularly since hypomania tends not to be reported by these patients as a problem. Mood charting should be used in this case, as well as information from collateral contacts.

In this light, maintaining a vigorous mood stabilizer regimen is the cornerstone of bipolar depression treatment. If there is any mood cycling (even if the patient never gets into the hypomanic range) then adding or switching mood stabilizers should be considered. If there is no apparent mood cycling, then adding non-TCA antidepressants (particularly bupropion or paroxetine) should be tried, with caution. Episodes of depression that "break through" vigorous mood stabilizer treatment tend to be more severe and recalcitrant than episodes that occur in the absence of mood stabilizers, and are more likely to require AD treatment, despite its dangers.

Sachs suggests that with patients having a predominantly monophasic depression (repeated depressive episodes with only a past history of manic symptoms) or a "MDE" pattern (mania followed by depression followed by euthymia), the antidepressant should be tapered starting 6-12 weeks following remission, and maintenance AD treatment should only be used if there are repeated recurrences of depression. For those patients with a "DME" pattern, rapid cycling, or history of iatrogenic mania, tapering should start 6-12 days after remission (usually upon the first visit the patient reports resolution of depression), and maintenance ADs should be avoided. Some have questioned this, eg, Amsterdam et al, with studies suggesting that maintenance AD treatment prevents relapse of bipolar depression better than off and on treatment. However, for a critique of this and similar research, see Pies R, "How Should We Treat Depressed Bipolar II Patients?" (July 1999, Psychiatric Times). Pies says about the Amsterdam study, "...the authors candidly acknowledge that 'manic switch episodes were ascertained retrospectively from patient and physician reports of treatment-emergent adverse events,' and that 'specific mania rating scales were not used.' Unfortunately, patients are famously incomplete in their retrospective reporting of hypomanic and manic symptoms." He urges caution in using antidepressants at all in bipolar patients, much less using them unopposed by a mood stabilizer.

On the other hand, Altshuler et al (2001) found that discontinuing antidepressants that had been added to mood stabilizers resulted in a significantly increased relapse rate for depression. Dilsaver et all (1997) found that continuous use of antidepressants while on standard mood stabilizers was not associated with an increased risk of inducing mania.

In short, whether to discontinue antidepressants or maintain them will be a clinical judgment that depends on individual history and requires careful monitoring, since there is no way to predict whether a given patient will require antidepressants to avoid relapse or develop accelerated cycling or dysphoric activation. One suggestion from Post et al (2003) is that if a patient responds to an antidepressant with improved mood and no switching or cycling for 2 months then s/he should be continued on that antidepressant for the longer term; in other words 2 months may be long enough to screen out those who will be destabilized by a given AD. It goes without saying that mood stabilizers should always be on board.

As mentioned above, the patient's thyroid status should always be evaluated, and if there is any hint of functional hypothyroidism augmentation with T3 or T3 + T4 should be considered, especially in female patients.

One good source of information is David Osser's interactive depression treatment algorithm, available online, which has a section on bipolar depression. His algorithm suggests SSRIs and even TCAs more often than I would do so, but it is generally very useful.

Generally, then, the following treatments chould be considered:

(to table of contents)


Akiskal and others have presented data that suggests that a significant proportion of cyclothymics will go on to develop more severe mood episodes and be rediagnosed as bipolar II (or even I). Early treatment is probably worthwhile. Valproate may be the treatment of choice here, and doses of 125-500 mg/day (see, e.g., Jacobson 1993) seem to be effective in many cases, with serum levels of ~35 ug/ml.

(to table of contents)

Pregnancy & breastfeeding

The incidence of mood disruption in pregnancy and the postpartum period is very high. Managing bipolar disorder through pregnancy is a matter of weighing risks and benefits. In considering the risks, it is appropriate to keep in mind (and discuss with the patient) the significant risk of harm to the patient and fetus of a major mood episode, whether depressed or manic. Depending on the patient's clinical condition and circumstances, a number of options are possible.

If medications are to be continued unchanged through pregnancy, fetal monitoring can be done with ultrasound and amniocentesis; this option makes the most sense when the patient is willing to consider abortion if a major malformation is discovered. Folate may be protective with regard to some developmental abnormalities, e.g., neural tube defects with CBZ and VPA.

If medications are to be lowered or discontinued, mood monitoring and frequent follow-up visits should be maintained, and at the first sign of "roughening" (resumption of mild cycling, occasional days of activation or depression) they can be restarted; this is especially true after the first trimester. They should absolutely be restarted just before or immediately after delivery, since the risk of postpartum mania (perhaps including psychosis) without treatment is probably 50-80% (!) for bipolar women with more than one prior episode.

If the pregnancy is planned, and mood stabilizers are to be discontinued, a gradual taper (25% q2-4 weeks) is preferable, since sudden discontinuation can induce mania. This must be weighed, however, against the need to minimize the total time off medications. Past history will influence the risk/benefit ratio. Sachs recommends:

The relative risk of various agents in pregnancy is not firmly established; obviously controlled studies are impossible. By compiling data from various registries, however, Lee Cohen's group at MGH tentatively has concluded: With regard to breastfeeding, obviously the safest thing to do is bottle-feed. Little is known about the effects of long-term early exposure of the developing nervous system to substances that affect neurochemistry. The data on any measurable effects is "dirty," as it is for pregnancy. Small (often just a handful of cases) studies have looked at either breast milk levels or infant serum levels. No firm consensus with regard to safety has emerged. Generally the SSRIs have shown very low levels in breast milk or in the infant. The problem with fluoxetine theoretically is its long half-life; this in combination with the dip in liver CYP system activity in infants between 2 and 4 months of age has caused concern about the possibility of significant fluoxetine build-up in nursing children. One case report found a very high fluoxetine level in a single nursing infant. This observation has never been repeated and it is generally suspected now that this one result was a lab error, and that fluoxetine may be relatively safe for breastfeeding. Sertraline might also be considered. Valproate seems to be secreted in low levels in breast milk, and is regarded as relatively safe; the status of other anticonvulsant mood stabilizers is not so clear.

What is clear is that going off mood stabilizers in order to breastfeed is not an option. Our ability to predict anything in psychiatry is very poor, but this situation is as close as we come to a sure thing: a bipolar woman with two or more prior episodes (and especially if any of them are post-partum episodes) who is not on her medication regimen after delivery will almost certainly relapse, and she may relapse catastrophically.If she decides that the risk of nursing while on meds is unacceptable, her only option is to forgo breastfeeding. To forgo meds is to court disaster.

See also Alan Gelenberg's summary of these issues from 1997.

None of these tentative recommendations should substitute for a full assessment of the relative risks; performing a current literature review should probably be the norm, and a consultation is advisable, as well as coordination with OB/GYN. Full discussion with the patient and informed consent are, of course, mandatory.

(to table of contents)


Feedback is appreciated

Please email me if you have additions, corrections, or suggestions. I'm at:


If you do email me, I'd be interested to know how you found this site...

© 2000-2007 Peter M. Brigham, MD
Last updated 5/20/07


Who am I?

I'm a psychiatrist in private practice in the Boston area, Massachusetts, USA. I received my
medical school training at George Washington University in Washington, DC,
and did my residency at McLean Hospital, Belmont, MA. My specialty is the
psychopharmacological treatment of mood disorders. I have no affiliations
with any pharmaceutical or biotech companies, and thus have no axe to grind
about any of this material.

Number of visitors to this page since 10/20/07: