Bipolar disorder

Bipolar disorder involves episodes in which mood, energy, activity and thinking change substantially. Mania can bring unusually elevated or irritable mood, reduced need for sleep, accelerated thinking and behaviour that carries serious consequences; some episodes include loss of contact with reality. Depression can bring profound low mood, loss of interest, slowed thinking and difficulty managing everyday life. Symptoms and recovery differ greatly between people. A diagnosis does not mean that someone continuously alternates between opposite moods. National Institute of Mental Health explanation of bipolar disorder, clinical overview in The Lancet.

Bipolar I requires a history of mania. Bipolar II involves hypomania, a less severe form of elevated or irritable activation, together with major depression and no history of mania. “Less severe” describes the hypomanic episode, not the person's overall illness: depression and disruption in bipolar II can be substantial. These distinctions matter because research and treatment recommendations do not apply equally to every form of bipolar disorder.

The best-supported account combines inherited susceptibility with changes in sleep and daily rhythms, experiences and stresses, and several interacting brain and bodily systems. There is no established single chemical imbalance that explains every case. Nor does knowing which medicine helps establish exactly what originally caused the illness. Treatment can be effective while the complete chain from susceptibility to an episode remains uncertain.

Inherited susceptibility does not mean a predetermined future

Family and twin studies provide strong evidence of inherited susceptibility. Heritability estimates are often around 60–80%, depending on the population and method. This is a statistical description of variation within populations, not a statement that 60–80% of one person's illness is genetic. It does not give a child's probability of developing the disorder or measure how changeable its course is. Review of bipolar genetics.

Most susceptibility involves many genetic differences, each contributing a small amount. A 2025 international study compared 158,036 people with bipolar disorder with approximately 2.8 million comparison participants. It identified 298 associated regions of the genome and prioritised 36 genes for further investigation. A genomic region is not the same as a gene, and an association does not identify a proven causal mechanism. The results point toward processes involved in communication between nerve cells, their electrical responsiveness and the regulation of activity. Some findings implicate cells using GABA, a messenger that generally restrains neural activity; they do not show that everyone with bipolar disorder has a simple GABA deficiency. Large genetic study of bipolar disorder.

Rare variants can have larger effects. Research on AKAP11 and HECTD2 implicates systems that regulate cellular signalling and proteins, including connections with a pathway influenced by lithium. This provides a useful biological lead, but most cases are not single-gene disorders. A connection between a risk gene and a drug target does not establish why that drug works for every responder or identify who should receive it. Study of rare variants in AKAP11 and HECTD2.

Genetic boundaries also overlap diagnostic boundaries. Bipolar I tends to share more inherited susceptibility with schizophrenia, while bipolar II has stronger overlap with major depression. These are population patterns, not clean divisions. Results depend partly on how participants were recruited and which ancestries were represented. Scores combining many genetic variants remain research tools; they cannot currently determine an individual's diagnosis, personal cause or inevitable future.

Experiences, susceptibility and episode triggers are different questions

An experience can contribute to long-term susceptibility, help trigger an episode in someone already vulnerable, or result from an emerging episode. These possibilities are easily confused. For example, a period of disrupted routines may precede mania, but early manic changes may also be responsible for the disruption.

Childhood maltreatment is consistently associated with bipolar disorder. A 2025 review comparing 70 previous reviews across bipolar, depressive and schizophrenia diagnoses found that this was a shared association rather than a bipolar-specific cause. Many underlying studies had methodological weaknesses, including retrospective reporting and difficulty separating linked family and social influences. Adversity deserves attention in care, but it is neither necessary nor sufficient for bipolar disorder, and an association cannot establish the cause of one person's illness. Review of environmental risks across diagnoses.

Cannabis use is associated with later manic symptoms and bipolar diagnoses. A review published online in October 2026 combined eight longitudinal studies of bipolar onset and found an unadjusted risk ratio of 2.46, with an uncertainty range of 1.57–3.84. In those studies, the cannabis-exposed group had roughly two and a half times the observed risk. “Unadjusted” matters: shared susceptibility, other substance use, adversity and early symptoms could explain part of the association. Cannabis is a relevant risk marker and plausible contributor, but that estimate is not a measurement of its isolated causal effect. Longitudinal review of cannabis and subsequent bipolar outcomes.

Stressful events, some medicines and disrupted daily routines can be associated with episode onset. Much of the trigger literature relies on small studies or case reports, so timing alone is not proof. Antidepressants deserve particular attention because they can be followed by activation in some circumstances, but the risk varies and is not inevitable. Assessment must also consider substances and physical conditions, including thyroid disease, that may produce or complicate mood symptoms; not every such episode has the same explanation as bipolar disorder. Review of possible episode triggers, NICE assessment guidance.

Sleep and daily rhythms connect vulnerability with everyday life

Sleep, activity and the body's roughly 24-hour timing system are closely linked to bipolar episodes. Sleep loss can destabilise mood. Reduced need for sleep can also be an early sign of mania, while an episode can further disrupt routines and prolong difficulties. This two-way relationship explains why monitoring sleep can be useful without proving that a faulty body clock is the single original cause.

In 2026, 30 specialists from 15 countries reached formal agreement on the clinical importance of sleep, timing and related treatments. This records actual expert consensus, but expert agreement is different from experimental proof. Better causal evidence requires repeated measurements before, during and after episodes, ideally showing whether changes in timing precede mood changes within the same person. International expert consensus on sleep and daily rhythms.

The period after childbirth is a particularly important risk period for people with bipolar disorder or a history of postpartum psychosis. Biological changes, sleep disruption and individual susceptibility may interact. This does not mean that all new mothers are at similar risk, or that one hormonal change explains every episode. It supports advance specialist planning for those with an established history. Review of postpartum relapse.

What brain and cell studies can tell us

Networks and chemical messengers

Brain-imaging research points to differences across connected systems involved in emotion, motivation, attention and control. A review of 205 functional-imaging publications found that results depended on the task and clinical circumstances. A brain response during emotional processing may differ from one measured during attention or rest. Findings cannot be reduced to one permanently overactive emotional centre or one underactive control region. They describe group patterns with considerable individual overlap. Large review of functional brain imaging.

Dopamine helps the brain assign importance to events, anticipate rewards and organise action. It is a plausible part of the explanation for increased motivation and the unusual significance that experiences may acquire during mania. Medicines affecting dopamine can treat mania, giving this system clear treatment relevance. That still does not establish a simple model in which mania always means “too much dopamine” and depression “too little”, or show that the medicine reverses the original cause. Review of the dopamine hypothesis.

Glutamate, another major messenger, is also implicated. Across 40 studies involving 1,135 people with bipolar disorder and 964 comparison participants, brain spectroscopy found higher combined glutamate-and-glutamine concentrations by about 0.42 standard deviations in the anterior cingulate, a region contributing to emotion and control. A standard deviation describes the spread of measurements, allowing differences to be compared across studies. The finding was especially evident in adults during depression. This technique measures chemical concentrations in tissue, not moment-to-moment communication between nerve cells. The figure does not mean a 42% excess, and it cannot establish a universal imbalance. Review of brain glutamate-related measurements.

Calcium signalling and cellular energy

Calcium inside cells helps regulate electrical activity, the release of chemical messengers and adaptation to repeated activity. Genetic findings and laboratory studies both point toward calcium regulation as a useful line of investigation. A review of 32 studies, 21 of which could be statistically combined, found moderately higher calcium measurements at rest and after stimulation. Most measurements came from blood cells, such as platelets and lymphocytes, rather than living human neurons. This concerns signalling inside cells; it is not evidence that bipolar disorder is caused by too much dietary calcium or an abnormal routine blood calcium level. Review of cellular calcium in bipolar disorder.

Mitochondria are structures that supply cells with usable energy and help regulate calcium. In a 2026 laboratory study, neurons grown from cells donated by lithium-responsive patients showed unusually mobile mitochondria and reduced retention of these energy sources near sites where nerve cells communicate. During activity, this could make energy supply and signalling less stable. Mouse experiments involving AKAP11 reproduced aspects of the problem, and lithium reversed some abnormalities in experimental models. Experimental study of mitochondria and neural communication.

This kind of work moves beyond simply observing a difference: manipulating a system can help test whether it contributes to a cellular problem. Nevertheless, cultured cells and mice do not experience the full human disorder. Results from particular donor samples do not show that every person has the same mitochondrial difficulty, that it explains every episode, or that it predicts an individual's treatment response.

Inflammation and the stress response

On average, some studies find higher blood levels of substances associated with inflammation, including C-reactive protein, interleukin-6 and tumour necrosis factor. Results vary with mood phase and other circumstances. For C-reactive protein, a synthesis of 37 studies estimated a difference of 0.70 standard deviations, with a broad uncertainty range of 0.31–1.09. This is a comparison relative to the variation in measurements, not a 70% increase. Some associations remain after accounting for body mass, so obesity alone is not a complete explanation. Review of inflammatory markers across mood states.

Blood inflammation does not automatically demonstrate inflammation within the brain or identify the direction of cause. Sleep disruption, smoking, metabolic health, infection, medicines and the consequences of episodes can all influence these measurements. The findings do not establish bipolar disorder as one autoimmune disease or justify treating everyone on the assumption that inflammation is its cause.

The body's stress system provides another possible connection. A review of 41 studies found higher cortisol and ACTH levels, hormones involved in coordinating responses to challenge. Mood state and medication influenced the results. As with inflammation, a measurement made during established illness cannot determine whether it preceded the illness, helped maintain it or followed it. Review of stress-hormone regulation.

Sleep regulation, calcium signalling, cellular energy, inflammation and stress responses can interact. It is reasonable to investigate these connections. It is not yet justified to present them as one demonstrated sequence that starts with a gene and ends in mania or depression for everyone.

Illness course and the limits of biological tests

Symptoms often become apparent in late adolescence or early adulthood, but this does not point to one shared developmental injury. The course varies. Some people experience repeated severe episodes or lasting cognitive difficulties; others remain stable for long periods. Whether episodes themselves contribute to later biological changes is an important question, sometimes described as “neuroprogression”. The evidence does not establish inevitable brain degeneration.

A study following 307 people with bipolar disorder and 925 comparison participants found that more intervening manic or hypomanic episodes were associated with greater thinning in parts of the frontal brain. However, the bipolar group did not show generally accelerated thinning of the brain's outer layer. Faster enlargement of fluid-filled brain spaces was observed, but observational findings still cannot separate the contributions of episodes, treatment and other differences. Longitudinal international brain-imaging study.

A separate review of ten studies followed 540 people with bipolar disorder and 644 comparison participants for an average of 8.9 years. It found no significantly greater overall decline in thinking-test performance in the bipolar group. Difficulties in particular subgroups remain possible, and loss of participants during follow-up can affect results. The conclusion is that decline is not inevitable, not that nobody experiences it. Review of long-term cognitive change.

Diagnosis still depends primarily on the history of episodes, functioning, family history, physical health, substances and changes in sleep and activity. A scan, hormone measurement or genetic score does not explain the individual cause. A 2026 study of EDIT-B, a test combining blood RNA-editing measurements with clinical information, externally assessed its ability to distinguish bipolar disorder from major depression in 393 adults currently experiencing depression. RNA editing concerns changes to molecular messages made from genes. The study is promising for a particular diagnostic problem, but its selected population does not establish a general screening test, prove causation or remove the need for clinical history. Broader prospective validation, including first presentations, remains important. External European validation of EDIT-B.

Treatment depends on the phase of illness

Treatment has three related tasks: settling mania, relieving bipolar depression and preventing further episodes. A medicine may help with one task much more than another. Decisions also depend on previous response, adverse effects, physical health, the pattern of episodes and the person's preferences. The evidence below mainly concerns adults. Children, pregnancy and complex medical circumstances require separate consideration. A 2026 umbrella review brought together 77 reviews covering 116 interventions, but its underlying search ended in November 2024; subsequent studies add information rather than making the whole field uniformly up to date. Broad review of bipolar treatments.

Treating mania

Medication is the main established treatment for acute mania. Evidence supports several antipsychotic medicines, including quetiapine, risperidone, olanzapine, aripiprazole, asenapine and cariprazine, as well as lithium and valproate. “Antipsychotic” is a drug-class name; these medicines can treat mania even when psychosis is absent. A review of 72 double-blind trials involving 16,442 participants supports several effective options, but the average trial lasted only about four weeks. Short-term symptom control is not the same outcome as long-term recovery. Comparative review of medicines for mania.

Severe mania, psychosis or major behavioural disturbance may require an antipsychotic combined with lithium or valproate, balancing additional benefit against a greater burden of adverse effects. Short-term sedating medicines such as benzodiazepines may help manage agitation or insomnia within clinical care. Hospital treatment may be needed when safety, judgement or basic care is seriously compromised. Lamotrigine is not an effective treatment for acute mania. NICE bipolar treatment guidance, CANMAT and ISBD guideline summary.

Treating bipolar depression

Bipolar depression requires evidence specific to bipolar illness. A review of 101 trials involving 20,081 participants found moderately confident support for seven medication options: olanzapine combined with fluoxetine, quetiapine, olanzapine, lurasidone, lumateperone, cariprazine and lamotrigine. Average differences from placebo ranged from about 0.16 standard deviations for lamotrigine to 0.41 for olanzapine–fluoxetine. These are small to moderate average effects, not guaranteed individual outcomes or a reliable ranking for every patient. Side effects and the way trials selected participants affect the comparison. Comparative review of bipolar depression medicines.

Quetiapine has consistent support across major guidelines. NICE includes quetiapine or olanzapine–fluoxetine among options for moderate or severe bipolar depression in someone not already taking bipolar medication; existing treatment changes the decision. Sedation, weight gain and effects on metabolic health can materially affect suitability. A 2026 update identified 24 additional trials, but only seven had published results suitable for analysis; ongoing or unavailable studies cannot be treated as established evidence. NICE recommendations, 2026 update on bipolar depression treatment.

Bipolar II has fewer dedicated trials. CANMAT and ISBD identify quetiapine as the only first-line option specifically for bipolar II depression, while considering lithium, lamotrigine and other options at different evidence levels. Findings drawn mainly from bipolar I should not simply be assumed to apply identically to bipolar II. CANMAT and ISBD guideline summary.

Guidelines also differ over lamotrigine for an acute depressive episode. CANMAT and ISBD include it among first-line bipolar I options, while the US Veterans Affairs/Department of Defense guideline considers the evidence insufficient to recommend it alone for acute depression and supports selected combinations. The disagreement reflects its small acute effect, differences between improvement and full remission, and different weighting of tolerability and longer-term usefulness. Its role in preventing depressive recurrence is clearer than its ability to rapidly relieve a severe current episode. CANMAT and ISBD summary, VA/DoD bipolar guideline.

Preventing further episodes and supporting daily life

Lithium remains central to long-term treatment and is NICE's first-line maintenance option, with particularly strong protection against mania. Other effective options have different strengths. Quetiapine can help prevent both manic and depressive episodes. Lamotrigine is more useful for preventing depressive than manic recurrence. Some antipsychotics, including long-acting forms of aripiprazole and risperidone, are particularly relevant to preventing mania. Selected combinations can help when one medicine is insufficient. These are choices within a treatment plan, not interchangeable versions of one effect. NICE guidance, review of maintenance medicines.

A 2026 analysis of 44 trials involving 10,867 participants broadly supported existing maintenance guidance, while highlighting differences in dose, population and trial design. Many trials first stabilise people on a medicine and then compare continuing it with stopping or switching. This answers an important question for people who already benefit and tolerate it, but can make results look more favourable than they would in an unselected group starting treatment. Updated maintenance-treatment analysis.

Across 22 trials involving 5,462 stable, previously treated participants, continuing medication reduced six-month recurrence relative to discontinuation: the risk ratio was 0.61, with an uncertainty range of 0.54–0.70. This means an average relative reduction of about 39%. In those trial populations, approximately five people needed to continue treatment to prevent one additional recurrence over that period. It is not a lithium-only estimate, a result for every newly diagnosed person, or a reason to change medication without a planned clinical review. Review of continuing versus discontinuing treatment.

Structured psychological care adds benefit to medication. Psychoeducation helps people understand the illness and recognise early warning signs; cognitive behavioural therapy examines unhelpful patterns of thinking and behaviour; family work can improve communication and shared responses to difficulties. A review of 39 trials involving 3,863 participants found lower recurrence in the usable direct comparisons, with an odds ratio of 0.56. An odds reduction is not the same as a risk reduction. As an illustration, if recurrence without the additional therapy were 40%, that odds ratio would correspond to about 27%, or roughly 13 fewer recurrences per hundred people. Those percentages illustrate the calculation; they are not pooled event rates from the review. Review of psychological treatment alongside medication.

Psychological treatment commonly addresses sleep, regular routines, stress, relationships and early action when symptoms emerge. Interpersonal and social rhythm therapy brings relationships and daily timing together, although precise comparisons with other therapies are less certain. Help with substance use, physical health, employment, housing and family strain can be important to maintaining stability. These supports address the person's life as well as symptom scores.

Severe or resistant episodes require specialist options

Electroconvulsive therapy, or ECT, is an established specialist treatment when bipolar depression is severe, has resisted other treatment, or requires a rapid response. Examples include catatonia, in which movement and responsiveness can become profoundly disturbed, serious medical compromise or substantial suicide risk. ECT can also be considered for severe resistant mania. It is given under anaesthesia and requires discussion of benefits, alternatives and memory effects. Short-term memory problems are common during treatment, and some gaps in autobiographical memory can persist. VA/DoD guideline, Royal College of Psychiatrists explanation of ECT.

In a trial of 73 inpatients with resistant bipolar depression, ECT produced greater symptom improvement than a structured medication strategy. Among participants completing treatment, 17 of 23 receiving ECT responded, compared with 7 of 20 receiving medication. Full remission was less clearly different: 34.8% versus 30%, a statistically inconclusive result. The large number missing from the completer comparison limits certainty. Meaningful improvement and complete resolution of symptoms are different outcomes. Randomised comparison of ECT and medication.

Newer options have promising but more limited evidence. A September 2026 trial assigned 68 adults with resistant bipolar depression to intravenous ketamine or midazolam, a comparison sedative, alongside existing medication. The final efficacy analysis included 63 people after five withdrew. At day fourteen, ketamine produced an additional improvement of 7.3 points on a standard depression-rating scale, with an uncertainty range of 2.5–12 points. This small, short trial cannot establish lasting benefit, relapse prevention or uncommon harms, including the longer-term risk of mood destabilisation. Trial of repeated ketamine infusions.

Repetitive transcranial magnetic stimulation uses magnetic pulses to influence activity in targeted brain regions. A 2026 review found a small average benefit for bipolar depression, but results did not remain consistently convincing when the researchers varied their analytical assumptions. Its place is therefore more conditional than that of established medicines. Clinically supervised bright light therapy also has support as an addition for acute depression, with monitoring for activation and antimanic protection particularly in bipolar I. Whether it prevents future episodes remains unestablished. Review of magnetic stimulation, specialist recommendations on light therapy.

Safety is part of effectiveness

A treatment's average benefit matters only if it can be used safely and sustained. Lithium requires blood-level checks and monitoring of kidney function, thyroid function and calcium. The gap between an effective and a toxic level can be narrow; dehydration and interactions with other medicines can increase toxicity risk. Monitoring, clear information and prompt clinical assessment of suspected problems are integral to treatment. This is separate from the research on calcium inside cells and does not imply that the illness is a lithium deficiency. NHS Specialist Pharmacy Service lithium monitoring guidance.

Antipsychotic choices require attention to sedation, weight and metabolic effects. Lamotrigine needs gradual dose increases, and a new rash requires prompt assessment because some reactions can be serious. Relatively small differences in average trial efficacy may be outweighed by these practical considerations. VA/DoD bipolar guideline.

Valproate has major reproductive safety restrictions in the UK. For a first prescription in someone under 55, current rules generally require two specialists independently to document that alternatives are ineffective or not tolerated, or that compelling circumstances mean the reproductive risks do not apply. Additional pregnancy-prevention requirements apply to women and girls who could become pregnant. Precautions also concern men and their partners, although the evidence behind paternal-risk concerns is less certain than the established harms of exposure during pregnancy. These are rules about prescribing and review, not a reason to stop existing treatment abruptly; the official guidance advises discussing changes with the treating clinician. MHRA guidance on valproate reproductive risks.

Antidepressants and lithium need careful interpretation

Antidepressants alone are not recommended for bipolar I depression. In selected circumstances, an antidepressant may be added to an appropriate bipolar treatment after preferred options have been insufficient. Particular caution is needed when depression includes manic symptoms, when previous antidepressants have been followed by activation, or when episodes occur frequently. Some guidelines allow more selective antidepressant use in bipolar II depression without mixed symptoms. CANMAT and ISBD guideline summary, 2026 clinical guideline update.

This does not mean that antidepressants invariably cause mania. A 2025 review of 13 trials involving 1,362 participants found no statistically significant increase in switching for an individual antidepressant compared with placebo. However, confidence was low, the estimates were imprecise and evidence about use without accompanying bipolar treatment was limited. Venlafaxine produced the most concerning signal. Failure to detect a clear increase does not establish an absence of risk. Review of antidepressants and switching into mania.

Lithium's possible protection against suicide also needs a distinction between different forms of evidence. A Swedish observational study followed 51,535 people and found fewer suicide-related events during lithium-treated periods than during untreated periods. Comparing the same person across time removes some differences between individuals, but illness severity, adherence and clinical contact can change at the same time as treatment. Swedish study of lithium and suicidal behaviour.

A 2025 analysis of 15 randomised trials involving 3,036 participants produced an estimated suicide odds ratio of 0.61, but with an uncertainty range from 0.25 to 1.48. That range includes substantial benefit, no benefit and possible harm. Suicide is a rare outcome in individual trials, making reliable estimates difficult. Lithium's specific suicide-prevention effect therefore remains less certain than its established ability to prevent mood episodes. Suicide prevention cannot be reduced to choosing one medicine; risk assessment, ongoing care and a practical safety plan remain important. Review of randomised lithium evidence on suicidal outcomes.

What further research needs to resolve

Research on causes often measures people after years of illness and treatment. Differences may reflect original susceptibility, an active episode, adaptation, medication or accumulated consequences. Studies that mix mania, depression and stable mood can blur these possibilities. Large samples improve precision but do not automatically solve recruitment bias, limited ancestry representation or uncertainty about timing. Laboratory models can identify plausible mechanisms without establishing that the mechanism explains a person's illness.

Treatment studies have different limitations. Acute trials are often brief and do not settle long-term functioning, recurrence or uncommon harms. In one major maintenance review, 25 of 41 trials selected people who had already responded to the study medicine, and 29 had industry sponsorship. This does not invalidate the results, but selection, reporting and independent replication need scrutiny. Comparisons across networks of trials depend on whether participants, previous treatment, starting mood phase and outcome definitions are sufficiently similar. Missing participants, unpublished results and repeated inclusion of the same original trials in several reviews also limit certainty.

The most useful next studies would repeatedly measure sleep, activity and biological changes before and through episodes; test each proposed connection from genes to cells to symptoms; and examine whether a proposed biological subgroup predicts benefit from a specific treatment in new patients. Better representation of bipolar II, younger and untreated people, non-European populations and people with complex accompanying conditions is needed. Long-term direct treatment comparisons should measure relationships, work, physical health and adverse effects as well as symptom scores. Ketamine, magnetic stimulation and other newer approaches need stronger evidence about durability and mood destabilisation, while suicide outcomes require very large, carefully designed studies.

None of these uncertainties removes the value of established care. The broad evidence supports treatment matched to the current phase and previous course, combined with psychological and practical support, attention to sleep, appropriate monitoring and planned review. It supports neither a single-cause story nor a promise that one treatment will suit everyone. The aim is sustained safety, stability and a life the person can recognise as their own.