Highs without lows

The elusive search for unipolar mania

Melancholy and Raving Madness by Caius Gabriel Cibber at Bethlem

One of the most remarkable phenotypes in psychiatry, maybe in the whole of medicine, is bipolar disorder. It is defined by an episodic course with episodes at either ‘poles’ of the mood spectrum.

At one extreme are manic episodes, consisting of elated, euphoric or irritable mood, accompanied by increased drive and racing thoughts. People experiencing mania have decreased need for sleep and increased motor activity. They may behave unlike their usual selves; taking risks, embarking on business ventures or spending excessively. They may exhibit ‘grandiosity’, elevated self-esteem beyond normal limits. They may experience psychosis which typically fits with their mood state. This could manifest as grandiose delusions, such as believing they are a celebrity, a royal or even God himself.

Invariably, a manic episode will bring a person to the attention of the police or health services. The disinhibition and impulsivity makes them extremely vulnerable which can result in detention under the Mental Health Act for their own safety. I have met many people experiencing mania. I can’t think of any cases in which it looked enjoyable. These aren’t pleasurable ‘highs’. They are the kind of episodes that leave you at risk of financial and reputational ruin. Recovering from a manic episode may require rebuilding relationships, careers, or even self-worth.

At the other extreme are depressive episodes. These eventually predominate the clinical picture and are associated with profound debilitation, as well as increased risk of suicide. Symptomatically, they are similar to episodes of major depressive disorder (sometimes called ‘unipolar depression’). Bipolar depression, though, is more likely to include psychotic features and more somatic symptoms like increased appetite and hypersomnia. It has treatment implications too: we worry about antidepressants triggering mania (‘manic switch’) and generally expect a poorer response. Instead, antipsychotics or mood stabilisers are first-line.

We know that depressive episodes can take place in the absence of manic episodes. What has not been recognised in either DSM (North American) or ICD (rest-of-the-world) diagnostic systems is unipolar mania. Does the phenotype of manic episodes in the absence of depression exist? If so, we might have to reconsider whether mania is really synonymous with bipolar disorder.

Kraepelinian dichotomy

Our conceptualisation of bipolar disorder goes back to Emil Kraepelin. Who else? He divided the major psychoses into ‘dementia praecox’ (literally ‘premature dementia’ but what we now recognise as schizophrenia) and manic depressive insanity (which is now called bipolar disorder).

The distinction between these two was the marked progressive decline and cognitive impairment of dementia praecox contrasted with the episodic affective disturbance associated with manic depression. Kraepelin didn’t get everything right, it is still debated whether schizophrenia shows progressive cognitive decline, but for better or worse, his dichotomy persists more than a hundred years later.

The boundary between schizophrenia and bipolar disorder is not clear-cut. There is substantial genetic overlap, with a genetic correlation of around 0.6. While a group of patients exhibit major affective disturbance with more schizophrenia-like features, forming the diagnostic category of schizoaffective disorder.

Nevertheless, the dichotomy does have explanatory power. For example, copy number variants (deletion or duplication of large parts of DNA) are associated with schizophrenia and schizoaffective disorder but not bipolar. It makes sense that these types of genetic changes are implicated in neurodevelopmental problems and cognitive impairment.

While the broad category of bipolar disorder has remained largely intact from Kraepelin’s description, over time and subsequent DSM editions, this has expanded to include what I will call bipolar-spectrum disorders. Notably this spectrum does not include unipolar mania, despite this being first described (yes, by Kraepelin) as periodic mania in 1889.

The bipolar spectrum

Kraepelin’s manic-depression is now known as bipolar type I. A diagnosis requires at least one manic episode but does not require an episode of depression. Bipolar 1 has high inter-rater reliability; as might be expected, after all mania should be a barn-door diagnosis for a competent clinician.

The DSM chapter called Bipolar and related disorders now encompasses much more than the Kraepelinian manic-depression. In DSM-IV, bipolar type II was introduced. Diagnostic criteria requires a depressive episode but accompanied by at least one hypomanic episode (literally meaning below-mania) and no previous episodes of mania.

Hypomanic episodes, by definition, are less severe than manic ones. They might still involve increased drive, energy and risk-taking but without psychosis or an inability to function in society. These episodes probably wouldn’t come to the attention of health services on their own. The concept of bipolar II is controversial and involves recategorising a subgroup of depressive patients as having bipolar disorder, usually by retrospective description of hypomania. It is thus much more subjective, with lower reliability than bipolar I.

Surprisingly to me, depressive episodes massively outnumber hypomanic episodes, at a ratio of 39:1. In some ways, you might think of bipolar II as a specific type of depression. One in which a history of hypomania dictates treatment choices. Another disorder included in this spectrum is cyclothymic disorder. For this diagnosis, there has to be fluctuation in mood but not enough to meet criteria for clear-cut hypomanic or major depressive episodes.

While the diagnosis of bipolar has undergone expansion over time, it has not included the subtype of unipolar mania. Patients experiencing mania will be classified as having bipolar disorder whether or not they have experienced episodes at the depressive pole. Whether this is warranted rests on mania always heralding an episodic illness encompassing both poles of the mood spectrum.

Defining unipolar mania

Simply put, unipolar mania is at least one manic episode occurring in the absence of depressive episodes. However, to be scientific, we need a more precise definition. If we want to show that manic illnesses can take place without depression, we should exclude mania which is secondary to other causes. This might include drugs, either prescription (the textbook example is corticosteroids) or recreational. We should exclude an underlying medical/neurological condition (this should be suspected if the first episode presents late in life).

Furthermore, we should probably require substantial follow-up before declaring that the person doesn’t experience depressive episodes. After all, a proportion of those with a bipolar illness course (up to 40% in some studies) will first present with mania, later followed by depressive episodes. We should stipulate that manic symptoms can’t be as part of a schizophrenic or schizoaffective illness, as these conditions do not require distinct depressive episodes. Episodes which include both manic and depressive symptoms, known as mixed affective states, should be excluded too.

When taking all of the above into account, and bearing in mind that bipolar I is relatively rare (~0.5% global prevalence), finding reliable estimates of mania in the absence of depression is a challenge that will require large sample sizes.

Searching for unipolar mania

There have been various efforts to identify cases of unipolar mania. One is to simply ask patients diagnosed with bipolar I whether they have had previous depressive episodes. The problem with this approach is that retrospectively reporting anything to do with health is unreliable. When cases are defined in this way and then prospectively monitor them, up to 80% go on to develop at least mild depressive episodes over a 20 year follow-up.

Another method is record-linkage, whereby the hospital records containing diagnostic codes and prescriptions are obtained, often for a whole country. This was done in Taiwan and found that 14.9% of patients with bipolar 1 had no depressive episodes and no prescription of antidepressants. However, this kind of study relies on (i) depressive episodes coming to the attention of health services, (ii) depressive episodes being correctly coded on the system, (iii) antidepressant medication being used to treat bipolar depression. I think each of these assumptions is questionable.

Another approach is to search electronic health records of patients with bipolar disorder. This is a slightly enhanced version of a simple record-linkage study. Not only can researchers search for patients who have not had depressive episodes coded on their health record, they can search the text of patients de-identified medical notes for mention of depressive symptoms.

This method was used by a team at the Maudsley, led by Paul Stokes and Allan Young. They identified records of patients who had a diagnosis of mania or bipolar but never a diagnosis of schizophrenia, schizoaffective disorder or depressive episode. They then searched the text of these notes for words associated with depression. To ensure an adequate length of time was available to develop depressive episodes, patients had to have at least 10 years of notes available. The team identified 17 patients who experienced manic episodes, in the absence of depressive episodes. This represented 1.2% of 1458 patients with bipolar 1. Put differently, 98.8% of patients who experience a manic episode also suffer a depressive episode.

While I expect this proportion will be refined by future replications in larger electronic health record samples, this is the most reliable estimate that I’ve come across to date.

Nothing is 100%, but…

It is a truism that nothing in medicine can be said with 100% certainty. We have all heard stories of cancer going into spontaneous remission. Or of patients being cured by unconventional treatments. Psychiatry is particularly bad at prediction when it comes to prognosis. So it’s pretty remarkable that we can say with ~99% certainty that if someone experiences an episode of mania, they will experience an episode of depression at some point as well.

It is astounding that we can tell people who have experienced a manic episode - we know that around 99% of people who have experienced an episode like yours will experience episodes of depression too. This has massive prognostic and treatment implications.

You may quibble with Stokes and Young’s estimation; it is certainly lower than other (though I’d argue less rigorous) studies. However, it’s also possible that their 98.8% would creep toward 100% using a longer timeframe of say 20 or 50 years, or using stricter definitions of mania to exclude any evidence of schizophrenia. It’s possible that some of that 1.2% did experience mild depressive episodes that just weren’t recorded in their health record. Equally, it’s possible that initiatives to improve services for first episode mania might reveal different patterns of illness when patients are systematically and periodically assessed after an index manic episode.

For now though, there is no need for an urgent DSM revision.

Depressive episodes are ubiquitous in patients who experience mania.

The concept of bipolar disorder is robust.