Bipolar II Disorder Is NOT a Myth

The challenge to define and refine psychiatric diagnostic categories has led researchers to a healthy debate about the usefulness or lack thereof of particular diagnoses, especially bipolar II disorder.¹ The National Institute of Mental Health (NIMH) and others have challenged the overall validity of the Diagnostic and Statistical Manual of Mental Disorders rubric, at times dismissing it as an arbitrary set of symptoms, observations, severity, duration, and impairment, which fail to “carve nature at its joints.” As an alternative, NIMH proposed and implemented the multidimensional Research Domain Criteria (RDoC), a translational project to better understand psychopathology from molecular to cellular to animal model to clinical phenomena.² Yet, after considerable investments, RDoC has yet to deliver on its promise as a top-down deterministic paradigm to uncover previously unknown pathophysiology to supplant the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). Ultimately, the overarching project of understanding the physiologic basis of complex psychiatric disorders and developing better treatments, including bipolar disorder, will most likely be more difficult than anyone had anticipated and will need to use all of the tools available, including categorical and dimensional approaches when appropriate to the task at hand.

Which brings us back to the controversy around the diagnosis of bipolar II disorder. Ultimately, while reversing the causal chain of research from molecules to diagnoses implied by the RDoC, will the diagnosis of bipolar II disorder prove pathophysiologically or clinically distinct from bipolar I? The history of how the diagnosis of bipolar II was established and defined is well covered elsewhere.¹ Hypomanic episodes, rather than manic episodes, distinguish bipolar II from bipolar I. Malhi and colleagues¹ are correct in pointing out the vagaries of how the DSM-5 and International Classification of Diseases, 11th Revision (ICD-11) define hypomania—and one could argue that hypomania differs from mania as a matter of degree (severity, duration, and impairment, along with the presence or absence of psychosis).³ More important, episodes of major depression and subthreshold depressive symptoms appear to be worse in bipolar II versus bipolar I. In the now-classic naturalistic longitudinal Collaborative Study of Depression, Judd and colleagues^4,5 found that people with bipolar II spent over 50% of their time with depression compared to those with bipolar I who spent about 30% of time. Malhi et al.¹ argue that this distinction should be folded into a spectrum of bipolar disorder rather than a categorical difference, letting the data speak for themselves (i.e., just like light, bipolar disorder should be considered a “wave” rather than distinct “particles”). I argue that, like light, one should consider bipolar disorder as a spectrum and as distinct categories depending on the purpose, be it exploring pathophysiology, longitudinal course, or response to treatment.

Seeking Definition

Malhi et al.¹ argue that no biological factors separate bipolar I from bipolar II disorder—but this argument ignores our ignorance about bipolar disorder overall (i.e., we do not know the fundamental pathophysiology of bipolar disorder).⁶ If we don’t know the fundamental pathophysiology, how can the lack of biological factors between bipolar I and bipolar II be considered a problem in the legitimacy of bipolar II as a distinct diagnosis? In contrast, Judd et al.^4,5 show different longitudinal courses. Should this distinction be considered a mere spectrum of bipolar disorder? If so, how would one go about trying to define the spectrum if the longitudinal course of bipolar II disorder has shorter periods of hypomania and a greater burden of depression? And if some people with hypomania thrive during their periods of hypomania,^7,8 how would that be taken into account? Here, again, perhaps we should go beyond the debate of spectrum versus category and go back to the data. As shown by Kohler-Forsberg and colleagues,⁹ trajectory analyses can use longitudinal data to construct clinically useful categories of bipolar disorder (in these studies, they focused on suicidal ideation and depression), which can help with prognosis and lead to clinically useful studies (e.g., treatments for those with persistent symptoms and disability).¹⁰ It is not a question of whether or not we should pathologize periods of increased productivity, as argued by Malhi et al.,¹ but instead we need to define these “good” periods followed by difficult impairing periods of depression. What, then, is the best treatment to prevent or manage those subsequent depressive episodes?

Malhi et al.¹ also argue that bipolar II can overlap with borderline personality disorder and that this somehow negates the legitimacy of the diagnosis of bipolar II. However, borderline personality disorder is characterized by an early onset of persistent dysregulations of affect, irritability, impulsiveness, interpersonal relationships, and distress intolerance along with functional impairment and a high risk of suicide,¹¹ while in contrast, the hypomania of bipolar II is episodic with periods of increased energy plus the concomitant other characteristic symptoms distinct from one’s normal self. No diagnostic system requires that one must have one or another diagnosis; they can co-occur. Correct diagnosis is critical for effective treatment planning, and nothing can substitute for a skilled systematic and meticulous longitudinal history confirmed by informants. Otherwise, Malhi and colleagues¹ are correct that the wrong diagnosis can lead to the wrong treatment.

Malhi et al.¹ argue that expanding the diagnosis of bipolar disorder to include bipolar II has led to nefarious responses by pharmaceutical companies to increase sales of their products and has led to an overdiagnosis of bipolar disorder in adults, adolescents, and children in the United States. Yet, this argument ignores the epidemiological data about the prevalence of bipolar disorder (as high as 4.5% of adults if the full spectrum is included and 4% of adolescents and children).^12–14 These data bring up the following question: overdiagnosis compared to what? It is highly likely that some people are misdiagnosed as having bipolar II who do not have it. But the bigger problem is underdiagnosis and undertreatment as shown by epidemiological data, which showed that 38% of youth with bipolar disorder did not receive any treatment.¹⁵ Furthermore, if, as Malhi et al.¹ argue, bipolar II should be subsumed into a spectrum of true bipolar,¹⁶ then what is the problem if physicians are appropriately treating a spectrum of bipolar disorder?

As for the argument that the distinction between bipolar II and bipolar I is without a difference in pathophysiology, genetics, brain function, and so on, here, again, is a problem. To reiterate: while multiple reviews list the multiple studies showing physiological differences between those with bipolar disorder compared to healthy controls,¹⁷ no overarching pathophysiology of bipolar disorder and no cause of bipolar disorder have yet to be identified. Yet, a recent study of the psychiatric disorders of the offspring of parents with bipolar I and bipolar II did find a clear difference: offspring of bipolar I parents were more likely to have major depressive disorder and higher scores on attention-deficit/hyperactivity disorder (ADHD) and trauma scales.¹⁸ Whether these differences are due to differences in the spectrum or categories of the parental diagnoses may be irrelevant.

Malhi et al.’s final argument¹ is that the diagnoses are not separated by response to medication, a distinction that is akin to Quitkin and colleagues’ attempt¹⁹ to “pharmacologically dissect” typical from atypical depression. Yet, surprisingly little is known about the optimal treatment of those with bipolar II disorder. Should they be on long-term lithium or other mood stabilizers? Should they be treated with antidepressant monotherapy? What are the benefits and risks of dopamine-blocking agents for short- and long-term treatment? Again, it may be irrelevant if one considers bipolar II disorder as a dimensional spectrum of bipolar disorder or as a separate category—those questions still need clinical answers.

A New Dimension

Observation and bottom-up science have a long and hallowed tradition as, for example, practiced by Charles Darwin²⁰ to then support and construct larger theories of evolution. Psychiatry was plagued by the lack of objective observation and theoretically driven top-down approaches during the height of psychoanalysis. The basic building block and “bits” of theory-driven data were contemporaneous notes or reconstruction of sessions after the fact. For studies of bipolar disorder, we have mostly relied on patients’ abilities to observe themselves, store those observations, and then recall them upon questioning. Even the classic studies by Judd et al.^4,5 referenced earlier to support the distinction between bipolar I and bipolar II disorder used a highly flawed method of longitudinal follow-up—for most of the 25-year follow-up, researchers used the Longitudinal Interval Follow Up Evaluation (LIFE)²¹ instrument to reconstruct mood events for the prior year without any contemporaneous documentation. But we are learning that bipolar disorder and its associated mood states can be associated with cognitive dysfunction,²² which can interfere with each of these processes. How, then, can we have the precise data we need to develop contemporary longitudinal dimensions of the complex and multidimensional aspects of bipolar disorder? One new dimensional approach is by using ecological momentary assessments^23–25 and digital phenotypes from pervasive smartphones.²⁶ Once we gather these data, we will have a much better basis to “carve nature at its joints.” Similarly, we can analyze these data with network and systems approaches using graph theory to better understand the dynamic relationships between depressive, manic, anxious, and psychotic symptoms as well as psychiatric and medical comorbid conditions and their impact on functioning.²⁷ We can also learn more about the nature of long-term responses to treatment using real-time, real-world, medicine-based evidence.²⁸ Whether these data will lead us to understand bipolar disorders I and II as clinically relevant categories with distinct pathophysiologies or disorders along a more dimensional spectrum remains for us to determine.

Conclusion

Is bipolar II a valid category or is it a spectrum of bipolar disorder? Yes. Or maybe. We do not know. But we do know that we need better methods to track symptoms precisely (smartphones, accelerometers) and then we can use those data to determine when bipolar II acts more as a wave, a particle, or both.