Don’t throw the baby out with the bathwater: A case for bipolar II

We proposed that the construct of bipolar II disorder (BD-II) is understudied; Malhi et al. (2018) responded that BD-II does not warrant investigation and should be retired. While we share their scepticism around psychiatric diagnoses, we maintain that BD-II has ongoing scientific and clinical relevance as we move towards stratified approaches to psychiatry. Fortunately, recent good quality data speak to the question of whether the BD-II diagnosis warrants discontinuation. Here, we focus on longitudinal course, aetiology and treatment specificity supporting bipolar subtyping distinctions.

Longitudinal course: depressive polarity

The BD-II diagnosis does not rest solely on hypomania (vs mania) severity. Major depression is required for BD-II but not for the BD-I diagnosis, and the classic BD-II pattern is characterised by repeated depressive episodes with much less frequent hypomanic episodes. As noted in our earlier commentary (citing two large-scale prospective studies), depressive predominant polarity distinguishes BD-II from BD-I yet remains poorly understood. Data from a more recent prospective 4-year naturalistic follow-up study contrasting BD-II and BD-I further support course differences, whereby the BD-II group (1) relapsed more frequently (most commonly into depression) and (2) experienced more depressive episodes (first and index episode; Simhandl et al., 2014). The polarity concept is gaining momentum as a key predictor of recurrence to a specific affective episode: ignoring this defining BD-II characteristic risks inadequate, non-individualised treatment selection.

Aetiological heterogeneity

Building on family studies (cited in our earlier commentary) demonstrating aetiological heterogeneity between BD-II and BD-I, a large-scale study of BD familial transmission showed specificity of familial aggregation of mania but not hypomania (Merikangas et al., 2014). These data suggest that the BD sub-types do not simply represent increasingly severe manifestations along a shared continuum. Further supportive data come from the genome-wide associated study (6447 BD cases) from the International Cohort Collection for Bipolar Disorder (Charney et al., 2017). Here, differences were reported in heritability estimates of BD-II and BD-I and their genetic architecture, along with evidence of a molecular correlate supporting their division. Taken together, these data provide a compelling rationale for dissecting the broad bipolar phenotype in family studies to enable exploration of distinct, novel biologic pathways underlying BD.

Treatment specificity

The third major component of validity of a psychiatric diagnosis relates to treatment specificity. Treatment studies comparing BD-II and BD-I are needed: the knowledge base here is severely lacking. The CANMAT BD treatment guidelines note that studies to date preclude us from determining whether clinically meaningful differences exist in treatment response between BD-II and BD-I and that while response to mood stabilisers and antipsychotics may be similar, ‘there are enough exceptions to suggest this should not be taken for granted’ (Yatham et al., 2018: 128). These exceptions remain poorly understood. The guidelines further note that antidepressants may have a more favourable risk-benefit ratio in BD-II. Ultimately, more information is needed to determine which treatments are optimal. We suggest that identified differences between BD-II and BD-I with respect to illness course and aetiology provide enough of a signal to warrant further investigation into treatment specificity. Likewise, the lived experience must be respected: anecdotally, those with BD-II do not relate to the manic behaviours (and psychosis accompanying mania) typically seen in BD-I. Treatments (whether pharmacological or psychological) and outcome studies must respect these differences as part of a targeted (and face-valid) approach.

Why the hype about sub-type?

Classification systems help clinicians identify those most at risk for complications, paving the way for personalised, effective treatments. Diagnoses enable a common clinical language and provide patients with useful templates to understand their experiences. Deepening our understanding of BD heterogeneity is necessary: prematurely shifting the dividing line within the spectrum does not resolve the problem, and it just locates it in a different place. While the cleavage remains imprecise, these distinctions provide a practical way of moving forward. Questions regarding the validity/utility of the BD-II diagnosis may best be pursued in two ways: long-term, reliable illness course and aetiological differences between BD subtypes enable useful hypotheses to navigate the complex multi-dimensional space that may best describe mood pathology. Short term, we swim with the international tide in retaining the BD-II construct, if nothing else because it helps highlight the urgent need for treatment specificity research in BD-II. There are clear empirical hints that the pressing challenge in mood disorders – improving treatment outcomes – may benefit from responding to demonstrable course and aetiological differences between BD subtypes. We propose the bar should be set low for any conceptualisation that might achieve this, while holding diagnostic constructs with appropriate scepticism as Malhi et al. (2018) remind us.

Above all, and moving beyond academic debates, treatments for bipolar disorder (whether I, II or part of the BD spectrum) remain sub-optimal. Throwing the (BD-II) baby out with the bathwater does not move scientific understanding nor treatment forward.