Predominant polarity in bipolar disorders: refining or redefining diagnosis?

The major criticism that is usually posed against categorical diagnosis in psychiatry concerns their lack of theoretical basis with lack of biologic underpinning, poor specificity and alleged overinclusiveness. We do not wish to enter this controversial debate, but there is some agreement amongst clinicians and researchers that psychiatry badly needs more accurate and specific diagnostic categories. The inclusion of DSM course specifiers is thought to help refine diagnostic categories in order to make them more reliable and adapted to the almost endless clinical presentations of any psychiatric condition. Hence, a course specifier should define a clearly different subpopulation within a diagnostic boundary, provide some clinically relevant information and have therapeutic and prognostic implications. Unfortunately, many of the existing specifiers do not fulfil many of these criteria. In a review on DSM-IV modifiers performed to identify which changes should DSM-5 undertake, it became evident that most of the existing modifiers were not particularly evidence-based, did not clearly define any specific subgroup of individuals and/or would not have treatment and prognostic relevance 1. Specifically, some widely used specifiers such as postpartum episode (but not postpartum onset) failed to show therapeutic and prognostic implications 2 whilst others needed to be completely redefined as they ignored some clinical presentations. The last is the case of seasonal pattern, which until DSM-IV only took account of depressive – but not (hypo)manic – episodes, with little prognostic implications according to this extremely narrow definition 3. This was fortunately changed in DSM-5. Last, the vast majority of current specifiers are not longitudinal but actually cross-sectional, with few exceptions such as rapid cycling 4. Cross-sectional features may be sometimes helpful, but course specifiers have a plus in terms of predictive value. Indeed, there is a need to include longitudinal specifiers that inform the clinician about long-term course of illness and impact treatment choices and approach. One focus that received little attention in the development of DSM-5 is that many bipolar patients have a clear tendency towards one pole or another (mania/hypomania vs depression), a tendency that has been managed intuitively by the treating physicians for many years. However, in a similar manner that physicists aim at representing natural phenomena in mathematical formulations, physicians would ideally tend to better understand and make the best treatment choices for complex illness phenomena by operationalizing them. In fact, some attempts have been made to achieve some degree of operationalization of this tendency towards one pole or another. The first studies on this issue were performed by Jules Angst, who described clear groups of patients classified as 'preponderantly' manic (more manic and hypomanic episodes than depressive, lifetime) or depressed (more depressive episodes) or 'nuclear' (half and half) type in a long follow-up that exceeded 10 years 5. Angst works stressed the importance of repetitive binomial patterns (Md, Dm, dM, mD) rather than predominance of a polarity. Even then, his seminal studies pointed out at a close relationship between depressive preponderance and female gender, whilst males would be equally distributed amongst the three categories. More recently, two different proposals have been tested. The Barcelona proposal 6 is defined as at least two-thirds (2/3) of the total number of past episodes being from the same polarity. Hypomanic and manic episodes sum together against depressive episodes, and mixed episodes are not included in the definition. It has been tested in large samples in Spain, Italy and Brazil 6, 7. It has also been considered in some post hoc RCT studies 8 and was included in the ISBD Nomenclature Taskforce recommendations for the DSM-5 9 but unfortunately not included in the current version of DSM-5. A second, broader, proposal is the Harvard index that results from a quotient between the number of (hypo)manic episodes and the number of depressive ones. If the score is above one, the patient is considered as 'predominantly manic' and if it is below one as 'predominantly depressive'. Hence, it asks for one half plus one of the episodes to be of a certain polarity. The authors also experimented with other possible criteria ranging from 51% to 67% of episodes of a certain polarity – which would make it 'user-unfriendly'. It has been tested in large samples in the USA and beyond 10. Regardless of the definition used, the concept of predominant polarity also has a strong genetic correlation 11. As it can be easily understood, the main difference between the two definitions relies on their stability. Whilst the Harvard definition may be more dynamic, allowing one single patient to swap polarity simply by having two repeated episodes of a certain polarity, the Barcelona definition is more time-stable, accounting for some long-term therapeutic implications. As the two definitions are sensibly different, their clinical and sociodemographic distributions are also different (for a review, see 12). As a weakness, the Barcelona definition may be specific but too restrictive, as between 38 and 44% of patients remain without an assigned predominant polarity 6, 7. Neither definition considers mixed episodes, although the Harvard group experimented adding the mixed episodes to the predominantly depressed, which proved to be highly specific to predict suicidal risk. It is unlikely that a 'predominantly mixed' scenario can be identified, but perhaps the presence/absence of mixed features lifetime may have prognostic and therapeutic implications as well 13. Some proposals to include predominant polarity in the therapy decision-taking process have already been made by means of the 'polarity index' 14, a numeric expression of the efficacy profile of a given drug obtained from number needed to treat (NNT) for prevention of depressive episodes and NNT for mania prevention ratio. This metric has been used to rank operationally both pharmacological and non-pharmacological prophylactic treatments 15, and although it has received some methodological criticisms 16, it has also been tested in large naturalistic samples 17. The biology underlying a high polarity index appears to be closely related to strong D2 antagonism, in the sense that drugs that strongly block the D2 receptor tend to have higher PIs, whereas dopamine release inhibitors (such as most mood stabilizers) and weak D2 antagonists (such as quetiapine) have lower PIs 14. At the end of the day, long-term course specifiers will only be meaningful if they have therapeutic and prognostic implications; a patient known to have presented with rapid cycling will be a case for avoiding antidepressant use at all costs or for potentially using a certain type of stabilizers 18. Similarly, a patient with a clearly depressive predominant polarity would be a good candidate for lamotrigine and, probably, an unlikely candidate for predominantly mania–prophylaxis agents. It is important to realize that long-term specifiers help clinicians to better understand a particular patient and make decisions regarding therapy. The clinical usefulness of terms as 'rapid cycling', 'predominant polarity', 'lifetime history of attempted suicide' or 'early onset' remains out of question. One step forward would be to test whether these items just refine the diagnosis or really redefine it, as for some patients it could be much more relevant to account for predominant polarity or rapid cycling rather than, for instance, bipolar subtype.