ICD Future

Scrying

Recently, a depressed patient who had agreed to participate in an imaging research project asked me, quite reasonably, ‘What does my brain scan tell you about my depression?’ In addition to a structural and functional MRI scan of her brain, the study had also required genotyping and the administration of neuropsychological tests, alongside extensive assessment using self-report questionnaires. Her question reflected her curiosity and perfectly logical desire to further understand her depression. This patient had never doubted the accuracy of her diagnosis. She ‘knew’ when she had depression because her brain would ‘stop functioning’. Over the course of her life she had experienced several bouts of depression and recognised it simply as an inability to get out of bed, to go to work or to tend to her needs. In contrast, the detailed research assessment deconstructed her experience into sets of symptoms that ensured that she met the criteria for major depression. It failed to capture, however, the fact that her symptoms varied, sometimes quite substantially, both from one episode to the next and even within an episode. The symptoms that ‘defined’ her depression often varied in nature, intensity or both, and perhaps this partly explained why her diagnosis had been delayed considerably, despite seeking advice from several doctors on a number of occasions.

I would have liked to have been able to tell this patient that the MRI scan confirmed her diagnosis of depression and that key changes observed in the images indicated that she had a specific subtype. Furthermore, I would have liked to say that the neuropsychological and clinical tests revealed that she would respond to a particular type of medication in conjunction with psychological treatment, and that soon after commencing treatment she would start to feel much better. Ideally this interview would then have concluded with the reassurance that, if she responded to treatment and remained engaged in follow up, her prognosis would be good. Alas, in reality, all I could say was that the MRI scan data could only be meaningfully interpreted when combined with data acquired from groups of people, and that even then the information couldn’t be used to confirm a diagnosis or to predict her response to treatment. In essence, the information, though useful for addressing some research questions, was of no real clinical value.

This scenario highlights the enormous gap between our clinical aspirations as psychiatrists and the reality of modern day psychiatric practice.

Classify my eye!

A key part of a psychiatrist’s role is to arrive at a diagnosis and to assign it a name (Malhi, 2008). This is a crucial clinical step that provides patients with a formal term to communicate their experience to others and also to acknowledge that they are suffering from an illness. For example, in the absence of a diagnosis, our study patient was unable to relate her problems to others succinctly and instead had to describe her symptoms repeatedly. Obtaining a recognised diagnosis also validated her complaints, especially in the eyes of her friends and family, and also enabled her to participate in our research study. Indeed diagnosis, and grouping of conditions according to diagnoses, is critical for clinical research as well as for communication in clinical practice. But simply naming a problem does not connote knowledge; and yet, in psychiatry, the term ‘disorder’ has become synonymous with ‘disease’ (Malhi, 2013a). Remarkably, this has occurred in spite of the fact that psychiatrists are first and foremost physicians, and as such are fully aware of the limitations of relying solely on signs and symptoms to make medical diagnoses. Taking a history, conducting a physical examination and synthesising this information to inform decision-making, which is further enriched by the results of pertinent investigations, is a process ingrained into our way of thinking through a long and arduous education and apprenticeship. We learn that signs and symptoms in medicine are rarely pathognomic and that even common and characteristic symptoms are usually not specific to a particular disease process; for example headache, fever and nausea routinely herald the onset of a plethora of illnesses. In psychiatry, this coupling between symptoms and pathology is even more tenuous and varied, and the majority of psychiatric symptoms are best regarded as features, which can arise in a multitude of settings and which reflect a variety of underlying causes. Even individual symptoms, for example changes in motivation, enjoyment and worry, have to be gauged as to whether they are abnormal or simply at the extreme of normalcy. Consequently, in the absence of any definitive neurobiological underpinning for neuropsychiatric diseases, psychiatric classification remains dependant on eliciting signs and symptoms of mental illness. This then is the central problem from which many other difficulties in psychiatry originate.

Layers within lairs

The observation by Donkey in the movie Shrek ‘You’re so wrapped up in layers, onion boy, you’re afraid of your own feelings’ highlights beautifully a key problem that researchers face when investigating the fundamental pathophysiology of mental phenomena (Shrek, 2001). Translational research is the current zeitgeist; much has been written about the application of clinical wisdom to basic scientific endeavours and, conversely, the construction of clinical studies using bench-top knowledge. These are ambitious goals, and possibly the correct approach to champion, but in the absence of any real connection between our knowledge of neuroscience and clinical experience in psychiatry, it remains an assertion that lacks a clearly articulated argument. It is therefore imperative, perhaps now more so than ever before, that pathways that link mental phenomena to their underlying neural substrata be found.

At some time or another during a high-school science class, we have all peered through the lenses of a microscope and spent what seemed like an eternity at the time twiddling various knobs to visualise the structure of cells, such as those of an onion. Too much magnification and everything on the slide became a blur; too little and it lacked sufficient detail. Unknown to us then was the fact that this phenomenon is not unique to cell biology. In physics, for example, the model of the atom has been immensely useful for understanding the constituents of our normal physical world. But delve a little deeper, to the sub-atomic level of quarks and other miniscule particles, and reality reveals itself as quirky and unpredictable – if not almost unbelievable. Similarly, light, a familiar and ubiquitous entity that is usually thought of as a wave, actually consists of elementary particles called photons. In reality it behaves both as a wave and as a particle and this ‘wave-particle duality’ can vary according to context and perspective. Thus the world as we perceive it to be operates differently at the atomic scale, and differently again at the cosmic scale. Different ‘layers of existence’ appear to have different rules.

In attempting to map the connections between psychiatric phenomena and neural substructures, a major difficulty is that, as yet, we do not know the number of layers that exist, whether they observe similar principles and which ones are important with respect to pathology. These are things we need to know before we can even embark upon our quest of translation. Hence we remain chained to classification systems that rely on sets of questions to identify and group symptoms into syndromes, which in turn form the basis of psychiatric disorders.

Clinically we function on a large scale – a macro level that is perhaps equivalent to cosmology in the aforementioned example. It has provided adequate descriptions of psychiatric manifestations, but fails to advance cause and effect, which possibly operate at much smaller (micro) levels. However, when we limit ourselves to this clinical layer, our methods have significant limitations and thus our picture of psychiatric disorders is incomplete. For example, our inability to define structure and functioning within mental processes means that we continually struggle to delineate psychopathology from normalcy and find it difficult to distinguish disorders both in cross-section and longitudinally. This has meant that, both clinically and in research, the diagnoses that we have been trying to pin down have behaved as ‘moving targets’, literally.

A disorder defined three decades ago would probably be defined differently today. This in itself is not unique to psychiatry and, with advancing knowledge, one would hope that definitions of diagnoses are regularly updated. The problem in psychiatry is that some complaints that were regarded as normal 30 years ago may now be seen as abnormal (Maj, 2014); this is a serious concern. For example, depression has been sliced dimensionally on the basis of severity and subtyped on the basis of patterns of symptoms. It has been defined on the basis of whether the symptoms are typical or not, and when it occurs (e.g. post partum, involutional and adolescent depressions). Depression has been regarded as a standalone entity or one that is inherently enmeshed with anxiety and personality. It has been conceptualised as emerging from an anxious milieu and as a reaction to environmental factors, or to occur irrespective of circumstance – a consequence of genetics and fate. In recent years, the porous, malleable and incomplete borders of depression are gradually melding with those of bipolar disorder, such that many individuals initially described as having depression are now thought to have ‘latent bipolarity’, presenting forme fruste and emerging only when triggered to do so.

So what could our future hold?

The fact that DSM-5 has altered little and remains wedded to its pragmatic foundations (Ghaemi, 2014) means that our chances of finding meaningful connections between neurobiology and clinical disorders using DSM-5 are slim. The eleventh revision of the ICD therefore provides an opportunity to rectify this unhappy situation. Can it do so? Are our expectations realistic?

The reality is that ICD-11 is designed for coding and capturing what happens in clinical practice, even more so than DSM-5. Thus far its revision appears to be following trends in DSM-5 and hence it is likely to offer few new insights, especially given that one of its aims is to harmonise its diagnostic categories with those of DSM-5 (McGuffin and Farmer, 2014). Professor Carroll in this Journal (Carroll, 2014) has put forward a number of suggestions for improving ICD as it transitions from version 10 to 11, and the majority of these are in alignment with proposals from learned others (Goldberg, 2014; Gitlin and Miklowitz, 2014, McGuffin and Farmer, 2014; Maj, 2014; Ghaemi, 2014 and Carroll, 2014).

Amongst the various ideas put forward, one that is particularly important and relatively easy to implement is that of adopting a longitudinal perspective when considering diagnosis (Carroll, 2014; Goldberg, 2014). Diagnoses are innately fluid and should, especially at the outset, be regarded as provisional. In fact, in practice some diagnoses are already used in this manner, such as schizoaffective disorder, which can also be regarded as the diagnosis of ‘schizophrenia-on-hold’ (Malhi, 2013b). But the intent here is to allow revision of a diagnosis that is essential in light of new information, so that diagnoses are eventually anchored more securely.

To this end, the pillars upon which diagnoses could be more securely placed, originally put forward by Robins and Guze (1970), need to be resurrected. This would mean that, within the realm of neuro-scientific research, family history, along with genetic information, need to be integrated into both diagnosis and management (McGuffin and Farmer, 2014) and biomarkers from several ‘levels’ of enquiry, such a neuroimaging and cell biology, need to be better defined and linked to clinical psychopathology. Similarly, researchers residing in clinical realms could focus on examining the cause of illnesses, their trajectories and how these are modified by treatment. Response to treatment could in and of itself be used to define subgroups, which are then characterised on the basis of neurobiology.

When deliberating what needs to be done and where to begin, it is perhaps natural to feel somewhat overwhelmed and wonder whether it is even possible. There is no doubt that, with the tools at hand, linking psychiatric disorders to disease is likely to pose multiple challenges, many of which we cannot even foresee; but if we wish to advance our craft and to achieve a meaningful understanding of mental phenomena, then such steps will be necessary. Though no one (that we know of) can see into the future, through our efforts we can certainly attempt to define it. For now, perhaps we can persuade ICD-11 to consider undertaking some bold steps that will benefit all of us in this endeavour.