Compared with the other branches of medicine, psychiatry has long been the poor relation. Whereas purely ‘physical’ conditions have tangible causes – a fractured femur, a kidney problem – mental problems can be harder to pin down. Rather than referring to something visible, diagnoses tend to look at patterns of behaviours and leave the physical etiology untouched.
Take a patient with a heart complaint. This can be diagnosed definitively, with the chest pain easily traceable to its source. The doctors here would not merely ask questions, but follow up the symptom check with the relevant cardiological tests. Now take a patient presenting with depression. On visiting his GP, he is not apt to be sent out for a brain scan. Rather, his assessment and ensuing treatment hinges on self-reported symptoms, a subjective – and at times unreliable – list.
The common problem of misdiagnosis
For the purposes of argument, let’s say this patient is diagnosed with major depressive disorder and treated with a course of SSRIs. Initially, he seems to get better; his low mood replaced with near-euphoria and his energy levels soaring. But in an all-too common tale, the depression soon returns. He switches medication, the cycle continues, and the next few years are ridged with peaks and troughs.
As becomes clear over this longer period, his symptoms are consistent with a different condition altogether, bipolar disorder II. Characterised by periods of mania or hypomania as well as depression, bipolar disorder is the sixth most common non-communicable condition in medicine. While it can be extremely severe, patients tend only to seek help during their lower ebbs. This can exacerbate the problem – when the patient takes antidepressants, the oscillations may grow steeper still.
"Unless the healthcare professional has the time or the ability to delve deep, it’s difficult to pick up on these mania-like tendencies," says psychiatrist Professor Mary Phillips. "Up to a third of people with bipolar disorder have to wait eight to ten years before getting the correct diagnosis. So the challenge is to do a much better job of detecting the signs of the illness, to prevent people suffering without proper treatment."
Until recently, there was little that could be done, aside from asking more incisive questions. But over the last 15 years, scientists have begun to take strides in their knowledge of neural circuitry and the patterns involved in emotional regulation. Ultimately, this will pave the way for biologically based diagnostic tools.
The physiology of mood disorders
Phillips has devoted her career to this cause. Professor of psychiatry at the University of Pittsburgh, as well as director of the Mood and Brain Lab, and the Clinical and Translational Affective Neuroscience Programme, her major research interest is the physiology of mood disorders. Together with her team at Pittsburgh, she uses multimodal imaging techniques to discover the associated abnormalities.
Bipolar disorder has long been a top priority. "We’re hoping new imaging and medical techniques will allow us to pick up bipolar disorder long before the disease takes hold," she says. "And while the scan is still relatively expensive, think how expensive it can be to misdiagnose someone and treat them for months or years. It won’t be too long before we can take the science into clinical practice and have it as part of a routine investigation."
In September 2008, a study by Phillips’ team shed new light on the condition. Using diffusion tensor imaging (DTI), they demonstrated that each half of the brain may be associated with a different ‘pole’ of the illness, the symptoms attributable to a form of faulty wiring.
A brief overview: both sides contain an almond-shaped cluster called the amygdala which has a seminal role to play in emotion processing. This links to the orbitofrontal cortex, a kind of regulatory centre that helps ‘slam the brakes’ on our behaviour, make decisions and mete out rewards (we might characterise the two, loosely, in Freudian terms, as the difference between the id and superego). One key connecting tract is the uncinate fasisculus, a bundle of fibres implicated in mood disorders.
According to Phillips’ study, bipolar sufferers may have a thicker-than-average tract in the right side of their brain, the part associated with negative emotions. It is easy to see how extra cross-connections feed into excessive rumination; the dark, maladaptive thought vortices typical of depression. But in the left side of the brain the tract is thinner, suggesting that highly positive feelings may come with less of a filter.
Polar pattern plotting
Several years later, another study pitted bipolar disorder against unipolar depression and found some striking neurological contrasts. "These DTI findings are probably the most compelling so far," affirms Phillips. "We’re finding a pattern of relatively widespread structural abnormalities, which are much more apparent in people with bipolar disorder. With unipolar depression we don’t see anywhere near the amount of disruption in the white matter tracts involved in rewards."
The potential here is obvious. Through using neural biomarkers, doctors could eradicate ambiguity, determining categorically whether a patient has major depression or bipolar. "Our goal would be to have a brain scan as a second stage exam like a chest X-ray," continues Phillips. "That perhaps sounds far-fetched, but I really think that this could be in place in the not-too-distant future – we certainly have enough clue about imaging for us to be able to put this into clinical practice."
And it would appear there are even brighter possibilities ahead. Neural imaging techniques may soon be used not just for diagnosis, but also to identify sufferers even before the condition takes hold. Through early intervention, they can receive the help they need, delaying or preventing onset and giving them the tools to cope.
February 2012 saw the release of a paper that suggested precisely this. Using a combination of functional magnetic resonance imaging (fMRI) and pattern-recognition software, researchers showed that computers could be trained to predict a person’s risk. The software could say with some accuracy whether that person would be neurologically susceptible in future.
Phillips’ team, led by Dr Janaina Mourao-Miranda of University College London, took 16 children and adolescents belonging to a genetically high-risk group, and matched them up against a control group with no family history of the disorder. All 32 participants were healthy, displaying no signs of any psychiatric condition and scoring similarly in terms of emotional lability. But because bipolar disorder tends not to appear until early adulthood, it was possible that their profile would differ a few years down the line.
Although neuroimaging had previously been used to differentiate between low and high risk groups, it hadn’t helped on an individual level. "Bipolar is highly heritable," points out Phillips, "but just because you have a parent with the disorder doesn’t mean your chances of getting it are 100%. You would still want to know ‘is it going to be me, or my brother or sister?’."
For those with afflicted relatives, the actual risk is around 10%; a further 10-25% will go on to develop another major affective disorder such as anxiety or depression. Compare this with the average population prevalence (2% for bipolar) and two things become clear. First, the figure is epidemiologically important; second, it’s predictively irrelevant.
Facial emotion processing
In this study, subjects underwent an fMRI scan, where they were presented with a series of faces whose expressions ranged from happiness to fear. This was already a well-validated paradigm. Because depression and bipolar are known to affect facial emotion processing, the technique can reliably be used to examine mood dysfunction, with depressed people more likely to misconstrue neutrality as a threat.
The resulting brain scans were fed into a computer, which went on to assign the subjects to the low or the high-risk group. "The algorithm works out a pattern from the whole brain, a kind of curvilinear hyperplane whereby it forms a template," explains Phillips. "So in future if an unknown child comes through, it can say, ‘if the pattern looks like this, the child is likely to have a parent with bipolar disorder, and if it doesn’t, they’re a healthy control’. Essentially the algorithm was pretty accurate around 75% of the time without knowing beforehand which child was likely to be at risk."
While this was just a proof-of-concept study, a further exploratory test provided some promising results. "We wanted to know whether that pattern could tell us something about the future," says Phillips. "So we followed these children up to about four years later, by which time some high-risk children had gone on to develop an illness. Remarkably, the more confident the computer algorithm was in placing the child in a high-risk group, the more likely that child was to become ill."
This second part of the study is where the real grounds for excitement lie, suggesting that imaging may take us beyond the here and now and inform us about the years ahead. The team is now looking in detail at larger subject groups, trying to replicate some of their findings on a larger scale.
Towards full clinical applicability
Within the next decade, many of these developments look set to attain full clinical applicability. Our hypothetical depressive, visiting his GP, is less likely to be prescribed a ceaseless parade of inappropriate treatments. Rather, he will be referred for a brain scan, or even a blood test, with facilities growing more widespread as their use becomes more routine. Better still, he can be tested early in life to nip any signs in the bud.
"The precise timeframe depends on the country, but there’s a lot of positive feedback about the potential to implement these changes," says Phillips. "We just have to persuade clinicians that they are useful."
As she takes pains to point out, these measures are no substitute for a thorough psychiatric assessment. That should be there in parallel – nobody is advocating a one-sided reductionist approach. All the same, the value of neural biomarkers ought not to be understated. Our present approach may one day seem as primitive as using soreness to diagnose a broken bone.