Large banner
ECNP in depth: John Geddes


Applying the without to within in bipolar disorder

The next article in our 'in depth' series features John Geddes, 2016 ECNP Neuropsychopharmacology Award winner. Geddes received the award for his exceptional research contributions to bipolar disorder and mood instability.

For over 15 years, John Geddes, professor of epidemiological psychiatry at the University of Oxford, has been getting to grips with the state of bipolar disorder treatment. Today, his research studies existing treatments that could give novel insights into disease mechanisms, generating paths to the discovery of new treatments.

He is doing this by bringing the very best cutting-edge science from across medicine – and beyond – to bear on the problem. “We have a tough problem,” he summarised in an interview with ECNP. “People have tended to think that mental disorders will be very complicated to crack. What that has meant is that the level of resource applied to them has lagged way behind other areas of medicine.

“Not surprisingly, then, progress has not been as impressive as in some other areas. But our patients’ needs are the same or more, and they deserve the same level of investment applied to treatment discovery.”

Failure must be expected in scientific investigation, he noted, and difficult problems demand greater firepower, not less. “In science, quite often you will find something you are not looking for through serendipity – and that has certainly been the history of drug discovery in our area. Just about every drug in psychiatry has been discovered by serendipity. We need to guide serendipity by identifying some starting points, providing the right level of investment, the right infrastructure and the right scientific leadership. Even if you are trying to find a needle in a haystack, you need an effective and properly resourced strategy for needle detection in a haystack, not a random search around a car park!”

And starving an area of resources is fatal, he continued. “When you have resources that are very limited, the competition for those resources becomes intense, and the a priori case for committing the resources has to be perceived as very low risk – and decision-making becomes ultraconservative. The fear of failing becomes a real impediment to investing in high risk, high pay-off areas. Greater resources will increase innovation and progress.”

A world of greater research resources is certainly appealing. But, argued Professor Geddes, this needs to be combined with increased efficiency. “We need to identify when a track of research isn’t really leading anywhere. We are not very good at doing that. We need to learn to recognise failure quickly – what’s the killer experiment that answers go or no-go? And then divert resources to new areas.”

"We find ourselves in an epoch characterised by an ever-increasing volume of information, which holds enormous potential to the research community but also creates the challenge of separating signal from noise."

Indeed, Professor Geddes, Scott Monteith, and others recently discussed the outstanding analytical challenges that this presents, looking forward to the advent of personal biological monitoring integrated with other more familiar data sources such as electronic medical records and smartphone activity[1].

Our ability to capture data on behaviour and biological processes from patients has been completely transformed by remote monitoring and wearable devices. As mental disorders are characterised as problems of thinking, feeling and behaviour, for the first time this provides an opportunity to measure the manifestations of illness accurately and much more reliably.

“If you consider our conventional ways of assessing and measuring mental disorder, the principal method was to ask the patient to recall their feelings or thoughts over a period of time – which could be months or years. That is going to be quite unreliable. For example, in bipolar disorder, mood is variable and many people can’t remember their feelings of a day or two ago. So you get a very, very imprecise, noisy measure of what the course of the disorder is.”

Connected devices such as wearables and smartphones, though, have the potential to capture a person’s feelings and behaviour in a continuous stream, allowing for a more complete understanding of how these might stray from healthy behaviours. But analysis of these data presents numerous challenges, noted Professor Geddes: “First of all, how do you capture those data reliably? Secondly, how do you start analysing them in a way that is actually going to produce useable and robust ’signatures’?

“Our standard approaches struggle when we want to use multiple streams of complex, time stamped data that reflect behaviours, thoughts, feelings, physiology including neural activity. We need innovative mathematical approaches to analyse these data.”

Looking outside of psychiatry – in the financial world, for example, where a lot of money is spent on predictive research – has provided some much-needed input to the problem: “We are working closely with some of the people who have previously worked in the financial industry, looking at market fluctuations and patterns. They shared some of the same goals – for example, classification and prediction of future outcomes. If you predict the financial markets, you get very rich; if we predict, we can transform our ability to help patients.”

While we aren’t there yet, the derivation of signatures from the data may provide a path to discovering new clinical approaches or drug treatments. Several groups are now pursuing this approach; Professor Geddes and his colleagues have developed the True Colours system, initially for bipolar disorder, over ten years ago. True Colours captures self-reported mood daily or weekly, and some patients have been using it since its inception.

“True Colours provides a way of visualising mood data over time,” explained Professor Geddes. “And because the data are then made available to the patient, it gives them a dashboard of how they are going on over time, which they can provide to the physician. The patient becomes much more aware of their mood variability, and then how things such as lifestyle, activity and sleeping may have an impact. They can monitor how much they are helped by a new medicine or a particular psychological treatment. They can often see very clearly whether it is having any impact.

“This feedback allows the patient to get control of the disorder – and they can then self-monitor and self-manage to a varying, but sometimes quite remarkable degree. Some people who start the system become – for the first time – empowered to look after themselves to a large extent. They can begin to identify patterns of when their mood might be deteriorating, and they can access help earlier.”

But, continued Professor Geddes, creating an automatic system that picks up early signs of deterioration from self-report early from patients is quite a tough problem. “This is where Big Data analytics will play a role. In the meantime, put simply, the very act of giving someone a useable system (and one that is usable across multiple platforms including smartphone apps, text or web/email) seems to be very valuable.

“It already seems clear that some people can use the system and find it helpful immediately, and others can’t. We think that some people need help to use the system effectively and we have tested some educational courses that the patient can use in combination with that approach to see if this helps.

“How to integrate mood monitoring into overall care remains an important question. We have a system that really works very well for some patients, but needs development to allow us to precisely map it to the needs of individual patients.”

This approach is not just of promise for people with chronic unstable mood, because the difficulty for patients and clinicians assessing the course of a disorder extends far beyond mental disorders. It applies to any chronic disorder, whether it is inflammatory bowel syndrome, musculoskeletal disorders, migraine, epilepsy – anything where the longitudinal course is quite complicated with temporal variability and deteriorations and improvements, and hence difficult to remember retrospectively. “If you can put a system in like this to monitor it, it can actually transform the way we practice medicine,” said Professor Geddes. “Within Oxford, we are increasingly using the system with our colleagues in other areas of medicine.”

In bipolar disorder, with Wellcome Trust funding via the CONBRIO programme, Professor Geddes and colleagues are adding further data streams to the True Colours platform – including measurements of activity, sleep, heart rate, etc. They are also using the approach to characterise the effects of existing treatments more precisely. For example, the OxLith trial is a high intensity randomised clinical trial which seeks to investigate lithium[2]: “By capturing multiple data streams from patients treated with lithium, the specific effects of the drug can be identified, providing insights into mechanisms of the disorder, the mechanisms of drug action and, with luck, generating some new leads for drug discovery.”

John Geddes gave a Plenary Award Lecture at the 29th ECNP Congress in Vienna on the topic of 'Rediscovering drug discovery in Bipolar disorder'. You can watch the lecture here.

1. Monteith S et al. Big data for bipolar disorder. Int J Bipolar Disord. 2016; 4: 10.
2. Saunders KE, Cipriani A, Rendell J, Attenburrow MJ, Nelissen N, Bilderbeck AC, Vasudevan SR, Churchill G, Goodwin GM, Nobre AC, Harmer CJ, Harrison PJ, Geddes JR Oxford Lithium Trial (OxLith) of the early affective, cognitive, neural and biochemical effects of lithium carbonate in bipolar disorder: study protocol for a randomised controlled trial. Trials. 2016 Mar 2;17(1):116.


Contact ECNP
To the website
Share this on:
Follow us:
Facebook Twitter mail to a friend Facebook Twitter