Wednesday, November 02, 2005

Psychiatry and Neuroimaging: great expectations but little patience.

While proponents of Neuromarketing are planning to uncover the mysteries of the consumer's mind with Neuroimaging techniques (Time magazine Oct. 24, 2005, Getting inside your head), Psychiatrists (New York Times, Oct. 18, 2005, Can brain scans see depression ?) are complaining that the techniques, as of the year 2005, did not live up to their promises or, more exactly, to their expectations.

Why these two fallouts are contradictory in terms of the application of Neuroimaging ? Since Neuromarketing is not under the constraints of the classical explanation - prediction - modelisation and, in the end, the curative therapy application, it does not have to bridge the gap between reality and expectations. It is the same with the science of Economics, it can make predictions and design interventions based on expectations but it does not have to meet reality in its implications.

Lets then examine the gap between reality and expectations in Psychiatry and Neuroimaging.
A New York Times article (Oct. 18, 2005, Can brain scans see depression ?) voices the frustration of some psychiatrists:
After almost 30 years, researchers have not developed any standardized tool for diagnosing or treating psychiatric disorders based on imaging studies.

That's a hasty judgment pulling toghether more than three different approaches to brain imaging; ElectroEncephaloGraphy and Event Related Potentials - EEG and ERP, Positron Emission Tomography- PET (which the article mistakingly calls Positron Emission Topography) and Magnetic Resonance Imaging - MRI, both anatomical and functional. These approaches were historically introduced in brain research and clinical research in the order shown above with the last being the more recent and only some 10 years in the work. Note that while the first approach is completely non invasive, the second requires the injection of radioactive material in order to visualise brain activity and the last, while non invasive, is applied with some restrictions. Moreover, there are restrictions - common to all three approaches - inherent to research on humans. They are related to the fact that researchers must make sure that the criteria for their studies are normative before applying them in clinical settings and to pathological conditions.

Setting the norms

The number factor: Research methodology requires large populations with matching and homogenous criteria, so research on humans is very limited because of the difficulty in recruitment.

The time factor: Time is another limiting factor in research on humans specially when it comes to the Mind sciences. Most mental illnesses developp over a long period of time. Some of them start in childhood and result in overt symptoms only during adolescence or adulthood. To study the final phenotype of an illness is to study the complex results of many subphenotypes changing over time under the influence of the environment. In some cases like ADHD, there is an effort from granting agencies to finance longitudunal studies but most of the time, it is not the case in research where projects are financed for 3 to 5 years at best.

The method factor: The oldest in Neuroimaging techniques performed in specific behavioral contexts or cognitive events is The Event Related potentials approach. It is more than 30 years old. But research in this area stayed focused for a long time on normal events in behavior and cognition. The PET technique stood alone in its applications because of its mild invasiveness; no pressure and no unrealistic expectations accompanied the emergence of the technique in clinical settings. Nothing of the sort was allowed for MRI techniques. Because of the non-invasiveness of these techniques and the good and precise observation they offer of the anantomy of the brain, they created instantly great expectations about their potential fallout and they were immediately used in clinical research on mental illnesses. However widely these techniques are used now, they still lack normative studies - studies on normal populations and their common features. The normative studies are made complex by interindividual differences in brain anatomy and functioning which can be evaluated sometimes at more than 10% for features like hippocampal volume. Despite all this, I am confident that, with more time, Neuroimaging techniques will be able to achieve normative studies that can be applied to diagnostic and treatment of mental illnesses . EEG and ERP techniques were able to achieve this goal and are increasingly used now to study, predict or diagnose mental illnesses including those who are difficult to predict like Alzheimer (New York Times, Oct. 25, 2005, Predicting Alzheimer's is more wish than reality).

However, when the methodological limitations of Neuroimaging techniques will be overcome there are still some important things to fix in order to fill the gap between Neuroimaging applications in Psychiatry and their related expectations among practitioners. The most important of them is what I may call the explanatory and interpretation gap.

Filling the explanatory gap

What to make of brain scans, images of brain slices and images of brain activity ? An important thing to understand is that these images don't stand alone in the explanation, they need and interpretational framework. The framework is that of the brain with its many levels of organisation and of the biological system in which the brain is embedded. This system is made of the body, its environment and the ever changing interactions which continually affect the system. The whole thing contribute to an external and observable phenotype that is related to many subphenotypes at all the organisational levels of the brain and the organims. It is true that it is difficult to study complexity and change, two features of livings which that are closely related. However, in the life of an organism, there are biologically defining moments and biologically defining interactions between the organism and its environment that contribute to the final complexity of the phenotype. Longitudinal studies are made to track these defining moments and to systematise them and translate them into norms, despite interindividual differences. The biologically defining moments can be located at different levels of the phenotype. Neuroimaging shows only some levels of the phenotype but not all of them, namely, the level of the structure and of the network. No wonder Dr. Mayberg's studies of depression with PET and their fallouts were so successful (New York Times, Oct. 18, 2005, Can brain scans see depression ?), it is because they referred to an interpretational framework that goes beyond the single structure to reach the network. Although Dr. Mayberg's studies still lack placebo groups, there is a strong indication that the interpretation and its subsequent application is powerful since the patients report immediate change in their mood as soon as the electrical device slowing Brodmann area 25 - whose overactivation is at the center of a network involved in depression - is implanted and working in the brain.

In order to extract a reliable, normative, predictive and applicable interpretation of Neuroimaging techniques, researchers still have to connect the network to the behaviour of its cellular, chemical and molecular components and, beyond, to the physiology of the whole organism in its interactions with its environment.

Non invasive Neuroimaging techniques still have a long way to go in their own development and, at the conceptual level, we still have a long way to go in the Mind Sciences in order to connect the many dots and the many levels of organisation sustaining a behaviour or a specific brain structure and activity.

My advice to psychiatrists: please be understanding and, more importantly, be patient !

No comments: