One of the practical issues in doing neuroscience in humans is that you have a problem determining causation.

Say I do an imaging study with a neurological disease and find that the activity in a certain brain region is consistently lower. Do I know whether that reduced activity is causing the disease or whether it is just secondary? It is often really hard to tell in humans. If it were an animal study, we could just lesion the area in experimental animals and see if they were still capable of getting the disease. However, there are diseases where good animal models are lacking, and one of those diseases is Post-Traumatic Stress Disorder (PTSD).

Koenigs et al., publishing in Nature Neuroscience, take an interesting approach to this problem in PTSD. They looked at a large number of Vietnam War veterans, some of whom had damage to their brains in combat. Then the looked at the ratio of individuals who later got PTSD and compared that with where in their brains they were injured. They show that injuries to two areas of the brain -- the ventromedial prefrontal cortex (vmPFC) and the amygdala -- lower the risk for PTSD, strongly implicating these regions as causative for the disease.

PTSD is a psychiatric disorder characterized by the patient re-living a traumatic event. Patients avoid reminders of the event and often suffer from hyper-arousal and flashbacks. It's prevalence in war veterans can be quite high, and it's prevalence world-wide is about 15 million people so it is a major psychiatric issue.

We know from imaging studies that patients with PTSD have higher activation the amygdala and lower responses in the vmPFC. Since the amygdala is involved in learned fear responses and the vmPFC is involved in inhibiting learned fear responses, these observations caused researchers to speculate that PTSD is caused by persistent change to the system for fear learning in the brain. However, it is really hard to know that for certain, right? I mean, it is possible that the changes in activity are the consequence of reliving a painful experience rather than the cause of the disease.

To address this issue, Koenigs et al. looked at the histories and brains of 193 Vietnam veterans with lesions throughout the brain and 52 veterans with no lesions as controls. The groups were balanced as well as possible for duration of combat, whether they were drafted or not, etc. (The non-injured group, as expected, had longer duration in combat.)

Then they compared the PTSD prevalence -- whether the veterans had ever shown PTSD symptoms -- for individuals in four groups: non-injured, injuries to brain regions other than the amygdala and vmPFC, amygdala lesions, and vmPFC. The results are shown below (Table 2 from the paper):

As you can see, the amygdala and vmPFC lesioned individuals had significantly lower prevalence of PTSD than the other veterans. Importantly, the rates of PTSD in veterans that had non-amygdala and vmPFC lesions was the same as that for the non-injured veterans. This is important because it addresses the issue of total combat duration. The injured veterans had seen less total combat, and hence we would expect them to have a lower risk for PTSD. That these two numbers are equivalent -- and similar to other values for PTSD risk in Vietnam veterans -- suggests that the main effect i.e. where the lesion was is significant.

These results strongly suggest that the amygdala and vmPFC are involved in the formation of PTSD, but this is not the end of the story. For example, it is possible that brain regions very close to these regions are actually involved rather than the regions per se. This is especially important for the amygdala because another brain region called the hippocampus -- which is involved in learning and memory -- is really close to the amygdala and has also been implicated in PTSD. (I wrote an article last year about how stress precedes volume reduction to the hippocampus in PTSD.) To address this issue, the authors look at veterans who had injuries to the medial temporal lobe but not the amygdala and veterans with injuries to the hippocampus. Neither group had a significant reduction in the prevalence of PTSD, suggesting that these regions are not involved.

The results suggest that changes in the amygdala and vmPFC but not the hippocampus that cause PTSD. However, I would note that the group of veterans with hippocampal lesions only had 11 people, so it is possible that a larger data set would show a significant effect. Also, interpreting lesion studies in humans is always a bit dodgy because you don't really have control over the extent of the lesion. You try your best with what you have, and hope that larger numbers of patients balances out the variability in lesion size.

To take a step back, I chose to talk about this paper for a couple reasons.

1) It may seem gruesome, but neurology moves forward in part by understanding brain injuries. The consequences of brain injuries are devastating and terrible, but neurologists need to evaluate and deal with patients with brain injuries because they provide the only clues for what parts of the human brain actually do. In this sense, sadly enough, the history of neurology has been subsidized by war.

It is certainly my hope that someday knowledge garnered from these patients will help us learn to make them better, and I am certain that I speak for all neurologists (or neurologists-to-be) when I say that we deeply appreciate the ability to work with patient populations like this. So gruesome as it may be, studies like this are still very important.

2) This paper is very clever in the way that they employ lesions to highlight whether a region is necessary for the formation of disease. I don't think that I have ever seen a study that looked at disease prevalence in a neurologically injured group.

This is because in many cases you are dealing with a small group of patients rather than one this large. When you only have 10 patients to work with that have a particular lesion, you have to operate by observation rather than by statistics. The authors of this study get to do statistics on prevalence of PTSD partly because they have so many patients and partly because the prevalence in this population is so high. This is hardly ever the case, so I thought it was notable to televise this paper.

For example, they show this figure in the paper. On the right is a picture of sections of a normal brain. On the left is a colored histogram of the difference between the number of patients showing PTSD and the number of patients not-showing PTSD for lesions in that particular region of the brain. More orange regions indicate a higher prevalence for PTSD for that region. The red part at the front of the brain on the right is the vmPFC. (Figure 1 from the paper. Click to enlarge.)

Frankly, I have never seen a diagram like this one, and I think it is pretty cool.

The full citation is below:

Michael Koenigs, Edward D Huey, Vanessa Raymont, Bobby Cheon, Jeffrey Solomon, Eric M Wassermann & Jordan Grafman. "Focal brain damage protects against post-traumatic stress disorder in combat veterans." Nature Neuroscience Published online: 23 December 2007; | doi:10.1038/nn2032