Commentary
Summary
Bit of a 'hey ho, not much to write home about' study. The biggest weakness is the use of empiric criteria so these are probably mostly chronic fatigue cases rather than CFS. The biggest strength is that they pursued a specific (and plausible) hypothesis that the problem was in the basal ganglia, and used a gambling test that had already been shown to activate the basal ganglia - so this wasn't a fMRI fishing trip looking for random differences between patients and controls. The correlation between a clinical feature - fatigue - and brain differences during the gambling test also adds weight to the findings.
However, the differences they found were small and patchy, which, combined with the broad empiric criteria makes this study, in my view, of limited interest. Even though these findings are consistent with an inflammatory model of mecfs, which has growing evidence for it.
Background biology
The basal ganglia are a set of structures deep in the brain that have been implicated in fatigue both in neurological diseases and in response to inflammation (eg after a
dministration of the cytokine interferon-alpha as treatment of Hepatitis C patients). The basal ganglia are also a site of reward processing, and this study looked at changes in basal ganglia during a gambling test, a classic test of reward processing.
The basal ganglia has three components, as you can see in this wikimedia pic: the caudate (long, tail-like structure) the putamen and (hidden behind the putamen) the globus pallidus. As for much of the brain, this structure is symmetrical, with left and right caudate, putamen and globus pallidus. This fMRI study looked at each structure separately
Gambling test
participants had to guess which of two cards presented face-down on a screen was “red” (hearts or diamonds) by pressing one of two buttons on a MRI response box held in their right hand. Two seconds into the trial, the selected card was turned over, and, depending on its color, the participant either won (red card) or lost (black card) one dollar
Unbeknown to the subjects the test was rigged:they started with $16 and all ended with $23 at the end so they had a consistent 'gambling' experience.
Weak definition of CFS
As @
alex3619 and @
Valentijn have pointed out, this study uses the empiric criteria, which in a poplulation study found a rate of 2.5% for CFS, compared with around 0.3% for a similar earlier study using the Fukuda criteria. The particular issue with the empiric criteria is it uses very 'soft' thresholds: in this sample the average SF36 Physical Function score of 64/100, compared with around 40/100 in outpatient clinic trials. Additionally the sample only had 4.6 CDC symptoms (8.1 total symptoms) suggesting this might be a group of people with general health problems inc fatigue rather than even Fukuda CFS.
Oh, to add to the fun, 39% had done post-graduate study, showing this sample to be somewhat unrepresentative.
Patchy results and Statistical issues
I know, stats are so dull, but so important too, especially in fMRI studies that generate shed-loads of data and so are prone to finding false positive.
Looking at the whole of the basal ganglia there was a difference between patients and controls, but it wasn't big (p=0.02). Figure 2 shows data for individual areas within the basal ganglia and as you can see there is substantial overlap between patient and control scores.
As the authors point out:
Thus, not all persons in our study exhibited reduced neural activation in the basal ganglia, although all CFS subjects exhibited significant fatigue. Therefore, although alterations in basal ganglia function may characterize a subgroup of CFS patients, such alterations do not typify all CFS subjects and do not account for all instances of fatigue.
Back to the detail, and the study found:
CFS subjects exhibited significantly reduced activation for winning versus losing trials compared to controls in the right caudate (.., p = 0.014) and right globus pallidus ..., p = 0.019
Again, not hugely significant results. They also found differences for these structures using bilateral data (combined left and right): p=0.047 for caudate and p=0.036 for globus pallidus. However, note no differences in the
left caudate or globus pallidus, and no differences anywhere in the putamen. So these results are a bit patchy.
[
correction: - bilateral data
was in fact significant, but I'm trying to establish if appropriate statistical corrections for multiple comparisons were made].
Also, why was the effect seen in the right structures but not the left? The authors did not predict this, and while they speculate as to why it might be, I suspect if they'd found the reverse pattern they could have happily speculated in the opposite direction too.
Looking at correlation between fatigue and fMRI-measured response to reward - the most interesting part of the study - the results were again patchy. The authors only looked where they'd found results above and found correlations found in the right globus palllidus, but not the right caudate - and didn't mention the bilateral pallidus or caudate. However, they have said that the effects they found in the right globus pallidus were large effects.
The correlations that were found, in the right globus pallidus, were that increased fatigue was correlated with decreased response to reward (the direction that would be expected). It was strongest for mental fatigue (as opposed to general or physical), which again makes sense and the correlation of r-squared = 0.49 (p=0.001) is pretty impressive: equivalent to a correlation coefficient r of 0.7 (0-1.0 scale). This one result does look very good and suggests they might be on to something.
Fatigue vs motivation
Another slightly odd aspect of this study is that it sets out to investigate fatigue and the basal ganglia role in fatigue, but focus on motivation (measured by the gambling test), which ain't quite the same thing.
If the authors had used a better sample, these results would have been more interesting, especially as I think other studies have indicated difference between basal ganglia in CFS patients v controls. But fMRI is a field noted for its false positives so I think a lot more work is needed before too much weight can be put on the role of basal ganglia in mcfs.