For me the key findings are the reduction in Natural Killer Cell Cytotoxicity (NKCC) rather than the weaker findings about DPIV/CD26. The difference in NKCC levels is large: 28% for controls vs only 12% for CFS patients, and highly significant with a stated "p=0.000" (which I assume means p<0.0005).
However, its not, unfortunately, as simple as All CFS patients have much lower NKCC than controls. What complicates things is that theres a lot of variation in the NKCC levels of both controls and patients, with a substantial overlap between the two. (For instance, although patients on average have much lower NKCC levels, a quarter of them scored
above 21% while a quarter of controls had NKCC
below the same level)
Not really a biomarker
from the paper:
NKCC was a good predictor of CFS status...
...By ROC analysis, NKCC (and dipeptidyl peptidase/CD26) were identified as potential biomarkers for CFS through their demonstrated accuracy in discriminating CFS patients from healthy controls.
Unfortunately the data in their paper suggests NKCC would not be a clinically useful biomarker.
According to Leonard Jason a test needs an AUC threshold of between 90-100% to have diagnostic meaning, and 95% or above to be considered a good diagnostic tool.
The NK Cell Cytotoxicity test had an AUC of 0.776 (77.6%) so on this basis would not make a good diagnostic tool. This isnt surprising given the overlap in NKCC levels between patients and controls discussed above.
background: Understanding ROC analysis
The two measures of accuracy of any test or biomarker are ability to pick out true positive (CFS) cases while also avoiding false positives. This is usually evaluated with ROC (
Receiver Operator Characteristic) Curves and the key measure is AUC (
Area Under Curve) which is pleasingly simple to understand. An AUC score of 1.0 equates to a perfect test (correctly detects all true cases with no false positives) while 0.5 is no better than flipping a coin.
One slight cause for concern is that none of the assays for DPPIV/C26 (the other proposed biomarker, see next post) were significantly correlated with NKCC. If both NKCC and DPPIV/CD26 were really measuring underlying CFS-ness you would expect there to be a strong correlation between the two. This suggests that at least one of the two is not really measuring CFS-ness.
It's also worth noting that - as is the case in ANY study of biomarkers for CFS - the comparison with healthy controls isn't that helpful. Even an idiot like me could walk into a room of healthy controls and CFS patients and separate the CFS from the healthy with, I modestly suspect, pretty close to 100% accuracy using just a couple of questions. (The situation is different for something like HIV, where there may be no obvious signs of an illness.) What would be altogether more useful would be to distinguish between chronic fatigue of many different causes and ME/CFS itself and that would be an interesting next step, building on these findings.
Biological significance of reduced NKCC levels in CFS patients
Assuming the authors really have nailed a clear biological difference between CFS patients and controls, what does it mean? Is it just a helpful association or does this factor help explain the illness itself?
A
smaller study from 2006 indicated that patients with poor NKCC were more impaired, indicating the NKCC levels may be directly relevant to the illness rather than just being associated with it. This current study is continuting to collect data over 18 months which should throw more light on the situation.
Obvious limitations of this study are that each patient sample represents a single point in time. To address this, we are conducting a large longitudinal study to follow 150 subjects over 18 months. Samples are collected during times of relative symptom remission and exacerbation. Completion of the study will allow the correlation of CFS related symptoms with lymphocyte function and activation.
The authors also suggest that the reduced NKCC activity found in the current study indicates reduced innate immunity and a possilbe switch from Th1 to Th2-type, or humeral immunity.
Although the precise significance of reduced NKCC levels is not yet clear, I think it is itself enormously important that the authors have pinned down a clear biological abnormality that has been reproduced both by their own group and others.