By Dr. Alan R. Light, professor of anesthesiology, neurobiology & anatomy, University of Utah
Question from Claire (posted to Research1st summary of the Lights’ biomarker study published in the Journal of Internal Medicine): What are “published biomarker criteria?” Is this specific to CFS or recognized generally?
The review article cited here (as noted by Kim) is very useful, both for its explanation of technical details of how to evaluate a biomarker of any kind, but also in explaining the different uses of biomarkers and how this affects the way in which they should be evaluated. (Ray P, Le Manach Y, Riou B, Houle TT. Statistical evaluation of a biomarker. Anesthesiology. 2010 Apr;112(4):1023-40. [Free full text PDF])
The criteria for a biomarker in general actually can be quite variable. Much depends on the projected use of the biomarker. For example, a biomarker can be used for
- disease diagnosis,
- determining the severity of a disease,
- determining risk for contracting a disease,
- predicting the response to a specific treatment or
- determining if a treatment is having the desired effects.
In addition, how useful a biomarker is depends on factors not necessarily involving the science behind the test. As summarized in the article cited above, these include “…cost, invasiveness, technical difficulties, rapidity, prevalence of the disease, and consequences of the outcome of the test.” Finally, the acceptability of a test by physicians and patients can be a major factor in determining the usefulness of a biomarker.
For a CFS biomarker, one that would satisfy all or most of the criteria above would be ideal. However, a CFS biomarker that is good for diagnosis alone, while less than ideal, would be very useful for both patients and researchers.
Regardless of the criteria listed above, the usefulness of a biomarker also depends on how accurate, sensitive and specific it is. However, there are no agreed-upon set points for these measures, as much depends on how important it is to know that a patient definitely has a specific disease, for example ME/CFS, or if it is more important to know that they definitely do not have this specific disease.
One way to graph the overall accuracy of a biomarker is with a Receiver Operating Curve, or ROC. The ROC for the blood-based CFS data in our paper is graphed here. The accuracy for this curve was 0.80 and the area under the curve is 0.91, both attributes of a very good biomarker. If we choose the left-most point as the cutoff for the biomarker, then we would misidentify four of 49 normal subjects as having CFS (high specificity). However, we would misidentify 10 patients out of 28 as not having CFS, when in fact they do. If we choose the right-hand point as the cutoff, we would correctly identify all but two of the 28 CFS patients as having CFS, but would incorrectly identify 10 normal subjects as having CFS. The question is: which of these errors is more important? This may depend on whether you are a patient with CFS or a physician trying to treat CFS patients. Some point between these two extremes might turn out to be a good compromise. It is important to note that single-study ROC curves tend to overestimate the accuracy of potential tests, so the analysis I include here is only an example, not proof, that this test would be a marketable product.
At the present time, there are no published, agreed-upon criteria for a CFS biomarker. In addition to the accuracy of a biomarker, the acceptability by physicians, the cost of the test and the accuracy of the test will all play a role in dictating the final availability and utility of any biomarker for CFS.
Dr. Alan R. Light is a professor of anesthesiology, neurobiology and anatomy at the University of Utah. He is a member of the University of Utah Program in Neuroscience, the Brain Research Institute and the Pain Research Center. Dr. Light has published more than 90 peer-reviewed research articles, including seven on the mechanisms of CFS. Dr. Light is a member of the international panel that developed new consensus criteria for myalgic encephalomyelitis (ME).