Sense and nonsense of biomarkers in pain medicine
Updated: Jan 14
"Biomarker for pain": a frequently used but nonsensical construct
Again and again, we read of the need of having "objective" biomarkers for pain. When considering the definition of pain by the International Association for the Study of Pain, the nonsense of this construct is obvious:
An unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage.
The IASP adds key notes to the definition of pain, among them:
Pain is always a personal experience
A person’s report of an experience as pain should be respected
Because the diagnosis of pain is purely based on the patient's report of pain (learn more here), the search for an "objective" measure or biomarker for pain remains illogical.
There is hardly a more misused term in pain than "biomarker"
Let's start with the definition of biomarker by the FDA-NIH Biomarker Working Group. It's not as complicated as it may seem.
A biomarker is a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions; biomarkers are not measures of how an individual feels, functions, or survives.
This definition tells us few things:
A biomarker must be measurable
Biomarkers are not a unique category but serve different purposes, such as identifying a disease (diagnostic biomarker), predicting an outcome (prognostic biomarker), predicting the efficacy of a treatment (predictive biomarker), and others
Measures on how an individual feels or functions are not biomarkers. Therefore, patient-reported measures of psychosocial characteristics or physical function (such as questionnaires for depression or disability) are not biomarkers
Biomarkers are not endpoints or outcomes, which are typically patient-relevant measures such as pain, physical function, survival, and adverse events.
Finding a measure that is related to pain does not make it a biomarker
For example, let's assume that we want to know whether a medication will provide pain relief or not in a specific patient, based on a blood test that measures a genetic characteristic (genotype) related to the action of the medication. In this case, pain relief with the medication will be the endpoint, and the genotype would be a biomarker that hopefully predicts whether the medication will produce pain relief or not in individual patients.
Let's assume that we find a statistically significant association between genotype (potential biomarker) and the effect of the medication on pain. This finding alone would not qualify the genotype as biomarker. A statistically significant associations does not answer questions that are crucial for biomarker definition, and in fact the NIH-FDA has determined that several steps are necessary for a measure to be validated as biomarker. While a thorough discussion of biomarker validation is outside the scope of this blog, I will provide some points for reflection.
Is the association strong enough to make the measure useful? Specifically:
In the example above, what percentage of patients who have pain relief will have the genotype? (Sensitivity). A sensitivity less than 100% implies that some patients who would have pain relief will not receive the medication because the genotype test failed to identify them as responders to treatment.
What percentage of patients who do not have pain relief will have a "negative" genotype test? (Specificity). A specificity less than 100% implies that some patients who do not have pain relief will receive the medication because of a positive genotype test that would wrongly identified them as responders to treatment.
Because a 100% sensitivity and specificity is largely unrealistic for any biomarker, the next essential questions are:
What is the risk or disadvantage of not offering the medication to patients who have a negative test, but would actually have pain relief?
What is the risk or disadvantage of offering the medication to patients who have a positive test, but will not have pain relief? Incidence and severity of side effects of the medication are important considerations in this regard.
What is the net benefit of drawing the blood to measure the genotype, vs. just trying the medication and see if the patient benefits or not?
What about cost / benefit and applicability to clinical practice?
This is not a trivial detail that is however frequently neglected. Many of the potential biomarkers being investigated would involve additional costs, work, and inconvenience for patients. A blood draw or urine sample are mostly feasible, if the benefits outweigh the costs, but cost/benefit analyses are uncommonly done in research. Several other potential biomarkers are more demanding and costly. Examples are quantitative sensory testing and brain imaging. Such potential biomarkers will have to demonstrate clear clinical benefits to justify time and costs in clinical environments, where every minute counts and financial pressure is a major determinant of care. For instance, research on the ability of brain imaging to predict chronic pain is of great value in terms of gaining knowledge of pathophysiology, but the potential clinical implementation of brain imaging as prognostic biomarker remains highly elusive.
We do need biomarkers in pain medicine
This is not in contradiction with the introductory section of this blog. Several areas of pain medicine would enormously benefit from the availability of biomarkers. For example, we do need biomarkers as indicators of pain-related mechanisms / nociceptive processes. If a biomarker told us which molecular or functional mechanisms is involved in a patient's pain, treatments that target that specific mechanism could be developed and applied to those patients who test positive to the biomarker.
The FDA/NIH working group has defined categories, based on the context of use of biomarkers:
Pharmacodynamic / response biomarker
Susceptibility / risk biomarker
Research for biomarkers is new in pain medicine, but is ongoing also thanks substantial NIH funding. Hopefully we will see the fruits in the next few years.