Physicians have at their disposal a battery of tests to measure heartrate, blood pressure, and temperature—objective, quantifiable indicators of disease. But no such biomarkers exist yet for chronic pain. Without them, assessing pain and the response to treatment must rely on patient self-report, which is often distilled to choosing a number on a pain scale of one to 10 or an emoticon ranging from smiling to frowning to crying.
The lack of pain biomarkers not only impedes treatment for the estimated 25 million Americans who live with daily chronic pain but also hinders the quest for new pain therapies. According to the US National Institutes of Health (NIH) data, only 2 percent of candidate therapies for pain in Phase 1 clinical trials ultimately get approved by the US Food and Drug Administration (FDA), whereas that number is about 10 percent across all other diseases.
In November 2018, the NIH held a two-day workshop in Washington, DC, to discuss the current state and future outlook for the development of biomarkers for pain and pain treatment response. The meeting was co-chaired by Mary Ann Pelleymounter, a program director at the National Institute of Neurological Disorders and Stroke (NINDS), and Simon Tate, founding partner of Bridge Valley Ventures, Cambridge, UK, and formerly of Biogen and Convergence Pharmaceuticals.
Researchers at the workshop presented data and discussed a range of different types of biomarkers, from electrophysiological readouts to machine-learning analysis of population data. The attendees agreed that developing validated biomarkers of pain and treatment responses will be key to developing new therapies and generally improving the outlook for patients with chronic pain.
Moving beyond opioids
NINDS Director Walter Koroshetz described the purpose of the meeting and of pain research in general as “the scientific quest to free humanity from our dependence on the poppy plant. Can we displace the poppy from the [pain treatment] armamentarium?” Perhaps, Koroshetz said, “but without a biomarker to see what’s working, we will be in trouble.”
“Good science depends on good metrics,” said Dave Thomas, a program officer at the National Institute on Drug Abuse (NIDA) and a member of the NIH Pain Consortium. Of current metrics for pain, he said, “smiley faces, self-reporting—we can do better.”
That’s not to say that patient self-report is not valid and even critical to understanding chronic pain conditions. “People are concerned that, will we ignore patients, or [wonder], will [biomarkers] be used in court? That’s not what this is about,” Thomas said. “These [biomarkers] are [there] not to measure the totality of pain, but to advance better treatments, enhance translation from animals to humans, and get better decision-making support for early treatments. Pain is real, and by getting better metrics, we hope to advance the field.”
The effort to develop new biomarkers of pain is a major aim of the Helping to End Addiction Long-term (HEAL) Initiative announced by the NIH last year (see PRF related news). HEAL will provide more than a billion dollars over two years to fund research on pain and addiction, mainly through grants awarded by NINDS, NIDA, and the National Center for Complementary and Integrative Health (NCCIH).
What is a biomarker?
A biomarker, a mash-up of “biological marker,” was defined in a document from the World Health Organization (WHO) in 2001 as “any substance, structure, or process that can be measured in the body or its products, and influence or predict the incidence or outcome of disease.” That definition has expanded to include markers of the response to treatments and interventions (Strimbu and Tavel, 2010).
More recently, an online resource co-created by the FDA and the NIH called BEST (Biomarkers, EndpointS and other Tools) defined a biomarker as “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.” BEST also categorizes biomarkers as susceptibility/risk, diagnostic, monitoring, prognostic, predictive, pharmacodynamic/response, and safety biomarkers. Types of biomarkers include molecular, histological, radiographic, or physiologic characteristics.
So what biomarker could best address the needs in chronic pain treatment and research? That depends on the context—whether the aim is to diagnose a pain condition, predict a patient’s likelihood to develop that condition, or predict how someone will respond to a particular drug.
Christopher Leptak of the FDA differentiated between biomarkers—objective medical signs—and symptoms, which rely on the patient’s perception. Clinical trials would benefit greatly from pain biomarkers. For example, these biomarkers could be used to help stratify patients into subgroups of those most likely to respond to a treatment, or by specific characteristics of their pain condition. Ideally, validated biomarkers would indicate whether a potential therapy is effective in altering the course of chronic pain.
But biomarkers are not themselves clinical endpoints, Leptak said. “There can be all kinds [of biomarkers], using histology, algorithms—but [they are] not how a patient feels, functions, and survives. They can predict that, though.”
Rather, validated biomarkers can be used as surrogate endpoints in clinical trials. Leptak suggested that researchers refer to the BEST resource to increase the cohesiveness of biomarker development.
New biomarkers could speed the translation of preclinical studies to the clinic as well. In animal studies, researchers typically use withdrawal thresholds as an indication of nociception, or slightly more sophisticated tests that can capture pain-like behaviors or memories. But those measures are still very simplistic, and probably don’t do a very good job of modeling the complex disease of chronic pain in humans. But objective biomarkers could provide a reliable measure of pain, and biomarkers that are consistent in multiple species could go a long way toward finding mechanisms underlying chronic pain.
“It’s important that we discover but also validate new biomarkers,” Koroshetz said.
How to get there from here
When it comes to developing and implementing new biomarkers, Tate said, “Oncology is really leading the way; we could learn a lot from that field,” which has used genetic, immune, and other biomarkers to create more personalized cancer treatments.
One type of biomarker could be used to improve pain treatment immediately: pharmacokinetic/pharmacodynamic biomarkers, which indicate how patients metabolize or otherwise react to certain medications. Some genetic markers can also predict how a patient will respond to certain drugs. Such markers are currently being used to improve treatment for people with depression, and could help patients optimize their regimen of medications more quickly, which can often take months or years. They might also be used to predict who might become addicted to opioids and who could safely take long-term opioids.
Some data sets are so big that they defy analysis by traditional processing software, and new computer algorithms must be employed to make sense of the information. So-called “big data,” describing patient phenotypes, for example, can offer insights about the course of chronic pain conditions and how to intervene. In partnership with the NIH, Sean Mackey and colleagues at Stanford University, Palo Alto, US, have developed an open-source platform called the Collaborative Health Outcomes Information Registry (CHOIR) to collect detailed information beyond an electronic medical record. For example, questions meant to assess the extent to which patients catastrophize over their pain could provide a predictive marker of those most likely to suffer from chronic pain. Though not strictly a “biomarker,” objective data collected from patients could be used to chart the best course of treatment for an individual, or identify who is most vulnerable to worsening chronic pain.
“From [CHOIR], we get a deep, comprehensive phenotype of each patient,” Mackey said. “It has changed the notion of how we care for patients and conduct research; we need to directly integrate this to patient care.” Mackey and colleagues are exploring how such data could help make predictions and improve treatment.
A brain imaging-based biomarker
Brain imaging holds potential as a biomarker of pain. But Karen Davis, University of Toronto, Canada, says it’s important to consider the individual patient and not to overinterpret data collected from large groups of people.
People with chronic pain seem to differ in their brain connectivity, Davis said. “We look at the balance between the health and strength of the [pain-]sensing versus the [descending pain-] modulation pathways. If the sensing [circuitry] is dominant, we refer to that as pro-nociceptive, and if modulation is dominant, that’s an anti-nociceptive brain.” (See Cheng et al., 2015.)
Indeed, researchers have already discovered a number of different prognostic indicators using brain imaging that might help to predict who will develop chronic pain. A study from A. Vania Apkarian, Northwestern University, Chicago, US, found that patients who would go on to develop chronic pain following an acute incident of low-back pain could be identified in the subacute phase of pain development by a pattern of magnetic resonance imaging (MRI) activity in the brain’s limbic system (Mansour et al., 2013). When it comes to the transition from subacute to chronic pain, Apkarian said, “all the [contributing] factors have traditionally been explained by nociceptive input and spinal cord processing. But limbic brain circuitry can predict risk.” (See PRF related news stories here and here.)
Similarly, Tor Wager, University of Colorado Boulder, US, has found evidence for a “neurologic pain signature,” a complex pattern of brain activation detected by a computer algorithm (Wager et al., 2013PRF related news). While the signature is not affected by some pain treatments, it does shift following cognitive-behavioral therapy (CBT) and some other interventions, Wager said. “This could be an interesting treatment target for [drug] discovery.”
Practically speaking, MRI will likely remain too expensive to be used as a pain biomarker for patients, but cheaper modalities such as electroencephalography (EEG) and magnetoencephalography (MEG) could be put to use, Davis said. Ultimately, brain imaging biomarkers might be best used to discover brain processes underlying chronic pain development.
In any case, Davis said, imaging “should be used as a predictive tool, rather than as a detector.” Use of brain imaging or other biomarkers to “prove” the existence of pain in a patient would be ethically questionable and could leave people with pain even more vulnerable to denial of care or financial hardship.
The search for pain biomarkers based on brain imaging sparked a bioethical controversy in 2015 when functional MRI (fMRI) was used in a court of law as evidence of a pain condition. That raised concerns among many pain researchers, including Davis and Mackey.
Mackey testified in the court case against the use of fMRI as evidence of pain and, together with Davis and other researchers, subsequently published a consensus statement of recommendations in Nature Reviews Neurology regarding the use of brain imaging as a pain biomarker (Davis et al., 2017PRF related news).
Other potential biomarkers
Sleep patterns also have potential as a pain biomarker, according to Clifford Woolf, Boston Children’s Hospital and Harvard Medical School, Boston, US. Sleep patterns can reflect spontaneous pain, which is thought to arise from aberrant firing of nociceptive neurons and is a key feature of neuropathic pain.
Considering the relationship between sleep and pain, Woolf and colleagues wondered whether the aberrant nociceptive firing, which occurs during sleep as well as wakefulness in the setting of spontaneous pain, might cause a detectable disruption in sleep patterns, as measured by EEG. In previously published work (Alexandre et al., 2017PRF related news), they detected very brief awakenings from non-REM sleep states in mice with spared nerve injury (SNI), a model of neuropathic pain. These awakenings were absent in control animals, and correlated temporally with the onset of pain hypersensitivity in the injured mice. Further, analgesic drugs reduced the sleep disruptions. “We believe these brief awakenings may be a potential biomarker of spontaneous pain,” Woolf said.
Researchers at the meeting also discussed potential biomarkers including quantitative sensory testing (QST), genetic variations, RNA sequencing, machine learning to decode facial expressions, and even proton magnetic resonance spectroscopy (MRS), a noninvasive, quantitative technique to assess regional brain biochemistry. But, the attendees agreed, these various biomarkers of pain should not be in competition with one another. Instead, a suite of biomarkers will be needed to understand the mechanisms driving chronic pain, better predict who will develop pain or respond to certain drugs, and track the effects of pain treatments. Similarly, new biomarkers will need to be cross-validated by groups not involved in their development to establish them as standard measurements.
A subgroup of workshop attendees plans to publish a white paper detailing the findings of the meeting later this year and meet regularly to follow up on progress in pain biomarker development. Stay tuned.…
Stephani Sutherland, PhD, is a neuroscientist and freelance journalist in Southern California. Follow her on Twitter @SutherlandPhD.
Image credit: Novi Elysa/123RF Stock Photo.