This is Part 2 of a two-part report on selected talks from the NIH Pain Consortium’s 7th Annual Symposium on Advances in Pain Research, and from an FDA CDER public scientific workshop held in conjunction with the symposium. See also Part 1.
In a departure from previous symposia organized by the NIH Pain Consortium, the 2012 version was held in conjunction with a US Food and Drug Administration (FDA) Center for Drug Evaluation and Research (CDER) workshop focused mainly on whether opioids relieve chronic non-cancer pain. The workshop took place against a backdrop of increasing concern, by experts and the lay public alike, about opioid misuse and abuse. However, the workshop focused primarily on the question of efficacy, sometimes to the frustration of patients, patient advocates, and other stakeholders who spoke of their personal experiences with opioids during an open public hearing during the workshop, as well as during the scientific sessions’ discussion periods. The workshop comprised talks on the basic science of pain, the epidemiology of chronic pain in the US, and, most of all, the randomized controlled trial (RCT) and non-RCT evidence concerning opioid analgesic efficacy. A final session provided additional patient and clinician perspectives on drug treatment of chronic pain.
Basic science
Treatments approved for neuropathic pain do not provide clinically meaningful improvements for the majority—75 percent, in fact—of patients, said Clifford Woolf, Children’s Hospital Boston and Harvard Medical School, US, in his talk that opened the basic science session. To improve that success rate, Woolf said, pain researchers and clinicians need to better understand individual pain phenotypes—the unique constellation of symptoms characteristic of each patient—and then link those phenotypes to their underlying neural mechanisms. That is in contrast to the pain field’s current practice of classifying patients according to disease etiology. Summarizing the conceptual framework he published recently with colleagues (von Hehn et al., 2012), Woolf emphasized that pain patients are not all alike, even those who receive the same clinical diagnosis. Individuals with diabetic neuropathy, for example, differ widely in the degree to which positive (evoked pain) versus negative (e.g., changes in temperature sensitivity) versus spontaneous pain symptoms affect them. Furthermore, any single symptom can result from different underlying mechanisms, including injury to sensory nerves, peripheral sensitization, central sensitization, ectopic activity of neurons, as well as changes in the brain. Woolf’s message was that connecting symptoms to the molecular machinery driving them will help the pain field to more clearly grasp why so many patients fail to respond to current treatments, and to better target treatments to the right patients.
Quantitative sensory testing (QST) is one promising route to answering the call for mechanism-based pain diagnosis, said Roger Fillingim, University of Florida, Gainesville, US. One mechanistic question that QST can address is how people with pain differ from those without pain. For instance, one of Fillingim’s recent studies used QST to assess pressure, mechanical, or heat pain sensitivity, and the technique uncovered differences between patients with temporomandibular disorders (TMD) and healthy controls (Greenspan et al., 2011; also see PRF related news story). Recent QST studies by Fillingim and colleagues also unearthed hypersensitivity to nociceptive stimuli in patients with irritable bowel syndrome (Zhou et al., 2010), as well as deficiencies in endogenous pain inhibitory function in IBS and TMD (King et al., 2009). Another group documented local and spreading central sensitization in knee osteoarthritis patients (Arendt-Nielsen et al., 2010), all compared to healthy controls. Fillingim has shown that QST can also be used to group pain-free people based on their responses to experimental pain modalities like heat, pressure, and ischemia (Hastie et al., 2005), and another group has found that QST can also categorize patients with different neuropathic pain syndromes (Maier et al., 2010). Fillingim described work from others showing how QST can illuminate another mechanism-based issue: the response to treatment. For instance, a 2006 study found that baseline heat pain thresholds predicted the response to opioids in post-herpetic neuralgia patients (Edwards et al., 2006), and a more recent investigation of diabetic neuropathy patients showed that patients’ baseline conditioned pain modulation predicted the response to duloxetine (Yarnitsky et al., 2012). The recent use of QST to reveal different pain mechanisms underlying right- and left-sided pain in a spinal cord injury patient, even though the patient reported the same burning and prickling pain on both sides, is a tantalizing example of the promise of QST to uncover pain mechanisms (Westermann et al., 2012), Fillingim said. The pain on each side exhibited different responses to drug treatment, revealing how a better understanding of mechanisms could lead to more targeted treatments.
In the next talk, William Maixner, University of North Carolina at Chapel Hill, US, used his work on the catechol-O-methyltransferase (COMT) gene as a vignette to illustrate the path by which the discovery of genetic associations might lead to new molecular targets for novel pain treatments, as well as to better use of existing therapies. In 2005, Maixner and colleagues identified three haplotypes of COMT conferring low, average, or high experimental pain sensitivity and an increased risk of TMD (Diatchenko et al., 2005). The identification of those genetic variants led to studies indicating that the variants had functional significance, with differences in the messenger RNA structure corresponding to each haplotype predicting COMT protein levels and enzymatic activity (Nackley et al., 2006). Those in-vitro functional studies led to in-vivo investigations showing that decreasing COMT activity by pharmacologically using a COMT inhibitor in rats increased mechanical and thermal pain sensitivity, and that the effect required activation of β-adrenergic receptors (Nackley et al. 2007). Finally, from rodents to people: a randomized, double-blind, placebo-controlled crossover pilot study of 40 subjects with TMD found that COMT haplotypes in those subjects predicted the analgesic effect of treatment with the β-adrenergic antagonist propranolol (Tchivileva et al., 2010). Propranolol is approved for use in a variety of conditions, including high blood pressure, angina, and migraine headaches, and Maixner has also found it improves clinical pain measures not only in TMD patients, but also in people with fibromyalgia (Light et al., 2009), at least over the short term. Thus, his work offers the possibility of using genetics to identify individuals most likely to benefit from therapy.
In the final talk of the basic science session, Sean Mackey, Stanford University, Palo Alto, California, US, focused on the uses of neuroimaging to advance knowledge of pain. First, Mackey said that neuroimaging is relevant in trying to understand differences in pain sensitivity from one person to the next. For instance, it has long been known that normal volunteers with high pain sensitivity show increased activity, as measured by functional MRI (fMRI), in the primary somatosensory, anterior cingulate, and prefrontal cortical areas compared to volunteers with low sensitivity to pain (Coghill et al., 2003), and one of Mackey’s own fMRI studies of healthy subjects pinpointed associations between pain-related fear and anxiety and activity in distinct brain regions (Ochsner et al., 2006). Second, Mackey said that neuroimaging reveals how pain changes the nervous system. For example, he pointed to one group’s study of chronic back pain patients that found decreases in the gray matter density of the prefrontal cortex and thalamus (Apkarian et al., 2004). In addition to changes in brain structure, neuroimaging has documented alterations in brain connectivity in patients with diabetic neuropathic pain, for example, and in patients with fibromyalgia (Cauda et al., 2009; Napadow et al., 2010). Third, neuroimaging indicates the reversibility of pain-induced changes in the brain over time; a study last year of CLBP patients, for instance, found that aberrations in brain anatomy and function of the dorsolateral prefrontal cortex had reversed six months after successful spinal surgery or facet joint block treatment (Seminowicz et al., 2011). Mackey’s own work shows how pain treatment alters the brain. His group has reported correlations between morphine use and volumetric increases or decreases in a number of brain regions in CLBP patients after just one month of treatment; these morphine-induced alterations persisted for four and a half months after treatment was stopped (Younger et al., 2011). In the future, Mackey is particularly interested in learning whether neuroimaging can be developed to replace self-report—the latter is considered one of the weaknesses of current clinical studies—and serve as an objective biomarker of chronic pain.
Epidemiology suggests a wide lens is needed...
The prevalence of common, recurrent pain conditions like migraine and back pain that are severe enough to disable reaches 10 percent in the general US population, said Walter Stewart, Geisinger Health System, Danville, Pennsylvania, US, during a session on epidemiology after the basic science discussion. Meanwhile, in the US, 4.5 million adults—including those with severe pain but also increasing numbers of people who use them for non-medical purposes—take opioids regularly. As a result, the rate of fatal drug overdoses involving prescription opioids has increased fourfold over the last decade, to 16,000 deaths in 2009, said Michael Von Korff, Group Health Research Institute, Seattle, Washington, US. Those striking facts—millions of people with disabling pain, millions using opioids, including those without pain, and thousands dying—were recognizable to the speakers at the public hearing following the epidemiology session. The statistics were familiar not only to those who shared negative experiences concerning opioids, but also to those with positive stories; speakers who fall into the latter category worried about the condemnation of an entire class of drugs that they have learned to use safely and appropriately.
But the focus narrows: do opioids relieve chronic non-cancer pain?
The frustration of those who spoke at the public hearing was not entirely addressed in the following day’s scientific session, which focused narrowly on the fundamental question of opioid efficacy in chronic pain. Data from RCTs of chronic nociceptive pain do show that opioids alleviate pain, at least over the short term, said Srinivasa Raja, Johns Hopkins School of Medicine, Baltimore, Maryland, US. For example, he cited studies of oxymorphone and buprenorphine showing that those opioid analgesics decreased pain in patients with CLBP over 12 weeks of treatment (Katz et al., 2007; Steiner et al., 2011). Similar results come from RCTs of opioids for treatment of osteoarthritis (OA)—a 2009 Cochrane Review of 10 trials with more than 2,000 subjects found that oral or transdermal opioids relieved pain more effectively than did control interventions in hip and knee OA (Nüesch et al., 2009). Raja said that RCTs also support the short-term efficacy of both peripheral and central opioids for neuropathic pain conditions, such as post-herpetic neuralgia and diabetic neuropathy, with recent meta-analyses of RCTs finding that opioids had consistent short-term efficacy in the neuropathic pain setting (Finnerup et al., 2010; Papaleontiou et al., 2010). A recent meta-analysis also concluded from its examination of 62 RCTs that opioids were equally effective for both nociceptive and neuropathic chronic non-cancer pain (Furlan et al., 2011).
In contrast to the evidence base for short-term efficacy of opioids, that for long-term efficacy is almost non-existent, according to Raja. Another Cochrane Review did find that long-term opioid use produced clinically significant reductions in chronic non-cancer pain (nociceptive, neuropathic, or both) (Noble et al., 2010). Yet, of the 26 studies of at least six months of opioid treatment included in the analysis, 25 were either case studies or open-label extensions of RCTs, and the studies altogether included only about 5,000 patients. Raja concluded that opioids provide short-term relief for nociceptive and neuropathic pain, but the data are too sparse to make definitive pronouncements about long-term efficacy.
By looking to observational and epidemiological data, Jane Ballantyne, University of Washington, Seattle, US, next addressed the matter of whether opioids improve pain. Her work and that of others shows that observational data (e.g., case studies and open-label follow-up studies to RCTs) do suggest pain relief from opioids in chronic pain patients who take moderate doses of drug, and who were usually followed no more than two years (Ballantyne and Shin, 2008). However, the data do not allow for clear pronouncements on functioning or quality of life, nor do they provide clear findings on addiction risk (Noble et. al., 2010). Epidemiological data tell a much different story: they support neither pain relief nor functional improvement. In fact, opioids have been linked to increased self-reports of severe pain and poor health, as well as a poorer quality of life, lack of employment, and increased use of the healthcare system (Eriksen et al., 2006). Meanwhile, studies of workers’ compensation claims have revealed an association between early prescription of opioids and a delayed return to work in patients with low back pain/injuries (e.g., Webster et al., 2007; Franklin et al., 2008). Ballantyne explained that epidemiological studies include patients who are sicker, take higher doses of drug, do not receive treatment in careful practice settings, and receive opioids for a much broader assortment of diagnoses, compared to patients in observational studies, and that such differences may explain the striking disparity between the two datasets. The problems observed with opioids appear to be concentrated in high-dose users. The data also suggest that tolerance is linked closely to dependence and is the gateway to use of higher doses. In addition, pain and function often get better in opioid users who are able to taper off of the drugs. Areas for future research include the risk factors for drug escalation—does it have to do with the drug, the patient, or both? Also unclear is which patients do well on high doses, or whether a cutoff dose of opioids can be identified beyond which safety will be compromised.
While Raja and Ballantyne focused on what the data say, Dennis Turk, University of Washington, Seattle, US, discussed how the limitations of research on the pain-relieving effects of opioids make gathering meaningful data difficult. Many of the weaknesses of RCTs in general, such as difficulties in blinding, a reluctance of patients to be randomized to placebos, and the mismatch between what is statistically significant in the carefully controlled setting of an RCT—efficacy—and what is clinically significant in the “real-world” environment of physician offices—effectiveness—also apply to opioid RCTs. A perfect illustration of the problem is that the maximum dosages of opioids used in RCTs are much lower than those seen in clinical practice. Other limitations commonly seen in opioid RCTs, including their short duration (a five-week average), increased side effects and dropout rates, and a reliance on self-report, are also not unique to those RCTs. Turk said the interpretation of all RCTs is further complicated by the bias to publish positive findings. Furthermore, RCTs financially supported by industry are more likely to report positive results, and 80 percent of all RCTs are, in fact, funded by industry. Turk also stressed that, in general, systematic reviews and meta-analyses of RCTs often confuse clinical decision making rather than aid it. Actually, such investigations often reach different conclusions even though they rely on the same data, an unsurprising finding in light of the fact that 49 different systems are available to reviewers to rate the strength of evidence from RCTs. In fact, he said it is increasingly common for a literature search on a particular topic to uncover more systematic reviews/meta-analyses than RCTs upon which the former are based. Turk stressed the importance of being aware of all of the biases that make the interpretation of RCTs so challenging, and concluded that the data now available simply do not allow for confident assertions about the long-term efficacy of opioids.
In a discussion that followed the session on whether opioids relieve pain, Marc Hochberg, University of Maryland School of Medicine, Baltimore, US, said that “patients don’t fail drugs; drugs fail patients.” The scientific consensus of the workshop was that opioids are not failing patients over the short term, though the long-term picture is unclear, if the sole criterion is whether the drugs alleviate pain. But the public consensus was that the issue of opioid efficacy cannot be divorced from the complex problems of misuse and abuse that those powerful analgesics carry with them.
Note: The symposium and workshop are available to view on the NIH website as videocasts: Click to view Day 1, Day 2, or Day 3.
