Central pain mechanisms in chronic pain states--maybe it is all in their head.

Rheumatologists might best understand pain and sensory processing by considering that this type of processing is controlled in a manner very similar to immune function. Just as high levels of pro-inflammatory cytokines, or low levels of anti-inflammatory cytokines, can move an individual towards hyperimmune function, there are neurotransmitters that are similarly known to either increase or decrease pain transmission in the CNS. Overall, the analogy of an increased “volume control” or gain“ setting on pain and sensory processing is supported by studies from a variety of sources. Similar to essential hypertension, where a variety of root causes can lead to elevated systemic blood pressure, these disorders represent ”essential hypertension of pain and sensory processing pathways“. Elevated levels of neurotransmitters that tend to be pronociceptive (i.e. on the left side of Figure 3 below) or reduced levels of neurotransmitters that inhibit pain transmission (i.e. on the right side of the figure) have a tendency to increase the volume control, and drugs that block neurotransmitters on the left or augment activity of those on the right will typically be found to be effective treatments, at least for a subset of individuals with this spectrum of illness. The arrows on Figure 3 indicate the direction of the abnormalities in these neurotransmitter levels (either in the CSF or brain) that have been identified to date in fibromyalgia. As noted, in FM, there is evidence for increases in the CSF levels of Substance P, glutamate, nerve growth factor, and brain derived neurotrophic factor, and low levels of the metabolites of serotonin, norepinephrine, and dopamine, any of which could lead to an ”increase in the volume control“ and augmented pain and sensory processing ( 72 - 75 ). The only neurotransmitter system that has been studied to date and not found to be out of line in a direction that would cause augmented pain transmission is the endogenous opioid system. Both CSF levels and brain activity by functional neuroimaging appears to be augmented, not reduced (as would cause augmented pain processing) in FM, which may be why opioidergic drugs do not work well to treat FM and related pain syndromes ( 76 , 77 ). Although the CSS conditions all were originally felt to be autoimmune or inflammatory diseases and then later felt not to be, recent findings are leading to a re-consideration of whether subtle inflammatory changes may be responsible for some of the symptoms seen. Immunological cascades have a role in the maintenance of central sensitivity and chronic pain which is enhanced through release of pro-inflammatory cytokines by CNS glial cells; thus, the traditional paradigm regarding inflammatory versus non-inflammatory pain may gradually become less dichromatic. As may be expected in any complex biological system, a delicate apparatus of checks and balances is at work in the spinal transmission of pain. Multiple inhibitory transmitters act at the spinal level to reduce the ”volume“ of pain transmission. Serotonin, norepinephrine, enkephalins, dopamine and gamma-ammino-butyric – acid (GABA) ( 75 ) are among the better known players in this balance.

Pain is ultimately experienced in the brain, not in the peripheral tissues, and the function of pain processing systems throughout the body markedly influences who has pain and how much pain an individual experiences. Just as we know there is tremendous variability in pain sensitivity between strains of rodents, there is similarly great variability in pain sensitivity among humans. Evidence exists for a strong familial component to FM and all other CSS. Arguably, this component has been best studied in twin studies comparing a variety of functional somatic syndromes, and in fibromyalgia. Regarding the development of FM, Arnold and colleagues showed that the first degree relatives of individuals with FM had an eight-fold greater risk of developing FM compared with those in the general population ( 20 ). Family members of individuals with FM are more sensitive to pressure stimulation (i.e. have a lower pain threshold) than family members of controls, irrespective of the presence of pain. Furthermore, family members of individuals with FM are also much more likely to have other pain syndromes, such as IBS, TMJD, headaches, and other regional pain syndromes ( 40 , 41 ). Similarly, strong genetic predisposition to chronic pain, and to nearly all of the CSS syndromes, has similarly been noted. These observations are congruent with the twin studies that suggest that approximately 50% of the risk of developing one of these disorders is genetic, and 50% environmental. Zubieta et al first showed that the COMT Val158Met polymorphism was responsible for differential pain sensitivity in humans, working in part by modulating the endogenous-opioid system (42)-3}. Diatchenko and colleagues demonstrated that subsets of individuals could be identified based on the findings in 4 COMT single-nucleotide polymorphisms (SNPs), termed low pain sensitive (LPS), average pain sensitive(APS), and high pain sensitive (HPS) groups ( 43 ). These three subgroups are highly predictive of pain sensitivity in a variety of different experimental pain-inducing tasks ( 28 ). A prospective cohort of 240 pain-free individuals phenotyped at baseline and followed for 3 years demonstrated that individuals in the HPS group were 3 times as likely as the others to develop TMJ disorder ( 43 ). A differential effect of COMT in male and female patients was subsequently noted by van Meurs et al ( 44 ) is also not surprising given what is known about COMT. COMT is regulated by estrogen and has been shown to be at least partly responsible for sex differences in several different “phenotypic” characteristics of women and men, including pain sensitivity. Because these types of genetic factors may play a strong role in determining an individual’s experience with pain, it is equally as likely that genetic and epigenetic factors also play key roles in most rheumatic diseases and may help account for sex differences in patient reported outcomes noted in many rheumatologic disorders. Other candidate genes have been identified which may play a role in the susceptibility to central pain mechanisms. This area is constantly evolving, with potential genes identified each year. Studies in a cohort of patients with chronic widespread pain did not show an association between the frequency of a GCH1 “pain-protective” haplotype of GTP cyclohydrolase when compared with controls and no significant associations were observed between the OPRM1 (mu opiod receptor) SNP ( 45 ). Once fibromyalgia is established, by far the most consistently detected objective abnormalities involve pain and sensory processing systems. Evidence of augmented pain and sensory processing is the most reproducible pathogenic feature of these illnesses. Since FM is defined in part by tenderness, considerable work has been performed exploring the potential reason for this phenomenon. The results of two decades of psychophysical pressure pain testing in fibromyalgia have been very instructive ( 46 ). One of the earliest findings in this regard is that tenderness in fibromyalgia is not confined to tender points, but instead extends throughout the entire body ( 47 - 49 ). Theoretically, such diffuse tenderness could be either primarily due to a psychological factor (e.g., hypervigilance, where individuals are too attentive to their surroundings), or neurobiological influence (e.g., the plethora of factors that can lead to temporary or permanent amplification of sensory input) factors. Early studies typically used dolorimetry to assess pressure pain threshold, and concluded that tenderness was in large part related to psychological factors, because these measures of pain threshold were correlated with levels of distress ( 34 , 49 - 52 ). To minimize the biases associated with “ascending” (i.e., the individual knows that the pressure will be predictably increased) measures of pressure pain threshold, Petzke and colleagues performed a series of studies using more sophisticated paradigms using random delivery of pressures ( 53 - 56 ). These studies showed that: 1) the random measures of pressure pain threshold were not influenced by levels of distress of the individual, whereas tender point count and dolorimetry exams were; 2) fibromyalgia patients were much more sensitive to pressure even when these more sophisticated paradigms were used; 3) fibromyalgia patients were not any more “expectant” or “hypervigilant” than controls; and 4) pressure pain thresholds at any four points in the body are highly correlated with the average tenderness at all 18 tender points and 4 “control points” (the thumbnail and forehead). In addition to the heightened sensitivity to pressure noted in fibromyalgia, other types of stimuli applied to the skin are also judged as more painful or noxious by these patients. Fibromyalgia patients also display a decreased threshold to heat ( 56 - 59 ), and electrical stimuli ( 60 ). Gerster and colleagues were the first to demonstrate that fibromyalgia patients also display a low noxious threshold to auditory tones, suggesting that this was a more global problem in sensory processing in some ( 61 ). A recent study by Geisser and colleagues used an identical random staircase paradigm to test fibromyalgia patients’ threshold to the loudness of auditory tones, and to pressure ( 59 ). This study found that fibromyalgia patients displayed low thresholds to both types of stimuli, and the correlation between the results of auditory and pressure pain threshold testing suggested that some of this was due to shared variance, and some unique to one stimulus or the other. The notion that fibromyalgia and related syndromes might represent biological amplification of all sensory stimuli has significant support from functional imaging studies that suggest that the insula is the most consistently hyperactive region. This region has been noted to play a critical role in sensory integration, with the posterior insula serving a purer sensory role, and the anterior insula being associated with the emotional processing of sensations ( 38 , 62 , 63 ). These same findings of hyperalgesia and allodynia have been noted in most of the other conditions acknowledged to be part of this continuum, including IBS, TMJD, tension type headache, idiopathic low back pain, vulvodynia, and interstitial cystitis ( 25 , 64 - 68 ). Brain imaging studies also the existence of central pain augmentation in FM, IBS, low back pain, and several other of these conditions ( 68 - 71 ).

Several drug and non-drug therapies have been shown to be effective for nearly any of the CSS disorders, further reinforcing that this may well be a large overlapping disorder rather than several separate ones. Amongst classes of drugs, substantial data suggest that tricyclic compounds are effective for treating most of the conditions noted ( 78 - 80 ). Newer serotoninnorepinephrine re-uptake inhibitors such as duloxetine and tramadol have similarly been shown to be effective across a broad range of these conditions ( 81 ), and interestingly duloxetine had much earlier been shown to be helpful in treating the pain associated with depression, which is not surprising. The alpha-2-delta ligands such as pregabalin and gabapentin are also being shown to be efficacious in a wide range of these entities ( 81 , 82 ). Figure 4 lists the classes of drugs and their level of evidence in fibromyalgia, but in general those drugs with the highest level of evidence in fibromyalgia are also being shown to work in subsets of individuals with CSS. More importantly, drugs such as duloxetine are being shown to be effective in conditions such as osteoarthritis and low back pain, pointing out the these central mechanisms that are ”front and center“ in patients with syndromes such as fibromyalgia may be also palying prominent roles in conditions heretofore thought to be peripheral pain syndromes. But we have known for some time that hyperalgesia and other central factors, as well as various other indicators of a wide range of ”fibromyaglia-ness, is present in conditions such as osteoarthritis and low back pain. It is of note that any one of these classes of drugs only works well in about a third of patients, which is entirely consistent that this is a strongly genetic – but polygenic – disorder and thus will need different treatments in different individuals. Going back to the “essential hypertension of pain processing pathway” analogy, just as we use 8 – 10 classes of drugs acting in different body systems and at different molecular targets to control hypertension, and individuals may respond very well to one class of anti-hypertensive drug but not another, the same is likely true of CSS syndromes. Individuals may only respond to one of these classes of drugs or may often be on several classes of centrally-acting analgesics (e.g. a low dose of cyclobenzaprine at bedtime, pregabalin or gabapentin either just at bedtime or twice daily, and a serotoninnorepineprine reuptake inhibitor such as duloxetine or milnacipran during the day. However, our current pharmacological armamentarium is not nearly as well developed for central pain as for essential hypertension, which is likely one of the reason that these syndromes are often still difficult to treat. Figure 4 also points out that classes of drugs that are quite effective for “peripheral” pain due to damage or inflammation in peripheral tissues such as NSAIDs and opiods are not effective analgesics in central pain states. There are even some data suggesting that giving opioids to individuals with central pain states may worsen their pain, by leading to opioid-induced hyperalgesia that could augment and worsen the baseline hyperagesia that may be playing a central pathogenic role in these conditions. Just as many pharmacological therapies work across all or most of these conditions, similarly non-pharmacological therapies such as education, exercise, and cognitive behavioral therapy have been demonstrated to be effective across nearly all of the CSS conditions ( 83 - 85 ).

Until recently, pain in OA was largely been attributed to joint damage, and nearly all previous therapies were aimed at treating pain localized to the joint, including exercise, topical analgesics, oral nonsteroidal anti-inflammatory drugs (NSAIDs), injections of various preparations, opioids, and joint replacment. It has been evident for some time that peripheral factors can, at best, only partially explain the pain and other symptoms suffered by individuals with OA. Population-based studies consistently show a poor relationship between the degree of “pathology” in OA and reported pain intensity. In fact, in population-based studies approximately 30 – 40% of KOA patients with the most severe forms of radiographic knee OA (KOA; Kellgren-Lawrence grades III and IV) have no pain.( 86 , 87 ). Furthermore, many OA patients that theoretically have a pathological process confined to one or only a few joints experience symptoms that cannot be explained by the “peripheral” model of pathogenesis. Multifocal pain in areas not commonly affected by OA is common in individuals identified as having KOA.( 88 ). Similarly, other somatic symptoms that would not be explained by a purely peripheral problem are often seen. For example, a recent study from our group showed that fatigue was a prominent problem in individuals with KOA, and in many individuals it was a more functionally limiting symptom than was pain ( 89 ). Insomnia likewise occurs in a sizable proportion of OA patients, and has been shown to improve along with pain in the small proportion of individuals with KOA that responded to opioids, suggesting that insomnia and pain may have overlapping biological underpinnings ( 90 , 91 ). Peripheral factors are clearly important in OA, as seen in a recent study by Neogi and colleagues that elegantly demonstrated that for individuals with asymmetric KOA, pain levels in each knee were strongly related to joint space narrowing in the affected knee ( 92 ). However, some individuals have central factors that are superimposed upon the more traditional peripheral factors leading to the need for a broader and more flexible approach to diagnosis and treatment of pain in OA. OA, like nearly all other chronic pain states is likely a “mixed pain state,” with individual variability in the relative balance of peripheral (i.e. nociceptive) and central elements of pain. Thus, until recently it had been clear that entirely “peripheral” theories regarding OA were incomplete, and only recently have studies begun to explore the potential CNS contributions to this condition. Several groups have demonstrated that small groups of OA patients display diffuse hyperalgesia to mechanical or heat stimuli ( 93 - 95 ). Kosek demonstrated that individuals with OA of the hip had reduced descending analgesic activity, which partially normalized following hip arthroplasy( 96 , 97 ). Those having baseline hyperalgesia required more opiods in the perioperative setting. Harden and colleagues ( 98 ) compared 37 subjects with KOA and 35 controls on a battery of psychophysical tests and found that the OA subjects demonstrated lower overall mechanical pain thresholds compared to controls, as well as greater mechanical and thermal temporal summation than controls, whereas there were no differences among the subjects for any of the remaining experimental pain testing paradigms testing. Most recently, Gwilym and colleagues used both experimental pain testing and functional neuroimaging procedures to show augmented CNS processing of pain in 20 OA patients ( 99 ). Perhaps the strongest evidence of CNS factors being important in OA is the finding that the drug duloxetine, a centrally acting analgesic, is efficacious in individuals with KOA ( 100 ).

In the past few decades, our understanding of chronic pain has evolved tremendously. The study of the heretofore considered “functional” or “psychosomatic” conditions such as fibromyalgia and IBS has taught us about the mechanisms that may underlie chronic pain or other somatic syndromes, in individuals without fibromyalgia per se . A better understanding of the underlying mechanisms and most effective treatment for this spectrum of illness is critical to rheumatologists, because as Wolfe has taught us, many patients with chronic pain disorders have variable degrees of “fibromyalgia-ness”. When this occurs, we need to treat both the peripheral and central elements of pain along with other somatic symptoms. The era of evidence-based, individualized analgesia in chronic pain is upon us.

The large numbers of studies that have directly compared the rate of FM in other related CSS, and vice-versa, or rates of co-morbidities between syndromes, will not be reviewed because these have been covered elsewhere. Instead, we will focus on describing several lines of research that have better clarified the overall picture with respect to the inter-relationships of these symptoms and syndromes. Kato and colleagues, using a large Swedish twin registry, have performed a series of studies first showing the co-morbidities with chronic widespread pain, and then later examined a number of functional somatic syndromes and the relationship of these symptoms to those of depression and anxiety ( 18 , 26 ). These studies clearly demonstrated that functional somatic syndromes such as FM, CFS, IBS, and headache have latent traits (e.g. multifocal pain, fatigue, memory and sleep difficulties) that are different than (but overlap somewhat with) psychiatric conditions such as anxiety and depression. Interestingly, the findings are exactly those found in functional neuroimaging studies, where, for example, individuals with FM alone primarily have increased activity in the regions of the brain that code for the sensory intensity of stimuli (e.g., the primary and secondary somatosensory cortices, posterior insula, thalamus) whereas the FM patients with co-morbid depression also have increased activation in brain regions coding for the affective processing of pain, such as the amygdala and anterior insula ( 27 ). The notion that there are two overlapping sets of traits, one being pain and sensory amplification, and the other being mood and affect, is also seen in other genetic studies of idiopathic pain syndromes ( 28 , 29 ). Twin studies have also been useful in helping examine potential underlying mechanisms versus “epiphenomena”. Buchwald and colleagues have compared identical twins with and without symptoms and have found that, in many cases, the two twins share abnormalities in sleep or immune function, yet have markedly different symptom profiles. These investigators have likewise suggested that this is evidence of a problem with perceptual amplification in the affected twins ( 30 ). As with most illnesses that may have a genetic underpinning, environmental factors may play a prominent role in triggering the development of central sensitivity syndrome related conditions. Environmental “stressors” temporally associated with the development of either fibromyalgia or chronic fatigue syndrome include early life trauma, physical trauma (especially involving the trunk), certain infections such as Hepatitis C, Epstein Barr virus, parvovirus, Lyme disease, and emotional stress. The disorder is also associated with other regional pain conditions or autoimmune disorders ( 31 - 33 ). Of note, each of these stressors only leads to CWP or fibromyalgia in approximately 5 – 10% of individuals who are exposed; the overwhelming majority of individuals who experience these same infections or other stressful events regain their baseline state of health. Although an early advocate of the FM construct, more recently, Wolfe has become critical of the construct, arguing that FM is not a discrete illness but rather the end of a continuum ( 34 ). A contrary opinion is that FM is both a discrete illness (i.e., an individual with FM) and the end of a continuum of pain processing. Wolfe has performed seminal work in showing that the degree of fibromyalgia symptoms an individual with any rheumatic disorder has (including individuals with osteoarthritis, rheumatoid arthritis, regional pain syndromes, etc.) as measured by his Symptom Inventory is closely correlated with their level of pain and/or disability, even if they do have a “peripheral” cause for their pain ( 35 ). The Symptom Inventory (or other measures of the current or lifetime level of somatic symptoms or “somatization”) is a very good measure for whether an individual has a CSS or an element of central sensitivity, no matter if they also have a peripheral cause for their pain or not. We predict that genetic factors such as those discussed below will be shown to be highly predictive of these measures, and that functional imaging (e.g., hyperactivity or increases in excitatory neurotransmitters in brain regions such the insula that code for the intensity of all sensory information) and other research methods will similarly show that these self-report measures will have strong biological underpinnings ( 36 - 39 ).