The common link between functional somatic syndromes may be central sensitisation

Highlights

• Functional somatic syndromes may be related to central sensitisation (CS).

• CS may be a common neurophysiological process that explains functional symptoms.

• CS may represent endophenotypic vulnerability to the development of these syndromes.

• Investigation of CS in these syndromes may lead to new treatment options.

Abstract

Objectives

Functional somatic syndromes are common and disabling conditions that all include chronic pain, and which may be related to central nervous system sensitisation. Here, we address the concept of central sensitisation as a physiological basis for the functional somatic syndromes.

Methods

A narrative review of the current literature on central sensitisation and physiological studies in the functional somatic syndromes.

Results

Central sensitisation may be a common neurophysiological process that is able to explain non-painful as well as painful symptoms in these disorders. Furthermore, central sensitisation may represent an endophenotypic vulnerability to the development of these syndromes that potentially explains why they cluster together.

Conclusions

Further research is needed to verify these findings, including prospective studies and the standardisation of combined methods of investigation in the study of central sensitisation in functional somatic syndromes. In turn, this may lead to new explanatory mechanisms and treatments being evaluated. Our conclusions add to the debate over the nomenclature of these syndromes but importantly also provide an explanation for our patients.

Introduction

Functional somatic syndromes (FSS) are conditions in which physical symptoms are not fully explained by an established, alternative medical disorder (Table 1). FSS are common, disabling, and incur a significant use of health care resources and are a source of financial burden borne by patients, their families and society [1], [2]. The conceptualisation of these disorders as being ‘unexplained’ is not only a potential barrier to improved care [3], but it is also challenged by our increasing understanding of them from a neurophysiological perspective.

The nosology of FSS is undetermined. This is in part due to uncertainty regarding both pathophysiology and aetiology as reflected in the DSM V review of the classification of somatic syndromes [4] and the borderland between mental and physical health that they inhabit [5]. FSS encompass a wide range of apparently distinct syndromes (Table 1). However, symptoms overlap between the disorders to such an extent that some have suggested that there is one generic FSS, rather than many [6], whilst others have proposed that FSS are discrete disorders, grouped together by common symptoms, but separated by individual pathophysiologies [6]. These opposing arguments have led to the coloquialised question as to whether we should ‘lump’ these disorders together, or ‘split’ them apart [6], [7].

Patients with one FSS are more likely to also suffer from another [8]. For example 51% of patients with chronic fatigue syndrome (CFS) and 49% of patient with fibromyalgia (FM) also have irritable bowel syndrome (IBS) [9]. This may be because these disorders seem to share common predisposing risk factors (but not precipitants) [10], [11]. These cited studies examined patients with either IBS or CFS and reported predisposing factors common to both conditions (other FSS and mood disorders), in addition to precipitating factors that differentiated them (different infectious agents) [10], [11].

Research into the pathophysiological mechanisms of FSS is made difficult by the nature of these disorders, which represent symptom clusters across a wide array of different systems. One approach has been to look at the individual symptoms that predominate in a given FSS, e.g. fatigue in CFS or bowel symptoms in IBS. These symptoms generally involve a combination of different physiological, cognitive and affective drivers, representing an interaction between the central nervous system (CNS) and the dominant system to which the defining symptoms are attributed. Defining and then modelling diffuse symptoms such as fatigue is particularly difficult. Pain, on the other hand, represents an easier target. It is a diagnostic criterion of all FSS and the commonest symptom overall [12]. As such, it represents a unifying symptom, a better understanding of which may improve our understanding of FSS as a whole. To this end, some authors have considered these syndromes together as disorders that reflect a state of CNS mediated hypersensitivity to painful stimuli [13], [14] — ‘central sensitisation’ (CS).

Pain represents an experience that is influenced not only by sensation, but also by context and prior experience. The experience of pain is distinct from nociception [15]: nociception describes afferent neural activity transmitting sensory information about stimuli that have the potential to cause tissue damage [16], whereas pain is an emergent phenomenon, a conscious experience that requires cortical activity [17], and which can occur in the absence of nociception. Acute pain originates from nociceptors, activating two types of nerve fibre (low threshold fast A delta fibres and high threshold slow unmyelinated C fibres). These fibres synapse onto both wide dynamic range (those that are responsive to all sensory modalities and to a broad range stimulus intensity) and high-threshold neurons in the dorsal horn of the spinal cord, depending on the nature of the painful stimulus encountered.

The superficial dorsal horn is comprised of lamina I and II. Lamina I plays a key role in the modulation of pain transmission with 85% of its neurons being nociceptive with a high threshold for excitation, whilst only 15% have a wide dynamic range, responding to lower thresholds. In comparison, the majority of neurons in lamina V, located in the neck of the dorsal horn, also have a wide dynamic range, and so are predominantly non-nociceptive. Neurons from lamina II terminate locally within the dorsal horn, whilst those from lamina I have long axons that travel in the parabrachial pathway [18]. This pathway along with the spinothalamic tract, carries information to the parabrachial area in the midbrain from where it projects to higher cortical centres, including the hypothalamus, the amygdala and parts of the thalamus [19]. These higher centres form part of a non-specific network of structures previously referred to as a ‘neuromatrix’ [20] that is thought to be involved in, but not specific to the perception of pain [21]. The different structures within this network contribute to the evaluative, affective and sensory interpretation of the painful stimuli. The parabrachial pathway also projects to other centres involved in the descending control of pain, including the rostro-ventral medulla and the periaqueductal grey area [22]. These descending control mechanisms are responsible for a reduction in the sensation of pain and the inhibition of its spread upon the perception of pain (see Fig. 1a).

Pain perception exhibits neuroplasticity, whereby repeated nociception may result in either habituation (reduced response) or sensitisation (increased response). As such, continued stimulation may result in increased neuronal responsivity or sensitisation [23], depending on the intensity and temporal features of the stimulation. CS refers to a ‘pain phenotype’ generated by processes that result in the amplification of the pain, which are thought to underlie many chronic pain states. This sensitised state involves both spinal mechanisms, at the level of the dorsal horn and below and supraspinal mechanisms that involve the disproportionately augmented response of a network of higher brain centres [20], with the latter subsequently contributing to the descending control of afferent spinal neurotransmission. CS therefore involves a dual process of spinal sensitisation and augmentation of this neural network. It is important to note, however that these processes occur in parallel rather than in series and that there is an overlap in the pathways involved in spinal and supraspinal mechanisms of pain perception (see Fig. 1a and b).

Spinal sensitisation may follow the activation of nociceptive pathways by peripheral injury, such as that caused by trauma or inflammation, resulting in the release of substance P and pro-inflammatory cytokines [24] and the activation of spinal cord glial cells [25]. In turn, this may result in the increased release of glutamate [26] and the long-term potentiation of such changes at the level of the dorsal horn. This dorsal horn sensitisation causes the ‘gate’ to spinal pain pathways to remain open even without further noxious stimulation. This effect is accentuated by a reduction in descending inhibition that would ordinarily dampen the sensation of pain [27]. The overall effect is to generate a state of increased activity in ascending (activating) pathways and a reduction in activity in descending (inhibitory) pathways. The resultant hypersensitivity is clinically evident as increased sensitivity to painful (hyperalgesia) and non-painful (allodynia) stimuli (see Fig. 1a & b) [15], [28]. The loss of descending inhibition also means that the normal process by which the spread of pain is inhibited is lost, resulting in both secondary hyperalgesia (hypersensitivity in neighbouring dermatomes) and widespread hyperalgesia (hypersensitivity at remote sites). When combined with increased ascending pathway activity, this results in the slow summation of stimuli, with stimuli of the same intensity becoming more painful (enhanced temporal summation or wind-up) [29], [30]. These phenomena have been demonstrated in a variety of FSS, including IBS [31], [32], fibromyalgia [33], [34] and CFS [35], [36] (see Table 2).

These manifestations of CNS sensitisation can be assessed using quantitative sensory testing (QST). QST is a well-validated research tool involving psychophysical methods that systematically document alterations and reorganization of nervous system function and, in particular, of the nociceptive system [37]. As such, QST provides insights into underlying pain mechanisms and is able to identify central nervous system sensitisation [38], [39]. The most commonly applied tests are measures of enhanced temporal summation, widespread hyperalgesia and conditioned pain modulation (CPM). It is beyond the scope of this article to detail the execution of these tests and their various modalities, which have been reviewed in detail elsewhere [37]. Although QST is able to discriminate between CS and non-CS, it is generally accepted that QST is a measure of predominantly spinal mechanisms, albeit that these are influenced supraspinally by higher cortical centres and so does not exclude the possibility of further modulation. Furthermore, the utility of QST in detecting CS is determined, in part, by the battery of tests employed with some providing insufficient evidence for a sensitised state when used in isolation [40].

During the normal perception of pain, the sensory, evaluative and affective attributes of a stimulus are applied by a network of higher brain centres [20]. In CS, this neural network becomes hyper-responsive. Each region of this network, or combinations thereof, serve different functions in the experience of pain [41] which, when combined, contribute to different aspects of the experience of pain, such as unpleasantness. One such subdivision provides a triumvirate made up of lateral (sensory and discriminative), medial (affective and motivational) and frontal (cognitive and evaluative) systems [41], [42]. The relative contribution of these components to the pain experience may vary between individuals and pain disorders, potentially contributing to whether a chronic pain disorder develops, the nature of that disorder, and possibly the individual variance in the experience of pain. This is important in understanding how chronic pain differs from acute pain, with the focus being more on the sensory component in acute pain, and the motivational and evaluative systems being predominant in chronic pain. This neural network therefore demonstrates a multimodal response to nociceptive, somatosensory, auditory and visual stimuli [43]. Furthermore, these responses are more dependent on novelty and context than the nature or modality of the stimulus [43].

Augmented responses to painful stimuli have been shown in several different FSS (see Table 2) and the concept of multimodality is evident in these conditions. For example, in both FM [44] and CFS [45], neuroimaging studies have shown increased neural recruitment (as demonstrated by increased regional activation) with cognitive tasks, reflecting a centrally mediated state of multimodal hyper-responsivity, whilst in FM, low thresholds to auditory tones correlate with pressure pain thresholds suggesting that these are in part due to a shared variance between pressure pain and auditory thresholds [46].

The insula and striatum are two of the most commonly identified components of the neural network activated by pain and contribute to the affective/motivational and sensory/discriminative subsystems of this network. The insula appears to act as the integration centre for multimodal inputs involved in pain [47], interoception [48], threat detection associated with bodily arousal [49], emotional regulation [50] and motivational salience in learning models [51]. The integration of this information enables the insula to serve as a site for comparison between expected and perceived sensory mismatches (‘mismatch negativity’), that in turn allows for accurate self-monitoring and the preferential processing of salient (important) sensory information [52].

Salience does not refer to the attributes of a given stimulus alone but rather their comparison with those of surrounding stimuli and the ability of a stimulus to stand out from others [53]. The salience of a stimulus is dependent upon speed of onset [54], novelty [55], context and past experience [56] and its application is a function of part of the neural network that responds to pain — the ventral striatum [57]. The detection of salience allows for the assessment of mismatch negativity based upon prior exposure [56], which in turn facilitates attentional processing and response generation [58]. Both the application and recognition of salience are therefore a vital part of threat detection essential to an organism's survival [59]. Nociception is vital for threat detection, representing the most proximal and dangerous of threats; nociceptive stimuli are consequently highly salient. Given that the neural network activated by pain is also involved in threat detection and that it can be influenced by context and novelty, it has been proposed that this system's primary role is in salience detection and threat appraisal [43], [60].

The initial processing of salience may occur as early as the spinal cord; peripheral nociceptors respond preferentially to highly salient stimuli [61], including pain. This detection of salience is facilitated by mechanisms of descending inhibition, responsible for the suppression of less relevant stimuli, so that only the most salient information is processed within this neural network. When this mechanism fails, such as in spinal sensitisation, non-salient information is processed to the same extent as salient information, generating an attentional bias towards ordinarily non-salient stimuli and resulting in sensory amplification [62]. In short, the combination of spinal and supraspinal mechanisms involved in CS allows the processing of all information as salient, with threat detection becoming unfiltered with a near-permanent state of response readiness. In turn, this would have a direct effect upon higher cortical areas responsible for the assessment of both mismatch negativity and response prediction as all stimuli may generate a greater activation and response, than would be predicted based upon prior experience of a similar context.

In the context of the FSS, therefore, an explanation for the disparity between stimulus and perception, and the lower thresholds for stimulus tolerance in multiple modalities typical of these disorders, is related to the following: a loss of censorship of peripheral nociceptors, afferent messages at the level of the spinal cord, an increase in mismatch negativity, and the application of salience and threat detection at the level of the neural network.

Evidence for CS in the FSS is growing (Table 2) [13], from various different forms of investigation, including QST [31], [32], [33], [34], [35], [36], [63], [64], [65], [66], [67], [68] and functional neuroimaging [66], [69], [70], [71], [72], [73]. Further support comes from the successful use of centrally acting agents to treat the disorders [74], [75], [76], [77], [78], [79], [80], implying the stabilisation of dysregulated CNS mechanisms (Table 2). However, not all FSS demonstrate sensitisation to all modalities of stimuli [81]. This variation might suggest heterogeneity amongst FSS; it is potentially explained by the fact that standardised protocols are often lacking, making comparisons between the individual FSS difficult. In particular, the majority of earlier studies in this area have relied on tests of temporal summation alone to assess for the presence of CS. Although the facilitated degree of temporal summation indicates an enhanced central integrative mechanism (i.e. CS) [30], this may be insufficient to adequately distinguish between CS and non-CS when used in isolation [40] rather than, for example, in combination with an assessment of descending inhibition such as CPM.

A variety of environmental and psychosocial factors are recognised as risks for FSS [82]. These include childhood adversity and personality traits as predisposing factors, and physical trauma and infections as triggers [83]. Any kind of early life adversity is more commonly reported in all the FSS [10], [11], [84], although this association may vary across FSS [85]. This predisposition may be governed by the effects of such events on the neuroendocrine stress axis [86], as demonstrated by the association between adverse childhood experiences with hypocortisolism in FSS and chronic pain [87], [88]. HPA axis involvement has been implicated by nucleotide polymorphisms in sub-groups of CFS [89], which supports its role in predisposition or cause rather than effect of these disorders. Such a mechanism may affect the consequences of events experienced as an adult, such as a precipitating infection or other stressors. The biological mechanism involved appears related to centrally sensitising phenomena, with early adversity being associated with neuroendocrine sensitisation [87], limbic system hyper-responsiveness and reduced hippocampal volumes [90]. These may be governed by epigenetic changes [91], [92] and not only confer susceptibility to infections or their immune response, but also to stress related illnesses such as affective disorders and the FSS [87], [93].

A genetic predisposition may exist for the FSS (e.g. FM [94]), which is independent from the mood disorders that frequently co-occur. The Swedish twin studies [82], [95], [96] examined the symptom profile in over 30,000 identical twins for latent traits in CFS, FM, IBS, recurrent headache, major depressive disorder and generalised anxiety disorder. Two vulnerabilities were reported for the development of FSS; one was more heritable and related to the development of a mood and anxiety disorders, whilst the other was acquired and specific to the particular FSS developed. This overlap has been noted in other studies of chronic disorders of idiopathic pain [97]. These data suggest that genes may “lump” these disorders together, whilst environmental influences split them apart into their individual syndromes.

Sensitivity to pain of itself may be a heritable trait [98]. Specifically, catecholamine-O-methyl transferase (COMT) val/met polymorphisms may explain individual differences in sensitivity to pain [99]. The association between COMT polymorphisms and neural network activation in response to painful stimuli correlates with the stimulus intensity required to induce a constant pain sensation [100]. COMT polymorphisms define low, average and high pain sensitive groups and the high sensitivity group is more likely to develop chronic disorders of idiopathic pain [101]. COMT is also influenced by oestrogen. Oestrogen receptor density is high in brain areas involved in pain perception [102] and in women, pain sensitivity varies during the menstrual cycle with lower pain thresholds being present during the luteal phase [103]. This may provide part of the explanation firstly, for the sex differences seen in these conditions, where there is typically a female predominance and secondly, for the symptom exacerbation seen in some disorders during menstruation (e.g. IBS [104]).

CS, or susceptibility to it, also demonstrates heritability; both secondary hyperalgesia and allodynia are heritable with an estimated 50% genetic contribution to the pain variance [105]. This is supported by animal models suggesting a genetic contribution to central sensitisation through hypersensitivity to calcitonin gene-related peptide [106]. These heritability studies are yet to be applied to the FSS.

Is CS the cause or effect of FSS? Thermal and ischaemic stimuli and tolerance to temporal summation, in asymptomatic individuals, in conjunction with measurement of the three COMT haplotypes (incorporating low, average and high pain sensitivities) can predict the subsequent onset of temporomandibular joint dysfunction [101]. This would not only suggest that CS predates the onset of FSS but also that it can be used to predict the onset of FSS, in combination with COMT analysis. COMT polymorphisms therefore may provide a link between CS, the neural network involved in the perception of pain, and the FSS. However, although COMT polymorphisms have been associated with IBS, including specific IBS related bowel patterns [107], evidence is weaker for FM [108] and CFS [109]. Negative findings here may further reflect the heterogeneity of these conditions [110] resulting in a dilution effect, as has been demonstrated in CFS through latent class analysis and genetic evaluation of monoamine oxidase and HPA-related genes [89].

Comorbid mood disorders are more likely in those with FSS compared to those without [111]. Susceptibility to major depressive disorder (MDD) and anxiety disorders is a known risk marker for the development of FSS [112]. The difficulty here is that a previous mood disorder may represent the risk of developing a FSS when accompanied by a comorbid mood disorder, rather than being specific for a FSS alone [113]. The risk might be related to abnormal central processing leading to both mood disorders and FSS. This would seem feasible, with their shared neuroanatomical and neurobiological substrates [114].

Pain perception in MDD is mostly increased [115], [116], but decreased [117] and normal pain thresholds have also been demonstrated [114], [117]. The use of non-standardised protocols may have made the results incomparable. A more recent study [118] using standardised QST in depressive disorder without painful symptoms, as compared with FM, found a reduced threshold to cold and increased temporal summation in depressive disorder, suggesting that CS is a factor in pain perception in depression. The study sample included many subtypes of ‘depression’, including adjustment disorder, and the presence of enhanced temporal summation alone to determine CS is debated [40].

Chronic pain is a common symptom in post-traumatic stress disorder (PTSD) [119], with sufferers often also having FM [120]. PTSD and chronic pain may share cognitive and emotional risk markers. One such example is ‘anxiety sensitivity’ [121], which is thought to amplify both the intensity of emotional reactions and sensitivity to pain [121]. PTSD patients paradoxically appear hyper-responsive to suprathreshold noxious stimuli (i.e. they rated them as being more intense than did healthy controls or those with other anxiety disorders) but were hyposensitive to noxious stimuli with higher thresholds for rating these stimuli as painful [119]. These authors suggest that hyper-responsiveness may be mediated by anxiety and hyposensitivity mediated by dissociation, sometimes found in PTSD [122]. Temporal summation was normal, leading the authors to conclude that altered pain sensitivity in PTSD cannot be simply due to CS [123], although this conclusion is provisional being based solely on temporal summation. The experience of peritraumatic pain and the early provision of adequate analgesia are important factors in determining PTSD [124], [125]; these factors may need to be separated in the examination of central pain processing in PTSD.

Subjective reports of sleep difficulties are common in all FSS [126], [127], [128], even when other causes of poor sleep such as comorbid depression and anxiety are controlled for [129]. In particular, sleep is described as being non-refreshing, with consequent fatigue, daytime somnolence, irritability, word finding difficulties, poor short-term memory in addition to physical symptoms [130], including effects on bowel function [131] as well as being strongly associated with chronic pain, as discussed below. Non-restorative sleep (NRS) implies self-reported restlessness or poor quality sleep [132]. Studies looking at different components of insomnia, such as NRS, difficulty initiating (early insomnia) or maintaining sleep (middle insomnia) have suggested that NRS is more likely to be coupled with daytime problems than early or middle insomnia [133].

Sleep studies suggest that increased alpha activity (light sleep) during slow wave (deep) sleep (SWS) induces NRS, commonly referred to as alpha–delta sleep [134], which has been demonstrated in FSS [135], [136]. These alpha waves are conceived of as ‘mini arousals’ that are responsible for the consequences of NRS. The subjective report of poor sleep correlates with unexpected alpha activity [136].

Experimental disruption of SWS has also been shown to induce pain and fatigue in healthy volunteers [137], [138], [139] and this also increases pain, depressive symptoms and disability in FSS [140]. Experimental sleep deprivation has been used to reduce pain thresholds, with these only returning to baseline with a restoration of SWS [141]. SWS disruption also interferes with the normal CNS inhibitory responses to painful nociception in healthy subjects [142], resulting in a hypersensitive state resembling CS [143].

During sleep, the brain does not discriminate between sensory inputs, which require consciousness to translate them into an emotional and physiological experience, but it remains ‘aware’ of and reacts to threatening stimuli [144]. This allows a ‘threat detection’ system to remain active, but prevents awakening due to irrelevant stimuli. That is to say that highly salient stimuli, including pain [144], are still able to promote arousal. Arousals are detectable as transient rises in autonomic nervous system (ANS) activity, already demonstrated in some FSS [110], [145]. These are able to activate low intensity responses, which typically remain unconscious, but may result in motor restlessness. CS results in a multi-sensory hyper-responsivity in the neural network, responsible for multimodal threat detection. During sleep, the CNS is arousable by threatening, highly salient stimuli that promote wakefulness. This threat detection system is less discriminate in CS and so mini-arousals may be more frequent, resulting in poorer sleep, more symptoms and greater disability [140]. The connection here may be substance P, which is raised in these disorders [146], [147]. Its role in pain is well known, however animal models have also implicated its involvement in sleep disturbance [148]. This suggests that NRS may be a consequence of CS, rather than a factor that maintains it.

Disturbed sleep/wake functions are closely associated with dysregulated circadian rhythms, such as those of the tubero-infundibular and HPA axes. Growth Hormone (GH) is released at the onset of SWS; waking halts its secretion and part of its function is cell growth and repair. This not only ties GH to abnormalities of sleep but also, through disruption of normal healing and repair, to enhanced pain, muscle fatigue and the daytime consequences of NRS [133]. Reduced GH secretion has been demonstrated in FM [149], with one study suggesting that GH is an effective treatment [150]. Thus, through a centrally sensitised disruption of SWS, dysregulation of neuroendocrine function is also implicated in CS.

HPA axis involvement has been implicated in sleep abnormalities [151] and at a genetic level in FSS [89], whilst hypocortisolism resulting from childhood adversity may increase vulnerability to FSS [87], [152]. Cortisol has pain-inhibiting effects, whilst corticotrophin releasing factor (CRF) also leads to the release of β-endorphin, explaining in part the role of this hormone and releasing factor in pain sensitivity [153], [154]. However, reports of HPA axis dysfunction vary between FSS, with a recent meta-analysis reporting hypocortisolism in CFS and in women only with fibromyalgia but normal cortisol levels in IBS [155]. This may reflect either methodological heterogeneity or the heterogeneity of FSS. However, the HPA axis is closely regulated through negative feedback mechanisms and the neuroendocrine state in the chronic phase of illness does not necessarily reflect its status at onset. Some have suggested that there exists a greater rise in cortisol response to acute stress in IBS, on a background of normal or low baseline cortisol, and reduced or raised adrenocorticotrophic hormone (ACTH), whilst in CFS and FM, delays in the decline in cortisol after the experience of stress have been reported on a background of reduced ACTH [156]. This might infer abnormalities in corticotropin releasing hormone (CRH) release — relevant to exacerbations in pain [153], [154] in response to stress in FSS but also to other symptoms such as those affecting the bowel, where intestinal permeability increases in response to CRH release [157].

Dysregulated tubero-infundibular and HPA axes may therefore be associated with CS in FSS through disruption of sleep and enhanced pain in addition to other symptoms, including bowel dysfunction. Although the direction of HPA axis abnormalities and their location in the feedback loop are yet to be adequately determined.

Autonomic dysregulation (underactivity in the parasympathetic nervous system and/or sympathetic overactivity) has been implicated in the FSS by demonstration of reduced heart rate variability during sleep and when awake [158], [159], [160] and may contribute to symptoms such as sleep disturbance and fatigue [161]. The autonomic nervous system (ANS) may be associated with abnormal stress responses and contribute to pain. This may be either an initiating or maintaining factor in the central augmentation of the neural network responding to pain, notably the insula [162], in centrally sensitised individuals.

Fatigue is a ubiquitous symptom in FSS. As a chronic symptom, it resembles pain in so far as that it is not a unitary phenomenon but likely reflects the involvement of multidimensional processes, irrespective of the FSS in which it arises. Fatigue is associated with disruptions of SWS [138], ANS dysfunction [163], neuroendocrine dysregulation [164], including genetic polymorphisms [89], as well as with CS [14], [165]. Evidence to date increasingly suggests that as a symptom of FSS it is associated with CS [14].

The component parts of CS have been demonstrated in the FSS (Table 2) and suggest that CS may represent an underlying vulnerable pathophysiology for FSS in general. By way of its effects on sleep, neuroendocrine axes and the ANS, CS may provide the physiological basis for symptoms other than pain in FSS. Furthermore, although further work is required in those that recover from FSS, CS in FSS appears to be a constant state. This is important, as CS has been demonstrated in other disorders, the pathophysiology of which has been more clearly defined. Two prominent examples are migraine and osteoarthritis [166], [167]. As such, CS would not seem to be sufficiently specific to FSS. However, in both migraine and osteoarthritis, CS seems to be an induced and as such a temporary state. In migraine, 79% of sufferers have been reported to demonstrate allodynia, predominantly ipsilateral to the migrainous head pain during an attack, with no sufferers demonstrating allodynia between attacks [166]. As such, CS in migraine seems dependent upon the presence of active pathophysiology. In osteoarthritis, CS appears to return to a non-centrally sensitised state after successful surgical replacement of the affected joint and elimination of the osteoarthritis pain [167]. This would suggest that CS in osteoarthritis is an effect rather than a cause. Currently it is therefore unclear as to whether an ‘over-processed’ input in the periphery is required for a centrally sensitised state to prevail. Data from osteoarthritis patients [167] might suggest that this is essential and that CS can be reversed with the removal of this input. However, in FSS it is potentially possible that normal pathways exist to the level of the brainstem (by way of example) and that it is the supraspinal networks that are dysfunctional, without the need for an over-processed peripheral input. In turn, this might allow for a concept of central augmentation of higher brain centres that is autonomous and independent of peripheral input — a central rather than peripheral over-processing. Further work is required in this area in order to elucidate the final contribution of supra versus subtentorial components to the experience of pain.

Therefore, in approaching the FSS as disorders united by CS with different phenotypes determined by triggering environmental exposures, CS may represent a feasible endophenotype for these syndromes (Table 3) [14], [168], [169], [170]. In this respect, further support exists in that CS is manifested in both those with and without FSS, being more common in unaffected family members of those with FSS [171]. This would also explain why the search for biomarkers specific to the individual syndromes has been unfruitful — vulnerability to the development of FSS may be marked by CS, whereas precipitating events such as environmental exposures might thereafter mark the development of specific syndromes or their sub-phenotypes. In short, this endophenotype might suggest that we should lump first, then split afterwards.

Future work will need to adopt an integrated, biopsychosocial approach and focus on both further case control and prospective studies of CS, measured in different modalities, in different FSS, including family and twin-studies, and particularly test the persistence of CS during remission of these disorders, in order to see whether this endophenotype is state or trait related. Prospective studies might involve seeking associations with both childhood adversity, the relationship with physiological stress responses and exposure to environmental precipitating factors, regressed against the development of CS and subsequent evolution of FSS. Given the variability of results from genetic studies to date, it may be that polymorphisms are associated with a susceptibility to the development of CS, to the development of an individual FSS, or to both.

Standardised QST protocols will be needed with appropriate control for the multitude of psychophysical variables that are present in the FSS, which are known to independently affect pain perception (e.g. mood and sleep), and these will need to be combined with robust functional neuroimaging studies in order to fully define the neurophysiology of these syndromes and the relative contributions of supra- and sub-tentorial components to the final experience of pain.

The gradual emergence of neural correlates and biological mechanisms for functional somatic syndromes has the potential to alter not only our understanding but also the nomenclature by which we define our understanding. Terms such as psychosomatic, ‘medically unexplained’, somatisation and somatoform seem to beg more questions than answers, and have lost their clinical and scientific utility [172], resulting in significant recent debate surrounding new classifications [5], [173]. Perhaps more importantly, central sensitisation provides healthcare professionals with a model of understanding that can be readily shared with and understood by patients, which moves us on from the sterile and sometimes pejorative language of the past. It has the ability to define not only a better understanding of these most difficult and complex disorders but also, through a clearer understanding of their pathophysiological basis, how they might be better treated. In short, it may help to render the medically unexplained, explicable.