The American College of Rheumatology criteria for FM require that an individual has a certain degree of tenderness. A tender point count is performed by applying 4 kg pressure manually to 18 predefined tender points, and then asking the patient whether these areas are tender. A positive response is considered a tender point; if an individual has 11 tender points or more, this element of the case definition is satisfied.

The apparent close link between tenderness and FM has been well studied in both clinical trials of new therapies and in mechanistic studies. In a number of longitudinal randomized, placebo-controlled trials, improvements in clinical pain have corresponded with a significant change in tender point counts or in the tender point index 121314. In contrast, other studies did not show a correspondence between improvements in clinical pain and tender point counts 151617181920.

The discrepancies between studies could either be because the therapies did not improve tenderness or because tender points are not a good measure of tenderness. Both factors are likely to play a role since, in certain studies where multiple measures of the pain threshold were used, tender point counts did not significantly improve whereas other measures did 2122. Moreover, other studies have shown that tender points are not a pure measure of tenderness. For example, there is a strong correlation between tender point counts and measures of distress in population-based studies 23. Tender points have also been demonstrated to be biased by cognitive and emotional aspects of pain perception, whereas other measures of tenderness are much less so (see below) 24. Improvements in tender point counts in some previous FM trials therefore possibly occurred because of improvements in distress, rather than because of inherent improvements in pressure pain threshold. Finally, tender points are often not continuously distributed in samples; rather, most people have either very few or nearly 18 tender points. As such, many investigators do not feel that tender point counts are useful to assess tenderness, and have instead turned to psychophysically and statistically superior measures.

Directly measuring pressure pain thresholds is an alternative method of documenting tenderness. Devices that measure pressure pain thresholds have been used to demonstrate a left-shift and lowered pressure pain thresholds in patients with FM compared with control individuals, and this finding is noted anywhere in the body, both at tender points and in areas previously considered control points (Table 2). These findings suggest to many investigators that the term control points should be abandoned, or replaced by a term such as high-threshold tender point, since FM patients are just as tender in these regions relative to healthy control individuals.

Many of these studies initially used commercial devices or dolorimeters to deliver continuously increasing pressure via blunt probes. These measures were found to be sensitive to psychophysical and psychological biases, however, slightly similar to tender point counts using digital palpation (reviewed in 25). For instance, the rate of increase of stimulus pressure, controlled by the operator, and patient distress were both shown to influence the pain threshold 2426. To minimize the bias, more sophisticated paradigms using random delivery of pressures have been developed and investigated 2728 (Table 3). Random delivery may be less sensitive to certain influences, but it is not free of bias. For instance, in a study by Petzke and colleagues, FM patients reported higher pain during random delivery than during ascending – possibly due to a perceived lack of control 28.

A recent longitudinal study compared the three different evoked measures – tender point counts, the dolorimeter (ascending pressure paradigm), and the multiple random staircase (random pressure paradigm) – with clinical reports of pain improvement 21. Although both clinical pain measures improved during the course of the study involving acupuncture, only one of the evoked measures – the multiple random staircase measure, which presented stimuli to individuals in an unpredictable fashion – improved after treatment. These results suggest that, of the different methods, the random stimuli paradigm may be more likely to systematically change over time. Interpretation of the results is nonetheless limited and will need to be reproduced and examined using other treatment modalities.

In addition to the heightened sensitivity to pressure noted in FM, other types of painful stimuli also are judged more painful by these patients. A decreased heat pain threshold in FM patients as compared with control individuals has been shown by multiple groups 282930 (Table 4). A reduced cold pain threshold has been reported by one group in two different studies 3031. Sensitivity to warmth and the ability to detect electrical stimuli do not appear to be discriminative measures at this time.

In the process of understanding altered evoked pain sensitivity present in FM, evaluation of the intrinsic analgesic systems has uncovered another potential biomarker: diminished diffuse noxious inhibitory control (DNIC). DNIC testing in both animals and humans involves testing the pain threshold at baseline, and then administering an acutely painful stimulus that leads to a systemic analgesic effect, presumably by activating endogenous analgesic systems.

Several studies by different groups, using different conditioning stimuli (the acute noxious stimulus) and test stimuli (the stimulus used to measure pain threshold at baseline and following the acute, noxious stimulus), have indicated a deficiency of DNIC in individuals with FM. Diminished DNIC was observed in four cross-sectional studies by different groups that used variable test and conditioning stimuli 31323334 (Table 5). Diminished DNIC has also been noted in other types of chronic pain; that is, temporomandibular disorder and hip osteoarthritis 3536. The normalization of DNIC after hip osteoarthritis surgery suggests it may be an objective measure of chronic pain that can change over time with treatment 36.

Functional neural imaging enables investigators to visualize how the brain processes the sensory experience of pain. The primary modes of functional imaging that have been used in FM include functional magnetic resonance imaging (fMRI), single-photon emission computed tomography (SPECT), and positron emission tomography.

fMRI studies evaluating pain processing have the strongest current evidence of the functional imaging studies, because they corroborate this left-shift in stimulus–response function (that is, hyperalgesia/allodynia) noted in FM. Specifically, several areas of the brain consistently show greater activation in FM patients than in control individuals given the same objective stimulus intensity – especially the secondary somatosensory cortex, insula and the anterior cingulate cortex. These findings have been noted in five cross-sectional studies by two different groups, using both pressure and heat stimuli 3738 (Table 6). In the study by Giesecke and colleagues, the clinical pain intensity corresponded with an increase in the evoked regional cerebral blood flow 37. The resting regional cerebral blood flow was evaluated by a third group in a longitudinal study using fMRI, and showed change after drug treatment 39. These studies have also been useful in identifying differences in pain processing in individuals with and without psychological comorbidities, showing for example that depression does not seem to be influencing the magnitude of neuronal activation in sensory pain regions such as the secondary somatosensory cortex, whereas cognitive factors such as catastrophizing did influence the sensory intensity of pain 3740.

Positron emission tomography imaging in FM has been reported in only a few studies with inconclusive results. The only positive study is a recent one showing there may be altered dopaminergic activity in FM 41.

SPECT imaging has been studied in four cross-sectional studies by different groups that consistently found reduced regional cerebral blood flow in the right thalamus of patients with FM (three of the four studies) 42434445. No correlation between symptoms and findings were noted in the SPECT studies.

The consistent abnormalities seen in fMRI and SPECT studies suggest either of these methods might be useful to use as a biomarker, but longitudinal studies showing that improvements in symptoms coincide with normalization of functional imaging findings would be necessary to establish this role. The advantages of fMRI imaging over positron emission tomography and SPECT include the less invasive nature and the higher temporal and spatial resolutions of fMRI. Disadvantages of fMRI include the cost and practicability as well as the inability to perform receptor–ligand studies that are possible with positron emission tomography and SPECT.

Cerebral potentials evoked by noninvasive stimulation provide a unique opportunity to investigate the functional integrity and magnitude of brain processing pathways. Expressing the ability of the human brain to discriminate, classify, and memorize the significance of exogenous stimuli, event-related potentials (ERPs) have been used as a marker of cognitive function in patients with psychiatric and neurological disorders. The electrical waveforms generated can be divided into late and early components, and the waveforms are designated by their polarity (P-positive, N-negative) and latency (timing of peak) after stimulus onset. Additionally, the amplitude – the size of the voltage difference between the component peak and a prestimulus baseline – is also quantified. Auditory, somatosensory, and visual ERPs have been evaluated in patients with FM in a few studies.

Among the ERPs evaluated to date, the P300 potential (most commonly generated by an auditory consciously attended stimuli) appears to be the most promising to differentiate FM patients from control individuals. The P300 wave is a late cortical neuropsychological event, the latency of which reflects the information processing speed and the amplitude of which expresses memory functions. A reduced P300 amplitude during an auditory discriminated-task paradigm has been significantly noted in FM patients as compared with control individuals in three cross-sectional studies by two different groups 464748 (Table 7). All three studies also evaluated the P300 latency, but only the largest study by Alanoglu and colleagues noted an increase in P300 latency, a finding that may have not been found in the prior studies due to lack of power 46. In the one of these three studies by Ozgocmen and colleagues that performed ERPs before and after treatment, 8 weeks of sertraline treatment led to an increase in the P300 magnitude 48.

These studies generally failed to show an association between the ERP findings and symptom severity, although there was an association noted with the total myalgic score. Although the change in the P300 potential after sertraline treatment was attractive, the authors agreed that – given the corresponding significant clinical improvement in pain, fatigue, or depression – the mechanism for the change remained unclear, and they acknowledged it may represent regression to the mean. Larger studies by different groups with an attention to standardizing methods are essential prior to mainstream use of this marker.

In contrast to auditory potentials, there are few and varied studies evaluating somatosensory and visual ERPs. The assorted protocols used in the studies investigating somato-sensory and visual ERPs may have contributed to the lack of consistently demonstrated differences in FM and normal individuals. The lack of an established standardized methodology makes direct comparison difficult and may limit the evidence of reproducibility.

A method of motion assessment that infers sleep and wakefulness from the presence of limb movements, actigraphy is increasingly being used as a surrogate marker for both sleep and activity. The actigraph typically combines a movement detector and memory storage on a watch-like device. The device can be worn on the wrist or the ankle continuously for long periods of time. Sleep-pattern measures available via actigraphy analyses include sleep latency, the wake time after sleep onset, and the total sleep time; sleep architecture cannot be measured, as with polysomnography. Compared with polysomnography, however, actigraphy is less expensive, less invasive, and more conducive to repeated measures, resulting in extensive use in intervention studies 54.

Actigraphy is being increasingly used in FM studies and appears promising, but has not yet proven adequately sensitive to stand alone in clinical evaluation or treatment trials 505556. As a measure of sleep quality there have been inconsistent actigraphy results, with one group noting increased levels of activity at night in FM (also noted in patients with major depression) 55 and another group noting no difference 50. Edinger and colleagues used actigraphy as an outcome measure in an intervention trial comparing cognitive behavior therapy intervention with sleep hygiene and usual care in the treatment of insomnia 57. Deriving an actigraphic improvement criterion, the investigators showed a greater number of patients receiving cognitive behavior therapy had clinically significant improvement in the total wake time compared with sleep hygiene therapy. No statistical difference between cognitive behavior therapy and usual care was able to be demonstrated, even though a statistical difference between the groups was shown using sleep log data in the same study.

As an objective measure of functional status, actigraphy might hold more promise as a surrogate outcome measure, because it allows the direct recording of activity levels, rather than relying on patient self-report 58. Kop and colleagues demonstrated that although patients with FM have 36-Item Short Form health survey scores nearly two standard deviations below the population average, they have the same average activity level as a group of sedentary control individuals 58. The FM patients had much lower peak activity levels, however, suggesting that the problems in function that FM patients report might be more due to an inability to rise to the intermittent demands of day-to-day life than due to overall reduced function.

In basal and diurnal cortisol studies, the most consistently found measure is a flattened diurnal plasma cortisol level with an elevated trough, found in three of four cross-sectional studies by two out of three groups 606162 (Table 8). Studies evaluating basal plasma cortisol levels, salivary basal and diurnal cortisol levels, and urinary cortisol levels have shown inconsistent results, but they generally demonstrate normal to reduced basal levels. Since atypical depression can show a reduced cortisol level, biopsychological factors that influence cortisol levels may be contributing to the inconsistent results currently found in the literature 63. These factors need to be better elucidated and accounted for in future studies. Nonetheless, a flattened diurnal cortisol level is a promising objective measure.

Evaluation of other components of the hypothalamic–pituitary–adrenal axis has been relatively unrevealing. Basal and diurnal adrenocorticotropic hormone shows no difference in FM patients versus healthy control individuals 626465 (Additional file 1). Provocative hypothalamic–pituitary–adrenal studies utilizing the cosyntropin test have shown inconsistent results 62666768 (Additional file 2).

Results of the dexamethasone suppression test have been reported in a number of studies by different groups, and the results reveal normal to high levels of cortisol following infusion of the corticosteroid 6064666970 (Additional file 3). Depression also typically follows a pattern of resistance to the dexamethasone test, and therefore is a confounding factor in a large number of these evaluations.

Studies have also been completed to assess the cortisol response to exogenous corticotropin-releasing hormone or endogenous activators of corticotropin-releasing hormone (that is, hypoglycemia, IL-6) in FM. Investigators found normal to reduced cortisol levels in patients with FM after an increase in corticotropin-releasing hormone, but these results were not reproduced in other similar studies. Further investigation taking into account psychological factors as well as doses of different drugs will be prudent.

Tilt table testing and heart rate variability have been evaluated in patients with FM. The consistent and reproducible finding of lower heart rate variability in FM patients compared with control individuals (in three cross-sectional studies by two different groups) makes it a more useful measure than tilt table testing 717273. An abnormal drop in blood pressure or an excessive rate of syncope during tilt table testing has been noted in two out of three cross-sectional studies completed by three different groups 747576. One study noted no difference in normal individuals and control individuals using univariate analysis 76. Moreover, recent findings also suggest that aberrations in heart rate variability may predispose to fibromyalgia symptoms 7778, possibly identifying patients at risk.

FM syndrome is more prevalent in women than in men, suggesting a role of sex hormones in the pathophysiology of FM 79. To date, two studies have failed to show an association between sex hormones and pain sensitivity 7980. The reason for a female predominance in FM is complex and warrants further investigation.

The search for representative autoantibodies is a predictable step for a disease like FM, often evaluated by rheuma-tologists and coexisting with autoimmune diseases. Antiserotonin antibody, antiganglioside antibody, and antiphospholipid antibody have been shown to be different in FM patients and control individuals, but the applicability of these findings is not yet clear 81 (Table 9). Antiserotonin antibody has been shown to be increased in FM in three cross-sectional studies by two different groups 818283. Antiganglioside antibody and antiphospholipid antibody have each been shown to be increased in FM in two cross-sectional studies by the same group 8182. A different group evaluating antiganglioside antibody in a third cross-sectional study was unable to reproduce the results 83. Antithromboplastin antibody 83, antipolymer antibody 84, and anti-68/48 kDa and anti-45 kDa 85 have each been evaluated in one cross-sectional study and have shown increased levels in FM. A review of the literature demonstrates that antinuclear antibodies, antithyroid antibodies, antisilicone antibodies, and antiglutamic acid decarboxylase are not informative in FM.

The nonspecific increase in antibodies to a number of antigens may be a nonspecific finding that arises from a subtle shift in immune function in this spectrum of illness. In the closely related chronic fatigue syndrome, investigators have noted a shift from a T1 to a T2 immune response, which would be expected to lead to increased production of nonspecific antibodies. Any antibody or autoantibody proposed as either a diagnostic test for FM or a biomarker of FM must therefore be carefully tested using various control individuals to ensure its authenticity.

Substance P is a neuropeptide released in spinal fluid when axons are stimulated. Four different cross-sectional studies by various groups in FM patients noted an elevation of substance P in cerebrospinal fluid 86878889. In contrast, a normal substance P level has been noted in the cerebrospinal fluid of patients with chronic fatigue syndrome 90. Although these results appear promising, elevated substance P is not specific for FM but rather has been shown to occur in other pain states such as chronic, daily headaches and chronic neck or shoulder pain associated with whiplash injury 9192. A high level of substance P therefore seems to be a biological marker of the presence of chronic pain.

Nerve growth factor and calcitonin gene-related peptide are additional neuropeptides that have been evaluated in FM. Nerve growth factor was shown in one study to have increased levels in FM and not in FM/rheumatoid arthritis overlap, therefore presenting inconclusive results 93. Cerebrospinal fluid and serum calcitonin gene-related peptide have been studied and not found to be different in FM patients and control individuals 9495.

The amino acid tryptophan and the cytokine IL-8 have both been shown to be different in patients compared with control individuals in a couple of studies, but neither have been evaluated in longitudinal studies 969798. A low tryptophan level has been found in two of three studies by three different groups 9699100. IL-8 has been consistently demonstrated in three studies by two different groups 9798101. Moreover, IL-8 has been shown to correlate with symptoms of FM and not to be associated with depressed FM 98. Serum IL-6 was evaluated and found to be normal in FM patients 98101.