In this study, two group comparisons were conducted using structural, DWI and functional MRI data: Firstly, we compared FMS patients to healthy controls, secondly, we divided the FMS group into two subgroups with and without PNS pathology (PNS and noPNS groups) and compared these subgroups with each other. In the structural analysis, we separately compared the two subgroups with the healthy control group.
We show that in FMS 1) cortical thickness is decreased in the left pars opercularis of the inferior frontal gyrus, 2) FA is generally increased in corticospinal tracts and regions of the limbic system and 3) functional connectivity is reduced between the right midfrontal gyrus and the posterior cerebellum as well as the right crus cerebelli.
Importantly, we demonstrate that structural and functional brain changes are more prominent in those patients with objective evidence of small fiber impairment. 1) A detailed subgroup analysis of cortical thickness revealed two clusters with reduced thickness in the left frontal cortex in the PNS group compared to the control group, but no differences in the subgroup of patients without reduced skin innervation (noPNS subgroup). 2) The PNS subgroup, compared to the noPNS subgroup, showed hypoconnectivities between several brain regions that process and evaluate sensory information.
4.1 Comparison of the present findings with published data
Our cortical thickness data are in line with the results of a recent meta-analysis on voxel-based morphometry studies, which compared FMS patients to healthy controls, regarding the left rostral middle frontal cortex 5, and it has also been shown that the left superior frontal cortex of FMS patients decreases in thickness with increasing duration of the disease 36. The literature concerning the left pars opercularis in FMS is not congruent. In our study, this part of the inferior frontal gyrus shows a decrease in cortical thickness in the FMS group compared to controls. One previous study showed a correlation of pars opercularis cortical volume loss with duration of disease 36, and a SPECT study showed symptom improvement and stronger activation of the pars opercularis in FMS patients after therapy with hyperbaric oxygen 37.
Other studies that have evaluated the relationship between clinical parameters and structural measurements of individual pain processing brain regions showed correlations in areas of the prefrontal cortex, the anterior cingulum or the cerebellum 38,39. In our study the severity of symptoms, such as disability caused by pain or depressiveness, also correlated negatively with the cortex thickness of regions in the left prefrontal cortex (left medial/lateral orbitofrontal cortex and frontal pole). However, these regions of the correlation analysis do not exactly correspond to the clusters where decreased volume was found in the group comparisons (pars opercularis and the left rostral middle/ superior frontal cortices). Thus, because the cortex areas correlating with depression or pain are not the same as those with reduced cortex thickness in our group comparisons, the hypothesis that the CNS changes are a consequence of symptom severity is not supported by this finding. The prefrontal cortex is a known site of pain modulation. Indeed, a dual role has been described including antinociceptive effects by modulating sensory afferent influx, as well as the furthering of chronic pain via corticostriatal projections. Interestingly, decline of prefrontal cortex volume in chronic pain can be reversed with successful biopsychosocial therapy, be it cognitive behavioral therapy, exercise or transcranial magnetic stimulation 40.
Regarding FA, a marker for the integrity of the white matter, we found an increase in FA in the corona radiata and regions of the limbic system (e.g. fornix and cingulate cortex) in the FMS group compared to controls. Some DTI studies with a smaller number of FMS patients found, consistent with our outcomes, increased FA in right and left ACC 39 or in the anterior thalamic radiation, anterior limb of internal capsule up to the putamen 41. Other studies found a decrease in FA in FMS, especially in the corpus callosum 7. It is assumed that the increase in FA is related to an increased number of oligodendrocytes, the myelin-producing cells of the CNS 42,43. There is evidence that these play a role in the modulation of pain mechanisms by producing and modulating chemokines and cytokines and interacting with microglia 44,45. A PET study in FMS patients found increased microglial activity in the medial and lateral areas of the temporal and frontal lobes overlapping with our areas of FA increase 46. Our results might therefore indicate an inflammatory process in the white matter in FMS patients, particularly in those without peripheral nerve pathology. Further studies combining FA, PET, and possibly CSF measures of inflammation are needed to confirm this. It has also been shown specifically in FMS patients that white matter pathways, whose FA increases after a prolonged period of increased activity 47, in this case in pain processing regions, decrease again after pain chronification and show lower values than healthy controls 41.
Regarding the functional connectivity, the posterior cingulate cortex, as the central hub of the DMN, has been identified in several FMS studies as the area for processing pain catastrophizing 48,49. Thalamic hyperconnectivity to parts of the DMN, e.g. the posterior cingulate cortex or the precuneus cortex, was shown after application of acute pain stimuli in FMS patients 50. It has been hypothesized that these enhanced connections are associated with increased activity of the hippocampus, which results in increased alertness and anxiety of patients in response to pain 51. Other preliminary work on connectivity analysis also finds changes in structural connectivity analysis (DTI) in regions such as the parahippocampal gyrus or the supramarginal gyrus, which showed changes in our fMRI connectivity analysis 52. Similar findings to those in our present study have been published using other methods to quantity connectivity measurements, such as electroencephalography and magnetoencephalography 53,54. In our subgroup comparison, we showed greater differences in functional connectivity between PNS and noPNS patients than between patients and controls, however, the question of cause and effect cannot be answered by our data. Intriguingly, a recent study, using DTI, found reduced connectivity between the thalamus and pain-related areas in patients with reduced skin innervation due to small fiber neuropathy 55, in contrast to the hyperconnectivity found in our PNS subgroup. Thus, the relation of small fiber pathology and CNS connectivity seems to be disease specific.
4.2 Are the findings specific for FMS?
Some of the findings we report here have also been observed in other chronic pain disorders 56. Changes of connectivity in the DMN were also found in patients with chronic back pain, complex regional pain syndrome and knee osteoarthritis 57. It has already been suggested that a lower activity of the prefrontal cortex, a well-known pain modulation area, could lead to a failure in the elimination of subcortically driven fear behaviors, thereby resulting in pain chronification 58. It is currently unclear whether these processes are adaptive, maladaptive or cause some of the symptoms. In order to better understand the pathophysiology of FMS, it is therefore important to first understand the role of brain neuroplasticity in chronic pain, as a brain signature of pain appears to be found across various pain syndromes 59.
4.3 Limitations of our study
Our study has some limitations. Because the PNS subgroup with damage to the small nerve fibers also had more severe neuropathic pain and fibromyalgia-specific disease symptoms compared with the noPNS subgroup, it is not clearly distinguishable whether the subgroup effects were driven by disease severity or neuropathic changes in the small nerve fibers.
The healthy controls in our study did not receive a skin biopsy, so we cannot rule out that some persons with reduced IENFD might have been in this group. However, in our previous study 4, only 2% of normal controls had reduced IENFD at the lower and upper leg, so that it is highly unlikely that a large number of our present controls would have had this finding.