Autoantibody Correlation Signatures in Fibromyalgia and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Association with Symptom Severity

Recent studies provide some evidence for the contribution of antibody-mediated autoimmune mechanisms to the nature of fibromyalgia (FM) and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). Much attention was paid to the autoantibodies (AAb) targeting G protein-coupled receptors as natural components of the immune system. However, the natural AAb network is much more extensive, and has not been previously investigated in these disorders. The enzyme immunoassays ELI-Viscero-Test and ELI-Neuro-Test were used to determine changes in serum content of 33 natural AAb to neural, organ-specific and non-tissue-specific autoantigens (a) in 11 ME/CFS patients with comorbid FM; (b) in 11 ME/CFS patients without FM; (c) in 11 healthy controls. Individual AAb profiles and their correlation with some clinical symptoms were analyzed. Both patients with ME/CFS(−)FM and ME/CFS(+)FM were characterized by more frequent and pronounced deviations in the immunoreactivity to GABA-receptors than healthy controls. Although the level of other natural AAb did not differ between study groups, AAb correlation signatures were altered in patients compared to healthy controls. Both in patients and healthy controls the level of natural AAb to various neural and tissue-specific antigens correlated with the severity of fatigue, bodily pain, depression, anxiety, physical and mental health-related quality of life. Notably, widely different correlation patterns were observed between study groups. Findings from this pilot study provide some evidence that the homeostasis of autoimmune relationships, which are possibly a physiological part of our immune system, may be altered in FM and ME/CFS. The correlation of disease-induced perturbations in individual AAb profiles with some clinical symptoms may arise from the immune system’s ability to reflect qualitative and quantitative changes in antigenic composition of the body.


Introduction
Fibromyalgia (FM) is a common cause of chronic, widespread, musculoskeletal pain with unclear etiopathogenesis. However, it was demonstrated in a recent breakthrough study that IgG from the patients with FM produced painful sensory hypersensitivities by sensitizing peripheral nociceptive afferents in mice [1]. IgG from patients in this study labeled satellite glial cells and neurons in vivo and in vitro, as well as myelinated fiber tracts and a small number of macrophages and endothelial cells in mouse dorsal root ganglia. Furthermore, FM IgG was bound to human dorsal root ganglia. The same author

Questionnaires for Symptom Scoring
The common ME/CFS and FM symptoms (including depression and anxiety) as well as baseline health status were assessed in both patients and HC using the following instruments: the Short Form 36 Health Survey (SF-36), the Multidimensional Fatigue Inventory (MFI), DePaul Symptom Questionnaire-Short form (DSQ-SF), and Hospital Anxiety and Depression Scale (HADS). The SF-36 includes the following subject-reported evaluations about current health status: physical and social functioning, physical and emotional limitations, vitality, pain, and general and mental health [16]. The MFI comprises a 20-item self-reported questionnaire focused on general, physical and mental fatigue, reduced activity, and reduced motivation [17]. Cognitive symptoms were assessed based on the data from the DSQ-SF questionnaire. In particular, composite score was calculated for the cognitive symptoms by averaging scores for the frequency and severity (ranged from 0 to 4) of two symptoms of cognitive dysfunction from the DSQ-SF questionnaire ("Problems remembering things" and "Difficulty paying attention for a long period of time"). HADS is a reliable scale for identifying and assessing the symptom severity in anxiety disorders and depression, both in patients with somatic diseases and in patients with mental disorders [18]. The score 0-7 for each subscale (depression and anxiety symptoms) indicated "normal results"; 8-10-"borderline results, or doubtful case of anxiety/depression"; 11 or more-"probable case of anxiety/depression".

Autoantibody Quantification-ELISA
We defined individual normalized levels of AAb against 21 organ-specific and nonorgan-specific antigens and 12 neural antigens using standardized ELISA test systems for semi-quantitative serum AAb evaluation (ELI-Viscero-test-24 and ELI-N-test-12 by Medical Research Center "Immunculus", Moscow, Russia). The antigens used in the test systems are listed in Table 1. Table 1. List of antigens, included in the test systems ELI-Viscero-test-24 and ELI-N-test-12.

No
Antigen Abbreviation 1 Double stranded deoxyribonucleic acid dsDNA 2 β2-glycoprotein-I β2-GP 3 Fc-fragments of IgG Fc-IgG 4 Collagen type IV Collagen 5 Membrane antigen of cardiomyocytes CoM 6 β1-adrenergic receptors of cardiomyocytes β1AdrRe 7 Platelet membrane antigen TrM-03 8 Cytoplasmic antigen of neutrophils ANCA 9 Membrane antigen of renal glomerular cells KiM-05 10 Membrane antigen of pulmonary alveolocytes LuM-02 11 Membrane antigen of gastric wall cells GaM-02 12 Membrane antigen of small intestine wall cells ItM-07 13 Membrane antigen of colon wall cells ScM 14 Cytoplasmic antigen of hepatocytes HeS-08  15  Membrane antigen of hepatocyte mitochondria  HMMP  16  Human insulin  Insulin  17  Insulin receptors  Ins-Re  18  Thyroglobulin  TG  19  Thyrotropin receptor  TSH-Re  20  Membrane antigen of adrenal medulla cells  Adr-D\C  21  Membrane antigen of sperm and prostate cells  SPr-06  22  Neurofilament protein 200  NF-200  23 Glial fibrillary acidic protein GFAP 24 S100 protein S100 25 Myelin The pooled control serum was a preparation of polyclonal immunoglobulins of the IgG class, synthesized by B-lymphocytes in response to antigenic stimuli that occurred throughout the life of donors. Immunoglobulins in the control serum were obtained from the blood serum of more than 5000 healthy donors, and brought to a concentration close to physiological (16 mg/mL). Thus, pooled control serum contained population-normalized IgG class polyclonal antibodies to each of the studied antigens, and was used as a universal standard for all tested antigens. Depending on the studied antigen, the pooled control serum was diluted to a final concentration, which was derived on the basis of studies of the level of AAb of a large cohort of individual serum samples from healthy donors. The content of AAb to the studied antigens was evaluated in the conventional units of optical density, and compared to their content in a control pool of sera from healthy donors (taken for 100%). Then, the average individual immunoreactivity (AIR) of the studied samples for each of the antigens in comparison with the pooled control serum was calculated according to the formula: AIR-the average reactivity of an individual patient's serum to all studied antigens, expressed as a percentage of the average reactivity of the pooled control serum with the same antigens. R (ag1, 2, N)-reactivity (in units of optical density) of the patient's serum with studied antigens. R(k1, k2, N)-reactivity (in units of optical density) of the pooled control serum with studied antigens.
The normal (physiological) levels of individual AIR are restricted by the range from −30% to 0% (or conditional units ) of the control sample AIR.
To construct immunoreactivity profiles, the deviation (as a percentage of the individual AIR) of the patient's serum reactivity with each of the antigens was calculated using specialized software according to the formula: The normal (physiological) R(dev)agN for each AAb is restricted by the range from −15% to +10% (or conditional units) from the individual AIR.

Statistical Analysis
Statistical processing was performed with the Statistica 10.0 software package. A chisquared test was applied to compare the prevalence of non-physiological AAb deviations from the individual AIR between patients and HC. The Mann-Whitney U test was applied to compare mean R(dev)agN values between patients and HC. Spearman correlation analysis was performed in each study group in order to compare AAb correlation signatures between the groups. We used Chord diagrams to visualize the patterns of Spearman's rank correlation coefficients between AAb. Spearman correlation analysis was also performed to study the relationship between the severity of different symptoms and R(dev)agN values. Differences were considered significant at p < 0.05.

Subject Characteristics
The study involved 11 patients with ME/CFS+FM, 11 patients with ME/CFS who did not suffer from FM, and 11 healthy controls. Patient characteristics are shown in Table 2. Patients from ME/CFS(+)FM and ME/CFS(−)FM groups did not differ from controls in terms of gender and age. While BMI in ME/CFS(−)FM group was similar to the BMI in HC, it was significantly lower in ME/CFS(+)FM group. Regarding race and ethnicity, all patients and HC were Caucasian and not Hispanic.
The illness duration varied, with a range of 1 to 35 years and median value of 6.0 years in the ME/CFS(+)FM group and with a range of 2.5 to 35 years and median value of 7.0 years in the ME/CFS(+)FM group. Four patients (33%) with ME/CFS and comorbid FM and eight patients (73%) without comorbid FM reported an infection-triggered onset of disease.
All scales from SF-36 and MFI were significantly different between ME/CFS(+)FM and HC, as well as between ME/CFS(−)FM and HC (Table 3). Both physical and mental component scores (PCS and MCS, respectively) derived from the SF-36 short survey were, as expected, higher in the control group (p < 0.05), indicating better health. Table 3. Fatigue, depression, anxiety, and baseline health status assessment in patients and healthy controls. FM-fibromyalgia, HADS-Hospital Anxiety and Depression Scale, HC-heathy controls, IQR-interquartile range, PCS-physical component score; MCS-mental component score, ME/CFS-myalgic encephalomyelitis/chronic fatigue syndrome, MFI-multidimensional Fatigue Inventory. It should be noticed that all patients in our study presented with clinically significant post-exertional malaise (according to the DSQ-SF, they had frequency and severity ratings ≥2 in questions number 2 or 3), which has been considered a key symptom of ME/CFS for over three decades [10][11][12].
Depression and anxiety levels were significantly higher in ME/CFS(−)FM and ME/CFS(+)FM groups. The median HADS-D subclase score indicated probable comorbid depression only in the ME/CFS(+)FM group, and the median HADS-A score corresponded to borderline anxiety in both groups of patients.
A significantly higher proportion of patients in ME/CFS(−)FM and ME/CFS(+)FM groups presented with abnormal levels of AAbs against GABA receptors compared to HC (Table 4). At the same time, the proportion of patients with abnormal peaks of any other AAb in groups of patients did not differ significantly from the controls.  In order to assess the extent of abnormalities in the immunoreactivity profiles of patients, we calculated median absolute deviations of the participants' serum reactivity towards each of the underlying antigens from AIR (median R(dev)agNs) in each of the three study groups, and then compared the obtained values between patients and controls. No significant differences have been observed.
Based on the concept of the antibodiome as a functional and physiological network of AAb, which reflects the exposome and could be disturbed in the disease process [19], we performed a correlation analysis between R(dev)agNs and some clinical characteristics (BMI, age, SF-36 and MFI-20 subscales, depression, anxiety, composite score for cognitive symptom frequency and severity) in each of the three study groups. The significant correlations are shown in Tables 5-7.   To identify changes in the AAb relationships in patients suffering from ME/CFS with and without concomitant FM, we analyzed AAb correlations in the two patient groups separately. A number of correlations between the absolute R(dev)agNs values in both groups were revealed. Disease-specific changes among the studied AAb were identified (Figure 1a,b).  The aforementioned data on the correlation between R(dev)agNs and some symptom scores, even in HC, suggested the presence of physiological relationships among natural AAb. Based on these findings, we expanded our study to include the analysis of HC sera for correlations between the studied AAb ( Figure 1c). As previously shown for a few types of AAb (especially those targeting GPCR) [10], we observed that a number of AAb targeting various organ-specific, neural, and non-organ-specific antigens correlated with each other in HC.
To gain better insights into the antibodiome, we presented the significant AAb correlations for each study cohort in chord diagrams. The relationship between the Aab from three antigen groups (organ-specific, neural, and non-organ-specific ones), the number of AAb correlations in patients and HC, and the loss of normal correlation signatures in the disease are more descriptive in these plots (Figure 2a-f). AAb. Based on these findings, we expanded our study to include the analysis of HC sera for correlations between the studied AAb ( Figure 1c). As previously shown for a few types of AAb (especially those targeting GPCR) [10], we observed that a number of AAb targeting various organ-specific, neural, and non-organ-specific antigens correlated with each other in HC.
To gain better insights into the antibodiome, we presented the significant AAb correlations for each study cohort in chord diagrams. The relationship between the Aab from three antigen groups (organ-specific, neural, and non-organ-specific ones), the number of AAb correlations in patients and HC, and the loss of normal correlation signatures in the disease are more descriptive in these plots (Figure 2a-f).   Table 1 for abbreviations employed), which are grouped according to the target antigen (red/pink = organ-specific, green = neural, blue = non-organ-specific antigens). Chords linking AAb indicate significant correlations (at least p < 0.05) according to Spearman rank correlations, while chord thickness is directly proportional to correlation coefficient.

Discussion
The role of autoimmunity in ME/CFS and FM is widely discussed now, largely due to the emerging data on functional anti-GPCR AAb in ME/CFS [8], and successful passive transfer of FM pain from patients to mice [1]. In this context, it has been emphasized that effort and positive resources should be placed into exploring the role of functional AAb, their trigger factors, and their potential evolutional purpose in conditions characterized mostly by somatic symptoms, rather than objectively identifiable signs [20]. It has been suggested that autonomic nervous system imbalance might be causing the symptoms in these conditions and, in turn, might be triggered by the pathogenic functional AAb [21]. At the same time, it should be kept in mind that the presence of AAb does not imply the presence of an autoimmune condition, as AAb are also recognized in non-autoimmune diseases [22,23]. According to the modern interpretation of the phenomenon of physiological autoimmunity, AAb act as adaptive bioregulators of cell functions and growth along with neurotransmitters and hormones, both in health and disease [24]. Moreover, there is a concept of "Immunculus" or "Immune Homuculus" based on the assumption that the network of physiological autoreactive antiidiotypic AAb may dynamically reflect the whole individual antigenome as a totality of internal immunological images of the autoantigens [25][26][27].
The results obtained in our study suggest that ME/CFS and FM are rather not autoimmune diseases, but conditions with dysregulated natural autoimmunity. In particular, while 54.5% and 45.5% of patients in ME/CFS(+)FM and ME/CFS(−)FM groups presented with abnormal absolute deviations of serum reactivity against GABA Re compared to none in the control group, these changes were not very pronounced (since no significant differences have been observed in median R(dev)agNs between the three groups). Classical autoimmune diseases with pathogenic AAb are characterized with a significant  Table 1 for abbreviations employed), which are grouped according to the target antigen (red/pink = organ-specific, green = neural, blue = non-organ-specific antigens). Chords linking AAb indicate significant correlations (at least p < 0.05) according to Spearman rank correlations, while chord thickness is directly proportional to correlation coefficient.

Discussion
The role of autoimmunity in ME/CFS and FM is widely discussed now, largely due to the emerging data on functional anti-GPCR AAb in ME/CFS [8], and successful passive transfer of FM pain from patients to mice [1]. In this context, it has been emphasized that effort and positive resources should be placed into exploring the role of functional AAb, their trigger factors, and their potential evolutional purpose in conditions characterized mostly by somatic symptoms, rather than objectively identifiable signs [20]. It has been suggested that autonomic nervous system imbalance might be causing the symptoms in these conditions and, in turn, might be triggered by the pathogenic functional AAb [21]. At the same time, it should be kept in mind that the presence of AAb does not imply the presence of an autoimmune condition, as AAb are also recognized in non-autoimmune diseases [22,23]. According to the modern interpretation of the phenomenon of physiological autoimmunity, AAb act as adaptive bioregulators of cell functions and growth along with neurotransmitters and hormones, both in health and disease [24]. Moreover, there is a concept of "Immunculus" or "Immune Homuculus" based on the assumption that the network of physiological autoreactive antiidiotypic AAb may dynamically reflect the whole individual antigenome as a totality of internal immunological images of the autoantigens [25][26][27].
The results obtained in our study suggest that ME/CFS and FM are rather not autoimmune diseases, but conditions with dysregulated natural autoimmunity. In particular, while 54.5% and 45.5% of patients in ME/CFS(+)FM and ME/CFS(−)FM groups presented with abnormal absolute deviations of serum reactivity against GABA Re compared to none in the control group, these changes were not very pronounced (since no significant differences have been observed in median R(dev)agNs between the three groups). Classical autoimmune diseases with pathogenic AAb are characterized with a significant increase in the level of the AAb. While immunological and inflammatory changes have been re-peatedly reported in many ME/CFS patients, these changes tend to be most pronounced during the first 3 years of the disease [28]. However, according to the longitudinal studies of varying duration, less than 10% of patients with ME/CFS fully recover [3]. These findings support the hypothesis that not pathogenic AAb, but dysregulated natural autoimmunity is of crucial importance in the pathogenesis of ME/CFS. According to the concept of physiological autoimmunity, quantitative changes in the content of natural AAb are related to variations of expression and secretion of the relative antigen, reflecting functional state of the corresponding cell type [19,27]. GABA Re is an element of endogenous stress-regulating mechanisms preventing distress [29]. The last factor has long been considered as increasing the risk of ME/CFS [30]. An imbalance between excitatory and inhibitory neurotransmission has been linked to ME/CFS and FM [31]. Interestingly, pregabalin, which is one of the three FDA-approved drugs for the treatment of FM, is a lipophilic analogue of GABA. At the same time, it neither acts like GABA nor binds to GABA receptors, but binds strongly to the auxiliary alpha-2 delta subunit of the presynaptic voltage-gated calcium channel receptor to reduce the activation of postsynaptic neurotransmitter release [31]. Regarding some potential substances for future clinical research in ME/CFS and FM, it should be mentioned that a number of complementary dietary supplements have been reported to rebalance glutamate:GABA neurotransmission, namely Omega-3 PUFAs, CoQ10, Withania Somnifera (Ashwagandha, Indian Ginseng), N-acetylcysteine, vitamin B12, curcumin (contained in turmeric), zinc, magnesium, 2-aminoethanesulfonic acid (L-Taurine), and carnitine (L-Carnitine) [32].
The majority of AAb, for which significant correlations with the symptom scores were identified in HC, target neural antigens, which was expected as the analyzed symptoms are neuropsychological. This association was disturbed in both ME/CFS(+)FM and ME/CFS(−)FM groups-the number of AAb targeting internal organ-specific and nontissue specific antigens correlated with the symptoms scores in these groups. Thus, our findings allow us to assume that ME/CFS and FM are organic multisystem diseases, rather than psychological disorders.
The pattern of correlations found in ME/CFS(−)FM (according to the manufacturer of the ELI-tests) suggests the roles of the gut microbiome, disturbed detoxification mechanisms (namely, liver and kidney functioning), and an inflammatory process in the pelvic organs in the development of symptoms. At the same time, the mechanisms underlying the observed correlations remain largely unclear and can differ between patients and HC. For example, anti-GABA-Re AAb were associated with a lower mental component score, i.e., worse self-perceived mental health in HC, while in patients from the ME/CFS(−)FM group, these AAb were inversely correlated with depression.
Changes in immunoreactivity to β-endorphin were associated with more pronounced fatigue in both ME/CFS(−)FM and ME/CFS(+)FM groups, but not in HC. These findings suggest the dysfunction of the endogenous opioid system in ME/CFS and FM. Interestingly, a significant factor that differentiates β-endorphin from other endogenous opiates is its high affinity for and lasting effect on µ-opioid receptors [33]. When comparing patient groups and HC, with regard to the associations of bodily pain with the studied AAb, it can be noted that patients from both groups abrogated the normal interconnection between anti-µ-Opioid-Re AAb and pain intensity. It was shown by Schrepf et al., in the study employing PET and fMRI together, that dysregulation of the endogenous opiate system in FM could lead to less excitation in antinociceptive brain regions by incoming noxious stimulation [34]. Hyperalgesia and allodynia commonly observed in this population were consistent with the imaging findings.
Increasing evidence suggests that GFAP might be a biomarker for a number of neurological conditions, which is characterized by strong brain-specificity and high expression level in the brain. Brain injury causes the release of GFAP and its breakdown products from injured astrocytes to the extracellular space, where these proteins equilibrate into the subarachnoid CSF compartment, then release into the circulating blood by the glymphatic pathway or by diffusing past the (possibly compromised) brain blood barrier [35]. It has been also reported that GFAP can serve as an autoantigen, triggering AAb response in a subset of patients [35]. In our study, changes in anti-GFAP AAb level were positively associated with worse mental component score in the ME/CFS(+)FM group and with illness duration in the ME/CFS(−)FM group.
The network-based analysis has been recently implemented in the study of physiological autoimmunity and its disturbances. In particular, distinct signatures of anti-GPCR AAb in HC, which were influenced by age, gender, and various diseases have been revealed [10].
In our study, we determined the correlation signatures of some AAb targeting neural, internal organ-specific and non-organ-specific in HC, and in patients suffering from ME/CFS with and without comorbid FM. AAb to β2-GP, Fc-IgG, CoM, LuM-02, and GaM-02 had obviously more associations with other AAb in the ME/CFS(−)FM group than in HC. Danilenko et al. [36] showed, with the same method as in our study, that anti-β2-GP AAb were increased in ME/CFS patients, but not in healthy participants, which alluded to the link between ME/CFS and antiphospholipid syndrome, earlier suspected by Berg et al. in 1999 [37].
The ME/CFS(+)FM group was characterized by an increase in the number of associations of anti-collagen AAb and anti-TrM-03 AAb with other AAb. It has been recently reported that 81% of patients with ME/CFS and/or FM met Brighton criteria for hypermobility syndrome, and 18% met 2017 hypermobile Ehlers-Danlos syndrome criteria. Hypermobility scores significantly predicted symptom levels in these patients [38]. Notably, a high titer of AAb to type I collagen was found in patients with undifferentiated connective tissue dysplasia and joint hypermobility by other group of authors [39]. Earlier, such individuals were also demonstrated to be predisposed to anti-thyroid autoimmunity [40]. Another feature of the AAb correlation pattern in the ME/CFS(+)FM group was the abolished intragroup correlations between anti-neural AAb.
A major limitation encountered in this study was the small sample size. The other limitation was the absence of one universally accepted case definition of ME/CFS. When the underlying pathology of an illness is unknown, as with ME/CFS, there is no gold standard against which to assess the sensitivity or specificity of a case definition [12]; therefore, we had to use three sets of diagnostic criteria to confirm ME/CFS in our patients. We also did not have an opportunity to check the patients for other frequent comorbidities of ME/CFS and FM (e.g., irritable bowel syndrome, Ehlers-Danlos syndrome, postural orthostatic tachycardia syndrome etc.). One could suggest that these comorbidities may also have an impact on the natural AAb patterns. Further research is needed to confirm this suggestion.
Based on our observations, we assume that AAb are natural components of the immune system and may become dysregulated, not only in classical autoimmune conditions, but also in FM and ME/CFS. This assumption is in accordance with the perception of the role of the immune system in homeostatic regulation beyond host defense.