1. Introduction
Fibromyalgia (FM) is a chronic syndrome of unknown pathophysiology characterized by widespread musculoskeletal pain coupled with fatigue, stiffness, disordered sleep, perceived cognitive dysfunction, and mood disturbances [
1]. One of the current hypotheses suggests that an unbalanced immune system may play a role in the pathophysiology of FM [
2]. Several studies evaluated the role of interleukins (IL), chemokines, and immune-related substances in FM, resulting in inconsistent results. For example, increased levels of IL-6, CXCL-8, high-sensitivity C-reactive protein (hs-CRP), and IL-10 were found in some [
3,
4,
5], but not all studies [
6,
7,
8]. Moreover, other results supported the view that FM may be accompanied by suppression of the immune-inflammatory response system (IRS) [
8]. In addition, in order to obtain a more comprehensive picture of immune function in FM, it is important to consider not only biomarkers of the IRS, including CXCL-8 and hs-CRP, but also immune regulatory or anti-inflammatory cytokines, including IL-10, and cytokines with combined IRS and regulatory properties such as IL-6 [
9].
Results of published systematic reviews and meta-analyses indicate that methodological differences among immune studies are a main source of inconsistent results [
10,
11]. Major sources of heterogeneity might be due to uncontrolled variables including age, sex, body mass index (BMI) [
12], sleep disorders [
13,
14], or comorbid medical disorders [
15,
16]. In addition, there is a wide range of clinical extraneous variables that were constantly disregarded, including depressive and anxiety symptoms [
17,
18], intake of antidepressants (ADs) [
19,
20], menopause [
21], use of anticonvective hormones [
22], aerobic exercise [
23,
24], seasonality, and diurnal variations [
25,
26].
Hence, in this case-control study we examine serum levels of pro-inflammatory markers, including CXCL-8 and hs-CRP, as well as IL-6 and the major negative immune-regulatory cytokine IL-10 in FM as compared with healthy controls, while adjusting for most of the above-mentioned extraneous variables. Based on the immune-inflammatory hypothesis of FM, the specific hypothesis is that - after controlling for relevant socio-demographic and biomedical confounders - FM is characterized by higher levels of IL-6, CXCL-8, and hs-CRP and lowered or unchanged levels of IL-10. Additionally, the predicting ability of the inflammatory biomarkers along with other predictors (e.g., severity of depression and anxiety) was examined using neuronal network (NN) machine learning. NN is well suited to discover more complex nonlinear relationships between input variables (i.e., cytokines, depression, anxiety) predicting FM diagnosis and widespread pain scores (the outcome variables) while examining multiple interrelated factors. Additionally, NN allows to rank the input variables in order of predictive power of the features.
3. Discussion
The major finding of this case-control study is that, contrary to our a priori hypothesis, IL-6 and CXCL-8 levels were significantly decreased in patients with FM compared with controls. Moreover, IL-10 was also significantly lower in FM patients, while there were no significant differences in hs-CRP after controlling for confounding variables. Our results on IL-6 levels are not in agreement with most published papers, which reported higher levels of IL-6 in FM [
6,
28,
29,
30], although some studies found no significant differences between FM patients and controls [
4,
8,
31,
32,
33,
34]. Previously, some authors reported lower levels of CXCL-8 in FM [
28,
30]. Nevertheless, as with IL-6, there is a large heterogeneity between published studies with a number of studies reporting no significant differences in CXCL-8 [
3,
33,
35] and some even increased CXCL-8 levels [
30]. In accordance with a previous study, we found lowered levels of IL-10 in FM [
6], although other studies reported higher levels of IL-10 [
3,
4]. Similarly, previous results on hs-CRP in FM were mixed with some studies reporting higher levels [
34,
36], while other studies could not find any differences between FM patients and controls [
37,
38].
All in all, the results show that there is a large heterogeneity among studies in cytokine-chemokine results. This is further corroborated by the results of a recent systematic review [
10], which was unable to delineate a specific immune profile in FM and concluded that the methodological quality of immune studies in FM should be improved. A major source of heterogeneity is that most studies did not control or only partially controlled for the potential important confounders, which may modulate immune biomarkers. In the present study, we found that after controlling for BMI, comorbid illnesses, aerobic exercise, anxiety, depression, perceived stress symptoms, and quality of sleep, the differences in IL-6, IL-10, and CXCL-8 between groups remained significant, whereas hs-CRP was no longer significantly increased in FM. Since patients with FM usually present with higher BMI values than observed in the general population [
39,
40], it is plausible that the differences in hs-CRP levels reported in some studies are attributable to effects of BMI. Previously, it was shown that BMI is also a major determinant of increased hs-CRP values in bipolar disorder and major depression [
41,
42]. In this work, we have found that two other confounders, namely age and quality of sleep, may interfere with cytokine-chemokine measurements in FM. Firstly, age was significantly and negatively correlated with IL-6 levels, although some but not all studies showed that IL-6 may increase with age [
43]. Secondly, we found a very modest inverse association between quality of sleep and IL-10 levels. Moreover, FM is a syndrome with a high prevalence of comorbidities with medical and psychiatric conditions (e.g., obesity, arthrosis, depression, anxiety), which can influence peripheral levels of immune-inflammatory biomarkers and which should, therefore, be considered when interpreting cytokine results. As such, the effects of comorbidities combined with other variables including quality of sleep, BMI, age, sex, and exercise could explain the discrepant results reported in the literature.
Another important issue is that it is difficult to interpret results on cytokines-chemokines without considering their pro- versus anti-inflammatory profile, because the balance between pro-/anti-inflammatory biomarkers determines the functional state of the immune system [
44]. As such, we have assayed 2 pro-inflammatory mediators (CXCL-8 and hs-CRP), the major negative immune-regulatory cytokine (IL-10) and a cytokine (IL-6) that may exhibit pro-inflammatory as well as anti-inflammatory effects [
44,
45]. In this regard, z unit weighted composite scores may be computed which denote different functional immune ratios including pro/anti-inflammatory or IRS/regulatory [
44]. In our study, we expected to find immune activation as characterized by an immune profile with increased levels of the four biomarkers or an inflammatory response as indicated by increased levels of hs-CRP, IL-6 and CXCL-8, and possibly a relative decrease in IL-10, an anti-inflammatory and regulatory cytokine [
44]. Nevertheless, the immune profile observed in our study showed a reduction in IL-6, IL-10, and CXCL-8 indicating that there is no clear immune activation and no inflammatory response (according to these classic biomarkers) in our FM patients. This is important as our findings do not corroborate the major hypothesis that IL-6 and other cytokines, which may induce pain, mental, and cognitive impairments, are increased in that illness. Recently, IL-6 has emerged as a key regulator of pathologic pain and administration of IL-6 may cause allodynia and hyperalgesia [
46]. Pain models are frequently accompanied by increased levels of IL-6 and the IL-6 receptor (IL-6R) in the spinal cord and dorsal root ganglia [
46]. This hyperalgesic effect of IL-6 is probably mediated via IL-6 trans-signaling through increased levels of the soluble IL-6 receptor (sIL-6R), which binds with IL-6 to form an active pro-inflammatory complex [
47].
The second major finding of our study is that regression analysis showed that, taking into consideration the measured immune biomarkers, IL-10 is the best predictor of FM diagnosis, while NN analysis showed that the three cytokines measured herein are the best predictors of the WPI score and IL-10 the best immune biomarker predicting the SSS score. Comparable results indicating lowered levels of anti-inflammatory cytokines were reported in a study that examined IL-4 and IL-10 gene expression patterns and serum concentrations in chronic widespread pain patients including FM [
48]. IL-4 is a T helper-2 (Th-2) cytokine, and IL-10 is a T regulatory (Treg) cytokine, and both may suppress an immune response [
9]. Thus, these authors found lower levels of the anti-inflammatory cytokines IL-4 and IL-10 in patients with chronic pain, whereas there were no significant differences in pro-inflammatory cytokines-chemokines including IL-2, CXCL-8, tumor necrosis factor (TNF)-α, and transforming growth factor (TGF)-β1, another immune-regulatory cytokine. The authors concluded that chronic pain is associated with lowered levels of anti-inflammatory and analgesic Th-2 (and Treg) cytokine activity [
48]. Both Th-2 and Treg cytokines attenuate the production of pro-inflammatory M1 macrophage and Th-1 cytokines, suggesting that lower levels of IL-10 in FM may attenuate the negative feedback on the inflammatory response system. Nevertheless, in our study some other immune markers were suppressed (IL-6 and CXCL-8) or unchanged (hs-CRP), while in the study of Üçeyler et al. [
48] reduced Th-2 and Treg activities were not accompanied by signs of immune activation in patients with chronic pain.
The above findings show that there should be another explanatory hypothesis, which combines the effects of lowered IL-10 and IL-6 in explaining FM symptoms. One hypothesis is that this immune profile may indicate loss of neuroprotection because IL-10 [
49] and classical IL-6 signaling [
45,
50] have (neuro)protective properties. For example, the lack of IL-10 may accelerate lesion expansion following excitotoxic injuries in association with the onset of pain behaviors, while IL-10 may protect against this type of lesions and the accompanying pain [
51]. Furthermore, FM is accompanied by highly increased serum levels of sgp130, the natural inhibitor of IL-6 trans-signaling and an inhibitor of the activity of leukemia inhibitor factor (LIF) and oncostatin [
52]. LIF is a growth factor that plays a role in genomic plasticity; it has regenerative effects and promotes T regulatory functions, and repair and tolerance in the central nervous system [
53,
54]. Oncostatin is produced by myofibers and maintains the regenerative capacity of stem cells in the muscle [
55]. As such, it appears that FM may be accompanied by disorders in the IL-6–sgp130 system and in regenerative IL-6-related cytokines.
Last but not least, our neural networks analysis showed that widespread pain symptoms (WPI subscale score) are strongly predicted by sleep quality, perceived stress and anxiety, and that the FM diagnosis is significantly predicted by sleep quality, perceived stress, anxiety, and aerobic activities indicating that FM is much more than a pain condition. These results extend the findings of Wolfe et al. [
56] who reported that the original FM criteria (1990) [
57] underestimated the prevalence of FM by approximately 25% and that therefore it is more appropriate to use the revised ACR diagnostic criteria, which comprise sleep problems, depression, fatigue, gastro-intestinal, and cognitive symptoms.
Our findings should be interpreted with regard to its limitations. Firstly, this study recruited only females who represent the majority, but not all patients with FM. According to the last ACR criteria, there are around 33% males in the FM population. Secondly, the sample size was rather small particularly the normal control group. Therefore, our findings deserve replication in males using a larger study sample. Thirdly, future research should include of a more comprehensive list of cytokines, including M1 macrophage, Th-1, Th-2 and Th-17 cytokines altogether with other proteins implicated in the immune status, such as sgp130, LIF, and oncostatin.