We aimed to assess the effects of non-pharmacological interventions on biomarkers (specifically cytokines, neuropeptides, and CRP) in patients with FM. Despite the importance of non-pharmacological interventions in FM patients, few studies have focused on changes to biomarkers after non-pharmacological intervention. In fact, to our knowledge, this is the first systematic review on this subject. We found only 12 articles that fulfilled our inclusion criteria. Of these articles, only one trial examined psychological interventions .
Our findings indicated that the levels of serum IL-8 in FM patients decreased after exercise or multidisciplinary interventions and those of IL-6 (spontaneous, LPS-induced, or serum) in FM patients tended to decrease after exercise or dietary interventions. For IL-8, three trials revealed a reduction in serum IL-8 levels between pre- and post-intervention, including 3 weeks of multidisciplinary exercise or 4 or 8 months of pool-aquatic exercise [28, 34, 35]. IL-8 is both a pro-inflammatory cytokine that activates neutrophils and a chemokine that plays an important role in the infiltration of neutrophils at inflammation sites . For IL-6, one trial demonstrated a reduction of the spontaneous and LPS-induced production of IL-6 between pre-intervention and 8 months post exercise within group . Moreover, another trial showed that the levels of serum IL-6 in the dietary therapy group decreased between baseline and after 6 months of intervention within group . Taking into consideration these findings, exercise, in particular, could act as an anti-inflammatory influence in FM patients, although it could act as a pro-inflammatory influence in healthy individuals . In other words, exercise interventions could improve the inflammatory status in FM patients, reaching values close to the basal levels in the healthy population, by adjusting the inflammatory-stress feedback mechanism .
In terms of anti-inflammatory cytokines, however, our findings were inconsistent. Ortega et al.  reported that the spontaneous production of IL-10 in FM patients decreased between pre-intervention and 8 months post exercise within group, whereas LPS-induced IL-10 to increase within group. It remains unclear if this discrepancy is related to only the diverse functions of IL-10 and other factors, or the smaller sample size.
In addition, with regard to growth factors, Bjersing et al.  demonstrated that the levels of total serum IGF-1 in lean FM patients increased after 15 weeks of exercise within group, whereas these levels were unchanged in overweight or obese FM patients within each group. They also found evidence of a positive influence of IGF-1 on fatigue. Given that the fatigue response exercise is related to levels of body mass index (BMI), changes in IGF-1 might indicate a beneficial role for fatigue in obese FM patients.
Neuropeptides and CRP
We found no consistent changes in the levels of neuropeptides or CRP due to non-pharmacological interventions. One trial  reported that the levels of serum NPY in obese FM patients significantly increased after 30 weeks (including 15 weeks of exercise and a 15-week follow-up) (p <0.05) within group, whereas in the other trial  they increased both in the dance/movement group and the control group compared to baseline after 14 months treatment, within each group.
With respect to other neuropeptides, however, some notable results have been observed, although only one trial was conducted to study each neuropeptide. Bjersing et al.  also reported that the levels of serum resistin in all FM patients increased after 30 weeks (including 15 weeks of exercise and 15-week follow up) within group, which correlated with decreased fatigue. In view of the abovementioned results, they noted that IGF-1 and resistin were involved in the mechanism that reduced fatigue after moderate exercise in FM patients.
Of the other trials, Lund et al.  demonstrated that the concentrations of urinary CRF-LI in the massage group decreased after 6 weeks of massage treatment and 1 month after completion of the treatments, within group. Furthermore, Bazzichi et al.  reported that the concentrations of serum BDNF significantly decreased in both 2-week balneotherapy and 2-week mud bath therapy after 12 weeks, within each group. Although these two trials [32, 33] were not exercise intervention but manipulative and body-based non-pharmacological interventions, CRF-L1 and BDNF could be used as indicators of the effects of each intervention in FM patients.
In terms of CRP, as with the neuropeptides, our findings were inconsistent. In three trials [28–30] the levels of serum CRP in the intervention group decreased compared to the baseline level after 4 or 8 months of pool-aquatic exercise (p <0.05, each), within group [28, 29] or 6 months of dietary therapy within group . In contrast, another trial  noted that there were no significant differences in the levels of plasma CRP between the guided imagery group and the group receiving usual care at baseline, 6 weeks or 10 weeks. Further research comparing each non-pharmacological intervention, specifically psychological and non-psychological intervention, is required to clarify the changes in CRP levels in FM patients.
The results of this review have several limitations. The primary limitation is the paucity and low quality of the included studies. We only included 12 articles (two trials were conducted with almost the same population), and of these, there were only 8 RCTs. The quality of the included trials has been assessed with the Cochrane risk-of-bias tool , which showed that only two studies [30, 31] were considered high quality, whereas the others were low quality with a high risk of bias. With regard to publication bias, we did not conduct the investigation using funnel plot analysis due to the high levels of heterogeneity among studies, and small sample sizes , and unfortunately we were unable to make any inference from this.
Another limitation is the high heterogeneity of non-pharmacological interventions. They consisted of three types: CAM, exercise, and multidisciplinary therapy. Of these types of treatment, CAM was also divided into the following categories: mind-body therapies (e.g., guided imagery, dance/movement therapy), manipulative and body-based therapy (e.g., balneotherapy, massage, mud-bath), and biological-based therapies (e.g., dietary therapy) . Furthermore, the duration of each intervention and each study had various patterns. In particular paying attention to follow up, of our included trials, six [26, 27, 32–34, 36] were conducted with a follow-up period, although two [26, 27] were based on almost the same population. Of these six trials, only two [32, 34] demonstrated that the effect of each intervention in the levels of biomarkers continued over the follow-up period within the intervention group. Lund et al.  showed that the concentrations of urinary CRF-L1 decreased over the 10-week study duration within the intervention group, and Wang et al.  reported that the levels of serum IL-8 decreased over the 6-month study duration within the intervention group. On the other hand, although we only included the studies with non-pharmacological interventions, there are few studies on the effect of pharmacological treatments in inflammatory biomarkers in FM, whereas it appears that antidepressants can normalize levels of some biomarkers such as ACTH . Our results for each intervention should be interpreted carefully.
An additional limitation is the high variability of materials. Biomarkers were investigated in diverse materials: serum, plasma, cerebrospinal fluid (CSF), urine, and saliva. Thus, a direct comparison of the results is hardly possible even when the same biomarker was compared.
Moreover, it is unclear whether our findings are exclusive to FM patients . Of all of our included studies, only FM patients or FM patients and healthy subjects were included. The non-pharmacological interventions that were conducted in our selected articles might affect other inflammatory diseases. Ploeger et al.  reported that the inflammatory response in patients with chronic inflammatory diseases compared to the healthy population generally increased after acute exercise. Further research is needed to elucidate whether non-pharmacological interventions are effective only for patients with FM.
In addition, there are not enough studies on the effect of non-pharmacological interventions on inflammatory biomarkers in healthy individuals, and we are unable to compare these effects and be certain that the effects reported here are related to FM. For instance, exercise could act as an anti-inflammatory influence in FM patients, whereas it could act as a pro-inflammatory influence in healthy individuals . Further research is required to explore whether the effects of non-pharmacological interventions on inflammatory biomarkers are different in FM patients compared to the healthy population.
Our findings, above all else, highlight the need to conduct further RCTs focused on biomarker changes after non-pharmacological interventions in FM patients. The following provisions would be helpful to improve future research in FM patients. First, non-pharmacological interventions should include not only exercise but also psychological treatment. Only one trial, guided imagery has been conducted as a psychological intervention . Of note, no study has assessed the efficacy of mindfulness on biomarkers except for cortisol  in FM patients. We need to elucidate the relationships between psychological interventions and biomarkers in patients with FM.
Second, biomarkers should be measured under the same conditions (e.g., controlling for menstrual phase and medications) to obtain homogeneous patient samples. As recommended in RCTs, control groups should be randomized and active, using standardized low-intensity non-pharmacological interventions (and not waiting list or non-active interventions). Follow up is mandatory and standardized periods should be accepted by researchers (for instance: post-treatment, 3-month and 12-month follow up). In addition, more neuropeptides should be included as biomarkers. Finally, the relationship between these biomarkers and pain-related measures deserves to be studied. Third, participants should be homogeneous, specifically for disease duration and age. Given that many of FM patients are female, more attention should be paid to the effects of menstruation. Moreover, participants who do not regularly take some kinds of medications (e.g., analgesic or psychotropic drugs) would be desirable. Future trials should be conducted with the aim of reducing the influence of these medications. On the other hand, participants would include not only patients with FM but also patients with other inflammatory diseases. Few studies have focused on comparing the effects of non-pharmacological interventions between FM and other inflammatory diseases.