Fatigue in chronic inflammation - a link to pain pathways

Fatigue is a multidimensional concept and has various causes. In rheumatic diseases, the association between fatigue and pain has been well established [2–4]. High fatigue is most often associated with high pain, and fatigue and pain seem to be synchronous [2, 5]. The link between fatigue and disease activity is less clear. First, fatigue is clearly a symptom included in rheumatic diseases: in rheumatoid arthritis (RA), it is an important outcome to evaluate according to OMERACT [6], and it has been associated with the Disease Activity Score in 28 joints (DAS28) and the Clinical Disease Activity Index [7]. In spondyloarthritis (SpA), fatigue is part of the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and appears more strongly related to the disease process than patient-related variables [8]. Furthermore, in anti-neutrophil cytoplasmic antibody-associated vasculitis, fatigue was associated with increased levels of C-reactive protein (CRP) [9]. However, a systematic review of fatigue in RA found no link between it and some characteristics of inflammatory activity, such as erythrocyte sedimentation rate or DAS28 [2]. Of note, Lee et al. [10] described a subgroup of RA patients with well-controlled disease but high persistent levels of fatigue (34 %).

These discrepancies could be explained by the variable definitions of fatigue or its multidimensional aspect. Indeed, several aspects can affect fatigue: illness-related characteristics (pain, inflammation, disease activity and joint damage), physical functioning (disability, health-related quality of life, sleep quality), cognitive and emotional impairment (anxiety and depression) and personal components (gender, age, social support, work and environment) (Fig. 1a) [2, 11–14]. The multidimensional nature of fatigue was well described by Hewlett et al. [13] in their conceptual model of the interaction between fatigue and three components - disease process, cognitive and behavior aspects, and personal life issues - with a bidirectional path suggesting interrelationships among these components. In this model, pain was included in the disease process and could cause fatigue but could also interact with other factors such as the inflammatory process (responsible for pain, joint damage and disability), anemia and sleep disturbance [2, 9, 13–15]. Conversely, in RA, fatigue seemed associated more with the global assessment of RA or pain than inflammatory components such as erythrocyte sedimentation rate or swollen joint count, perhaps because of the effect of sleep disturbance or reduced physical activity on fatigue [3, 4, 16, 17]. Among the other components of the Hewlett et al. model, cognitive and behavioral factors interacted with thoughts, feelings, behaviors, and symptoms [13]. More recently, Rongen-van Dartel et al. [18] showed a high level of daily physical activity was associated with reduced fatigue even after adjustment for pain and other confounding factors. In SpA, high fatigue was associated with the disease process, such as high disease activity (BASDAI), but also personal components, such as poor quality of life, and other disease activity (bowel symptoms) [19, 20]. Therefore, fatigue is a subjective symptom that interacts with the multiple aspects involved in inflammatory diseases.

Fatigue is common among individuals living with a chronic illness, particularly a disease with an overriding inflammatory process: rheumatologic diseases such as RA or SpA, cancers, inflammatory bowel diseases, connective tissue diseases such as systemic sclerosis, systemic autoimmune diseases, autoimmune type 1 diabetes and infections [14, 21, 22]. In a descriptive study, fatigue was more severe in patients with inflammatory bowel disease than in controls whatever their age [23].

In patients with cancer (acute myelogenous leukemia and myelodysplastic syndrome), fatigue severity was correlated with serum levels of the inflammatory cytokines interleukin (IL)-6, tumor necrosis factor (TNF)-α and the IL-1 receptor antagonist (IL-1RA). In lung cancer patients, IL-8 was a relevant genetic factor of pain and fatigue [24, 25]. Moreover, the expression of IL-6 and nuclear factor-kappa B (NFκB) was increased in oncology patients with sleep disturbance [26]. In these patients, inflammation-induced fatigue could be explained by cancer treatments (radiation, chemotherapy) but also by secretion of inflammatory cytokines by the tumor [27]. Even after chemotherapy or radiation treatments, fatigue could persist for up to 10 years, when levels of a number of pro-inflammatory plasma markers, including IL-6, IL-1RA, CRP and soluble TNF receptor type II, were high [27, 28]. A cumulative effect of levels of cytokines corresponded to the number of days with fatigue in both breast and prostate cancer patients [28]. As well, circulating T lymphocytes were increased in number, with no alteration in immune cells [28]. A recent review described fatigue occurring with inflammation before, during and after treatment with several cancers. Fatigue was well correlated with high levels of inflammatory peripheral cytokines (IL-6, IL-1 and TNF), which could signal the central nervous system (CNS) and generate fatigue or other behavioral symptoms [29].

For some authors, the link between inflammation and fatigue is less clear. A subgroup of patients with high fatigue and high depression harbored a minor allele for the anti-inflammatory IL-4 [30]. Moreover, fatigue was associated with some immunotherapies for HIV/AIDS, such as interferon (IFN)-α and IL-2; about half of 317 patients living with HIV/AIDS reported high levels of fatigue. In this study, plasma TNF levels were lower with antiretroviral therapy, CD4+ T-cell counts of at least 200 cells/mm³ and undetectable viral loads; however, on multivariable analyses, plasma levels of none of the cytokines evaluated was significantly associated with fatigue [10].

However, the replication of studies showing an association between expression of genes or inflammation cytokines and fatigue regardless of chronic illness suggests that inflammation could play a role in the fatigue experience (Fig. 1b).

Unusual and chronic fatigue with various etiologies was described for 27 % of patients in a primary care clinic [31]. Fatigue is common in RA, SpA, Sjögren syndrome, systemic lupus erythematosus and vasculitis, although most publications concerned fatigue in RA or SpA [9, 14, 32–35]. In RA and SpA, the frequency of fatigue ranged from 42 % to 80 % depending on the definition and methods of assessment [1, 4, 8, 36–39]. For 75 % of patients with ankylosing arthritis and 50 % of those with RA, fatigue was considered severe [4, 39].

Several methods of evaluation have been used to investigate fatigue in rheumatologic diseases [14]. The simplest and quickest scale is the visual analog scale (VAS; scores from 0 to 100; the higher the score, the greater the fatigue), but this is a unidirectional scale. Multidirectional scales developed to include the different aspects of fatigue are the Medical Outcomes Study Short Form 36 (SF-36) vitality subscale (four questions; scores from 0 to 100; the higher the score, the lower the fatigue), the Functional Assessment of Chronic Illness Therapy Fatigue Scale (domains physical, social/family, emotional and functional; scores from 0 to 52; the higher the score, the lower the fatigue), the Profile of Mood States, the RA-specific Multidimensional Assessment of Fatigue (MAF) scale, the Multidimensional Fatigue Inventory, the Brief Fatigue Inventory, and the Fatigue Severity Scale [40–42]. With the VAS scale, the mean fatigue level in patients with RA was 42.1 [43].

Classical treatments of RA and SpA consist of pharmacological treatments to control inflammation and multidisciplinary interventions such as cognitive behavioral therapy and physical exercises to reduce symptoms and maximize self-management [44]. Among pharmacological treatments known to reduce disease activity in RA, disease-modifying antirheumatic drugs (DMARDs), such as methotrexate and leflunomide, or biologic treatments, such as anti-TNF (infliximab, adalimumab, etanercept, golimumab and certolizumab), anti-IL-6 (tocilizumab), CTLA4 immunoglobulin (abatacept) and anti-CD20 (rituximab), have improved pain and mood disorders as well as fatigue, although the overall effect size of biotherapies on fatigue was small (effect size = 0.45; 95 % confidence interval 0.31 to 0.58) [4, 45, 46]. More recently, new biologic therapies have shown efficacy for fatigue: FACIT-Fatigue and SF-36 vitality scales were improved with secukinumab, an antibody against IL-17, and tofacitinib, an oral Janus kinase inhibitor [47, 48]. The placebo effect on fatigue was seen in our meta-analysis of the effect of biotherapies on fatigue: scores ranged from 1.04 ± 22.6/100 (DEO19 study) to 11.57 ± 21.92/100 (REFLEX study) [46]. Wells et al. [49] found that 69, 84 and 90 % of American College of Rheumatology 20/50/70 responders receiving abatacept, respectively, had a >20 % improvement in fatigue according to the VAS. So even if the level of concordance between fatigue and disease activity was high, 10 % to a third of patients had fatigue whereas disease activity improved. With secikinumab, however, patients reporting increased pain showed worsened fatigue according to both fatigue scores [48]. After 3 months of anti-TNF therapy for RA, fatigue was decreased in patients but was independent of the level of CRP: on multiple regression, only global health and tender joint count explained 34 % of the variance in fatigue [50]. Otherwise, a randomized controlled trial showed significant benefits of an exercise program on fatigue scores, quality of life, pain and sleep quality [51]. After six weekly sessions of behavioral therapy and a consolidation session, RA patients with initial VAS fatigue score ≥6/10 reported better fatigue scores than controls (MAF and VAS scales) and better perceived fatigue severity, coping, disability, depression, helplessness, self-efficacy and sleep [52].

Fatigue and pain are two common symptoms in RA, and the link between fatigue and pain in it has been described [2–4, 53]. In a cross-sectional study, VAS fatigue score was mainly correlated with VAS pain score; pain was most strongly associated with the five variables explaining fatigue [4]. Garip et al. [53] confirmed this association, showing that fatigue intensity in RA patients was strongly correlated with VAS pain score and DAS28, with greater correlation between fatigue and pain scores than between fatigue score and DAS28. Moreover, after treatments (DMARDs or biologic agents), the decrease in VAS fatigue score was correlated with ameliorated pain and improved DAS28 score [4]. For Wolfe et al. [11], pain was a strong independent predictor of fatigue with sleep disturbance, depression, tender joint count and disability by the health assessment questionnaire. Among the variables usually linked to fatigue, pain was a better predictor of fatigue [54].

To examine the bidirectional effect of pain and fatigue, van Dartel et al. [5] conducted a prospective study of patients with established RA who received DMARDs and/or biologic agents; pain and fatigue were measured monthly for 1 year. Pain and fatigue levels fluctuated, and the change in fatigue level was positively associated with change in pain level during the same month. However, change in fatigue level was not related to a change in pain level that occurred 1 month earlier, and change in pain level was not related to a change in fatigue level 1 month earlier [5]. Moreover, in a cohort of patients with fibromyalgia, usually considered a non-inflammatory disease, pain and fatigue scores were correlated (r = 0.45, P < 0.001) and, in the prospective analysis at 1 week, daily pain evaluation predicted increased fatigue level reported the next day (more than depression or daily sleep quality) [16].

Although fatigue, stress or depression, and pain have complex and various mechanisms of action, some inflammatory cytokines are found associated with these three domains, so inflammation may be their potential link. The association between these symptoms has been documented in various medical conditions: classical inflammatory diseases such as rheumatic diseases, cancer or infections but also CFS, metabolic disorders or depression, which exhibit low-grade inflammation [77].

Some authors found that in RA, fatigue, mood disorders and pain are clinically associated [2]; IL-1 level was higher in cerebrospinal fluid of patients than controls and the increased level was correlated with fatigue [78]. Biologic treatments are efficacious and can decrease levels of markers of inflammation such as CRP; tocilizumab is especially efficacious because of its action on IL-6 involved in the synthesis of CRP [79]. Thus, decreased systemic inflammation could be one of the mechanisms of action improving fatigue, pain and mood disorders.

Moreover, in a cohort of 1,466 European patients with advanced cancer, increased CRP level was associated with pain and fatigue (rho = 0.154 and 0.197, respectively) [80]. In ovarian cancer patients, diurnal and nocturnal cortisol and plasma IL-6 levels decreased (became normal) during the first year following surgery, and this decrease was associated with decreased fatigue and depression [81].

Another model of the association of fatigue, pain and depression is CFS. Recently, a review showed that chronic inflammation could explain, in part, the sickness behavior [57]. In this pathology, with increased sensitivity to pain and with sickness behavior, inflammatory cytokines could have an effect on nociception. Such cytokines appeared to be critical mediators of hyperalgesia in a lipopolysaccharide-induced animal model [82, 83]. In CFS, the levels of neopterin, a marker of cellular immune system activation, IL-1 and TNF were correlated with fatigue and depression [84]. Brain inflammation could have a role in CFS, in part by activating microglia or astrocytes. Indeed, the density of 11C-(R)-(2-chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline-carboxamide, a marker of neuroinflammation, was increased in some areas of the brain, in particular the cingulate cortex, in patients with CFS compared with healthy patients. Its increased level in the thalamus was correlated but not significantly with pain score and fatigue sensation (P = 0.0683) [85].

We have emerging evidence of the role of microbiota in the pathogenesis of autoimmune disease, particularly in rheumatologic disease [86]. Recently, Galland [87] proposed a schema in which the gut microbiome could affect CFS or fibromyalgia: the bacterial components could excessively stimulate the innate immune system and induce systemic and CNS inflammation by producing neurotoxic metabolites or could directly stimulate afferent neurons of the nervous system to send signals to the brain via the vagus nerve. Then, the gut microbiome could affect the HPA axis and be responsible for fatigue and pain in these diseases.

Moreover, it is interesting to highlight brain areas involved in these various domains. With the Kana Pick-out Test, in the model of induced-fatigue previously described, the dorsolateral prefrontal cortex and cingulate cortex were activated as seen on functional MRI [64, 65]. In adults, noxious stimulation increased MRI activity in primary somatosensory cortices, the anterior cingulate cortex, bilateral thalamus, and divisions of the insular cortices [88]. There is a common area for neural activity in experiences of physical pain, induced fatigue, depressive mood and bacterial endotoxin-induced inflammation (with increased IL-6 level): the anterior insula, believed to play a role in consciousness and in emotion or regulation of the body’s homeostasis, and anterior cingulate cortex, known to be involved in autonomic and cognitive functions [64–66]. A common CNS pathway would be a link between fatigue and pain, and among the different mechanisms of fatigue, the action of inflammation on the CNS could be one of these pathways.