The High Costs of Low-Grade Inflammation: Persistent Fatigue as a Consequence of Reduced Cellular-Energy Availability and Non-adaptive Energy Expenditure

1Chronic or persistent fatigue is a common, debilitating symptom of several diseases. It is one of the most frequently reported symptoms of cancer and cancer treatment (Servaes et al.,; Abrahams et al., ) and is highly prevalent in several chronic diseases, such as multiple sclerosis, diabetes, and rheumatoid arthritis (Wolfe et al.,; Drivsholm et al.,; Induruwa et al.,; Sanoobar et al., ). In addition, it is the hallmark symptom of chronic fatigue syndrome, a condition in which severe persistent fatigue is experienced in absence of a diagnosed disease (Fukuda et al.,; Afari and Buchwald, ). Persistent fatigue is distinct from acute fatigue. Acute fatigue is a healthy, adaptive response to physical or mental exertion, inducing metabolic signaling to prevent further energy consumption (Keyser, ). Acute fatigue typically resolves after rest or sleep. In contrast, persistent fatigue is often disproportional to exerted activities and is generally not completely alleviated by rest. No treatments for persistent fatigue have been approved by the US Food and Drug Administration, in part because the underlying mechanisms are still poorly understood.

2Activation of inflammatory pathways has been suggested to underlie persistent fatigue in many patient populations (Bower,; Karshikoff et al.,; Lasselin et al.,; Montoya et al., ) and animal models (Krzyszton et al.,; Mahoney et al.,; Bonsall et al.,; Norden et al.,; Zhang et al.,; Vichaya et al., ). Indeed, it is well-known from experimental studies that acute severe inflammation, such as induced by lipopolysaccharide (LPS), causes acute sickness behavior, including fatigue. This response has been interpreted as an adaptive process leading to the conservation of energy and reduction of the risk of further dissemination of pathogens (e.g., by withdrawing from social interactions) (Dantzer et al.,; Engler et al., ). Moreover, in autoimmune diseases such as multiple sclerosis, diabetes, and rheumatoid arthritis, the level of fatigue is associated with an increase in plasma cytokines, especially during symptom relapse (Lasselin et al.,; Malekzadeh et al.,; Patejdl et al.,; Choy and Calabrese, ). Associations between small, prolonged increases in plasma inflammatory cytokines and chemokines and persistent fatigue have also been reported in cancer survivors (Bower and Lamkin, ) and in individuals with chronic fatigue syndrome (Montoya et al., ). In these patients, the levels of plasma cytokines are generally much lower than those detected in patients with autoimmune diseases. The mechanisms by which subtle increases in inflammation induce fatigue are still unclear, and could be different from what has been shown for acute severe inflammation. However, the number of associations reported in the literature suggest that the effect is biologically significant.

3We propose that chronic low-grade inflammation induces and/or maintains persistent fatigue by inducing an imbalance between cellular-energy availability and cellular- and behavioral-energy expenditure (Figure 1). Inflammation increases the need of immune cells for rapid generation of cellular energy. To meet this need, immune cells shift to aerobic glycolysis for energy production, a less-efficient, but fast-acting pathway (Kominsky et al.,; McGettrick and O'Neill, ). During chronic low-grade inflammation, the extended reliance on aerobic glycolysis would be expected to lead to reduced nutrient availability and thus to less energy availability for demanding organ systems. The organismal energy balance can further be encumbered by changes in circadian rhythms and sleep. In addition, there is evidence suggesting that low-grade or chronic inflammation (but not acute severe inflammation) can be linked to increases in behavioral-energy expenditure (Vichaya et al.,; Lasselin et al., ), contributing to the imbalance between energy availability and expenditure and, thereby, leading to fatigue.

4Adenosine triphosphate (ATP) is the primary form of energy “currency” utilized by the cells. The generation of ATP involves the catabolism of macronutrients (carbohydrates, lipids, and proteins), each starting from a unique metabolic pathway but ultimately being shuttled to the cells for ATP production (Figure 2). Factors that can negatively affect ATP production include reduced intracellular glucose availability through either alterations in macronutrient metabolism or reduced glucose uptake by the cells; reduced functioning of the mitochondria for aerobic energy production and subsequent increased dependence on less-efficient aerobic glycolysis.

5Inflammation requires a change in metabolism, and these changes differ between acute and chronic or prolonged inflammation (Figure 3). During acute inflammation, rapid ATP production in immune cells is required for the multitude of immune responses. Immune cells (lymphocytes and leukocytes) switch from oxidative phosphorylation to an increased reliance on aerobic glycolysis for rapid ATP production (Kominsky et al.,; McGettrick and O'Neill,; Kelly and O'Neill, ). This shift is in favor of precipitous ATP production while catabolic efficiency is sacrificed, resulting in decreased mitochondrial function and increased lactate production.

6Additionally, prolonged inflammation is associated with increased insulin resistance and reduced glucose tolerance, leading to reduced glucose uptake by the immune cells and, consequently, less overall ATP production from glucose (Shoelson et al.,; Asghar and Sheikh, ). Therefore, during ongoing low-grade inflammation, energy and glucose are diverted from other systems to support the immune response. To compensate, the body increases lipid and protein metabolism for ATP production (Liu et al., ). Lipid metabolism is a slow process and thus cannot respond to rapid energy requirements. Additionally, increasing reliance on protein metabolism means other functions, such as promotion of growth stimulation, are compromised (Le Floc'h et al., ). Inflammation also leads to increased reactive oxygen species (ROS) to combat invading pathogens (Liu et al.,; Maitra et al., ). However, ROS can be harmful to mitochondria and mitochondrial DNA (Sinha et al., ) and thus, prolonged exposure to ROS during chronic inflammation contributes to mitochondrial dysfunction.

7The effects of chronic inflammation are especially detrimental for neurons that heavily depend on oxidative phosphorylation for their ATP needs (Hall et al., ). In normal brain metabolism, neurons rely heavily on astrocytes. Astrocytes use glycolysis to convert lipids and glucose to pyruvate and then lactate. Lactate is then shuttled to neurons where it is enters the tricarboxylic acid (TCA) cycle. During chronic inflammation, insulin resistance decreases astrocytic glucose and protein metabolism, resulting in reduced lactate availability for neurons (Blázquez et al., ), forcing the neurons to rely more heavily on the slower process of lipid metabolism. Further, inflammation is able to low mitochondrial efficiency (Yan et al., ), increasing the metabolic burden on neurons. Alterations in neuronal metabolism will ultimately affect neuronal functioning.

8Chronic low-grade inflammation has been estimated to increase systemic energy expenditure by up to 10% (Straub, ). Given the increased energy demand of chronic inflammation, and the resulting reduced energy availability, one would expect that behavioral-energy expenditure (i.e., amount of energy spent on activities) would decrease. Several studies in the context of acute, severe inflammation corroborate this hypothesis showing reductions in willingness (motivation) to exert effort, both in rodent models (Larson et al.,; Felger et al.,; Nunes et al.,; Yohn et al., ) and in humans (Draper et al., ) (Table 1). However, results from these and other studies also suggest that motivational behavior might be differentially affected in conditions of low-grade inflammation.

9Studies on animal models show a reduction in effort expenditure for food only after administration of higher doses of inflammatory mediators (Larson et al.,; Felger et al.,; Nunes et al.,; Yohn et al., ). For the lowest doses, not only was there no reduction in activity, there also seemed to be a trend for increased activity (Larson et al., ). Further, Vichaya et al. tested animals 24 h after an acute inflammatory insult (LPS injection), at which point the acute inflammatory response and sickness behavior had subsided. A shift was observed in motivational priority toward choices that required a higher effort for a more salient reward.

10Comparing the results from the two available studies in humans also point to distinct effects of high vs. low-grade inflammation on effort expenditure choices. Draper et al. tested healthy individuals 2 and 5 h after LPS/saline injection and found a reduced willingness to perform high-effort tasks in LPS-treated participants at 2, but not 5, h post injection. At 2 h, inflammation was as its peak. In contrast, Lasselin et al. tested participants 5 h post LPS/saline injection and observed an increased willingness to perform high effort tasks in LPS-treated participants (but only when the conditions to win were favorable). Lasselin et al. further noted that participants were too weak to be tested between 3 and 4 h post injection. Of note, the two studies differed in several important aspects including sample characteristics (only males vs. mix of males and females) and task design (choice between accepting or rejecting a task vs. choice between a high effort/high reward or a low effort/low reward task) (see Table 1 for study specifics). Nevertheless, we hypothesize on the basis of these human and murine experimental data that acute severe inflammation leads to overall adaptive reductions in effort expenditure, whereas mild inflammation can lead to relative increases in effort expenditure provided the incentive is strong enough. This notion is supported by our findings that low-grade inflammation was associated with increased high-effort choices in a sample of cancer patients and survivors (Lacourt et al., under review). It is still to be determined if these behavioral shifts in effort expenditure correspond to “recovery” or to a switch in immune cells back from a glycolytic to a more oxidative-driven metabolic profile. In addition, behavioral shifts during chronic low-grade inflammation have not yet been studied.

11The effects of inflammation on effort expenditure in motivational tasks appear to be mediated by the dopaminergic system (Felger et al.,; Yohn et al., ). Although the mechanism by which inflammation is capable of altering dopamine neurotransmission is still unclear, the literature on Parkinson's disease indicates that chronic neuroinflammation leads to oxidative stress and mitochondrial damage in dopaminergic neurons (Niranjan,; Blesa et al., ), leading to an impairment in dopaminergic neurotransmission. Dopaminergic neurons are particularly vulnerable to inflammation (de Pablos et al.,; Park et al., ). Although the cause of this vulnerability is not fully understood, the higher basal oxidative phosphorylation of these cells and the auto-oxidation of excess extracellular dopamine may contribute (Asanuma et al.,; Pacelli et al., ).

12Inflammation can additionally lead to more-sustained alterations in metabolism via changes in circadian rhythms and sleep. Chronic low-grade and severe inflammation has been shown to affect sleep quality and circadian rhythms (Pollmächer et al.,; Haspel et al.,; Anderson et al., ), and vice versa (Leproult et al.,; Rahman et al.,; Wright et al., ). Alterations in circadian activity rhythms have been linked to fatigue (Payne,; McHill and Wright, ). Sleep disturbances or reduced sleep quality—often reported by individuals with persistent fatigue—can also lead to disorganized or misaligned circadian rhythm through daytime napping and structural alterations in nighttime sleep onset.

13Most metabolic processes are regulated by the circadian clock, and changes in circadian rhythm or sleep are associated with metabolic changes, such as increased circulating glucose and decreased insulin sensitivity (Depner et al.,; Potter et al., ). Specifically, expression of mitochondrial proteins involved in many metabolic processes undergo posttranslational modifications regulated by the Clock gene (Masri et al., ). In addition, mitochondrial oxidative metabolism is also controlled by the circadian clock through oscillations in biosynthesis of nicotinamide adenine dinucleotide (NAD+) and availability of rate-limiting mitochondrial enzymes (Peek et al.,; Neufeld-Cohen et al., ).

14Insufficient sleep and changes in circadian rhythms can also alter cellular-energy and behavioral-energy expenditure. Preclinical studies show increased energy expenditure after total sleep deprivation in rats (Caron and Stephenson, ). Likewise, experimentally induced sleep deprivation in healthy human volunteers led to increased daily energy expenditure, assessed via whole-room calorimetry (oxygen consumption and carbon dioxide production). Although this was accompanied by a compensatory increase in energy (food) intake (i.e., caloric intake exceeded caloric expenditure), physical exhaustion increased significantly (Markwald et al., ). As the above described studies rely on experimentally induced alterations in circadian rhythm and sleep, it is unclear to what extent these findings are relevant for naturally occurring changes. In addition, whether more subtle, sustained changes in sleep and circadian rhythm, such as observed in fatigued individuals, also alter energy expenditure remains to be shown.

15Chronic low-grade inflammation is related to a persistent metabolic switch to faster, but less efficient, glycolytic energy production, and increased ROS production. This in its turn affects mitochondrial function, thereby increasing the reliance on glycolytic pathways. In addition, although chronic inflammatory processes demand more energy, this increased need for energy is not always accompanied by decreases in behavioral-energy expenditure. This seems particularly the case in the context of low-grade inflammation, which has been shown to increase motivation for effort expenditure. Inflammation can also affect circadian rhythms and sleep (and vice versa), which can further exacerbate the effects on energy availability and expenditure.

16We propose that this imbalance of energy availability vs. expenditure underlies the experience of fatigue induced by chronic low-grade inflammation. Below, we discuss the evidence for these metabolic pathways in persistent fatigue.

17Associations between low-grade inflammation and persistent fatigue have been reported predominantly for individuals with CFS or CRF. In patients with CFS, fatigue severity has been associated with elevated cytokine concentrations in plasma (e.g., Montoya et al., ) and spinal fluid samples (e.g., Hornig et al., ). Alterations in physiological stress-response (predominantly hypothalamic-pituitary-adrenal axis output) in patients with CFS have also been reported (e.g., Silverman et al., ), suggesting that low-grade inflammation might be the result of reduced sensitivity to immune-regulating stress hormones such as cortisol. For CRF, fatigue usually starts during cancer treatment but persists into survivorship in a significant subset of patients (Servaes et al.,; Goedendorp et al., ). Most first-line treatments such as chemotherapy and surgery are known to induce an inflammatory response (e.g., due to tissue damage) (Fitzpatrick and Wheeler,; Boomsma et al.,; Wang et al., ). Increases in inflammatory markers have indeed been associated with increased fatigue severity in patients actively undergoing treatment, although not all inflammatory markers associated with fatigue show a response to cancer treatment, suggesting the presence of other causal factors (Reed et al., ). In cancer survivors, despite the cessation of the immediate inflammatory effects of cancer treatment, associations between elevated levels of inflammatory markers and increased fatigue have also been observed (Bower, ). CRF in survivors has in addition been associated with psychosocial stress factors, such as fear of cancer recurrence (Young and White, ), suggesting that inflammatory processes during survivorship might (in part) be stress-related (Lacourt and Heijnen, ).

18Evidence of reduced energy availability in persistent fatigue comes mainly from metabolomic studies (e.g., Fluge et al.,; Naviaux et al.,; Yamano et al., ) and from studies of mitochondrial dysfunction (e.g., Filler et al., ).

19Mitochondria are often represented as the “powerhouses of the cell.” Mitochondrial function can be compromised by a multitude of factors, including reduced availability of necessary metabolites and mitochondrial damage through, for example, oxidative stress (Box 1). The notion that mitochondrial dysfunction can lead to fatigue is indicated by the observation that fatigue is a common symptom of mitochondrial disease (Gorman et al., ) and that fatigue-like behavior in animal models is associated with reduced expression of mitochondrial complexes, reductions in metabolic activity, and alterations in mitochondrial morphology in the brain (Surapaneni et al.,; Wang et al.,; Vichaya et al., ).

20Healthy mitochondrial functioning depends on a multitude of factors. Mitochondria themselves produce reactive oxygen species (ROS) as a byproduct of metabolism. In normal conditions, antioxidants are sufficient at protecting against damage from ROS, and the cell remains in equilibrium. However, if the balance skews toward ROS production during excessive and/or dysfunctional mitochondrial activity, ROS can induce the production of several toxins and can damage the cell; this is referred to as oxidative stress.

21Most patient studies on biomarkers of mitochondrial function have focused on circulating levels of coenzyme Q10 (CoQ10) or L-carnitine. CoQ10 is a component of the electron transport chain of the mitochondria and is thus important for mitochondrial energy production. It is assumed that CoQ10 serum levels mirror levels in the mitochondria, although to the best or our knowledge this has not been formally studied. The antioxidant properties of CoQ10 make it an anti-inflammatory agent as well as an important mitochondrial protectant (Genova et al.,; Sanoobar et al., ). In vitro and animal studies have indeed shown a beneficial effect of CoQ10 on mitochondria (Bergamini et al.,; Jiménez-Santos et al., ). L-carnitine can be either made by the body or acquired in the diet. L-carnitine is important for transport of fatty acids across the mitochondrial membrane for β-oxidation and adenosine triphosphate production. Thus, reductions in L-carnitine indicate reduced availability of fatty acids in the mitochondria and reduced tricarboxylic acid activity (Porter et al., ).

22The association between mitochondrial dysfunction and persistent fatigue has been most widely studied in patients with CFS (Table 2). A systematic review reported that 21 of 25 papers were restricted to CFS (Filler et al., ). Lower serum levels of the antioxidant CoQ10 was the most consistent finding of the review, with limited evidence for lower levels of other antioxidants and increased levels of oxidative stress markers. Lower carnitine levels were observed in some studies, but results depended on the type of carnitine assessed. Later published studies on CFS confirm increased oxidative stress levels, possible decreases in ATP production efficiency, and reductions in mitochondrial energy production in patients with CFS (Ciregia et al.,; Tomas et al., ). The assessment of oxygen consumption in PBMCs by Tomas et al. is promising, indicating a reduced capacity of mitochondria to meet increased energy demands during cellular stress, a replication is warranted based on the use of both frozen and fresh blood samples and the limited sample size.

23Studies on CRF have thus far been limited to the immediate effects of cancer treatment, and it is unclear to what extent these effects are still present into survivorship. Preclinical research has shown that both cancer and cancer therapy are associated with mitochondrial dysfunction (Dumas et al.,; Tzika et al.,; Luo et al.,; Gouspillou et al.,; Sridharan et al.,; Tabassum et al.,; Chiu et al.,; Gilliam et al.,; Vichaya et al., ). It is unknown from preclinical models whether treatment-induced or cancer-induced changes in mitochondrial function are associated with fatigue. However, a handful of observational patient studies from one research group report an association in prostate cancer patients undergoing radiation therapy. Their results point to reduced mitochondrial functioning in association with fatigue, as evidenced by reduced expression of mitochondria-related genes (Hsiao et al.,, ), decreased expression of mitochondrial electron transport complex II enzymes (Filler et al., ), and increased apoliprotein A1 (ApoA1) in association with higher fatigue (Lukkahatai et al., ). ApoA1 is important for lipid scavenging, and an increase in ApoA1 is indicative of a reduced use of lipids for ATP synthesis.

24A handful of studies and clinical trials corroborate the above-described observational findings. Two experimental studies in patients with CFS showed alterations in metabolic processes. Results of in vitro stimulation of skeletal muscle cells from patients with CFS were indicative of insulin resistance or decreased sensitivity of insulin receptors (Brown et al., ). Further, patients with CFS had a slower recovery from an initial maximal exercise test, leading to a more rapid reliance on less-efficient glycolytic metabolism during a subsequent test (Snell et al., ).

25Antioxidant supplementation can be effective in reducing fatigue, suggesting a causal relationship between reduced availability of antioxidants and fatigue. In patients with CFS, supplementation with CoQ10 and nicotinamide adenine dinucleotide (NADH) reduced fatigue, whereas a placebo treatment did not (Castro-Marrero et al., ). Supplementation has also been shown to improve CRF. Cachexic patients with advanced-stage gynecological cancer seemed to benefit from the addition of several metabolism-related supplements to their standard cachexia treatment with synthetic progestogen, showing decreases in fatigue, inflammation, and resting-state energy expenditure as compared to standard treatment alone (MacCiò et al., ). In breast cancer patients undergoing chemotherapy, supplementation with an amino-acid jelly that contained CoQ10 and L-carnitine led to less-severe fatigue (Iwase et al., ). As this study included a standard-of-care condition as control, a placebo effect cannot be ruled out. In contrast, a large study of breast cancer patients undergoing chemotherapy using supplementation of vitamin E with or without CoQ10 did not find any effect of the added CoQ10 on fatigue development (Lesser et al., ). In animal models, it has also been shown that various agents with antioxidant properties can reduce fatigue-like behavior (Singh et al.,, Davis et al.,; Fu et al.,; Zhuang et al., ). For these studies, it should be noted that the models of fatigue pose a significant limitation. As the mechanisms underlying CFS are yet ill-defined, the validity of animal models of CFS are questionable. Further, many of these studies induce and quantify fatigue in the same intervention (e.g., forced swim), which is inherently problematic. Therefore, much work is needed to improve our models systems and validate these findings.

26In sum, most evidence for an association between fatigue and mitochondrial functioning comes from CFS, indicating lower levels of antioxidants and possible reductions in mitochondrial ATP production. While lower antioxidant levels were not found in cancer patients (Filler et al., ), alterations in mitochondrial gene expression do indicate a role for mitochondrial functioning in fatigue. Results from clinical and pre-clinical studies point toward possible beneficial effects of mitochondria-supporting supplements. However, the majority of observational and intervention studies suffer from severe limitations in either sample size or study design. In addition, the exact causes for reduced antioxidant levels or ATP production have not been studied. While inflammation is a likely cause, it is definitely not the only candidate. Thus, additional preclinical research to identify mechanisms, as well as clinical replications in larger samples using placebo controlled-designs and relevant biomarkers as output, are urgently needed.

27Metabolomic studies allow for a broad assessment of alterations in metabolism (Table 1). Three such studies have been reported in CFS, each with different results, but all pointing toward reduced metabolic activity. Yamano et al. showed evidence for decreased activity in the TCA cycle and the urea cycle (Yamano et al., ). Naviaux reported downregulated metabolites of two classes of lipids in their CFS group (Naviaux et al., ) and Fluge reported reduced concentrations of the amino acids that act as precursors for acetyl-Coenzyme A, one of the primary inputs of the TCA cycle, in CFS patients (Fluge et al., ). While Yamano et al. and Naviaux et al. did not report on associations between metabolites and fatigue severity, Fluge et al. reported that there were no associations. Rather, they observed associations with age and body mass index, suggesting that their findings were not specific for fatigue in the CFS patients.

28There has been limited work on metabolomics in animal models of fatigue; however, in one study of exercise-induced fatigue, several potential “antifatigue” metabolomic biomarkers were identified (Ma et al., ). One was geranyl diphosphate, which can indirectly regulate lipid synthesis and protein degradation. Sebacic acid, a consequence of fatty acid metabolism that can be oxidized into metabolic intermediates for the TCA cycle, was another.

29In summary, results from the three metabolomic studies point to alterations in lipid and fatty acid metabolism and decreased TCA activity in relation to fatigue, which resemble metabolic changes during chronic low-grade inflammation (Figure 3). As the TCA cycle is needed to create the precursors for the oxidative phosphorylation process, decreased activity would indicate reduced ATP production via reduced oxidative phosphorylation. As lipid metabolism, or lipolysis, generates the most energy per gram of substrate, it is the most efficient manner of energy storage. As noted in Figure 2, the catabolism of lipids to usable energy is a slow process and mainly utilized during extended energy need, such as chronic inflammation. Thus, alterations in this metabolic pathway might result in reduced availability of efficient energy sources, creating an increased reliance on carbohydrate-based metabolism. In addition, reduced fatty acid availability would interfere with aerobic energy production, increasing the need for anaerobic glycolytic ATP production.

30Several findings point to behavioral-energy expenditure exceeding energy availability in patients with CFS and CRF. For example, one study reported that about half of a group of patients with CFS had perceived energy expenditure levels that exceeded their perceived energy availability (Jason et al., ). In addition, a non-pharmacological intervention aimed at decreasing fatigue in patients with CFS was effective only in those patients in whom perceived energy availability and expenditure were matched (Brown et al., ). In CRF, we have recently shown that cancer survivors reporting more-severe fatigue exhibited an increased tendency to exert effort (Lacourt et al., under review). Cancer patients actively undergoing treatment did not show this association, but rather showed a decreased inclination for exerting effort. Similarly, Mortimer et al. reported significant positive associations between fatigue severity and average daily caloric expenditure in breast cancer patients after the fourth cycle of chemotherapy. Thus, there is preliminary evidence suggesting that energy expenditure does not match (perceived) energy availability.

31As mentioned above, low-grade inflammation-induced changes in metabolism can be mediated by changes in circadian rhythms and sleep, possibly through changes in melatonin rhythmicity (Box 2). Below, we discuss the findings on sleep, circadian rhythm, and melatonin in relation to chronic fatigue (see also Table 3).

32Melatonin moderates both sleep and circadian rhythm and an important biomarker for sleep and circadian rhythm. Melatonin (N-acetyl-5-methoxytryptamine) is a hormone which moderates both sleep and circadian rhythms and is specifically known for its importance in sleep onset. Melatonin concentrations show a circadian pattern with slowly increasing levels during the day, leading to maximal levels at the time of sleep onset, followed by a slow decrease until it reaches a minimum level in the morning when it is time to wake up. In addition to its function for sleep, melatonin can also act as an antioxidant and immune regulator, playing a role in mitochondrial DNA protection (Ramis et al., ). Studies in rodents confirm that inflammation and stress are capable of modulating melatonin levels (Persengiev et al.,; Huang et al., ) and that melatonin can reduce inflammation and oxidative stress associated with sleep deprivation (Kim et al.,; Zhang et al., ). However, melatonin administration may be insufficient to restore sleep-wake rhythmicity (Mirmiran and Pevet, ).

33In CFS, poor sleep and altered melatonin rhythmicity, but not alterations in circadian rhythmicity have been reported. Patients with CFS report more sleep disturbances (Russell et al., ) and display abnormal sleep progression (i.e., greater cyclic alternating pattern rate) (Guilleminault et al., ). In addition, fatigue severity was associated with patient-reported poor sleep (Milrad et al., ), which was in turn associated with minor increases in plasma levels of the proinflammatory cytokines interleukin-1β and tumor necrosis factor-α. While no evidence has been found for altered circadian rhythm in CFS (Hamilos et al.,; Rahman et al., ), there is evidence for altered melatonin rhythmicity. For example, significantly higher nocturnal salivary melatonin levels were seen in adolescents with CFS who report unrefreshing sleep compared to healthy subjects (Knook et al., ). Further, the association between body temperature circadian rhythm and melatonin onset observed in healthy individuals was absent in patients with CFS (Williams et al., ). Although treatment with melatonin or phototherapy (daylight therapy) did not alleviate fatigue in CFS patients in one study (Williams et al., ), melatonin treatment was successful in reducing fatigue when participants were selected for a later-than-usual evening-melatonin onset (van Heukelom et al., ). The latter study did not include a placebo-control but fatigue reductions were significantly more pronounced in patients with very late onset vs. relatively early onset, suggesting that the effects are not likely explained by placebo effect alone.

34In cancer patients, associations between CRF and disrupted circadian rhythm and sleep have been reported before, during, and after cancer treatment (Miaskowski et al.,; Payne,; Ancoli-Israel et al., ). In a longitudinal study of breast cancer patients, increases in fatigue during chemotherapy were related to increased reports of sleep disturbances as well as increased nap time and decreased wake time during the day (Liu et al., ). In the subset of these patients, for whom information on inflammation was available, increases in either fatigue or disturbed sleep were associated with increased inflammation (Liu et al., ). Interventions aimed at normalizing circadian rhythm have proven successful in alleviating CRF. Bright white light therapy was effective in both preventing and treating circadian-rhythm desynchronization in patients with breast cancer (Neikrug et al., ) and improved fatigue in breast cancer survivors (Redd et al., ) and in a mixed sample of nonmetastatic cancer survivors (Johnson et al., ). Light treatment during chemotherapy for breast cancer was shown to prevent increases in fatigue, although surprisingly, these effects were not mediated by concurrent effects on circadian rhythm (Ancoli-Israel et al., ).

35In summary, disturbed sleep and altered circadian or melatonin rhythmicity occur frequently among individuals with persistent fatigue, although persistent fatigue can be experienced in absence of sleep and circadian-rhythm alterations. Only a few studies have assessed the associations between low-grade inflammation and sleep or circadian rhythm in the context of fatigue, and the results are inconsistent. In addition, associations between sleep or circadian rhythm and metabolic outcomes have not been studied in fatigued populations.

36We have here proposed a model of persistent fatigue as a consequence of chronic low-grade inflammation leading to an imbalance in energy availability and expenditure, which can be mediated and maintained by changes in circadian rhythms and sleep. The here discussed literature indeed points to associations between persistent fatigue in CFS and CRF with alterations in cellular metabolism, disturbed sleep, and, to a lesser extent, disruptions in circadian rhythm. For both CRF and CFS, multi-causal models are generally suggested (Papadopoulos and Clear,; Moss-Morris et al.,; Bower, ); not only including low-grade inflammation and disturbed sleep, but also alterations in stress physiological responses, genetic vulnerability, and sociodemographic factors, among others. These causes are not mutually exclusive and feasibly point to a more limited number of final end points. We have here described such a final common pathway, incorporating several recognized contributors to persistent fatigue.

37There are several limitations to the current state of knowledge. First, most studies have been carried out in patients with CFS patients, and this is especially the case for studies on metabolism and energy production. It is thus far unclear to what extent findings on metabolism and energy production also pertain to patients with CRF. Further, few reports are available that included more than one of the suggested parameters of our proposed model. Only two studies explored associations between low-grade inflammation and sleep, one in CFS patients and one in fatigued breast cancer patients. To the best of our knowledge, no studies have been published that describe associations between inflammation and metabolism in patients with persistent fatigue. Despite this lack of more-encompassing studies, the evidence thus far suggests that reduced cellular-energy availability might play a role in chronic fatigue when the fatigue is associated with low-grade inflammation. In addition, there is limited evidence for increased behavioral-energy expenditure suggesting that energy expenditure is not adapted to the level of fatigue or of energy availability.

38Intervention studies aimed at improving aspects of the model presented, with the ultimate goal to alleviate fatigue, are still scarce. Nevertheless, it is already apparent that participants for such studies should be selected carefully and not solely on the basis of their fatigue experience. This is nicely illustrated by the study by van Heukelom et al., in which careful selection of individuals with CFS with later-than-usual evening-melatonin onset revealed significant beneficial effects of melatonin supplementation. Such effects were not observed when participants were selected solely on the basis of their fatigued state (Williams et al., ).

39In sum, more research is needed to piece together the puzzle explaining how chronic low-grade inflammation can lead to the experience of persistent fatigue. Nevertheless, reduced cellular energy paired with increased or maladaptive changes in energy expenditure poses a potentially important explanation. Evidence for this model can be found for both low-grade inflammation and fatigue.

40TL, EV, and CH: contributed to conception of the model proposed in the review; TL: wrote the first draft of the manuscript; EV and GC: wrote sections of the manuscript; RD and CH: reviewed the manuscript at several stages and contributed to fine-tuning of the ideas presented in the review. All authors contributed to manuscript revision, read, and approved the submitted version.

41The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.