Conceptualising the neurobiology of fatigue

Introduction

Fatigue is a common symptom reported by patients. Its assessment depends on the ability to distinguish subjective feelings of fatigue from other ‘inner states’ such as tiredness, sleepiness, exhaustion, feeling stressed, anhedonia, lassitude, fatigability, weariness, depressed mood, lack of drive, psychomotor retardation and muscle weakness. The ability to separate these feelings meaningfully and use the appropriate semantics will inevitably differ significantly between subjects and also across various cultures and languages. Indeed, in many languages (e.g. French and English), fatigue is a common expression with daily colloquial usage, but in others, such as German, it is only used as a technical term in psychology and medicine. Therefore, the term fatigue lacks specificity and its use within psychopathological and diagnostic classifications would be more useful if the underlying pathophysiology was known or it had predictive value – for example, concerning treatment response. Thus, it is important to better define fatigue and to this end after briefly summarising findings regarding its putative neurobiology we posit in this Viewpoint that fatigue should be partitioned into two subtypes, namely: ‘hypoaroused fatigue’ – associated with a propensity for increased daytime sleep, as found in the context of inflammatory or immunological processes and ‘hyperaroused fatigue’ – characterised by difficulties in falling asleep, as in the case of major depressive disorder (MDD).

Neurobiological correlates of fatigue

Chronic fatigue occurs in the context of idiopathic chronic diseases (such as chronic fatigue syndrome (CFS) and fibromyalgia) or a variety of other diseases such as cancer, rheumatoid arthritis, multiple sclerosis, chronic infection, sleep disorders or stroke. Fatigue is also an integral part of the symptomatology of MDD (Demyttenaere et al., 2005), as well as a precursor and a residual symptom of this prevalent disorder. ‘Reduction of energy’, which is closely related to fatigue, is indeed one of the three core items of MDD in the International Classification of Diseases (ICD)-10.

Given the broad spectrum of disorders associated with fatigue, it is no surprise that a common underlying neurobiology has not been found. Even in conditions such as CFS, which are characterised by an enduring state of disabling fatigue (> 6 months), identifying the pathophysiology of fatigue has proven difficult.

Abnormalities of the hypothalamic-pituitary-adrenal (HPA) axis

One of the more consistent findings in CFS is hypocortisolism and/or blunted responsivity of the HPA axis (Silverman et al., 2010), especially in CFS patients with traumatic childhood experiences (Heim et al., 2009); however, to what extent such hypo-cortisolism is caused by fatigue per se is still a matter of considerable debate. If HPA axis abnormalities are found to be causal these could provide potential targets for therapeutic interventions (Cleare et al., 1999; McKenzie et al., 1998), but equally they could be consequential to sleep problems, physical de-conditioning and other factors associated with CFS (Cleare, 2004). However, another unknown is how generalisable the HPA axis findings in CFS are to fatigue in other severe somatic disorders. Regarding fatigue that occurs as part of clinical depression (MDD), it would seem reasonable to assume that the nature of fatigue is different to that in CFS. Specifically, instead of hypocortisolism there is hypercortisolism along with other signs of HPA axis hyperactivity (Pariante and Lightman, 2008).

Inflammatory and immunological processes

There is also a large body of literature indicating an important role of inflammatory, post-inflammatory and immunological factors in the pathophysiology of different fatigue syndromes, including CFS. Fatigue is part of the common sickness behaviour occurring together with bacterial or viral infections and triggered by proinflammatory cytokines such as interleukin (IL)-1α and β, IL-6 and tumour necrosis factor α (TNF-α). Vagal afferents, macrophage-like cells in circumventricular organs and endothelial cells of brain vessels have been shown to play an important role in immune-to-brain communications and in inducing sickness behaviour. The possible role of such immunological processes in depression has become a dynamic area of research that has been fuelled by findings of elevated levels of circulating proinflammatory cytokines such as TNF-α (Himmerich et al., 2008) in depression, and the fact that repeated injections of IL-2 or interferon-α can induce initially sickness behaviour, including fatigue, and later depressive syndromes (Arnett et al., 2011).

In CFS it has recently been proposed (Arnett et al., 2011) that there is deficient early infection control that results in chronic inflammatory responses to infectious agents (for review, see Arnett et al., 2011). Indeed, the hypoactive HPA axis found in many CFS patients (Silverman et al., 2010) is associated with immunosuppressive effects that are thought to provoke an overproduction of inflammatory cytokines such as IL-1β and lead to fatigue (Arnett et al., 2011).

A plausible link between depression-related fatigue and immunological processes is provided by the findings that proinflammatory cytokines (TNF-α, interferon-gamma) reduce the bioavailability of tryptophan, which is the precursor of serotonin (Müller and Schwarz, 2007). In vulnerable subjects, tryptophan depletion can induce depression (Neumeister, 2003) and in cancer patients treated with cytokines a drastic decrease of plasma tryptophan has been shown to occur that is associated with high depression scores at 4 weeks of treatment (Capuron et al., 2002).

Summating thus far, fatigue in general cannot be linked to a common underlying neurobiology. In part this no doubt reflects the clinical heterogeneity of fatigue and its lack of specificity as a descriptive term. However, neurobiologically, two distinct subtypes of fatigue can be conceptualised, as discussed below.

Hypo- versus hyperaroused fatigue

A distinction relevant for specifying fatigue concerns regulation of wakefulness and vigilance (‘brain arousal’). Fatigue and tiredness can be associated with increased daytime sleepiness (unstable vigilance regulation), but also with the opposite, a reduced sleep propensity (hyperstable vigilance regulation), as typically found in MDD. Patients with MDD are not sleepy, but exhausted and describe a state with high inner tension. Traditionally, psychopathology makes the clinically useful distinction between lack of drive (anergic state) and inhibition of drive (retardation), the latter being more typically observed in MDD (for review, see Hohl-Radke, 2008). We therefore suggest that hypo- and hyperaroused fatigue differ, not only phenomenologically, but also across other parameters, including treatment response (Table 1).

Table 1.

Proposed features distinguishing between two types of fatigue according to the regulation of vigilance (‘brain arousal’).

Features	Hypoaroused fatigue	Hyperaroused fatigue
Related disorders (Demyttenaere et al., 2005)	Cancer (Olbrich et al., 2012)	Major depressive disorder
	Immunological disorders (Silverman et al., 2010)
	Postinfectional fatigue (Silverman et al., 2010)
	CFS (Neu et al., 2008)
Drive	Lack of drive (anergic state)	Inhibition of drive (retardation)
Type of fatigue	Sleepy	Exhausted, high inner tension
Wakefulness regulation	Unstable (Olbrich et al., 2012), short sleep latency	Hyperstable (Hegerl et al., 2012), prolonged sleep latency
HPA axis activity	Decreased	Increased
Positive treatment response	Psychostimulants	Antidepressants (Shen et al., 2011)

CFS: chronic fatigue syndrome; HPA: hypothalamic-pituitary-adrenal axis.

‘Hypoaroused fatigue’

Wakefulness regulation can be objectively measured. The best established, but time-consuming instrument is the Multiple Sleep Latency Test (MSLT), which measures iteratively the sleep onset latency. A newly developed and validated instrument for assessing vigilance stages and their regulation is the Vigilance Algorithm Leipzig (VIGALL). Within a 15-minute EEG (electroencephalogram) recording under quiet rest, transitions are assessed from high vigilance stages (vigilance stage 0) during full wakefulness to relaxed wakefulness (stages A1–A3) and proceeding on to lower vigilance stages that are associated with increasing drowsiness (B1–B3) until the onset of sleep (stage C) (Günther et al., 2011; Olbrich et al., 2011, 2012).

Using VIGALL in cancer-related fatigue, an unstable vigilance regulation is found. Patients show more frequent and earlier lower vigilance stages (Figure 1) and drop to vigilance stages associated with drowsiness and sleep (vigilance stages B2, B3 and C) more so than healthy controls (Olbrich et al., 2012). Using the MSLT, an increased sleep propensity with reduced sleep latencies was also found in CFS (Neu et al., 2008; Spitzer and Broadman, 2010) and in fatigue with immunological disorders (Iaboni et al., 2006).

Figure 1.

EEG vigilance regulation in patients with major depressive disorder (MDD) and patients with cancer-related fatigue syndrome. (a) Mean proportion of the low vigilance stages (stages B2/3 and sleep stage C) in the resting EEG of 30 unmedicated patients suffering from MDD in comparison to 30 healthy age- and sex-matched subjects (see Hegerl et al., 2012). Differences between patients with MDD and controls were tested by Mann–Whitney tests: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. (b) Statistically significant differences between patients with cancer-related fatigue syndrome (CRF) and healthy controls (HC) in the time it took them to attain low vigilance stages (stages B2/3) (see Olbrich et al., 2012). In patients with cancer-related fatigue, stages B2/3 were significantly more frequently present than in healthy controls, as opposed to patients with MDD. (min = minutes.)

For these fatigue syndromes with unstable vigilance and wakefulness regulation, drugs with wakefulness stabilising properties like psychostimulants seem to be a better option as indicated by several studies and meta-analyses of randomised controlled trials (Minton et al., 2008, 2010; Olbrich et al., 2012). Specific effects on cancer-related fatigue have also been shown for modafinil in a placebo-controlled randomised controlled trial involving over 600 patients (Jean-Pierre et al., 2010).

‘Hyperaroused fatigue’

The fatigue and tiredness reported by patients with typical MDD is not associated with sleepiness and increased sleep propensity. On the contrary, MDD is associated with prolonged sleep onset latencies at night (Hubain et al., 2006) and normal daytime sleep propensity in the MSLT (Nofzinger et al., 1991). Using VIGALL for assessing vigilance regulation in 30 unmedicated patients with MDD compared to healthy controls, a hyperstable wakefulness regulation with higher proportions of high vigilance stages and fewer declines to low vigilance stages was found over a 15-minute recording period (Hegerl et al., 2012) (Figure 1). High inner tension and inhibition of drive (not lack of drive), as reported by patients with MDD (Hohl-Radke, 2008), and signs of a hyperactive HPA axis with hypercortisolism (Pariante and Lightman, 2008) are in line with this dysregulation of vigilance.

Such patients are unlikely to be responders to psychostimulants. Indeed, several placebo-controlled studies with psychostimulants and also more recently with modafinil failed to show clear evidence for an antidepressant effect in patients with MDD (for review, see Candy et al., 2008). Furthermore, the meta-analysis by Candy et al. (2008) did not demonstrate significant antidepressant effects with respect to response rates. Only when decline in a validated rating scale for depression was used as a primary outcome measure did the significant superiority of psychostimulants to placebo regarding short-term antidepressant effects emerge. However, two of these trials focused on patients with secondary depressive syndromes in the context of traumatic brain injury and HIV with hypersomnia (for details, see Candy et al., 2008). These secondary depressive syndromes are likely to correspond to anergic states with sleepiness and lack of drive (‘hypoaroused fatigue’) and not to typical MDD. First-line treatment options for fatigue in the context of depression are antidepressants and even sedating antidepressants have been found to be beneficial (Shen et al., 2011). In this context it is of interest that nearly all antidepressants and electroconvulsive therapy reliably reduce the tonic as well as phasic firing rate of neurons in the locus coeruleus (West et al., 2009). The high noradrenergic activity is thought to be a pathogenetic factor in MDD that mediates anhedonia, inhibition of behavioural activation (retardation) and has wakefulness-promoting and arousing effects (Stone et al., 2011). A vigilance regulation model of affective disorders linking the high vigilance stages associated with withdrawal and ‘sensation avoidance’ in depression to response to antidepressants, and the unstable vigilance regulation found in mania and attention deficit hyperactivity disorder (ADHD) to response to psychostimulants has been presented elsewhere (Hegerl and Hensch, 2012).

In summary, emerging evidence supports the empirical distinction between hypoaroused fatigue (sleepiness, lack of drive) that is characteristic of CFS (Neu et al., 2008; Spitzer and Broadman, 2010), cancer-related fatigue (Olbrich et al., 2012) and fatigue with systemic lupus erythematosus (Iaboni et al., 2006), and hyperaroused fatigue (reduced sleep propensity, exhaustion and inhibition of drive) as found in MDD. The heuristic value of this distinction lies in the generation of a testable hypothesis that hypo- and hyperaroused fatigue respond differentially to treatment.