Introduction

Attention-deficit/hyperactivity disorder as one of the most common psychiatric conditions in childhood and adolescence affects about 5 % of the young population (Polanczyk et al. 2007) and furthermore shows a distinct tendency to persist throughout the lifespan (50–60 %; Fayyad et al. 2007). Symptoms from the domains of inattention, motor hyperactivity and maladaptive impulsivity as well as emotional dysregulation constitute a serious obstacle to academic and professional accomplishment and the establishment of a stable and supportive social environment (Barkley et al. 2006; Hoza et al. 2005). ADHD presents with a high rate of comorbidity such as oppositional defiant disorder, conduct disorder, substance related or affective disorders (IMpACT et al. 2012; Wilens et al. 2009; The MTA Cooperative Group 1999), and it possesses a strong genetic component (Faraone et al. 2005). Genes from a wide range of neurotransmitter and other systems have been found to be associated with ADHD (Arcos-Burgos et al. 2010; Gizer et al. 2009); however, endeavours into identifying the exact genetic pathomechanisms are complicated by the fact that phenotypical presentations are very heterogeneous, and the high heritability is best accounted for by a large number of genes exerting small effects and in sum bringing out the phenotype.

Numerous abnormalities in brain anatomy have been shown in ADHD, e.g. volume reductions in dorsolateral prefrontal cortex, striatum, corpus callosum and cerebellum for childhood ADHD (Cao et al. 2010; Castellanos and Tannock 2002; Hutchinson et al. 2008; Seidman et al. 2005; Valera et al. 2007) or decreased volumes of dorsolateral and orbitofrontal prefrontal cortex and anterior cingulate cortex for adult ADHD (Hesslinger et al. 2002; Schneider et al. 2006; Seidman et al. 2006). Deviations in developmental cortical thickness (Makris et al. 2007; Shaw et al. 2006, 2007, 2009) and altered structural and functional connectivity are also observed in ADHD (for a review, see Liston et al. 2011).

The aim of this paper was to present the vigilance regulation model of ADHD and to discuss it in the context of earlier arousal dysregulation theories of ADHD. The vigilance regulation model of ADHD interprets the hyperactivity and sensation seeking of patients with this disorder as an autoregulatory attempt at stabilizing vigilance (vigilance in the sense of central nervous arousal) by creating a stimulating environment. Before elaborating on this model, an overview of current neurobiological models of ADHD with a focus on those relating ADHD to physiological hypoarousal will be provided.

Former and recent concepts of hypoarousal in ADHD

With regard to putative neurobiological bases of an arousal deficit in ADHD, Sikstrom and Soderlund (2007) attributed central hypoarousal to a dopamine deficiency along the lines of the dopamine deficit theory (Levy 1991; for a discussion regarding the validity of the theory, see Gonon 2009). According to this theory, as a consequence of low tonic dopamine levels and the subsequent up-regulation of autoreceptors, the phasic dopamine response is exaggerated, making patients both chronically underaroused and phasically hypersensitive to environmental stimulation. Other authors suspected structural alterations in regions involved in modulating arousal (e.g. formatio reticularis; Halperin and Schulz 2006) to be disruptive to vigilance regulation in ADHD. Additionally, the vigilance deficit in ADHD might be rooted in reported altered low-frequency connectivity (Fair et al. 2010; Tian et al. 2008) between regions of the default mode network (DMN; Raichle et al. 2001; Sonuga-Barke and Castellanos 2007), a distributed set of brain regions in frontal (DLPFC, inferior frontal cortex), parietal (precuneus, inferior parietal cortex), temporal (inferior temporal gyrus, amygdala) and medial regions, which is active during resting condition and associated with idling. Lapses of attention owing to intrusions of this network could manifest in behavioural variability and unstable vigilance.

The role of physiological hypoarousal in the aetiopathogenesis of ADHD was first commented upon by Satterfield and Dawson (1971). These authors, and subsequent studies (Barry et al. 2009, 2012; Broyd et al. 2005; Conzelmann et al. 2014; Hermens et al. 2004; Lazzaro et al. 1999), illustrated a lower general skin conductance level (SCL)—an established indicator of autonomic arousal—in ADHD at rest.

Several resting EEG studies have also suggested ADHD-associated hypoarousal by demonstrating excess of slow frequencies, especially increased theta, and reduced faster alpha and beta activity (Barry et al. 2003, 2009). The increased ratio of theta-to-beta activity during resting conditions has received extensive attention as a putative marker of the disorder (indicative of hypoarousal), in both childhood (Monastra et al. 1999, 2001; Clarke et al. 2001a, b, 2002) and adult ADHD (Bresnahan et al. 1999; Bresnahan and Barry 2002). Interestingly, EEG profiles point to different subgroups of ADHD. For example, a subgroup with an excess in beta activity was identified, which was suspected to be hyperaroused (Clarke et al. 2013). To further clarify arousal in ADHD, Clarke et al. (2013) additionally recorded SCL as a measure of autonomic arousal, and unexpectedly found the excess theta and excess beta groups were both characterized by low SCL, indicating hypoarousal in both groups. It was therefore concluded that the theta/beta ratio, although strongly associated with ADHD and its subgroups, may not indicate arousal, but possibly task-related activation (Barry et al., 2009; Clarke et al. 2013). Arousal had previously been shown to be more linked to alpha activity, which appears to be equally diminished in both the excess beta and excess theta ADHD group (Clarke et al. 2013), so regardless of the proportion of beta and theta activity, EEG studies support the assumption of hypoarousal in ADHD.

Sharing the assumption of low central nervous arousal, subsequently emerging theories reflect different assumptions about the pathway from hypoarousal to phenotype. According to the optimal stimulation theory (Zentall and Zentall 1983), ADHD patients actively seek out individually appropriate environmental input by widening their attentional focus through expansive behaviour (motor hyperactivity) and short devotion of attention to a large range of stimuli (distractibility, inattention) to compensate for an innate hypoarousal, qualifying behavioural symptoms as a method for self-stimulation and sensation seeking.

On a more cognitive level regarding the management of central nervous arousal, Lubar (1991) suspected patients to show fundamental impairments in summoning the necessary amount of arousal required for the completion of a given task (Lubar 1991). Electrophysiologically, this would manifest in an inability to shift into a task mode with the associated levels of alpha and beta activity required for focused reactions. The cognitive-energetic model (CEM, Killeen et al. 2013; Sergeant 2000, 2005) took up the importance of arousal on ADHD presentation, however, without assuming a trait over- or underarousal for ADHD. Instead they postulated deficits in the three interlinked energetic pools—arousal, activation and effort—to be at the heart of ADHD symptoms. Arousal in this context refers to a response to a specific stimulus, governed chiefly by amygdala and formatio reticularis. Fine-tuning Sergeant’s model, Sonuga-Barke et al. (2010) argue that not physiological arousal per se is affected, but instead the management of arousal into task-related activation is malfunctioning in ADHD (Sonuga-Barke et al. 2010). One possibility is pathologically altered activation as a subcomponent of arousal, and ADHD deficits are due to an unstable activation level that is more variable and less reliable across contexts. Under favourable circumstances, patients manage to muster a task-adequate level of activation at the cost of the executive control system continuously requesting allocation of resources and in sum investing a disproportionate amount of effort. As a consequence, motivation to invest that effort especially under monotonous conditions will diminish. Alternatively, if effortful control and allocation of activation in response to external stimulation is impaired, arousal and activation are fundamentally functional but their potential might not be used to the fullest, and suboptimal environmental stimulation levels cannot be countered successfully. Crucially, the relationship between arousal and performance follows an inverted U-shape, where only medium levels of arousal provide suitable conditions and overarousal as well as underarousal interferes with performance particularly in complex tasks.

In 2007, Sonuga-Barke and Castellanos came forward with the default mode interference hypothesis, attributing the unstable performance and high reaction time variability consistently observed in ADHD on periodic intrusions of the default mode network (Sonuga-Barke and Castellanos 2007). As mentioned above, Tian et al. (2008) uncovered increased low-frequency connectivity of those resting state associated brain areas in ADHD. Furthermore, patients deactivate the DMN to a lesser extent during inhibition or working memory tasks, and those deviations can be successfully rescued by methylphenidate administration (Fassbender et al. 2009; Peterson et al. 2009). Since an increase in autonomic arousal correlates with a deactivation of the task negative (default mode) and engagement of the task positive network (Fan et al. 2012), this could corroborate the notion of hypoarousal as a core characteristic of ADHD pathophysiology (Table 1).

Table 1 Theories of arousal as pathogenetic factor in ADHD
Full size table

Assessment of vigilance (arousal) by EEG

As outlined above, hypoarousal in ADHD has been documented for many years by measuring SCL and quantitative EEG (QEEG). The hypothesis of hypoarousal contributing to ADHD is picked up and expanded upon by the recently introduced vigilance regulation model of affective disorders and ADHD. In this context, an EEG-based algorithm has been developed and validated, which allows classifying EEG segments into different vigilance stages, and therefore provides an objective tool to study the regulation of CNS arousal (Hegerl and Hensch 2012; Hegerl et al. 2010).

Within the frame of the vigilance regulation model, the term vigilance is used as a synonym for CNS arousal. Thus, the term vigilance refers to different global functional brain states, as they can be discerned with the EEG. Vigilance in this context is not understood in terms of sustained attention, which is normally inferred from performance-based variables such as reaction times or errors in “vigilance tests” (Oken et al. 2006).

Vigilance regulation describes the adjustment of vigilance to situational demands, such as a required increase in cases of novelty or danger or when a task is to be performed. In contrast, a decrease of vigilance is required in situations associated with rest (e.g. going to bed). EEG-assessed vigilance regulation shows stable inter-individual differences (Van Dongen et al. 2004): under resting conditions without major external stimulations, most subjects show a gradual decline to lower vigilance stages associated with drowsiness or sleep, within 20 min. Some subjects, however, exhibit an unstable vigilance regulation with rapid declines to low vigilance stages after a very short time, whereas others display a hyperstable vigilance regulation without such declines to lower vigilance stages, even after recording periods of more than 20 min.

In order to assess the tendency towards a vigilance decline, subjects are required to lie on an upholstered couch in a dimly lit and sound-attenuated room with their eyes closed for 15–20 min. The procedure is similar to the multiple sleep latency test (MSLT, Carskadon and Dement 1977; Sullivan and Kushida 2008) insofar as the MSLT also involves recording a 20-min eyes-closed resting EEG. Unfortunately, the MSLT has a time-consuming testing protocol, due to the fact that the EEG recordings have to be repeated several times. The most important difference between the MSLT and the vigilance regulation paradigm is, however, that the MSLT only scores the EEG-defined sleep onset, ignoring all the fluctuations of vigilance before sleep onset.

The recently developed Vigilance Algorithm Leipzig (VIGALL) allows to objectively assess fine-graded vigilance fluctuations before sleep onset (Hegerl and Hensch 2012). VIGALL is based on previous studies illustrating EEG changes in the transition between wakefulness and sleep onset (Benca et al. 1999; Bente 1964; Cantero et al. 2002; Corsi-Cabrera et al. 2000; De Gennaro and Ferrara 2003; De Gennaro et al. 2001, 2004, 2005; Kaida et al. 2006; Marzano et al. 2007; Roth 1961; Strijkstra et al. 2003; Tsuno et al. 2002) and takes into account the spatiotemporal pattern of the EEG activity using EEG source localization approaches and additionally the electrooculogram (EOG).

VIGALL automatically attributes to short EEG segments (1 s by default) one of the following seven vigilance stages (Hegerl and Hensch 2012; Olbrich et al. 2009, 2011):

  • Stage 0 is characterized by low amplitude, non-alpha activity and is observed during an activated state, such as mental effort with high alertness.

  • Stage A demonstrates dominant alpha activity corresponding to relaxed wakefulness. Stages A is further subdivided into substages A1, A2, A3, according to the degree of spreading of alpha activity from occipital to more anterior cortices.

  • Stage B shows low amplitude non-alpha (substage B1) and increasing theta and delta activity including vertex waves as correlates of drowsiness (substage B2/3).

  • Stage C is defined by sleep spindles or K-complexes characterizing sleep onset.

Figure 1 shows the course of vigilance within the EEG recordings (20 min) of an unstable (a) and a hyperstable (b) subject.

Fig. 1
figure 1

Course of vigilance within the EEG recordings (20 min) in a subject with an unstable vigilance regulation (a) and a subject with a hyperstable vigilance regulation (b). On the left, the vigilance stages are plotted for every 1-s segment in a VIGALLogram. EEG epochs containing artefacts are indicated by grey lines. Additionally, on the right, the time series of vigilance stages are plotted by assigning a numerical value to each of the vigilance stages (from “1” for the lowest stage “C” to “7” for the highest vigilance stage “0”). Then a simple moving average (SMA) is calculated for each second including the 15 preceding and the 15 subsequent seconds

Full size image

The EEG is known to show a high heritability and intraindividual stability (De Gennaro et al. 2008; Landolt 2008; van Beijsterveldt et al. 1996) and, at the same time, a large interindividual variability. To account for the latter, VIGALL performs individual adaptation procedures, e.g. concerning several amplitude and frequency parameters. The VIGALL is not suited for subjects with the rarely occurring alpha variant rhythms, i.e. rhythms most prominently recorded over posterior regions, which resemble reactivity of the alpha rhythm, but differ in frequency (theta or beta instead of alpha range; Noachtar et al. 1999). Additionally, VIGALL is not to be used in subjects with major morbogenic (e.g. Alzheimer’s disease, confusional states) or pharmacogenic (e.g. anticholinergic drugs) modifications of the EEG.

More details about VIGALL and the standard operating procedures (SOPs) can be found in the manual, which can freely be downloaded at www.uni-leipzig.de/~vigall. The free software is also available on the website for download and assessment. The current version VIGALL v2.0 is implemented as an add-in in the Brain Vision Analyzer 2 software (subject to licence, Brain Products, Gilching, Germany).

Using VIGALL in combination with the vigilance regulation paradigm has several advantages over former QEEG studies:

  1. 1.

    VIGALL does not only estimate mean vigilance of one QEEG segment of arbitrary length, but takes into account the dynamic of vigilance regulation. Vigilance stages are assigned to 1-s EEG segments, which allows for the assessment of the dynamics of vigilance during the 15–20-min EEG recordings with high temporal resolution. Important temporal aspects, such as the stability of vigilance over time, are not normally assessed by QEEG studies on ADHD, which focus only on short EEG recordings of, for instance, 2 min (Snyder and Hall 2006).

  2. 2.

    The vigilance approach provides a framework for meaningful data reduction and facilitates the interpretation of EEG signals. In addition to the already mentioned adaptation of the algorithm to individual amplitude and frequency characteristics, VIGALL also reduces problems arising from the fact that certain frequency patterns can occur both during high and low vigilance stages. For example, low-voltage EEG can be seen in very different vigilance states: slight drowsiness (VIGALL stage B1) and focused wakefulness during intense mental activity (VIGALL stage 0). VIGALL discriminates both stages by additionally detecting slow eye movements (SEM), which only occur in B1.

  3. 3.

    Theta and alpha both not only indicate arousal, but are also considered to reflect cognitive processes, for example during a working memory task (Loo and Makeig 2012; Klimesch 1999). Depending on experimental set-up, this might confound the assessment of arousal. In former QEEG studies, there is a worrying lack of standardization concerning subjects’ cognitive engagement: in some studies no task was given, whereas in others, tasks such as reading, drawing or listening during EEG recording were applied (Snyder and Hall 2006). It is beyond the scope of this review to provide an overview regarding which kinds of tasks might interfere with vigilance assessment by producing cognitively induced frequency bands. Eliminating this potential confound, the vigilance regulation paradigm can expected to be largely free of cognitively induced frequency bands, since it does not include a cognitive task and participants are expressly instructed to permit their natural course of vigilance. Although this does not exclude the risk of uncontrolled cognitive processes (e.g. rumination) taking place while the subject is supposed to be resting, it is plausible to assume that not presenting any cognitive task will at least reduce the impact of cognition-related frequencies, as these have been reported to occur in cognitive tasks compared with such no-task resting conditions (Klimesch 1999).

  4. 4.

    Some former QEEG studies were recorded with eyes open, and some with eyes closed, which can further complicate interpretation of alpha activity: opening of the eyes leads to alpha reduction (desynchronization). However, with increased drowsiness, this alpha reduction is dampened (Michimori et al. 1994). Thus, directly following eyes opening, higher alpha activity indicates lower arousal. In the vigilance regulation paradigm, after Berger Manoeuvres at the beginning, eyes are closed throughout the whole recording session, thereby reducing difficulties in interpreting alpha activity.

Unstable vigilance as pathogenetic factor in ADHD

Vigilance and compensatory behaviour

The recently introduced vigilance regulation model of affective disorders and ADHD (Hegerl and Hensch 2012; Hegerl et al. 2009, 2010) suggests that the vigilance level is not only influenced by the environment, but that we may also produce a more or less “arousing” environment by our behaviour. In an autoregulatory manner, a more or less stimulating environment might be actively created in order to stabilize or reduce vigilance levels. Besides exposing the subject to arousing stimuli, the behaviour itself (such as fidgeting) may have arousing effects. A good non-pathological example for such an autoregulatory attempt to stabilize vigilance by increasing stimulation may be overtired children, who often develop a hyperactive, talkative and sensation-seeking behaviour.

Similar concepts of vigilance and autoregulatory behaviour have already been proposed earlier, including their potential relevance for affective disorders (Bente 1964; Ulrich 1994) and ADHD (Weinberg and Brumback 1990; Zentall and Zentall 1983). In the same way, related concepts have been presented concerning personality traits such as sensation seeking (Zuckerman 1979) and extraversion (Eysenck 1990). These personality traits were also interpreted as autoregulatory behaviour in order to achieve an optimal level of arousal and were suggested to reflect vulnerability to affective disorders and ADHD (Hensch et al. 2007; White 1999).

In the following, arguments supporting the vigilance regulation model of ADHD and affective disorders will be presented.

Unstable vigilance in ADHD

As reported above, several QEEG studies indicate an unstable vigilance regulation in ADHD (Barry et al. 2003, 2009). Furthermore, stronger daytime sleepiness has been found in ADHD using the MSLT in two studies (Golan et al. 2004; Lecendreux et al. 2000) and a third small study (Palm et al. 1992) comprising of children “with deficits in attention, motor control and perception” (no diagnosis according DSM was given in this study). A fourth study by Prihodova et al. (2010) could not confirm significantly higher overall sleepiness in ADHD. However, the authors found shorter sleep latency at the first and third but not the other MSLT sessions in ADHD. These findings were interpreted by the authors as a possible sign of dysregulated arousal. Recently, further data in support for lower vigilance in ADHD have been provided using VIGALL (Sander et al. 2010). Complementing these EEG studies indicative of an unstable vigilance in ADHD, subjective sleepiness was also reported to be higher in ADHD, and sleepiness was associated with symptom severity (Cortese et al. 2009; Gamble et al. 2013; Yoon et al. 2012).

This unstable vigilance resulting in attention deficits with or without compensatory behaviour may have different causes: a subgroup of ADHD patients may be characterized by unstable vigilance as an inherent trait (e.g. due to abnormalities in neurotransmitter systems responsible for arousal regulation). Other subgroups may present ADHD symptoms due to hypoarousal, which is secondary to sleep and circadian disorders, such as sleep-disordered breathing (SDB), restless legs syndrome (RLS), periodic limb movements and circadian phase delay. Sleep and circadian disorders are highly prevalent in ADHD. About 30 % of children and 60–80 % of adults with ADHD exhibit symptoms of sleep disorders (Cortese et al. 2009; Yoon et al. 2012). It should be noted that many of these studies were done in unmedicated patients; therefore, the disturbed sleep cannot be explained by stimulant use alone. In contrast, some studies even point to a possible improvement of sleep quality in ADHD treated with stimulants (Kooij et al. 2001; Sobanski et al. 2008). To illustrate the central role and high prevalence of sleep problems in ADHD, symptoms of disturbed sleep were initially even included in the DSM as diagnostic criteria, but later dropped due to being non-specific (Gruber et al. 2009). Nonetheless, it has been discussed that sleep and circadian disorders might play a pathogenetic role in ADHD symptomatology (Arns and Kenemans 2012; Beebe 2011; Yoon et al. 2012). Experimental studies on sleep restriction demonstrated induction or aggravation of some ADHD symptoms by sleep reduction (Fallone et al. 2005; Gruber et al. 2011).

An unstable vigilance regulation provides an explanation for the attention deficits in ADHD, especially the well-documented deficits in continuous performance tasks (Nichols and Waschbusch 2004). Furthermore, the vigilance model of ADHD can also explain the ADHD presentation specifiers according to DSM-V and the ADHD subtypes as their predecessors in the DSM-IV-TR (see Fig. 2). In the predominantly inattentive presentation (formerly named predominantly inattentive subtype), the deficits are explained by the instability of the vigilance regulation (Fig. 2a). In the combined presentation (subtype) with attention deficits and hyperactivity, additional autoregulatory aspects come into play with hyperactivity, sensation and novelty seeking as an attempt to stabilize vigilance (Fig. 2b).

Fig. 2
figure 2

Unstable vigilance trait explaining ADHD presentation without (a) and with (b) vigilance stabilization syndrome (predominantly inattentive presentation vs. combined presentation, respectively). The attention deficits become especially evident in monotonous low-stimulus situations

Full size image

The presented model is also capable of explaining why studies across the board not only report substantially lower prevalence rates for the predominantly hyperactive-impulsive subtype but also generally call into question the general validity of this subtype (Hurtig et al. 2007; Willcutt et al. 2012): the unstable vigilance is suggested to be a core pathogenetic factor, which results in attention deficits. In this line of thought, hyperactivity, in contrast, does not represent a primary disorder per se, but an autoregulatory response, which may or may not be present. Following this aetiological consideration, a “pure” hyperactive subgroup should not exist. It should be noted that according the DSM-IV-TR and DSM-V, the predominantly hyperactive subtype/presentation actually is not “pure”, but allows substantial and impairing subthreshold attentional deficits to co-occur. According to DSM-IV-TR/V, the predominantly hyperactive subtype can be diagnosed even in the presence of substantial attentional deficits, as long as the hyperactive symptoms reach the criterion while the sum of attention-related problems does not reach the critical number of six.

Unstable vigilance as common factor in ADHD and mania

Mania and ADHD show striking similarities at the symptom level, and ADHD and bipolar disorder (BD) display high comorbidity. In adults, the rate of comorbid ADHD in bipolar patients was 9–35 % (Bernardi et al. 2012; Klassen et al. 2010; McIntyre et al. 2010). In paediatric BD, however, comorbidity with ADHD is a matter of debate with estimates varying substantially; it was estimated that up to 98 % of children with BD also have ADHD, and up to 22 % of children with ADHD have BD (Faraone et al. 2012; Klassen et al. 2010). However, in Europe, consistently low rates of paediatric bipolar disorder are reported, owing to the fact that precursor symptoms are generally unspecific (Holtmann et al. 2007, 2010). Particularly with regard to the controversies about the diagnosis of paediatric bipolar disorders, a discussion about the validity of these comorbidity figures is ongoing (Youngstrom et al. 2010). Family studies can contribute to elucidate the question of a real comorbidity of both diseases: Faraone et al. (2012) conducted a meta-analysis of family genetic studies of ADHD and bipolar I patients. The authors found a significantly higher prevalence of bipolar I among relatives of ADHD patients, and a significantly higher prevalence of ADHD among relatives of bipolar I patients. Most importantly, the relative risk for bipolar disorder was very similar for adult relatives (2.2) as for child relatives (2.1). This suggests that the link between bipolar disorder and ADHD cannot be explained alone by misdiagnosis of bipolarity in children.

These and other findings suggest that ADHD and BD share not only similarity on the symptom level, but also possibly common underlying pathomechanisms (as reviewed in Hegerl et al. 2010). Unstable vigilance regulation is supposed to be such a common pathogenetic mechanism. Surprisingly, during manic episodes patients typically also show an unstable vigilance regulation, which at a first glance seems to be in contrast to their highly energetic behaviour. However, when studied in a quiet environment with low external stimulation and eyes closed, manic patients often show rapid declines to low vigilance stages and microsleeps with sleep spindles already within the first minute of EEG recording (Schoenknecht et al. 2010; Small et al. 1999; Ulrich 1994; Van Sweden 1986). Several lines of evidence suggest that this unstable vigilance is not only a consequence of sleep deficits on account of mania, but plays a causal role in the pathogenesis of mania.

Numerous studies show that factors causing sleep deficits are among the strongest triggers of mania and contribute to the exacerbation of manic symptoms (Harvey 2008; Wehr 1992). Sleep deprivation in bipolar depression can induce hypomania and mania (Colombo et al. 1999; Kasper and Wehr 1992; Wu and Bunney 1990), and sleep deprivation has also been proposed as a promising method in the search for an animal model of mania (Benedetti et al. 2008; Gessa et al. 1995). The causal relevance of sleep reduction (e.g. due to bereavement, newborn infants, shift work, travel, obstructive sleep apnoea) for triggering mania was reviewed by Plante and Winkelman (2008). Furthermore, with the exception of bupropion (Moreira 2011), all antidepressants can impair vigilance (Bull et al. 2002; Cascade et al. 2009; O’Hanlon et al. 1998; Riedel et al. 2005) and are associated with an increased risk of switching to mania in bipolar patients. Compared to SSRI treatment, the switch rate is higher in those antidepressants with particularly sedating properties due to additional anticholinergic and antihistaminic properties, such as tricyclic antidepressants or trazodone (Gijsman et al. 2004; Jabeen and Fisher 1991; Peet 1994; Terao 1993). While sleep deficits can trigger mania, stabilization of sleep-wake rhythm is an important element in behavioural therapies for bipolar disorder (Frank et al. 2005; Leibenluft and Suppes 1999; Riemann et al. 2002). Accordingly, extended bed rest and darkness as an add-on to the usual treatment of acute mania resulted in a faster reduction of symptoms (Barbini et al. 2005; Nowlin-Finch et al. 1994; Wehr et al. 1998).

To conclude, there are several lines of evidence supporting the hypothesis that both in ADHD and mania unstable vigilance regulation may result in compensatory autoregulatory behaviour contributing to the overlapping phenotypes. Both disorders, however, differ concerning the trait/state characteristic of the unstable vigilance regulation. In ADHD, the unstable vigilance is supposed to be a stable trait, be it acquired, genetic or resulting from sleep disorders. In contrast, in affective disorders the unstable vigilance appears to be associated with the manic disease state. In mania, an exaggerated autoregulatory behaviour is thought to override the physiological tendency to seek sleep, thus aggravating the sleep deficits, and as a consequence, the instability of vigilance. A vicious circle is started, which then contributes to full-blown mania.

For the sake of completeness, it should be mentioned that according to first data, a typical unipolar depressive episode is characterized by the opposite, compared with ADHD and mania, namely a hyperstable vigilance (Hegerl et al. 2011; Olbrich et al. 2012). Again, the vigilance regulation model is also able to explain certain symptoms and treatment effects in depressive episodes: the tendency to withdraw and to avoid loud music, social interactions and other external stimulations, which can be observed within a typical depressive episode, could also be interpreted as autoregulatory behaviour. Factors which further stabilize vigilance (increase arousal), such as long sleep or early time of the day, further aggravate the depression (Wilk and Hegerl 2010), whereas therapeutics, which reduce vigilance, such as sleep restriction (Benedetti and Colombo 2011), anticholinergic drugs (Drevets and Furey 2010) and antidepressants, ameliorate depression (see for details Hegerl and Hensch 2012). It is of interest in this context that in preclinical studies, it is a consistent finding that all standard antidepressants and also electroconvulsive therapy reduce the firing rate of neurons in the noradrenergic locus coeruleus (West et al. 2009), which plays a pivotal role in vigilance regulation (see Hegerl and Hensch 2012).

The vigilance regulation model and stimulant treatment

Stimulants in ADHD

The therapeutic effects of different classes of psychostimulants in ADHD are well established. Stimulants strongly reduce attention deficits, sensation-seeking behaviour, and hyperactivity in patients with ADHD (Pietrzak et al. 2006; Riccio et al. 2001; Spencer et al. 2005) with the first effects of medication usually manifesting rapidly (within 30–45 min for methylphenidate, Greydanus et al. 2007). The reduction of EEG slow wave activity under treatment with psychostimulants is in line with their vigilance stabilizing properties (Bresnahan et al. 2006). Thus, the therapeutic effects of stimulants in ADHD could partly be explained by their vigilance stabilizing effects, which could interrupt the autoregulatory hyperactivity and sensation-seeking behaviour.

Stimulants in mania as new treatment option?

In mania, stimulants are considered as contraindicated by many clinicians. However, this is not based on empirical data. A reanalysis of randomized clinical trials with psychostimulants in ADHD by the Food and Drug Administration (FDA) revealed that psychotic or manic-like reactions occurred rarely (in about one of 400 treated patients), and in the majority of cases (55 of 60), the symptoms resolved within 2 days (Gelperin and Phelan 2006; Mosholder 2006; Phelan 2006a, b; Ross 2006). Additionally, in a controlled trial in children with ADHD and severe mood dysregulation, an improvement according to the Young Mania Rating Scale was observed under methylphenidate (Waxmonsky et al. 2008). Stimulants as an add-on to mood stabilizers have already been prescribed to bipolar patients. In children and adolescents, open trials (Kowatch et al. 2003; Kummer and Teixeira 2008) and three recent controlled trials (Findling et al. 2007; Scheffer et al. 2005; Zeni et al. 2009) showed that adding a stimulant did not worsen, but often improved manic symptomatology. Mirroring these findings, in adults neither uncontrolled studies (Carlson et al. 2004; El-Mallakh 2000; Fernandes and Petty 2003; Lydon and El-Mallakh 2006; Nasr et al. 2006) nor two controlled trials (Calabrese et al. 2010; Frye et al. 2007) could detect a greater risk for (hypo)manic symptoms in bipolar depressed patients treated with stimulants as add-on to mood stabilizers.

There is more evidence suggesting that psychostimulants might be beneficial not only in patients with ADHD but also in those with mania. According to the vigilance model presented here, vigilance stabilizing drugs should be able to interrupt the vicious circle of mania. Indeed, rapid antimanic effects of psychostimulants have quite consistently been shown in case reports and case series. A substantial improvement of manic symptoms within 1 h was observed in these studies, comparable to the effects in treatment of ADHD. For example, Beckmann and Heinemann (1976) showed a pronounced decrease in mania ratings by the end of a 30-min d-amphetamine infusion in all of the six manic patients. Garvey et al. (1987) reported a reduction of at least 50 % on a mania scale in five of six patients, rated by clinicians blind to the d-amphetamine treatment. Schoenknecht et al. (2010) also reported a rapid response of an acutely manic patient to monotherapy with the vigilance stabilizing drug modafinil. Taken together, the converging evidence that methylphenidate is a safe treatment option in bipolar patients and possibly also possesses antimanic effects was strong enough to start an international randomized placebo-controlled clinical trial on this intriguing question (Kluge et al. 2013; EudraCT: 2010-023992-24; ClinicalTrials.gov: NCT01541605).

Treatment response prediction by EEG vigilance regulation?

ADHD is a heterogeneous disease and accordingly treatment response varies between patients. EEG vigilance may help to predict who is likely to profit from stimulant treatment. EEG vigilance regulation might also be able to predict possible antimanic effects of stimulants in mania. A first case report by Bschor et al. (2001) points in this direction. Improvement of manic symptoms was shown about 2 h after intake of methylphenidate in a manic patient with signs of unstable EEG vigilance regulation. In contrast, no improvement was found in another manic patient without unstable EEG vigilance.

Summary

ADHD is a highly persistent psychiatric disorder with an early onset in childhood and adolescence. In the last decades, there have been several attempts to explain the symptomatology of this disorder by neurophysiological hypoarousal as assessed with SCL or EEG. Expanding upon these early neurophysiological arousal concepts of ADHD are newer findings using structural and functional brain imaging. For example, disturbed functional connectivity, e.g. in the default mode network, might lead to intrusions of this rest-associated network and result in the frequent lapses in attention and high variability of reactions ubiquitously reported in ADHD populations. Building on this body of research, the vigilance regulation model of ADHD and affective disorders has recently been proposed. Research in this area is greatly facilitated by a newly developed and validated EEG-based algorithm (VIGALL) which allows for classifying 1-s EEG segments into different vigilance stages and objectively assessing the regulation of vigilance during a 15–20-min EEG recording under quiet resting conditions with eyes closed. According to the vigilance regulation model, the hyperactivity and sensation-seeking behaviour observed in ADHD and mania can be interpreted as an autoregulatory attempt of the organism to create a stimulating environment in order to stabilize vigilance. According to this model, unstable vigilance is a basic dysfunction in ADHD, which always results in attention deficits especially in monotonous sustained attention tasks. A subgroup of patients develops the autoregulatory hyperactive and sensation-seeking behaviour in order to stabilize vigilance. Here, hyperactivity merely constitutes a secondary reaction to the unstable vigilance and is not expected to exist as an independent deficit in ADHD. In this sense, attention deficits are mandatory, whereas hyperactivity and impulsiveness are optional in ADHD. Mania and ADHD are characterized by symptom overlap and vigilance instability. There is mounting evidence that psychostimulants might be safe and effective in the treatment of acute mania. An international randomized placebo-controlled trial on the antimanic efficacy of methylphenidate has recently been started and will give first evidence whether or not this may be a new treatment option in mania. The vigilance regulation is a biomarker whose predictive and diagnostic validity concerning ADHD has to be further assessed in future studies.