Epidemiology of Comorbid ADHD and Epilepsy

Attention-deficit/hyperactivity disorder (ADHD) is a common comorbidity of epilepsy, causing a significant disruption in quality of life (Ettinger et al., 2015). ADHD occurs in up to 50% of children and up to 20% of adults with epilepsy, in contrast to 7%–9% of children and 2.5%–4% of adults in the general population (Williams et al., 2016). The epidemiology also shifts in terms of ADHD subtype and gender predominance. The three subtypes of ADHD, according to the Diagnostic and Statistical Manual of Mental Disorders (5th edition; DSM-V), are inattentive, hyperactive, and combined phenotypes (American Psychiatric Association and American Psychiatric Association, 2013). There is a disproportionately higher distribution of the inattentive subtype in patients with epilepsy compared to the general population (24%–52% vs. 3%–11%, respectively). In addition, males and females with epilepsy are equally affected by ADHD, whereas in the general population, ADHD is more commonly diagnosed in males (Lax-Pericall et al., 2019). The etiology of ADHD is unlikely due directly to seizures or antiseizure medications (ASMs) since the diagnosis of ADHD often precedes the diagnosis of epilepsy (Crunelli et al., 2020). Additionally, resolution of seizures in absence epilepsy, for example, has no significant effect on the prevalence of ADHD (Masur et al., 2013). Some genetic epilepsy syndromes, including absence epilepsy, fragile X syndrome, and tuberous sclerosis, have a particularly high association with ADHD, suggesting some degree of genetic predisposition to both epilepsy and ADHD.

Genetics of ADHD and Epilepsy

Despite the common occurrence of both ADHD and epilepsy, there is an overall fairly weak and nonsignificant genetic correlation between epilepsy (both focal and generalized) and ADHD (Brainstorm Consortium et al., 2018), suggesting a likely multifactorial etiology, including polygenic and environmental contributions in the majority of cases. However, a number of epilepsy syndromes are associated with candidate genes, which have a relatively higher prevalence of ADHD, suggesting a shared genetic predisposition (Table 58–1). The associated genes are often directly involved with cellular neurophysiology and synaptic transmission, including Scn1a, Chrna7, Slc6a1, SYNGAP1, GABRA1, Cacna1a, and SNAP25 (Lo-Castro and Curatolo, 2014). Of note, genes specifically regulating dopaminergic neurotransmission (e.g., Slc6a3, DRD4) have been associated with ADHD in the general population (Hawi et al., 2015). While one study found that polymorphisms in dopamine transporters and dopamine receptors were significantly associated with the severity and family history of epilepsy, there was no difference between the prevalence of these polymorphisms in patients with epilepsy and in healthy controls (Alcantara et al., 2018). Altogether, these studies indicate that the predisposition to having comorbid epilepsy and ADHD is likely polygenic, with additional influence from environmental factors.

Table 58–1

Epilepsy Syndromes Associated with ADHD.

Testing Attention Deficits in Rodents

Given a plethora of rodent models of epilepsy, behavioral testing of these models to identify comorbid attention deficits provides an opportunity to dissect and isolate the pathophysiological changes that may lead to the development of seizures, attention deficits, or both. The type of attention being tested can vary between sustained attention, the ability to maintain focus over long periods of time; selective attention, the ability to avoid distractions while performing a task; and attentional set shifting, the ability to shift attention on a task from one dimension to another. ADHD is classically associated with deficits in sustained attention, with relative preservation of selective attention (DeShazo Barry et al., 2001). However, patients with epilepsy and comorbid attention deficits may have varying degrees of dysfunction in both sustained and selective attention (Lee et al., 2018). Two of the most common tests that aim to isolate specific deficits in attention are the Attentional Set-Shifting task (ASST—primarily testing attentional set shifting but also indirectly tests selective attention at the final stage) and the 5-Choice Serial Reaction Time task (5-CSRTT—primarily testing sustained attention but can also be used to test selective attention and impulsivity; Fig. 58–1; Bushnell and Strupp, 2009). In both tasks, rodents are food deprived to ~85%–90% of their baseline body weight to ensure they are motivated to perform the task.

Figure 58–1.

Attention testing in rodents. A. In the Attentional Set-Shifting task, rodents learn to discriminate the location of a reward based on sensory cues along two different dimensions—here Dimension 1 is the digging medium (e.g., paper vs. felt, represented (more...)

The goal of the ASST is to assess the flexibility in employing attention, mediated by frontal lobe structures (Heisler et al., 2015; Tait et al., 2014). This task is analogous to the Wisconsin Card Sorting Task performed in the clinical setting. After first learning to discriminate by one dimension alone (simple discrimination), the rodent then learns to discriminate in the presence of an irrelevant dimension (compound discrimination). In Figure 58–1, for example, the relevant dimension is the digging medium (e.g., paper, felt, ribbon) and the irrelevant dimension is odor (e.g., cinnamon, lemon, vanilla). Next, the location of the reward is reversed, and the rodent must learn to associate the reward with the alternate item along the relevant dimension. There is then an intra-dimensional shift (IDS), where two new items in the same relevant dimension (digging medium) are presented, and the subject must learn to associate one of the new relevant items with the reward. The IDS can be repeated (twice in this example) before another reversal along the same relevant dimension is presented. Up until this point, the relevant dimension has always been the same, and the subject has been trained to continuously ignore the irrelevant dimension (odor). The final stage is an extra-dimensional shift (EDS), where the subject must now selectively pay attention to the previously irrelevant dimension (odor) and ignore the previously relevant dimension (digging media). A greater number of trials to find the correct reward at the EDS stage indicates cognitive rigidity, with a deficit in shifting selective attention. Recent advances using in vivo two-photon calcium imaging in prefrontal cortex of mice while performing an attention set-shifting task has confirmed that prefrontal cortex neurons enable attentional set-shifting by encoding feedback from recent trial outcomes (Spellman et al., 2021).

The 5-CSRTT generally takes a longer time for rodents to master than the ASST, but it can be used to assess multiple facets of attention, including sustained attention and impulsivity/response inhibition, in addition to selective attention (Bari et al., 2008). It was modeled after the human continuous performance task (CPT) (Robbins, 2002). After learning to associate a reward with a light stimulus in one of five holes on a curved wall, the duration of that stimulus is gradually reduced to a duration as brief as 0.5 seconds. The task requires the subject to sustain endogenous (non-stimulus-induced) attention in anticipation of the stimulus to be able to identify which hole is briefly illuminated. Sustained attention is measured by evaluating the relative proportion of correct, incorrect, and omission responses. Impulsivity and response inhibition are also measured by evaluating the number of perseverative and premature responses. Finally, selective attention can also be measured by assessing accuracy in the setting of distractors, such as irrelevant auditory cues. Because of these advantages, the 5-CSRTT has been the most widely used test for assessing measures of attention in rodents (Higgins and Silenieks, 2017).

Rodent Models of ADHD and Epilepsy/Seizures

In the last decade, these behavioral measures of attention have been applied to a multitude of seizure and epilepsy models, which range from exposure to chemoconvulsive agents to monogenic models of epilepsy (Table 58–2), which give progressively greater insight into the cell-type-specific pathophysiological development of comorbid epilepsy and ADHD.

Table 58–2

Rodent Models of Comorbid Epilepsy and ADHD.

The intrahippocampal administration of picrotoxin (PTX, a GABA_A receptor antagonist) at subconvulsive doses in rats was found to result in significantly reduced accuracy on the 5-CSRTT when compared with administration of saline (McGarrity et al., 2017). These results indicate that acute disinhibition in the hippocampus may result in remote cortically mediated deficits in attention through well-described hippocampal-prefrontal connectivity, but the generalizability of these results to patients with epilepsy is limited by the lack of seizure activity in this model.

Another study evaluated the performance of rats on the 5-CSRTT using the lithium-pilocarpine model of temporal lobe epilepsy (Faure et al., 2014). These rats have spontaneous, recurrent seizures and represent a well-validated model of patients with temporal lobe epilepsy (Turski et al., 1989). They found that epileptic rats had reduced accuracy, and neuronal loss in the hippocampus and frontal lobe significantly correlated with attention deficits. Only anti-choline-acetyl transferase (ChAT) and antineuronal nuclear marker (NeuN) antibodies were used to identify neurons, so a more detailed cell-type-specific dissection of neuronal loss was not performed. Another limitation of this study is that electroencephalogram (EEG) was not recorded during the attention testing, so it is not clear if the poor performance may have been influenced by the presence of nonmotor seizure activity. Nonetheless, reduced attention performance was reproduced by another laboratory around the same time with the same model (lithium-pilocarpine) using a similar task (Pineda et al., 2014), so this model provides a reliable substrate to screen the effects of novel therapies against both epileptic seizures and attention performance.

The genetic absence epilepsy rat from Strasbourg (GAERS) model of childhood absence epilepsy has also undergone attention testing with the 5-CSRTT (Marques-Carneiro et al., 2016). This inbred line was tested against two controls (nonepileptic controls [NECs] inbred to suppress seizure activity, and the Wistar rat line from which the original GAERS line was derived). The GAERS rats had fewer correct responses than the NECs, but not the Wistar line. These findings suggest the attention deficits may be subtle in rodent models of absence epilepsy, but there are important limitations to this study, including a lack of littermate controls, no evaluation of seizure burden before or during tasks, and limited insight into the pathophysiology of epilepsy-related attention deficits.

Finally, a growing number of monogenic models of epilepsy have been evaluated with various degrees of attention testing. Monogenic mutants have the benefit of comparing behavior against littermate controls and greater insight into the pathophysiology underlying phenotypes. If the mutation results in an abnormal phenotype, the cause and effect are established a priori; however, primary changes must be carefully evaluated alongside the possibility of indirect or compensatory changes secondary to the mutation. Monogenic models of three ADHD and epilepsy related genes—SNAP25, Scn1a, and Grm7—have been generated and tested for both seizures and abnormal behavior. Mice heterozygous for the SNAP25 mutation have frequent interictal epileptiform discharges, a lowered threshold for kainate-induced seizures, a hyperactive phenotype, and deficits in associative learning tasks (Corradini et al., 2014; Ohira et al., 2013). However, these mice did not have spontaneous seizures, and attention was not specifically tested. Similarly, in the Scn1a rat model, heat-induced seizures, hyperactivity, and spatial learning deficits were all present, but attention testing was not specifically performed (Ohmori et al., 2014). Finally, the Grm7 homozygous knockout mouse model, which normal encodes the metabotropic glutamate receptor 7, demonstrated handling-induced seizures, and the closely related Elfn1 knockout mouse model demonstrated hyperactivity (Dolan and Mitchell, 2013; Fisher et al., 2020; Tomioka et al., 2014). Both mutations have been found in human cohorts with epilepsy and ADHD, but again, attention was not specifically tested in any of these studies. Without spontaneous seizures or specific deficits in attention, the translational utility of each of these models is limited. However, they all point to some degree of inhibitory neuron dysfunction: SNAP25 has an important role in regulating the development of MGE-derived (parvalbumin/PV- and somatostatin/SST-expressing) interneurons (Pai et al., 2020); Scn1a mutations cause dysfunction in both PV- and SST-expressing interneurons (Tai et al., 2014); and Elfn1 causes loss of trafficking of mGluR7 specifically in SST-expressing interneurons (Tomioka et al., 2014).

To test the specific effect of an epilepsy and ADHD-related genetic mutation on interneuron subtypes, a recent study selectively deleted one copy of Cacna1a from PV-interneurons (PV^Cre;Cacna1a^+/–) and then evaluated the mice for spontaneous seizures and selective attention deficits using the ASST (Lupien-Meilleur et al., 2021). While PV^Cre;Cacna1a^–/– mice have severe generalized epilepsy, PV^Cre;Cacna1a^+/– have a reduced threshold for pentylenetetrazol (PTZ)-induced seizures but no spontaneous seizures. Attention testing with the ASST was performed and revealed a specific deficit in the final stage associated with the extradimensional shift, indicating cognitive rigidity and a deficit in regulating selective attention. These findings were reproduced with regional expression of PV^Cre;Cacna1a^+/– in the medial prefrontal cortex, lending further evidence for interneuron dysfunction as a potential etiology underlying both epilepsy and comorbid attention deficits. Of note, PV-interneuron dysfunction need not be limited to neocortical regions; there are additional links between PV-interneuron dysfunction in the reticular thalamic nucleus (RTN), absence epilepsy, and attention deficits that have also been extensively reviewed elsewhere (Fogerson and Huguenard, 2016).

One of the most extensively studied models of comorbid epilepsy and ADHD is the FMR1 mutant model (FMR1^–/–) of Fragile X syndrome, first described in mice and then more recently in rats. The mouse model has reliable audiogenic seizures, which presumably arise from subcortical regions since cortical EEG recordings during audiogenic seizures typically show no epileptiform activity (Maxson, 2017). FMR1^–/– mice exhibit hyperactivity, which is improved with STEP (STriatal-Enriched protein tyrosine Phosphatase) inhibition (Chatterjee et al., 2018). The mechanism for STEP-inhibition mediated improvement is postulated through reduction of abnormally elevate spine density on neocortical pyramidal cells, but its effect on inhibitory neurons was not investigated. These mice were also previously shown to have impaired attentional set formation using the ASST (Casten et al., 2011). In addition, a rat model of FMR1^–/– has also been generated, which showed significantly increased omissions during the 5-CSRTT, but audiogenic seizures were not able to be induced. Similar to the cell-type-specific dissection of Cacna1a mutations, an effort has been made to understand the role of specific cell types with the FMR1 mutation. In one study, in vivo two-photon imaging of primary visual cortex in FMR1^–/– mice was performed during a Go/No-Go task, in which mice learn to lick a lickometer with one stimulus (S1) but withhold licking with a different stimulus (S2) (Goel et al., 2018). This is a frontal lobe-mediated task requiring executive function, assessing impulsivity (when there is a lick after S2) and deficits in sustained attention (when there is no lick after S1). FMR1^–/– mice showed significant impulsivity and a nonsignificant trend toward a reduction in sustained attention. This was associated with reduced evoked responses in PV-interneurons, and the impulsivity normalized with targeted chemogenetic activation of these interneurons. One major concern about the translational validity of this model remains that audiogenic seizures are not associated with cortical epileptiform activity, so its generalizability as a seizure model for human epilepsy is somewhat diminished.

While data thus far have largely implicated dysfunction in inhibitory neurons, especially PV-interneurons, more work is still necessary to find an ideal model of comorbid epilepsy and attention deficits, which would include (1) a genetic mutation that has been described in patients with both epilepsy and attention deficits, (2) specific testing of the multiple facets of attention (e.g., using the 5-CSRTT), (3) simultaneous EEG recording to disentangle the relationship between seizures and task-based attention, and (4) cell-type-specific imaging and manipulation to identify potential insights for novel therapeutic strategies based on targeting the underlying pathology.

The Effect of Pharmacotherapy for ADHD on Seizures

In the absence of contraindications such as cardiovascular risk factors and history of drug abuse, the first-line therapy for ADHD is stimulant medications. Typical stimulants include methylphenidate (Ritalin) and amphetamine (Adderall), which increase synaptic levels of dopamine and norepinephrine by blocking their reuptake. These neuromodulators generally act with an inverted-U relationship with attention—with either too little or too much monoaminergic activity leading to inattention (Cools and D’Esposito, 2011; Datta et al., 2019). Similarly, dopamine and norepinephrine have both been linked to both proepileptic and antiepileptic effects, depending on the receptor involved (Akyuz et al., 2021). For example, animal studies have shown that D1-receptor signaling is generally proepileptogenic while D2-receptor signaling is generally antiepileptogenic (Bozzi and Borrelli, 2013). Although historically there has been a theoretical concern that stimulants may lower seizure threshold, most clinical evidence and international guidelines support stimulant therapy as both safe and effective in patients with epilepsy (Auvin et al., 2018; Leeman-Markowski et al., 2021). However, large, randomized, clinical trials are still needed to confirm this notion. The effect of stimulant use on seizures in rodent models of epilepsy has not been extensively tested. However, in the Scn1a rat model of epilepsy, methylphenidate not only improved spatial learning and reduced hyperactivity, but it also reduced the threshold for heat-induced seizures (Ohmori et al., 2014). Attention metrics, however, were not specifically tested. Further evaluation of stimulant therapy in rodent models of epilepsy would help to understand further the potential risks and benefits for seizures and attention deficits, respectively.

Atomoxetine (Straterra) is a nonstimulant medication, which selectively blocks the reuptake of norepinephrine (Zhang et al., 2017). It is typically not as effective as stimulant therapy, so it is reserved for those with potential contraindications or intolerance to stimulants. Because of the relationship between norepinephrine, seizure-induced respiratory arrest, and sudden unexpected death in epilepsy (SUDEP), a series of studies have investigated the potential for the use of atomoxetine to prevent SUDEP in the setting of audiogenic and maximal electroshock mouse models (Kruse et al., 2019; Zhang et al., 2017). These studies showed that atomoxetine does not exacerbate seizures and may have a promising protective effect against SUDEP—but further study is warranted to understand the underlying mechanism of this effect.

Treatment of Concurrent Seizures and Attention Disorders

There is currently no ASM that clinically improves attention deficits, and there is no treatment for attention deficits that clinically improve seizures. This observation suggests that there is an underlying pathology from which seizures and ADHD independently arise, such that treatment of one has no effect on the other because the underlying pathology remains untreated. For example, in both the stargazer and tottering mouse models of absence epilepsy, ethosuximide effectively treats seizures (Maheshwari et al., 2016). However, alterations of phase-amplitude coupling (PAC) in the background EEG of these mice remain aberrant even after seizures have been treated (Maheshwari et al., 2017). These background alterations may represent a biomarker for the underlying pathology that remains untreated by ethosuximide. It is tempting to speculate that these retained alterations may be an independent biomarker for retained attention deficits in these mutant mouse models, similar to the attention deficits that remain after seizures are treated in patients with absence epilepsy (Masur et al., 2013). However, further testing is needed to evaluate aberrant PAC as a potential biomarker for attention deficits.

Another approach to the concurrent treatment of seizures and ADHD may be dietary as opposed to pharmacological. The ketogenic diet—with high fat and very low carbohydrate intake—has been used as a dietary therapy for patients with epilepsy since the 1920s (Wheless, 2008). It has been shown to reduce hyperactivity in nonepileptic rats (Murphy and Burnham, 2006). More recently, the use of the ketogenic diet was evaluated in dogs diagnosed with idiopathic epilepsy and shown to improve seizures, trainability (scored based on attentiveness to owners), and hyperactivity (based on degree of chasing small animals) (Packer et al., 2016). The major drawback to both of these studies with the ketogenic diet is that attention was not formally tested, so the efficacy of the ketogenic diet on attention still requires further investigation.

Future Screening of Antiseizure Drugs

Current strategies for the treatment of patients with epilepsy involve a strong consideration of psychiatric comorbidities (García-Morales et al., 2008). For example, patients with comorbid depression or anxiety should consider mood-stabilizing ASMs such as lamotrigine or carbamazepine and avoid mood destabilizing ASMs such as levetiracetam (Chen et al., 2017). Some medications are known to exacerbate attention deficits more than others; for example, with CAE, valproic acid causes significant aggravation of underlying attention deficits when compared with either lamotrigine or ethosuximide (Masur et al., 2013). However, there are no current FDA-approved ASMs that are known to improve deficits in attention. These findings in patients are consistent with testing of ASMs on rodents using the 5-CSRTT, where valproic acid, phenytoin, pregabalin, vigabatrin, and lacosamide all had a dose-dependent increase in incorrect or omission responses, whereas levetiracetam had no significant effect compared to vehicle alone (Higgins et al., 2010; Mazurkiewicz et al., 1992). These data are consistent with clinical findings suggesting that levetiracetam uniquely spares attention with some subtle signs of cognitive improvement at therapeutic levels (López-Góngora et al., 2008; Piazzini et al., 2006). Given the excellent translational correlation of ASM-related patient outcomes and rodent performance on the 5-CSRTT, there is significant merit to screening of novel ASMs using the 5-CSRTT with both wild-type and rodent models of epilepsy.