A wide range of comorbidities affects patients with epilepsy, including psychiatric, neuropsychological, social, and systemic disorders. Depression is one of the most prominent and disabling. Patients with depression as well as epilepsy have lower income, QOLIE-89 scores, as well as marriage and employment rates. Some studies suggest a relation with poor response to antiepileptic drugs or even surgical outcome.1
Estimates of depression prevalence in epilepsy vary widely, from about 10–40%, depending on whether data come from community surveys or clinical samples.2 Rates are usually highest in patients being evaluated for drug-resistant epilepsy, either because of the severity of the seizure disorder or thorough investigation. Several studies based on community health surveys have been performed. In the Canadian Community Health Survey and the US Centers for Disease Control Health Styles Survey, the risk for depression in epilepsy patients at any time point was about twice that of the overall population.3, 4 A recent review suggested that the lifetime risk for experiencing any psychiatric disorder, of which depression and anxiety are the most common, is 30–35%.1 Patients often describe prominent anxiety, as a well as an intermittent, waxing and waning course described as a “dysthymic” or “interictal dysphoric” disorder.5, 6 Patients may report depression, irritability, or less commonly elation, hours to days before a seizure. Postictal depression and anxiety can last up to 14 days.7, 8
The etiology of depression in patients with epilepsy most likely is multifactorial, including both social and biological factors1 (Table 19.1). Imaging studies have the potential to help elucidate the pathophysiology of some of these conditions, and suggest treatment approaches.
|Age of onset and duration|
|Seizure frequency, severity|
|Temporal lobe epilepsy (TLE)|
|Left focus localization|
|Lack of control|
|D. Antiepileptic drugs (AEDs)|
19.2 MRI and PET Glucose Metabolism in Primary Major Depressive Disorders
In patients with primary major depressive disorders (MDD) but not epilepsy, MRI studies have shown atrophy in orbitomedial prefrontal cortex, as well as hippocampus and other limbic regions. Reduced hippocampal volume is the most consistent funding, and may be restricted to subfields, particularly cornu ammonis and dentate gyrus.12 The patients do not have increased signal consistent with the focal gliosis found in epilepsy, however. There is some evidence for progressive atrophy, possibly related to stress-induced increased glucocorticoid release and inhibited hippocampal neo-neurogenesis.13 Positron emission tomography (PET) studies of cerebral metabolism using [18F]-2-deoxy glucose showed decreased cerebral metabolic rate for glucose (CMRglc) in dorsomedial and anterolateral prefrontal cortex and cingulate gyrus.14
19.2.1 MRI, Depression, and Epilepsy
Structural Imaging studies have shown a number of possible findings that may be associated with depression in people with epilepsy. Patients with mesial temporal sclerosis (MTS) have been reported to have higher depression rating scores, either without a laterality effect,15 with right MTS,16 or left MTS.17 In contrast, however, several investigators reported no relation of depression to MTS.18, 19 One study reported an increased incidence of depression in nonlesional compared with lesional focal epilepsy.20 Two others reported that depressed patients with a right temporal focus, with or without MTS, had greater left hippocampal volume reduction than patients without depression.17, 21 In contrast, in a study of 28 patients, those with TLE and depression had larger hippocampal volumes than those with TLE alone.22
One study found positive relations between bilateral amygdala volume and depression.23 In contrast, patients with the “dysphoric” disorder of epilepsy, including emotional instability, dysphoria, irritability, and aggression, had significantly reduced amygdalar volumes, compared to patients with epilepsy alone and those with depression.24
Several new MRI-based approaches have been used to examine relationships between structural and functional networks and depression in epilepsy. TLE patients were compared with healthy controls using voxel-based morphometry (VBM).25 Patients received both the Structured Clinical Interview for DSM-IV and the Beck Depression Inventory (BDI). Those with depression had more areas of gray matter volume loss, compared with control values, than patients with TLE alone. In a study using quantitative, surface-based MRI analysis, increasing severity of depression was associated with orbitofrontal thinning in controls, but thickening in TLE patients.26 Right and left mesial TLE patients and controls were investigated using resting-state blood-oxygen-level-dependent functional MRI for functional connectivity.27 Right TLE patients showed a relation between high depression and anxiety scores and normal hippocampal to amygdala functional connectivity values, while left TLE patients showed the reverse relationship. A study comparing patients with TLE to healthy controls compared BDI-II scores with functional connectivity (FC) between medial temporal and prefrontal regions as well as diffusion tensor imaging for uncinate fasciculus fractional anisotropy as well as amygdalar and hippocampal mean diffusivity to predict depressive symptoms.28 Higher depression rating scores were associated with stronger connectivity of hippocampus ipsilateral to the seizure focus with anterior prefrontal cortex (this measure was the strongest predictor of depressive symptoms), lower and bilateral uncinate fasciculus fractional anisotropy. Patients with left temporal foci had higher ipsilateral hippocampal mean diffusivity, while those with right temporal foci had lower contralateral uncinate fasciculus fractional anisotropy.
Seventeen TLE patients with a lifetime affective disorder diagnosis, 31 without psychiatric history, and 30 healthy controls performed a visuospatial working memory paradigm to investigate the default mode network.29 TLE patients with lifetime affective disorder had significantly greater deactivation in subgenual anterior cingulate cortex than either patients with TLE but not depression or healthy controls. The result was not affected by current psychiatric drug treatment or the severity of current depression or anxiety measured by the BDI. In this study there were no significant between group differences in gray matter volume or connectivity.
Overall these disparate data do suggest that depression in people with epilepsy is associated with deficits in cortical structure and connectivity, particularly affecting limbic regions. However, many of the studies are small, and methodology differs. Other factors such as AED response, that may themselves be related to depression, could have affected the results of many studies.30 A large body of work has shown that in patients with TLE, hippocampal volume ipsilateral to the seizure focus is inversely associated with epilepsy duration, and may be affected by a history of complex or prolonged febrile seizures. Association studies cannot examine cause-effect relationships. The complexity of interpreting these results is increased even more by data, for example, showing that an “unhealthy western diet” is associated with lower left hippocampal volume.31
Studies in patients with major depressive disorders (MDD) but not epilepsy have shown results paralleling some findings in epilepsy but also illustrating the difficulties of interpretation. A review of task-based and resting-state fMRI studies of patients younger than 25 years old found increased activity in anterior cingulate, ventromedial and orbitofrontal cortex, and amygdala most consistently and commonly reported in studies involving emotional and reward processing, and affective cognition, as well as resting-state connectivity.32 Meta-analysis by another group found increased activation in some of the same regions, but more task-specific changes including hypoactivation in caudate, hyperactivation in thalamus and parahippocampal gyrus, hypoactivity in posterior insula, and hyperactivity in dorsolateral prefrontal and superior temporal cortex.33 A second meta-analysis found frontoparietal hypoconnectivity and default mode network hyperconnectivity.34 Some of these differences could have been due not only to variation in the studies included, although there was substantial overlap, but also to the vagaries of the analytic approaches. It is likely that in patients with epilepsy only large multicenter studies that will have the power to account for a variety of factors including seizure frequency, epilepsy duration, and AED treatment, will lead to firmer conclusions.
19.2.2 Depression, Imaging, and Epilepsy Surgery
The effect of temporal lobectomy on depression in epilepsy is variable and not always predictable.18 In a retrospective study, 30 patients with left MTS had gray matter volumes measured using voxel-based morphometry (VBM).35 Development of postoperative depression, not necessarily in the immediate perioperative period, was associated with reduced preoperative gray matter volume in orbitofrontal cortices, ipsilateral cingulate gyrus and thalamus. The results were not affected by outcome for seizure control. In another study, patients with mesial and lateral temporal lobe resections were compared.36 In the mesial but not lateral temporal group postoperative depression was associated with significantly smaller contralateral hippocampal volume. Postoperative BDI scores correlated significantly with hippocampal and amygdala resection extent in 35 patients.37
19.3 Cerebral Metabolism in Depression and Epilepsy
Using [18F]fluorodeoxyglucose PET to measure cerebral glucose metabolism, patients with reduced mesial temporal CMRglc ipsilateral to their seizure focus had higher BDI scores, and a study using 1H-magnetic resonance spectroscopy found a relation between the extent of the temporal lobe abnormality and depression rating scores.38 Several studies found that bilateral or unilateral frontal hypometabolism was associated with depression in epilepsy.39, 40 These results are interesting in light of the more recent data suggesting alterations in frontal lobe cortical thickness and connectivity in patients with TLE and depression.26 In each of these studies, patients with drug-resistant epilepsy being considered for surgery were evaluated. These patients may be more likely to have both depression and extensive PET hypometabolism. A study of 11 patients using the N-methyl d-aspartate receptor agonist [18F]GE-179 VT found that patients with focal epilepsy overall had increased binding, but that there was reduced receptor availability in patients taking antidepressants, suggesting increased exogenous occupancy that might be related to drug effects.41
19.3.1 Serotonin and Epilepsy
The most interesting approaches to PET imaging of depression in epilepsy used serotonin (5HT) 1 A receptor ligands. The serotonin system includes at least 16 receptor subtypes, that are involved in modulation of a wide variety of physiologic processes including vascular and nonvascular smooth muscle contractility, platelet aggregation, appetite, wakefulness, sleep, mood, and anxiety.42 Activation of the G-protein coupled 5HT1A receptors in particular has antidepressant and anxiolytic effects. Serotoninergic neurons have cell bodies in the midbrain raphe and project widely to neocortical limbic regions; 5HT1A receptors are particularly abundant in the latter. Their activation leads to increased potassium and reduced calcium conductance, increased norepinephrine and decreased glutamate release, and CA1 membrane hyperpolarization. 5HT1A receptor activation has antiseizure effects in several epilepsy models.
19.3.2 5HT1A PET Imaging in Primary Major Depressive Disorders
Previous work in patients with primary MDDs but not epilepsy had shown reduced 5HT1A receptor binding in several brain regions, including lateral and mesial temporal and parieto-occipital cortex, insula, anterior and posterior cingulate in both treated with selective serotonin reuptake inhibitors (SSRIs) and untreated patients.43 The reductions are present when patients are in remission, suggesting a trait rather than state effect.44 Some monkeys have a naturally occurring “depressive disorder”; these animals show reduced 5HT1A binding compared to nondepressed monkeys in similar regions to those found in humans.45
In contrast, other investigators reported increased binding in the same regions in MDD patients who had never been exposed to SSRIs, and no significant abnormalities in those who did have previous treatment, even if they were not taking the drugs at the time of study.46 A study using [11C]WAY-100635 that compared patients before and after SSRI treatment showed no difference between the two states. However, before treatment, patients who proved later to be SSRI nonresponders had higher bilateral orbitofrontal binding than those who did respond.47
19.3.3 5HT1A Imaging in Epilepsy
Patients with mesial and lateral temporal foci have reduced 5HT1A receptor using [18F]-Trans-4-Fluoro-N-(2-[4-(2-methoxyphenyl)piperazin-1-yl]ethyl)-N-(2-pyridyl)cyclohexanecarboxamide ([18F]-FCWAY).48–51 These reductions are not due simply to tissue volume loss, as shown by studies that performed partial volume correction.51 5HT1A receptor imaging may help predict outcome after temporal lobectomy.52 Two studies investigated the relation of scales measuring concurrent depressive symptoms to 5HT1A receptor binding or number. [18F]-FCWAY ligand free fraction corrected volume of distribution (V/f1) in hippocampus ipsilateral to the seizure focus was inversely correlated with the BDI, with a nonsignificant trend for contralateral hippocampus53 (Figure 19.1). Patients with a BDI score more than 20, suggesting moderate to severe depression, had significantly lower [18F]-FCWAY binding than those with lower BDI (Figure 19.2). The presence of MTS, focus laterality, or gender did not have a significant effect on the BDI. Similar results had been found in ipsilateral insula using [11C]WAY and the Montgomery-Asberg scale.49 In a study relating 5HT1A receptor binding to hippocampal volume, BDI, and neuropsychological test scores, there was a significant effect of the interaction of left hippocampal [18F]FCWAY V/f1 and left hippocampal volume on delayed auditory memory, but not of either alone.54 Focus laterality and BDI did not affect memory in this study, but BDI did correlate inversely with 5HT1A receptor binding as in the previous reports.
Figure 19.1. Original and partial volume corrected (PVC) [18F]-FCWAY PET scans showing reduced binding in a patient with right temporal mesial temporal sclerosis and the effect of partial volume correction.
Figure 19.2. Correlation of Beck Depression Inventory with 5HT1A receptor binding measured with [18F]-FCWAY in hippocampus ipsilateral to the seizure onset zone (R = .38, p < .02).
In contrast, a study using a different ligand, [18F]-MPPF, in 24 patients with drug-resistant TLE and MRI evidence of hippocampal sclerosis, BDI scores ranging from 0 to 34, and no prior antidepressant exposure, found that total BDI score and symptoms of psychomotor anhedonia and negative cognition correlated positively with [18F]MPPF binding in raphe nuclei and insula contralateral to the seizure focus.55 Somatic symptoms correlated positively with [18F]MPPF binding in the hippocampus and parahippocampus ipsilateral to seizure onset, left midcingulate gyrus and bilateral inferior dorsolateral frontal cortex. There may be several potential explanations for the difference between this study and the previous ones that showed relatively decreased rather than increased binding in patients with epilepsy and depression. The authors suggested that there might be differences in the patient population, including uniform presence of MTS, and no prior exposure to antidepressant drugs (which would not have distinguished the patients from those in several previous studies). However, the most likely reason for the discrepancy is the relative difference in binding affinity between the PET ligands used. [18F]MPPF has much lower affinity for 5HT1A receptors than [18F]-FCWAY, and is very close to the binding potential of serotonin itself. Reduced synaptic 5HT availability, which presumably is present in patients with depression, could lead to relatively increased receptor occupancy by the exogenous ligand since more binding sites would be available, when depressed epilepsy patients are compared with nondepressed patients.55 [18F]FCWAY would not be affected by this mechanism, since it binds more avidly that 5HT itself. In fact, the studies using the two ligands complement each other, confirming the role of 5HT1A receptors in depression and epilepsy, while showing both reduced receptor availability ([18F]FCWAY), and also reduced synaptic 5HT availability. Moreover, they suggest widespread and partly bilateral abnormalities, involving not only mesial temporal structures, but cingulate cortex and insula as well.
Some studies in MDD suggested an effect of SSRIs on 5HT1A receptor binding.46, 47 Others however did not find an effect, possibly because of the relatively low occupancy of the receptor sites.43, 56 Some antiepileptic drugs (AEDs) may increase 5HT synthesis and release or block reuptake. However, no AED effects on [18F]-FCWAY binding were found in patients taking these drugs, once correction was made for the increase in FCWAY plasma free fraction, probably due to protein binding interactions with AEDs.57
In addition to studies using state-related depression scales such as the BDI, epilepsy patients may present with a comorbid MDD diagnosis based on the Structured Clinical Interview for DSM-IV, which identifies an underlying depressive illness even if patients are euthymic at the time of study.58 This study, using [18F]FCWAY, found relatively reduced binding in TLE patients with MDD, extending into nonlesional limbic brain areas outside the epileptic focus. After Bonferroni correction, the depressed group had lower values in anterior cingulate, hippocampus, and medial and superior temporal lobe (Figure 19.3). Seizure focus side was associated with ipsilateral decreases in [18F]-FCWAY V/f1 in hippocampus, parahippocampal gyrus, left amygdala, left fusiform gyrus, and right superior temporal lobe. The only significant group (depressed versus nondepressed) by focus (right versus left) interaction was lower [18F]FCWAY V/f1 raphe values in depressed patients with right temporal foci. However, this did not survive correction for multiple comparisons. Nevertheless, it is an interesting finding, as the [18F]MPPF study also found effects in raphe, where the 5HT cell bodies, as well as autoreceptors are located.55 The presence of comorbid anxiety, or MTS, did not affect the results. When only patients with concurrent depression were included, only the hippocampus remain significantly reduced after correction for multiple comparisons.
Figure 19.3. Free-fraction corrected 5HT1A serotonin receptor volume of distribution measured with [18F]FCWAY PET in healthy controls, patients with TLE alone, and TLE with depression, showing the additive effects of depression.
19.4 Serotonin Transport
The serotonin transporter (5HTT), responsible for reuptake of synaptically released ligand, has been implicated in the pathophysiology of depression. “Short” alleles of the encoding gene are associated with reduced transcriptional activity and 5HT reuptake. In patients with one or two short alleles, possible associations with depression, increased response to stress, fear, and “suicidality” in relation to stressful life events, in comparison to individuals homozygous for the long allele have been described.59
Some data suggest a role for 5HTT in epilepsy including a higher frequency of some alleles in TLE and MTS compared with healthy controls, and possibly differing AED response. Altered 5HT reuptake and thus synaptic availability might affect modulation of hippocampal excitability. However, not all studies have found a relation between 5HTT allelic variants and TLE.60
Several PET ligands have been used to image the 5HTT transporter in patients with depression. Some reported increased availability in patients with MDD and bipolar disease compared to controls, in thalamus, insula, prefrontal cortex, and cingulate gyrus.61, 62 However, others reported decreased availability in amygdala, hippocampus, thalamus, putamen, anterior cingulate, and midbrain.63, 64 A recent review found meta-analysis indicates a trend toward reduced serotonin transporter availability in patients with MDD. The authors suggested that inconsistencies in results among studies might be due to symptom heterogeneity and might therefore be important for stratification of patients into clinical subsets.65 Another implication is that much larger studies would be needed to elucidate fully patterns of altered transport (or indeed 5HT receptor binding, given the variability in results discussed above). Unfortunately, PET studies are expensive.
Thirteen patients with TLE who had been evaluated with the BDI as well as the SCID had imaging with the 5-HTT transporter ligand [11C]-3-amino-4-(2-dimethylaminomethylphenylsulfanyl)-benzonitrile ([11C]DASB) as well as [18F]FCWAY.66 Regional [11C]DASB binding did not differ between patients and controls. However, depression diagnosis had a significant effect on [11C]DASB asymmetry, with significantly lower [11C]DASB binding in insular cortex and a trend for fusiform gyrus ipsilateral to the seizure focus (Figure 19.4). [18F]FCWAY binding ipsilateral to the seizure focus was significantly lower for patients than controls in hippocampus, amygdala, and fusiform gyrus. And there was a significant correlation between [18F]FCWAY and [11C]DASB binding. These data suggest relatively reduced transporter activity ipsilateral to seizure foci in TLE patients with depression, compared to those with TLE alone or healthy controls. This mechanism might lead to reduced 5HT reuptake and thus increased synaptic availability. Since [11C]DASB and [18F]FCWAY binding were correlated, reduced transport might represent a compensatory mechanism for 5HT1A receptor loss.
Figure 19.4. (A) [11C]-DASB PET in a patient with no history of depression shows symmetrical insula binding. (B) [11C]DASB PET in a patient with a history of depression shows relatively reduced right insula binding.
Taken together, studies of both 5HT1A receptors and 5-HTT suggest a strong role of 5HT1A receptors in TLE patients with depression, and support the use of SSRIs in treating them. Moreover, a review of US Food and Drug Administration data from antidepressant drug trials showed that patients had fewer seizures during treatment than placebo arms, suggesting that the drugs may have antiseizure effects.67
Several lines of evidence, including structural and functional MRI studies, as well as PET studies of 5HT1A receptors and 5HT transport, suggest a role for alterations in limbic structures, as well as serotonergic neurotransmission, in the pathophysiology of depression in patients with epilepsy. At present, most of the data are cross-sectional. It is uncertain, for example, if patients with epilepsy are more likely to become depressed due to limbic dysfunction, or whether some exogenous factors, such as seizure frequency and epilepsy duration, in addition to focus localization, influence both clinical symptoms and anatomo-physiological correlates of depression in epilepsy.
It is encouraging that there are strong parallels between imaging results in patients with depression in epilepsy and MDD alone. These observations support the epidemiologic data showing reciprocal relations between the two disorders. The PET 5HT1A and 5HTT imaging data strongly support the use of SSRIs in treatment. Emerging data on the role of altered FC may lead to new therapeutic approaches involving direct modulation of neuronal activity.