Address correspondence and reprint requests to Dr. C. Baumgartner at Universitätsklinik für Neurologie, Währinger Gürtel 18-20, 1090 Vienna, Austria. The first two authors contributed equally to this work.
Summary: Purpose: Postictal psychosis is a well-known complication, occurring especially in patients with temporal lobe epilepsy. It usually runs a benign course. The literature on this topic is sparse, and the underlying pathogenic mechanisms are not known.
Methods: We report five patients with temporal lobe epilepsy in whom postictal psychosis developed during the course of video-EEG monitoring; they were studied with hexamethyl-propyleneamine-oxime single-photon emission computed tomography (HMPAO-SPECT) during and after the psychotic event.
Results: In comparison to the interictal state, all SPECT scans obtained during postictal psychosis were remarkable for bifrontal and bitemporal hyperperfusion patterns. Some studies also demonstrated unilateral left lateral frontal hyperperfusion. These cortical blood-flow patterns appeared to be distinct from those obtained during complex partial seizures.
Conclusions: Our data suggest that postictal psychoses in patients with temporal lobe epilepsy are associated with hyperactivation of both temporal and frontal lobe structures. This hyperperfusion may reflect ongoing (subcortical) discharges, active inhibitory mechanisms that terminate the seizure, or simply a dysregulation of cerebral blood flow.
The relation of psychosis and epilepsy has been an issue of interest since European psychiatrists in the early nineteenth century noted a high incidence of psychotic episodes in institutionalized patients with epilepsy (1,2). Although chronic interictal psychosis in patients with long-standing epilepsy has been studied in detail (3–6), postictal psychosis (PIP) has received relatively little attention (7–9). A clear temporal relation exists between the psychotic state and a precipitating series of tonic–clonic or even complex partial seizures, with a characteristic lucid interval lasting from several hours up to 6 days. The psychopathology of PIP is often polymorphic, with abnormal mood, paranoid delusions, and fluctuating impairment of consciousness and orientation being the symptoms most often reported (7). PIP accounts for ∼25% of psychoses in epilepsy (10) and was reported in 6.4% of all patients undergoing video-EEG monitoring at the Cleveland Clinic during a period of 18 months (11).
PIP usually runs a benign course, with remission of psychotic symptoms over several days, often without need for neuroleptic treatment (12). In rare cases, however, chronic psychoses may develop from recurrent or even a single episode of PIP (7,13). A possible relation between PIP and certain types of epilepsy has often been stressed but remains unclear. Whereas some authors reported a higher frequency of PIP in patients with focal epilepsies and complex partial seizures (7), others described a preponderance of generalized epilepsies (14).
PIPs have attracted some attention in the last decade because they are supposed to have a closer association with epileptogenesis than do interictal/chronic forms of epileptic psychoses. Nevertheless, the pathophysiology of PIP is not known.
Because single photon emission computed tomography (SPECT) offers the opportunity to visualize the three-dimensional dynamic changes of regional cortical blood flow (rCBF) associated not only with seizures but also with different types of psychiatric disorders, we think that it is especially suitable for the study of PIP, a condition in which epilepsy and psychosis are closely (temporally and maybe causally) related to each other.
We report five patients with temporal lobe epilepsy (TLE) in whom psychotic symptoms developed during video-EEG monitoring. In each patient, hexamethyl-propyleneamine-oxime (HMPAO)-SPECT scans were performed during PIP and after complete remission of psychotic symptoms to elucidate mechanisms possibly involved in the generation of PIP.
All patients were suspected to have focal epilepsies and therefore underwent intensive video-EEG monitoring for the purpose of surgical therapy for their refractory epilepsies.
The diagnosis was based on the findings of clinical seizure semiology, interictal and ictal EEG (recorded for an average of 5 days), high-resolution magnetic resonance imaging (MRI) scans, and in some patients, functional imaging studies, including SPECT and positron emission tomography (PET). Antiepileptic drugs (AEDs) were reduced at the beginning of video-EEG monitoring until three to eight seizures had been documented. Thereafter, either the same AEDs were restarted or a new therapy regimen was introduced. PIP was diagnosed by a board-certified psychiatrist (N.B.) with simultaneous EEG available during assessment.
SPECT studies were obtained after formal psychiatric assessment during psychosis (PIP study), and after complete remission of psychotic symptoms (interictal study). Video-EEG was performed while the tracer was administered in PIP studies but not in the interictal study. Patients had to be seizure free and of normal psychiatric assessment for ≥48 h before interictal studies were obtained. In two patients, an additional ictal study was performed within 60 s after EEG onset of a complex partial seizure of typical clinical semiology.
SPECT studies were performed with a three-headed rotating scintillation camera (Siemens Multispect-3) equipped with UHRES collimators. The spatial resolution of the system was 6.5- to 7-mm full-width half-maximum (FWHM) in the reconstructional plane. Thirty minutes after intravenous administration of 740 MBq (20 mCi) of [99mT]-labeled HMPAO, data acquisition was started by step-and-shoot mode. A total of 180 projections was recorded during 30 min (2-degree steps, 30 s/angle) in 128 × 128 matrices by using a diameter rotation of 280 mm. After acquisition, projections were filtered before reconstruction to improve the signal-to-noise ratio. After reconstruction (filtered backprojection, Butterworth filter; cut-off frequency, 0.9), cross sections were corrected for tissue absorption by Chang's method (15).
To ensure comparison of identical anatomic structures, both SPECT studies and MRI scans were realigned by using an interactive technique for three-dimensional image reconstruction (16). Thereafter, three 3.45-mm (voxel size) thick transverse cross sections were summed consecutively to obtain a set of overlapping 10.35-mm-thick cross sections. A total of 107 regions of interest (ROIs) covering the entire brain was drawn on the interictal SPECT images and then transferred to the corresponding cross sections of the PIP studies. Normalized rCBF values (regional indices, RIs) were calculated as the ratios between the mean counts per voxel of a specific ROI and the mean counts per voxel of all ROIs (RI, mean counts/voxel of ROI divided by mean counts/voxel of all ROIs). The relative changes of rCBF for a specific ROI between the PIP study and the interictal study were calculated as the ratio of the regional indices as follows: [(RI'PIP'/RI'interictal') × 100]–100. Thus a positive value of this ratio is equivalent to an activation of a specific ROI during PIP, whereas a negative value corresponds to a deactivation. According to a previous study on the reproducibility of rCBF in normal volunteers, an RI change of ≥12% between the two conditions was considered significant (17).
Patient 1 was a 31-year-old right-handed man whose history was notable for recurrent febrile generalized tonic–clonic seizures starting at age 8 months and lasting for ∼30 months. Thereafter he became seizure free until age 10 years, when he experienced his first complex partial seizure.
During video-EEG monitoring, we recorded a total of four seizures. The patient experienced a behavioral arrest with impaired consciousness, followed by oroalimentary and bimanual automatisms. During the postictal period, the patient was first aphasic and then confused for ≤10 min. All four seizures occurred within 22 h, with the last one evolving into a secondarily generalized tonic–clonic seizure followed by intravenous administration of 10 mg diazepam (DZP). Oral phenytoin (PHT) was reinstated. The patient recovered to baseline until 36 h after the last seizure, when he began to experience delusions of jealousy and illusions that his wife was deceiving him (for details of psychiatric symptoms, see Table 1). HMPAO-SPECT was performed 5 h after the onset of PIP and revealed bitemporal as well as bifrontal hyperperfusion (see Table 3). Scalp EEG performed immediately before administration of HMPAO showed slightly increased spike frequency as compared with the interictal period, with maximal amplitude over the left sphenoidal electrode but no rhythmic activity that could be interpreted as an ictal phenomenon. Overall, PIP lasted for 3 days and was treated with intravenous haloperidol. Previous history was unremarkable for psychotic symptoms; however, several months after video-EEG monitoring, the patient attempted suicide after a series of generalized tonic–clonic seizures, raising the suspicion of another PIP event. Similar to the first episode, the patient fully recovered within 1 week without ongoing antidepressant or neuroleptic medication. After selective amygdalohippocampectomy, the patient remained almost seizure free (Engel's class II) for 7 years.
Table 1. Psychiatric assessment during postictal psychosis
Stream of thought
Content of thought
Disoriented to situation
Coherent, accelerated, lost associations
Inconsistent, accelerated, lost associations
Blunted, euphoric, anxious
Coherent, slow, blocking
Inconsistent, lost associations
Table 3. Overview on hexamethyl-propyleneamine-oxime single-photon emission computed tomography results
Hyperperfusion during psychosis
Bitemporal (inferior and medial); left and right mediofrontal; left orbitofrontal; left lateral frontal posterior
Left mediotemporal, bilateral inferior temporal; left and right mediofrontal; left orbitofrontal; left lateral frontal posterior
Left > right temporal
Bitemporal (inferior and medial); bilateral orbito- and mediofrontal; left lateral frontal posterior
Bitemporal (inferior and medial); right mediofrontal; left orbitofrontal; left lateral frontal posterior
Bitemporal, left and right mediofrontal
Patient 2 was a 29-year-old woman in whom seizures developed several months after concussion at age 18 years. No previous episodes of psychiatric abnormalities were reported by the patient or her family. During intensive video-EEG monitoring, we recorded a total of seven complex partial seizures. The patient experienced an aura with a comfortable feeling in the head followed by a behavioral arrest and impairment of consciousness. Thereafter she performed oroalimentary and bipedal automatism. Postictally, she was confused for several minutes.
An ictal HMPAO-SPECT was performed during the third seizure on day 3 of the video-EEG monitoring and revealed bitemporal (right more than left) hyperperfusion (Fig. 1). On day 5, four complex partial seizures occurred within 24 h. At a time 16 h after the last seizure, while the patient was still not taking antiepileptic medication (AEDs), she became anxious and experienced delusions that her family tried to kill her (for details of psychiatric assessment, see Table 1). A repeated HMPAO-SPECT was performed during this episode and revealed pronounced bifrontal and bitemporal hyperperfusion (see Table 3 and Fig. 1). Scalp EEG obtained during the psychotic event was comparable to the interictal EEG. The psychotic symptoms completely resolved within 3 more days by simply reinstating the AED dosage used before, without the need for antipsychotic medication.
Patient 3 was a 36-year-old right-handed man whose history was remarkable for a prolonged delivery and peripartal hypoxia. After a single seizure at age 13 months, complex partial seizures with rare secondary generalization developed at age 3 years. Two years before he was admitted for intensive video-EEG monitoring, he had repeated short episodes with psychotic symptoms and was therefore treated with antidepressants and neuroleptics. During video-EEG monitoring, we recorded four complex partial seizures with behavioral arrest. Two seizures were remarkable for oroalimentary automatisms with impaired consciousness, and one seizure secondarily generalized with version and dystonic posturing of the right upper extremity. Postictally, the patient was confused for ≤5 min in three seizures. During the first complex partial seizure, an ictal SPECT scan revealed left temporal (mesial and lateral) hyperperfusion.
On the fifth day of video-EEG monitoring, the patient had three seizures within 4 h and was given his AEDs again. He fully recovered until 48 h later, when his behavior became suspicious because of severe anxiety, optic hallucinations, and delusions of being shadowed by strangers (see Table 1 for detailed psychiatric assessment). One day later, when he was still psychotic, we performed an HMPAO-SPECT, which showed bitemporal and bifrontal hyperperfusion (see Table 3 and Fig. 2). Scalp EEG during PIP was normal except for several bitemporal spikes with a frequency similar to the interictal period. He was given oral haloperidol for 5 days with complete resolution of psychotic symptoms. A left-sided amygdalohippocampectomy was performed 3 years ago, and the patient subsequently remained almost seizure free.
Patient 4 was a 30-year-old right-handed man in whom the first seizure developed at age 9 years, with the etiology unknown. During intensive video-EEG monitoring, we documented two complex partial seizures with behavioral arrest as the single clinical sign, with no notable postictal deficits. Because of the nonlateralizing EEG and absence of pathological findings on the MRI scan (for details, see Table 2), invasive EEG recordings were suggested, but refused by the patient. Two days after admission, the patient experienced two complex-partial and two secondarily generalized tonic–clonic seizures within 6 h at home, and he received 10 mg of rectal DZP. Twenty-four hours later, he was readmitted to our hospital because of strange feelings and hallucinations. On admission, he experienced acoustic hallucinations and illusions of people poisoning him (for details, see Table 1). Scalp EEG revealed single spikes over left temporal electrodes. HMPAO-SPECT performed during psychosis was remarkable for bitemporal and bifrontal hyperperfusion (see Table 3 and Fig. 3). The patient received oral haloperidol and regained normal psychopathological status within 3 days. No similar events were reported before or during a 4-year follow-up period.
Table 2. Patients' histories and results of video-EEG monitoring
CPS >> GTCS
98% left temporal; 2% right temporal
Left hippocampal sclerosis
Left mesial TLE
39% left temporal 61% right temporal
CPS >> GTCS
60% left temporal; 40% right temporal
Left hippocampal scerosis
Left mesial TLE
CPS >> GTCS
45% left temporal; 55% right temporal
CPS >>> GTCS
26% left temporal; 74% right temporal
Right hippocampal sclerosis
Right mesial TLE
Patient 5 was a 34-year-old right-handed woman whose seizure disorder started at age 12 years. Her history was remarkable for febrile convulsion at age 30 months. The patient reported the occurrence of acoustic hallucinations with God speaking to her several times in the history. During one of these episodes, she tried to commit suicide by jumping out a window. During video-EEG monitoring, five seizures could be documented, two of them secondarily generalized (for details, see Table 2). The last three seizures (one of them secondarily generalized) occurred within 24 h, and the patient was given her AEDs immediately after the last seizure. Twenty hours later, she became euphoric and started to exhibit auditory hallucinations and religious illusions. An HMPAO-SPECT, obtained on the next day while she was still psychotic, was remarkable for left more than right mesiotemporal and bilateral mediofrontal hyperperfusion (see Table 3). Simultaneous EEG during PIP was comparable to interictal recordings. Without the need for neuroleptic treatment, the patient regained normal psychopathologic status within 3 days. The patient remained seizure free after a right-sided amygdalohippocampectomy was performed 5 years ago.
We report five patients with TLE in whom PIP developed during or immediately after video-EEG monitoring. As compared with the symptom-free interval, individual SPECT scans apparently showed a global increase in CBF. After normalization of regional CBF to the global CBF, we compared individual ROIs and found a bitemporal and a mesial frontal and a unilateral left lateral frontal hyperperfusion pattern during PIP (see Fig. 4). The corresponding scalp EEG showed no significant change from the EEG of the interictal state and specifically no ictal EEG pattern. Therefore, our results suggest involvement of both temporal and frontal lobes in the generation of PIP.
Studies on the mechanisms underlying PIP are few, and the results are inconsistent. An HMPAO-SPECT study performed by Fong et al. (18) in two patients with right TLE in whom PIP developed after video-EEG monitoring reported a marked right temporal and left basal ganglia hyperperfusion. The authors concluded that postictal cerebral hypofunction, similar to Todd paralysis, might not be the underlying mechanism of PIP, as was suggested by Savard et al. (8), who noted the clinical analogy of psychoses after complex partial seizures to other postictal phenomena such as postictal paresis. In contrast, the findings of focal hyperperfusion on ictal SPECT scans led Fong et al. (18) to the hypothesis that cerebral hyperactivation may lead to PIP. Similarly, Boylan (19), in her reply to the work of Fong et al., argued that the regional cerebral hyperperfusion during PIP was in good accordance with another SPECT study showing regional cerebral hyperperfusion in patients with postictal hemiparesis (20). A vascular “blush,” perhaps representing loss of cerebrovascular autoregulation at the side of the epileptogenic foci, might be the explanation for this hyperperfusion, as suggested by an angiogram performed during postictal hemiparesis (21). However, Boylan (22) presented the strongest argument against the hypofunction theory of Todd paralysis in PIP, which is the delayed onset of the phenomenon, as compared with the decrescendo course of Todd motor, cognitive, and visual phenomena.
The few depth-EEG studies performed during epileptic psychosis have yielded inconsistent results. One patient documented by So et al. (23) showed frequent bitemporal independent interictal epileptiform discharges, maximally involving mesial limbic structures, accompanied by slow-wave abnormalities, but no electrographic seizure activity. In contrast, a case report by Takeda et al. (24) described an increase in the frequency and duration of seizure discharges in the left amygdala when psychotic symptoms deteriorated, and a gradual decrease of seizure discharges, with psychotic symptoms remitting. However, the clinical course of this patient does not discount the possibility that the patient indeed had an ictal psychosis. Difficulties in differentiating PIP from complex partial status epilepticus can sometimes occur, as it has been documented impressively by Wieser (25), whose patient experienced prolonged anxiety, auditory hallucinations of songs, and dreamy recall during right temporal lobe status epilepticus. Similarly, Stevens and Lonsbury-Martin (26) suggested that subclinical limbic seizures could mediate apparently interictal psychopathology. Our own results could not definitely rule out the possibility that PIP was due to ongoing discharges, because no depth EEG recordings were performed. However, PIP in our patients differed from the preceding seizures in several aspects: (a) surface EEG showed no ictal discharges, although there was a slight increase in spike frequency; (b) the clinical seizure semiology was rather stereotyped in all complex partial seizures of each patient, differing markedly from symptoms during PIP; and (c) in two patients additionally, an ictal SPECT was available, showing circumscribed temporal hyperperfusion patterns in contrast to the scans obtained during psychosis.
Landolt (27,28) first described the phenomenon by which, in patients whose seizures came under control and whose EEGs became normal, psychotic symptoms developed, and he later called the phenomenon forced normalization. Forced normalization thus was defined as an EEG phenomenon, whereas its clinical counterpart with patients becoming psychotic when their seizures came under control and their psychosis resolving with the reoccurrence of seizures was referred to as alternative psychoses by Tellenbach (29). This concept of an association between epilepsy and psychosis was in contrast to the earlier hypothesis that some kind of antagonism between epilepsy and psychosis exists, which originally had led von Meduna (30) to introduce convulsive therapy for the treatment of schizophrenia. This apparent discrepancy was at least in part resolved by the concept of Wolf (31), who proposed that ongoing subcortical epileptic activity with an inhibitory surround is related to the phenomenon of PIP. Fong (18), in his reply to Boylan, suggested that ongoing abnormal electrical activities propagating via subcortical networks might cause PIP. Increased CBF, as found in our study in this context, may reflect such subcortical electrical activity as well as some inhibitory mechanisms terminating the cortical epileptic discharges either via increased active inhibition or via dysregulation of CBF. Postictal SPECT scans in patients without clinical relevant findings after seizure offset would add some valuable information to this hypothesis. However, such studies have not been performed because of ethical concerns.
Left lateralization of temporal lobe dysfunction was mentioned as a risk factor for epileptic psychosis, originally by Flor-Henry (32) in the late 1960s. Studies supporting the laterality hypothesis were later conducted by using SPECT (33–35), PET (36), and surface (37) and depth (5) EEG recordings, as well as neuropathologic (38) and neuropsychological (39) examinations. For PIP, our results suggest an involvement of both hemispheres in the development of PIP, because all patients had bilateral hyperperfusion patterns, although left lateralization in the lateral frontal region also was apparent in four of our five patients, regardless of the hemisphere of seizure onset. In addition, four of five patients showed bitemporal independent interictal epileptiform discharges, which points to involvement of both hemispheres. Although our patient group is too small to allow any generalizations, our results are supported by other studies that found an incidence of bilateral interictal epileptiform discharges in patients with PIP of ≤77%(8,23). Similarly, the interictal EEG in two patients with right-sided mesial TLE studied by Fong et al. (18) again showed bilateral abnormalities.
To conclude, our HMPAO-SPECT findings in five patients obtained during PIP suggest a hyperactivation of temporal and frontal lobe structures as one mechanism involved in the generation of PIP. Our findings enhance previous work on this area, supporting the hypothesis that PIP might be related to an increase in CBF. Elevated CBF in turn might be associated with inhibitory activity either through increased active inhibition or through dysregulation of cerebral blood flow. Alternatively, increased CBF might be associated with some ongoing electrical activity propagating via subcortical networks after termination of the seizure and suppression of its corresponding cortical discharges.
Acknowledgment: This research was supported by the Jubiläumsfonds der Österreichischen Nationalbank (projects 5792 and 8135).