- Email: [email protected]
Parietal lobe source localization and sensitivity to hyperventilation in a patient with subclinical rhythmic electrographic discharges of adults (SREDA)
Clinical Neurophysiology 117 (2006) 2257–2263 www.elsevier.com/locate/clinph
Parietal lobe source localization and sensitivity to hyperventilation in a patient with subclinical rhythmic electrographic discharges of adults (SREDA) Dominik Zumsteg, Danielle M. Andrade, J. Martin Del Campo, Richard Wennberg
*
Division of Neurology, Krembil Neuroscience Centre, University of Toronto, Toronto Western Hospital, 399 Bathurst St., Toronto, Ont., Canada M5T 2S8 Accepted 10 July 2006 Available online 23 August 2006
Abstract Objective: Subclinical rhythmic electrographic discharges of adults (SREDA) is currently considered a benign EEG pattern of uncertain significance. The underlying cortical sources and generating mechanisms are unknown. We performed a source localization analysis of SREDA with the aim of better understanding this unusual EEG pattern. Methods: Multiple spontaneous episodes of typical SREDA were recorded in a patient during continuous EEG monitoring. Additional SREDA episodes were induced by hyperventilation. Source localization was carried out using statistical non-parametric mapping (SNPM) of low resolution electromagnetic tomography (LORETA). Results: SNPM of both time- and frequency-domain LORETA revealed a widespread biparietal cortical origin of SREDA, the anatomical distribution of which included the parietal operculum and the known vascular watershed areas between anterior, middle and posterior cerebral arteries. Vigorous deep hyperventilation induced SREDA on three of four attempts. Mean duration of the hyperventilationinduced SREDA was approximately three times longer than spontaneous events. Conclusions: Investigations in this patient with typical SREDA revealed hyperventilation sensitivity and a posterior hemispheric source localization maximal in the parietal cortex bilaterally, in large part overlying the anatomical distribution of the vascular watershed areas. Significance: The source localization results and sensitivity to hyperventilation suggest some sort of association between cerebral vascular supply and SREDA, as originally proposed by Naquet et al. [Naquet R, Louard C, Rhodes J, Vigouroux M. A propos de certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital. Leur activation par l’hypoxie. Rev Neurol 1961;105:203–207.]. Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: SREDA; EEG; Epilepsy; Electroencephalography; Source localization; LORETA
1. Introduction Subclinical rhythmic electrographic discharges of adults (SREDA) is a rare and distinctive EEG pattern seen primarily in older people, usually over the age of 50 years (mean age 62 years, range 35–98 years in a large series reported by Westmoreland and Klass, 1997). The pattern in its most typical form has a duration of 40–80 s and is characterized by rhythmic, sharply contoured 5–7 Hz theta waves, often with intermingled upper delta/lower theta fre*
Corresponding author. Tel.: +1 416 603 5402; fax: +1 416 603 5768. E-mail address: [email protected] (R. Wennberg).
quency components, the amplitude ranging from 40 to 100 lV with a widespread and bilaterally synchronous distribution maximal over the parietal regions (Westmoreland and Klass, 1981; De Falco et al., 1983; Miller et al., 1985; O’Brien et al., 1998). SREDA may have an abrupt onset, or, in about half of cases, the pattern may begin with one or more monophasic discharges of high amplitude, followed one to several seconds later by the typical predominant theta discharge. SREDA is mostly observed during wakefulness and tends to occur more than once during a single EEG recording, particularly during or shortly after hyperventilation, and recurs in subsequent EEGs obtained from the same person (Westmoreland and Klass, 1981;
1388-2457/$32.00 Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2006.07.137
2258
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
Miller et al., 1985). Unusual or atypical forms with predominant delta frequencies, notched waveforms, more focal distribution, atypical evolution and presence in younger individuals or during sleep have been described (Silva et al., 1995; Westmoreland and Klass, 1997; Nagarajan et al., 2001; Fleming et al., 2004). The typical pattern has not been associated with observable clinical changes or subjective symptoms and, although it may resemble a subclinical ictal epileptiform discharge, seems not to have any correlation with clinical seizures (Thomas et al., 1992; Westmoreland and Klass, 1997). SREDA is probably identical to an EEG pattern described by Naquet et al. (1961, 1965) as ‘‘paroxysmal discharges of the temporo-parieto-occipital junction’’. In addition to hyperventilation, these authors demonstrated that pure relative hypoxia (associated with nitrogen inhalation) or mild relative ischemia (from carotid artery compression) could reliably induce the paroxysmal pattern in susceptible patients. Accordingly, the occurrence of SREDA has been postulated to be associated with cerebrovascular disease by some authors (Naquet et al., 1961; Thomas et al., 1992), although others have not found epidemiological support for this hypothesis (Westmoreland and Klass, 1997). With the vascular hypothesis of Naquet et al. (1961) in mind, we performed a source localization analysis of SREDA with the aim of better understanding this unusual EEG pattern.
No clinical events were noted during the two days of monitoring. A therapeutic trial of sodium divalproex, 750 mg/day, was initiated, and relatives reported a possible slight decrease in frequency of the patient’s ‘‘disconnection’’ episodes over the subsequent month. However, a follow-up 48-h 21-channel ambulatory EEG at this time showed no decrease in frequency of the patient’s SREDA discharges, which occurred more than 20 times per day. One possible clinical event was witnessed by a family member during this recording period, wherein the patient appeared to not correctly identify a common object in his hand, followed by a short period (5–10 s) during which his conversation was not entirely of its normal character. The patient denied knowledge of anything unusual having happened during the entire recording. According to the family member’s recollection, the time of occurrence of this clinical episode, based on a wall clock, coincided with a 25 s epoch of SREDA on the ambulatory EEG. However, the event marker was not activated on the EEG headbox, such that definitive time synchronization was not possible. To further investigate the possibility of a clinical correlation to the SREDA episodes, the patient was admitted to hospital five months later for continuous video-EEG recording. The sodium divalproex had been tapered and discontinued during the previous months. The patient’s family reported a decrease in frequency of the clinical episodes since the initial investigations.
2. Patient and methods
2.2. Video-EEG recording and LORETA analysis
2.1. Case report
In-hospital EEG recording was performed using an XLTEK (Oakville, ON, Canada) recording system with a digitization rate of 200 Hz. Twenty-nine conventional AgCl surface electrodes were placed according to the international 10–20 electrode system, with additional electrodes placed at F9/10, T9/10, P9/10, supraorbitally, and over the zygomatic arches. The reference electrode was placed between Fz and Cz, the ground electrode at Fpz. The band pass was set at 0.1 to 70 Hz. Twenty-five episodes of SREDA were recorded during 24 h of monitoring (see Section 3). Time-domain LORETA analysis was carried out on 64 suitable EEG segments from SREDA episodes (free of movement and eye artifacts). These episodes were identified off-line by visual inspection of the EEG using INSIGHT software (Persyst Development Corporation, Prescott, AZ, USA; 2004) and then subjected to alignment (centered on parietal peak activity (P3, Pz, P4) of the SREDA discharges using longitudinal bipolar and referential derivations, 1000 to +1000 ms). Frequency-domain LORETA analysis was carried out on cross spectra of 72 suitable EEG segments of 2.5 s length (512 sample points) from SREDA episodes (free of movement and eye artifacts; not identical to those used for time-domain analysis). Cross spectra of normalized EEG segments were calculated using the LORETA software package (KEY Institute, Zu¨rich).
A 65-year-old man presented with a three-month history of brief unexplained episodes of ‘‘disconnection’’, which, according to his relatives, were characterized by subtle motor arrest and inappropriate speech. The patient was subjectively unaware of the episodes, which had occurred once every 2–3 weeks and were described to last 5–15 s. There was no history of neurological or cardiovascular disease, and the patient had no cardiovascular risk factors. Neurological examination was normal. A routine outpatient EEG including hyperventilation was normal. Brain MRI (FLAIR sequence) revealed a few non-specific focal high signal abnormalities in the hemispheric white matter bilaterally, interpreted as probable microangiopathic changes (Fig. 1). Carotid Doppler studies were unremarkable. Continuous 48-h 21-channel ambulatory EEG recording documented 28 separate episodes of typical SREDA (13 on day one and 15 on day two), all of which occurred during wakefulness. The EEG was otherwise normal during wakefulness, drowsiness and sleep. The mean SREDA episode duration was 51 s (range 25–100 s). An EKG lead showed no significant changes during the SREDA discharges, although many of the episodes were associated with a mild increase in heart rate of less than 10–15 beats per minute.
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
2259
Fig. 1. Axial brain MRI images (FLAIR) showing few non-specific subcortical punctate bilateral white matter hyperintensities.
These EEG segments and cross spectra were available for source analysis with low resolution electromagnetic tomography (LORETA; Pascual-Marqui et al., 1994, 1999; Frei et al., 2001). The selected segments were subjected to statistical non-parametric mapping (SNPM) of LORETA, as previously described (Zumsteg et al., 2005a,b, 2006). LORETA values represent the power (the squared magnitude of the computed intracerebral current density) within 2394 voxels of the cortex including the amygdala and the hippocampus, corresponding to a voxel size of about 7 mm (Pascual-Marqui et al., 1994; Pascual-Marqui, 1999). The grid in use is based on the digitized Talairach atlas, as provided by the Brain Imaging Center of the Montreal Neurological Institute. The SNPM method calculates the t-values using the permutation approach and offers, for each voxel, the exact (i.e., corrected for multiple comparisons) critical t-values for p < 0.05 and p < 0.01 as well as the exact (i.e., corrected for multiple comparisons) randomization probabilities for testing cluster size (Nichols and Holmes, 2002). 3. Results Twenty-two spontaneous episodes of SREDA (mean duration 53 s; range 15–137 s) were recorded during the video-EEG investigation, the discharges characterized by widespread generalized rhythmic theta activity between 4 and 7 Hz, of medium or medium–high amplitude, with relatively abrupt onset and offset and with no electrographic changes after offset. Some of the longer events showed intermingled delta frequency slow wave activity, especially during the later stages. Occasionally the episodes were preceded (and at times followed) by a small number of low or low–medium amplitude periodic slow waves or slow sharp waves appearing maximal over the mid-parietal regions. Fig. 2 depicts a representative example. During three of these spontaneous SREDA episodes, the patient could be
seen and heard on the simultaneous video recording to be conversing normally either on the telephone or with a nurse in his room. He was able to remember and accurately describe these conversations when questioned later. During two other spontaneous SREDA episodes, the patient was able to respond appropriately to questions from a physician in the room, with complete subsequent recollection of the interaction. Attempts to induce SREDA with vigorous deep hyperventilation were carried out four times over a 30 min period, with SREDA discharges elicited following three of the four attempts (mean duration 158 s; range 40–225 s). The hyperventilation was stopped in each case upon recognition of SREDA onset in the EEG. With the first hyperventilation a 40 s episode of typical SREDA started 60 s after commencement of hyperventilation. With the second hyperventilation a longer SREDA event started 90 s after commencement of hyperventilation, lasting 210 s. With the third hyperventilation (2 min) no SREDA occurred. With the last repetition of hyperventilation SREDA started 75 s after commencement of hyperventilation, lasting for 225 s. The three hyperventilation-induced SREDA episodes had onset within 27 min of each other. In comparison, six spontaneous SREDA episodes occurred during the six hours after awakening (median one SREDA/hour), before hyperventilation. The video-EEG monitoring was stopped 10 min after the final hyperventilation, with no further recorded SREDA. During each of the hyperventilation-induced SREDA episodes the patient was fully conscious and able to interact with the physicians in the room as well as perform bedside neuropsychological tests including correctly drawing a clock face and putting the clock hands on the figure, accurately drawing a cube, bisecting a line correctly, correctly performing serial 7 s and serial 3 s, as well as repeating and remembering three words and naming the days of
2260
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
Fig. 2. Typical episode of spontaneous SREDA, approximately 60 s duration, recorded during in-hospital investigation. Anterior–posterior bipolar montage. Bandpass 0.5–70 Hz.
the week and months of the year sequentially. The only questionable finding was that during the third hyperventilation-induced SREDA episode he correctly repeated a five-digit span forward but would not attempt to produce the numbers in reverse order. He had full recollection of the tests performed during the hyperventilationinduced SREDA episodes. Symptomatically, he felt only lightheadedness as would be expected with the extreme hyperventilation. Time-domain (Fig. 3) and frequency-domain (Fig. 4) source localization using SNPM of LORETA revealed a significant increase of cortical activity over both posterior hemispheres, maximal over the parietal regions. The area of significant increase extended to the parietal operculum and parieto-occipital border, with relative sparing of the parasagittal region on both sides. There was no clear evolution of the spectral pattern with respect to frequency over the course of the SREDA episodes (Fig. 4A). Likewise, there were no significant differences in cortical activation patterns obtained during early (first third) and late (last third) phases of SREDA episodes, neither with time-domain (not shown) nor with frequency-domain analysis (Fig. 4). No significant differences in cortical activation patterns were found between SREDA episodes occurring spontaneously or induced by hyperventilation. 4. Discussion In their initial studies of the EEG discharges that would later come to be known as SREDA, Naquet et al. (1961, 1965) proposed a vascular etiology, based on identification
of the EEG pattern in older individuals, some with other evidence of cerebral vascular disease, as well as localization of the discharges to the vascular watershed area and the ability to reliably reproduce the EEG pattern in these individuals by hyperventilation, nitrogen inhalation and carotid artery compression. The common mechanism proposed for these three activation techniques was relative cerebral hypoxia: ‘‘pure’’ in the case of nitrogen inhalation; ischemic in the case of carotid compression; and associated with decreased cerebral blood flow caused by vasoconstriction in response to hypocapnia in the case of hyperventilation (Naquet et al., 1961). However, Westmoreland and Klass (1981) found that the incidence of cerebral vascular disease, seizures and other neurologic disorders in a large series of patients with SREDA was no greater than that in a control group of age matched subjects with normal EEGs. SREDA is currently viewed as a benign non-specific EEG phenomenon of unknown cause and mechanism (Westmoreland and Klass, 1997; O’Brien et al., 1998). Our source localization results demonstrate a clear regional predominance for the generation of SREDA discharges localized to the biparietal areas, in large part overlying the known anatomical distribution of the vascular watershed areas between anterior, middle and posterior cerebral arteries. The source localization results suggest some sort of relationship between cerebral vascular supply and SREDA, as initially proposed by Naquet et al. (1961). The sensitivity to hyperventilation is also consistent with the vascular hypothesis as originally posed by Naquet et al. (1961). It seems unlikely that hyperventilation-induced respiratory alkalosis and hypocapnia would be themselves
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
2261
Fig. 3. Top left, overlay of 64 individual SREDA discharges in a bipolar montage, aligned to the peak activity of the discharge (red line). Middle, averaged EEG of the 64 individual SREDA discharges depicted at left using a different time scale (above) and spatiotemporal source maps showing the color coded SNPM LORETA t-values of individual Brodmann areas (below). Right, three-dimensional time-domain SNPM LORETA reconstructions at peak activity of the SREDA discharge (red lines A, B, and C from middle plots) representing distributed, color coded t-values (red for positive t-values above level of significance) projected onto generic brain templates.
Fig. 4. (A) Spectrogram [window length 2.5 s, Kaiser window, 90% overlap] of a single representative SREDA episode showing an irregular increase of spectral density power in lower delta and theta EEG frequency bands during the course of the pattern. Note that, with respect to frequency, there is no clear evolution of the spectral pattern over the course of the SREDA event. (B) Three-dimensional frequency domain (theta cross spectra) SNPM LORETA reconstructions for three different periods of 10 SREDA episodes representing distributed, color coded t-values (red for positive t-values above level of significance) projected onto generic brain templates. Early, 21 epochs of 2.5 s (512 sample points) obtained from the first third of the SREDA events; all, 72 epochs obtained from the entire course of the SREDA events; late, 26 epochs of the last third of the SREDA events. Note that there is no significant difference of cortical activation patterns for the three periods analyzed.
2262
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
the triggers for SREDA, given the previously demonstrated ability to similarly induce SREDA with nitrogen breathing, such that the more likely inciting factor associated with hyperventilation is a mild relative hypoxia due to cerebral vasoconstriction and decreased cerebral blood flow in susceptible individuals. Functional imaging studies using positron emission tomography or functional MRI have revealed a decrease in global cerebral blood flow of as much as 50 percent during hyperventilation (Bednarczyk et al., 1990; Posse et al., 1997; Ma¨kiranta et al., 2004), and although arterial oxygen pressure and cerebral oxygen extraction increase during hyperventilation, these compensatory mechanisms might, hypothetically, be insufficient in certain older individuals to overcome such a reduction in cerebral blood flow, resulting in a transient state of relative cerebral hypoxia. That such a state might have its earliest effects in the most tenuously vascularized regions of the cerebral cortex is plausible. In our patient, mild hyperventilation performed during his routine EEG did not elicit SREDA, whereas vigorous deep hyperventilation performed in hospital did, perhaps related to a greater effect on cerebral blood flow in the latter case. Moreover, hyperventilation-induced SREDA tended to be much longerlasting than spontaneous episodes of SREDA. It is also of interest that typical SREDA does not occur during sleep. In our patient, not a single episode of SREDA was recorded during five nights of sleep, in contrast to the multiple daily episodes recorded during wakefulness. One may speculate that a more stable cerebral blood flow during recumbent sleep could be responsible. One component of our source localization findings, however, is not so easy to reconcile with the vascular hypothesis, and that is the involvement of the parietal opercular regions. These areas do not lie within the anterior– posterior circulation watershed zone, but are instead within the territory of the middle cerebral artery. We do not have a ready explanation for this finding. Whereas the vascular watershed hypothesis seems to be the most parsimonious explanation for hyperventilation sensitivity and the previously demonstrated ability to evoke SREDA with carotid compression and nitrogen breathing, one cannot easily envisage a unified vascular hypothesis for SREDA that would invoke the parietal operculum. It is possible that this finding might be an atypicality specific to our patient, despite the typical visual EEG appearance of his SREDA. On a technical level, it is possible that the earliest focal source localization to the left parietal operculum with time-domain LORETA may represent an artificial construct arising from the ‘‘jittering’’ nature of these earliest waveform deflections, remembering that the averaging process was performed on the parietal peak component occurring approximately 50 ms later. Nevertheless, the opercular activation is present throughout the peak SREDA activity and is also evident in the frequency-domain analysis. Replication of the source localization findings in other patients with SREDA will be needed to see if the opercular activation is a uniform feature.
Our patient was investigated extensively because of a question of unexplained clinical episodes, for which no definite diagnosis was found. After the first ambulatory EEG recording, and in the absence of any other identified explanation, it was considered that the electrographic SREDA discharges might be a neurophysiologic correlate of the clinical events, notwithstanding the current consensus that these discharges are benign and unassociated with clinical signs or symptoms. Although suggestive evidence that SREDA may not always be entirely subclinical came from the second ambulatory EEG recording, a clinical component could not be definitively proven with direct testing during the SREDA events in hospital. SREDA is evidently subclinical from the patient’s perspective, as our patient never once activated the EEG event marker. Nevertheless, given the presumptive mechanism of mild relative cerebral hypoxia, it may be that rare SREDA episodes could be associated with minor transient behavioral alterations. However, there is certainly no evidence that SREDA represents an epileptic phenomenon (Westmoreland and Klass, 1981, 1997; Thomas et al., 1992; O’Brien et al., 1998). One limitation of our source localization results that should be acknowledged is the low number of scalp electrodes (29) used for recording, as the localization accuracy of LORETA and other distributed source models has been shown to increase substantially up to 64 electrodes. However, we have previously demonstrated that a low number of electrodes (23) is sufficient for the SNPM method of LORETA to reliably localize even small areas of increased activity associated with mesial temporal interictal epileptiform discharges, as validated by simultaneous intracranial EEG recording (Zumsteg et al., 2005a, 2006). Finally, the association between cerebral vascular supply and SREDA seems obvious but nevertheless cannot be considered proven. Future studies of such patients using simultaneous EEG and functional MRI will hopefully provide further insight into this unusual EEG pattern. References Bednarczyk EM, Rutherford WF, Leisure GP, Munger MA, Panacek EA, Miraldi FD, et al. Hyperventilation-induced reduction in cerebral blood flow: assessment by positron emission tomography. Ann Pharmacother 1990;24:456–60. De Falco FA, Vacca G, Fels A, Natale S, Striano S. An unusual EEG pattern in elderly subjects: subclinical rhythmic EEG discharge of adults (‘‘SREDA’’). Electroclinical study of six cases. Acta Neurol 1983;5:373–9. Fleming WE, Avidan A, Malow BA. Subclinical rhythmic electrographic discharge of adults (SREDA) in REM sleep. Sleep Med 2004;5:77–81. Frei E, Gamma A, Pascual-Marqui R, Lehmann D, Hell D, Vollenweider FX. Localization of MDMA-induced brain activity in healthy volunteers using low resolution brain electromagnetic tomography (LORETA). Hum Brain Mapp 2001;14:152–65. Ma¨kiranta M, Ruohonen J, Suominen K, Sonkaja¨rvi E, Saloma¨ki T, Kiviniemi V, et al. BOLD-contrast functional MRI signal changes related to intermittent rhythmic delta activity in EEG during voluntary hyperventilation-simultaneous EEG and fMRI study. Neuroimage 2004;22:222–31.
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263 Miller CR, Westmoreland BF, Klass DW. Subclinical rhythmic EEG discharge of adults (SREDA): further observations. Am J EEG Technol 1985;25:217–24. Nagarajan L, Gregory PB, Hewitt IK, Gubbay SS, Parry TS. Subclinical rhythmic EEG discharge of adults: SREDA in two children. Pediatr Neurol 2001;24:313–6. Naquet R, Franck G, Vigouroux R. Donne´es nouvelles sur certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital rencontre´es chez l’homme. Zentralbl Neurochir 1965;25:153–80. Naquet R, Louard C, Rhodes J, Vigouroux M. A propos de certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital. Leur activation par l’hypoxie. Rev Neurol 1961;105:203–7. Nichols TE, Holmes AP. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum Brain Mapp 2002;15:1–25. O’Brien TJ, Sharbrough FW, Westmoreland BF, Busaker NE. Subclinical rhythmic electrographic discharges of adults (SREDA) revisited: a study using digital EEG analysis. J Clin Neurophysiol 1998;15:493–501. Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain. Int J Psychophysiol 1994;18:49–65. Pascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MCG, Hell D, et al. Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia. Psychiatry Res 1999;90:169–79. Pascual-Marqui RD. Review of methods solving the EEG inverse problem. Int J Bioelectromagn 1999;1:75–86.
2263
Posse S, Olthoff U, Weckesser M, Ja¨ncke L, Mu¨ller-Ga¨rtner H-W, Dager S. Regional dynamic signal changes during controlled hyperventilation assessed with blood oxygen level-dependent functional MR imaging. Am J Neuroradiol 1997;18:1763–70. Silva DF, Zanoteli E, Lima MM, Anghinah R, Lima JG. Subclinical rhythmic electrographic discharge in adults: an atypical evolution. Arq Neuropsiquiatr 1995;53:266–9. Thomas P, Migneco O, Darcourt J, Chatel M. Single photon emission computed tomography study of subclinical rhythmic electrographic discharge in adults. Electroencephalogr Clin Neurophysiol 1992;83:223–7. Westmoreland BF, Klass DW. A distinctive rhythmic EEG discharge of adults. Electroencephalogr Clin Neurophysiol 1981;51:186–91. Westmoreland BF, Klass DW. Unusual variants of subclinical rhythmic electrographic discharge of adults (SREDA). Electroencephalogr Clin Neurophysiol 1997;102:1–4. Zumsteg D, Friedman A, Wennberg RA, Wieser HG. Source localization of mesial temporal interictal epileptiform discharges: correlation with intracranial foramen ovale electrode recordings. Clin Neurophysiol 2005a;116:2810–8. Zumsteg D, Wennberg RA, Treyer V, Buck A, Wieser HG. H215O- or 13NH3-PET and low resolution electromagnetic tomography (LORETA) during partial status epilepticus. Neurology 2005b;65:1657–60. Zumsteg D, Friedman A, Wieser HG, Wennberg RA. Source localization of interictal epileptiform discharges: comparison of three different techniques to improve signal to noise ratio. Clin Neurophysiol 2006;117:562–71.
Parietal lobe source localization and sensitivity to hyperventilation in a patient with subclinical rhythmic electrographic discharges of adults (SREDA) Dominik Zumsteg, Danielle M. Andrade, J. Martin Del Campo, Richard Wennberg
*
Division of Neurology, Krembil Neuroscience Centre, University of Toronto, Toronto Western Hospital, 399 Bathurst St., Toronto, Ont., Canada M5T 2S8 Accepted 10 July 2006 Available online 23 August 2006
Abstract Objective: Subclinical rhythmic electrographic discharges of adults (SREDA) is currently considered a benign EEG pattern of uncertain significance. The underlying cortical sources and generating mechanisms are unknown. We performed a source localization analysis of SREDA with the aim of better understanding this unusual EEG pattern. Methods: Multiple spontaneous episodes of typical SREDA were recorded in a patient during continuous EEG monitoring. Additional SREDA episodes were induced by hyperventilation. Source localization was carried out using statistical non-parametric mapping (SNPM) of low resolution electromagnetic tomography (LORETA). Results: SNPM of both time- and frequency-domain LORETA revealed a widespread biparietal cortical origin of SREDA, the anatomical distribution of which included the parietal operculum and the known vascular watershed areas between anterior, middle and posterior cerebral arteries. Vigorous deep hyperventilation induced SREDA on three of four attempts. Mean duration of the hyperventilationinduced SREDA was approximately three times longer than spontaneous events. Conclusions: Investigations in this patient with typical SREDA revealed hyperventilation sensitivity and a posterior hemispheric source localization maximal in the parietal cortex bilaterally, in large part overlying the anatomical distribution of the vascular watershed areas. Significance: The source localization results and sensitivity to hyperventilation suggest some sort of association between cerebral vascular supply and SREDA, as originally proposed by Naquet et al. [Naquet R, Louard C, Rhodes J, Vigouroux M. A propos de certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital. Leur activation par l’hypoxie. Rev Neurol 1961;105:203–207.]. Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: SREDA; EEG; Epilepsy; Electroencephalography; Source localization; LORETA
1. Introduction Subclinical rhythmic electrographic discharges of adults (SREDA) is a rare and distinctive EEG pattern seen primarily in older people, usually over the age of 50 years (mean age 62 years, range 35–98 years in a large series reported by Westmoreland and Klass, 1997). The pattern in its most typical form has a duration of 40–80 s and is characterized by rhythmic, sharply contoured 5–7 Hz theta waves, often with intermingled upper delta/lower theta fre*
Corresponding author. Tel.: +1 416 603 5402; fax: +1 416 603 5768. E-mail address: [email protected] (R. Wennberg).
quency components, the amplitude ranging from 40 to 100 lV with a widespread and bilaterally synchronous distribution maximal over the parietal regions (Westmoreland and Klass, 1981; De Falco et al., 1983; Miller et al., 1985; O’Brien et al., 1998). SREDA may have an abrupt onset, or, in about half of cases, the pattern may begin with one or more monophasic discharges of high amplitude, followed one to several seconds later by the typical predominant theta discharge. SREDA is mostly observed during wakefulness and tends to occur more than once during a single EEG recording, particularly during or shortly after hyperventilation, and recurs in subsequent EEGs obtained from the same person (Westmoreland and Klass, 1981;
1388-2457/$32.00 Ó 2006 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2006.07.137
2258
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
Miller et al., 1985). Unusual or atypical forms with predominant delta frequencies, notched waveforms, more focal distribution, atypical evolution and presence in younger individuals or during sleep have been described (Silva et al., 1995; Westmoreland and Klass, 1997; Nagarajan et al., 2001; Fleming et al., 2004). The typical pattern has not been associated with observable clinical changes or subjective symptoms and, although it may resemble a subclinical ictal epileptiform discharge, seems not to have any correlation with clinical seizures (Thomas et al., 1992; Westmoreland and Klass, 1997). SREDA is probably identical to an EEG pattern described by Naquet et al. (1961, 1965) as ‘‘paroxysmal discharges of the temporo-parieto-occipital junction’’. In addition to hyperventilation, these authors demonstrated that pure relative hypoxia (associated with nitrogen inhalation) or mild relative ischemia (from carotid artery compression) could reliably induce the paroxysmal pattern in susceptible patients. Accordingly, the occurrence of SREDA has been postulated to be associated with cerebrovascular disease by some authors (Naquet et al., 1961; Thomas et al., 1992), although others have not found epidemiological support for this hypothesis (Westmoreland and Klass, 1997). With the vascular hypothesis of Naquet et al. (1961) in mind, we performed a source localization analysis of SREDA with the aim of better understanding this unusual EEG pattern.
No clinical events were noted during the two days of monitoring. A therapeutic trial of sodium divalproex, 750 mg/day, was initiated, and relatives reported a possible slight decrease in frequency of the patient’s ‘‘disconnection’’ episodes over the subsequent month. However, a follow-up 48-h 21-channel ambulatory EEG at this time showed no decrease in frequency of the patient’s SREDA discharges, which occurred more than 20 times per day. One possible clinical event was witnessed by a family member during this recording period, wherein the patient appeared to not correctly identify a common object in his hand, followed by a short period (5–10 s) during which his conversation was not entirely of its normal character. The patient denied knowledge of anything unusual having happened during the entire recording. According to the family member’s recollection, the time of occurrence of this clinical episode, based on a wall clock, coincided with a 25 s epoch of SREDA on the ambulatory EEG. However, the event marker was not activated on the EEG headbox, such that definitive time synchronization was not possible. To further investigate the possibility of a clinical correlation to the SREDA episodes, the patient was admitted to hospital five months later for continuous video-EEG recording. The sodium divalproex had been tapered and discontinued during the previous months. The patient’s family reported a decrease in frequency of the clinical episodes since the initial investigations.
2. Patient and methods
2.2. Video-EEG recording and LORETA analysis
2.1. Case report
In-hospital EEG recording was performed using an XLTEK (Oakville, ON, Canada) recording system with a digitization rate of 200 Hz. Twenty-nine conventional AgCl surface electrodes were placed according to the international 10–20 electrode system, with additional electrodes placed at F9/10, T9/10, P9/10, supraorbitally, and over the zygomatic arches. The reference electrode was placed between Fz and Cz, the ground electrode at Fpz. The band pass was set at 0.1 to 70 Hz. Twenty-five episodes of SREDA were recorded during 24 h of monitoring (see Section 3). Time-domain LORETA analysis was carried out on 64 suitable EEG segments from SREDA episodes (free of movement and eye artifacts). These episodes were identified off-line by visual inspection of the EEG using INSIGHT software (Persyst Development Corporation, Prescott, AZ, USA; 2004) and then subjected to alignment (centered on parietal peak activity (P3, Pz, P4) of the SREDA discharges using longitudinal bipolar and referential derivations, 1000 to +1000 ms). Frequency-domain LORETA analysis was carried out on cross spectra of 72 suitable EEG segments of 2.5 s length (512 sample points) from SREDA episodes (free of movement and eye artifacts; not identical to those used for time-domain analysis). Cross spectra of normalized EEG segments were calculated using the LORETA software package (KEY Institute, Zu¨rich).
A 65-year-old man presented with a three-month history of brief unexplained episodes of ‘‘disconnection’’, which, according to his relatives, were characterized by subtle motor arrest and inappropriate speech. The patient was subjectively unaware of the episodes, which had occurred once every 2–3 weeks and were described to last 5–15 s. There was no history of neurological or cardiovascular disease, and the patient had no cardiovascular risk factors. Neurological examination was normal. A routine outpatient EEG including hyperventilation was normal. Brain MRI (FLAIR sequence) revealed a few non-specific focal high signal abnormalities in the hemispheric white matter bilaterally, interpreted as probable microangiopathic changes (Fig. 1). Carotid Doppler studies were unremarkable. Continuous 48-h 21-channel ambulatory EEG recording documented 28 separate episodes of typical SREDA (13 on day one and 15 on day two), all of which occurred during wakefulness. The EEG was otherwise normal during wakefulness, drowsiness and sleep. The mean SREDA episode duration was 51 s (range 25–100 s). An EKG lead showed no significant changes during the SREDA discharges, although many of the episodes were associated with a mild increase in heart rate of less than 10–15 beats per minute.
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
2259
Fig. 1. Axial brain MRI images (FLAIR) showing few non-specific subcortical punctate bilateral white matter hyperintensities.
These EEG segments and cross spectra were available for source analysis with low resolution electromagnetic tomography (LORETA; Pascual-Marqui et al., 1994, 1999; Frei et al., 2001). The selected segments were subjected to statistical non-parametric mapping (SNPM) of LORETA, as previously described (Zumsteg et al., 2005a,b, 2006). LORETA values represent the power (the squared magnitude of the computed intracerebral current density) within 2394 voxels of the cortex including the amygdala and the hippocampus, corresponding to a voxel size of about 7 mm (Pascual-Marqui et al., 1994; Pascual-Marqui, 1999). The grid in use is based on the digitized Talairach atlas, as provided by the Brain Imaging Center of the Montreal Neurological Institute. The SNPM method calculates the t-values using the permutation approach and offers, for each voxel, the exact (i.e., corrected for multiple comparisons) critical t-values for p < 0.05 and p < 0.01 as well as the exact (i.e., corrected for multiple comparisons) randomization probabilities for testing cluster size (Nichols and Holmes, 2002). 3. Results Twenty-two spontaneous episodes of SREDA (mean duration 53 s; range 15–137 s) were recorded during the video-EEG investigation, the discharges characterized by widespread generalized rhythmic theta activity between 4 and 7 Hz, of medium or medium–high amplitude, with relatively abrupt onset and offset and with no electrographic changes after offset. Some of the longer events showed intermingled delta frequency slow wave activity, especially during the later stages. Occasionally the episodes were preceded (and at times followed) by a small number of low or low–medium amplitude periodic slow waves or slow sharp waves appearing maximal over the mid-parietal regions. Fig. 2 depicts a representative example. During three of these spontaneous SREDA episodes, the patient could be
seen and heard on the simultaneous video recording to be conversing normally either on the telephone or with a nurse in his room. He was able to remember and accurately describe these conversations when questioned later. During two other spontaneous SREDA episodes, the patient was able to respond appropriately to questions from a physician in the room, with complete subsequent recollection of the interaction. Attempts to induce SREDA with vigorous deep hyperventilation were carried out four times over a 30 min period, with SREDA discharges elicited following three of the four attempts (mean duration 158 s; range 40–225 s). The hyperventilation was stopped in each case upon recognition of SREDA onset in the EEG. With the first hyperventilation a 40 s episode of typical SREDA started 60 s after commencement of hyperventilation. With the second hyperventilation a longer SREDA event started 90 s after commencement of hyperventilation, lasting 210 s. With the third hyperventilation (2 min) no SREDA occurred. With the last repetition of hyperventilation SREDA started 75 s after commencement of hyperventilation, lasting for 225 s. The three hyperventilation-induced SREDA episodes had onset within 27 min of each other. In comparison, six spontaneous SREDA episodes occurred during the six hours after awakening (median one SREDA/hour), before hyperventilation. The video-EEG monitoring was stopped 10 min after the final hyperventilation, with no further recorded SREDA. During each of the hyperventilation-induced SREDA episodes the patient was fully conscious and able to interact with the physicians in the room as well as perform bedside neuropsychological tests including correctly drawing a clock face and putting the clock hands on the figure, accurately drawing a cube, bisecting a line correctly, correctly performing serial 7 s and serial 3 s, as well as repeating and remembering three words and naming the days of
2260
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
Fig. 2. Typical episode of spontaneous SREDA, approximately 60 s duration, recorded during in-hospital investigation. Anterior–posterior bipolar montage. Bandpass 0.5–70 Hz.
the week and months of the year sequentially. The only questionable finding was that during the third hyperventilation-induced SREDA episode he correctly repeated a five-digit span forward but would not attempt to produce the numbers in reverse order. He had full recollection of the tests performed during the hyperventilationinduced SREDA episodes. Symptomatically, he felt only lightheadedness as would be expected with the extreme hyperventilation. Time-domain (Fig. 3) and frequency-domain (Fig. 4) source localization using SNPM of LORETA revealed a significant increase of cortical activity over both posterior hemispheres, maximal over the parietal regions. The area of significant increase extended to the parietal operculum and parieto-occipital border, with relative sparing of the parasagittal region on both sides. There was no clear evolution of the spectral pattern with respect to frequency over the course of the SREDA episodes (Fig. 4A). Likewise, there were no significant differences in cortical activation patterns obtained during early (first third) and late (last third) phases of SREDA episodes, neither with time-domain (not shown) nor with frequency-domain analysis (Fig. 4). No significant differences in cortical activation patterns were found between SREDA episodes occurring spontaneously or induced by hyperventilation. 4. Discussion In their initial studies of the EEG discharges that would later come to be known as SREDA, Naquet et al. (1961, 1965) proposed a vascular etiology, based on identification
of the EEG pattern in older individuals, some with other evidence of cerebral vascular disease, as well as localization of the discharges to the vascular watershed area and the ability to reliably reproduce the EEG pattern in these individuals by hyperventilation, nitrogen inhalation and carotid artery compression. The common mechanism proposed for these three activation techniques was relative cerebral hypoxia: ‘‘pure’’ in the case of nitrogen inhalation; ischemic in the case of carotid compression; and associated with decreased cerebral blood flow caused by vasoconstriction in response to hypocapnia in the case of hyperventilation (Naquet et al., 1961). However, Westmoreland and Klass (1981) found that the incidence of cerebral vascular disease, seizures and other neurologic disorders in a large series of patients with SREDA was no greater than that in a control group of age matched subjects with normal EEGs. SREDA is currently viewed as a benign non-specific EEG phenomenon of unknown cause and mechanism (Westmoreland and Klass, 1997; O’Brien et al., 1998). Our source localization results demonstrate a clear regional predominance for the generation of SREDA discharges localized to the biparietal areas, in large part overlying the known anatomical distribution of the vascular watershed areas between anterior, middle and posterior cerebral arteries. The source localization results suggest some sort of relationship between cerebral vascular supply and SREDA, as initially proposed by Naquet et al. (1961). The sensitivity to hyperventilation is also consistent with the vascular hypothesis as originally posed by Naquet et al. (1961). It seems unlikely that hyperventilation-induced respiratory alkalosis and hypocapnia would be themselves
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
2261
Fig. 3. Top left, overlay of 64 individual SREDA discharges in a bipolar montage, aligned to the peak activity of the discharge (red line). Middle, averaged EEG of the 64 individual SREDA discharges depicted at left using a different time scale (above) and spatiotemporal source maps showing the color coded SNPM LORETA t-values of individual Brodmann areas (below). Right, three-dimensional time-domain SNPM LORETA reconstructions at peak activity of the SREDA discharge (red lines A, B, and C from middle plots) representing distributed, color coded t-values (red for positive t-values above level of significance) projected onto generic brain templates.
Fig. 4. (A) Spectrogram [window length 2.5 s, Kaiser window, 90% overlap] of a single representative SREDA episode showing an irregular increase of spectral density power in lower delta and theta EEG frequency bands during the course of the pattern. Note that, with respect to frequency, there is no clear evolution of the spectral pattern over the course of the SREDA event. (B) Three-dimensional frequency domain (theta cross spectra) SNPM LORETA reconstructions for three different periods of 10 SREDA episodes representing distributed, color coded t-values (red for positive t-values above level of significance) projected onto generic brain templates. Early, 21 epochs of 2.5 s (512 sample points) obtained from the first third of the SREDA events; all, 72 epochs obtained from the entire course of the SREDA events; late, 26 epochs of the last third of the SREDA events. Note that there is no significant difference of cortical activation patterns for the three periods analyzed.
2262
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263
the triggers for SREDA, given the previously demonstrated ability to similarly induce SREDA with nitrogen breathing, such that the more likely inciting factor associated with hyperventilation is a mild relative hypoxia due to cerebral vasoconstriction and decreased cerebral blood flow in susceptible individuals. Functional imaging studies using positron emission tomography or functional MRI have revealed a decrease in global cerebral blood flow of as much as 50 percent during hyperventilation (Bednarczyk et al., 1990; Posse et al., 1997; Ma¨kiranta et al., 2004), and although arterial oxygen pressure and cerebral oxygen extraction increase during hyperventilation, these compensatory mechanisms might, hypothetically, be insufficient in certain older individuals to overcome such a reduction in cerebral blood flow, resulting in a transient state of relative cerebral hypoxia. That such a state might have its earliest effects in the most tenuously vascularized regions of the cerebral cortex is plausible. In our patient, mild hyperventilation performed during his routine EEG did not elicit SREDA, whereas vigorous deep hyperventilation performed in hospital did, perhaps related to a greater effect on cerebral blood flow in the latter case. Moreover, hyperventilation-induced SREDA tended to be much longerlasting than spontaneous episodes of SREDA. It is also of interest that typical SREDA does not occur during sleep. In our patient, not a single episode of SREDA was recorded during five nights of sleep, in contrast to the multiple daily episodes recorded during wakefulness. One may speculate that a more stable cerebral blood flow during recumbent sleep could be responsible. One component of our source localization findings, however, is not so easy to reconcile with the vascular hypothesis, and that is the involvement of the parietal opercular regions. These areas do not lie within the anterior– posterior circulation watershed zone, but are instead within the territory of the middle cerebral artery. We do not have a ready explanation for this finding. Whereas the vascular watershed hypothesis seems to be the most parsimonious explanation for hyperventilation sensitivity and the previously demonstrated ability to evoke SREDA with carotid compression and nitrogen breathing, one cannot easily envisage a unified vascular hypothesis for SREDA that would invoke the parietal operculum. It is possible that this finding might be an atypicality specific to our patient, despite the typical visual EEG appearance of his SREDA. On a technical level, it is possible that the earliest focal source localization to the left parietal operculum with time-domain LORETA may represent an artificial construct arising from the ‘‘jittering’’ nature of these earliest waveform deflections, remembering that the averaging process was performed on the parietal peak component occurring approximately 50 ms later. Nevertheless, the opercular activation is present throughout the peak SREDA activity and is also evident in the frequency-domain analysis. Replication of the source localization findings in other patients with SREDA will be needed to see if the opercular activation is a uniform feature.
Our patient was investigated extensively because of a question of unexplained clinical episodes, for which no definite diagnosis was found. After the first ambulatory EEG recording, and in the absence of any other identified explanation, it was considered that the electrographic SREDA discharges might be a neurophysiologic correlate of the clinical events, notwithstanding the current consensus that these discharges are benign and unassociated with clinical signs or symptoms. Although suggestive evidence that SREDA may not always be entirely subclinical came from the second ambulatory EEG recording, a clinical component could not be definitively proven with direct testing during the SREDA events in hospital. SREDA is evidently subclinical from the patient’s perspective, as our patient never once activated the EEG event marker. Nevertheless, given the presumptive mechanism of mild relative cerebral hypoxia, it may be that rare SREDA episodes could be associated with minor transient behavioral alterations. However, there is certainly no evidence that SREDA represents an epileptic phenomenon (Westmoreland and Klass, 1981, 1997; Thomas et al., 1992; O’Brien et al., 1998). One limitation of our source localization results that should be acknowledged is the low number of scalp electrodes (29) used for recording, as the localization accuracy of LORETA and other distributed source models has been shown to increase substantially up to 64 electrodes. However, we have previously demonstrated that a low number of electrodes (23) is sufficient for the SNPM method of LORETA to reliably localize even small areas of increased activity associated with mesial temporal interictal epileptiform discharges, as validated by simultaneous intracranial EEG recording (Zumsteg et al., 2005a, 2006). Finally, the association between cerebral vascular supply and SREDA seems obvious but nevertheless cannot be considered proven. Future studies of such patients using simultaneous EEG and functional MRI will hopefully provide further insight into this unusual EEG pattern. References Bednarczyk EM, Rutherford WF, Leisure GP, Munger MA, Panacek EA, Miraldi FD, et al. Hyperventilation-induced reduction in cerebral blood flow: assessment by positron emission tomography. Ann Pharmacother 1990;24:456–60. De Falco FA, Vacca G, Fels A, Natale S, Striano S. An unusual EEG pattern in elderly subjects: subclinical rhythmic EEG discharge of adults (‘‘SREDA’’). Electroclinical study of six cases. Acta Neurol 1983;5:373–9. Fleming WE, Avidan A, Malow BA. Subclinical rhythmic electrographic discharge of adults (SREDA) in REM sleep. Sleep Med 2004;5:77–81. Frei E, Gamma A, Pascual-Marqui R, Lehmann D, Hell D, Vollenweider FX. Localization of MDMA-induced brain activity in healthy volunteers using low resolution brain electromagnetic tomography (LORETA). Hum Brain Mapp 2001;14:152–65. Ma¨kiranta M, Ruohonen J, Suominen K, Sonkaja¨rvi E, Saloma¨ki T, Kiviniemi V, et al. BOLD-contrast functional MRI signal changes related to intermittent rhythmic delta activity in EEG during voluntary hyperventilation-simultaneous EEG and fMRI study. Neuroimage 2004;22:222–31.
D. Zumsteg et al. / Clinical Neurophysiology 117 (2006) 2257–2263 Miller CR, Westmoreland BF, Klass DW. Subclinical rhythmic EEG discharge of adults (SREDA): further observations. Am J EEG Technol 1985;25:217–24. Nagarajan L, Gregory PB, Hewitt IK, Gubbay SS, Parry TS. Subclinical rhythmic EEG discharge of adults: SREDA in two children. Pediatr Neurol 2001;24:313–6. Naquet R, Franck G, Vigouroux R. Donne´es nouvelles sur certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital rencontre´es chez l’homme. Zentralbl Neurochir 1965;25:153–80. Naquet R, Louard C, Rhodes J, Vigouroux M. A propos de certaines de´charges paroxystiques du carrefour temporo-parie´to-occipital. Leur activation par l’hypoxie. Rev Neurol 1961;105:203–7. Nichols TE, Holmes AP. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum Brain Mapp 2002;15:1–25. O’Brien TJ, Sharbrough FW, Westmoreland BF, Busaker NE. Subclinical rhythmic electrographic discharges of adults (SREDA) revisited: a study using digital EEG analysis. J Clin Neurophysiol 1998;15:493–501. Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain. Int J Psychophysiol 1994;18:49–65. Pascual-Marqui RD, Lehmann D, Koenig T, Kochi K, Merlo MCG, Hell D, et al. Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia. Psychiatry Res 1999;90:169–79. Pascual-Marqui RD. Review of methods solving the EEG inverse problem. Int J Bioelectromagn 1999;1:75–86.
2263
Posse S, Olthoff U, Weckesser M, Ja¨ncke L, Mu¨ller-Ga¨rtner H-W, Dager S. Regional dynamic signal changes during controlled hyperventilation assessed with blood oxygen level-dependent functional MR imaging. Am J Neuroradiol 1997;18:1763–70. Silva DF, Zanoteli E, Lima MM, Anghinah R, Lima JG. Subclinical rhythmic electrographic discharge in adults: an atypical evolution. Arq Neuropsiquiatr 1995;53:266–9. Thomas P, Migneco O, Darcourt J, Chatel M. Single photon emission computed tomography study of subclinical rhythmic electrographic discharge in adults. Electroencephalogr Clin Neurophysiol 1992;83:223–7. Westmoreland BF, Klass DW. A distinctive rhythmic EEG discharge of adults. Electroencephalogr Clin Neurophysiol 1981;51:186–91. Westmoreland BF, Klass DW. Unusual variants of subclinical rhythmic electrographic discharge of adults (SREDA). Electroencephalogr Clin Neurophysiol 1997;102:1–4. Zumsteg D, Friedman A, Wennberg RA, Wieser HG. Source localization of mesial temporal interictal epileptiform discharges: correlation with intracranial foramen ovale electrode recordings. Clin Neurophysiol 2005a;116:2810–8. Zumsteg D, Wennberg RA, Treyer V, Buck A, Wieser HG. H215O- or 13NH3-PET and low resolution electromagnetic tomography (LORETA) during partial status epilepticus. Neurology 2005b;65:1657–60. Zumsteg D, Friedman A, Wieser HG, Wennberg RA. Source localization of interictal epileptiform discharges: comparison of three different techniques to improve signal to noise ratio. Clin Neurophysiol 2006;117:562–71.