Early Thalamic Injury Associated with Epilepsy and Continuous Spike–Wave during Slow Sleep


Address correspondence and reprint requests to Dr. F. Guzzetta at Division of Child Neurology and Psychiatry, School of Medicine, Catholic University, Largo Gemelli I-00168 Rome, Italy. E-mail: fguzzetta@rm.unicatt.it


Summary: Purpose: Mechanisms inducing continuous spike–wave during slow sleep (CSWS) in encephalopathy with electrical status epilepticus during sleep are still unclear. Recently, some sporadic cases with early thalamic injury associated with CSWS have been reported. The aim of the study was to investigate in a population of patients with an early thalamic injury the presence of an activation of paroxysmal activities during sleep, their characteristics, and possible relations to neuroimaging and neuropsychological features.

Methods: Thirty-two patients with prenatal or perinatal thalamic injuries, mostly due to a vascular mechanisms, were fully examined, including neuroimaging, EEG monitoring, and cognitive follow-up.

Results and Conclusions: Twenty-nine of 32 patients showed major sleep EEG activation. Among these 29 patients, two different groups were distinguished: the first included the more or less typical CSWS (12 cases), generally with symmetry of spike and waves (SWs) and often with no spindle at all. The other cases had an usual asymmetry of SWs and presence or reduction of spindles, plus other atypical features concerning synchronism and morphology of SWs. Behavioral disorders were significantly more present in patients with a true CSWS; their improvement (and in one case of the three thoroughly followed the improvement of cognitive competence) paralleled the disappearance of CSWS. The generally predominant injury of the lateral aspect of the thalamus included reticular nucleus and ventral nuclei. An imbalance of γ-aminobutyric acid (GABA)B- versus GABAA -mediated receptors may be evoked as a cofactor predisposing to CSWS.

Epilepsy with continuous spike–waves (SWs) during slow sleep (CSWS) is an acknowledged epileptic syndrome characterized by a combination of focal epilepsy, neuropsychological impairment, and typical EEG findings with a pattern of diffuse SWs occurring in ≤85% of slow sleep (1). Recently, the possible role of primary early thalamic injury in generating CSWS has been suggested (2), and sporadic cases with similar lesions have been reported (3–8). Morphologic and metabolic thalamic abnormalities have been described in CSWS as well (9). Our study aimed at clarifying (a) the possible presence and eventually the characteristics of an activation of paroxysmal activities during sleep in a population of patients with an early thalamic injury; and (b) their relation to findings from neuroimaging, with thalamic injuries, and with neuropsychological features.


All the patients aged between 4 and 12 years (when CSWS usually develops), showing at magnetic resonance imaging (MRI) an early acquired thalamic lesion and consequently admitted to our hospital between 1999 and 2003, were enrolled in the study. Indication for MRI resulted from the clinical problems suggesting hospital admission [i.e., paroxysmal manifestations (19 cases) or follow-up of newborns at risk (13 cases)].

Brain injury co-involving thalamus was generally determined by an ischemic or hemorrhagic prenatal or perinatal insult; in only three cases were brain lesions sequelae of an infectious disease that occurred in the first months of life. All patients underwent a full clinical assessment that included a neuropsychological assessment, an EEG recording session with at least two sleep recordings performed over a 1-month period, and MRI. In all cases, apart from two without epilepsy, the follow-up was calculated from the onset of seizures to the last observation that had been carried out in our department.

Seizure types and epilepsies were classified according to current international classifications (10,11). To assess the severity of epilepsy, we considered both the frequency and the intensity of seizures, as well as the seizure type and the duration of the active period. The most severe degree (+++) consisted of: daily seizures, drop attacks, secondary generalization, status, and polymorphic seizures. The degree immediately below (++) consisted of: weekly to monthly seizures, absence of drop attacks, no status, and possible polymorphic seizures. When the second degree (++) lasted <1 year, we rated epilepsy with only one plus (+)


All patients underwent MRI on a 1.5-Tesla MR system (Horizon Echospeed/Excite; General Electric, Milwaukee, WI, U.S.A.). MRI was performed in all patients by using a high-resolution protocol, consisting of: (a) sagittal spin-echo T1-weighted sequence (TR, 650 ms; TE, 15 ms; two excitations; 24-cm field of view; 4-mm thickness; 10% intersections gap; 256 × 256 acquisition matrix); (b) axial and coronal fast spin–echo T2-weighted sequences (TR, 2,750 ms; TE, 100 ms; two excitations; 24-cm field of view; 3-mm thickness; 10% intersections gap; 256 × 320 acquisition matrix).

MRI was repeated several times (from two to five times), especially when the first examinations did not allow us to evaluate the lesions properly. Neuroimaging records included the description of thalamic lesions according to the involvement of different nuclei, as well as regarding any other possible brain lesion. An analysis of the thalamus was performed by defining, whenever possible, the ischemic and/or hemorrhagic character of the lesion and its location. Because of possible image distortions caused by the different types of destructive injury, especially when hydrocephalus was present, the location was only grossly defined as regarding (a) the lateral aspect of the thalamus including the reticular nucleus (nRT), (b) the medial nuclear mass, including the mediodorsal nucleus, (c) the ventral nuclear mass comprising the ventral posterior, lateral and anterior nuclei; and (d) the lateral-posterior nuclear mass (the pulvinar). The last two categories may possibly include the intralaminar nucleus.

EEG recording and polysomnography

Videopolygraphic study was performed by using 21 EEG electrodes according to the 10/20 International System. In the youngest infants, 11 electrodes were used. Deltoid surface electromyogram (EMG) also was recorded.

Every patient underwent numerous recordings when awake (≤24) and repeated, sometimes prolonged, sleep recordings (≤16); 14 patients had at least one nocturnal polysomnography video-EEG recording lasting ≥24 h.

Two groups of patients on the base of the duration of slow sleep abnormalities were distinguished: those with >85% SWs (CSWS) and those with <85% (not classifiable as CSWS). The former were separated into typical and atypical CSWS. Typical CSWS was characterized by a pattern persisting for more than a month, consisting of diffuse, rhythmic, synchronous, and symmetrical SWs occurring in >85% of slow sleep, with typical morphology and with the absence of physiologic sleep figures (Fig. 1). In atypical CSWS, the morphology of the anomalies was variable, consisting of SWs associated with spikes (S), polyspikes (SS), sharp waves, often being asynchronous, asymmetrical, or even unilateral and nonrhythmic (Fig. 2). In the group without CSWS, the anomalies were more heterogeneous, often multifocal, noncontinuous (occupying ≥50% of slow sleep), and sometimes similar to those seen in Lennox–Gastaut syndrome; the physiologic sleep elements were usually present. We named these aspects “sleep overactivation” of epileptic paroxysms (Fig. 3).

Figure 1.

Case 2. a: Awake EEG: focal right centrotemporal spikes and synchronous bilateral frontocentrotemporal SW discharges. b: Diffuse continuous SW during slow sleep with typical SW morphology (synchronous, rhythmic, and symmetrical). c: Well evident at 400-mV sensitivity.

Figure 2.

Case 4. a: Awake EEG: multifocal spikes. b: Continuous irregular, asynchronous spikes and SW during slow sleep. c: More evident at 200-mV sensitivity.

Figure 3.

Case 13. a: Awake EEG: strongly asymmetrical background activity with left parietooccipitotemporal spikes. b, c: Subcontinuous spikes and sharp waves on the left hemisphere and polyspikes discharges predominant on the left regions.

Neuropsychological assessment

Neuropsychological skills and cognitive development were assessed with techniques relevant to the age of the children examined [Griffiths developmental scales (12), WPPSI (13), WISC-R (14)], as well as with tests concerning specific abilities. Patients were classified according to the following criteria:

  • IQ or DQ >85, normal mental development

  • Between 75 and 85, borderline mental development

  • Between 50 and 75, mild mental retardation

  • Between 25 and 50, moderate mental retardation

  • <25, severe mental retardation.

The behavioral disturbances were defined as (a) severe (+++) when associated with major disorders in communication and in social interaction, with restricted and stereotyped patterns of behavior as well as severe symptoms such as extreme hyperactivity and aggressiveness or self-injurious behavior; (b) moderate (++) when characterized by a lesser degree of these major behavioral troubles; and (c) mild (+) when minor social disorders and abnormal adaptive problems such as aggressiveness, hyperactivity, or major inhibition were present.


Main clinical data are reported in Table 1. Twenty-nine of the 32 patients showed a dramatic activation of paroxysmal activities during sleep, shaping a pattern of CSWS in 12 cases; only three were the patients with thalamic injury and without paroxysmal activation during sleep. The follow-up mean after the first seizure was 4 years 9 months (range, 7 months to 13 years 10 months). Most of the cases had a hemiplegia; nine cases had a double hemiplegia (cases 9, 11, 15. 20, 23, 25, 27, 28, and 30), and only one patient (case 7) had a spastic diplegia. Cases 14 and 22 were neurologically normal. All the patients but two (cases 31 and 32) had epilepsy; 12 of them were seizure free for at least the last year of the follow-up.

Table 1. Main clinical data
No.DiagnosisAge last
duration (yr. mo)
Development before
  1. PHH, posthemorrhagic hydrocephalus; IVH, intraventricular hemorrhage; PVL, periventricular leukomalacia; HIE, hypoxic–ischemic encephalopathy; MCA, medial cerebral artery; SPS, simple partial seizure; CPS, complex partial seizure; A, absence; SGP, secondarily generalized partial; GTCS, generalized tonic–clonic seizure; TS, tonic seizure; NC, neonatal convulsion; IS, infantile spasms; Drop att., drop attacks; FS, febrile seizures; Unilat.S, unilateral seizures; hemip., hemiparesis.

  2. aAfter hemispherectomy.

 1Neonatal stroke (MCA)16.10/13.10Left hemipMild RetCPS, SGP, A, Drop att.2 yr 3 mo+++Seizure free (3 yr)
 2Periventricular hemorrhagic infarction15.1/11.1Mild left hemipNormalCPS, SGP, GTCS, A4 yr+++Seizure free (2 yr)
 3Neonatal stroke (MCA)9.1/8.9Right hemipNormalSGP4 mo++Seizure free (1 yr)
 4PVL + IVH/PHH11.1/8.1Mild right hemipSevere RetSGP, A, Drop att.3 yr+++Several s./day
 5Thalamus hemor. + PVL + IVH/PHH11.1/9.2Right hemip, tremorNormalSPS, head drops23 mo+++Several s./day
 6Neonatal stroke (MCA)11.1/10.8Left hemipBorderlineI S, SPS, CPS5 mo++Some seizures/yr
 7Myelomeningocele7.7/3.2Spastic diplegiaMild RetNC, FS, CPS, SPS, GTCS4 yr 5 mo++Seizure free (1 yr 7 mo)
 8PHH + PVL7.2/2.9Left hemipBorderlineSGP4 yr 5 mo1 single statusSeizure free (2 yr 10 mo)
 9Postmeningitic hydrocephalus5.10/5.5Double hemipSevere RetSPS, CPS, SGP5 mo+++Several s./day
10HIE4.8/1.5Right hemipNormal (?)SPS, CPS (?)3 yr 3 mo+Three s. in a year
11Postmeningitic hydrocephalus6.11/6.2Double hemipSevere RetSPS, CPS, SGP9 mo++Monthly s.
12HIE4.6/0.7Left hemipMild RetCPS (?)3 yr 11 mo+Two s. in a year
13Neonatal stroke (MCA)5.1/2.5Right hemipNormalSPS, CPS2 yr 8 mo++Monthly s.
14HIE6.1/4.1NormalNormalUnilat. S2 yr3 statusSeizure free (7 mo)
15Periventricular hemorrhagic infarction + PHH7.7/7.8Double hemip, blindnessSevere RetTonic spasms1 day+Seizure free (6 yr)
16Neonatal stroke (MCA)15.1/9.8Right hemipMild RetCPS,SGP, TS, A, Drop att.5 yr 5 mo+++Several s./day
17Periventricular hemorrhagic infarction5.9/5.4Right hemipMild RetCPS, IS, Drop att.5 mo+++Seizure free (9 mo)a
18PHH6.1/5.7Left hemipMod-Sev RetI.S.6 mo++Seizure free (4 yr)
19Neonatal stroke (MCA)4.1/3.5Right hemip, hemianopsiaBorderlineSPS8 mo++Several s./day
20Grade 4 IVH, PHH6/5.5Double hemip (right+)Mod-Sev retSPS ++Seizure free (2 yr)
21Neonatal stroke (MCA)6/3.8Right hemipNormalSGP, CPS2 yr 4 mo+Seizure free (1 yr)
22PHH4.1/2.7NormalNormalCPS2 yr 6 mo+Monthly s.
23Postinfectious hydrocephalus7.4/6.10Double hemipNormalGTCS, SPS, CPS,6 mo+++Several s./day
24PVL + IVH/PHH7.5/5.10Left hemipNormalGTCS1 yr 7 mo+Seizure free (6 yr)
25HIE6/6Double hemipSevere RetNC, GTCS, SGP, IS1 day++Monthly s.
26Neonatal stroke (MCA)9/7.9Right hemipMild RetSPS, SGP1 yr 3 mo+Seizure free (5 yr)
27PVL + IVH/PHH7.3/5.11Double hemipModer RetIS1 yr 4 mo+Seizure free (5 yr)
28PVL + IVH/PHH9.9/9.8Double hemipSevere RetMal status, SGP, TS0 yr 1 mo+++Several s./day
29PHH, cerebellar hemorrhage7/3Left hemipMild RetCPS4 mo++Monthly s.
30PVL + IVH/PHH6.6/6.1Double hemipSevere RetIS, CPS5 mo++Monthly s.
31Neonatal stroke (MCA)6.11/1Right hemipNormalNo seizures
32Neonatal HIE5.11/2Left hemipMild RetNo seizures

At the onset of sleep EEG disorders, we avoided using antiepileptic drugs (AEDs) possibly supporting them, such as carbamazepine (CBZ), phenytoin (PHT), and phenobarbital (PB). CSWS was treated with usual specific drugs [valproate (VPA), ethosuximide (ESM), benzodiazepines (BZDs), and steroids[; two patients (cases 1 and 17) were given neurosurgical treatment: extensive corticectomy around a large porencephaly at 7 years and a hemispherectomy at 4 years, respectively.

Neuroimaging data

The results of MRI examination are shown in Table 2. Various cortical/subcortical injuries, consistent with different etiologies, were observed in all cases. The most common causes were ischemic or hemorrhagic in origin, with unilateral, or predominantly unilateral, locations. Vascular causes were common in 13 patients: with perinatal ischemic infarction in nine term newborns and with hemorrhagic venous infarction in four preterms. In other 15 patients, a more diffuse ischemic mechanism can be evoked, such as neonatal hypoxic–ischemic encephalopathy, periventricular leukomalacia, or posthemorrhagic hydrocephalus; the only case with a developmental disorder of the CNS (myelomeningocele) showed ischemic injury also, presumably linked to a vascular mechanism involving the posterior thalamic perforating arteries originating from the precommunicating portion of posterior cerebral artery (PCA). The remaining three patients had postmeningitic hydrocephalus.

Table 2. Neuroimaging
No.Thalamic lesionsCortical/subcortical injury location
SideInvolved nucleiLocationOrigin
  1. nRT, reticular nucleus; V, ventral mass; P, posterior mass; DM, dorsomedial nucleus; F, frontal; P, parietal; T, temporal; O, occipital; WM, white matter; PHI, periventricular hemorrhagic infarction; PHH, posthemorrhagic hydrocephalus; PVL, periventricular leukomalacia; IVH, intraventricular hemorrhage; PMH, postmeningitic hydrocephalus; HIE, hypoxic–ischemic encephalopathy; MCA, medial cerebral artery; MMC, myelomeningocele.

 1RnRT-V-P-DMRight F-T-PStroke (MCA)
 2RnRT-VRight periventricularPHI/PHH
 3LnRT-V-P-DMLeft F-P-T-OStroke (MCA)
 4BIL (L+)DM-PBil periventricPVL + IVH/PHH
 5LnRT-V-DMLeft periventricularPVL + IVH/PHH
 6RnRT-V-P-DMRight F-T-PStroke (MCA)
 7BIL (L+)nRT-V-P-DMLeft subcortical, right T-OPost. ischemia, MMC
 8RnRT-V-PPeriventricular WMPVL + IVH/PHH
 9RnRT-V-PPosterior WMInfectious/PMH
11LnRT-VLeft F-O-T, right P-T-O WMInfectious/PMH
12RnRT-V-PBil F-P-T-O, basal gangliaHIE
13LnRT-V-P-DMLeft F-T-PStroke (MCA)
14BIL (R+)Right V
Left nRT-V
Periventricular WMHIE
15BIL (L+)nRT-V-P-DMPeriventricular WMPHI/PHH
16LnRT-VLeft F-T-PtStroke (MCA)
17LnRT-V-P-DMLeft periventricular (F-P-O)PHI/PHH
18LnRT-V-P-DMBil PeriventricularPHI/PHH
19RnRT-V-DMF-T-P-O rightStroke (PCA)
21LnRT-V-PLeft T-P-OStroke (MCA)
23RnRT-V-PRight F-P-T-OInfectious/PMH
25BILnRT-VBilateral WMHIE
26LnRT-VL subcortical, basal gangliaStroke (MCA)
27RnRT-PRight F, diffuse WMPVL + IVH/PHH
28BIL (L+)nRT-V-PBilateral WMPVL + IVH/ PHH
29RPDiffuse WMCerebellar hemorrh, PHH
30RnRT-V-PPosteriorly diffusePVL + IVH/PHH
31LnRT-VLeft PStroke (MCA)
32BIL (L+)nRT-VLeft periventricular WMHIE

Characteristically, thalamic lesions were unilateral or predominantly unilateral, always ipsilateral to vascular infarction (Figs. 4 and 5). In eight of nine cases with bilateral thalamic injury, one side was predominantly involved (Fig. 6). Among thalamic nuclei, the lateral part of thalamus, including nRT, and the ventral mass were commonly injured; the other nuclei were differently affected according to the kind of injury. In one patient (case 15), the left thalamus was completely destroyed. (The percentage of the involvement for each considered part of thalamus is reported in Fig. 7.)

Figure 4.

Case 3. a: Axial fluid-attenuated inversion recovery image (FLAIR; TR, 8,800 ms; TI, 2,200 ms; TE, 140 ms). b: Axial turbo spin-echo T2-weighted image (TR, 2,750 ms; TE, 100 ms). The images show an extensive area of parenchymal loss involving the left temporooccipital lobes, with consequential dilation of the homolateral ventricle. Along the medial margins of the lesion, a mildly hyperintense signal on T2-weighted and FLAIR images is evident (arrow), involving the lateral aspect of the left thalamus, the reticular nucleus, and internal capsula, related to parenchymal gliotic changes.

Figure 5.

Case 5. a: Axial unenhanced computed tomography (CT) scan image. b: Axial spin-echo T1-weighted image (TR, 650 ms; TE, 15 ms). A focal hyperdense area (arrow), related to acute hemorrhage, is evident along the lateral aspect of the left thalamus, also involving the reticular nucleus. Coexistent hydrocephalic dilation of both ventricles, particularly evident on the left side. The MR examination, obtained after the ventriculoperitoneal cerebrospinal fluid shunt, confirms the parenchymal hemorrhagic injury (arrow) with perilesional edema. The inner white matter of the both cerebral hemispheres is reduced in thickness and mildly hyperintense on T2-weighted images with asymmetrical dilation of both ventricles (left>right), related to a previous diffuse parenchymal injury.

Figure 6.

Case 10. a: Axial fluid-attenuated inversion recovery (FLAIR) image (TR, 8,800 ms; TI, 2,200 ms; TE, 140 ms). b: Axial turbo spin-echo T2-weighted image (TR, 2,750 ms; TE, 100 ms). A focal area of mildly hyperintense signal on T2-weighted and FLAIR images is evident, involving the ventral aspect of the right thalamus (arrow) related to parenchymal gliotic changes. Along the lateral aspect of the ventral left thalamus, a small area with similar features is evident (small arrow), also involving the reticular nucleus. Some small parenchymal foci with similar features also are evident in the inner white matter of both cerebral hemispheres.

Figure 7.

Oblique dorsolateral view and medial coronal section of the thalamus, with the percentage of injury involvement of each area considered in the study.

Electrophysiologic results

EEG data are presented in Table 3. In all patients, focal and/or multifocal EEG abnormalities were observed, generally consistent with the location of cortical injuries. In all the 30 epilepsy patients but one (case 30), a strong activation of paroxysmal abnormalities was seen during slow sleep. Twelve patients had CSWS; three of these showed the typical form (patients 1–3), and nine, the atypical one (patients 4–12). The remaining 17 (patients 13–29) had only a “sleep overactivation.” Among them, two had features that were similar to Lennox–Gastaut syndrome, in particular, rapid recruiting rhythms and low frequency of SW (patients 16,17). Cases 31 and 32 had neither sleep overactivation nor epilepsy. The EEG, at awake and sleep, of patient 32 was normal in the first 3 years. Successively, the EEG showed rare right frontal spikes and rare generalized SWs during sleep. At 4 years, a worsening was observed, with the appearance of frequent spikes and SWs in the awake EEG in the right hemisphere, mildly activated by sleep.

Table 3. EEG features
No.Awake EEG abnormalitiesSlow sleep abnormalitiesDetection age
of sleep abnormalities
Detection age
of sleep recovery
  1. SW, spike–wave; S, spike; SS, polyspikes; SSW, polyspikes–waves; slow W, slow waves; sharp W, sharp waves; ND, not determined.

 1Sharp W, S, SWS, SS, SWYesYes≥85%No       ND    15 yr
 2SWS, SWYesYes≥85%No       ND    12 yr
 3SW parietooccipitalSWYesYes≥85%No       7 yr    Ongoing
 4S, SW, SSW bilatS, SW, SSNoYes/No≥85%No       4 yr 5 mo    Variable
 5S, slow W SWSW, SSWNoNo≥85%Unilateral (R)       3 yr    Ongoing
 6S, SS, SSWS, SWNoYes≥85%Bilateral (R+)       ND    Variable
 7Sharp W, SW, SSW multifocalS, slow WNoNo≥85%No       ND    5 yr 6 mo
 8S, SW multifocalS, SS, SWNoNo≥85%Unilateral (L)       4 yr 3 mo    Ongoing
 9S, SW multifocal, diffuseS, SWNoYes≥85%Bilateral       4 yr 5 mo    Ongoing
10S, SW multifocalS, SWYesNo≥85%Bilateral       6 yr 1 mo    Ongoing
11S, SW multifocalS, SWYesYes≥85%Bilateral       4 yr    Ongoing
12S, SW (R+)SW, S, OLNoNo≥85%Unilateral (L)       6 yr 6 mo    Ongoing
13Slow W, S, SWSNoNo≥50%≤ 80%Unilateral (L)       3 yr    Ongoing
14Slow W, sharp W, S, SWSWNoNo/Yes≥50%≤ 80%Unilateral (L)       4 yr 5 mo    Ongoing
15S, SW multifocalS, SW,OLNoNo≥50%≤ 80%Unilateral (L)       4 yr 6 mo    Ongoing
16SWS,SS,SW,SSWYesYes≥50%≤ 80%Bilateral (R+)       ND    Ongoing
17S, slow SW, SSS,SW,SSNoYes≥50%≤ 80%Unilateral (L)       ND    Ongoing
18S, SWS,SWNoNo≥50%≤ 80%Bilateral       3 yr 6 mo    Ongoing
19S, SWS,SWNoNo≥50%≤ 80%Bilateral (L+)       3 yr    Ongoing
20S, SWS,SWNoNo≥50%≤ 80%No       2 yr 8 mo    Ongoing
21S, SWS, SWNoYes≥50%≤ 80%Bilateral       5 yr    Ongoing
22S, SWS, SWNoYes≥50%≤ 80%Bilateral       3 yr 4 mo    Ongoing
23S, SS, SWS, SS, SWNoYes≥50%≤ 80%No       1 yr 7 mo    Ongoing
24S, slow WS, slowW, SWNoYes≥50%≤ 80%Bilateral       6 yr 2 mo    Ongoing
25S, SW, multifocalS, SWNoNo≥50%≤ 80%No       1 yr 8 mo    Ongoing
26S, SWS, SWNoNo≥50%≤ 80%Bilateral       8 yr 5 mo    Ongoing
27SSNoNo≥50%≤ 80%Bilateral       3 yr    Ongoing
28HypsarrhythmiaS, SS, SWNoNo≥50%≤ 80%No       4 yr    Ongoing
29SS, SS, SWNoNo≥50%≤ 80%Bilateral       5 yr 8 mo    Ongoing
30S, SW, left frontotemporalBilateral 
31S, SWBilateral 

For six children (patients 1, 2, 6, 7, 16, and 17), it was difficult to determine the very onset of sleep activation because they were referred to our center at a later date. At follow-up, we confirmed that the CSWS disappeared in only three patients, at 15 years for case 1, 12 years for case 2, and 5.6 years for case 7. In two patients, CSWS outcome was still unstable, and transitory recovery was dependent on the response to treatment (ESM, corticosteroids). These were qualified as “variable.” In the remaining cases, CSWS or “overactivation” was still ongoing.

Cognitive and behavioral disorders

Data in relation to sleep abnormalities (29 cases) are shown in Table 4. We performed cognitive tests ≥3 days after the last seizure. However, in cases 4, 5, 9, 16, 19, and 28, several seizures per day occurred; in these cases (three with borderline or mildly delayed IQ and three with severe mental retardation), it was not possible to exclude a role of the seizures in a worsening of the test performances. We should also consider that each patient, with the exclusion of the three without epilepsy, was administered ≥1 AED.

Table 4. Cognitive/behavioral disorders in relation to sleep abnormalities
CaseBefore sleep abnormalitiesDuringAfter
  1. N.M.D., normal mental development; B.M.D., borderline mental development; Mi.M.R., mild mental retardation; Mo.M.R., moderate mental retardation; S.M.R., severe mental retardation.

  2. For symbol explanations, see text (Methods).

 1??S.M.R. (IQ 19)+++S.M.R+−−
 2??Mi.M.R. (IQ 69)++−B.M.D. (IQ 77)−−−
 3N.M.D. (IQ 95)−−−B.M.D. (IQ 76)−−−
 4S.M.R. (IQ 30)++−S.M.R. (IQ 30)+++
 5B.M.D. (IQ 78)−−−Mi.M.R. (IQ 60)+−−
 6??Mi.M.R. (IQ 56)+−−
 7??Mo.M.R. (IQ 29)+++Mo.M.R. (IQ 30)++−
 8B.M.D. (IQ 72)−−−B.M.D (IQ 71)−−−
 9S.M.R. (IQ 25)−−−S.M.R. (IQ 27)−−−
10??B.M.D. (IQ 82)−−−
11S.M.R. (IQ 32)−−−S.M.R. (IQ 30)−−−
12??B.M.D. (IQ 84)−−−
13N.M.D. (IQ 102)−−−B.M.D. (IQ 83)−−−
14N.M.D. (IQ 90)−−−N.M.D. (IQ 88)−−−
15S.M.R. (IQ 27)−−−S.M.R. (IQ 28)−−−
16??Mi.M.R. (IQ 55)++−
17??Mi to Mo MR (IQ 73->47->44)++−
18Mi.M.R. (IQ 51)−−−Mo.M.R. (IQ 40)−−−
19B.M.D (IQ 81)−−−Mi.M.R. (IQ 48)−−−
20S.M.R. (Bayley MD <50)−−−S.M.R. (IQ 26)−−−
21N.M.D. (IQ 101)−−−?−−−
22N.M.D. (IQ 110)−−−N.M.D. (IQ 89)−−−
23S.M.R. (IQ 18)−−−S.M.R. (IQ 19)−−−
24N.M.D. (IQ 92)−−−B.M.D. (IQ 78)−−−
25S.M.R. (IQ 28)−−−S.M.R. (IQ 28)−−−
26N.M.D. (IQ 91)−−−B.M.D. (IQ 75)+−−
27??N.M.D. (IQ 86)−−−
28S.M.R. (IQ 25)−−−S.M.R. (IQ 24)−−−
29B.M.D. (IQ 81)−−−Mi.M.R. (IQ 71)−−−

In the 20 epilepsy patients, where reliable information before the onset of sleep abnormalities was available, cognitive development was normal or borderline in 11 cases; the remaining nine cases had mild retardation in one, and severe, in the other eight. After the onset of sleep abnormalities, a definite deterioration of the cognitive skills was observed in nine of 19 patients. The remaining 10 patients also generally showed signs of mental retardation, but these were severe in eight patients and borderline in one; in one case, development was normal before and remained normal after the onset of the sleep disorder.

Only in one case (patient 2) of three with recovery from CSWS, was an improvement in cognitive competence seen. In all but one of the patients with longitudinal assessment, the comparison between DQ was reliable because the successively performed scales were the same. Behavioral disorders were present before the onset of sleep abnormalities in one patient with severe mental retardation; he belonged to the group of patients that would thereafter have CSWS. During the period of sleep abnormalities, six patients of 12 with typical or atypical CSWS and only three of 17 cases with a sleep overactivation of paroxysms, had behavioral disorders. All the three patients followed up after CSWS recovery showed an improvement of behavior.


CSWS is a specific pattern of electrical status “epilepticus” during slow sleep originally described by Tassinari et al. (15). In the latest scheme proposed by the International League Against Epilepsy (ILAE), CSWS, associated with various seizure types and a prominent neuropsychological disorder, has been considered to be a unique epileptic encephalopathy in children, called “encephalopathy with electrical status epilepticus during slow sleep (ESES), or ESES syndrome” (16). It is an age-related and self-limited disorder generally occurring between ages 4 and 12 years and has typical EEG features during slow sleep, consisting in almost continuous (>85%) bilateral and diffuse slow SWs. However, after several hundred patients, we should consider many variants of the typical slow-sleep features (focal to diffuse paroxysmal activities, various percentages of SWs <85, morphology of the SWs), and realize that the typical characters of slow-sleep abnormalities are only the “tip of an iceberg” (1).

An intriguing point has been raised by a recent study carried out by Monteiro et al. (2) that reports a case in which primary neonatal thalamic hemorrhage was associated with a strong activation of focal paroxysmal activities in the left parietotemporal region; they also pointed out a similar case previously reported in the literature (3). To explain this association, the authors drew inspiration from the experimental work of the Steriade school concerning the role of the thalamus in generating physiologic sleep oscillations and their possible changes into SW seizures (17). In particular, they built on experimental studies carried out by Steriade and Contreras (18) that described “long sequences of continuous spike–wave activity of 2–4 Hz, with a spectacular synchronization across all cortical leads” in unilateral athalamic cats. Afterward, some sporadic cases of early thalamic lesions associated with slow sleep paroxysmal activation were reported (4–8). Furthermore, if the possible role of corticothalamic injuries as a cofactor predisposing to CSWS (4–6) must be considered; the cases of CSWS described by Veggiotti et al. (19) in early shunted hydrocephalic children could be explained by the involvement of thalamocortical circuitries determined by the ventricular dilatation.

Our series comprises 32 patients with early thalamic injury, 29 of which had epilepsy with CSWS or paroxysmal overactivation during sleep.

The peculiarity of our sample, consisting of early clastic encephalopathies, could explain the high frequency of the epileptic disorder associated with thalamic injuries. Conversely, several were cases in whom thalamic lesions were detected since birth or occasionally in routine diagnostic procedures. Thus even though without epidemiologic value, given the selected origin of the sample, it is noteworthy that only in three of 32 cases, thalamic lesions were not associated with sleep paroxysmal activation. The fact that sometimes in our series we detected thalamic injuries after the diagnosis of sleep activation may suggest that these lesions could be not always accurately looked for and therefore were underestimated in children with CSWS.

Apparently no significant difference of clinical and neuroimaging features exists in the three patients without paroxysmal EEG activation during slow sleep. Other coexisting mechanisms may play a role, as the lack of epilepsy in two of them may suggest. We should point out the relatively young age of the three patients, between 6 and 7 years, and thus liable to a possible subsequent onset of the disorder.

Thalamic lesions in our series were unilateral in most cases; in other cases, an evident predominance of lateralization was noted. The involvement of the reticular nucleus and of the ventral nuclear mass was almost constant, and consistent with the following types of injury mechanisms: ischemic infarction of the medial cerebral artery territory in nine cases, periventricular hemorrhagic infarction in four, and an underlying ischemic pathogenesis in all the remaining patients but four, including those patients with postinfectious hydrocephalus or with a primary malformation (myelomeningocele). The pathogenic vascular mechanism in our series frequently involves the anterior choroidal artery for patients with ischemic infarction in the middle cerebral artery (MCA) or thalamostriate vein area that confluences with the terminal vein for patients with venous infarction; it may account for the generally predominant injury of the lateral aspect of the thalamus, including the reticular nucleus and the ventral nuclei. In only three patients was a different underlying (infectious) pathogenesis found: however, because of the association with hydrocephalus, we cannot exclude an ischemic mechanism in determining thalamic injuries. The analysis of electrophysiologic data showed an array of “iceberg” manifestations concerning the slow sleep activation of interictal EEG abnormalities. They extend from a simple strong activation of paroxysmal activities (with <85% and often with a focal predominance or morphologic atypical aspects) to an activation that was >85% shaping a true CSWS. Matching neuropathologic and electrophysiologic data, we first remark that all patients had corticosubcortical lesions, in two of them associated with basal ganglia injuries. Furthermore, we observed that all patients with bilateral thalamic injury (eight patients) had less typical forms of the epileptic sleep disorder. This suggests that the lack of at least one intact thalamus seems incompatible with typical CSWS. Finally, in cases with only a “sleep EEG overactivation,” the paroxysmal anomalies were generally asymmetrical (excepted case 16). We can thus distinguish the series in two different groups; the first included typical and almost all atypical CSWS with >85% of SWs, possibly symmetrical, and often with no spindles at all; the second one presented a usual asymmetry of SWs and an absence or reduction of spindles, plus other atypical features concerning the synchronism and morphology of SWs. This distribution, in at least two distinct groups with coherent characteristics, casts some doubt on the continuity of the “iceberg” and may suggest in our series different pathogenic mechanisms.

As regards cognitive features, we frequently observed cognitive deterioration or behavioral disorders or both at the onset of the sleep EEG activation in our series, and this is consistent with general findings in cases of CSWS (20,21). Furthermore, we note that behavioral disorders were significantly more present in patients with true CSWS. Only in one case of the three patients in whom CSWS disappeared was a definite improvement of cognitive competence noted, whereas behavior improved in all three patients. We were not able to draw any information from our series about the possible role of injury, epileptic disorder, or EEG focal location on the behavioral or cognitive profile of the disorder.

A possible explanation of the mechanism underlying the EEG activation in our series could be an imbalance between GABAB- and GABAA-mediated receptors due to thalamic injury, especially of the GABAergic reticular nucleus. Evidence exists that physiologic oscillations during sleep, such as spindles, and paroxysmal epileptic disorders, like absence seizures, are related phenomena (22,23) whose cellular mechanisms are based on the cyclical interaction between excitatory thalamocortical cells and inhibitory GABAergic nRT cells (24,25). In CSWS, epileptic discharges during slow sleep seem related to spindle-inducing mechanisms (25). The switch from physiologic oscillation frequency at 6–10 Hz to a slow paroxysmal (SW) oscillation frequency of 3–4 Hz could be related to the passage from GABAA receptor–mediated inhibitory postsynaptic potentials (IPSPs), setting the network oscillation frequency at 6–10 Hz, to GABAB receptor–mediated IPSPs changing the spindle-wave frequency to 3–4 Hz (26,27). The switch mechanism can possibly be linked to the increased firing of the cortex, leading to increased firing in the GABA cells of the thalamus. The release of higher amounts of GABA into thalamic synapses activates GABAB receptors and changes the normal GABAA receptor–mediated oscillation setting (28). An imbalance of GABAB- versus GABAA-mediated receptors, such as determined by GABAA-antagonist blockade of fast IPSPs, produces three to four large oscillations independent of the kind of cortical firing (29).

In some way similar, an ill-balanced condition due to thalamic injury, in particular of nRT, could be a predisposing cofactor of CSWS. In conclusion, our series is a selected cohort of children with thalamic early injury associated with symptomatic CSWS or sleep SW overactivation. A comparative study of clinical features between CSWS epileptic children with and without early thalamic injuries could better highlight the role played by thalamic lesions in predisposing to the sleep electrical disorder and the consequent clinical picture. If our data are confirmed, early thalamic injury may be considered a cofactor predisposing to CSWS, and its detection in infancy or early childhood could be considered a warning sign of the onset of CSWS. It should indicate an accurate EEG sleep monitoring for an early treatment, to prevent the cognitive deterioration and the behavioral disorders that are usually associated with the sleep EEG disorder.


Acknowledgment:  We warmly thank Professor C.A. Tassinari and Professor R. Guerrini for reading the paper and giving precious comments and suggestions.