Idiopathic epilepsy with generalized tonic–clonic seizures only versus idiopathic epilepsy with phantom absences and generalized tonic–clonic seizures: One or two syndromes?


  • Michael Koutroumanidis,

    1. Department of Clinical Neurophysiology and Epilepsies, Guy's, St. Thomas' and Evelina NHS Foundation Trust, London, United Kingdom
    Search for more papers by this author
  • Konstantinos Aggelakis,

    1. Department of Clinical Neurophysiology and Epilepsies, Guy's, St. Thomas' and Evelina NHS Foundation Trust, London, United Kingdom
    Search for more papers by this author
  • Chrysostomos P. Panayiotopoulos

    1. Department of Clinical Neurophysiology and Epilepsies, Guy's, St. Thomas' and Evelina NHS Foundation Trust, London, United Kingdom
    Search for more papers by this author

Address correspondence to Michael Koutroumanidis, Department of Clinical Neurophysiology and Epilepsies, Lambeth Wing, 3rd floor, St. Thomas' Hospital, London, UK. E-mail:


Purpose: To define the relationship between two syndromes of idiopathic generalized epilepsy (IGE) with apparently similar phenotypes: The form with generalized tonic–clonic seizures only (IGE-GTCS) and that with phantom absences (IGE-PA).

Methods: We compared the electroclinical features of 33 consecutive patients with GTCS and generalized spike wave (GSW); 18 had only GTCS and were diagnosed as IGE-GTCS, and 15 had hitherto unnoticed mild absences on the electroencephalography (EEG) and were diagnosed as IGE-PA. All patients were subjected to the same diagnostic workout, including video EEG during hyperventilation with breath counting (HBC). Patients with a clinical history of absences or myoclonic seizures were excluded.

Results: PA were easily identified with the first or second EEG in 14 of 15 patients with IGE-PA and always with sleep-deprived EEGs; conversely, PA did not occur in the IGE-GTCS patients despite using more EEGs. GTCS were twice as frequent in the IGE-GTCS group and tended to occur on awakening, whereas episodes of absence status affected twice as many patients with IGE-PA. The hereditary risk was 30% in the IGE-GTCS and 6.7% in IGE-PA. GSW had a strong polyspike component in IGE-PA and were briefer in IGE-GTCS. There is no evidence for a maturational influence on the duration of GSW in either syndrome.

Conclusion: Our findings clearly indicate that IGE-GTCS and IGE-PA are two distinct IGE syndromes and emphasize the role of PA for patients' diagnosis and management and for syndromic classification. They also appear to validate HBC as a simple, sensitive, and pragmatic method for the clinical identification of typical absences.

The syndrome of “idiopathic generalized epilepsy (IGE) with generalized tonic–clonic seizures (GTCS) only” (IGE-GTCS) is a form of the so-called awakening epilepsy with GTCS mainly within the first 1–2 h after waking up (Janz, 1953, 2000); the concept was recognized as “IGE with GTCS on awakening” (Commission on Classification and Terminology of the International League Against Epilepsy, 1989) and was broadened in the 2001 International League Against Epilepsy (ILAE) diagnostic scheme to include also patients with random and nocturnal GTCS (Engel, 2001). Despite the apparent strictness of the term, there has been no consensus on whether or not infrequent and mild typical absences or myoclonic seizures were to be accepted (Panayiotopoulos, 2007). Their acceptance would result in potentially significant overlap with other IGE syndromes that share the same seizures and activation on awakening, such as juvenile myoclonic epilepsy (JME); on the other hand, their formal exclusion would not prevent over diagnosis of IGE-GTCS, because truly mild absences or myoclonic seizures may be easily overlooked in a busy general neurology outpatient clinic and when conventional electroencephalography (EEG) recordings are used. From the EEG standpoint, interictal generalized spike wave (GSW) discharges may or may not occur in a given recording; considering them as absolute prerequisite for the diagnosis might be too restrictive as possibly missing out true cases. The converse would allow potential overlap with poorly identifiable epileptic conditions, and robust EEG criteria to filter out such patients (in terms of type and number of recordings per patient and activation methods) have not been standardized. Even in the presence of GSW, the possibility of cryptogenic focal epilepsies with fast secondary generalization remains open (Raymond et al., 1994; Ferrie et al., 1995). Some of these diagnostic shortcomings and pitfalls that can blur the electroclinical and genetic identity of the IGE-GTCS may have supported the decision of the ILAE Classification Core Group to dismiss this entity from the catalogue of epileptic syndromes (Engel, 2006).

In 1997, we described a syndrome of IGE with phantom absences (IGE-PA) in a group of 13 adult patients (Panayiotopoulos et al., 1997b). They all had GTCS of usually late onset, and about half of them also had frequent episodes of absence status (ASE). Because of their mildness and brevity, PA had passed unnoticed by all these patients and their relatives alike and were only documented with our routine video EEG practice of breath counting during hyperventilation (HBC). Complete lack of myoclonic seizures is a diagnostic prerequisite that differentiates IGE-PA from myoclonic IGE syndromes with brief and often unnoticed absences, such as JME. We attempted to convey the clinical imperceptibility of these absences by the term “phantom” that was recently adopted by the ILAE classification core group to denote a type of absence “likely to be a result of brain maturation” (Engel, 2006). True, absences may become shorter and milder with time, and such absences may coexist with more severe and prolonged ones in childhood absence epilepsy (CAE) and juvenile absence epilepsy (JAE). However, the essence of what we described as IGE-PA is quite different: PA were the only absence seizure in our patients and were never more conspicuous; patients and relatives were invariably unaware of them before diagnosis was made and therefore, the age of their onset is by definition unknown. Therefore, the IGE-PA syndrome—as we define it—is another listing in the differential diagnosis of IGE with GTCS.

The patient with a first GTCS presents a diagnostic challenge. The demonstration of GSW would suggest IGE, particularly when intellect, neurologic examination, and brain imaging are normal, and absences or myoclonic seizures are identified by history or meticulous follow-up. In the absence of such evidence though, a question arises: Is there a single homogenous condition with its characterization as either IGE-GTCS or IGE-PA being merely depended on the methodology and the depth of clinico-EEG investigation, or are there distinct IGE phenotypes with important clinical and EEG differences that may require specific management?

In patients with GTCS and GSW discharges, sleep-deprived video EEG with HBC on awakening may reveal hitherto unnoticed PA pushing the diagnosis away from IGE-GTCS; showing that in most of these patients PA are easy to elicit and frequent enough might even argue for amalgamation of the two conditions (Koutroumanidis & Smith, 2005). Alternatively, persistent lack of PA or other possible differences would verify the existence of two distinct syndromes. Because at St. Thomas' we have been using the HBC technique as tool to study absences since 1989 (Panayiotopoulos et al., 1993; Giannakodimos et al., 1995), we analyzed prospectively collected clinical and EEG data on consecutive patients with GTCS and GSW, diagnosed with IGE-GTCS and IGE-PA, in an attempt to better understand these two conditions and their relationship.


Our tertiary epilepsy clinic of Guy's and St. Thomas' Hospital, London, accepts referrals from the area's general practitioners, the neurological and other specialist services of our hospital including the pediatric neurology/epilepsy services, and the accident and emergency department through a dedicated “first seizure” clinic. Approximately 10% are tertiary referrals from outside the catchment area of the two hospitals.

Our clinical approach of patients with epilepsy has been detailed elsewhere (Panayiotopoulos et al., 1992, 1997b). For the adolescents who are transferred from our pediatric epilepsy service, full information on their earlier clinical course, including the actual video EEGs, is readily available, whereas for patients who had been seen and followed-up elsewhere, previous medical correspondence and old EEG reports or actual records are systematically sought. For patients with IGE, we attempt to ascertain the type of their early seizures (absence, myoclonic, GTCS, ASE) and the order of appearance, frequency, and diurnal variation, with particular emphasis on absences and their possible bold and consistent clinical characteristics, such as eyelid or other facial myoclonias, and possible triggers such as photic stimuli and TV, before initiation of treatment. We diagnose episodes of ASE on clinical grounds when a clear history of prolonged confusional state is available in patients with clinical and EEG evidence of IGE (Anderman & Robb, 1972; Panayiotopoulos et al., 1992; Agathonikou et al., 1998; Koutroumanidis, 2008).

All patients have a routine or sleep video EEG after partial sleep deprivation, tailored to the clinical information (Koutroumanidis & Smith, 2005). When IGE is clinically suspected, the sleep EEG is extended by a prolonged recording on awakening with one or multiple sessions of hyperventilation to reveal possible absences. All patients with IGE have repeat video EEG studies, particularly when the initial routine or sleep video EEG has failed to record clear ictal events.

Assessment of behavioral changes and cognition during GSW discharges

We routinely use the HBC in all IGE patients (Panayiotopoulos et al., 1993; Giannakodimos et al., 1995). This method uses hyperventilation, by a default absence-activating technique, during which patients are asked to count their breaths on expiration, aloud and sequentially. Hence, they are engaged in a continuous, semiautomatic (and therefore not expected to suppress GSW through intense concentration) mental and motor process, simple to perform by children and elderly patients alike, and subject to interruption by GSW with their performance in the absence of GSW acting as the patients' own control. Any faltering associated with GSW (hesitations, repetitions, or misses) is interpreted as cognitive impairment, and consequently the electroclinical event as absence seizure.

Possible cognitive impairment associated with spontaneous 3- to 4-Hz GSW discharges is typically tested with numbers or words called as quickly as possible after the onset of the discharges, which patients are asked to repeat after the end of the discharge. Concurrent behavioral changes are identified and analyzed by carefully reviewing the video EEG.

Diagnostic criteria for IGE-GTCS

Patients were classified in this group if they satisfied all three diagnostic criteria as follows: (1) had suffered GTCS no matter their timing or number/frequency (Fisher et al., 2005); (2) had GSW discharges on EEG at ≥2.5 Hz and of any duration, provided that they were not associated with evidence of impairment of consciousness (IoC) even during HBC; and (3) had no other—independent of the GTCS—seizures (myoclonic, absences including PA, or focal) that occurred at any stage of the natural history according to detailed anamnesis and review of previous medical correspondence and EEG reports, and during the follow-up period in our clinic. Episodes of ASE leading to GTCS were accepted (Koutroumanidis, 2008).

Diagnostic criteria for IGE-PA

The IGE-PA group includes patients in whom mild IoC associated with >2.5-Hz GSW (phantom absences) was first appreciated on EEG/video EEG after a GTCS that made them seek medical advice. We consider IoC as the defining, sine qua non element of the typical absence, therefore GSW discharges associated with subtle motor changes such as eyelid flickering, but not with IoC that our methodology could unveil, were not diagnosed as PA. There should be no previously experienced, observed, or diagnosed absences or myoclonic seizures according to detailed history, and previous medical correspondence/EEG reports. However, some patients with PA may retrospectively admit momentary lack of concentration and forgetfulness, which, in their opinion, was not of practical relevance (see below).

All patients fulfilled the diagnostic requirements for IGE, including normal examination and intellect, normal background EEG activity and brain magnetic resonance imaging (MRI). All EEGs of each patient were reviewed and analyzed, including the number, duration, and timing of the GSW discharges in wakefulness and sleep. Videos were reviewed for possible clinical correlates of all discharges that occurred during hyperventilation and also of those that occurred spontaneously and were tested for IoC. Patients in whom not all their video EEG recordings were available for review, and those with incomplete or insufficient EEG evaluation, including suboptimal testing for IoC during GSW, were excluded from the study. We also excluded a patient with IGE-PA who by her last follow up had suffered no GTCS, but was seen after a prolonged episode of ASE that prompted the diagnostic video EEG (Panayiotopoulos et al., 2001); such patients would serve no purpose in this IGE-GTCS versus IGE-PA comparison.

Patients were classified as seizure-free when they remained without any GTCS and ASE at least for twice as long as the time that corresponds to the average frequency for each patient or longer than the longest ever period of seizure freedom, as indicated by patients' detailed seizure histories, and as non–seizure-free when they continued with seizures after the most recent antiepileptic drug (AED) change or if they declined treatment. A third group included those who were seizure-free at their last assessment, but the time since the most recent AED change was either too short or the seizure frequency was too low to assess outcome. The response of PA to AED was mainly assessed with follow-up video EEG, but was more difficult to ascertain.

Nonparametric characteristics were compared using two-tailed Mann–Whitney U Test. Nominal clinical characteristics were analyzed by Chi square (χ2) test with Yates' correction. Statistical significance was set at p < 0.05.


The clinical and EEG characteristics of the 18 patients with IGE-GTCS and the 15 patients with IGE-PA who satisfied the inclusion criteria are shown in Table 1, and those of the individual patients in Tables 2a and 2b. Seven of the patients with PA (1, 3, 4, 6–8, and 14) had been included in our original report (Panayiotopoulos et al., 1997b). There were no significant differences between the two groups with regard to gender distribution, age at referral, age at last follow-up, duration of the epilepsy, and the length of follow-up (Table 1).

Table 1.   Clinical and EEG characteristics
 IGE with GTCS onlyPhantom absencesStatistics
  1. GTCS, generalized tonic–clonic seizures; PA, phantom absences; NS, not significant; R, routine EEG; S, Sleep EEG facilitated by partial sleep deprivation; GSW, generalized spike wave discharges; GPPR, generalized photoparoxysmal response.

  2. aMann–Whitney U Test (P/Z).

  3. bFisher's exact test.

Patients (men/women)18 (7/11)15 (8/7) 
Age (mean; range; median)
 At onset (GTCS)20; 7–44; 19.524.3; 13–56; 19NSa
 At clinical/EEG evaluation32; 16–69; 2937.0; 22–68; 34NSa
 At last follow-up37.7; 22–71; 36.542.2; 29–73; 36NSa
 Duration16; 3–38; 1617; 1–40; 17NSa
 Lifetime number13; 1–50; 76; 1–35; 30.008/2.58a
 Frequency (per year)0.87; 0.1–4; 0.50.34; 0.28; 0.08–0.8; 0.200.025/2.24a
 Circadian variation and timingIn 16 on awakening or a.m. also during relaxation; in 2 mainly nocturnalUsually variable; only in two consistently on awakening; in none nocturnal 
 GTCS status1 patientNone 
 Absence status4 pts (22%)8 (53.3%)NSb (0.055)
Typical absencesNone recorded or admittedPA in all; 7 thought might had before their first GTCS. 
Precipitants10 (55.5%)12 (80%)NSb
Family history of epilepsy7 (39%)/5 (28%) with 1st degree2 (13.5%)/1 with 1st degreeNSb
Diagnostic yield of 1st EEG (type)
 Subclinical GSWD16 (14R; 2S)6 (R) 
 PA8 (5R; 3S) 
 Normal2 (R)1 (R) 
Additional yield from 2nd EEG
 Subclinical GSW2 (S)1 (R) 
 PA6 (2R; 4S) 
Additional yield from 3rd EEG +
 PA1 (R) 
No of GSWD21; 1–70; 12.533; 8–88; 28NSa
Maximal GSWD duration3.5; 1–8; 36.7; 2–23; 50.022/2.28a
Min duration of absenceN/A2 s 
Polyspikes6 (33.3%)12 (80%)0.008b
Photosensitivity5 (27.8%; 3 posterior/2 GPPR)2 with GPPR (13.5%)NSb
Table 2a.   Patients with GTCS only
Pt/sex/ageGTCSGTCS no/ASEDuration  Longest AED onAED onResponse to
at last FUonsetfrequencyaonset/numberyearsF-U yearsEEGGSWD (s)Polyspikesreferrallast FUtreatmentb
1/F/322011/0.90 127RRRS3+ (w + s)Not SF
2/M/22165/0.80 77RRS4c+ (w)VPASF (4.5 years)
3/F/4820>10/0.36 2717RS1+ (s)Not SF
4/M/71443/0.10 274SS2+ (s)VPAVPA/LTGInsufficient FU
5/M/319∼50/2.27Teens/>1/y1711RRRRS8VPA/LTGNot SF
6/F/381810/0.50 2020RS1cVPAVPA/LTGSF (5 years)
7/M/36197/0.4019/once1310RSRR4VPAVPASF (4 years)
8/F/2274/0.27 142RS3cNot SF
9/F/28244/1.00 44RRS3+ (w + s)VPAVPANot SF
10/M/45266/0.3226/5198RSS3+ (s)Not SF
11/M/502915/0.70Early 30s/<52215RRRS8Not SF
12/F/441614/0.50 2813RRS2GBPVPAd/LEVInsufficient FU
13/F/37175/0.25 103RRR3CBZCBZNot SF
14/F/469∼50/1.35 3713RRRR1TPMTPM/LEVSF (4 years)
15/F/35234/0.40 108RS6CBZCBZ/CNZSF (8 years)
16/F/28107/0.40 1919SS5CBZCBZSF (7 years)
17/F/27244/1.30 32.5RS5LEVSF (2 years)
18/M/383224/4.0 66RSS2CBZVPANot SF
Table 2b.   Patients with PA
Pt /sex/ageGTCSGTCS No/ASEDuration  Longest AED onAED onResponse to
at last FUonsetFrequencyaonset/numberyearsFU yearsEEGGSWD (s)Polyspikesreferrallast FUtreatmentb
  1. aGTCS frequency is expressed as ratio of total number of GTCS over epilepsy duration.

  2. bSeizure-free state (SF) requires a period without any seizures at least twice as long as the time that corresponds to the average frequency for each patient, or longer than the longest ever period of seizure freedom as indicated by patients' detailed seizure history.

  3. cDid not have discharges during hyperventilation.

  4. dVPA was replaced because of side effects.

  5. GSWD, the longest generalized spike wave discharges of each patient, including those during sleep. In Table 2b (patients with IGE-PA), these durations do not refer to phantom absences; phantom absences were associated with discharges 2 s or longer, but longer discharges were not necessarily associated with impairment of consciousness (see also text); FU, follow-up; ASE, absence status; R, routine EEG; S, sleep EEG facilitated by partial sleep deprivation; w, wake; s, sleep; VPA, sodium valproate; LTG, lamotrigine; LEV, levetiracetam; TPM, topiramate; CBZ, carbamazepine; GBP, gabapentin; CNZ, clonazepam; PHT, phenytoin.

1/M/341515/0.8015/<2/year263RR4+ (w + s)PHT/VPAVPA/LTGNot SF
2/F/41172/0.08 2414S6+ (s)PHTPHTSF (20 years)
3/F/733035/0.8030/>1/year406RRRR3CBZVPASF (6 years)
4/F/73562/0.1256/51817RR15+ (w + s)VPANot SF
5/M/35242/0.1824/once99R6VPASF (9 years)
6/M/32281/N/A 11R3+ (w)VPALost to FU
7/M/44185/0.2018/>3/year188RRR5+ (w)VPA/LTGNot SF
8/M/59561/N/A 11RS2+ (s)Lost to FU
9/F/36222/0.14 99S5+ (s)CBZVPASF (9 years)
10/F/291310/0.62 1717S23+ (w + s)CBZCBZSF (8 years)
11/M/31195/0.4219/twice126RS5+ (s)CBZVPASF (4 years)
12/F/30203/0.30 1212RS8+ (w + s)Not SF
13/M/37133/0.1211/once2616RS5+ (w + s)CBZVPAInsufficient FU
14/F/25172/0.67 38RSR7+ (w + s)VPASF (8 years)
15/M/54167/0.2024/7382RS2PHTLEVInsufficient FU

Age at seizure onset

This could be ascertained for the GTCS in all patients, and could be determined with reasonable accuracy for the ASE (Tables 1, 2a, and 2b), but remained unknown for PA. Apart from patient 13 with IGE-PA in whom his sole episode of ASE preceded his first GTCS by 2 years, the first ASE was usually associated with, or occurred after the first GTCS.


Generalized tonic–clonic seizures

These were almost twice as frequent and twice as many in the IGE-GTCS group (Table 1). In this group, they tended to occur mainly within 1–2 hours after awakening, but could also occur later on in the day, particularly when they followed episodes of ASE. Two patients had mainly nocturnal GTCS. A different diurnal pattern was noted in the IGE-PA group: GTCS timing was generally more variable during the day with only patients 7, 12, and 13 having seizures consistently in the morning after awakening, while none had nocturnal seizures.

Absence status epilepticus

Episodes of ASE occurred in both groups, but affected twice as many patients with IGE-PA (Table 1). In these patients ASE also appeared to occur more frequently than in IGE-GTCS: Episodes would recur more than once per year on average in three patients with PA (patients 1, 3, and 6; Table 2b) being sometimes difficult to terminate (Fig. 1) and outnumbering their GTCS, as opposed to only one patient with IGE-GTCS (patient 5; Table 2a) with yearly ASE that were still less frequent than his GTCS. On most occasions, episodes of ASE were temporally associated with a GTCS, usually ending in one.

Figure 1.

 Upper trace: Absence status epilepticus (ASE) in patient 3 with IGE-PA (Table 2b). The patient is mildly confused, with good speech. She had 35 episodes of ASE from the age of 30 years onwards; all occurred without identifiable precipitants and invariably ended in GTCS. Interictal video EEG on other recordings confirmed PA, of which the patient and her family were completely unaware. Lower trace: Treatment with 5 mg diazepam intravenous (IV) within the EEG department abolished discharges immediately and completely, but transiently; 3 min later they reappeared, first as abortive generalized, then in duplets, triplets, and becoming denser until the full continuous unabated pattern of the upper trace was reestablished. Another 5 mg diazepam IV produced exactly the same response, although GSW discharges reappeared a bit later (5 min). The lower trace depicts the return of the ASE shortly before it became fully continuous again. Final resolution of ASE was achieved much later after IV sodium valproate (VPA), although spontaneous termination mechanisms might have also been involved (from Koutroumanidis, 2008 with the permission of the editors).


No absences were admitted, observed, or diagnosed in any of the patients of either group before their referral to our clinic. After the clinical EEG diagnostic workout was concluded and the nature of their epilepsy was explained (including examples of PA on video clips) all patients with IGE-GTCS once again denied past absences. On the other hand, seven patients of the IGE-PA group thought that they might have had similar events earlier, although they could not even approximate their onset or frequency. In some patients, these were also retrospectively identified by their relatives or witnesses.

Other seizures

There were no other seizures, and in particular, no patient had independent myoclonic or focal seizures.

Family history of epileptic seizures

A positive family history excluding febrile seizures was obtained from seven patients with IGE-GTCS and from two patients with IGE-PA. From the former group, patients 3, 5, 9, 11, and 16 (Table 2a) had one first degree relative each with epilepsy (three had absence syndromes and two had GTCS), while patients 5, 9, 10, and 15 had second degree relatives with seizures (one with absences and GTCS, all other with unknown seizure types). Patient 4 with IGE-PA had three nephews from different sisters with infrequent GTCS of late onset, while the father of patient 14 also had PA and three GTCS from age 51 years.

Seizure precipitants

Seizure precipitants were reported by 10 patients with IGE-GTCS and 12 patients with IGE-PA and included sleep deprivation in 14 patients (five with IGE-GTCS and nine with PA), alcohol consumption in 12 patients (five with IGE-GTCS and seven with IGE-PA), and stress in seven patients (five with IGE-GTCS and two with IGE-PA). In addition, two patients with IGE-GTCS reported seizures during relaxation, but none with IGE-PA did so. No patient from either group reported any seizures triggered by visual stimuli, reading, or other linguistic activities or other external stimuli.

Diagnosis on referral

All patients were referred with a diagnosis of either “grand mal or generalized epilepsy” or “focal—usually temporal lobe epilepsy (TLE)—with secondary generalization.” The latter was based on misinterpreted EEG asymmetries or focalities, and in some patients on symptoms implying “focal” onset. In addition, when not missed in history, and despite GSW discharges in previous EEGs, prolonged confusional episodes were characterized as complex partial status epilepticus (CPSE) in patients 7 and 10 with IGE-GTCS and patients 1, 4, 5, and 7 with IGE-PA, luring into the diagnosis of focal epilepsy. Patients 9 and 10 with IGE-PA had been diagnosed with focal epilepsies with fast secondary generalization, because they both had solitary rolandic seizures at age 8 and 9 years. Past episodes of ASE were altogether missed in two patients with IGE-GTCS (patients 5 and 11) and in four with IGE-PA (patients 3, 11, 13, and 15).

Two patients with IGE-GTCS (patients 2 and 17) and six with IGE-PA (patients 4, 6–9, and 14) were referred by their general practitioners with newly diagnosed GTCS. The remaining patients had been followed-up elsewhere. Old EEG reports, obtained through correspondence for most of the patients, invariably described GSW activity, but none suggested absences. Patients 9 and 10 with IGE-PA had EEG rolandic spikes in childhood.

EEG/video EEG studies

The types (routine or sleep) of the recordings and their chronological order for each patient are shown in Tables 2a and 2b, and their diagnostic yield and the overall results of the EEG analysis are summarized in Table 1. In total, our patients had 82 EEG recordings; those with IGE-GTCS had 53 (mean 2.95, routine 33, sleep 20), and those with IGE-PA had 29 (mean 1.9, routine 20, sleep 9).

All patients with IGE-PA had at least one PA during hyperventilation, manifested by hesitation, mistake, or omission of a number during HBC, and on a few occasions by concurrent mild eyelid or lip flutter detectable only on video review. PA were associated with GSW discharges as brief as just 2 s (Fig. 2). The same methodology in the group of patients with IGE-GTCS elicited only subclinical GSW discharges (Fig. 3) in all but patients 2, 6, and 8, who had GSW discharges only during sleep. No patient from either group ever failed to recall verbal stimuli during GSW discharges of any length.

Figure 2.

 Phantom absences on video EEG manifested with brief hesitations on breath counting during hyperventilation in patients 11 (upper trace) and 1 (bottom trace) of Table 2b. Note the brief duration of both discharges (about 2 s) and the onset with generalized polyspikes of the discharge in the lower trace.

Figure 3.

 Subclinical GSW discharges on video EEG during hyperventilation and breath counting in patient 15 with IGE-GTCS (Table 2a). Upper trace: Recording of 2003 on awakening after partial sleep deprivation. Note that despite the marked activation, discharges remained subclinical. Lower trace: Routine recording of January 2008. The patient has been completely seizure (GTCS)-free for the last 8 years, but GSW discharges persist.

In the IGE-GTCS group, the first EEG recorded GSW in 16 patients, while the second showed GSW in patients 6 and 18, in whom the first awake recordings were normal (Tables 1 and 2a). In contrast, the first EEG in the IGE-PA group detected PA in eight patients (patients 2, 4–6, 9–11, and 14), the second detected PA in six others (patients 1, 3, 8, 12, 13, and 15) who had subclinical GSW in their first recordings, and the third detected PA in patient 7 whose first routine EEG was normal and the second—also routine—showed GSW (Tables 1 and 2b). In summary, it was possible to detect PA with the first or the second EEG in 14 of 15 patients of this group, and with routine only recordings in eight. In addition, sleep-deprived EEG recordings with HBC on awakening were always successful in detecting PA in the IGE-PA group and always unsuccessful in patients with IGE-GTCS (Tables 1, 2a, and 2b).

With the total numbers of GSW discharges being comparable between the groups, GSW were significantly longer in the IGE-PA group than in the IGE-GTCS (Table 1), including those during sleep. There was no difference of the maximal discharge duration between younger and older patients with either IGE-PA or IGE-GTCS, using a cutoff at 30 years at the time of the EEG assessment that split patients into fairly equivalent groups.

Patients with IGE-PA were significantly more likely to show polyspikes or polyspike and wave complexes within the generalized discharges (Tables 1, 2a, and 2b). However, none of the patients of either group had sustained trains of polyspikes. Five patients with IGE-GTCS and two patients with IGE-PA were photosensitive, but none reported or admitted any symptoms when exposed to environmental light sources, including TV and video games.

Treatment and outcome

Almost half of the patients (eight with IGE-GTCS and seven with IGE-PA) were taking no AED on referral. After the implications of the diagnosis and the management options were discussed, five patients with IGE-GTCS and two with IGE-PA declined again treatment with AED (Tables 2a and 2b).

The follow-up period ranged from 2 to 20 years (mean 9.4 years, median 8 years) for the IGE-GTCS group and from 1 to 17 years (mean 8.6 years, median 8 years) for the group with PA. At their last assessment, seven patients of each group could be characterized as seizure-free, yielding for both groups a 60% chance for fair outcome when patients with insufficient follow-up and those who declined AED treatment were excluded (Tables 2a and 2b). We did not record PA in the follow-up EEGs of the seven seizure-free patients with IGE-PA, but all had subclinical GSW discharges. GSW discharges also occurred in the follow-up EEG recordings of all patients with IGE-GTCS, including those who were seizure-free (Fig. 3).


We analyzed prospectively collected clinical and EEG data of 33 consecutive adults with IGE and GTCS to study the relationship between the subsyndromes of IGE-GTCS and IGE-PA. There were no myoclonic seizures or absences admitted, observed, or diagnosed in any of these patients at referral, and all patients were assessed with the same EEG protocol for IGE that includes HBC, routinely used in our center to document absences. Using this technique, we diagnosed 15 patients with brief absences as IGE-PA and 18 patients without as IGE-GTCS. The following clinical and EEG differences between the two groups consolidate a useful distinction between the two syndromes and describe their most important characteristics.

Clinical aspects

The most striking finding was the easiness in detecting PA in the IGE-PA group (with the first or second recording in 14 of 15 patients, and always with a sleep EEG and HBC on awakening) as opposed to the IGE-GTCS group, in which PA were not recorded despite the larger number of EEGs per patient. PA were associated with GSW as brief as 2 s (Fig. 2), and apart from the momentary faltering during HBC, associated sometimes with subtle eyelid or lip flutter, there were no other clinical manifestations.

Second, GTCS—the seizure-symptom that brought all these patients to medical attention—were more frequent in the patients with IGE-GTCS with a circadian pattern more readily referable to what Janz has called awakening epilepsy (Janz, 2000). In contrast, GTCS showed a more variable timing in the IGE-PA group, as they frequently followed episodes of ASE in most of these patients.

Third, although they occurred in both groups, episodes of ASE showed a tendency to affect more patients with PA, in whom they also seemed to be more frequent. Episodes of ASE had been identified in half of the affected patients in each group (in two of four with IGE-GTCS and in four of eight with IGE-PA) before they were referred to our center, but had been invariably characterized as CPSE.

In common, most patients from both groups had a relatively benign course with infrequent seizures, and only a few from each group appeared resistant to AED that are deemed appropriate for IGE (Tables 2a and 2b).

EEG aspects

We note two important findings: First, polyspikes appeared in both conditions, but significantly more so in IGE-PA (Table 1). This is a rather unexpected finding for a purely absence syndrome because polyspikes are traditionally associated with myoclonic seizures, which are hereby excluded by definition. The relatively late median age of EEG recording in this group (34 years) can hardly be a reason, as two of the three patients without polyspikes (patients 3 and 15 in Table 2b) were in the seventh and fifth decade of life, respectively. Therefore, our finding indicates that a polyspike component appears to be an EEG characteristic of the IGE-PA and also reinforces Niedermeyer's view that polyspikes are not necessarily associated with myoclonic seizures or severe epilepsies of adulthood (Niedermeyer, 1966).

Second, GSW discharges were significantly longer in IGE-PA (average duration 6.7 s versus 3.5 s in IGE-GTCS; Tables 1, 2a, and 2b). However, we do not necessarily interpret this difference as relevant to the mechanism that underlies impairment of cognition, the very essence of PA: Solid clinical experimental evidence suggests that IoC tends to occur within the initial few seconds of the GSW discharge, even by the time of the paroxysm onset (Porter & Penry, 1973; Browne et al., 1974), and this accords with the brief PA we recorded on video (Fig. 2); the length of the discharge matters more for other clinical correlates of the typical absence seizure, such as automatisms (Penry et al., 1975; Panayiotopoulos et al., 1989) or autonomic changes (Bogacz & Yanicelli, 1962; Mirsky & van Buren, 1965). Once IoC has occurred, the longer the duration of the seizure the more clinically apparent this unresponsiveness may become (notwithstanding the effect of possible automatisms that also attract attention), appreciated by patients, bystanders, and doctors. The notion of disturbance of consciousness from the very start of the GSW discharge comfortably explains the phenomenon of PA (despite their brevity) and also their imperceptibility (because of their brevity), particularly when compared with the lengthy and clinically bold absences of CAE and JAE. Therefore, the difference between IGE-GTCS and IGE-PA with respect to the duration of the GSW discharges appears to better characterize the former syndrome: Plainly, GSW discharges in IGE-GTCS inherently tend to be very brief (Fig. 3), even more so than those in the IGE-PA.

Why then patients with IGE-GTCS and GSW discharges ≥2 s during HBC consistently do not show IoC, and also why patients with IGE-PA, do not have demonstrable IoC with every GSW discharge ≥2 s? If technical factors like mistiming of the discharge during a very slow rate of HBC are excluded, one may hypothesize that 3-Hz GSW oscillations are initially handled or sustained by different thalamocortical circuits, some of which may be more important for consciousness than the other. Demonstration of such possible differences would probably need coregistration of sophisticated functional imaging techniques such as functional MRI (fMRI) with EEG or magnetoencephalography (MEG-EEG).

Phantom absences and the question of brain maturation: a seizure type, a characteristic of a syndrome, or both?

Our results do not support the recently promulgated by the ILAE notion that PA are “likely to be a result of brain maturation” (Engel, 2006). Statistical analysis showed no difference in the duration of the longest discharges between younger (22 to ≤30 years) and older patients (>30 to 73 years) in either group. This series does not include patients younger than 20 years, but is well representative of IGE-PA, which mainly concerns adults with IGE; even in the sole so far reported child with PA (Panayiotopoulos et al., 2001) none of the recorded discharges (including her PA) lasted more than 5 s, which is exactly the median maximum duration of the GSW discharges in our present IGE-PA group (Table 1).

Brief GSW discharges associated with mild and imperceptible (to testing with loud verbal stimuli) impairment of cognition may coexist with more severe absences in the prototype absence IGE syndromes of JAE and CAE, in which they are simply considered as subclinical; similar inconspicuous absences can also occur in predominantly myoclonic syndromes, such as JME where they are the characteristic absence type. There seems to be no clear effect of brain maturation here. Certainly, lengthy absences during childhood and adolescence may also become shorter in adulthood (Gastaut et al., 1986; Panayiotopoulos et al., 1992; Michelucci et al., 1996), and this may indicate a maturation process. Therefore, PA as a particular seizure (absence) type may exist in different IGE syndromes and ages, but do not necessarily relate to brain maturation. However, in the syndrome of IGE-PA, phantom absences are by any available clinical and EEG standards the only absence type from the very beginning, and remain stable in terms of length and clinical expression and severity for many years. It appears that in IGE-PA, thalamocortical oscillations are systematically cut short by some strictly set time mechanism whose faltering may lead not to longer (and clinically obvious) absences, but straight to ASE.

On the occurrence of ASE in patients with IGE-GTCS (without absences)

The observation that our patients 5, 7, 10, and 11 with IGE-GTCS also had episodes of absence status (Table 2a) generates important questions: From the clinical viewpoint, the concept of IGE patients without absences developing ASE appears as an oxymoron; from the taxonomy standpoint, and if we accepted that an ASE episode is a seizure type, the combination of GTCS and ASE would conflict with the strict definition of IGE-GTCS. Genton and colleagues (2008) described some patients with similar clinical features and proposed a distinct syndrome coining the term “absence status epilepsy” (Genton et al., 2008). Eight of their 11 patients had GTCS and ASE; in two GTCS followed numerous episodes of ASE by 18 and 29 years, respectively, and in six GTCS either preceded ASE by 1–36 years (patients 4 and 6–8) or begun at the same age (patients 3 and 11). Zambrelli et al. (2006) also reported on an 83-year-old man with a GTCS on awakening at age 20 years and several episodes of ASE culminating in GTCS and becoming frequent after his early 70s.

Our group and others have reported that some IGE syndromes are more closely associated with ASE than other syndromes (Panayiotopoulos et al., 1992; Agathonikou et al., 1998; Walker et al., 2005; Thomas et al., 2006), but none appears to be immune; moreover, some patients appear prone to recurrent episodes of ASE while others with the same absence syndrome (such as IGE-PA, JAE, and JME) never have one (Agathonikou et al., 1998; Koutroumanidis, 2008). Therefore, the responsible pathophysiological mechanisms for ASE can hardly lie with the mere existence of absences in the clinical syndrome. It also appears that ASE pathophysiology is not uniform; extreme cortical firing (Koutroumanidis, 2008) and defective termination mechanisms (Fig. 1) may be involved, influenced by environmental and circadian factors, and facilitated by some AED drugs, such as the GABAergic gabapentin, vigabatrin, or tiagabin (Panayiotopoulos et al., 1997a; Thomas et al., 2006; Panayiotopoulos, 2007; Genton et al., 2008; Koutroumanidis, 2008). In this sense, the mere existence of GSW discharges may carry an inherent (genetically influenced?) liability to ASE, and for the time being we may only predict the likelihood of ASE in the individual patient based on syndromic classification and not on pure EEG characteristics.

IGE patients with GTCS and ASE are rare (or perhaps still under diagnosed), and the available information may be insufficient to decide whether this phenotype is acceptable within the syndrome of IGE-GTCS (Koutroumanidis, 2008) or makes a distinct syndrome (Genton et al., 2008). In our opinion, definition of a syndrome on the basis of a seizure type (ASE) that shows significant electroclinical and pathophysiological diversity and is so frequently misdiagnosed (as CPSE) may be problematic.

Conclusive remarks

On the syndrome of IGE-GTCS

Strictly defined (without myoclonic seizures or absences, but with compulsory GSW on the EEG) and also comprehensively distinguished from PA, the IGE-GTCS shows a strong genetic background (39% with family history of epilepsy with 28% hereditary risk, i.e., first degree relatives with generalized seizures/epilepsies) and a clear tendency to occur on awakening. Unterberger et al., (2001), who evidently did not attempt to differentiate from PA and included few patients without GSW on their EEG, also reported a high (26%) hereditary risk. A clinical EEG syndrome with such a high electroclinical homogeneity and stability over time [without absences or myoclonic seizures for a median disease duration of 8.5 years in the series of Unterberger et al. (2001) and of 17 years in ours (Table 1), and for 63 years in the case reported by Zambrelli et al. (2006)] may still lie far from final dissolution.

On the syndrome of IGE-PA

Our findings confirm and expand our previous observations (Panayiotopoulos et al., 1997b) and consolidate a homogenous electroclinical phenotype. Genton et al. (2008) wrote that “PA and short subclinical discharges are similar findings and do not justify specific diagnosis of epilepsy,” yet our work shows that using the simple HBC technique, it is possible to draw a clinically pragmatic distinction between subclinical GSW and PA, and consequently between syndromes (IGE-PA and IGE-GTCS). Identification of PA is clinically important for two reasons: First, it triggers a diagnostic process that places the patient within the spectrum of absence IGE and fosters absence-specific treatment and management. Patients can be also advised about the increased risk for ASE and how they can prevent and also treat it out-of-hospital before it ends in a GTCS. Second, the diagnostic role of PA is analogous to that of the mild (and therefore also frequently unnoticed by patients and relatives alike) myoclonic jerks in patients with late onset GTCS; the delayed diagnosis of JME in these patients and the ensuing specific management and treatment with antimyoclonus drugs are based not on the first GTCS but on the electroclinical suggestion of preexisting myoclonic seizures no matter how mild they may be. Similar arguments apply to patients with mild epigastric or experiential sensations that long predate GTCS; here, diagnosis of TLE and specific management and treatment is based on the identification of the simple focal seizure and not of the (secondary) GTCS or a possible complex partial status.

Finally, the recognition of minor seizures is also a principal tool for the classification of epilepsies and epileptic syndromes. In this paper for example, PA not only defined the homonymous IGE-PA syndrome, but—by virtue of their own absence—they also contributed to the better definition of another syndrome (IGE-GTCS), yielding a high hereditary risk that beckons genetic analysis.

On some limitations, a possible overlap and a methodological proposal

It is possible that some of our patients with IGE-GTCS might have shown PA if they were subjected to prolonged video EEG telemetry or long ambulatory recordings. Prolonged monitoring however is not indicated for the clinical assessment of patients with infrequent GTCS as in this series, and certainly detailed testing of possible impairment of cognition during GSW over such a time scale is not pragmatic. Also, despite using HBC, we cannot exclude some possible overlap between the two syndromes. This would not be an argument against their distinction, as limited overlaps occur between all IGE syndromes; examples include JME and JAE, JAE and CAE, and JME and praxis-induced epilepsy or reading/language-induced epilepsy or eyelid myoclonia with absences. It is also likely that some of our patients with IGE-GTCS might show mild impairment of cognition during their discharges if more refined testing were used; sophisticated methods based on complicated behavioral tasks have been used on an experimental basis (Porter & Penry, 1973; Browne et al., 1974). However, for clinical purposes, the optimal method to detect absences must be sensitive to distinguish between useful clinical syndromes, but also simple, so that the required task can be routinely executed by all patients and in all EEG laboratories. In addition, it must be capable of setting a socially acceptable cutoff point; driving restrictions for example should not be based on EEGs just showing interictal epileptiform activity. The HBC method utilizes the activating potential of hyperventilation and has all these advantages. In addition, being a rather continuous mental and motor process as it involves sequential planning and execution, is less likely to miss the important for consciousness first 1–2 seconds of the GSW (Porter & Penry, 1973; Browne et al., 1974). Notably, none of our patients with IGE-PA had any problem to recall loud verbal stimuli given during GSW; these are often pronounced well after GSW onset, even by the most able technologists, and therefore are only able to detect lasting disturbance of cognition during prolonged discharges. Without routinely applying HBC, all our patients with IGE-PA would have been diagnosed as IGE-GTCS (blurring a high hereditary risk and a clear occurrence on awakening) or (those with additional ASE) as “absence status epilepsy” (Genton et al., 2008). Endorsement of HBC would standardize the diagnosis of absences and improve interrater reliability and communication between epileptologists and homogenize protocols for much needed future clinical research.


We would like to express our gratitude for the generous moral support and the charitable donation kindly bestowed to us by Mrs. Ellis Stewart in honor of the memory of her belated son Stewart; her support encourages us to continue our research on epilepsies.

Conflict of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors has any conflict of interest to disclose.