- Top of page
- PATIENTS AND METHODS
Summary: Purpose: To evaluate how diagnostic criteria influence remission rates for patients with childhood absence epilepsy (CAE) and to assess clinical and EEG parameters as predictors of outcome.
Methods: One hundred nineteen patients were diagnosed with CAE, according to International League Against Epilepsy (ILAE) classification criteria. They were subsequently evaluated according to stricter diagnostic criteria. Sixty-two subjects fulfilled these criteria as group 2; 57 did not and constituted group 1. Diagnostic parameters that prevented patients of group 1 from entering group 2, and variables such as sex, familial history of generalized epilepsy, and personal history of febrile convulsions also were tested as prognostic factors for terminal remission.
Results: Compared with those in group 1, patients of group 2 had significantly higher rates of seizure control (95% vs. 77%), higher rates of terminal remission (82% vs. 51%), fewer generalized tonic–clonic seizures (8% vs. 30%), and shorter mean periods of treatment (2.2 vs. 3.8 years). Significantly fewer patients were receiving polytherapy in group 2 than in group 1 (11% vs. 47%), and fewer patients had seizure relapses at antiepileptic drug discontinuation (0 vs. 22%).
Conclusions: Remission rates of patients with CAE are greatly influenced by the classification criteria used for selection. Stricter diagnostic criteria allow the definition of a homogeneous group of patients with excellent prognosis. Factors predicting unfavorable prognosis were generalized tonic–clonic seizures in the active stage of absences, myoclonic jerks, eyelid myoclonia or perioral myoclonia, and EEG features atypical for CAE.
Childhood absence epilepsy (CAE) is a well-known syndrome with onset in middle childhood and is characterized by multiple typical absences (TAs) per day (1–4). TA also may occur in other recognized idiopathic generalized epilepsies (IGEs), such as juvenile absence epilepsy (JAE), juvenile myoclonic epilepsy (JME), and epilepsy with myoclonic absences (5), as well as in the newly described syndromes of eyelid myoclonia with absences and perioral myoclonia with absences (6,7). Because these syndromes have different prognoses and outcomes, it is important to define the syndromic classification for every patient rigorously (6). In this context, the literature regarding the evolution and prognosis of patients with CAE is considerable but still inconclusive (8) because of the wide range of diagnostic criteria used (9). Unfavorable prognostic factors have been considered to be age at onset, appearance of generalized tonic–clonic seizures (GTCSs) (3,10), types of absences (11), response to treatment (12,13), and types of abnormalities in the EEG background (8).
We aimed here to evaluate remission rates in patients with CAE in relation to different diagnostic criteria. We also analyzed multiple clinical and EEG factors to attempt to predict the outcome of patients with CAE.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
All patients with a diagnosis of TA seizures treated at the Department of Pediatrics, University of Siena, Italy, between January 1982 and December 2000, were retrospectively evaluated. We analyzed clinical information including age, sex, birth history, family history, neurology, age at the absence seizure onset, pattern of typical absences before treatment, additional seizure types, their onset, antiepileptic drug (AED) treatment, and treatment responses. EEGs, video-EEGs, and camera recordings were reevaluated. Computed tomography (CT) scan or magnetic resonance imaging (MRI) investigations were reexamined. Information was obtained by chart review for those patients who had their last visit in our Department ≤4 months before study inclusion, and by personal follow-up interview in others. In all patients, treatment was initiated with one drug, usually valproic acid (VPA), and the clinical response was assessed every 1–3 months until complete control was achieved. Once clinical control was achieved, the EEG was performed after 3–6 months. If VPA was unable to control the seizures, it was replaced by ethosuximide (ESM) or by a combination of both drugs. Lamotrigine (LTG), clonazepam (CZP), and topiramate (TPM) also were used as add-on therapy in those patients with resistant seizures.
Patients were first evaluated in accordance with the diagnostic criteria proposed by the 1989 International League Against Epilepsy (ILAE) proposal for a revised classification of epilepsies and epileptic syndromes (5), as reviewed by Loiseau (3). They were (a) onset of TAs before puberty; (b) normal neuromotor development when TAs occurred; (c) absence seizures as the initial type of seizures; (d) very frequent TA seizures occurring many times per day; (e) TAs associated in the EEG with bilateral, symmetrical, and synchronous discharge of regular 3-Hz spike-and-wave complexes with normal or mildly abnormal background activity. Patients were further evaluated in accordance with stricter diagnostic criteria as proposed by Panayiotopoulos (14) and reviewed by Loiseau et al. (9), and thereafter defined in this article as Panayiotopoulos' criteria (Table 1). The patients who fulfilled the latter diagnostic criteria were classed as group 2, and those who did not were included in group 1.
Table 1. Diagnostic criteria of CAE described in Panayiotopoulos's proposal for absence epilepsies
| 1. Frequent (many per day), brief ∼10 s, >4 s) typical absences with abrupt and severe impairment of consciousness|
| 2. Age at onset between 4 and 10 years|
| 3. EEG ictal discharges of generalized high-amplitude spike (or no more than double) spike- and slow-wave complexes, which are rhythmic at ∼3 Hz, with a gradual and regular slowdown from the initial to the terminal phase of the discharge. Duration varying from 4 to 20 s|
| 1. Other than typical absences such as GTCSs, or myoclonic jerks before or during the active stage of absences|
| 2. Absences with marked eyelid or perioral myoclonus, single or rhythmic massive limb jerking, and single or arrhythmic myoclonic jerks of the head, trunk or limbs|
| 3. Absences with mild or not clinically detectable impairment of consciousness during the 3- to 4-Hz discharges|
| 4. Stimulus sensitive absences (photosensitive, pattern-sensitive, fixation-off, etc.)|
| 5. Discharge fragmentation (within 1 s) and multiple spikes;|
| 6. Irregular, arrhythmic spike and multiple spike and slow-wave discharges with marked variations of the intradischarge frequency or of the spike and multiple spike- and slow-wave relations|
| 7. Predominant brief discharges of 3- to 4-Hz spike-wave of <4 s|
| 8. Fixed “lead in” anomaly in the frontal region on EEG|
Analysis of prognostic factors
We defined terminal remission of CAE as complete absence of seizures without AEDs for ≥1 year before final follow-up. We analyzed the real predictivity of each of the diagnostic criteria used to exclude patients of group 1 from group 2 (Table 3). Variables such as sex, any family history of generalized epilepsy, and a personal history of febrile convulsions were also considered.
Table 3. Diagnostic criteria excluding patients of group 1 from entering group 2
|Diagnostic parameter||No. of patients|
|EEG features atypical of CAE||17|
|Myoclonic jerks during TA||4|
|GTCSs during the active stage of absences||6|
|TA triggered by photosensitivity||4|
Data processing and analysis were performed on STATA SE 8.2 (Stata Corp., College Station, TX, U.S.A.). We used χ2 and Fisher's exact tests for group comparisons at nominal-data level and analysis of variance for parametric data. Relations between outcome and variables or their association, and the identification of subsets of independent variables that could be good predictors of the dichotomous outcome variable were checked by using multivariate logistic regression models with forward step-wise conditional selection of variables.
- Top of page
- PATIENTS AND METHODS
One hundred eighty-nine patients were identified as affected by TAs. Seventy patients were excluded for the following reasons: lost at follow-up (n = 26); no clearly defined seizure semiology before treatment (n = 26); psychomotor retardation (n = 7); GTCSs occurring before TAs (n = 5); unknown age at TA onset (n = 3); or associated partial seizures (n = 3). One hundred nineteen patients were identified as affected by CAE in accordance with the ILAE diagnostic criteria as reviewed by Loiseau (3). Of these, 62 fulfilled Panayiotopoulos' criteria (9,14) and were therefore classed as group 2, whereas 57 did not and entered group 1. Table 2 summarizes the main findings of the two groups according to the different diagnostic criteria. In detail, the longer period of treatment observed in group 1 (p < 0.03) was mainly influenced by the time necessary to reach seizure control and normalization of EEGs. Discontinuation of AEDs (data not shown in Table 2) was attempted in 36 patients of group 1 and 48 of group 2. TA relapses occurred in eight (22%) patients, and none, respectively (p < 0.002). Significantly more patients still under treatment at the last visit were in group 1 (p < 0.001). In that group, 10 patients (three of whom had TAs only, and seven showed an association of TAs and GTCSs) were still receiving therapy although they had their seizures controlled. Eighteen patients still had with TAs only (n = 3), TAs plus GTCSs (n = 8), GTCSs only (n = 2), TAs plus myoclonic jerks (n = 2), and MJ only (n = 2). Irregularity in AED use was likely in seven of the latter patients. The epilepsy history of four (7%) patients of group 1 met the diagnostic criteria of JME. In group 2, six patients (four with TAs only and two with TAs plus GTCSs) were still receiving treatment, although they had their seizures controlled. In the remaining five patients (one with TAs only, one with TAs and GTCSs, and three with GTCSs), sporadic seizures were still present at the last visit. Table 3 lists the diagnostic parameters that prevented patients of group 1 from being classed as group 2. EEG features “atypical” for CAE included evident photoparoxysmal response (n = 9); irregular ictal 3- to 4-Hz spike–wave complexes (n = 6); abnormal EEG background, defined as excessive slowing during the waking record (n = 2); fixation-off (n = 2); and fixed “lead-in” anomaly in the frontal region (n = 1).
Table 2. Main characteristics of two groups of patients
|Clinical findings||Group 1||Group 2||p Value|
|No. of patients||57||62|| |
|Girls (%)||32 (56%)||40 (65%) ||NS|
|Family history of epilepsy/FC (%)||15 (26%)||8 (13%)||NS|
|Personal history of FC (%)||5 (9%)||9 (15%)||NS|
|Age at follow-up (yr)|
|Mean (range)||23.1 (14–27)|| 21.8 (14–25.5)||NS|
|TA as the sole type of seizure (%)||40 (70%)||57 (92%) || 0.004|
|Controlled TA (%)||44 (77%)||59 (95%) || 0.01 |
|Myoclonic jerks||4 (7%)||—||—|
|GTCSs||17 (30%)||5 (8%) || <0.004 |
|Mean (range) age at GTCS onset (yr)|| 14.8 (3.0–21)|| 13.5 (9.5–19.4)||NS|
|Age at TA onset|
| <4 yr||4 of 18 (22%) ||—||—|
| 4–8 yr||1 of 8 (12%) ||5 (8%)||NS|
| >8 yr||12 of 31 (39%) ||—||—|
|Patients off therapy|| 29 (51%) ||51 (82%)|| 0.001|
| Mean period of treatment (yr)|| 3.8 (2–5.2)|| 2.2 (2–2.8)|| 0.03 |
|Patients submitted to polytherapy|| 27 (47%)|| 7 (11%)|| 0.001|
Outcome (the dependent variable) was coded as 0 for terminal remission and 1 for no terminal remission. The following independent variables were considered: age at TAs onset; GTCSs during the active stage of the disease; EEG features atypical for CAE; myoclonic jerks while taking drugs; presence of eyelid or perioral myoclonias; TAs triggered by intermittent photic stimulation; sex; family history of generalized epilepsy; and personal history of febrile convulsions. These were checked by univariate analysis and considered in a multivariate logistic regression analysis with forward step-wise conditional selection of variables. The cutoff value of significance chosen for each variable was 0.05 to enter into the model and 0.1 for removal. Variables meeting these criteria were, in order of decreasing significance: GTCSs during the active stage of absences (model χ2= 31.9; p = 0.000); myoclonic jerks (model χ2= 17.8; p = 0.000); eyelid or perioral myoclonia (model χ2= 15.2; p = 0.000); EEG features atypical for CAE (model χ2= 9.5; p = 0.002); TAs triggered by IPSs (model χ2= 9.3; p = 0.002); age at TA onset (model χ2= 8.4; p = 0.004); and family history of generalized epilepsy (model χ2= 6.1; p = 0.013). The remaining variables considered were not significant.
Partial correlations between the dependent and each independent variable with the relative estimated coefficients, significance (p), and odds ratio, showed that the logistic model correctly predicted 79% of cases (Table 4).
Table 4. Logistic regression analysis results
|GTCSs during the active stage of disease||−3.8723||0.7293||−5.31||0.000||0.0208|
|Eyelid or perioral myoclonias||−3.6412||0.0985||−3.69||0.000||0.0262|
|EEG features atypical of CAE||−3.4621||0.0764||−3.43||0.006||0.0211|
- Top of page
- PATIENTS AND METHODS
Most of the available evidence is inconclusive regarding evolution and prognosis of CAE, and a wide range of remission rates (33%–79%) have been reported in the literature (4,7,10–13,15–18). This is because of the different classification criteria adopted in the studies and because of diverse follow-up periods (9,19). Moreover, most of the studies were performed before 1989, with possible difficulties in demarcating between patients with CAE and other types of epilepsy with TAs with less favorable prognosis (4). In one population-based study, 65% of patients with CAE were seizure free and off therapy, 17% had relapses at the discontinuation of AEDs, and 11% had continuing seizures despite medication. Fifteen percent of the total population had progressed to JME (20). Better results were observed in another population-based study in which absences persisted beyond the age of 20 years in only 10% of patients (21). Isolated or rare GTCSs occurred in 26% and were related to the patient's age at onset of TAs, being more common (44%) among patients with the onset of TAs between ages 9 and 10 years than in those for whom TA onset was before 9 years (16%). However, stricter diagnostic criteria than those previously proposed (3) were considered in that study (21). Indeed, patients with TA onset after age 10 years and those with a positive photoparoxysmal response were excluded from the analysis (21). A lower rate of total terminal remission (56%) was recently reported in a hospital-based study in patients with CAE. The predominant seizure pattern at onset in association with later development of myoclonic jerks or GTCSs was found to be useful in predicting terminal remission in patients with CAE and JAE (22).
We evaluated how different diagnostic criteria might influence the outcome for patients with CAE. For this, patients with CAE fulfilling diagnostic criteria proposed by ILAE, as reviewed by Loiseau (3), were compared with those who further satisfied the stricter Panayiotopoulos's diagnostic criteria (14), as recently reviewed (9). Of course, retrospective studies lack of the rigor of prospective investigations, but we believe they may provide useful insight into the outcome of CAE.
Some of the inclusion/exclusion criteria (Table 1) that we considered in group 2 deserve comments, and in particular, whether the term puberty used in ILAE diagnostic criteria is poorly informative. Age at onset of TAs (4–10 years) certainly excludes some patients with unquestionable CAE, which begin later than age 10 years or before 4 years. However, the lowest and highest age at onset compatible with CAE is uncertain (9,14). We excluded from group 2 those patients in whom intermittent photic stimulation (IPS) triggered TAs or an evident photoparoxysmal response on EEG recordings, because these features are not characteristic of CAE (9,23). IPS is inherited independent of EEG discharges evoked by hyperventilation (16). Focal interictal spikes are not related to an unfavorable outcome (9,24,25), but the prognostic value of the fixed frontal “lead in" anomaly is inconclusive. Indeed, patients with ictal EEGs showing this feature have been considered to belong to the “frontal absence group," with a higher incidence of learning problems and less-controllable absences (26). Although these findings must be confirmed (25), we considered ictal frontal lead-in as an exclusion criterion. Sixty-two patients of the present series (group 2) fulfilled Panayiotopoulos's diagnostic criteria; the remaining 57 (group 1) were considered affected by CAE only when the ILAE criteria, as reviewed by Loiseau (3), were applied. The two groups of patients showed differences (Table 2). The rate of TAs controlled by therapy was statistically higher in group 2, as was the rate of patients in terminal remission. This group of patients had a shorter mean duration of treatment (mainly influenced by the time necessary to obtain seizure control and EEG normalization) and lower percentage of patients receiving polytherapy, when compared with group 1. As previously reported (21), the incidence of GTCSs was higher in patients with TA onset beyond 10 years than in those for whom TA onset occurred earlier. Of interest, we found a high frequency of GTCSs (23%) in subjects with TA onset before age 4 years. Few studies describe the outcome of patients with TA onset earlier than this age (8,16,27). Similar findings have already been reported, albeit with lower incidence (16).
We also tested the real predictivity for CAE outcome of those diagnostic parameters that excluded patients of group 1 from entering group 2 (Table 3) together with sex, a family history of generalized seizures, and personal history of febrile convulsions. EEG findings considered atypical for CAE diagnosis were considered as a whole in the statistical analysis. We found by univariate analysis that all variables, except sex and a personal history of febrile convulsions, showed statistically significant differences between groups 1 and 2. This suggests their possible utility for recognizing patients with potential unfavorable prognosis. Of the subset of independent variables tested by multivariate analysis and multivariate logistic regression analysis for predictivity of CAE outcome, GTCSs in the active stage of absences ranked first in the hierarchy, followed by myoclonic jerks, eyelid or perioral myoclonia, and EEG findings “atypical” for CAE (considered as a whole). Our study confirms the ominous effect of GTCSs during the active stage of disease on terminal remission of CAE. Indeed, this criterion is strongly indicated as an unfavorable prognostic factor for patients with CAE (9,14). Four (7%) children from group 1 had an epilepsy course congruent with a diagnosis of JME. None of them had terminal remission. Unfortunately, as previously reported (20), remission in these patients could be predicted only by the course of their epilepsy. In none of the patients of group 2 did JME develop. This may have occurred by chance, as in theory, even when applying strict diagnostic criteria, the possible evolution of CAE to JME cannot be predicted (28). By contrast, strict classification parameters may allow one to exclude from CAE diagnosis those patients with eyelid or perioral myoclonia with absences who, as with patients with JME, notoriously show a less-favorable prognosis requiring lifelong medications (29). Slowing of the EEG background has predicted lack of remission of TAs in some studies (8,30), but not in others (8). Less studied is the predictivity for terminal remission of photoparoxysmal response on initial EEG. Our data clearly demonstrated that these EEG features are good predictors for lack of remission in CAE.
Many other studies have found that patients who have a prompt response to AEDs or those who need monotherapy tend to have a better outcome than do those whose disease remains uncontrolled for >1 year (16,20). Our study seems to indicate that these findings may be considered as epiphenomena of classification criteria used to select patients rather than real prognostic factors.
Finally, the application of strict diagnostic criteria allowed the identification of patients with specific epilepsy syndromes whose outcome is not as favorable as that of CAE. However, strict diagnostic parameters also exclude from CAE those patients with typical absences, whose electroclinical pattern is not evocative of recognized TA-related syndromes but with less-favorable outcome.
In conclusion, our data indicate that the outcome of CAE is greatly influenced by the diagnostic criteria used to select patients. Stricter diagnostic criteria allow us to define a homogeneous group of patients with an excellent prognosis, because the disease responds well to therapy (commonly monotherapy) and needs a shorter period of treatment with a low risk of TAs recurring when AEDs are withdrawn. Around this homogeneous group of patients, a gray zone includes subjects whose electroclinical patterns may be or may not be evocative of recognized TA-related syndromes. For these patients, terminal remission rates are lower, with a higher rate of relapses when AEDs are discontinued.