Epilepsiu, 41(10):1269-1275, 2000 Lippincott Williams & Wilkins, Inc., Baltimore 0 International League Against Epilepsy
How Well A
Can Epilepsy Reassessment
Syndromes Be Identified at Diagnosis? 2 Years After Initial Diagnosis
*Anne T. Berg, TShlomo Shinnar, $Susan R. Levy, $Francine M. Testa, $Susan Smith-Rapaport, and $Barbara Beckerman
*Depai%nent of Biological Sciences, Northern Illinois University, DeKalb, Illinois, U.S.A.; fDepaments of Neurology and Pediatrics and the Comprehensive Epilepsy Management Center, Montefore Medical Center, Albert Einstein College of
Medicine, Bronx, New York, U.S.A.; and Deparhents of $Pediatrics and Neurology, Yale University, New Haven, Connecticut, U.S.A.
Summary: Purpose: Epilepsy syndromes can be identified very early in the course of a seizure disorder. It is unclear how accurate and resilient such early classifications are. We compared the classification of epilepsy syndromes made pre- viously on the basis of information available at diagnosis with those made 2 years later in a cohort of children with newly diagnosed epilepsy.
Methods: Children (n = 613) were prospectively identified at the t ime of initial diagnosis by participating physicians in Connecticut between 1993 and 1997. Classification of epilepsy syndrome according to International League Against Epilepsy guidelines was made previously based on all relevant informa- tion available at diagnosis. All cases were reclassified again after 2 years of additional evidence had accumulated. The dis- tributions of syndromes at diagnosis and at 2 years are com- pared and reasons for changes examined.
Results: After 2 years, syndromes remained the same in 86.3% of the cohort and changes occurred in 13.7% (n = 84). Evolution of the syndrome occurred in 24 children (3.9%), and rectification to the initial diagnosis occurred in 60 children
(9.8%). The most common scenario for evolution of a syn- drome was from West syndrome (n = 5), undetermined (n = 4), or symptomatic localization-related epilepsy (n = 3) to the Lennox-Gastaut syndrome. The most common rectification of initial classifications involved incompletely classified syn- dromes (cryptogenic localization-related and undetermined syndromes; n = 36). In a few instances, a fully specified syn- drome was reclassified to another apparently unrelated syn- drome. In these cases, initial information at diagnosis had been difficult to interpret.
Conclusions: Epilepsy syndromes can, for the most part, be identified at the time of initial diagnosis. Two years la- ter, rectifications were made in only 9.8% of cases, and most of these involved syndromes that represented incomplete classifications in the f i s t place. Significant changes were rare. The International League Against Epilepsy classifi- cation of the epilepsies can be meaningfully applied in epide- miological studies of newly diagnosed pediatric epilepsy. Key Words: Childhood-onset epilepsy-Epilepsy syndromes- Classification-Epidemiology
The syndromiC approach to classlfymg epilepsy rec- ognizes epilepsy as a set of diverse disorders (1). Ideally, knowledge of an individuals syndrome should provide information about the underlying etiology and the prog- nosis with respect to seizures and epilepsy as well as help determine initial treatment and management strategies. To some extent, accurate classification may require ad- ditional evidence (including about the outcomes) accu- mulated over time to determine the syndrome. Three other studies, one from the Netherlands of children with
newly diagnosed epilepsy (2), one from France of chil- dren and adults with a single seizure or newly diagnosed epilepsy (3), and one from the United States of children identified at the time of a first unprovoked seizure (4), have examined this issue and found that, 2 to 5 years after diagnosis, approximately 10 to 12% of patients are reassigned to a different syndrome than was originally assigned at diagnosis. In this follow-up to our previous report (3, we examined the classification of pediatric epilepsy syndromes 2 years after initial diagnosis and the reasons for changes since diagnosis.
Acceoted June 12.2000. METHODS A prospective cohort of 613 children was recruited at
the time of the initial diagnosis with epilepsy (two or
Addrkss correspondence and reprint requests to Dr. Anne T. Berg at Department of Biological Sciences, Northern Illinois University, DeKalb, IL 601 15, U.S.A. E-mail: email@example.com
1270 A. T. BERG ETAL.
more unprovoked seizures on separate days). Full details of the study methods have been provided previously (5). The following provides a synopsis. Children were re- mited in Connecticut from 1993 through 1997. Initial diagnostic assessments were those done by the treating physician, usually a child neurologist. AU but five chil- dren had electroencephalograms (EEGs), and 79.6% of the cohort had some form of neuroimaging, including 63% who underwent magnetic resonance imaging (MIU). Syndromes, seizure types, and causes .were clas- sified according to International League Against Epi- lepsy (ILAE) published criteria (C8). For more detailed criteria for individual syndromes, we relied on Roger et al. (9). Etiology was separately categorized as idiopathic, cryptogenic, and remote symptomatic. Specific disorders were recorded for remote symptomatic causes. Previ- ously, syndromes and seizures had been independently classified by each of three child neurologists (S.S., S.R.L., F.M.T.) on the basis of all relevant clinical data obtained at the time of initial diagnosis ("initial classifi- cation"), and discrepancies and concerns were resolved in conference (5,lO). All pertinent information from the medical record was used.
Study personnel conducted follow-up calls with par- ents every 3 months, and medical records were reexam- ined at 6-month intervals for subsequently obtained clinical information and test results. Discrepancies be- tween medical records and the parents' reports were re- solved through discussion with the parents and sometime the neurologist.
After 2 years of follow-up, each child's etiology and syndrome were reassessed based on baseline information as well as all subsequent information accumulated during the 2 years of follow-up. Etiology could be reclassified because of a change in the syndrome to or from an id- iopathic syndrome or because of new evidence regarding an underlying etiology. Developmental regression or failure to develop after initial diagnosis in a child con- sidered normal at the onset of epilepsy was not grounds for reclassifying the initial etiology.
For syndrome, two types of potential changes were considered: First, the initial classification of the syn- drome was reassessed based on evidence accumulated during 2 years from additional imaging studies, EEGs, and the seizure history. For example, if a child with cryptogenic etiology was initially classified with a form of epilepsy that was undetermined to be either focal or generalized (essentially unclassifable) and a subsequent EEG revealed a clear seizure focus not related to an idiopathlc partial epilepsy, the syndrome would be re- classified as symptomatic localization-related epilepsy by virtue of localization. In such a case, there is no change in the underlying epilepsy, and this represents essentially a rectification of the initial classification. In addition, there were instances in which the initial clas-
sification was not rectified but during the course of 2 years the epilepsy evolved to a different syndrome. A well-known example of this is West syndrome, which often evolves into the Lennox-Gastaut syndrome (9). Reasons for changes in the syndrome were recorded. When changes were made, we determined as accurately as possible when the changes first became apparent.
After 2 years, any information that came to light that might alter the classification of the underlying syndrome or etiology was reviewed independently by each of the three study neurologists. Any discrepancies in their clas- sifications were discussed in conference, and a consensus classification was reached. When necessary and possible, original EEG and imaging studies were obtained for clarification of findings described in the reports.
For children who were lost to follow-up before 2 years, the syndromes were reassessed based on what was available. If critical information that resulted in a change in their syndromes came to light, their syndromes were reclassified. Otherwise, they were left unchanged but flagged as having less than 2 years of follow-up.
The median age at onset was 5.3 years, and the ethnic distribution was comparable to that in the state. These and other characteristics of the cohort have been de- scribed previously (5). Of the initial 613 children re- cruited into the study, 20 (3.3%) were not followed for a full 2 years (2 deaths, 3 refusals, and 15 currently lost to follow-up). A re-review of all information after 2 years was necessary in 219 children (35.7%).
Etiology There were 28 changes made to etiology (Table 1).
Fourteen were attributable simply to a change in the classification of the syndrome from cryptogenic to idio-
TABLE 1 . Etiology at initial diagnosis and as reassessed afier 2 years of follow-up
Etiology as assessed after 2 years
Remote At initial diagnosis Total Idiopathic Cryptogenic symptomatic
Idiopathic 185 183 1 1" Cryptogenic 317 14 292 1 1" Remote symptomatic 11 1 1" 0 110" Total 613 198 293 122
"One child initially classified as having idiopathic etiology was recognized in retrospect to have autism predating the onset of his epilepsy. Another child initially classified as having remote symptom- atic etiology secondary to a brain cyst was reclassified as having benign rolandic epilepsy with an incidental choroid fissure cyst. Eleven chil- dren initially classified with cryptogenic etiology had their etiology reclassified as intrauterine insult (n = 3). brain malformation (n = 2). tumor (n = 2), and neurodegenerative disorder (n = 4). Five children initially assessed with remote symptomatic etiology had their etiology reclassified from atypical trisomy 15 to Angelman's syndrome (n = l), from intrauterine insult, perinatal hypoxia, or trauma to neurodegen- erative disease (n = 3), and from vascular malformation to tumor (n = 1).
Epilepsia, Vol. 41, No. 10, 2000
HOW WELL CAN EPILEPSY SYNDROMES BE IDENTIFIED AT DIAGNOSIS? 1271
pathic, and one from idiopathic to cryptogenic. One child was reclassified from remote symptomatic based on the initial interpretation of an MRI scan as showing a cyst in the temporal lobe to idiopathic when, later, after initial classifications were complete, the MRI scan became available for re-review. Ln 12 children, evidence of a remote symptomatic etiology became apparent after the initial diagnosis. In another five children initially con- sidered to have a remote symptomatic etiology, a differ- ent etiology became evident. Of note, by 2 years, seven children were reassessed as having neurodegen- erative conditions.
Syndromes A change in the epilepsy syndrome occurred in 84
children, 60 because of a rectification to the initially assigned syndrome and 24 because of a true evolution of the syndrome. The initial distribution of syndromes and the distribution at 2 years are shown in Table 2 with a summary of the types of changes that occurred. Spe-
cific changes and reasons for changes are presented in Table 3.
Evolution of syndromes fell primarily into a few cat- egories (Table 3) . Half of the evolutions were to the Lennox-Gastaut syndrome. Half of the remaining evolu- tions were to another form of cryptogenic or symptom- atic generalized epilepsy. One child had multiple evo- lutions. The evolution to another syndrome initially became apparent during the f i s t 6 months of follow-up for 14 children (58%), during the second 6 months for 2 children (8%), and during the second year for 8 chil- dren (33%).
In 60 cases, we felt that the syndrome assigned at initial diagnosis was incorrect or partially incorrect, and the child's syndrome was reclassified at 2 years on the basis of information accumulated since diagnosis. Most rectifications to the initial classification of the syndrome fell into a limited number of categories. Almost half of the reclassifications involved children either (a) who were initially assigned to the cryptogenic localization-
TABLE 2. Distributions of syndromes at initial diagnosis and after 2 years and summary of changes
1 .Localization-related 1.1 Idiopathic localization-related
I. 1.1 Benign rolandic epilepsy 1.1.2 Childhood epilepsy with occipital paroxysms
Related with seizures characterized by specific modes of precipitation Symptomatic localization-related"
1.3 Cryptogenic localization-related
2.1 Idiopathic generalized
1.2 Symptomatic localization
2.1 .O Primary generalized not further classified 2.1.3 Benign myoclonic epilepsy in infancy 2.1.4. Childhood absence epilepsy 2.1.5 Juvenile absence epilepsy 2.1.6 Juvenile myoclonic epilepsy 2.1.7 Epilepsy with generalized tonic-clonic seizure 2.1.8 Other primary generalized 2.1.9 with seizures precipitated by specific modes of activation
2.2.0 Generalized cryptogenic/symptomatic not further classified 2.2.1 West syndfome" 2.2.2 LeMOx-GaStaUt syndrome 2.2.3 Epilepsy with myoclonic-astatic seizures (Doose) 2.2.4 Epilepsy with myoclonic absences
2.3.1 Symptomatic generalized epilepsy (nonspecific etiology) 2.3.2 Symptomatic generalized epilepsy (specific syndromes)
2.2 Cryptogenic/symptomatic generalized
2.3 Symptomatic generalized
3. Undetermined 3.1 With partial and generalized features
3.1.2 Severe myoclonic epilepsy in infancy 3.1.3 Epilepsy with continuous spike wave during slow-wave sleep 3.1.5 Other undetermined epilepsies not def ied above
3.2 Epilepsies without unequivocal generalized or focal features (unclassified)
At initial diagnosis
After 2 years
Evolution of syndrome
Into Out of
Correction of initial diagnosis
Added to Taken from
359 (58.6) 61 (10.0) 59 (9.6) 2 (0.3)
195 (3 1.8) 1 (0.2)
194 (31.6) 103 (16.8) 178 (29.0) 126 (20.6)
l(0.2) 1 (0.2)
74 (12.1) 15 (2.4) 12 (2.0) 2 (0.3)
18 (2.9) 3 (0.5)
43 (7.0) 3 (0.5)
24 (3.9) 4 (0.7)
10 (1.6) 2 (0.3) 9 (1.5) 4 (0.7) 5 (0.8)
76 ( I 2.4) 5 (0.8) 1 (0.2) 1 (0.2) 3 (0.5)
366 (59.7) 66 (10.8) 63 (10.3) 3 (0.5)
209 (34.1) 0 (0)
209 (34.1) 91 (14.8)
195 (31.8) 135 (22.0)
1 (0.2) 2 (0.3)
76 (12.4) 17 (2.8) 15 (2.4) 2 (0.3)
18 (2.9) 4 (0.7)
51 (8.3) 3 (0.5)
19 (3.1) 19 (3.1) 8 (1.3) 2 (0.3) 9(1.5) 3 (0.5) 6 (1.0)
52 (8.5) 2 (0.3) 1 (0.2) l(0.2) 0 (0)
1 9 0 0 0 0 0 0 1 8 0 0 1 8 0 1
13 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
18 9 2 0 3 9
13 0 0 0 0 0 2 0 0 0 2 0 2 5 0 0 0 0 0 0 0 0 2 5
21 6 5 0 4 0 1 0
29 8 0 1
29 7 4 15
10 4 11 2 0 0 1 0 2 0 2 0 3 0 0 0...