PEDIATRICS Vol. 107 No. 5 May 2001, pp. 1232-1235

A Long Letter and an Even Longer Reply About Autism Magnetoencephalography and Electroencephalography

To the Editor.

As discussed below, the paper by Lewine et al raises serious concerns, including ethical ones, which were not adequately addressed by Dr Neville (who also seems to be a proponent of neurosurgery for autism) in his commentary, and prompts the following comments (I wish it had been possible to reply more succinctly):

Lewine et al (Pediatrics 1999;104:405-418), citing "clinical parallels" between Landau-Kleffner syndrome (LKS) and "regressive autism," suggest that subclinical epilepsy in the form of "epileptiform activity" on electroencephalography (EEG) or magnetoencephalography (MEG) may cause the apparent regression of language and behavior seen in some children with autism. Although it is well-known that epilepsy occurs in some individuals with autism, the authors fail to emphasize that the significance of an epileptiform EEG in children without seizures is controversial. Moreover, they conflate "regressive autism" and LKS, although the latter is a distinct entity for which the etiology and significance of EEG changes remain uncertain.^1,2 The matter is further confused by using the term "LKS variant" in relation to 20 children for whom "language dysfunction was considerably more severe than deficits in social or other cognitive domains." There is no consensus as to a precise definition of "LKS variant."

They attempt to buttress their argument that epileptiform activity plays a primary, causative role in autistic regression by citing the frequent association of seizures and autism in a number of conditions (tuberous sclerosis, infantile spasms, etc.). However, the vast majority of persons with autism, 85% to 90%, do not have associated medical conditions predisposing to epilepsy.³

It is stated that "all" the children with autism spectrum disorders (ASDs) had "occasionally" shown "unusual behavioral episodes including unprovoked blinking, crying, and/or holding of the hands to the ears" which "would often be considered to be suggestive of a possible seizure disorder." It is facile to equate such "unusual paroxysmal behaviors" with epilepsy or an epileptiform EEG because it is a common observation that many autistic children appear to be sensitive to certain extraneous sounds and bright light, and are prone to temper tantrums and sensory self-stimulation.

Selection bias includes cases where "parental report indicated normal development up to 18 to 30 months of age ... " Of course, "normal" is in the eyes of the beholder. It is possible that some of these children may have had subtle atypical development earlier in life, rather than a period of normal development followed by true regression. A parent's retrospective recall of developmental gains in specific language milestones may not be reliable because, by the typical age of regression in autism (18 to 24 months), expressive language is still limited in many children. Furthermore, they state that 50 children were diagnosed with a "regressive ASD before the age of 6 years." However, the standard diagnostic manuals specify that the features of autism must be present before the age of 36 months.^4,5 Extending the period of observed regression to the age of 6 years overlaps with LKS, which has language regression after 36 months (typically at 5 to 7 years of age¹) and further detracts from their precision of diagnosis. This is especially the case in the instance of the child with "mild to moderate autistic features" (whatever that means), "significant language regression at 4.5 years," and an EEG "indicative of LKS" (Fig 4).

Their imprecision of diagnosis is compounded by the overlap of the DSM-IIIR and DSM-IV criteria during the time period of their study. The DSM-IV evolved because the DSM-IIIR had excessive sensitivity and overdiagnosed autism in up to 60%.³ In the absence of reascertainment of patients originally diagnosed under the DSM-IIIR, some may not have had autism at all. Two patients (ASD12 and ASD32, Table 3) are classified as "autism/PDD" although there is no such designation in the current diagnostic schema.

Whether or not the MEG changes described by these authors have clinical relevance, the suggestion that a disorder as complex as autism is amenable to a neurosurgical procedure ("multiple subpial transection" [MST]) advocated by some for LKS, at the presumed epileptic focus, is reaching well beyond the available data and current state of science regarding autism. It is naive to hold that the neurobiologic basis of social cognition, which has a multifocal topology extending over multiple brain regions, including those well beyond the reach of the scalpel blade, is amenable to such a "blunt instrument" approach. Moreover, there is no mention of institutional review board approval or informed consent for an experimental surgical procedure of dubious rationale. Serious ethical concerns arise if the parents of the 18 children with ASDs having MST (including 5 having more than 1 neurosurgical procedure) were not informed that their children were being subjected to an experimental procedure.

The authors claim that their subjects improved after MST based on parental reports of "dramatic improvement" in 4 children. The use of self-reporting by parents is a well-known source of nonobjective observations and bias. They also cite improvement in scores on the Childhood Autism Rating Scale (CARS), ranging between 1.0 and 9.0 (Table 3). They have left out the denominator. They fail to provide a statistical analysis as to the significance level of the change in scores. Nor do they state if the CARS observations were made by appropriately trained clinicians and if interrater reliability was assessed. The CARS has 15 separate items, each graded on a scale of 1 to 4. It is intended as an initial diagnostic instrument and has not been validated for serial testing as a measure of progress after a treatment intervention.

They also state that the Peabody Picture Vocabulary Test (PPVT) in 10 showed "improved" receptive language, ranging between 1.0 and 2.5 years (Table 3). The age at which the PPVT was administered is not given (the PPVT is intended for the age range 2.5 to 4.0 years). The authors do not give the time interval between baseline and follow-up measurements for either the CARS or PPVT. Gains in test scores would be expected to occur with the passage of time and as a consequence of enhanced test-taking ability based on previous experience with the test venue. Moreover, the PPVT does not say much about whether or not improvement occurred in the core social-affective disability of autism after neurosurgical intervention (MST). Table 3 also lists improvement in "eye contact" without their description of "Methods" stating the means of documenting how such observations were made and who made them. Other "improvements" are said to have occurred in expressive language, attention, and "behavior and social functioning." Casual observations of "behavior and social functioning" in children with ASDs are fraught with pitfalls and claims as to treatment-related improvement demand far greater rigor than given in their description of "Methods."

There is additional imprecision in grading the extent of cognitive improvement after high-dose steroid therapy (Table 3) as "mild," "moderate," or "major" without specifying criteria for each grade.

The authors fail to discuss, in the absence of controls, that there is a high probability of a placebo effect as an explanation of the alleged improvement. Moreover, as with all young children during an interval of follow-up after a treatment (MST, in this instance), ranging between 8 to 23 months, some "improvement" or behavioral change is likely to occur as part of the natural course of development, albeit atypical, in such children. Moreover, their claims of improvement consequent to MST are confounded by multiple variables because these children "continued to receive language, occupational, and behavioral therapy." The conflation of "autism" and "pervasive developmental disordersnot otherwise specified (PDD-NOS)" lends additional confusion because the latter includes higher-functioning individuals who, during the normal course of events, would be expected to show some progress with the passage of time, especially if other interventions are concurrently implemented.

The statement that "despite these impressive gains, these children are not now normal, epileptiform activity having robbed these children of years of unperturbed cognitive and social development" is wildly speculative and opinionated. It does not belong in a peer-reviewed medical journal.

Finally, to take the reductionist position that epileptiform activity is "causative" of ASDs and that the goal of treatment is to "eliminate epileptiform activity" is to sow confusion in a field where the significance of epileptiform activity is controversial and debate continues as to whether such changes are central to the disorder or an epiphenomenon.

This paper should have been confined to a report of MEG changes in autism. Regrettably, it was a "Dustbuster," carrying along with it MEG data in children with LKS, which was not directly relevant, undocumented "preliminary" data (for example, on page 413, " ... MEG study of 25 children with early-onset [before 18 months] ASDs has revealed sleep epileptiform activity in 70% ... "), and vague assertions, on page 415 (such as "Preliminary analyses suggest an interesting association ... " and, on the same page, "specificity seems to be high" for the "identified pattern of activity" and autism) that do not lend convincing support to their central thesis that epileptiform activity is the cause of regression in autism.

The management of these challenging children is ill-served by a report with much preliminary data that will be cited by parents as authoritative in pressuring their pediatricians to support a referral for neurosurgery.

Postscript. While I was in attendance at a workshop on the treatment of autism (November 8 and 9, 1999), under the auspices of the National Institute of Mental Health, one of the invited attendees, Dr Andres M. Kanner, a co-author of the above-referenced article, stated conclusions that were diametrically opposite to those in the published article; ie, he did not agree that multiple subpial transection was an effective treatment. The session was chaired by Edwin Cook, Jr, MD, a child psychiatrist and an autism researcher at the University of Chicago. He immediately urged Dr Kanner to state his differing conclusions in a letter to you. Donald J. Cohen, MD, head of child psychiatry at Yale, also advised Dr Kanner to do so. I believe such a disavowal is important to the readers of the journal. If you do not hear Dr Kanner's points of view, I believe that readers of Pediatrics will have been deprived of a significant bit of information in attempting to reach their own conclusions about this radical form of treatment.

Ronald J. Kallen
Children's Medical Group
Pediatric and Adolescent Medicine
Kenosha, WI 53142

REFERENCES

1. Rapin I Autism. N Engl J Med. 1997; 337:97 [Free Full Text]
2. Landau WM Landau-Kleffner syndrome. An eponymic badge of ignorance. Arch Neurol. 1992; 49:353 [Abstract/Free Full Text]
3. Barton M, Volkmar F How commonly are known medical conditions associated with autism? J Autism Dev Disorders. 1998; 28:273
4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th Revision. Washington, DC: American Psychiatric Association; 1994
5. American Academy of Pediatrics. The Classification of Child and Adolescent Mental Diagnoses in Primary Care. Diagnostic and Statistical Manual for Primary Care (DSM-PC), Child and Adolescent Version. Elk Grove Village, IL: American Academy of Pediatrics; 1996

In Reply.

Since the September 1999 publication of our article "Magnetoencephalographic Patterns of Epileptiform Activity in Children With Regressive Autism Spectrum Disorders" (Pediatrics. 1999;104:405-418), we have received a number of favorable comments, mixed with a few unfavorable critiques, as typified by the commentary submitted by Dr Kallen. This opportunity to clarify some of the relevant issues is very much appreciated. Our apologies concerning the length of this reply, but this was dictated by the length of Dr Kallen's critique.

There have been 3 critical issues raised by Dr Kallen and others concerning our report. These relate to 1) the definition, diagnosis, and selection of our subject population, 2) the relationship between MEG and EEG findings, and 3) the limitations of the surgical data presented in the "Discussion" section. The surgical data certainly have been the most controversial aspect of our report, but clarification of the first 2 issues is needed to help put the third issue into perspective.

As stated in our report, we evaluated 50 children who had a history of normal early development followed by an autistic regression of language and social skill between 18 and 30 months of age. In stating that "50 children were diagnosed with a regressive ASD before 6 years of age," we intended to indicate that formal diagnosis may not have been made until 6 years of age. But, in all cases, and as required by DSM criteria, symptom onset was always before 36 months of age. It is quite common that autistic children do not receive formal autism diagnoses before 4, 5, or even 6+ years of age, and this was the case for some of the children reported in our study. Even for the child reported in Fig 4, there was no extension of the critical symptom onset period beyond the DSM-dictated criteria of 36 months of age. This child showed normal early development followed by a loss of language skills and onset of autistic features between 26 and 30 months of age. She was diagnosed with PDD-NOS at 34 months of age. Between 3 and 4.5 years of age she showed some recovery of language (but not social) skills, but this was followed by another regression in language at 4.5 years of age. Although the second regression occurred after 36 months of age, initial symptom onset was before the DSM-specified time period. Furthermore, at the time that her MEG was performed (6 years of age), she still met DSM-IV diagnostic and research definitions for PDD-NOS. This child is one where a suggestion had been made of an LKS-variant, but, by DSM-IV criteria, PDD-NOS was the appropriate diagnosis. We concur with Dr Kallen that the term LKS-variant is highly problematic from a diagnostic perspective. Nevertheless, we felt that it was prudent to let readers know that this term had been suggested by referring doctors for 20 of our subjects. However, in each case, despite the fact that language skills were somewhat more severely compromised than social skills, the child met DSM criteria for an ASD. As far as our study goes, there was no confusion or overlap in the definitions or diagnosis for classic LKS versus an ASD.

Dr Kallen's letter raises the question of DSM-IIIR versus DSM-IV criteria. It should be noted that the DSM-IV was not available until 1994. Six of the children reported in this paper were studied before 1994, at a time when only DSM-IIIR criteria were available. Five additional children were initially diagnosed using DSM-IIIR criteria and then subsequently rediagnosed using DSM-IV criteria. This subset included subjects ASD12 and ASD32, who were listed in Table III as autism/PDD (indicating DSM-IV and initial DSM-IIIR diagnoses).

A final issue related to the entrance of children into this study concerns selection bias, and as we specified in our report, there were some biases. Our study consisted of a consecutive series of 50 regressive ASD children referred by physicians for MEG between 1992 and 1998. The fact that all cases were medically referred (rather than recruited in a nonselective fashion from an autism clinic) undoubtedly introduced some bias into the study, especially because the referring physician suspected clinical or subclinical seizures in 35 of the patients. On the one hand, as Dr Kallen suggests, it may be facile to equate unusual paroxysmal behaviors with epilepsy or an epileptiform EEG because it is a common observation that many autistic children appear to be sensitive to certain extraneous sounds and bright light, and are prone to temper tantrums and sensory self-stimulation, but ignoring the possibility that some of these behaviors do indeed reflect epileptiform activity could be viewed as negligent. Nevertheless, we are the first to admit that what is needed is a study beyond the present one that focuses on the correlation between specific behaviors and specific epileptiform patterns. Dr Kallen also suggests that there is a problem with the use of a parent's retrospective with respect to identifying a regression. This may indeed be true, but available literature offers no universally accepted alternative strategy, and if some parents are reporting regression when early symptoms were present (as Dr Kallen suggests), this would only mean that our data are of a broader significance than we suggest, with direct relevance to the more common developmental delay profile of autism.

In recent months, we have been contacted by several colleagues concerning our observation of spikes in the MEG in the absence of spikes in the simultaneous EEG, even though Ebersole and colleagues (1995, 1999) have made similar observations in patients with lateral neocortical temporal lobe epilepsies. The general finding has been surprising to some because it is generally believed that EEG is more sensitive than MEG because EEG sees both radial and tangential currents, whereas it is known that MEG is "blind" to radial source configurations. In stating that activity was seen in the MEG and not the EEG, we must clarify the manner in which the activity was recorded and analyzed. For the MEG, data were collected using a 122- or 306-channel sensor unit whereas EEG data were collected with only 19 to 30 electrodes. This most definitely raises the possibility that the EEG missed the MEG- identified spikes because of its sparse spatial sampling. One should note, however, that 19 to 30 electrodes are considered to be the clinical routine, and this is generally accepted to provide adequate spatial sampling for clinical purposes. Some have suggested that the MEG is seeing activity that "is not really there," but there is no foundation for such a suggestion. The biophysics of MEG is quite straightforward, and sensor artifacts are easily distinguished from real signals. Also, in the cases that went to surgery, intracranial EEG monitoring with subdural grids always confirmed epileptiform activity in the MEG-identified zones, without a single false-positive.

Forward simulation studies suggest that the failure of our scalp EEG to identify correlates of some of the MEG activity mostly reflected the recording montage and signal-to-noise factors, rather than spatial sampling issues, per se. EEG recordings involve assessment of the electrical potential difference between 2 scalp electrodes. The choice of the reference strategy and recording montage has dramatic effects on what can be identified by EEG. In this study, when spikes were seen in the MEG, we always examined the EEG using each of 4 montages (Cz referential, transverse bipolar, double-banana bipolar, and contralateral mastoid referential). These are montages commonly used in clinical practice, but it is possible that alternative, nontraditional montages might have revealed EEG correlates of the MEG-identified signals. In fact, in 4 of 7 cases where MEG showed spikes but initial assessment of the EEG did not, we were able to define nontraditional EEG montages that showed the activity. However, this was only accomplished in light of a priori knowledge of the time points for spikes, as identified in the MEG. Overall, our data should be interpreted as indicating that routine MEG sometimes identifies spikes that are not seen in simultaneous routine EEG, although it may be possible to increase EEG sensitivity to intra-sylvian activity using augmented sensor arrays, nontraditional montages, and nonroutine data analysis strategies.

Our presentation of preliminary surgical outcome data in Table 3 was undoubtedly the most controversial aspect of our report. Let us begin by stating clearly and unequivocally that no one on our team is advocating any form of "psychosurgery" for autism. However, we do advocate that some persons with ASDs demonstrate significant epileptiform activity that should not be ignored in the medical management of these patients. In most cases, this medical management should be via pharmacologic intervention, although, as outlined below, surgery may be needed in a small number of cases, just as it is needed in some children with intractable epilepsy in the absence of autism. We would also note that our initially submitted manuscript provided only brief mention of the surgical data and that Table 3 was added at the request of both reviewers. In providing the data it was solely our intent to provide preliminary data for discussion, knowing full well that the data were limited in terms of the type of testing performed and the lack of a double-blind design.

In general, there is strong agreement in our group that multiple subpial transection surgery is effective and warranted in cases of classic LKS where the aphasia has been present for more than 1 year and there has been a failure of sustained response to pharmacologic intervention. The positive benefits of surgical intervention in LKS are documented in the peer-reviewed literature, with Grote et al (1999) reporting presurgical and postsurgical data on language performance for 14 LKS cases that were surgically treated at Rush-Presbyterian Hospital in Chicago. The situation with respect to surgical management of epileptiform activity in ASDs is more controversial, even among the co-authors of our report. This is because the situation in ASDs is complicated by several features, including a lack of traditional clinical seizures in many cases and the tendency for the "subclinical" epileptiform activity in these cases to be multifocal in nature. For example, early experience by investigators in Chicago suggested that the unilateral, peri-sylvian multiple subpial transection (MST) strategy, which is effective in the treatment of LKS, was generally ineffective in providing seizure control and cognitive and behavioral improvements in ASDs. As a result, the surgical team at Rush-Presbyterian Hospital in Chicago does not consider surgery a viable option for treatment of epileptiform activity in children with ASDs, even given the surgical data presented herein (because of their preliminary nature and limitations as discussed). In contrast, the surgical team in Omaha has found that a multifocal MST approach can lead to control of epileptiform activity in children with ASDs. This team has documented the efficacy of multifocal MST surgery in a number of cases of complex adult epilepsy (Patil et al, 1997), and adaptation of this approach for children with multifocal epilepsies is straightforward. The critical point is that in all cases patients were identified as surgical candidates from the epilepsy perspective alone, with MEG helping to determine if the epileptiform activity is unifocal or multifocal in nature. There is nothing dubious or experimental in the rationale for surgical intervention in cases of significant epileptiform activity that cannot be controlled by medications, and, in fact, our report is the fifth report on the positive cognitive, behavioral, and social benefits of epilepsy surgery in a subpopulation of children with ASDs (see Nass et al, 1999; Patil et al, 1998; Neville et al, 1997; and Gillberg et al, 1996), although in one report (Nass et al, 1999) the improvements tended to be transitory.

With this in mind, let us now more closely examine the data presented in our Table 3. Seven of the children had clear partial complex seizures that were not controlled by medications. These clinical seizures, although rare for most of the children, had, in and of themselves, a negative impact on the child's quality-of-life. Exploration of a surgical option in cases like these is part of the standard of care for children with refractory epilepsy. That surgery was a good therapeutic strategy for the epilepsy in these children is clear in most of the cases. Four of the 7 children had seizure-free outcomes and 1 showed a reduction in seizure frequency of greater than 90%. The other 2 subjects were initially seizure-free but then showed seizure recurrence at about 50% the presurgical rate.

Eleven of the children who underwent surgical intervention had significant epileptiform activity on MEG examination, but no clear clinical seizures. In each case, the validity of the MEG observation was confirmed during intracranial monitoring with subdural grids. However, epileptologists frequently debate the clinical significance of "subclinical" epileptiform activity, and it is clear that there are some types of inter-ictal activity (eg, that seen in benign rolandic epilepsy) where aggressive treatment does not seem to be warranted. However, there is a considerable body of literature which indicates that nonrolandic inter-ictal activity during testing can have a significant negative impact on cognition, behavior, and social skills, and that treatment of this activity leads to improvement in these domains (Brinciotti et al, 1989; Matricardi et al, 1989; and Binnie, 1993).

The presented postsurgical outcome data in our report are fully admitted to be preliminary in nature and with some significant limitations. Parental report information is indeed limited and potentially biased, as Dr Kallen suggests, and we agree that data of this type must be viewed cautiously. However, this does not mean that such data should be completely disregarded and left unreported.

Regardless, we find it surprising that Dr Kallen criticizes our use of the CARS, an evaluation completed by a trained professional. The criticism stems from the fact that the CARS has not been validated for serial testing as a measure of progress after a treatment intervention, but no instrument has been fully validated for this important application. The CARS is a commonly used instrument for assessing autistic features, and it is known to have high interrater reliability when administered by appropriately trained personnel as was the case in our study. As with any scale, there is some difference expected between rater and between evaluation variability, but this is low for the CARS, on the order of 2 to 3 points. We did not present a statistical analysis in our report because our presentation of the surgical data was intended as a preliminary point of discussion, not as a core result, but it should be noted that, using a paired t-test, the postsurgical change is significant at P < .01.

Dr Kallen is correct in stating that the PPVT does not address social-affective disability in autism, but we never suggest that it does. The PPVT is a well-standardized test for evaluating basic receptive language skills; Dr Kallen's statement that the PPVT is intended for the age range 2.5 to 4.0 years is incorrect. The test has been normalized well into adulthood. For our reported cases, presurgical testing was performed no earlier than 4 months before surgery and postoperative testing was performed at the stated follow-up time. Dr Kallen correctly indicates that our data are complicated by the fact that some improvement in cognition and behavior is expected as part of the natural developmental process during the months between presurgical and postsurgical assessment, but an improvement on the PPVT of >2 years in a 9- to 15-month postsurgical period (as was seen in several cases) is beyond general developmental expectations and unlikely to be a placebo effect. We should point out, however, that PPVT is not a good measure of social-communicative functioning. The PPVT measures 1-word receptive language, not language production or utilizationthese later factors clearly being important aspects to consider in the evaluation of treatment efficacy. A method known as language sample analysis would have been most appropriate here, and it is something that we are now using in our investigations.

It is clear that, as we stated, a controlled, double-blind study is indeed needed to better evaluate the efficacy of surgery in the subgroup of autistic children that we describe. On the other hand, we would argue that one never gets to this stage without the type of preliminary observations that we presented. One purpose of a discussion section is to place experimental observations within a broader context, and in some situations this demands an explication of preliminary findings.

We, like Dr Kallen, hope that parents do not cite our findings as authoritative in pressuring pediatricians to support a direct referral for neurosurgery. However, we do hope that the parents of children with histories of autistic regression use our data to pry open the door to appropriate electrophysiological evaluation (EEG or MEG including stage III sleep) of their children. Whether or not one believes that epileptiform activity plays a causative or merely comorbid role for a subset of children with ASDs, it is appropriate to try and identify which children have such activity and to medically manage this activity in the same manner as would be applied to a child without autism.

Jeffrey D. Lewine
John T. Davis
Michael Funke
Greg Jones
Brian Chong
Sherri Provencal
William W. Orrison Jr
Department of Radiology
University of Utah
Salt Lake City, UT

Richard V. Andrews
Ra Neurological
Omaha, NE

Arun-Angelo Patil
Department of Neurosurgery
University of Nebraska
Omaha, NE

Michael Weisend
Roland R. Lee
Neuroradiology Section
Veterans Administration Medical Center
Albuquerque, NM

REFERENCES

1. Binnie CD Significance and management of transitory cognitive impairment due to subclinical EEG discharges in children. Brain Dev. 1993; 15:23-30 [CrossRef][Medline]
2. Brinciotti M, Matricardi M, Paolella A, Porro G, Benedetti P Neuropsychological correlates of subclinical paroxysmal EEG activity in children with epilepsy, 1: qualitative features (generalized and focal abnormalities). Functional Neurol. 1989; 4:235-239
3. Ebersole JS Noninvasive presurgical evaluation with EEG/MEG source analysis. Electroencephalogr Clin Neurophysiol Suppl. 1999; 50:167-174 [Medline]
4. Ebersole JS, Squires KC, Eliashev SD, Smith JR Applications of magnetic source imaging in evaluation of candidates for epilepsy surgery. Neuroimaging Clin North Am. 1995; 5:267-288
5. Gillberg C, Uverbrant P, Carlsson G, Hedstrom A, Silfvenius H Autism and epilepsy (and tuberous sclerosis?) in two pre-adolescent boys, neuropsychiatric aspects before and after epilepsy surgery. J Intellect Disabil Res. 1996; 40:75-81
6. Grote CL, Van Slyke P, Hoeppner JA Language outcome following multiple subpial transection for Landau Kleffner syndrome. Brain. 1999; 122:561-566 [Abstract/Free Full Text]
7. Matricardi M, Brinciotti M, Paolella A, Porro G, Benedetti P Neuropsychological correlates of subclinical paroxysmal EEG activity in children with epilepsy, 2: quantitative aspects. Functional Neurol. 1989; 4:241-246
8. Naas R, Gross A, Wisoff J, Devinsky O Outcome of multiple subpial transections for autistic epileptiform regression. Pediatr Neurol. 1999; 21:464-470 [CrossRef][Medline]
9. Neville BG, Harkness WF, Cross JH, Cass HC, Burch VC, Lees JA, Taylor DC Surgical treatment of severe autistic regression in childhood epilepsy. Pediatr Neurol. 1997; 16:137-140 [CrossRef][Medline]
10. Patill AA, Andrews RV Surgical treatment of autistic epileptiform regression. J Epilepsy. 1998; 11:368-373 [CrossRef]
11. Patil AA, Andrews RV, Torkelson R Surgical treatment of intractable seizures with multilobar or bihemispheric seizure foci (MLBHSF). Surg Neurol. 1997; 47:72-77 [CrossRef][Medline]

In Reply.

We have reviewed Dr Kallen's letter and have extensively discussed with our co-authors the issues raised in the letter. We agree with their reply to Dr Kallen's concerns over the diagnosis and selection criteria of our subject population and over the relationship between MEG and EEG data. On the other hand, we disagree with our co-authors' reply to Dr Kallen's concerns over the surgical data. Therefore, we present our views in a separate letter.

In our opinion, there are 2 important contributions of our manuscript. First, we demonstrated the advantages of MEG over EEG studies in the identification of epileptiform activity in children with autistic regression (AR). Second, we documented different location(s) of epileptogenic zones in LKS and in AR, as well as the different electrographic patterns of their recorded epileptiform activity. These differences, coupled with the very distinct clinical presentations, should help dispel the erroneous belief that AR is a variant of LKS or that autistic disorders may present as LKS.

Unfortunately, the inclusion of the surgical data in the discussion of our manuscript served as a distraction from the main thrust of our findings and in fact inadvertently may have communicated a sense of similarity, not dissimilarity, of AR and LKS. Of greater concern has been the feedback from colleagues that these surgical data may be interpreted by health professionals and parents of autistic children as a suggestion that MST is effective in children with AR. As stated in our co-authors' reply, the bulk of the surgical data, which were presented in Table 3, were added at the specific request of both reviewers. We also emphasize that such data are preliminary and we do not feel that the therapy effect was marked, or similar to results obtained in the treatment of LKS to date. We thought that readers would view these data in this light. Unfortunately, we were wrong in our assumptions. We agree that inclusion of the surgical data in our manuscript has obscured rather than illuminated the important issues.

In our opinion, the use of MST today should be restricted for patients with a classic form of LKS and for patients with intractable epilepsy originating in eloquent cortex. There are no data in the peer-reviewed literature indicating that MST yields significant and long-lasting benefits in children with AR. The largest series published to date is that of Nass et al.¹ from the New York University Epilepsy Center, with which one of us (O.D.) is affiliated. That article analyzed the therapeutic impact of MST in the cognitive and behavioral dysfunction associated with AR in 7 children, several of whom were operated on at other sites. A moderate improvement was noticed postsurgically, but in most cases it was only temporary. At the present time, we neither advocate nor use MST for the treatment of AR at our epilepsy centers.

We hope that readers will not be distracted by preliminary surgical data and focus on the results of our study, which we believe represent a valuable contribution to the literature of LKS and AR.

Andres M. Kanner
Michael C. Smith
Rush-Presbyterian-St. Luke's Medical Center
Chicago, IL

Orrin Devinsky
New York University Medical Center
New York, NY

REFERENCE

1. Nass R, Gross A, Wisoff J, Devinsky O Outcome of multiple subpial transection for autistic epileptiform regression. Pediat Neurol. 1999; 21:464-470