Secretin and autism: a two-part clinical investigation

Journal of Autism and Developmental Disorders, Vol. 30, No. 2, 2000 Secretin and Autism: A Two-Part Clinical Investigation
Michael G. Chez,1,2 Cathleen P. Buchanan,1 Bradley T. Bagan,1 Michael S. Hammer,1
Karla S. McCarthy,1 Irina Ovrutskaya,1 Caralynn V. Nowinski,1 and Zamia S. Cohen1
Recent anecdotal reports have touted the gastrointestinal (GI) hormone secretin as a treatmentmodality for autism, though there is little clinical evidence or literature to support its viability. Weundertook a two-part clinical trial to investigate these claims. Fifty-six patients (49 boys, 7 girls,mean age = 6.4 years, SD = 2.7) enrolled in an open-label trial of secretin, during which they re-ceived one injection of the hormone (2 IU/kg). All subjects were evaluated by their parents at base-line and follow-up visits (3–6 weeks later, M = 3.7, SD = 1.4 weeks) with Childhood Autism Rat-ing Scales (CARS). Thirty-four patients were labeled with Pervasive Developmental Disorder NotOtherwise Specified, and 22 met diagnostic criteria for Autistic Disorder. Forty-five patients wereconcurrently on other drug treatments. At follow-up, some reported minimal but potentially sig-nificant improvements including changes in GI symptoms, expressive and/or receptive languagefunction, and improved awareness and social interactions. No adverse effects were reported or ob-served. Subsequently, 17 of the most responsive patients from Study 1 began a double-blind trialthat also included 8 newly enrolled patients. Patients in this second study were alternatively en-tered into one of two groups and received injections of secretin or placebo with crossover at 4 weeks.
Patients from Study 1 entered into Study 2 at an average of 6.5 (SD = 0.8) weeks after begin-ning Study 1. Results of both inquiries indicate that although treatment with secretin wasreported to cause transient changes in speech and behavior in some children, overall it producedfew clinically meaningful changes when compared to children given placebo injections.
KEYWORDS: Secretin; autism.
INTRODUCTION
eral of the domains investigated. Moreover, the authorsfailed to include a control group, and the study did not Secretin, a gastrointestinal hormone traditionally present any quantifiable data concerning changes in be- used as a diagnostic tool in pancreatic function evalu- havior or symptoms. Nonetheless, the authors proffered ations, was recently reported to have a remarkable ef- the notion that secretin may play a neuroactive role and fect in reducing specific behavioral components of speculated that it may have a direct effect on the cen- autism (Horvath et al., 1998). However, that investi- tral nervous system, including the hippocampus, cere- gation was the only article published in a peer-reviewed bellum, and temporoparietal cortex, an hypothesis that scientific journal regarding the efficacy of this hor- has garnered much debate and controversy.
mone: It documented outcome in only three children Recent media attention sparked public interest in and was replete with missing or insufficient data in sev- secretin therapy; however, little is known at this timeto effectively render its anecdotal success a viable treat- 1 Pediatric Neurology, SC, Lake Forest Hospital, Lake Forest, Illi- ment for autism. A letter to the editor of Journal of Child and Adolescent Psychopharmacology by Perry and Ban- Address all correspondence to Michael G. Chez, Pediatric Neurol- garu (1998) reported their clinical experience with six ogy, SC, 800 Westmoreland Road, Suite 100, Lake Forest, Illinois60045; e-mail: mchezmd@interaccess.com children administered secretin at their practice. Re- 0162-3257/00/0400-0087$18.00/0 2000 Plenum Publishing Corporation Chez et al.
garding changes in autistic symptoms, the authors stated jects (34%). This percentage is slightly higher than that that only one of the six children showed “clinically sig- previously published (Volkmar & Nelson, 1990) yet nificant” improvements in language and relatedness as represents the subpopulation of patients we typically reported by mother and therapist. Three others showed evaluate (Chez & Buchanan, 1996, 1997; Chez, “subtle changes” in the areas of language and related- Buchanan, Zucker, & May, 1997; Chez, Buchanan, ness. Another became more active and aggressive, while Field-Chez, Loeffel, & Hammer, 1998).
another showed no improvement. In sum, Perry and History of treatment with drug therapies used for Bangaru cautioned that the effectiveness of secretin is abnormal EEG was noted in 37 patients. Such treat- “much more modest” than both the popular press and ment protocols (e.g., Stefanatos, Grover, & Gellar, 1995) included past or concurrent use of valproic acid, The present study was exploratory in scope and valproic acid plus steroids, or steroids alone. Nine pa- sought to objectively assess clinical efficacy following tients also had a history of treatment with antidepres- injection of secretin via alterations in the behavioral sants (i.e., 5-HT reuptake inhibitors; e.g., Chugani et al., characteristics of autism in children who presented at 1999), and 9 patients were concurrently being treated our clinic with prior diagnoses falling along the con- with CNS stimulants (i.e., methylphenidate) to address tinuum of autistic spectrum disorders (ASD; Happé and long-standing hyperactivity. Changes in concurrent Frith, 1991). Prior investigations have not thoroughly medications were not allowed during secretin admin- examined the nature or degree of potential response to istration to ensure that potential clinical improvements the gut hormone. We hypothesized that if changes in were not due to alternative therapies. All parents com- autistic symptomatology occurred in the manner pre- pleted informed written consent, enabling their child to viously described qualitatively by Horvath et al. (1998), improvements should be easily detected in our samplevia a standardized autism rating scale.
Childhood Autism Rating Scales (CARS; Schopler Reichler, & Renner, 1988) were completed by parentsat baseline and postinjection office visits (follow-up duration of 3–6 weeks, M = 3.7 weeks, SD = 1.4). Ac- Fifty-six children (49 boys, 7 girls, mean age = cording to its authors, the CARS was designed as a rat- 6.4 years, SD = 2.7) were enrolled in the study. All pa- ing system for identification of behavioral symptoms, tients were self-referred and met diagnostic criteria for for research purposes, and for classification purposes PDDNOS (34 patients—299.80) or Autistic Disorder by replacing subjective clinical judgments with objec- (22 patients—299.0) (DSM-IV; American Psychiatric tive and quantifiable ratings based on direct behavioral Association, 1994). All patients were previously as- observation. It is divided into 15 subcategories, each sessed by one of several local licensed clinical psy- scored by increments of 0.5 points on a scale of 1 through chologists and diagnosis was confirmed by the primary 4 which are cumulatively added resulting in a total score author at intake. Through our clinical experience we be- ranging from 15 to 60. The rating scale for total scores lieve that our sample represents a diverse cross-section is then divided into three classifications: Nonautistic of patients typically seen by our pediatric neurology (15–30), Mildly/Moderately autistic (30.5–36), and Se- practice, including a wide range of age, symptoms, and verely Autistic (36.5–60). Therefore, a lower total score behavioral presentations. Table I shows the descriptive and demographic information for all patients who par- Although the scale was intended to be completed by a clinician and parent in an interview format, we A history of chronic gastrointestinal (GI) distress adapted its format for use as a blind rating scale for each (i.e., constipation, loose stools, diarrhea, or vomiting) of the baseline and follow-up ratings. We asked that the concurrent with or preceding the onset of autistic fea- same parent complete the scale each time for the sake tures occurred in 33 children (59%), whereas normal of consistency and homogeneity in assessment and that GI history was seen in 23 subjects (41%). Twenty-four they fill it out with regards to the best behavior wit- hour EEG findings were available for 49 subjects, re- nessed during the time between office visits.
vealing abnormal EEG with primarily intermittent We designated a 6-point improvement in ratings sleep-activated centrotemporal sharp waves or spike as indicative of a “clinically significant” change be- waves in 31 subjects (63%) and normal EEG in 18 sub- cause, by virtue of the instrument’s design, a change Secretin and Autism
Table I. Description of 56 Patients in Study Ia
Chez et al.
Table I. (Continued)
a GI Hx indicates a history remarkable for GI problems. F/U represents the time (weeks) between the initial secretin injection and the follow- up office visit. Bold represents patients whose baseline CARS scores fell in the Severely Autistic range. Italics represent 13 patients whose
CARS scores changed by 6 points or greater, representing a clinically significant change.
b 17 patients who also participated in Study 2.
c 3 patients whose scores met the ≥ 6 point change, but who did not originally fall into the Severely Autistic range.
of 6 points or more reflects a change in diagnostic classifications in children who fall in the midrange The CARS total scores for the entire group “moderately autistic” category. Such a change may showed a statistically significant improvement from be considered a liberal definition of improvement baseline to follow-up (Baseline M = 36.5, SD = 6.4; since it represents only a 10–15% improvement over Follow-up M = 33.2, SD = 5.4; t (55) = 5.67, p <. 01).
Further inspection of the individual CARS categories Physicians or nurses completed detailed clinical showed that statistically significant improvements (a assessments of GI function, eye contact, expressive conservative alpha adjustment of p < .01 for multiple speech, and receptive language using a standardized comparisons) occurred specifically in Relating to People, questionnaire previously implemented by our practice Imitation, Emotional Response, Object Use, Adapta- for the purpose of tracking patients over time in inves- tion to Change, Visual Response, Listening Response, tigational drug studies. Finally, parents noted changes Taste/Touch/Smell, Activity Level, and Verbal Com- in eye contact, interpersonal relations, self-stimulatory munication. Areas that were not significantly different behavior, bowel movement consistency, toy play, ex- from baseline to follow-up included Level of Intel- pressive language/verbalizations, receptive language, lect, Fear or Nervousness, Body Use, and Nonverbal and overall behavior with observational diaries.
However, when improvement is viewed in terms of clinical significance, which we designated as achange of 6 points or greater, frequency distributions In accord with the dosage reported by Horvath among children take on a different meaning. Figure 1 et al. (1998), all children received 2 IU of secretin (Fer- shows each child’s raw score changes from baseline ring Laboratories, Inc.) per kilogram of body weight to follow-up on the CARS. It is important to note that intravenously over a period of 1 to 2 minutes. All chil- 10 of the 13 patients with changes of 6 points or dren were observed for 1 hour post injection. No al- greater originally were classified in the Severely Autis- lergic reactions occurred. Parents of patients were tic category. Moreover, it is apparent that while 13 billed during this study for the cost of secretin, since children had clinically meaningful changes, the ma- experimental investigations of such a nature are not jority (n = 32) had changes in the 1.0- to 5.5-point range. Two children showed no change, and 9 chil-dren had profiles rated as “worse” (range: -0.5 to -11).
Parental diaries of children who were rated as “no Data were analyzed using the SPSS Statistical change” or “worse” were notable for comments re- Package version 6.1. All t tests were two-tailed and garding increased hyperactivity in the days immedi- were run as either independent samples when groups ately following the injections. These were not children were compared or pairwise for within-subjects com- who had originally entered the study on methylphenidate.
parisons. When indicated, a conservative alpha was Other adverse effects noted by parents generally re- adopted to control for the inflated effects of multiple lated to increased agitation, decreased focusing, and diminished responsiveness to others.
Secretin and Autism
Fig. 1. This graph illustrates the change in CARS scores from baseline (M = 36.5, SD = 6.4) to follow-up (M = 33.2. SD = 5.4) in Study 1. Aver-
age duration of follow-up = 3.7 weeks (SD = 1.4). Positive numbers represent raw score improvements; negative scores represent decline or
worsening of autistic symptoms. Two patients showed no change (indicated by empty bars). Only 13 of 56 patients demonstrated changes suf-
ficient (≥ 6 points) to change diagnostic categories.
In addition, other parents reported minor im- were rated as the most severe at the start of the study.
provements in GI function and eye contact in 19 cases Certainly, we did not observe any instances of dramatic (34%), improvements in expressive speech in 30 cases improvement with regard to explosions of speech or (54%), and receptive speech improvements in 15 cases other autistic characteristics as previously highlighted (27%). GI improvements noted were “more consis- in the popular press and media. Although the change tent,” “more formed,” “less constipated,” and “firmer from an average baseline CARS score of 36.5 (SD = stools.” Gains in expressive speech included use of 6.4) to a follow-up score of 33.2 (SD = 5.4) was sta- more words, more complete sentences, and more spon- tistically significant, this difference did not meet cri- taneous speech, while receptive speech improvements teria of ³6 point improvement necessary to conclude included improved ability to follow complex com- that significant changes occurred in these patients.
mands and greater responsiveness. The improvements When statistically broken down by domain, no sin- noted in all of these categories were transient, and gle outcome category dramatically changed at the time lasted approximately 3 to 7 days. GI function and eye of retest, suggesting that the significant overall change contact were reportedly worse in one case, and ex- was an artifact of more subtle trends across categories.
pressive speech was worse in a separate case. Ac- To further investigate the potential benefits of secretin cording to parental report, the remaining children did and to examine whether the statistically significant im- not display any changes in GI function, eye contact, provements noted above were due to rater bias or true expressive speech, or receptive speech. The clinical drug effect, we selected a subgroup of children who had interview and parental diaries were also noteworthy perceived improvements in Study 1. This subject selec- for “improved awareness,” “improved social inter- tion was intended to maximize the probability of de- action,” “more affectionate,” “better play/activity level,” tecting a difference in a placebo-controlled crossover study. We predicted that second injections would mag-nify and enhance the subtle changes observed from thefirst injections, if the hypothesis of Horvath et al. (1998) Discussion
The results from Study 1 indicate that if improve- ments are observed in autistic features when patients are treated with secretin, they are most likely to be seen when a child falls in the more severe range at baseline,a phenomenon which may also be viewed as regression Twenty-five children (22 boys, 3 girls) with an to the mean (Maris, 1998). We make this conclusion average age of 6.0 years (SD = 2.4) enrolled in the based upon the finding that 77% of our children who double-blind crossover clinical trial. Some of these pa- showed any improvement on the CARS were those who tients completed the protocol from Study 1 and chose Chez et al.
to continue with the second phase of the investigation Table II. Study 2: Group Averages for Childhood Autism Rating
due to perceived benefit. These “responders” from Study 1 were classified by CARS scores and parental reports. These subjects represent a biased sample of pa-tients from Study 1. They began Study 2 an average of 6.5 (SD = 0.8) weeks after beginning Study 1.
Total group
Thirteen of these patients had an unremarkable GI history. The remaining patients noted some history of (secretin,
GI problems, the most prevalent being diarrhea and placebo)
loose stools (9/25). Ten patients (40%) had abnormal EEGs in the past. Expressive language, receptive lan- (placebo,
secretin)
guage, and eye contact were evaluated on an individ-ual basis for improvement because of the wide varia- a Average CARS scores for all patients in Study 2. Note the minimal difference in mean scores following either placebo or secretin.
b p £ .05, baseline vs. secretin.
Patients were selected to enter into either Group A from the table, the average CARS ratings for groups or Group B based on an alternating schedule at entry.
were remarkably similar for children from Baseline to Group A (n = 13) received an intravenous injection of Follow-up, regardless of the injection type. There were secretin (2 IU/kg) at the baseline visit, and an injection not any statistically significant overall differences be- of placebo (normal saline) at the 4-week follow-up.
tween groups at 4 weeks, t (24) = -0.24, p = .81, or 8 Group B (n = 12) patients received the injections in the weeks, t (24) = 1.17, p = .25. Similarly, only about half reverse order. One child was lost to follow-up from Group of the parents of children enrolled in this study were B and did not return to the office for the second in the able to correctly identify the order of the injections due series of injections. During this phase of the investiga- to their observations of the patients’ behavior (12/25), tion, parents were billed a one-time fee which covered indicating chance responding. Patients in both groups the cost of the secretin and investigators’ time for both did not experience any medically or neurologically ad- arms of the crossover trial. The group designation was verse effects after either injection. One child was noted unknown to both physicians and patients, as the encoded to become flushed a few minutes after injection, which patient records were kept in a database to which only the registered nurse (K.M.) and technician (B.B.) had access.
Group A (secretin, placebo) did not show any sta- tistically significant group differences in CARS scores following secretin injection (Baseline to 4 weeks), t (12) Detailed neurologic interviews, clinical assess- = -0.59, p = .57. Additionally, parental diaries were ments, and CARS forms were completed at the baseline, notable for adverse effects such as decreased attention 4-week, and 8-week visits. Physical examination looked span, loss of social skills, and increased aggression.
for any new deficits or gains in neurological function After the saline injection, several children continued to that could have resulted from secretin administration.
evidence slight worsening of symptoms, although in- Interim changes in eye contact, speech, and language dividual observations were not statistically significant, function were reviewed, along with investigator-patient t (12) = -0.31, p = .76, when viewed as group means.
interaction. Parents were asked to fill out the CARS Clinical interviews for Group A revealed 5 children scales based on their child’s best behavior during the with improvements in expressive language and with im- time between visits. They were later asked to guess the proved GI function after the secretin injection. There order of the injections based on observations of their were not any reports of improved eye contact or re- child’s behavior before the code was broken at the final ceptive language function. Following the placebo in- office visit (8-week follow-up exam).
jection in Group A, there were three reports of improvedexpressive language and one report of improved GIfunction. Two parents noted improved eye contact, and three noted improved receptive language.
Table II illustrates the average CARS scores for Several parents of children in Group B noted slight the entire cohort and Groups A and B. As may be seen improvement in their diary notes following their first Secretin and Autism
injection (saline), although this did not reach statisti- blinded Study 2 did not substantiate even mild im- cal significance, [t (11) = 0.59, p = .57] as a group for provements in children. In summary, the behavioral pro- the CARS. When administered secretin at the second file of children following secretin injection at the dosage injection, however, children were perceived by parents recommended by Horvath et al., (1998) strongly re- as improved, [t (11) = 2.39, p = .04] overall.
sembles that of a child following saline administration.
Following the secretin injection, parents of chil- Furthermore, in the quest to identify a potential sub- dren in Group B noted improvements in expressive lan- class of responder, we were unable to find any associa- guage (n = 7), GI function (n = 4), eye contact (n = 2), tions between prior and current medications, age of and receptive language (n = 3). Subsequent to the onset, gender, GI symptoms, or other past medical placebo in Group B, parents reported 2 improvements in history to explain the differences between those who expressive language, 1 improvement in GI function, and showed optimal response and those who failed to bene- 1 improvement in eye contact. None of the parents noted fit from the injections. Also, changes were not attributed any changes in the receptive language category.
to alteration of other medication regimens since treat-ments were held constant over the course of the entireinvestigation.
Discussion
Clearly, CARS rating scales are a crude form of The results from Study 2 do not demonstrate any evaluation. It was not designed to be a sensitive mea- significant difference in CARS scores in patients treated sure to detect subtle changes in autistic behavior, so we with secretin versus patients treated with saline, though are cautiously basing our conclusions on statistical eval- the majority of the cohort for this phase of the trial was uation of CARS scores. In order to detect those subtle purposefully selected through biased methods. For the changes, we need to develop more sensitive methods of most part, only those parents who perceived benefit after evaluating autistic spectrum disorders. Future research Study 1 chose to continue with the protocol. The only way to eliminate bias in their evaluation of secretin was Statistical evaluation of the data from CARS may to blind everyone involved with regard to the treat- not fairly substitute information garnered from parental ment. Yet the results show that the potential benefits diaries. We cannot rule out the fact that some parents reported after an initial injection of secretin mimic re- were able to discriminate between the two injections, or sults from a saline injection, which strongly suggests a the parental reports of language, behavioral, and social lack of clinical improvement provided by secretin. The changes. Nonetheless, parents’ qualitative reports of placebo effect, in traditional literature, has been reported improvement were not consistently observed by the in- to yield improvements in approximately 35% of the vestigators during the follow-up office exams, nor were cases (Kienle & Kiene, 1996). Since parents reported they reflected in the quantitative rating scale.
the perceived changes instead of the subjects who ac- At this time there are not enough data to suggest tually received the injections, the placebo effect in this any substantial benefit from secretin injections. Before case may be better explained as an expectancy effect.
further time and expense are incurred, physiological Expectancy theory maintains that conditioning trials mechanisms and neural pathways involved in the pur- produce placebo response expectancies, rather than ac- ported neuroactive response to secretin should be the tual placebo responses, and that the expectancies elicit subject of future inquiry. Such considerations are es- the responses (Montgomery & Kirsch, 1997).
pecially relevant considering the tremendous hetero-geneity of the autistic spectrum disorders. Our experi-ence has led us to conclude that additional research is GENERAL DISCUSSION
needed to identify subtypes of children who may re- The results of this two-part clinical investigation spond to therapies aimed at targeting the gut–brain axis.
of secretin demonstrate that overt behavioral changes There exists the potential to isolate a biochemical or do not occur following the hormone injection in chil- behavioral “clinical marker” that could help identify dren who manifest symptoms of varying autistic sever- those children who will benefit from this and other ther- ity at baseline. Our data do not provide any evidence of apies. Objective assessment of children is also war- a treatment effect and are strengthened by the study de- ranted to counter the expectancies that parents develop sign which allowed each child to serve as his or her own with any potential “cure” for their child’s disorder. Ad- control in the crossover phase. The transient improve- ditionally, contrasts of children undergoing efficacious ments in behavioral symptoms noted by parents in Study multiple drug regimens (e.g., combination valproic 1 may actually be the result of a reporting bias since the acid and steroid therapy) in combination with secretin Chez et al.
versus those maintained on secretin-only treatment Chez, M. G., Loeffel, M., & Buchanan, C. (1998). Pulse high-dose could serve as the focus of future study (Bardenstein steroids as combination therapy with valproic acid in epilepticaphasia patients with pervasive developmental delay or autism.
et al., 1998; Chez & Buchanan, 1996; Chez, Buchanan, Annals of Neurology, 44, 539.
Field-Chez, et al., 1998; Chez, Buchanan, Loeffel, et Chugani, D. C., Muzik, O., Behen, M., Rothermel, R., Janisse, J. J., al., 1998; Chez, Loeffel & Buchanan, 1998; Piloplys, Lee, J., & Chugani, H. T. (1999). Developmental changes inbrain serotonin synthesis capacity in autistic and nonautistic 1994; Stefanatos et al., 1995; Trauner et al., 1997).
children. Annals of Neurology, 45, 287–295.
Happé, F., & Frith, U. (1991). Is Autism a pervasive developmental disorder? How useful is the “PDD” label. Journal of Child Psy- REFERENCES
chology and Psychiatry, 32, 1167–1168.
Horvath, K., Stefanatos, G., Sokolski, K. N., Watchel, R., Nabors, American Psychiatric Association. (1994). Diagnostic and statistical L., & Tildon, J. T. (1998). Improved social and language skills manual of mental disorders (4th ed.) Washington, DC: Author.
after secretin administration in patients with autistic spectrum Bardenstein, R., Chez, M. G., Helfand, B. T., Buchanan, C. P., & disorders. Journal of the Association for Academic Minority Zucker, M. (1998). Improvement in EEG and clinical function in pervasive developmental delay (PDD): Effect of valproic Kienle, G. S., & Keine, H. (1996). Placebo effect and placebo con- cept: a critical methodological and conceptual analysis of reports Chez, M. G., & Buchanan, C. P. (1996). Clinical spectrum of patients on the magnitude of the placebo effect. Alternative Therapies referred for Landau-Kleffner syndrome previously diagnosed in Health Medicine, 2 (6, 39–54) with pervasive developmental delay or autism: EEG, HMPAO Maris, E. (1998). Covariance adjustment vs. gain scores-revisited.
SPECT, and steroid response data. Neurology, 46, A115.
Psychological Methods, 3, 309–327.
Chez, M. G., & Buchanan, C. P. (1997). Age of regression of lan- Montgomery, G. H., & Kirsch, I. (1997). Classical conditioning and guage and behavior in children referred for pervasive develop- the placebo effect. Pain, 72, 107–113.
mental delay correlates with severity of epileptiform activity on Perry, R., & Bangaru, B. S. (1998). Secretin and autism. Journal of 24-hour EEG. Epilepsia, 38, 50.
Child and Adolescent Psychopharmacology, 8, 247–248.
Chez, M., Buchanan, C., Zucker, M., & May, B. (1997). Value of 24 Plioplys, A. V. (1994). Autism: Electroencephalogram abnormalities hour EEG versus routine EEG in detecting occult epileptiform and clinical improvement with valproic acid. Archives of Pedi- activity in children with pervasive developmental delay. Annals atric and Adolescent Medicine, 148, 220–222.
Schopler, E., Reichler, R. J., & Renner, B. R. (1988). The Childhood Chez, M. G., Buchanan, C. P., Field-Chez, M., Loeffel, M. F., & Autism Rating Scale: CARS. Los Angeles, CA: Western Psy- Hammer, M. S. (1998). Treatment of electroencephalographic epileptiform activity on overnight EEG studies in children with Stefanatos, G. A., Grover, W., & Geller, E. (1995). Case study: Cor- pervasive developmental disorder or autism: Defining similari- ticosteroid treatment of language regression in pervasive de- ties to the Landau-Kleffner syndrome. Journal of Developmen- velopmental disorder. Journal of the American Academy of Child tal and Learning Disorders, 2, 217–229.
and Adolescent Psychiatry, 34, 1107–1111.
Chez, M., Buchanan, C., Loeffel, M., Field-Chez, M., Nowinski, Trauner, D. A., Nabangchng, C., Ballantyne, A., & Tecoma, E. (1997).
C., Bardenstein, R., Hammer, M. (1998). Practical treatment Developmental aphasia with epileptiform abnormalities on EEG: with pulse dose corticosteroids in pervasive developmental Clinical features and response to prednisone. Annals of Neurol- delay or autistic patients with abnormal epileptiform sleep EEG and language delay. In M. V. Perat (Ed.), New develop- Volkmar, F. R., & Nelson, D. S. (1990), Seizure disorders in autism.
ments in child neorology (pp. 695–698). Bologna, Italy: Mon- Journal of the American Academy of Child and Adolescent Psy-

Source: http://www.svsu.edu/~gmlange/461Chapter10article.pdf