Absence of Magnetic Resonance Imaging Evidence of Pontine Abnormality in Infantile Autism Melissa Hsu,
MS; Rachel Yeung-Courchesne; Eric Courchesne, PhD; Gary A. Press, MD
vivo studies involving magnetic resoimaging and studies of neuropathologic specimens have shown that autism is most consistently associated with developmental hypoplasia of the neocerebellum. We investigated whether the cerebellar hypoplasia was accompanied by gross structural abnormalities in the major input (cerebrocerebellar) and output (cerebrorubral) pathways to the cerebellum by measuring the area of the ventral pons (including the pontine nuclei and the transverse fibers) and the midbrain on midsagittal magnetic resonance images in 34 autistic and 44 subjects. The area of the entire pons and several regions of interest within the midbrain (including the superior and inferior colliculi) were also determined with midsagittal magnetic resonance images. We found no significant dif\s=b\ In
nance
ference between measurements of the pons and midbrain in autistic and control subjects. Our data show no evidence of gross anatomic abnormalities in the input and output pathways to the cerebellum in autism, a finding that is consistent with previous studies of
neuropathologic specimens; rather, the reduced size of the neocerebellum in autism appears to be the result of maldevelopment within the cerebellum itself. (Arch Neurol. 1991 ;48:1160-1163) Accepted for publication March 28,1991. From the Departments of Neurosciences (Ms Hsu and Dr Courchesne) and Radiology (Dr Press), School of Medicine, University of California, San Diego, La Jolla, and the Neuropsychology Research Laboratory, Children's Hospital Research Center, San Diego, Calif (Ms Yeung-Courchesne and Dr Courchesne).
to the Neuropsychology ReChildren's Hospital Research Center, 8001 Frost St, San Diego, CA 92123 (Dr
Reprint requests
search
Laboratory,
Courchesne).
studies have reported neuroanatomic abnormalities in many ' patients with autism. In a recent mag¬ netic resonance (MR) imaging study of 18 autistic subjects, developmental hy¬ poplasia of the cerebellum was identi¬ fied in three subjects by routine qualita¬ tive neuroradiologic reading.24 Quantitative analysis of the image data revealed that the neocerebellar vermis (lobules VI and VII) was reduced in size compared with that of controls.3 A sec¬ ond MR imaging study of 10 of these 18 autistic subjects showed that the cere¬ bellar hemispheres were also reduced in size in correlation with the degree of size reduction of the vermis.4 Older studies involving computed tomogra¬ phy have also reported cerebellar hypo¬ plasia in three severely retarded autis¬ tic children with purine metabolism disorders,' and cerebellar atrophy as well as increased width in the upper and middle portions of the fourth ventricle have been detected in an unspecified number of autistic children.1' These lat¬ ter findings suggest that alteration in the cerebellum and the pons, which form the boundaries of the fourth ven¬ tricle, may contribute to its enlarge¬ ment in autistic subjects.' Subsequent studies, however, have reported no ventricular size differences between au¬ tistic and control subject groups."'" Gaff-
Tmaging
al,1" in a study involving 13 highfunctioning autistic subjects, reported ney et
that the entire pons was smaller in the autistic group. In pathologic studies of brain speci¬ mens of autistic patients, abnormalities have been found in the neocerebellum in all cases studied.11"1'' (In Williams et al,15
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
marked
Purkinje
neuron
loss
was re¬
ported in one ofthe four cases studied; of the three patients for whom no cere¬ bellar abnormality was reported, one showed evidence of phenylketonuria,
another had suffered a cerebral concus¬ sion, and the third had some clinical fea¬ tures suggestive of Rett syndrome.) In most,12 but not all,15 cases studied, ab¬ normalities have also been found in por¬ tions of the limbic system. Atrophy of the neocerebellar cortex was accompa¬ nied by decreased numbers of Purkinje cells in 10 autistic patients11"1 "; there was also a decreased number of granule cells in five of these cases.11"13 In two brains that have been studied in greatest de¬ tail, a lack of retrograde cell loss in that part of the inferior olive related to the abnormal cerebellar cortex suggests that Purkinje cell and granule cell loss probably occurred prenatally, before projections from the inferior olive to the Purkinje cells were firmly estab¬ lished.1113 Although the olivary neurons were pale and small, they were normal in number. Additional abnormalities in¬ cluded widening of the fourth ventricle and thinning and elongation ofthe supe¬ rior cerebellar peduncles. In the most profoundly affected specimen, neuronal clusters were found at the edge of the olivary convolutions; moreover, dense¬
ly packed, enlarged neurons were pre¬ sent centrally within the base of the the pons. In other specimens, however, "12 pons appeared to be normal. In the present investigation, we test¬ ed the hypothesis that hypoplasia ofthe
cerebellum in autism occurs without gross structural abnormalities of its ma¬ jor input (cerebrocerebellar) and output
(cerebrorubral) pathways. Because pontine nuclei receive massive input from the frontal cortex and give rise to mossy fiber projections to the cerebel¬ lum (Fig 1), we measured the midsagit¬ tal area of the ventral pons (including the pontine nuclei and the transverse
fibers) in 34 autistic and 44 control sub¬
jects to determine whether the neocerebellar hypoplasia seen in autism is asso¬ ciated with pontine abnormalities. We
also measured the area of the entire pons and regions of interest within the midbrain. SUBJECTS AND METHODS
Subjects
Thirty-four autistic (mean age, 18.9 years; range, 2 to 39 years) and 44 control (mean age, 19.8 years; range, 3 to 39 years) subjects participated in the study, including the 18 autistic and 12 control subjects examined in our earlier studies of cerebellar structures."' The autistic group consisted of 27 male and seven female subjects; the control group con¬ sisted of 40 male and four female subjects. The autistic subjects included 11 mentally retarded subjects. Thirty-five control sub¬ jects were normal, healthy volunteers; the remaining nine were patients referred to the University of California, San Diego, Magnet¬ ic Resonance Institute for evaluation of head and neck pain, headache, depression, and neurogenic bladder, among other symptoms, but who exhibited no radiologie abnormali¬ ties. Wechsler Adult Intelligence Scale-Re¬ vised or Wechsler Intelligence Scale for Chil¬ dren-Revised IQ scores were available for 19 autistic and 19 control subjects. The autistic group had a mean ( ± SD) full-scale IQ score of 82.6 ±15.5, a mean verbal IQ of 78.7±18.1, and a mean performance IQ of 89 ± 13.7. The control group had a mean fullscale IQ score of 113 ± 11.4, a verbal IQ of 112 ±13.2, and a performance IQ of 112 ±10.3. Protocol for MR
Scanning
After appropriate informed consent was obtained, MR imaging scans were obtained with a 1.5-T magnet (GE Signa, General Electric, Milwaukee, Wis). The data matrix was 256 256 at a field of view of 16 or 24 cm, yielding a pixel size of 0.63x0.63 mm2 or 0.94 0.94 mm2, respectively. A 1- to 2-minute scout sequence (repetition time 400 ms; echo time, 20 ms; number of signals aver¬ aged 1) was performed in the axial and sag¬ ittal planes to verify precise subject position¬ ing before the acquisition of high-resolution =
=
sagittal images. Subsequently, a ,-weight¬ ed (repetition time 600 ms; echo time, 25 ms; number of signals averaged 1 or 2) se¬ quence centered precisely at the midline was used to acquire 5-mm-thick sections with 2.5-mm gaps between adjacent sections. =
=
Quantification of Pons and Midbrain Areas
Areas were quantified with computer-as¬ sisted planimetrie analysis ofthe median sag¬ ittal MR images described previously."'4 The area ofthe ventral pons included the pontine nuclei and the transverse fibers. This area is
Fig 1. —Schematic diagram of the cerebellar vermis and related cerebellar nuclei and brain-stem structures of the human brain. 1 indicates cerebropontocerebellar afférents that terminate in the neocerebellar vermis and hemispheres; 2, cerebellar efferente from the Purkinje cells via the deep nuclei to the brain stem and thalamic systems; 3, the red nucleus; 4, the cerebellar nuclei; 5. the nucleus reticularis tegmenti pontis; l-V, vermis lobule l-V; VI-VII, vermis lobules VI-VII; and VIII, vermis lobule VIII (adapted from Nieuwenhuys et al'6).
Fig 2. —Midsagittal ,-weighted images showing the regions of interest measured. VP indicates ventral pons; PONS, entire pons; MB, midbrain; and mb, midbrain excluding the superior and inferior colliculi. bound
by the
ventral border of the medial
lemniscus and the ventral border ofthe pons (Fig 2). We also measured the area of the entire pons and the midbrain, as described by
et al.1" The area ofthe entire pons included the region from the caudal boundary of the midbrain (defined by a line from the superior pontine notch to the junction be¬ tween the tectum and the superior medullary velum) to a line drawn from the inferior pon¬ tine notch to the fourth ventricle (Fig 2). The midbrain area included the tectum, the teg¬ mentum, and the crus cerebri. It was thus bound by the pons caudally (as defined above), the tectal laminar cistern dorsally, the interpeduncular cistern ventrally, and the diencephalon (including the mammillary bodies, hypothalamus, thalamus, and sub¬ thalamus) rostrally (Fig 2). According to Nieuwenhuys et al,"' we defined the posteri¬ or margin of the hypothalamus as "a vertical plane passing just caudal to the mammillary bodies" (Fig 2). Because this measurement of the midbrain might be influenced by the size of the cerebral aqueduct and the sizes of the colliculi, we also measured midbrain areas limiteli dorsally by the floor of the cerebral aqueduct, thereby excluding the colliculi and cerebral aqueduct. Each image was inter¬ preted and measurements were performed
Gaffney
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
by two ofthe investigators (G.A.P. ), one of whom (M.H. or R.Y.-C.) was "blind" to the diagnosis. The measurements were aver¬ aged to obtain mean values for each subject. from which group means were calculated. Measures of brain areas were analyzed with one-way analysis of variance for group ef¬ fects; age effects were determined with re¬ gression analysis.
RESULTS
In all 78
scans
for which
measure¬
performed, there was no visually detectable abnormality of the pons or the midbrain (Fig 3), even in autistic subjects with greatly reduced cerebellar size (eg, 2 to 5 SDs below the ments were
for controls). Measurements ofthe ventral pons re¬ vealed no statistical difference between the autistic group (mean, 350±44.5 mm2) and the control group (mean, 348+41.6 mm2) (F[l,61] 0.147, P>.7; Figs 4 and 5). There was also no signifi¬ cant difference between the area of the entire pons in the autistic group (mean, 540 ±57.6 mm2) and that in the control 5i (mean, group mean
=
500
500 D
400
>
CDcB
D
300
400
200
10
20
30
40
0-
300
>
200
0
10
Age, y 500
400
400
%
£
200
—
0
10
300
-1-1-1
20
Fig 3. —Magnetic resonance imaging scans of the brain stem and vermis of three patients with autism (right) and three age-matched controls (left). The top pair of scans are from children (control, aged 5 years; autistic, aged 5 years); middle pair, from adolescents (control, aged 13 years; autistic, aged 14 years); and bottom pair, from adults (control, aged 23 years; autis¬ tic, aged 22 years). The midline sagittal views show no qualitative evidence of brain-stem atrophy or other morphologic abnormality in the autistic patients compared with the controls in these pairs.
(F[l,ül | 1.07, P>.3; Figs 4 and 5). In the midbrain areas, there was no signifi¬ cant difference between the control group (mean, 240 ±29.9 mm2) and the autistic group (mean, 230±27.1 mm2) (F[l,561 0.526, P>A). There was also no significant difference between mid¬ brain areas that excluded the colliculi and cerebral aqueduct (177 ±25.2 mm2 for the control group and 170 ± 21.5 mm2 for the autistic group) (F[l,641 1.369, P>.\). Interrater reliability was corre¬ lated at .864 for measurements of the ventral pons and .83 for measurements of the entire pons. With increasing age in autistic and control subjects, there was a significant increase in the area measurements of the ventral pons and the entire pons (autistic group: F[l,30]>8.0, P9.3, P
1
10
Age, y Fig 4. —Individual (right) subjects.
30
Age, y
500 -
300
20
20
30
40
Age, y
measurements of the ventral pons and the entire pons in control
(left) and autistic
either the pons or the midbrain in mid¬ sagittal MR imaging scans of autisticsubjects. The absence of a significant difference in the area measurements of the ventral pons, entire pons, and mid¬ brain between autistic and control sub¬ jects argues against the theory that ab¬ normalities in the major input
(corticocerebellar) or output (cerebellorubral) pathways to the cerebellum
responsible for the neocerebellar hypoplasia of autism. are
Our results
are
consistent with
neu¬
ropathologic findings in three autistic patients in whom no midbrain or pontine
abnormalities were detected. The ab¬ sence of retrograde cell loss in the brain stem in these autopsy cases provides evidence that the reduction in Purkinje and granule cell numbers (which likely accounts for the decreased size of the neocerebellum) is due to prenatal maldevelopment within the cerebellum it¬ self, rather than to abnormalities in¬ volving the input/output pathways of the cerebellum. Our MR imaging findings and the re¬ sults of neuropathologic studies are also compatible with results from electro¬ physiologic studies of autism. Although early studies of autism reported abnor¬ mal brain stem —evoked responses,'''" more recent studies report no abnor¬ malities."'2" The recent electrophysio¬ logic studies were carefully constructed and specifically addressed important "'2
méthodologie considerations, including subject homogeneity, cooperation, and body temperature and gender matching of autistic and control subjects. Because
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
Ventral Pons
Fig 5.—Quantitative
Entire Pons
measurements
(mean-SD)
of the ventral pons and entire pons in autistic (solid bars) and control (open bars) subjects. Our data revealed no statistically significant differences between the group measurements.
there is no abnormality in the auditory brain stem-evoked responses, we believe that abnormalities of later com¬ ponents of the auditory-evoked re¬ sponse21"21 cannot be the consequence of abnormalities occurring early in the au¬ ditory pathways. This argument sug¬ gests that autism is unlikely to be asso¬ ciated with deranged auditory brain¬ stem pathways responsible for evoked responses. The involvement of other systems within the brain stem, howev¬ er, cannot be excluded as yet. Our results do not replicate the find¬ ings of Gaffney et al. '" In their study, the area of the entire pons in 13 autistic patients (mean, 466 ±20 mm2) was sig¬ nificantly smaller than that in 35 patient controls (mean, 534 ± 12 mnrj. Because
the mean entire pons areas for the con¬ trol groups are similar in the two studies (the value reported by Gaffney et al of 534 ± 12 mm2 vs our value of 535 ± 53.2 mm2), it is unlikely that differences in the MR imaging protocol between their study and ours could account for the discrepancy between the results ob¬ tained in the two studies. Although the control group means are similar in the two studies, the SD for our sample was approximately 10%, while theirs was remarkably small—only 2% in their con¬ trol group and 4% in their autistic group. The ventral pontine areas we measured are similar to those of Hayakawa et al25 in their study of 94 pediatrie patients, 49 adult patients, and seven adult volunteers, none of whom were autistic. They found the average pon¬ tine area in adult (21- to 40-year-old) men to measure 374.1 ± 36.0 mm2 and in
similarly aged
women
to
measure
346.3 ±41.9 mm2 vs our value of 348 ±41.6 mm2. Moreover, the SDs of 10% to 12% in our study are similar to their SDs and, perhaps, are more typi¬ cal of the variation in the general popu¬ lation than that represented by the sam¬ ple evaluated in the study by Gaffney et al.'" We speculate that a subpopulation of autistic subjects may indeed have a smaller pons and that this subpopula¬ tion might have been unduly represent¬ ed in the study by Gaffney et al. '" How¬ ever, in a larger sample of autistic subjects that may be more representa¬ tive ofthe autistic population, we found no tendency for the pons of autistic pa¬ tients to be smaller. The absence of a pontine abnormality in MR imaging data in the majority of autistic subjects is consistent with autopsy data that also reflect no such abnormality. CONCLUSIONS
Our findings indicate that the re¬ duced size ofthe neocerebellum in autis¬ tic subjects is unlikely to be associated with gross structural abnormalities in the major input (corticocerebellar) or
output (cerebellorubral) pathways in the majority of autistic subjects. There¬ fore, we speculate that diminished neocerebellar size in autism is most likely a result of maldevelopment within the cerebellum. To confirm this hypothesis, it will be important to determine wheth¬ er the pontine and red nuclei and the superior and middle cerebellar pedun¬ cles are normal in size in those autistic subjects with developmental neocerebellar hypoplasia. We suggest that the abnormal cerebellum in autism may be relaying incorrect information to nu¬ merous structures within the central
system processes, thereby tributing to some aspects of the symp¬ con¬
nervous
toms of autism.1
References 1. Courchesne E. In vivo neuroanatomical imaging in autism. Pediatrics. 1991;87(suppl):781-790. 2. Courchesne E, Hesselink JR, Jernigan TL, Yeung-Courchesne R. Abnormal neuroanatomy in a nonretarded person with autism: unusual findings with magnetic resonance imaging. Arch Neurol.
1987;44:335-341.
3. Courchesne E, Yeung-Courchesne R, Press GA, Hesselink JR, Jernigan TJ. Hypoplasia of cerebellar vermal lobules VI and VII in autism. N Engl J Med. 1988;318:1349-1354. 4. Murakami JW, Courchesne E, Press GA, Yeung-Courchesne R, Hesselink JR. Reduced cer-
ebellar hemisphere size and its relationship to vermal hypoplasia in autism. Arch Neurol. 1989;46:689-694. 5. Jaeken J, Van den Berghe G. An infantile autistic syndrome characterised by the presence of succinylpurines in body fluids. Lancet. 1984;2:1058\x=req-\ 1061. 6. Bauman ML, LeMay M, Bauman RA, Rosenberger PB. Computerized tomographic (CT) observations of the posterior fossa in early infantile autism. Neurology. 1985;35(suppl 1):247. Abstract. 7. Gaffney GR, Kuperman S, Tsai LY, Minchin S, Hassanein KM. Midsagittal magnetic resonance imaging of autism. Br J Psychiatry. 1987;151:831\x=req-\ 833. 8. Rumsey JM, Creasey H, Stepanek JS, et al. Hemispheric asymmetries, fourth ventricular size, and cerebellar morphology in autism. J Autism Dev Disord. 1988;18:127-137. 9. Garber HJ, Ritvo ER, Chiu LC, et al. A magnetic resonance imaging study of autism: normal fourth ventricle size and absence of pathology. Am J Psychiatry. 1989;146:532-534. 10. Gaffney GR, Kuperman S, Tsai LY, Minchin S. Morphological evidence for brainstem involvement in infantile autism. Biol Psychiatry.
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
1988;24:578-586. 11. Bauman M, Kemper T. Histoanatomic ob-
early infantile autism. Neurology. 1985;35:866-874. 12. Bauman ML. Microscopic neuroanatomical servations of the brain in
abnormalities
in
autism.
Pediatrics.
1991;87(suppl):791-796. 13. Kemper TL: Neuroanatomic studies of dys-
lexia and autism, in Swann JW and Messer A(eds), Disorders of the Developing Nervous System: Changing Views on Their Origins, Diagnoses, and Treatments. New York, Alan R. Liss, Inc, 1988, pages 125-154. 14. Ritvo ER, Freeman BJ, Scheibel AB, et al. Lower Purkinje cell counts in the cerebella of four autistic subjects: initial findings of the UCLA\x=req-\ NSAC autopsy research report. Am J Psychiatry.
1986;143:862-866.
15. Williams RS, Hauser SL, Purpura DP, DeLong R, Swisher CN. Autism and mental retardation: neuropathologic studies performed in four re-
tarded persons with autistic behavior. Arch Neurol. 1980;37:749-753. 16. Nieuwenhuys R, Voogd J, can Huijzen C. The Human Central Nervous System. New York, NY: Springer-Verlag NY Inc; 1988:226. 17. Ornitz EM. Neurophysiology of infantile autism. J Am Acad Child Psychiatry. 1985;24:251\x=req-\ 262. Special Article. 18. Ornitz EM, Atwell CW, Kaplan AR, Westlake JR. Brain stem dysfunction in autism. Arch Gen Psychiatry. 1985;42:1018-1025. 19. Rumsey JM, Grimes AM, Pikus AM, et al. Auditory brainstem responses in pervasive develdisorders. Biol Psychiatry. opmental
1984;191:1403-1418. 20. Courchesne E, Yeung-Courchesne R, Hicks G, Lincoln AJ Functioning of the brain-stem auditory pathway in non-retarded autistic individuals.
Electroencephalogr
Clin
Neurophysiol.
1985;61:491-501.
21. Courchesne E. A neurophysiological view of autism. In: Schopler E, Mesibov GB, eds. Neurobiological Issues in Autism. New York, NY: Plenum Publishing Corp; 1986:285-324. 22. Dawson G, Finley C, Phillips S, Galpert L, Lewy A. Reduced P3 amplitude of the event-related brain potential: its relationship to language ability in autism. J Autism Dev Disord. 1988;18:493\x=req-\ 504. 23. Ciesielski KT, Courchesne E, Elmasian RO. Effects of focused selective attention tasks on event-related potentials in autistic and normal individuals. Electroencephalogr Clin Neurophysiol.
1990;75:207-220.
24. Courchesne E, Lincoln AJ, Yeung-Courchesne R, Elmasian R, Grillon C. Pathophysiologic findings in nonretarded autism and receptive developmental language disorder. J Autism Dev Disord.
1989;19:1-17.
25. Hayakawa K, Konishi Y, Matsuda T, et al. Development and aging of brain midline structures: assessment with MR imaging. Radiology.
1989;172:171-177.
MS; Rachel Yeung-Courchesne; Eric Courchesne, PhD; Gary A. Press, MD
vivo studies involving magnetic resoimaging and studies of neuropathologic specimens have shown that autism is most consistently associated with developmental hypoplasia of the neocerebellum. We investigated whether the cerebellar hypoplasia was accompanied by gross structural abnormalities in the major input (cerebrocerebellar) and output (cerebrorubral) pathways to the cerebellum by measuring the area of the ventral pons (including the pontine nuclei and the transverse fibers) and the midbrain on midsagittal magnetic resonance images in 34 autistic and 44 subjects. The area of the entire pons and several regions of interest within the midbrain (including the superior and inferior colliculi) were also determined with midsagittal magnetic resonance images. We found no significant dif\s=b\ In
nance
ference between measurements of the pons and midbrain in autistic and control subjects. Our data show no evidence of gross anatomic abnormalities in the input and output pathways to the cerebellum in autism, a finding that is consistent with previous studies of
neuropathologic specimens; rather, the reduced size of the neocerebellum in autism appears to be the result of maldevelopment within the cerebellum itself. (Arch Neurol. 1991 ;48:1160-1163) Accepted for publication March 28,1991. From the Departments of Neurosciences (Ms Hsu and Dr Courchesne) and Radiology (Dr Press), School of Medicine, University of California, San Diego, La Jolla, and the Neuropsychology Research Laboratory, Children's Hospital Research Center, San Diego, Calif (Ms Yeung-Courchesne and Dr Courchesne).
to the Neuropsychology ReChildren's Hospital Research Center, 8001 Frost St, San Diego, CA 92123 (Dr
Reprint requests
search
Laboratory,
Courchesne).
studies have reported neuroanatomic abnormalities in many ' patients with autism. In a recent mag¬ netic resonance (MR) imaging study of 18 autistic subjects, developmental hy¬ poplasia of the cerebellum was identi¬ fied in three subjects by routine qualita¬ tive neuroradiologic reading.24 Quantitative analysis of the image data revealed that the neocerebellar vermis (lobules VI and VII) was reduced in size compared with that of controls.3 A sec¬ ond MR imaging study of 10 of these 18 autistic subjects showed that the cere¬ bellar hemispheres were also reduced in size in correlation with the degree of size reduction of the vermis.4 Older studies involving computed tomogra¬ phy have also reported cerebellar hypo¬ plasia in three severely retarded autis¬ tic children with purine metabolism disorders,' and cerebellar atrophy as well as increased width in the upper and middle portions of the fourth ventricle have been detected in an unspecified number of autistic children.1' These lat¬ ter findings suggest that alteration in the cerebellum and the pons, which form the boundaries of the fourth ven¬ tricle, may contribute to its enlarge¬ ment in autistic subjects.' Subsequent studies, however, have reported no ventricular size differences between au¬ tistic and control subject groups."'" Gaff-
Tmaging
al,1" in a study involving 13 highfunctioning autistic subjects, reported ney et
that the entire pons was smaller in the autistic group. In pathologic studies of brain speci¬ mens of autistic patients, abnormalities have been found in the neocerebellum in all cases studied.11"1'' (In Williams et al,15
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
marked
Purkinje
neuron
loss
was re¬
ported in one ofthe four cases studied; of the three patients for whom no cere¬ bellar abnormality was reported, one showed evidence of phenylketonuria,
another had suffered a cerebral concus¬ sion, and the third had some clinical fea¬ tures suggestive of Rett syndrome.) In most,12 but not all,15 cases studied, ab¬ normalities have also been found in por¬ tions of the limbic system. Atrophy of the neocerebellar cortex was accompa¬ nied by decreased numbers of Purkinje cells in 10 autistic patients11"1 "; there was also a decreased number of granule cells in five of these cases.11"13 In two brains that have been studied in greatest de¬ tail, a lack of retrograde cell loss in that part of the inferior olive related to the abnormal cerebellar cortex suggests that Purkinje cell and granule cell loss probably occurred prenatally, before projections from the inferior olive to the Purkinje cells were firmly estab¬ lished.1113 Although the olivary neurons were pale and small, they were normal in number. Additional abnormalities in¬ cluded widening of the fourth ventricle and thinning and elongation ofthe supe¬ rior cerebellar peduncles. In the most profoundly affected specimen, neuronal clusters were found at the edge of the olivary convolutions; moreover, dense¬
ly packed, enlarged neurons were pre¬ sent centrally within the base of the the pons. In other specimens, however, "12 pons appeared to be normal. In the present investigation, we test¬ ed the hypothesis that hypoplasia ofthe
cerebellum in autism occurs without gross structural abnormalities of its ma¬ jor input (cerebrocerebellar) and output
(cerebrorubral) pathways. Because pontine nuclei receive massive input from the frontal cortex and give rise to mossy fiber projections to the cerebel¬ lum (Fig 1), we measured the midsagit¬ tal area of the ventral pons (including the pontine nuclei and the transverse
fibers) in 34 autistic and 44 control sub¬
jects to determine whether the neocerebellar hypoplasia seen in autism is asso¬ ciated with pontine abnormalities. We
also measured the area of the entire pons and regions of interest within the midbrain. SUBJECTS AND METHODS
Subjects
Thirty-four autistic (mean age, 18.9 years; range, 2 to 39 years) and 44 control (mean age, 19.8 years; range, 3 to 39 years) subjects participated in the study, including the 18 autistic and 12 control subjects examined in our earlier studies of cerebellar structures."' The autistic group consisted of 27 male and seven female subjects; the control group con¬ sisted of 40 male and four female subjects. The autistic subjects included 11 mentally retarded subjects. Thirty-five control sub¬ jects were normal, healthy volunteers; the remaining nine were patients referred to the University of California, San Diego, Magnet¬ ic Resonance Institute for evaluation of head and neck pain, headache, depression, and neurogenic bladder, among other symptoms, but who exhibited no radiologie abnormali¬ ties. Wechsler Adult Intelligence Scale-Re¬ vised or Wechsler Intelligence Scale for Chil¬ dren-Revised IQ scores were available for 19 autistic and 19 control subjects. The autistic group had a mean ( ± SD) full-scale IQ score of 82.6 ±15.5, a mean verbal IQ of 78.7±18.1, and a mean performance IQ of 89 ± 13.7. The control group had a mean fullscale IQ score of 113 ± 11.4, a verbal IQ of 112 ±13.2, and a performance IQ of 112 ±10.3. Protocol for MR
Scanning
After appropriate informed consent was obtained, MR imaging scans were obtained with a 1.5-T magnet (GE Signa, General Electric, Milwaukee, Wis). The data matrix was 256 256 at a field of view of 16 or 24 cm, yielding a pixel size of 0.63x0.63 mm2 or 0.94 0.94 mm2, respectively. A 1- to 2-minute scout sequence (repetition time 400 ms; echo time, 20 ms; number of signals aver¬ aged 1) was performed in the axial and sag¬ ittal planes to verify precise subject position¬ ing before the acquisition of high-resolution =
=
sagittal images. Subsequently, a ,-weight¬ ed (repetition time 600 ms; echo time, 25 ms; number of signals averaged 1 or 2) se¬ quence centered precisely at the midline was used to acquire 5-mm-thick sections with 2.5-mm gaps between adjacent sections. =
=
Quantification of Pons and Midbrain Areas
Areas were quantified with computer-as¬ sisted planimetrie analysis ofthe median sag¬ ittal MR images described previously."'4 The area ofthe ventral pons included the pontine nuclei and the transverse fibers. This area is
Fig 1. —Schematic diagram of the cerebellar vermis and related cerebellar nuclei and brain-stem structures of the human brain. 1 indicates cerebropontocerebellar afférents that terminate in the neocerebellar vermis and hemispheres; 2, cerebellar efferente from the Purkinje cells via the deep nuclei to the brain stem and thalamic systems; 3, the red nucleus; 4, the cerebellar nuclei; 5. the nucleus reticularis tegmenti pontis; l-V, vermis lobule l-V; VI-VII, vermis lobules VI-VII; and VIII, vermis lobule VIII (adapted from Nieuwenhuys et al'6).
Fig 2. —Midsagittal ,-weighted images showing the regions of interest measured. VP indicates ventral pons; PONS, entire pons; MB, midbrain; and mb, midbrain excluding the superior and inferior colliculi. bound
by the
ventral border of the medial
lemniscus and the ventral border ofthe pons (Fig 2). We also measured the area of the entire pons and the midbrain, as described by
et al.1" The area ofthe entire pons included the region from the caudal boundary of the midbrain (defined by a line from the superior pontine notch to the junction be¬ tween the tectum and the superior medullary velum) to a line drawn from the inferior pon¬ tine notch to the fourth ventricle (Fig 2). The midbrain area included the tectum, the teg¬ mentum, and the crus cerebri. It was thus bound by the pons caudally (as defined above), the tectal laminar cistern dorsally, the interpeduncular cistern ventrally, and the diencephalon (including the mammillary bodies, hypothalamus, thalamus, and sub¬ thalamus) rostrally (Fig 2). According to Nieuwenhuys et al,"' we defined the posteri¬ or margin of the hypothalamus as "a vertical plane passing just caudal to the mammillary bodies" (Fig 2). Because this measurement of the midbrain might be influenced by the size of the cerebral aqueduct and the sizes of the colliculi, we also measured midbrain areas limiteli dorsally by the floor of the cerebral aqueduct, thereby excluding the colliculi and cerebral aqueduct. Each image was inter¬ preted and measurements were performed
Gaffney
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
by two ofthe investigators (G.A.P. ), one of whom (M.H. or R.Y.-C.) was "blind" to the diagnosis. The measurements were aver¬ aged to obtain mean values for each subject. from which group means were calculated. Measures of brain areas were analyzed with one-way analysis of variance for group ef¬ fects; age effects were determined with re¬ gression analysis.
RESULTS
In all 78
scans
for which
measure¬
performed, there was no visually detectable abnormality of the pons or the midbrain (Fig 3), even in autistic subjects with greatly reduced cerebellar size (eg, 2 to 5 SDs below the ments were
for controls). Measurements ofthe ventral pons re¬ vealed no statistical difference between the autistic group (mean, 350±44.5 mm2) and the control group (mean, 348+41.6 mm2) (F[l,61] 0.147, P>.7; Figs 4 and 5). There was also no signifi¬ cant difference between the area of the entire pons in the autistic group (mean, 540 ±57.6 mm2) and that in the control 5i (mean, group mean
=
500
500 D
400
>
CDcB
D
300
400
200
10
20
30
40
0-
300
>
200
0
10
Age, y 500
400
400
%
£
200
—
0
10
300
-1-1-1
20
Fig 3. —Magnetic resonance imaging scans of the brain stem and vermis of three patients with autism (right) and three age-matched controls (left). The top pair of scans are from children (control, aged 5 years; autistic, aged 5 years); middle pair, from adolescents (control, aged 13 years; autistic, aged 14 years); and bottom pair, from adults (control, aged 23 years; autis¬ tic, aged 22 years). The midline sagittal views show no qualitative evidence of brain-stem atrophy or other morphologic abnormality in the autistic patients compared with the controls in these pairs.
(F[l,ül | 1.07, P>.3; Figs 4 and 5). In the midbrain areas, there was no signifi¬ cant difference between the control group (mean, 240 ±29.9 mm2) and the autistic group (mean, 230±27.1 mm2) (F[l,561 0.526, P>A). There was also no significant difference between mid¬ brain areas that excluded the colliculi and cerebral aqueduct (177 ±25.2 mm2 for the control group and 170 ± 21.5 mm2 for the autistic group) (F[l,641 1.369, P>.\). Interrater reliability was corre¬ lated at .864 for measurements of the ventral pons and .83 for measurements of the entire pons. With increasing age in autistic and control subjects, there was a significant increase in the area measurements of the ventral pons and the entire pons (autistic group: F[l,30]>8.0, P9.3, P
1
10
Age, y Fig 4. —Individual (right) subjects.
30
Age, y
500 -
300
20
20
30
40
Age, y
measurements of the ventral pons and the entire pons in control
(left) and autistic
either the pons or the midbrain in mid¬ sagittal MR imaging scans of autisticsubjects. The absence of a significant difference in the area measurements of the ventral pons, entire pons, and mid¬ brain between autistic and control sub¬ jects argues against the theory that ab¬ normalities in the major input
(corticocerebellar) or output (cerebellorubral) pathways to the cerebellum
responsible for the neocerebellar hypoplasia of autism. are
Our results
are
consistent with
neu¬
ropathologic findings in three autistic patients in whom no midbrain or pontine
abnormalities were detected. The ab¬ sence of retrograde cell loss in the brain stem in these autopsy cases provides evidence that the reduction in Purkinje and granule cell numbers (which likely accounts for the decreased size of the neocerebellum) is due to prenatal maldevelopment within the cerebellum it¬ self, rather than to abnormalities in¬ volving the input/output pathways of the cerebellum. Our MR imaging findings and the re¬ sults of neuropathologic studies are also compatible with results from electro¬ physiologic studies of autism. Although early studies of autism reported abnor¬ mal brain stem —evoked responses,'''" more recent studies report no abnor¬ malities."'2" The recent electrophysio¬ logic studies were carefully constructed and specifically addressed important "'2
méthodologie considerations, including subject homogeneity, cooperation, and body temperature and gender matching of autistic and control subjects. Because
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
Ventral Pons
Fig 5.—Quantitative
Entire Pons
measurements
(mean-SD)
of the ventral pons and entire pons in autistic (solid bars) and control (open bars) subjects. Our data revealed no statistically significant differences between the group measurements.
there is no abnormality in the auditory brain stem-evoked responses, we believe that abnormalities of later com¬ ponents of the auditory-evoked re¬ sponse21"21 cannot be the consequence of abnormalities occurring early in the au¬ ditory pathways. This argument sug¬ gests that autism is unlikely to be asso¬ ciated with deranged auditory brain¬ stem pathways responsible for evoked responses. The involvement of other systems within the brain stem, howev¬ er, cannot be excluded as yet. Our results do not replicate the find¬ ings of Gaffney et al. '" In their study, the area of the entire pons in 13 autistic patients (mean, 466 ±20 mm2) was sig¬ nificantly smaller than that in 35 patient controls (mean, 534 ± 12 mnrj. Because
the mean entire pons areas for the con¬ trol groups are similar in the two studies (the value reported by Gaffney et al of 534 ± 12 mm2 vs our value of 535 ± 53.2 mm2), it is unlikely that differences in the MR imaging protocol between their study and ours could account for the discrepancy between the results ob¬ tained in the two studies. Although the control group means are similar in the two studies, the SD for our sample was approximately 10%, while theirs was remarkably small—only 2% in their con¬ trol group and 4% in their autistic group. The ventral pontine areas we measured are similar to those of Hayakawa et al25 in their study of 94 pediatrie patients, 49 adult patients, and seven adult volunteers, none of whom were autistic. They found the average pon¬ tine area in adult (21- to 40-year-old) men to measure 374.1 ± 36.0 mm2 and in
similarly aged
women
to
measure
346.3 ±41.9 mm2 vs our value of 348 ±41.6 mm2. Moreover, the SDs of 10% to 12% in our study are similar to their SDs and, perhaps, are more typi¬ cal of the variation in the general popu¬ lation than that represented by the sam¬ ple evaluated in the study by Gaffney et al.'" We speculate that a subpopulation of autistic subjects may indeed have a smaller pons and that this subpopula¬ tion might have been unduly represent¬ ed in the study by Gaffney et al. '" How¬ ever, in a larger sample of autistic subjects that may be more representa¬ tive ofthe autistic population, we found no tendency for the pons of autistic pa¬ tients to be smaller. The absence of a pontine abnormality in MR imaging data in the majority of autistic subjects is consistent with autopsy data that also reflect no such abnormality. CONCLUSIONS
Our findings indicate that the re¬ duced size ofthe neocerebellum in autis¬ tic subjects is unlikely to be associated with gross structural abnormalities in the major input (corticocerebellar) or
output (cerebellorubral) pathways in the majority of autistic subjects. There¬ fore, we speculate that diminished neocerebellar size in autism is most likely a result of maldevelopment within the cerebellum. To confirm this hypothesis, it will be important to determine wheth¬ er the pontine and red nuclei and the superior and middle cerebellar pedun¬ cles are normal in size in those autistic subjects with developmental neocerebellar hypoplasia. We suggest that the abnormal cerebellum in autism may be relaying incorrect information to nu¬ merous structures within the central
system processes, thereby tributing to some aspects of the symp¬ con¬
nervous
toms of autism.1
References 1. Courchesne E. In vivo neuroanatomical imaging in autism. Pediatrics. 1991;87(suppl):781-790. 2. Courchesne E, Hesselink JR, Jernigan TL, Yeung-Courchesne R. Abnormal neuroanatomy in a nonretarded person with autism: unusual findings with magnetic resonance imaging. Arch Neurol.
1987;44:335-341.
3. Courchesne E, Yeung-Courchesne R, Press GA, Hesselink JR, Jernigan TJ. Hypoplasia of cerebellar vermal lobules VI and VII in autism. N Engl J Med. 1988;318:1349-1354. 4. Murakami JW, Courchesne E, Press GA, Yeung-Courchesne R, Hesselink JR. Reduced cer-
ebellar hemisphere size and its relationship to vermal hypoplasia in autism. Arch Neurol. 1989;46:689-694. 5. Jaeken J, Van den Berghe G. An infantile autistic syndrome characterised by the presence of succinylpurines in body fluids. Lancet. 1984;2:1058\x=req-\ 1061. 6. Bauman ML, LeMay M, Bauman RA, Rosenberger PB. Computerized tomographic (CT) observations of the posterior fossa in early infantile autism. Neurology. 1985;35(suppl 1):247. Abstract. 7. Gaffney GR, Kuperman S, Tsai LY, Minchin S, Hassanein KM. Midsagittal magnetic resonance imaging of autism. Br J Psychiatry. 1987;151:831\x=req-\ 833. 8. Rumsey JM, Creasey H, Stepanek JS, et al. Hemispheric asymmetries, fourth ventricular size, and cerebellar morphology in autism. J Autism Dev Disord. 1988;18:127-137. 9. Garber HJ, Ritvo ER, Chiu LC, et al. A magnetic resonance imaging study of autism: normal fourth ventricle size and absence of pathology. Am J Psychiatry. 1989;146:532-534. 10. Gaffney GR, Kuperman S, Tsai LY, Minchin S. Morphological evidence for brainstem involvement in infantile autism. Biol Psychiatry.
Downloaded From: http://archneur.jamanetwork.com/ by a New York University User on 05/21/2015
1988;24:578-586. 11. Bauman M, Kemper T. Histoanatomic ob-
early infantile autism. Neurology. 1985;35:866-874. 12. Bauman ML. Microscopic neuroanatomical servations of the brain in
abnormalities
in
autism.
Pediatrics.
1991;87(suppl):791-796. 13. Kemper TL: Neuroanatomic studies of dys-
lexia and autism, in Swann JW and Messer A(eds), Disorders of the Developing Nervous System: Changing Views on Their Origins, Diagnoses, and Treatments. New York, Alan R. Liss, Inc, 1988, pages 125-154. 14. Ritvo ER, Freeman BJ, Scheibel AB, et al. Lower Purkinje cell counts in the cerebella of four autistic subjects: initial findings of the UCLA\x=req-\ NSAC autopsy research report. Am J Psychiatry.
1986;143:862-866.
15. Williams RS, Hauser SL, Purpura DP, DeLong R, Swisher CN. Autism and mental retardation: neuropathologic studies performed in four re-
tarded persons with autistic behavior. Arch Neurol. 1980;37:749-753. 16. Nieuwenhuys R, Voogd J, can Huijzen C. The Human Central Nervous System. New York, NY: Springer-Verlag NY Inc; 1988:226. 17. Ornitz EM. Neurophysiology of infantile autism. J Am Acad Child Psychiatry. 1985;24:251\x=req-\ 262. Special Article. 18. Ornitz EM, Atwell CW, Kaplan AR, Westlake JR. Brain stem dysfunction in autism. Arch Gen Psychiatry. 1985;42:1018-1025. 19. Rumsey JM, Grimes AM, Pikus AM, et al. Auditory brainstem responses in pervasive develdisorders. Biol Psychiatry. opmental
1984;191:1403-1418. 20. Courchesne E, Yeung-Courchesne R, Hicks G, Lincoln AJ Functioning of the brain-stem auditory pathway in non-retarded autistic individuals.
Electroencephalogr
Clin
Neurophysiol.
1985;61:491-501.
21. Courchesne E. A neurophysiological view of autism. In: Schopler E, Mesibov GB, eds. Neurobiological Issues in Autism. New York, NY: Plenum Publishing Corp; 1986:285-324. 22. Dawson G, Finley C, Phillips S, Galpert L, Lewy A. Reduced P3 amplitude of the event-related brain potential: its relationship to language ability in autism. J Autism Dev Disord. 1988;18:493\x=req-\ 504. 23. Ciesielski KT, Courchesne E, Elmasian RO. Effects of focused selective attention tasks on event-related potentials in autistic and normal individuals. Electroencephalogr Clin Neurophysiol.
1990;75:207-220.
24. Courchesne E, Lincoln AJ, Yeung-Courchesne R, Elmasian R, Grillon C. Pathophysiologic findings in nonretarded autism and receptive developmental language disorder. J Autism Dev Disord.
1989;19:1-17.
25. Hayakawa K, Konishi Y, Matsuda T, et al. Development and aging of brain midline structures: assessment with MR imaging. Radiology.
1989;172:171-177.
