Acta Neuropathologica

, Volume 118, Issue 5, pp 673–684

The anterior cingulate cortex in autism: heterogeneity of qualitative and quantitative cytoarchitectonic features suggests possible subgroups

Authors

    • Department of Anatomy and NeurobiologyBoston University School of Medicine
  • Thomas L. Kemper
    • Department of Anatomy and NeurobiologyBoston University School of Medicine
  • Clare M. Timbie
    • Department of Anatomy and NeurobiologyBoston University School of Medicine
  • Margaret L. Bauman
    • Department of Anatomy and NeurobiologyBoston University School of Medicine
  • Gene J. Blatt
    • Department of Anatomy and NeurobiologyBoston University School of Medicine
Original Paper

DOI: 10.1007/s00401-009-0568-2

Cite this article as:
Simms, M.L., Kemper, T.L., Timbie, C.M. et al. Acta Neuropathol (2009) 118: 673. doi:10.1007/s00401-009-0568-2

Abstract

Autism is a behaviorally defined disorder with deficits in social interaction, communication, atypical behaviors, and restricted areas of interest. Postmortem studies of the brain in autism have shown a broad spectrum of abnormalities in the cerebellum and neocortex, involving limbic regions such as anterior cingulate cortex (ACC, Brodmann’s area 24). Using stereological techniques, we analyzed quantitatively cytoarchitectonic subdomains of the ACC (areas 24a, b, c) with regard to cell packing density and cell size. Microscopic examination of the ACC was also done to identify any neuropathologies. Results showed a significant decrease in cell size in layers I–III and layers V–VI of area 24b and in cell packing density in layers V–VI of area 24c. Direct comparisons revealed irregular lamination in three of nine autism brains and increased density of neurons in the subcortical white matter in the remaining cases. Because previous studies have suggested that von Economo neurons (VENs) may be altered in autism, a preliminary study of their density and size was undertaken. VEN density did not differ between autism and control brains overall. However, among the nine autism cases, there were two subsets; three brains with significantly increased VEN density and the remaining six cases with reduced VEN density compared to controls. Collectively, the findings of this pilot study may reflect the known heterogeneity in individuals with autism and variations in clinical symptomotology. Further neuroanatomic analyses of the ACC, from carefully documented subjects with autism, could substantially expand our understanding of ACC functions and its role in autism.

Keywords

AutismACCvon Economo NeuronsNeuropathology

Introduction

Autism is a behaviorally defined disorder with a prevalence rate estimated at one in 500 [40]. Twin studies indicate a strong genetic component [8, 20]. Essential clinical features include impaired social interaction and communication and restricted repertoires of activities and interests [6]. Studies of the brains of individuals with autism have revealed abnormalities in brain size with cellular abnormalities in the brainstem, cerebellum, limbic system, and neocortex [9, 13, 30, 35, 50, 63]. The most consistently reported abnormal cortical region in behavioral, imaging, and pathological studies is the anterior cingulate cortex (ACC) [9], a key component of the limbic system with contributions to affective and cognitive behaviors, and motor activity [16].

The ACC, and other closely related neocortical areas, have been implicated in the monitoring of social interaction, including the detection of adverse outcomes and the initiation of modifying behaviors [1, 4, 14, 22, 38, 45, 5254, 70]. It is also believed to play a role in theory-of-mind where it is represented by a circuit linking the ACC with the adjacent frontal cortex and temporoparietal junction [21]. Mundy [45] and Ohnishi et al. [49] have suggested that this same circuitry is involved in joint attention, an area of deficiency frequently observed in many autistic individuals. Other behaviors attributed to the ACC include emotional learning, motivationally significant goal-directed behavior, long-term socio-emotional attachments, emotional self-control, adaptive responses to changing conditions, and spatially complex bimanual motor coordination [4, 31, 66]. Most recently, the ACC was associated with conflict processing via its effective contribution to neuronal activity in the caudal cingulate and other cortical regions [19]. Frith and Frith [21] have suggested that autism may provide a paradigm for the consequences of ACC dysfunction.

Several imaging studies have identified abnormalities in the ACC in autism. Using positron emission tomography (PET), Ohnishi et al. [49] reported a decrease in cerebral blood flow in the left ACC. Using similar techniques, Haznedar et al. [26] found a decrease in glucose metabolism throughout the entire cingulate gyrus and reduced volume of the right ACC. Welchew et al. [65], in a fMRI study, found evidence of abnormal connectivity of the ACC with multiple other cortical areas. Barnea-Goraly et al. [10], with diffusion tensor imaging, reported reduced fractional anisotropy in the anterior cingulate gyrus bilaterally that extended into the adjacent corpus callosum and ventromedial frontal area and the subgenual prefrontal region as well as the temporoparietal junctions bilaterally and adjacent superior temporal gyrus. Using single photon emission computed tomography and a 5HT2A receptor ligand, Murphy et al. [46] found reduced 5HT2A receptor density in the anterior and posterior cingulate gyrus, frontal lobe, superior temporal gyrus, and left parietal lobes in autism. The latter ACC receptor binding deficits were significantly correlated with abnormal reciprocal social interaction.

There have been few neuropathological investigations examining ACC cytoarchitecture in autism. In a comprehensive comparison, microscopic study of nine brains of individuals with autism and age- and gender-matched controls, the ACC was observed to be poorly laminated with small cells and altered cell packing density [3436]. To date, the cytoarchitecture of the ACC has not been systematically studied using modern stereological principles. A potential pathologic feature of the ACC in individuals with autism is a unique neuronal population, the von Economo neurons (VENs). These large, spindle-shaped bipolar projection neurons occur in layer V of ACC and of frontoinsular cortex (FI) [47, 48]. The late emergence of VENs in primate phylogeny and human development may render these cells more vulnerable to abnormalities during neuronal development. In terms of primate phylogeny, in the study of an ample array of primates, VENs have been found to occur only in humans and great apes [47, 48]. VENs are also present in large-brained, socially complex cetacean species [11, 28], and in African and Asian elephants [24] suggesting an adaptive convergence for VEN evolution [28]. In human brain development, VENs are first noted between the 35th gestational week reaching adult numbers by the fourth year of life [5]. In the autistic brain, the period of VEN development overlaps with the reported abnormal brain overgrowth in autism [57], a change that is most significant in frontal cortex [12, 27].

In a study utilizing autism and control subjects across a broad age range, abnormal brain overgrowth, specifically increased head circumference, was reported to be a ubiquitous feature of all individuals with autism with the distribution curve shifted to the right [41]. The same study observed that 12–20% of these autism subjects were macrocephalic by 3–5 years of age. The timing of these events suggests the possibility that VENs are more prone to neurodevelopmental abnormalities in individuals with autism than in the normally developing brain. VENs are further implicated in autism due to their speculated contributions to behaviors that are altered in the disorder. VENs have a postulated role in socio-emotional and higher order cognitive processing [35, 47] leading to the speculation that they may be abnormal in autism [5, 15, 37, 45]. VENs have also been implicated in social functioning as evidenced by their significantly lower numbers in subjects with frontotemporal dementia [60], a neurodegenerative disease marked by diminished social behavior and emotion, and in agenesis of the corpus callosum, a disorder marked by deficits in social and emotional behavior [33]. VENs have not been examined in the ACC of individuals with autism.

As noted above, in many studies, the ACC is thought of as a functional unit in a constellation of other brain areas, making it often difficult to be certain that abnormal functions attributed to the ACC are the result of pathology within it or possibly a reflection of abnormalities elsewhere. The present study was designed to qualitatively and quantitatively document possible structural abnormalities in the ACC in nine male patients with autism and four male control cases matched for age-range. Using modern stereological techniques, the size and packing density of the neurons in the three cytoarchitectonic subdomains of the ACC were quantified and surveyed for developmental abnormalities. Since the literature suggests that there may also be abnormality of VENs in autism, the current investigation also includes a preliminary study of their neuronal density, size, and distribution in the ACC. Furthermore, given the strong role of ACC in social interaction and postulated role of VENs in socio-emotional and higher order processing, severity of autism correlated with any neuropathology in the autism patients is discussed.

Materials and methods

Subjects and histology

We obtained a single block of the ACC, Brodmann’s area 24, taken from either the right or the left cerebral hemisphere from nine male autism and four male control cases matched for age-range (Tables 1, 2). All subjects met DSM-IV criteria for autistic disorder [6]. The classification of the severity of autism features was estimated from a review of patient medical records, telephone administration of the Autism Diagnostic Interview (ADI), and a telephone interview regarding patient history by one of the authors, MLB [69] (Tables 1, 3).
Table 1

Case number, medication history, severity of autism determined by author MLB in [69], hemisphere used, age, gender, postmortem interval (PMI), and cause of death of the nine autism and four control subjects

 

Severity of autism

Hemisphere

Age

Gender

PMI

Case of death

Control

 3706

 

N/A

20

M

21

Asphyxiation

 3849

 

R

27

M

16

Arteriolosclerosis of coronary arteries and aorta

 3865

 

N/A

55

M

12

Heart attack

 4334

 

R

53

M

24

Cancer

Autistic

 2403a,b

 

L

26

M

5

Cardio-respiratory arrest

 3804a,c

 

L

15

M

4

Acute pancreatitis

 3916

 

R

32

M

21

Congestive heart failure

 4099

+++

L

19

M

3

Bronchial pneumonia, Muscular Dystrophy

 4705b

 

R

20

M

18

Suicide

 3845a,d,e

+++/++++

L

32

M

28

Pancreatitis, Seizures

 3511a,e

++/+++

L

27

M

16

Trauma

 4414a,d

+/++

L

26

M

48

Seizure complication

 2431

+++

L

54

M

4.3

GI Bleed

N/A not available; +, mild severity of autism; ++, moderate severity of autism; +++, severe severity of autism; ++++, profound severity of autism; **Cerebellum not included in brain weight

aSeizure disorder

bHistory of carbamazepine (Tegretol®) use

cHistory of Depakote

dHistory of phenytoin (Dilantin®) use

eHistory of Phenobarbital use

Table 2

Mean neuron density (neurons/mm3) in layers I–III and layers V–VI of the three cytoarchitectonic subdomains of the anterior cingulate cortex in autism and control brains

Case

24a I–III

24a V–VI

24b I–III

24b V–VI

24c I–III

24c V–VI

Autism

 2403

87,363

85,714

74,227

57,143

133,516

115,727

 3804

69,674

51,724

125,714

112,892

103,136

102,948

 3916

41,143

49,286

64,690

82,540

69,024

72,078

 4099

85,947

69,916

87,651

46,043

147,966

105,934

 4705

39,514

62,222

73,950

80,000

113,349

111,688

 3845

a

a

129,796

102,298

106,620

118,452

 3511

92,497

46,938

103,120

88,490

a

111,089

 4414

81,922

94,033

100,799

65,438

180,633

88,031

 2431

65,886

100,373

96,153

95,238

162,637

102,539

 Means

70,493

70,025

95,122

81,120

127,110

103,165

 SEM

7,296

7,443

7,548

7,232

12,790

4,903

Control

 3706

5,413

52,698

143,799

77,460

126,917

131,764

 3849

56,131

99,884

70,726

54,082

59,864

78,224

 3865

88,260

91,996

96,190

78,571

211,744

177,232

 4334

91,321

97,479

136,720

89,209

218,509

205,714

 Means

72,462

85,514

111,859

74,831

154,258

148,233

 SEM

10,033

11,063

17,262

7,406

37,739

27,865

 P value

0.44

0.13

0.16

0.31

0.2

0.02

aCytoarchitectonic subdomains with poor tissue quality

Brain tissue, fixed in 10% formalin, was obtained from the Harvard Brain Tissue Resource Center, The Autism Research Foundation (TARF), the brain bank located at the University of Miami, and the NICHD Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, MD, USA via the Autism Tissue Program (ATP). The ACC was blocked 1 cm caudal to the anterior tip of the genu of the corpus callosum. Blocks were cryoprotected in 10% glycerol–2% DMSO for 3 days, followed by 5 days in 20% glycerol–2% DMSO. Cryoprotected blocks were flash-frozen using isopentane and dry ice. The blocks were cut on the frozen stage of a sliding microtome; sections were cut in the coronal plane at a thickness of 80 μm in rounds of 13 sections, yielding 5–10 rounds of tissue sections per brain. The first round from each brain was mounted on gelatin coated glass slides and was defatted in a 1:1 ethanol to chloroform solution. Sections were rehydrated in a series of graded alcohols and deionized water. The mounted sections were then stained with 0.05% thionin, dehydrated in an ascending series of alcohols, cleared using xylene, and coverslipped using Permount (Fisher, Pennsylvania).

Qualitative neuropathological analysis

A qualitative, blind analysis of possible additional neuropathology in the ACC sections was made using a comparison microscope. With this microscope, paired tissue sections, matched for age-range, from control and autism brains, were shown side by side in the same field of view at the same magnification, permitting direct comparisons of lamination patterns, white matter neurons, and other cytoarchitectonic features.

Counting methodology and statistical analysis

The major cytoarchitectonic subdomains of the ACC, areas 24a, 24b and 24c, were delineated using cytoarchitectonic criteria [64] (see Figs. 1, 2). All tissue sections used in this study met the strict cytoarchitectonic criteria of area 24a, b, and c [64]. Area 24a has the highest density of pyramidal neurons. It is characterized by definable layers II and III, a thin, prominent layer Va, and a cell sparse cell layer Vb with clumped neurons. This area has the second highest density of VENs [48]. Cytoarchitectonic area 24b has an overall “rain shower” effect. It has a clearly defined layer II, a distinct and broad layer III, a thick, prominent layer Va, and a broad layer Vb. This area has the highest density of VENs among all three cytoarchitectonic subdomains [48]. Cytoarchitectonic subdomain 24c is characterized by a thick superficial region (II–III) relative to a thin deep region (V–VI) and layer V is relatively homogeneous. Area 24c has the lowest density of VENs [48].
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig1_HTML.gif
Fig. 1

Medial view of the human cingulate gyrus. Area 24 samples were blocked 1 cm caudal to the anterior tip of the genu of the corpus callosum (arrow) (adapted from Gray’s Anatomy of the Human Body, 1918)

https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig2_HTML.jpg
Fig. 2

Nissl-stained coronal section through the anterior cingulate cortex showing subdomains and regions of interest, 24a, 24b, and 24c. Each cytoarchitectonic subdomain (delineated by lines) was overlaid with a rectangular contour, composed of two subcompartments, 1 mm in width (see 24c above). The top compartment sampled layers I–III. The bottom compartment sampled layers V–VI. The rectangular contour extended into the subcortical white matter for VEN sampling. See “Materials and methods”. Scale bar 1 mm

Stereological parameters were set to yield a coefficient of error (CE) of 0.1 or less. ACC neuronal nucleolar counts were made in a 7 μm deep counting box within the thickness of ACC in a 1 mm-wide column centered in each cytoarchitectonic subdomain. Sampling was from the center of each cytoarchitectonic subdomain in order to bypass the potential ambiguity in architectonic border delineation in ACC. For neuron counts, the column extended from layer one to the gray matter–white matter border; this border was drawn where cellular density dropped off. Each cytoarchitectonic subdomain was further delineated into superficial layers (I–III) and deep layers (V–VI), thus creating six distinct regions of interest in each ACC section. For VEN counts, the 1 mm-wide column was also centered in each cytoarchitectonic subdomain. The 1 mm-wide column extended from layer V through layer VI and into the underlying subcortical white matter, the corticosubcortical boundaries were drawn where the cell density dropped off. In this study, the underlying subcortical white matter zone contains widely scattered neurons below the compact cell layer of layer VI (layer Vla), corresponding to cortical layer VIb in classical cytoarchitecture. It corresponds to Cajal’s layer of white matter and fusiform cells [56].

All analyses were made on coded slides to keep observers blind to the subjects’ diagnoses. Neurons within the three cytoarchitectonic areas were counted using a Nikon Eclipse 80i microscope with a motorized stage computer assisted by StereoInvestigator Version 6.0 (MBF Biosciences, Willington, VT). ACC neurons were systematically and randomly sampled with the optical fractionator probe [68] from seven tissue sections per brain at a section interval of every 13th section. A 290 × 290 μm scan grid overlaid each region of interest. A 50 × 50 μm dissector box with a height of 7 μm within the tissue section and a top guard zone of 1 μm was used to count cells using exclusion–inclusion criteria [67]. A minimum of 200 neurons were counted per brain with a minimum of 20 sites sampled per tissue section. Neurons were identified by the presence of a prominent nucleus, a nucleolus, and stained cytoplasm, with the nucleolus used as the counting object. For all neuronal counts, the Vertical Nucleator with four rays was used to sample neuronal area and volume. The Nucleator was applied when the clearest profile of the nucleolus was in focus.

Because the VENs comprise a relatively small fraction of layer V cells and showed no significant difference in cell size, all VENs within the regions of interest were counted, and counted without the use of exclusionary planes. The VEN nucleolar counts were taken from two randomly selected sections from each case. For each section, section thickness was determined. VENs were identified by a slender elongated soma, lighter staining than pyramidal neurons, and a symmetrical morphology about the cell’s vertical and horizontal axis [47, 48] (Fig. 3). VENs were further characterized by apical and basal dendritic processes of the same thickness oriented perpendicular to the pial surface. VENs were distinguished from layer VI fusiform neurons and atypical pyramidal neurons by their overall larger size and distinct morphological criteria previously discussed [47, 48].
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig3_HTML.jpg
Fig. 3

Nissl stained von Economo neurons (VENs) and pyramidal neurons of layer V in the ACC. VENs have an elongated spindle shape, one apical and one basal process, and are larger on average than pyramidal neurons. Scale bar 10 μm

Neuron packing density in the ACC was extrapolated using the number of objects counted, the number of sites visited, and the volume of the dissector box and from section thickness and area sampled in the case of the VENs. Group differences between autism and control brains in the superficial layers I–III and deep layers V–VI of areas 24a, 24b, and 24c were assessed using a two-tailed Student’s t test assuming equal variances. The potential confounding variables of post mortem interval and age were also assessed using two-tailed Student’s t tests assuming equal variances and any significant interactions of hemisphere on ACC neuron and VEN density were assessed using a one-way ANOVA and the post hoc Tukey test. To explore if seizures had an effect on cell size and density, we ran the Mann–Whitney U Test, correcting for multiple comparisons, using the critical value P = 0.01. Due to the inconsistent use of seizure medication in autism subjects, and the dual presence of seizures and seizure medication in most patients, it was difficult to assess the effect of medication. Coefficient of error (CE) was calculated by dividing standard error of the mean by cell density [58].

Maps of VEN distribution were generated by exporting the StereoInvestigator tracings and adjusting them in Adobe Illustrator. These tracings show the contour of the region of interest, the 1-mm wide column centered in each cytoarchitectonic subdomain. The cell markers reflect the exact location of each VEN.

Results

In all Student’s t test comparisons, PMI and age did not have a significant effect on neuron size or density. In all Mann–Whitney U tests, there was no significant effect of seizures/medications on neuron size or density. With the ANOVA, there was one significant interaction where the neuronal density in area 24a layers I–III was decreased in the right hemisphere compared to the left hemisphere in the autism group (F(2,8) = 5.9, P < 0.05 and post hoc Tukey test). In the autistic brains, a significant decreased cellular packing density was noted in the layers V–VI of area 24c (P = 0.02) and there was a significant decreased cellular area and volume in layers I–III and layers V–VI of area 24b (P = 0.01–0.03; Table 2; Figs. 4, 5a, b). The CE was less than 0.1 in all cases. Direct histopathologic examination of age-range matched control and autism brains with a comparison microscope revealed irregular lamination in three of the nine patients with autism (2431, 3845, 2403) and increased density of neurons in the subcortical white matter in two autism brains (4414, 3511) (Fig. 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig4_HTML.gif
Fig. 4

Mean neuron density (±SEM) in layers I–III and V–VI of subdomains 24a, b, c in autism and control brains. There was a significant decrease (*P = 0.018) in the density of neurons in layers V–VI of region 24c in autism

https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig5_HTML.gif
Fig. 5

Mean neuron volume and area (±SEM) in layers I–III and V–VI of subdomains 24a, b, and c in autism and control brains. In both layers I–III and layers V–VI of area 24b there was a significant decrease in the autistic brains a neuronal area (*P = 0.021 and 0.032, respectively) and b neuronal volume (*P = 0.013 and 0.02, respectively)

https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig6_HTML.jpg
Fig. 6

Irregular lamination and increased number of subcortical white matter neurons. Top panel Representative irregular lamination in the ACC area 24a in a patient with autism (arrow) (2431, lower row) versus control (arrow) (3706, upper row) at low and high magnification. Scale bars 1 mm and 200 μm, respectively. Bottom panel Representative increased number of subcortical white matter neurons in the ACC area 24a in autism (4414, left column) versus control brain (4334, right column) at low and high magnification. Scale bars 200 μm and 100 μm, respectively

The hemisphere obtained for study did not affect VEN counts (P > 0.05). There was no overall significant difference between VEN density and size in the autism and control groups (P > 0.05). However, in the autistic brains there was a striking dichotomy in their densities. Three autism cases (3845, 4414, 3511) had an unusually high density of these cells whereas the remaining six (2403, 4705, 3804, 3916, 4099, 2431) had an unusually low density of these cells. There was a significant difference in VEN density between these two autism groups in cytoarchitectonic subdomains 24a, b and c (P = 7 × 10−4, 3 × 10−4, and 0.04, respectively). The autism cases with an abnormally low density of VENs were significantly different from controls in areas 24a, b, and c (P = 0.01, 0.01, and 0.004, respectively). The autism cases with an abnormally high density of VENs had a significantly lower density in areas 24a and 24b as compared to controls (P = 0.03 and 0.04, respectively) (Fig. 7). It is notable that the VEN density gradient toward area 24c (with 24b > 24a > 24c) noted in the literature [48] is present in the control cases; but was less striking in the autism cases with low VENs and reversed between areas 24a and 24b in the autism cases with high VENs (Fig. 7).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig7_HTML.gif
Fig. 7

Density of VENs in layers V–VI in subdomains 24a, b, c and in subcortical white matter in autism compared to control brains. There was no significant difference between the overall density in autism versus controls. However, within the autism group there were two subpopulations with markedly different VEN numbers indicated by a natural clustering and break in the density data. As compared to controls, three autism cases (3845, 4414, 3511) had a significantly higher mean density of VENs in subdomains 24a, b and six autism cases (2403, 4705, 3804, 3916, 4099, 2431) had a significantly higher mean density of VENs in subdomains 24a, b, c (red-coloured asterisk denotes statistical significance in the high VEN group compared to controls and blue-coloured asterisk denotes statistical significance in the low VEN group compared to controls, P < 0.05)

In the autism cases, VENs were distributed in layers V–VI and rarely in the subcortical white matter. In the control cases, VENs were primarily distributed in V-VI but in contrast to the autism brains they were also found, in very small number, in the subcortical white matter (Figs. 8, 9). VENs were not present in layers I–III in the control and autism cases.
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig8_HTML.gif
Fig. 8

Distribution of VENs in layers V–VI and subcortical white matter in representative sections from a control brain (3849) and autism brains (4414, 2403). Each cytoarchitectonic subdomain (24a, b, and c) was overlaid with a large, 1-mm wide rectangular contour composed of three smaller subdivisions. The top subdivision contained layers I–III, the middle subdivision contained layers V–VI and the bottom subdivision contained adjacent subcortical white matter. Each dot represents one VEN

https://static-content.springer.com/image/art%3A10.1007%2Fs00401-009-0568-2/MediaObjects/401_2009_568_Fig9_HTML.gif
Fig. 9

Representative distribution of VENs in layers V–VI in sections from control brain (3849) and autism brains (4414, 2403) in ACC subdomain 24a. The representative microphotographs reflect VEN distribution of ACC area 24a in each of the three brains (3849, 4414, 2403). Each red circle encompasses one VEN. Layers V, VI are indicated with lines. Scale bar 100 μm

Discussion

The present study documents a variety of neuropathologies in the ACC of male patients with autism, suggesting that malfunction of this key component of the limbic system may play a role in some of the clinical features of males with autism. The ACC showed a significant decrease in neuronal size in the layers I–III and layers V–VI of area 24b and a significant decreased neuronal packing density in layers V–VI of area 24c. There was no significant difference in VEN density between the control and autism cases. However, there was a natural clustering of autism cases according to VEN densities. Three autism cases had a significantly higher VEN density as compared to controls and six autism cases had a significantly lower density than the controls. Five autism brains showed evidence of minor malformations. Three of these showed abnormalities in cortical lamination and cell distribution and two cases had an increased number of subcortical white matter neurons.

The decrease in neuronal size in the ACC in autistic patients has been previously observed [3436]. The subjective impression of an increased neuronal packing density, noted in these reports, however, was not confirmed by the objective observations of the current study. Instead, a decrease in neuronal density was found in one of the six counting areas (24c). There was one interaction of cerebral hemisphere on neuron density within the autism brains: the left hemisphere, compared to right, showed a significantly greater neuronal density in layers I–III of area 24a—a situation which may warrant future investigation. While macroscopic studies of the cingulate gyrus show leftward cerebral hemispheric asymmetries of the paracingulate sulcus and area 24 [29, 42, 51] only one microscopic study has quantitated cellular density in ACC subdomains. Gittins et al. [23] quantitatively assessed five bilateral rostral–caudal levels of area 24b in normal brain specimens and found that neuronal density was higher in the left mid-supragenual level—a finding not born out in the present study. Furthermore, there is reported intersubject variability in macroscopic features of area 24 [64] that may indicate cytoarchitectural variability in the area. Nevertheless, the decreased neuronal density found here has also been noted in other stereological studies of the brain in autism; van Kooten et al. [63] reported it in layer III of the fusiform gyrus, along with smaller perikaryal volume in layers V and VI, and Schumann et al. [59] found it in the lateral nucleus of the amygdala. These changes in cell size and packing density presumably represent focal areas of altered neuronal development and/or circuitry, abnormalities that could not be elucidated in the present study.

The disordered lamination in the ACC had been previously noted by Kemper et al. [34] and Kemper and Bauman [35, 36]. The abnormal laminar pattern in the cerebral cortex and the increased number of neurons in the subcortical white matter appear to be due to two different pathological processes. One of these is the persistence of a normally transient fetal and neonatal neuronal circuit. In the earliest stages of cerebral cortical development, there is a transient zone called the primordial plexiform layer into which the definitive neurons of the cerebral cortex migrate, separating it into the future layer I and scattered neurons in a transient subplate zone deep to the definitive cerebral cortex [43, 62]. The subplate zone is unusually prominent in the human brain, reaching its peak development in the 24th week of gestation and largely disappearing by the 6th postnatal week [39]. It appears to play an essential role in the development of cerebral cortical circuits, particularly thalamocortical circuits, but it is also implicated in the development of other subcortical projections to the cerebral cortex and in the development of ipsilateral and contralateral corticocortical circuits [7, 32, 39, 61]. The presence of excess number of neurons in the subcortical white matter in individuals with autism suggests a lack of proper resolution of the transient zone in the brain in autism, with the implication that the attendant cortical circuits may be compromised. An arrest of these neurons during their migration through the prospective subcortical white matter could conceivably also account for the ectopic neurons in this location. This is less likely as these subcortical neuronal heterotopias generally occur as clusters of cells rather than scattered individual neurons [25]. A similar increased number of subcortical neurons have also been noted in the brains of patients with schizophrenia [2, 17, 18].

The disordered lamination within the ACC appears to be due to a different mechanism. Rakic [55], in his paper on malformations of the cerebral cortex, divides malformations into three categories; failure of migration from the germinal zone, neurons arrested during migration to the cerebral cortex, and aberrant placement of neurons in their target area. The ectopic distribution of these cells is consistent with their aberrant migration within the cortical plate. This suggests that this malformation arises at an earlier time during fetal development than the persistence of scattered neurons in the subcortical white matter. In normal development, these migratory neurons first appear at about 7 weeks of gestation with the migration continuing until about 15 weeks of gestation [61].

This is the first study to investigate the status of the VENs in the ACC in patients with autism. No significant overall difference between the density of VENs in autism and control brains was noted similar to the study of Kennedy et al. [37] that found no significant difference between controls and patients with autism in the total number of VENs in cortical area FI. However, in the present study and in the reports of Kennedy et al. [37] and Allman et al. [5], there is evidence that there may be subgroups with different densities and distribution of VENs within the patients with autism. In the present study, in areas 24a and 24b, three of the nine autism brains had a significantly higher density of VENs in comparison to controls and the other six a significantly lower density. In area 24c, only the low-density group was also significantly different from the controls. In Kennedy et al. [37], one of four autism brains in area FI showed a marked increased number of VENs as compared to the controls. Allman et al. [5], in a preliminary study of two autism brains, noted a heavy concentration of VENs in the white matter that extended through layers V and VI. The present study differs from Allman et al. [5], in that a small number of VENs were found in the subcortical white matter in the controls and rarely in the subcortical white matter of the autism cases. With regard to possible subgroups, it is of interest to note that three brains with a high density of VENs also showed evidence of abnormal cytoarchitecture. One brain had irregular lamination of the cerebral cortex and in second, there were an increased number of neurons in the subcortical white matter. Only two of the six brains with a low density of VENs showed a cytoarchitectonic abnormality (irregular lamination). Similarly, marked variability in the pathology of the brain in autism can also be seen in the detailed neuropathology reports of Bailey et al. [9], Hutsler et al. [30] and in the review paper of Palmen et al. [50].

The subsets of cerebral cortical abnormalities and VEN densities provide a unique opportunity to examine potential clinicopathological relationships within autism cases. Because the ACC is strongly implicated in modulating social interaction, a logical hypothesis is autism cases with cerebral cortical malformations will exhibit the most severe autism features. Furthermore, given the suspected role of VENs in higher-order cognitive and socio-emotional processing, a logical hypothesis is the cases with decreased VENs might exhibit the most severe autism symptomotology. However, it can be seen in Table 3 that the available clinical data do not support a relationship between cortical malformations and autism severity or a relationship between clinical features of autism subjects and subpopulations of VEN densities. Furthermore, there was no remarkable relationship between brain weight and neuropathology. While the available clinical data does suggest that seizures may be associated with cerebral cortical abnormalities, there does not appear to be a relationship between seizures and VEN densities in the ACC. Four of the five autism cases with cerebral cortical abnormalities had seizures. Seizures were present in all three of the autism cases with high VEN density and two of six autism cases with low VEN density. The possibility that male autism patients with seizures represent a subpopulation with distinct morphology should be explored in future studies. A future study utilizing rigorous stereological principles, in a larger and balanced number of specimens controlling for cerebral hemisphere, is also needed to confirm the other ACC findings reported here.
Table 3

Clinicopathologic relationships in autism subjects based on available clinical data (severity of autism determined by author MLB in [69])

 

2431

4414

4099

3916

3511

3804

4705

3845

2403

Clinicopathologic relationships

 Severity of autism

+++

+/++

+++

 

++/+++

  

+++/++++

 

 Seizures

Not noted

Yes

Not noted

Not noted

Yes

Yes

Not noted

Yes

Yes

 White matter neurons

 

Yes

  

Yes

    

 Irregular lamination

Yes

      

Yes

Yes

 High VENs

 

Yes

  

Yes

  

Yes

 

 Low VENs

Yes

 

Yes

Yes

 

Yes

Yes

 

Yes

 Brain weight (g) (left hemisphere)

646

710

864

 

732

  

699

388**

+, mild severity of autism; ++, moderate severity of autism; +++, severe severity of autism; ++++, profound severity of autism; **Cerebellum not included in brain weight

Acknowledgments

Tissue was provided by the Harvard Brain Bank Tissue Resource Center, the Autism Tissue Program and the NICHD Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, MD. Supported by the National Institutes of Health Studies To Advance Autism Research and Treatment (NIH STAART) U54 MH66398 and National Institutes of Neurological Disorders and Stroke (NINDS) NS38975-01A1. We gratefully acknowledge Hillary Kaplan for her help with tissue processing, Dr. Michael Bowley for helpful discussions regarding stereological principles and design, Jerry Skefos for his expert technical assistance, and Dr. Ron Killiany for his statistical expertise.

Copyright information

© Springer-Verlag 2009