Tissue processing, embedding, sectioning, and staining

Tissue samples, demographics of individuals with autism and control subjects, brain weight, and changes during processing were characterized in our previous report [35]. In short, the postmortem interval (PMI), corresponding to the period between death and autopsy, ranged from 6 to 28 hours in the control group, and from 3 to 50 hours in the affected group, but the difference in PMI between the two groups was not significant. The average weight of the brains in the cohort with autism (1,453 g) was not significantly different from that in the control group (1,372 g). The brain hemisphere was fixed with 10% buffered formalin for an average of 408 days in the control group and 905 days in the affected group. Brain dehydration in a series of ascending concentrations of ethyl alcohol for 36 and 38 days in the control and affected groups, respectively, resulted in reduction of brain hemisphere weight by 47% on average in the group of subjects diagnosed with autism and by 45% in the control group. Brain hemispheres were embedded in 8% celloidin [37]. Serial 200-μm-thick sections stained with cresyl violet and mounted with Acrytol were used for morphometric studies.

Delineation of regions of interest

The anatomical boundaries of the examined brain structures and their cytoarchitectonic subdivisions were described in detail in our previous report [35]. Delineation of structures requiring micrographs is illustrated with Figures 1 and 2, which show partitioning of the striatum (caudate nucleus, putamen, nucleus accumbens and globus pallidus); the amygdala (lateral, basal, accessory basal and central nuclei); layers of the entorhinal cortex; and sectors in the Ammons horn.

The rostral striatum is divided into the caudate nucleus and the putamen by the internal capsule. The border between the nucleus accumbens, the caudate nucleus and the putamen was arbitrarily defined by a line perpendicular to the midline of the internal capsule at its ventral end [38]. The number of striatal small neurons was estimated but a few dispersed large neurons [39] were not included in this study. The external medullary lamina marks the border between the putamen and the globus pallidus. The medial medullary lamina divides the globus pallidus into the external and internal pallidum [40]. The number of neurons was determined in both parts of the pallidum. The boundaries of the lateral, basal, accessory basal and central nuclei were delinated using cytoarchitectonic criteria described in detail by Schumann and Amaral [31],[41]. The entorhinal cortex (EC, Brodmann area 28) occupies the gyrus ambiens and the anterior portion of the hippocampal gyrus. The anterior border of the EC starts a few millimeters rostrally to the amygdala complex. The EC ends rostrally to the lateral geniculate body. The medial portion of the EC borders with the preamygdaloid cortex and caudally with the subicular complex. The lateral portion of the EC borders with Brodmann area 35, which lacks a distinct layer IV [42]. The transentorhinal zone [43] was not included in this study. Neurons were measured within four layers of the EC. Layer II consists of islands of large modified pyramidal and stellate neurons separated by an acellular gap from layer III. A broad layer III is built up of medium-sized pyramidal neurons separated from layer V by the lamina dissecans, labeled as layer IV. Layer V consists of the large pyramidal neurons, whereas Layer VI is built up of the smallest neurons with a decreasing gradient of neuronal density in the deep portion of this layer [42].

In Ammon’s horn the number of neurons was estimated in the pyramidal layer in sectors CA1–4. The CA1 sector pyramidal neuronal layer extends from the subiculum to the CA2 sector as a thick band of medium size pyramidal neurons. Two features distinguish the CA2 sector: a compact arrangement of large pyramidal neurons and the lowest thickness of this layer. The size of neurons in the CA3 and 4 is similar, but parallel arrangement of neurons in the CA3 sector distinguishes CA3 from randomly dispersed neurons in the hilus, identified also as CA4 sector [42],[44]. The mapping of sectors on serial sections was based on the Amaral and Insausti [42] classification, but the CA4 sector has been delinated as a separate subdivision.

The neurons were distinguished from glial cells by using histological features revealed by cresyl violet staining, including neuron size, shape, and spatial orientation typical for specific layers, sectors, and brain nuclei, as well as the pattern of staining of nuclear chromatin, the distinct nucleolus in the majority of examined neuronal populations, and cytoplasm morphology. Small and round nuclei with uniform, intense staining of nuclear chromatin distinguished oligodendrocytes from astrocytes with large, round nuclei with a small amount of dispersed chromatin and an undetectable nucleolus.

Stereological analysis

Volume evaluation and counting of neurons were performed using a workstation consisting of an Axiophot II (Carl Zeiss, Goettingen, Germany) light microscope with Plan-NEOFLUAR objectives—1.25×(numerical aperture, N.A., 0.035), 2.5×(N.A. 0.075), 5×(N.A. 0.15), 20×(N.A. 0.5), and 40×(N.A. 0.75); a specimen stage with a three-axis, computer-controlled stepping motor system (Ludl Electronics; Hawthorne, NY, USA); a CCD color video camera (CX9000, MicroBrightfield Bioscience, Inc., Williston, VT, USA); and stereology software (Stereo Investigator, MicroBrightfield Bioscience).

The mean volumes, number of neurons, and neuronal density of structures in autistic and control brains were compared in analyses of covariance (ANCOVAs) and repeated measures analysis of variance in which brain weight and, for neuronal counts and density, log age, post-mortem interval, and days of fixation were entered as covariates. Pearson correlations were computed for age and the total number and density of neurons in the autistic and control cohort.

Repeated measures analysis of variance

To test the hypotheses that structures differ in the degree to which autism affects, first, the number of neurons in the structure and, second, the volume of the structure, we conducted repeated measures analyses of variance with structure and autistic status as factors, with the potential confounders of brain weight, post-mortem interval, days of fixation, and log age entered as covariates. The significant interaction of structure with autistic status in each case indicates that autism's effect upon the number of neurons (F[df = 13] = 2.138, p<.018) and upon structural volume (F[df = 13] = 10.278, p<.001) differs among the structures examined.

Volume of brain subdivisions

In the control group, the volume of 14 brain structures and their 24 subdivisions ranged from the largest structures, such as the thalamus, putamen, and caudate nucleus (2,999 mm³, 2,365 mm³, and 1,727 mm³, respectively) to the smallest inferior olive (59 mm³) (Figure 3). Table 2 summarizes the results of estimation of the volume of brain regions and their cytoarchitectonic subdivisions as the absolute volume calculated using mounted section thickness (StereoInvestigator, Microbrightfield). To compare the results of volume estimates with other reports, the volume was corrected for 47% tissue shrinkage during dehydration for the autism group and 45% shrinkage for the control group. A significant difference between the volumes of the examined structures in the affected and control cases was detected only in the caudate nucleus and nucleus accumbens. In the group with autism, caudate nucleus volume was larger by 22% (2,107 mm³ in the autism cohort, and 1,727 mm³ in the control cohort; p < 0.036). The nucleus accumbens was larger by 34% (244 mm³ in the autism cohort, and 182 mm³ in the control cohort; p< 0.050). However, analysis of alterations in 4- to 8-year-old children did not reveal a significant difference between the affected and control groups. The finding of a significantly larger caudate nucleus and nucleus accumbens only in 9- to 60-year-old individuals with autism suggests that the difference may be a result of changes during adolescence/adulthood.

The mean volume of the examined cerebellar hemisphere was almost identical in the affected and control groups: 31,779 mm³ and 31,133 mm³, respectively. The mean volume of the molecular and granule cell layers (12,623 mm³ and 12,092 mm³, respectively), and the percentage of the cerebellar cortex occupied by these two layers (51% and 49%, respectively) in the group with autism did not reveal a significant difference in comparison to the control group (12,885 mm³ and 11,926 mm³, respectively; with the volume ratio 52% and 48%, respectively). The mean volume of white matter was almost the same in the affected and control cohorts (6,700 mm³ and 6,005 mm³, respectively), as was the percentage of the cerebellar volume occupied by white matter (21% and 19%, respectively). Also, the volume of cerebellar nuclei (364 mm³ and 317 mm³, respectively), including the largest dentate nucleus (244 mm³ and 232 mm³, respectively), did not reveal a significant difference between individuals with autism and control subjects.

Number of neurons

The numerical density of neurons was brain region—specific and varied in the control group from 2,339/mm³ in the globus pallidus to 40,031/mm³ in the nucleus accumbens. In the group with autism, a significantly lower numerical density of neurons was detected in three brain regions: the lateral nucleus in the amygdala (24,390/mm³ in the affected group and 27,446/mm³ in the control group; p < 0.017), putamen (23,749/mm³ in the affected group and 27,531 mm³ in the control group; p < 0.041), and nucleus accumbens (33,107/mm³ in the affected group and 40,031/mm³ in the control group; p < 0.018). Box plots demonstrate that in the majority of subjects with autism the numerical density of Purkinje cells, neurons in the lateral nucleus in the amygdala, nucleus accumbens and in the putamen is less than the median numerical density in the control group (Figure 4).

The volume of cerebellum and cerebellar subdivisions as well as the number of neurons was estimated after exclusion of records for brain samples of two subjects with autism: one with focal defects of cresyl violet staining related to poor fixation, and the second because of focal dysplasia. A 60-year-old subject with autism and clinically diagnosed disturbed movement coordination and postmortem-detected severe hypoplasia of cerebellar lobes I-IV [36] was excluded from statistical analysis to eliminate the impact of one case of focal developmental defect on estimates of developmental alterations in an examined group of subjects with autism. In this cerebellum, the numerical density of Purkinje cells was 2.3 times less (274/mm³) in the hypoplastic area than in the control cerebellar cortex. The total number of Purkinje cells was 42% less in this case (7.1 million) than the mean total number in the examined hemisphere of other subjects with autism and 65% less than in the control group.

The average total number of Purkinje cells within the examined cerebellar hemispheres was significantly less (p < 0.0001) in the group with autism (12.1 million) than in the control cohort (16. million) (25% deficit). The minimal number of Purkinje cells in the examined hemisphere in the control group was 11.4 million, but in five of 12 examined subjects in the affected group (42%), the number was less than in this control subject. The total number of Purkinje cells was less in all individuals with autism than the average number in the control group. The numerical density of Purkinje cells was also significantly less (p < 0.0024) in affected subjects (488/mm³) than in control subjects (645/mm³) (-24.4%).

The absence of a significant correlation between age and the total number or numerical density of neurons (N/mm³) in the cohort with autism suggests that the detected neuronal deficits do not decrease greatly, if at all, from 4 to 60 years of age among subjects diagnosed with autism. However, examination of only three subjects with autism 52 to 60 years of age and of three control subjects 51–64 years of age is not favorable for detection of aging-associated neuronal loss.

Cerebellum

Studies of the cerebellum of individuals diagnosed with autism provide inconsistent results, including smaller cerebellar hemispheres [45],[46] and smaller vermis [47]-[49] as well as increased cerebellar volume or volume proportional to brain size [50]. The study by Fatemi et al. [25] of the cerebellum of five males with autism with a mean age of 25.2 years and of five age-matched controls revealed a 24% decrease in mean Purkinje cell size but no differences in Purkinje cell densities. The study of the cerebellum of a 16-year-old female diagnosed with autism and severe intellectual deficit showed no alterations in the cerebellum [24]. Several reports suggested associations between the size of the vermis and attention deficits, enhanced stereotypic behavior, and reduced exploration in autism [51]-[53].

Our review of the literature indicates that the current study is the first to report on the total number of Purkinje cells and cerebellar global volumetry in autism. The study revealed a significantly lower total number of Purkinje cells in the cerebellum in the group of subjects who were diagnosed with autism (12.1 million per examined hemisphere, and 24.2 million calculated for entire cerebellum) than in control subjects (16.0 and 32.0 million, respectively). The total number of Purkinje cells in the entire cerebellum of control subjects in our cohort 4–64 years of age is comparable to estimates reported by other authors, with some modifications due to age and methods: 28.5 million in 10 control males, 28–84 years of age [54]; 28.6 million in 19 control subjects, 19–84 years of age [55], 29.8 million in 10 controls, 21–65 years of age [56]; and from 19.7 to 27.0 million in four controls, 41–84 years old [57].

Autism is associated with significant inter-individual differences in the number of Purkinje cells, but a smaller number of Purkinje cells was observed in the majority of 29 individuals diagnosed with autism (reviewed by Palmen et al., [58]). Recently, eight subjects with autism and seven control subjects from our cohort of 14 autistic and 14 control subjects were examined by Skefos et al. [59] to detect links between structural alteration of four cerebellar subdivisions and functional abnormalities typical for autism. The authors examined the crus I and II in the posterior lobe reciprocally connected with prefrontal cortical networks that modulate social behavior, and the hemispheric portion of lobules IV-VI known as a primary sensorimotor processing area. The lower Purkinje cell density in these regions supports the hypothesis that abnormal Purkinje cell density may contribute to selected clinical features of the autism phenotype. The reduced number of Purkinje cells and dysplastic changes in lobule X may have a direct contribution to alterations of eye movement in subjects with autism [60]. Some studies suggest that a regional decrease in the number of Purkinje cells may be the result of prenatal loss of Purkinje cells [27]. Some inter-individual differences in the number of Purkinje cells reported in the past might be related to treatment of epilepsy-associated loss of Purkinje cells [61]-[64] and/or pre-mortem changes related to the mechanism of death. Signs of structural developmental alterations coincide with a 40% decrease in the expression of glutamic acid decarboxylase 67 (GAD67) mRNA in autistic subjects and an increased GABAergic feed-forward inhibition to Purkinje cells by basket cells. These data suggest a disruption in the timing of Purkinje cell firing and altered inhibition of the cerebellar nuclei, which could directly affect cerebello-cortical output, leading to changes in motor behavior and cognition [65].

In the examined cohort, the difference between volumes of the cerebellar hemisphere in the group with autism and control groups (31,779 mm³ and 31,133 mm³, respectively) was not significant. This volume reflects a 45% reduction of brain weight during dehydration in ethyl alcohol of the brains of control subjects and a 47% reduction in individuals diagnosed with autism. The volume calculated for both cerebellar hemispheres of control subjects and corrected for 45% shrinkage is equal to 113.2 mm³ and is comparable to 124.8 cm³ in the study by Andersen et al. [55] of 19 control males from 19 to 84 years of age.

This study found a significant—25%—deficit in the number of Purkinje cells in the absence of significant differences between affected and control cohorts in the volume of the molecular and granule cell layer, cerebellar white matter, or total volume of cerebellar nuclei. Previous study revealed severe Purkinje cell soma volume deficits in the range of 31% in 4- to 8-year-olds and of 23% in 29- to 60-year-olds [35]. These findings, and the absence of abnormalities in the number and distribution of GABAergic basket and stellate interneurons in the cerebellar molecular layer reported by Whitney et al. [27], suggest that Purkinje cells are the major target of the genetic and/or epigenetic factors that define the contribution of Purkinje cells' pathology to the autistic phenotype.

Striatum

In our control cohort the volume of the caudate nucleus was estimated as 1,727 mm³, versus 2,212 mm³ in Kreczmanski et al. [66] study, and 2,043 mm³ in Bogerts study [67] (with Kreczmanski's et al. correction for shrinkage). The volume of the putamen was 2,365 mm³ in our material, 2,713 mm³ in Kreczmanski et al. study [66], and 2,979 mm³ in Bogerts study [67] (with Kreczmanski's correction for shrinkage). The volume of the nucleus accumbens was estimated as 182 mm³ in this study, 204 mm³ in Kreczmanski et al.[66], 147 mm³ in Lauer et al. [68] (with Kreczmanski's shrinkage factor), and 138 mm³ in Pakkenberg study based on examination of 4-μm thick paraffin sections [69].

The volume of the caudate nucleus was significantly greater in the group with autism than in the control group (2,107 mm³ and 1,727 mm³, respectively; p < 0.036). Several MRI studies of patients with autism have revealed increased volumes of the basal ganglia [20],[70]-[72] and a correlation between caudate nucleus volumes and repetitive-behavior scores [70],[71]. A direct relationship between repetitive behaviors and right caudate nucleus volume was reported in an cohort of 17–55 years of age patients diagnosed with autism [71], but not in a cohort including children [73]. Brain size increases to the age of 5 years, whereas the volume of the caudate grows until late childhood [18],[74] and decreases during adulthood [73],[75]. Regional reduction of the caudate nucleus head volume by up to 50% and the presence of foci of tissue loss, detected by repeated scanning with high-resolution MRI and use of continuum mechanical tensor maps, were interpreted as evidence of the pruning process at 9–13 years of age in control subjects [74]. The lack of differences on postmortem evaluation of caudate nucleus volume between 4- to 8-year-old autistic and age-matched control subjects, but the significantly larger caudate nucleus and nucleus accumbens and insignificantly higher (by 11%) total number of neurons in these structures in 9- to 60-year-old subjects diagnosed with autism in our cohort may indicate timely pruning in the control group, but regionally defective pruning in affected subjects, as postulated by Voelbel et al. [76].

A subcomponent of changes in the striatum is the significant deficit in the volume of neuronal soma in the nucleus accumbens (34%) and caudate nucleus (22%) in 4- to 8-year-old subjects with autism, but not in affected adolescents or adults [35]. These data suggest the deregulation of mechanisms that control the number and volume of neurons and the total volume of the caudate nucleus and nucleus accumbens. Correlations between MRI-detected alterations within the striatum and the severity of clinical changes [70],[71] imply that these cellular alterations may contribute to stereotyped and ritualistic behaviors in autism. The nucleus accumbens, which is considered a mixed structure with elements of the striatum and the amygdala [77], processes information about reward value and availability and is involved in reward-related behaviors [78]. Projections of nucleus accumbens neurons to the ventral pallidum translate reward information into motivated action [79]. A significantly larger nucleus accumbens (by 34%, p < 0.050) appears to be another deregulated component of striatal circuits that may enforce repetitive behaviors in autism.

The striatum alterations observed in idiopathic autism resemble some structural and functional changes observed in fragile X syndrome (FXS). Autism is diagnosed in 25–30% of individuals diagnosed with FXS [80]-[84]. Hatton et al. [82] suggested that lower levels of fragile X mental retardation protein (FMRP) expression contribute to autistic behavior and intellectual deficits in children with FXS. Gothelf et al. [85] revealed a correlation between lower levels of FMRP, behavioral alterations, and alterations of the size of some brain subdivisions, including increased size of the caudate nucleus but decreased size of the amygdala, superior temporal gyrus, and posterior cerebellar vermis in FXS. A 75% reduction of FMRP expression in vermis and a 50% reduction in the superior frontal cortex (BA9) in adults with idiopathic autism may contribute to both structural abnormalities and the autistic phenotype [86].

Amygdala

Neuropathological studies [87] and results of structural [10],[88] and functional [89] neuroimaging indicate that the amygdala is affected in autism. The study by Schumann and Amaral [31] of the amygdala in nine brains of individuals with autism 10–44 years of age and 10 brains of age-matched controls did not reveal a difference in the number of neurons in the basal, accessory basal, and central nucleus. However, the authors reported 13% fewer neurons in the lateral nucleus (3.47 million in the autistic group and 4.00 million in the control group; p < 0.05) and 12% fewer neurons in the entire amygdala (10.74 million in the affected group and 12.21 million in the control group; p < 0.05). The significantly (p < 0.017) reduced numerical density of neurons in the lateral nucleus in the amygdala of subjects with autism in our cohort of 14 subjects 4–60 years of age (24,390/mm³ in the affected group and 27,446/mm³ in the control group) and the absence of significant changes in other amygdala subdivisions in these subjects confirm the selective developmental abnormality of the lateral nucleus. In our study, the total number of neurons in the lateral nucleus in subjects with autism was also less (by 17%) (4,609,751) than in the control group (5,544,098), but after control for five covariates, the difference was found to be not significant. Differences in tissue preservation, the age of examined subjects, and selection of only epilepsy-free males in the study by Schumann and Amaral [31] may contribute to some differences in the number of neurons reported, but the general pattern of alterations detected in both studies is similar.

Reduced neuronal density in the absence of a change in the total number of neurons in the amygdala may suggest the involvement of non-neuronal factors, including glia. Ahlsen et al. [90] reported increased levels of GFAP, a marker of astroglial activation, in the CSF in subjects with autism. There are no data about astrocytosis within the amygdala, but Laurence and Fatemi [91] demonstrated by Western blots an increased level of GFAP in superior frontal cortex (BA9; by 45%), parietal cortex (BA40; by 75%) and in the cerebellum (by 49%) in subjects with autism. Vargas et al. [92] reported signs of selective activation of astroglia and microglia in several brain regions of subjects with autism. A new concept of tripartite synapses with astrocytes controlling synapses formation and function [93] and signs of astrocytosis may suggest a glial contribution to structural and functional changes in the brains of subjects diagnosed with autism.

Individuals with autism display deficits of social behavior, including abnormalities in social reciprocity, difficulties in the use of eye contact, and deficits in social motivation [1]. The theory of mind proposes that the amygdala is one of several brain regions that are parts of the social brain and are necessarily abnormal in autism [89]. Moreover, the amygdala controls behavioral responses such as the fear and anxiety observed in 84% of children with autism [94],[95]. The alterations in the amygdala detected in morphometric studies and in MRI-based volumetry suggest that developmental abnormalities of the amygdala may contribute to the autistic phenotype. Some differences of the estimated total number of neurons in the lateral nucleus in the amygdala in the control group in Schumann and Amaral [4] study (4.0 million), Kreczmanski et al. [66] (4.43 million) and 5.54 million in our study reflect mainly technical differences and in part interindividual differences.

Neocortex and archicortex

Stoner et al. study [96] revealed patches of disorganization in the prefrontal and temporal cortices in a majority of children diagnosed with autism. This focal disruption of cortical architecture suggests dysregulation of layer formation and layer-specific neuronal differentiation at prenatal developmental stages. Van Kooten et al. [32] showed significant reductions in neuron densities in layer III, total number of neurons in layers III, V, and VI in the fusiform gyrus, but none of these alterations were found in the primary visual cortex or the whole cerebral cortex. This study provides insight about the cellular basis of abnormalities in face perception in autism. Several other reports indicated the absence of alterations in the number of neurons in the cerebral cortex [97]; in the superior frontal cortex [98]; in the BA44 and BA45 [28]; and in the frontoinsular cortex [23]. A study of von Economo neurons revealed no significant difference in the mean number of these neurons in the frontoinsular cortex [26] and in the anterior cingulate cortex (BA24) [33]. A study of the frontoinsular cortex in four subjects with autism, (4–11 years of age) and three control (4–14 years of age) subjects revealed an increase in the ratio between von Economo neurons and pyramidal neurons, but the difference between the number of von Economo neurons in subjects with autism and control subjects fell short of statistical significance, and the total estimated number of pyramidal neurons did not reveal a significant difference [23].

Our study of the archicortex of subjects with autism revealed focal microdysgenesis with selective deficits of stellate neurons in irregular smaller islands in the second layer of the entorhinal cortex in 23- and 60-year-old affected subjects [36]. The reduced total number of neurons in the islands and in the third layer of the entorhinal cortex suggests developmental abnormalities within the perforant pathway projecting to the dentate gyrus and Ammons horn. These alterations may affect memorization and emotional behavior [99]. Similar deficits of stellate neurons in islands and a thicker molecular layer in the entorhinal cortex of subjects diagnosed with schizophrenia were described as evidence of defects of migration of neurons to their target destination during development [100],[101].

Developmental deficit and loss of neurons in autism

For much of the history of postmortem studies of ASD pathology, brain abnormalities have been viewed as static [102], but the regression observed in 15–62% of autism cases has been linked to neurodegeneration and neuronal loss [103]. However, morphological markers of neurodegeneration have been reported only sporadically, including cytoplasmic homogenous inclusions in Purkinje cells in affected child [98], Purkinje and granule cells' mineralization [87], and neurofibrillary degeneration in a 24-year-old autistic female with self-injurious behavior including head-banging [104].

The Fatemi et al. [105] review demonstrates agreement on a Purkinje cells deficit in the cerebellum of individuals diagnosed with autism. The lower number of Purkinje cells in subjects diagnosed with autism is considered mainly the result of the loss of these cells before 30 weeks’ gestation [87], but postnatal Purkinje cell death accompanied by astrocytosis has been also reported [98]. In some cases, loss of Purkinje cells might be the result of chronic treatment with anti-epileptogenic drugs such as phenytoin [61],[62]. The lack of correlation between the number of Purkinje cells and age in the examined cohort of autistic subjects 4–60 years of age suggests that the 25% deficit of Purkinje cells is mainly a developmental fetal defect, as postulated by Kemper and Bauman [87], but the combination of prenatal deficit and postnatal loss cannot be excluded.

Comparison of the total number of neurons in the Ammons horn in this control group and control groups in other reports reveals differences related to the methods of tissue preservation and morphometric evaluation, and the age of examined cohorts. Our study revealed that the total numbers of neurons in the CA1, CA2/3, and CA4 sectors of control subjects 4–64 years of age were 11.5, 2.1, and 1.1 million, respectively. The numbers of neurons in the CA1, CA2/3, and CA4 sectors in control subjects 21–56 years of age were determined to be 14.7, 2.7, and 2.12 million, respectively [106]; for 47– 85 years of age, 16.0, 2.7, and 2.0 million, respectively [107]; and for 72–96 years of age, 6.2, 1.7, and 0.7 million, respectively [108]. Similar, age-associated differences are observed in the substantia nigra, with the total number of neurons of 707,000 in our control group, and of 550,000 neurons in 71- to 87-year-old control subjects in the Pakkenberg et al. study [109].

Several studies have revealed neuronal pathology affecting cytoplasmic organelles and possibly sustaining functional abnormalities in autism, but there is no direct evidence that they cause significant neuronal death. An abnormal electron transfer chain in neuronal mitochondria may alter energy generation, metabolism, and cell structure and function; however, these changes prevail only in children, not in adults [110],[111]. An imbalance between mitochondria biogenesis and degradation increases the mass of compromised mitochondria in neurons in the temporal cortex, but this pathology was also demonstrated only in young individuals diagnosed with ASD [112]. Currently, there is no evidence that the elevated oxidative stress detected as an increase of 3-nitrotyrosine in the cortex, cerebellum, and pons of subjects with autism [113] and the two-fold increase of DNA oxidation in the frontal and temporal cortex and in the cerebellum [114] cause neuronal loss. In idiopathic autism and autism associated with dup15, a self-enhancing cascade of pathological changes alters amyloid precursor protein (APP) metabolism, and increases cytoplasmic accumulation of amino-terminally truncated amyloid-β, which is the source of reactive oxygen species and which increases the formation of lipid peroxidation products. These changes enhance amyloid-β deposition and sustain the cascade of changes contributing to metabolic and functional impairments in neurons [115],[116]. These alterations may contribute to activation of astrocytes and microglial cells, enhanced astrocytosis, and astrocytes' death [115]; however, the potential neuronal loss associated with this pathology appears to be hidden within a range of interindividual differences and was not detected as an age-associated correlate in the examined cohort.