Article type: Review Article

Accepted Article
Autism Spectrum Disorders: Linking Neuropathological Findings to

Treatment with Transcranial Magnetic Stimulation

Manuel F. Casanova, M.D.1, Estate Sokhadze, Ph.D.1, Ioan Opris, Ph.D.2, Yao Wang1,3, and

Xiaoli Li, Ph.D.3

1
Department of Psychiatry, University of Louisville
2
Department of Physiology and Pharmacology, Wake Forest University School of

Medicine
3
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal

University

Running title: Autism and TMS

Correspondence to:

Manuel F. Casanova, M.D.

Gottfried and Gisela Kolb Endowed Chair in Psychiatry

University of Louisville

Department of Psychiatry

500 South Preston Street, Bldg A, Rm 217

Louisville, KY 40202

Email: m0casa02@louisville.edu
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1111/apa.12943
This article is protected by copyright. All rights reserved.
 ABSTRACT
Accepted Article
Postmortem studies in autism spectrum disorder (ASD) individuals indicate the presence of

abnormalities within the peripheral neuropil space (PNS) of cortical minicolumns. The

geometrical orientation of inhibitory elements within the PNS suggests using repetitive

transcranial magnetic stimulation (rTMS) to up-regulate their activity. Several rTMS trials in ASD

have shown marked improvements in motor symptomatology, attention, and perceptual binding.

Conclusion: rTMS is the first therapeutic attempt at ASD aimed at correcting some of its core

pathology.

Keywords

cerebral cortex, minicolumn, transcranial magnetic stimulation, autism spectrum disorder

Key notes

 The presence of heterotopias and dysplasias in the brains of autistic individuals clearly

frame the condition as a neurodevelopmental disorder.

 The smallest unit of information processing within the modular organization of the

cerebral cortex is the minicolumn.

 The presence of neuronomorphometric abnormalities within the periphery of

minicolumns in the brains of autistic individuals suggests using low frequency repetitive

transcranial magnetic stimulation (rTMS) as a possible treatment intervention.

INTRODUCTION

Bauman and Kemper’s pioneering histoanatomic studies briefly described the presence

of cerebral cortex pathology within their series of autism spectrum disorder (ASD) individuals.

Described changes included an indistinct laminar pattern and minor cortical malformations or

This article is protected by copyright. All rights reserved.

 dysplasias (1). Nevertheless, the first group to emphasize the role of the cerebral cortex in the
Accepted Article
neuropathology of ASD was that of Bailey and colleagues (2). Since then many studies have

underlined individual aspects of this pathology including changes in neuronal size and density

(3), abnormalities of minicolumnar morphometry (4, 5), heterotopias (6), the presence of

supernumerary cells in both the molecular layer and the subplate region (7), and dysplasias (8).

The prominence and prevalence of cortical pathology along with reports of widespread

heterotopias has prompted some researchers to suggest that, in some cases, ASD results from

the heterochronic divisions of germinal cells (periventricular and rhombic lip) giving rise to the

desynchronization in maturation between radially derived pyramidal cells and tangentially

derived interneurons (9). This abnormality of germinal cell division and later migration appears

to explain many of the reported neuropathological and clinical aspects of both idiopathic and

syndromic autism. In this regard autism can be considered a neurodevelopmental disorder as it

impairs the growth and development of the brain.

The cerebral cortex provides the mainboard of crucial circuitry that allows for the

emergence of higher cognitive functions such as attention, memory, judgment, language and

empathy. Impairments of corticogenesis usually manifest themselves early in life as varying

degrees of cognitive dysfunction for which they are traditionally labeled as neurodevelopmental

disorders. Understanding the mechanisms for the development of the cerebral cortex is

therefore of major importance when discussing the pathogenesis of autism.

In this article we will summarize recent findings regarding the organization of the

cerebral cortex, its ontogenetic origin, and related neuropathological findings in the brains of

ASD individuals. Subsequently we will describe how the topographical specificity of these

abnormalities, being primarily within the periphery of minicolums, is amenable to treatment with

repetitive transcranial magnetic stimulation (rTMS). We finalize the article by summarizing the

results of different clinical trials using rTMS in ASD.

This article is protected by copyright. All rights reserved.

 CORTICAL MODULARITY
Accepted Article
The cerebral cortex is organized into different areas that process information in a

hierarchical feedforward manner. Although it would be profitable to differentiate these areas in a

quantitative and unbiased manner this endeavor has been fraught with difficulties. At present,

techniques for visualizing cell morphometry can only poorly demarcate large portions of the

cortex. About two thirds of the cerebral cortex lies buried in folds under the surface of the brain.

The curvature and repositioning of cellular elements in these abstruse areas has confounded

researchers to such an extent that their existence has been ignored in different neuroanatomical

treatises. Indeed, the limitation of cytoarchitectural techniques led neuroanatomists to conclude

that attempts at subdividing the cerebral cortex by histological methods would prove

unprofitable, if not impossible. For many decades it has thus been recommended to

complement cytoarchitectural surveys with other techniques, notably those based on fiber

connectivity. Indeed the lack of “cookie cutter borders” between cytoarchitecturally-defined

parcellations of the cerebral cortex has led to an alternate affiliation system based on functional

connections.

Attempts at parsing the cerebral cortex by means of functional connections have yielded

a limited number of discrete areas (10, 11, 12) where Information travels in a stereotyped

manner going from the extrapersonal space to idiotypic cortex (primary sensory and motor

areas) and from there to homotypical (unimodal and heteromodal), paralimbic, and limbic

cortices. Since there is an infinite number and variety of inputs the proposed computation

system begs the question as to how intrinsic cortical circuits are adapted to so many different

tasks. According to Creutzeldt (13) the cerebral cortex overcomes this obstacle by instantiating

a canonical circuit that provides for similar transformations throughout the cerebral cortex

(Figure 1). This view upholds that functional differences among brain regions depends on

This article is protected by copyright. All rights reserved.

 variations in input sources, output targets, interconnectivity, and the modulatory effect of
Accepted Article
inhibitory influences.

Common patterns in the organization of nerve cells and their connections (i.e.,

microcircuits) have been described in the cerebral cortex of all examined mammalian species.

Lorente de Nó was the first to discuss the physiological role of “vertical cylinders” of cells when

he famously said: “All the elements of the cortex are represented in it, and therefore it may be

called an elementary unit, in which, theoretically, the whole process of transmission of impulses

from the afferent fiber to the efferent axon may be accomplished” (14). Later on

electrophysiological studies by Mountcastle helped define a minicolumnar organization of

neurons that encoded similar features and shared nearly identical terminal fields (15). Over the

years it has become clear that the repetitive and “on-demand” event-based nature of the

proposed canonical circuit is an enabler of neuroplasticity that allows different areas of the

cerebral cortex to process information stemming from varied sources, including some that are

artificial in nature.

In recent years Opris and colleagues (16) have implemented the use of custom-

designed conformal (i.e., following the longitudinal contour of the cortex) multielectrode

recording arrays to describe the coding of interlaminar circuitry within and between minicolumns

(Figure 2). Their group has found that substituting firing patterns of layer 5 neurons with

electrical stimulation during minicolumnar transmission from layers 2/3 improved task

performance (target selection) in primates (17). More importantly, when microelectrode

stimulation was applied following the pharmacological ablation of a cortical region, the

researchers observed recovery of performance (18). This finding, based on encoding the

pattern of signals of individual minicolumns, is the first reported intervention where a

neuroprosthesis has been used to successfully restore diminished cognitive function (19).

This article is protected by copyright. All rights reserved.

 Construct validity for the minicolumn, and the appropriateness of inferences based on
Accepted Article
their existence, is derived from multiple sources. Population analysis of neurons within the same

receptive field in primary somatosensory cortex (SI) shows that diversity within segregates is

attributable to variability among minicolumns rather than within minicolumns (20, 21). Metabolic

studies (2-deoxyglucose) of the receptive fields of monkey’s SI evoked by skin stimuli show

column-shaped patches of label comprised of active minicolumns interspersed with less active

minicolumns (22). Two-photon imaging of calcium influxes evidences an organization of cortical

response fields comprised of columns 1 to 2 cell wide (23). Similarly, studies using labeled

retroviruses and multiple cell recordings illustrate a propensity for pyramidal cells to develop

connections with sister cells rather than to neighboring non-siblings (24). This affinity may be

the result of the close apposition of neuroblasts as they migrate radially towards the cortical

plate. Indeed, early during brain development these cells are connected to each other via gap

junctions thus procreating a suitable scenario for Hebbian reciprocity. Minicolumns are therefore

highly integrated structures where connections within the modules far exceed those between

modules. In this regard minicolumns are modules within a hierarchical system whose parts are

interconnected by anatomical and physiological relationships fulfilling the function of an

information-processing unit.

It is important to note that although different minicolumns appear to convey the same

canonical circuit they are not clonal elements. Anatomical studies have found significant

minicolumnar variability within and across areas in individual brains as well as between

mammalian species. This variability (e.g., number of neurons and synaptic elements) between

minicolumns may be an adaptive feature that is responsible for both fault tolerances within

larger networks (e.g., macrocolumns) as well as for the competition among modules that is

necessary for circuit optimization (25).

This article is protected by copyright. All rights reserved.

 CORTICOGENESIS AND AUTISM
Accepted Article
During corticogenesis germinal cells surrounding the ventricles divide and migrate to the

cortical plate to form a consecutive series of strata one on top of another following an inside–out

configuration (Figure 3). Neuroblasts derived from periventricular germinal cells migrate radially

towards the cortical plate using a glia projection as a scaffold. Computerized image analysis of

the radial translaminar arrays of pyramidal cells has shown a chronological continuity from early

on in gestation (i.e., the ontogenetic minicolumn) through postnatal maturation and aging (26).

These neurons mature in synchrony with cells that are themselves derived from a tangential

migratory stream stemming primarily from the ganglionic eminences. The radially migrating cells

(future pyramidal cells) and tangentially derived neurons (future interneurons) provide for

functional units called dyads, e.g., pyramidal-double bouquet cell dyads, pyramidal-basket cell

dyads. Deficits of neuronal migration are likely to impair the formation of these dyads thus

promoting an excitatory/inhibitory bias capable of explaining the high prevalence of epilepsy

among neurodevelopmental disorders.

Some abnormalities of the cerebral cortex pinpoint to a developmental malformation

deficit. Occasional brains exhibit an excessive number of small convolutions (gyri) and, in

others, sulci have been noted to run in an abnormal direction (27). Microscopic examination has

revealed indistinct or disordered lamination, small neurons, and changes in cell-packing density

(2, 28). Bailey and colleagues noted the presence of increased number of neurons in both Layer

1 and the subcortical white matter (2). These findings, indicative of a migrational disorder, have

been reproduced by Avino and Hutsler in a study using computerized image analysis and in a

qualitative study that surveyed whole brain serial sections by Wegiel et al. (7, 8). More recently,

the presence of focal cortical dysplasias was quantitated in serial sections of ASD individuals as

compared to neurotypicals (29). Morphometric analysis of neurons within involved patches

This article is protected by copyright. All rights reserved.

 suggested a reduction in the overall number of interneurons and a reduction in size of remaining
Accepted Article
pyramidal cells.

Several studies have used unbiased semiautomated imaging methods to analyze

minicolumnar morphometry in ASD. In one study, photomicrographs from Brodmann areas 9,

21, and 22 were decomposed according to a gray-scale distribution and subsequently

thresholded to define cellular kernels in 9 ASD individuals and an equal number of neurotypicals

(4). The least square method was used to fit a radial axis through the clusters of Nissl stained

elements in order to define a mean tangential axis-to-axis distance, i.e., minicolumnar width.

The results showed significant narrowing (p = 0.034) of minicolumns in ASD with the greatest

decrease found in the PNS compartment. Thinning within the PNS of minicolumns in ASD was

observed across layers (i.e., supragranular, granular, infragranular) suggesting involvement of a

common anatomical element throughout the cortical width of ASD individuals. The findings were

latter corroborated using the same patient population but a different parcellation technique: the

gray level index (GLI) (3).

Neuropathological findings regarding a minicolumnopathy in ASD was reproduced in an

independent population with investigators blind to diagnoses (5). An algorithm based on the

Delaunay triangulation was used to define the distribution of edges between image kernels

(Figure 4). The results were fitted to a bimodal distribution of interneuronal distances within and

between core cell columns. The findings indicated that in ASD the number and arrangement of

pyramidal cells within a minicolumn was preserved but minicolumnar width was reduced. The

researchers also reported a small but significant reduction in the size of pyramidal cell perikarya

that would serve to bias brain connectivity in favor of short corticocortical projections at the

expense of longer ones.

This article is protected by copyright. All rights reserved.

 Studies of minicolumnar morphometry in autism indicate a relative preservation in the
Accepted Article
size of the core compartment alongside a significant reduction of the peripheral neuropil space.

The periphery of the minicolumn is the site that provides for lateral or surround inhibition (30).

Mountcastle described this compartment as conferring a strong flow of vertical inhibition to the

minicolumn (15). Szentágothai stressed the anatomical role of this peripheral compartment by

calling it a “shower curtain of inhibition” (31). Both EEG and tactile processing studies

corroborate the presence of an abnormality of lateral inhibition in the cerebral cortex of

individuals with ASD (32, 33). Flutter stimuli capable of differentiating between active and

inactive inhibitory surrounds are now being used as potential biomarkers in the diagnosis of

autism (33).

CLINICAL SIGNIFICANCE OF NEUROPATHOLOGICAL FINDINGS

The inhibitory defect in the cerebral cortex of ASD individuals focuses around its

minicolumnar organization (30). Positive feedback without inhibitory modulation serves to drive

a system harder towards signal saturation and deformation. For most intelligent systems it is

desirable to control the gain of a signal through negative feedback rather than allowing it go

towards its extremes. A defect of lateral inhibition would enable individual minicolumns to

coalesce into islands of excitatory activity. The inhibitory deficit procreated by the

minicolumnopathy, alongside gray matter heterotopias, offers a suitable clinicopathological

correlate to both the cognitive deficits and seizures observed in many autistic individuals. A

similar imbalance in the excitatory/inhibitory bias (i.e., stochastic resonance) may account for

abnormalities in multiple sensory domains (see below) (34).

Stochastic resonance is a phenomenon where a weak signal, normally below levels of

detection, can be recognized by adding noise to the system. In the case of ASD the noise (i.e.,

signals that add to the “intelligent” input) is inherent to the system in the form of a hyperexcitable

This article is protected by copyright. All rights reserved.

 cerebral cortex (Figure 5). Under an optimal noise level small perturbations increase the
Accepted Article
sensitivity of the system. On the other hand, non-optimal noise levels decrease its sensitivity,

blurring signal from noise (e.g., the background “hiss” in communication systems). This

mechanism, already demonstrated and applied in neurophysiology (36, 37), could help explain

features of concurrent hypo- and hypersensitivity in ASD, e.g., heightened apprehension to

external stimuli and decreased registration of internal body stimuli.

The interpretation of many of the clinical findings in ASD appears related to the

previously described minicolumnar pathology (see above). This core pathological feature of

autism is amenable to intervention when considering the nature of the anatomical elements

involved, their location, and their geometrical orientation in regards to the cerebral cortex. Some

inhibitory elements within the PNS stand at direct angle to the pial surface. Axons of double

bouquet cells transverse the width of the cerebral cortex forming tightly interwoven horsetail-like

bundles. The regular spacing of double bouquet cells serves to frame the minicolumn in an

inhibitory surround. The large number of axons stemming from each cell, their length, as well as

their geometrical orientation within the cerebral cortex suggests their susceptibility to low

frequency TMS stimulation (Figure 6). The axonal ramifications of these cells would provide an

additive effect to their overall diameter thus decreasing their resistance and facilitating their

proclivity towards being induced by an external magnetic field. However, despite the apparent

susceptibility of certain cells to TMS, especially at lower frequencies, it seems possible that

stimulation results in a compound action potential comprised of individual pulses in many

different types of axons.

This article is protected by copyright. All rights reserved.

 TRANSCRANIAL MAGNETIC STIMULATION
Accepted Article
Whenever current flows through a conductor a magnetic field is generated. For a straight

wire the resultant magnetic field has no polarity and the field strength tends to be small and of

little practical use. Twisting the wire into a loop forces the flux lines closer together and adds

polarity to the magnetic field. Sliding a bar of a ferromagnetic material through the coil changes

the core’s permeability to flux lines thus greatly increasing the strength of the magnetic field

produced. A coil with a ferromagnetic core is called an electromagnet. In contrast to a

permanent magnet the field around the coil or electromagnet exists only when current flows

through the loops of wire. The strength of the magnetic field thus produced depends on the

number of turns of the coil and the amount of current flowing through the same. The

conformation of the electromagnet serves to direct the magnetic field lines to a particular foci

thereby increasing flux density and strength but decreasing or limiting its penetration. For this

reason many machines use an air core with their coiled inductor. Current research is being

directed towards testing new cores and varied shaped coils that may induce stronger magnetic

fields while improving penetration. At present, depth of penetration of the induced magnetic field

is approximately 30 mms. Since the surface of the cerebral cortex lies about 20 mms below the

skin, brain stimulation with TMS targets primarily the crest of gyri.

In transcranial magnetic stimulation (TMS) a bank of super- or ultracapacitors discharge

stored current across an inductor coil. As the capacitors discharge, a current of high magnitude,

usually several kA, passes through the stimulating coil causing at first a magnetic field to rapidly

expand and then to collapse. For practical purposes stimulation of tissue elements (alterations

in membrane potential) with a single pulse of TMS is thought to occur during the rising edge of

the applied magnetic field. The resultant magnetic field is on the order of 1 T, which is

approximately 20,000 times the Earth’s magnetic field.

This article is protected by copyright. All rights reserved.

 Electromagnetic induction is the process of producing an electromotive force (voltage)
Accepted Article
through a conductor by either moving the same through a stationary magnetic field or by leaving

the conductor still and changing the magnetic field. In TMS the process of relative motion is

caused by the expanding magnetic field, otherwise the anatomical elements acting as

conductors remain stationary. The induced voltage is influenced by the strength of the magnetic

field applied, the speed of the relative motion between the conductor and magnetic field, the

length of the conductor, and the angle at which the conductor cuts the magnetic field. The

maximum induced voltage occurs in those conductors standing at ninety degrees to the

magnetic field.

Low frequency (less than 1 Hz) is considered to have an inhibitory effect on the cerebral

cortex, for the most part (38), as it suppresses the excitability of the motor or visual cortex for

motor or phosphene thresholds. Depending on the pre-existing baseline excitability level, higher

frequencies (considered as 5 Hz and greater) are considered excitatory. At different frequencies

TMS alters the expression of immediate early gene proteins and GABAergic neurotransmission

(39, 40). It has therefore been hypothesized that slow rTMS increases the activation of inhibitory

circuits. Indeed, slow rTMS significantly increases high frequency oscillations and have no effect

on somatosensory evoked potentials over the primary somatosensory cortex. These oscillations

are a reflection of the activity of intracortical inhibitory neurons (41). The results suggest that

slow rTMS affects cortical excitability by stimulating interneurons, an effect which lasts beyond

the time of stimulation (42, 43). Hoffman and Cavus in their review of the literature regarding

slow rTMS proposed long-term depression and long-term depotentiation for its mechanisms of

action (44). Changes accrued to these mechanisms appear to accumulate in additive fashion

over repeated sessions.

This article is protected by copyright. All rights reserved.

 TRANSCRANIAL MAGNETIC STIMULATION AND AUTISM
Accepted Article
Several studies have examined the effects of low frequency (0.5 Hz–1.0 Hz) rTMS on

behavioral, EEG, and ERP outcome measures in children with ASD (45, 46). The studies have

been performed as outpatient procedures usually lasting 20 to 30 minutes. Since there may be

a risk for inducing seizures, especially in patients with a hyperexcitable cortex, our group and

others have excluded participants with seizures or a family history of the same. As an additional

precaution in rTMS studies we have adjusted the stimulation intensity (energy of the pulses)

below the patient’s motor threshold. Given the susceptibility of metal objects to the effects of a

magnetic field, individuals who use intracranial metallic pieces, pacemakers and/or other

implantable devices have been excluded from participating in the studies.

Our studies have focused on stimulating the dorsolateral prefrontal cortex (DLPC) for 6,

12, and 18 weeks. The tremendous output territory of the DLPC makes it a connection hub

within the small-world network of corticocortical connectivity. It was thought that modulating the

output of the DLPC would procreate a beneficial cascading to the many areas connected to the

same. The age range of participants in the various studies was 8-27 with a mean of 14 years. In

our first study (N = 8 children with ASD, N = 5 wait-list participants, N = 13 age-matched

controls) we measured the EEG gamma band during a visual attention task (i.e., recognizing a

Kanizsa stimuli as either an illusory figure or not) and then measured the EEG gamma band

after 6 sessions of rTMS treatment (47, 48, 49). Binding of widely distributed cell assemblies by

these high frequency oscillations or gamma frequencies (30 Hz–80 Hz) has been associated

with top-down attentional processing and object perception. In this study gamma activity in our

control group was found to increase during the presentation of target-stimuli as compared to

non-target stimuli. The power of the gamma oscillations was higher in the ASD group and had a

shorter latency as compared to controls. After six sessions of slow rTMS the power of gamma

oscillations decreased over the frontal and parietal locations (on the same side of the

This article is protected by copyright. All rights reserved.

 stimulation), and the difference between gamma responses to target and non-target stimuli
Accepted Article
significantly increased (Figure 7). Clinical assessments revealed significant improvements in

repetitive and stereotyped behaviors.

Gamma band activity has been investigated after 12 sessions of bilateral slow rTMS to

the dorsolateral prefrontal cortex (50). In these studies 16 ASD patients and 9 age-matched

controls were assessed while using Kanizsa illusory figures. Following rTMS treatment ASD

individuals showed significant improvements between relevant and irrelevant visual stimuli.

Other studies have followed using independent populations showing that rTMS significantly

improves ERP indices of selective attention, motor response errors, error detection and

correction, and both repetitive behaviors and irritability (50, 51, 52, 53). Evidence from our trials

of compromised error monitoring was deemed of importance as it may underlie an impairment

of self-monitoring which is part of the supervisory attentional system that reviews when and how

to use specific strategies and adjust the same according to environmental changes and task

demands (51, 52). Individuals with error monitoring dysfunction may thus show impairments in

set shifting abilities and inflexibility in their capacity to change goals. In schizophrenia, an

impairment in error monitoring predicts poor executive function as tested with the Penn

Conditional Exclusion test (PCET) (54). TMS may therefore prove to be an intervention that

improves executive functions in addition to behavioral performance in ASD.

The effects of rTMS have been investigated after 18 weekly session of bilateral DLPC

stimulation. This study used two groups of children with autism (TMS and wait list group, N = 27

per group) (55). Following the study sessions there was a significant decrease in amplitude and

prolonged latency in the frontal and fronto-central N100, N200, and P3a ERP components to

non-targets in the TMS group as compared to the wait-list group. These changes along with an

increase in P2d, centro-parietal P100 and P3b suggest a greater efficiency of information

processing after TMS treatment.

This article is protected by copyright. All rights reserved.

 Bilateral prefrontal rTMS has been used to examine the effects on evoked and induced
Accepted Article
EEG gamma phase coherence between frontal and parietal sites during performance on a

Kanizsa illusory figures oddball task (56). The study consisted of 18 sessions with 32

participants (TMS and wait list, 16 individuals per group). Results indicated that TMS had its

most significant effect on induced gamma in the frontal region as suggested by a significant

increase after treatment in frontal gamma coherence in response to target stimuli compared to

the coherence observed in the wait list group. TMS also increased induced gamma phase

coherence between ipsi- and contra-lateral frontal and parietal regions. The data shows that

TMS does improve gamma coherence in autism, especially in the induced gamma time frame.

The combined use of rTMS and EEG neurofeedback (i.e., the use of brain activity

parameters as feedback to regulate a function we want to control) has been used to operantly

condition post-TMS EEG changes (57). The underlying hypothesis was that combined TMS and

neurofeedback therapy would improve executive functions and behavior in the treatment group

(N = 20) as compared to the wait list group (N = 22). Results of the integrated neuromodulation

treatment supported the initial hypothesis by demonstrating significant improvements in the

behavioral and ERP measures of executive functions, as well as significant changes in EEG

outcomes of neurofeedback training such as frontal theta-to-beta ratio and an increased relative

power of gamma activity.

Heart rate variability and electrodermal activity have been used as noninvasive

measures of autonomic nervous system activity during rTMS therapy in autism (58). The study

found that an accelerated heart rate in association with lower heart rate variability (HRV)

indexed by low frequency (LF) to high frequency (HF) ratio (LF/HF of HRV, so-called cardiac

autonomic balance index) and low standard deviation of HR (SDHR) along with high

electrodermal activity (skin conductance level -SCL) were good indicators of excessive

sympathetic and reduced parasympathetic activation at the pre-treatment stage (58). Results of

This article is protected by copyright. All rights reserved.

 the study showed that, except for a reduction in LF power of HRV, all dependent HRV variables
Accepted Article
changed in the predicted way (decreased heart rate, increased standard deviation of HR,

increase of HF of HRV, decreased SCL).

Besides the DLPC other areas of the brain have been stimulated with improved

performance in behavioral tasks in ASD individuals. Fecteau and colleges applied a single

session of rTMS (1Hz) to the left and right pars triangularis and pars opercularis in 10

individuals with ASD and 10 matched neurotypicals (59). Compared to the sham controls, all

ASD individuals showed reduced latency to name objects on the Boston Naming test. A similar

study using a single session of 1 HZ rTMS was applied to the left primary motor region (M1) and

the supplementary motor area (SMA) in 11 individuals with ASD (60). Activation of the

sensorimotor system while observing another person’s actions is referred to as interpersonal

motor resonance and is considered an index of mirror neuron activity. The study reported

significant improvements in late movement related cortical potential (MRCP) following

stimulation of M1 and of the early MCRP following stimulation of the SMA.

In summary, contrary to clinical interventions that target behavioral manifestations of

ASD, TMS focuses on what may be considered a core pathology of the condition, that is, a

minicolumnopathy that manifests marked deficits of lateral inhibition. It is the only intervention

for ASD that has shown improvements in executive functions as evidenced by normalization of

ERP responses, reaction time, and accuracy during tests of executive function. Although

applied primarily to the DLPC, the effects, according to behavioral performance and

electrophysiology, are not limited to this region. Contrary to other interventions, repetitive pulse

TMS is fairly safe and offers few side effects to treatment, e.g., a transient tension type

headache, discomfort due to the sound of the pulses. Many clinical trials have been performed

This article is protected by copyright. All rights reserved.

 in children attesting to both its safety and efficacy. Future studies should address the long-term
Accepted Article
effectiveness of the technique and the possible synergism when used in conjunction with

behavioral and pharmacological therapies in larger sample sizes.

ACKNOWLEDGEMENTS

This article is based on several studies partially supported by a grant from the National Institutes

of Health (MH86784) awarded to Manuel F. Casanova.

LIST OF ABBREVIATIONS

ASD: autism spectrum disorder

DLPC: dorsolateral prefrontal cortex

EEG: electroencephalography

ERP: Event-related potential

GABA: gamma-aminobutyric acid

HF: high frequency

HRV: heart rate variability

kA: kilo-amps

LF: low frequency

M1: primary motor cortex

MRCP: movement related cortical potential

PNS: peripheral neuropil space

SCL: skin conductance level

SMA: supplementary motor area

rTMS: repetitive Transcranial Magnetic Stimulation

S1: primary somatosensory cortex

TMS: Transcranial Magnetic Stimulation

This article is protected by copyright. All rights reserved.

 REFERENCES
Accepted Article
1. Kemper TL, Bauman M. Neuropathology of infantile autism. J Neuropathol Exp Neurol 1998;

57:645–52

2. Bailey A, Luthert PJ, Dean AF, Harding B, Janota I, Montgomery M, et al. A

clinicopathological study of autism. Brain 1998; 121:889–905

3. Casanova MF, Buxhoeveden DP, Switala AE, Roy E. Neuronal density and architecture

(grey level index) in the brains of ASD patient. J Child Neurol 2002; 17:515–21

4. Casanova MF, Buxhoeveden DP, Switala AE, Roy E. Minicolumnar pathology in autism.

Neurology 2002; 58:428–32.

5. Casanova MF, Van Kooten IAJ, Switala AE, Van Engeland H, Heinsen H, Steinbusch HWM,

et al. Minicolumnar abnormalities in autism. Acta Neuropathol 2006; 112:287–303

6. Casanova MF. The neuropathology of autism. In: Volkmar FR, Rogers SJ, Paul R, Pephrey

KA, editors. Handbook of pervasive developmental disorders. 4th ed. New York: Wiley,

2014:497–531

7. Avino TA, Hutsler JJ. Abnormal cell patterning at the cortical gray-white matter boundary in

autism spectrum disorders. Brain Res 2010; 1360:138–46

8. Wegiel J, Kuchna I, Nowicki K, Imaki H, Wegiel J, Marchi E, et al. The neuropathology of

autism: defects of neurogenesis and neuronal migration, and dysplastic changes. Acta

Neuropathol 2010; 119:755–70

9. Casanova MF. The minicolumnopathy of autism. In: Buxbaum J, Hof P, editors. The

neuroscience of autism spectrum disorders. Oxford: Academic Press, 2013:327–33

10. Yakovlev PI. Pathoarchitectonic studies of cerebral malformations. III. Arrhinencephalies

(Holotelencephalies). J Neuropathol Exp Neurol 1959; 18:22–55

11. Sanides F. Functional architecture of motor and sensory cortices in primates in the light of a

new concept of neocortex evolution. In: Noback CR, Motagna W, editors. The primate brain.

New York: Appleton-Century-Crofts, 1970:137–208

This article is protected by copyright. All rights reserved.

 12. Jones EG, Powell TP. An anatomical study of converging sensory pathways within the
Accepted Article cerebral cortex of the monkey. Brain 1970; 93:793–820

13. Creutzfeldt OD. The neocortical link: thoughts on the generality of structure and function in

the neocortex. In: Brazier M, Ptesche H, editors. Architectonics of the cerebral cortex. New

York: Raven Press, 1978:367–84

14. Lorente de Nó R. The cerebral cortex: architecture, intracortical connections, and motor

projections. In: Fulton JF, editor. Physiology of the nervous system. London: Oxford

University Press, 1938:291–339

15. Mountcastle VB. Perceptual neuroscience: the cerebral cortex. Cambridge: Harvard

University Press, 1998

16. Opris I, Fuqua JL, Huettl PF, Gerhardt GA, Berger TW, Hampson RE, et al. Closing the loop

in the primate prefrontal cortex: inter-laminar processing. Front Neural Circuits 2012; 6:88

17. Song D, Opris I, Chan RH, Marmarelis VZ, Hampson RE, Deadwyler SA, et al. Functional

connectivity between layer 2/3 and layer 5 neurons in prefrontal cortex of nonhuman

primates during a delayed match-to-sample task. Proc Conf IEEE Eng Med Biol Soc 2012;

34:2555–8.

18. Hampson RE, Gerhardt GA, Marmarelis V, Song D, Opris I, Santos L, et al. Facilitation and

restoration of cognitive function in primate prefrontal cortex by a neuroprosthesis that utilizes

minicolumn-specific neural firing. J Neural Eng 2012; 9:056012.

19. Opris I, Hampson RE, Gerhardt GA, Berger TW, Deadwyler SA. Columnar processing in

primate pFC: evidence for executive control microcircuits. J Cogn Neurosci 2012; 24:2334–7

20. Favorov O, Whitsel BL. Spatial organization of the peripheral input to area 1 cell columns: I.

The detection of “segregates”. Brain Res Rev 1988; 13:25–42

21. Favorov OV, Diamond M. Demonstration of discrete place-defined columns, segregates, in

cat SI. J Comp Neurol 1990; 298:97–112

22. Tommerdahl M, Favorov OV, Whitsel BL, Nakhle B, Gonchar YA. Minicolumnar activation

This article is protected by copyright. All rights reserved.

 patterns in cat and monkey SI cortex. Cereb Cortex 1993; 3:399–411
Accepted Article
23. Ohki J, Chung S, Ch′ng YH, Kara P, reid RC. Functional imaging with cellular resolution

reveals precise micro-architecture in visual cortex. Nature 2005; 433:597–603

24. Yu YC, Bultje RS, Wang X, Shi SH. Specific synapses develop preferentially among sister

excitatory neurons in the neocortex. Nature 2009; 458:501–4

25. Casanova MF. The significance of minicolumnar size variability in autism. In: Zimmerman

AW, editor. Autism: current theories and evidence. Totowa: Humana Press, 2008:349–60

26. Casanova MF, Trippe J, Switala A. A temporal continuity to the vertical organization of the

human neocortex. Cereb Cortex 2006; 17:130–7

27. Weidenheim KM, Goodman L, Dickson DW, Gillberg C, Rastman M, Rapin I. Etiology and

pathophysiology of autistic behavior: clues from two cases with an unusual variant of

neuroaxonal dystrophy. J Child Neurol 2001; 16:809–19

28. Bauman ML, Kemper TL. Neuroanatomic observations of the brain in autism. In: Bauman

ML, Kemper TL, editors. The neurobiology of autism. Baltimore: Johns Hopkins University

Press, 1994:119–45

29. Casanova MF, El-Baz AS, Kamat SS, Dombroski BA, Khalifa E, Elnakib A, et al. Focal

cortical displasias in autism spectrum disorders. Acta Neuropathol Commun 2013; 1:67

30. Casanova MF, Buxhoeveden D, Gomez J. Disruption in the inhibitory architecture of the cell

minicolumn: Implications for autism. Neuroscientist 2003; 9:496–507

31. Szentágothai J, Arbib MA. Conceptual models of neural organization. Cambridge: MIT

Press, 1975

32. Keita L, Mottron L, Dawson M, Bertrone A. Atypical lateral connectivity: a neural basis for

altered visuospatial processing in autism. Biol Psychiatry 2011; 79:806–11, 2011.

33. Puts NAJ, Wodka EL, Tommerdahl M, Mostofsky SH, Edden RAE. Impaired tactile

processing in children with autism spectrum disorder. J Neurophysiol 2014; 111:1803–11.

34. Davis G, Plaisted-Grant K. Low endogenous neural noise in autism. Autism 2014; in press

This article is protected by copyright. All rights reserved.

 35. Gerstner W, Kistler W. Spiking neuron models: single neurons, populations, plasticity.
Accepted Article Cambridge: Cambridge University Press, 2002

36. Douglass JK, Wilkens L, Pantazelou E, Moss F. Noise enhancement of information transfer

in crayfish mechanoreceptors by stochastic resonance. Nature 1993; 365:337–40

37. Levin JE, Miller JP. Broadband neural encoding in the cricket cercal sensory system

enhanced by stochastic resonance. Nature 1996; 380:165–8

38. Caparelli EC, Backus W, Telang F, Wang GJ, Maloney T, Goldstein RZ, et al. Is 1 Hz rTMS

always inhibitory in healthy individuals? Open Neuroimag 2012; 6:69–74

39. Aydin-Abidin S, Trippe J, Funke K, Eysel UT, Benali A. High- and low-frequency repetitive

transcranial magnetic stimulation differentially activates c-Fos and zif268 protein expression

in the rat brain. Exp Brain Res 2008; 188:249–61

40. Trippe J, Mix A, Aydin-abidin S, Funke K, Benali A. Theta burst and conventional low-

frequency rTMS differentially affect GABAergic neurotransmission in the rat cortex. Exp

Brain Res 2009; 199:411–21

41. Ogawa A, Ukai S, Shinosaki K, Yamamoto M, Kawaguchi S, Ishii R, et al. Slow repetitive

transcranial magnetic stimulation increases somatosensory high-frequency oscillations in

humans. Neurosci Lett 2004; 358:193–6

42. Filipović SR, Rothwell JC, Bhatia K. Slow (1 Hz) repetitive transcranial magnetic stimulation

(rTMS) induces a sustained change in cortical excitability in patients with Parkinson’s

disease. Clin Neurophysiol 2010; 121:1129–37

43. Casula EP, Tarantino V, Basso D, Arcara G, Marino G, Toffolo GM, et al. Low-frequency

rTMS inhibitory effects in the primary motor cortex: insights from TMS-evoked potentials.

Neuroimage 2014; 98:225–32

44. Hoffman RE, Cavus I. Slow transcranial magnetic stimulation, long-term depotentiation, and

brain hyperexci ility disorders. Am J Psychiatry 2002; 159:1093–102

45. Casanova MF, Sokhadze EM. Transcranial magnetic stimulation: application in autism

This article is protected by copyright. All rights reserved.

 treatment. In: Hu VW, editor. Frontiers in autism research: new horizons for diagnosis and
Accepted Article treatment. Hackensack: World Scientific, 2014: 583–606

46. Oberman L, Rotenberg A, Pascual-Leone A. Use of transcranial magnetic stimulation in

autism spectrum disorders. J Autism Dev Disord 2013; doi: 10.1007/s10803-013-1960-2

47. Sokhadze E, El-Baz A, Baruth J, Mathai G, Sears L, Casanova MF. Effects of low frequency

repetitive transcranial magnetic stimulation (rTMS) on gamma frequency oscillations and

event-related potentials during processing of illusory figures in autism. J Autism Dev Disord

2009; 39:619–34

48. Basar E. A review of gamma oscillations in healthy subjects and in cognitive impairment. Int

J Psychophysiol 2013; 90:99–117

49. Casanova MF, Baruth J, El-Baz AS, Sokhadze GE, Hensley M, Sokhadze ES. Evoked and

induced gamma frequency oscillations in autism. In: Casanova MF, El-Baz AS, Suri JS,

editors. Imaging the brain in autism. New York: Springer, 2013:87–106

50. Casanova MF, Baruth JM, El-Baz A, Tasman A, Sears L, Sokhadze E. Repetitive

transcranial magnetic stimulation (rTMS) modulates event-related potential (ERP) indices of

attention in autism. Transl Neurosci 2012; 3:170–80

51. Sokhadze E, Baruth J, Tasman A, Sears L, Mathai G, El-Baz A, et al. Event-related potential

study of novelty processing abnormalities in autism. Appl Psychophysiol Biofeedback 2009;

34:37–51

52. Sokhadze E, Baruth J, El-Baz A, Horrell T, Sokhadze G, Carroll T, et al. Impaired error

monitoring and correction function in autism. J Neurother 2010; 14:79–95

53. Sokhadze EM, Baruth JM, Sears L, Sokhadze GE, El-Baz AS, Casanova MF. Prefrontal

neuromodulation using rTMS improves error monitoring and correction function in autism.

Appl Psychophysiol Biofeedback 2012 37:91–102

54. Silver H, Goodman C. Impairment in error monitoring predicts poor executive function in

schizophrenia patients. Schizophr Res 2007; 94:156–63

This article is protected by copyright. All rights reserved.

 55. Sokahdze EM, El-Baz AS, Sears LL, Opris I, Casanova MF. rTMS neuromodulation
Accepted Article improves electrocortical functional measures of information processing and behavioral

responses in autism. Front Syst Neurosci 2014; 8:134

56. Hensley MK, El-Baz AS, Sokhadze E, Sears L, Casanova MF. Effects of 18 session TMS

therapy on gamma coherence in autism. Psychophysiology 2014; 51:S16

57. Sokhadze EM, El-Baz AS, Tasman A, Sears LL, Wang Y, Lamina EV. Neuromodulation

integrating rTMS and neurofeedback for the treatment of autism spectrum disorder: an

exploratory study. Appl Psychophysiol Biofeedback 2014; 39:237–57

58. Casanova MF, Hensley MK, Sokhadze EM, El-Baz AS, Wang Y, Li X, et al. Effects of rTMS

on autonomic functions in autism spectrum disorder. Front Hum Neurosci 2014; 8:851

59. Fecteau S, Agosta S, Oberman L, Pascual-Leone A. Brain stimulation over Broca’s area

differentially modulates naming skills in neurotypical adults and individuals with Asperger’s

syndrome. Eur J Neurosci 2011; 34:158–64

60. Enticott PG, Rinehart NJ, Tonge BJ, Bradshaw JL, Fitzgerald PB. Repetitive transcranial

magnetic stimulation (rTMS) improves movement-related cortical potentials in autism

spectrum disorders. Brain Stimulation 2012; 5:30–7

FIGURE CAPTIONS

Figure 1. Schematic display of principal cell types and their interconnections found in the

minicolumn, the canonical microcircuit of six-layered cerebral cortex (neocortex). Ch: chandelier

cell; DB: double bouquet cell; LB: large basket cell; SB: small basket cell; SS: spiny stellate cell.

Figure 2. Interlaminar recording and MIMO stimulation model: (left panel) rasters and peri-event

histograms depict the activity of supra-granular (blue) and infra-granular (red) cell layers.

Overlay cross-correlograms show inter-laminar firing increase following the presentation of

targets compared to pre-target epoch. Recording array with the MIMO model for recording in

This article is protected by copyright. All rights reserved.

 layer 2/3 and stimulation in layer 5. Stimulation effect (on the right panel) compares the
Accepted Article
cumulative effect in single session and population tuning for MIMO stimulation (red) vs. layer5

prefrontal cortical activity (dark blue dotted line). Overall MIMO stimulation effect (red) is

significantly greater than no-stimulation and the chance level. Asterisks **p < 0.001, ANOVA.

Figure 3. Coronal section of the human brain during embryonic development. The clear

arrowheads illustrate the tangential migratory pathway of cells (future interneurons) from the

lateral and medial ganglionic eminences (LGE and MGE, respectively) to the cortical plate.

Interneurons may also originate from the retrobulbar neuroepithelium of the lateral ventricle and

from the cortex itself. Those interneurons fated for the cortex acquire a superficial and deeper

pathway in order to avoid the embryonic striatum. A set of dark arrowheads illustrates how

neuroblasts migrate out of the ventricular zone and into the cortical plate (future pyramidal cells)

following a radial pathway. An excitatory/inhibitory imbalance may result from desynchronization

of cells as they undertake their radial and tangential migrations. Because interneurons are

generated at such distant sites their migration may be susceptible to disruption from a large

variety of sources. Reprinted from Medical hypotheses, volume 83, Casanova et al., “Autism as

a sequence”, pp. 32–38, copyright 2012, with permission from Elsevier.

Figure 4. Minicolumnar fragments visible in a 35 μm thick section from Brodmann area 22 of a

98-year-old male. These were identified by our automatic image analysis methods (5). The

highlighted fragment spans laminas III and IV (from top to bottom).

Figure 5. Stochastic resonance illustrated in a model neuron, SRM0 (35). The model was

stimulated with a combination of subthreshold pulses at 25 ms intervals or 40 Hz, within the

gamma band, and a background of random postsynaptic potentials (PSP) distributed as a

Poisson process. In the absence of background, the subthreshold pulses fail to produce action

potentials in the model neuron. When noise is present at a rate of 500 PSP per second (A), the

This article is protected by copyright. All rights reserved.

 neuron spikes occasionally with the gamma-band pulses. At a higher noise rate of 1000 PSP
Accepted Article
per second (B), the neuron fires in synchrony with the 40 Hz input signal. At still greater noise,

2000 PSP per second (C), random firing drowns out the signal. Vertical bars in each panel

indicate the timing of the input pulses.

Figure 6. The principle of rTMS is that a magnetic field (B), varying in strength over time,

induces electric currents (j) parallel to the direction of B. The greatest effect will be seen in

superficial layers of the cortex and outside of sulci, since B loses strength rapidly with

increasing distance from the scalp. This part of the cerebral cortex includes double bouquet

cells, whose long, straight axon bundles synapse on the basal dendrites of pyramidal cells,

creating the inhibitory surround that isolates a particular minicolumn from its neighbors.

Reprinted from Cutting-edge therapies for autism 2010–2011 by permission of Skyhorse

Publishing, Inc.

Figure 7. rTMS modulates evoked EEG gamma frequency oscillations to target and non-target

illusory figures at frontal site F7. Following a 12 session-long TMS course, early gamma

response to non-target decreased, whereas gamma oscillation power to target illusory figures

increased. Reprinted, with permission, from Frontiers in autism research, ed. Valerie W. Hu,

copyright 2014 World Scientific Publishing.

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.

Accepted Article

This article is protected by copyright. All rights reserved.