Open Med.

2018; 13: 90-95

Research Article

Lin Zhang, Kailong Li, Chengqi Zhang, Xianlong Qi, Ning Zheng, Guangbin Wang*

Arcuate Fasciculus in Autism Spectrum Disorder

Toddlers with Language Regression
https://doi.org/10.1515/med-2018-0014
Received September 29, 2017; accepted November 13, 2017
1 Introduction
Abstract: Language regression is observed in a subset of Autism spectrum disorder (ASD) is a range of conditions
toddlers with autism spectrum disorder (ASD) as initial classified as neurodevelopmental disorders in the
symptom. However, such a phenomenon has not been American Psychiatric Association’s Diagnostic and
fully explored, partly due to the lack of definite diagnostic Statistical Manual of Mental Disorders [1]. Individuals
evaluation methods and criteria. Materials and Methods: diagnosed with autism spectrum disorder are
Fifteen toddlers with ASD exhibiting language regression characterized by deficits in social communication and
and fourteen age-matched typically developing (TD) social interaction, as well as restricted, repetitive patterns
controls underwent diffusion tensor imaging (DTI). DTI of behavior, interests and/or activities.
parameters including fractional anisotropy (FA), average Although language deficits are not observed in all ASD
fiber length (AFL), tract volume (TV) and number of individuals, Barger et al. have suggested that language
voxels (NV) were analyzed by Neuro 3D in Siemens syngo regression occurs as an initial symptom at an early age in
workstation. Subsequently, the data were analyzed by toddlers (around 2 years old) with ASD [2]. However, most
using IBM SPSS Statistics 22. Results: Compared with studies focused on children above 2 years of age during the
TD children, a significant reduction of FA along with an past two decades, and alterations of the cerebral language
increase in TV and NV was observed in ASD children with area in ASD patients at an early age (<2 years old) remain
language regression. Note that there were no significant largely unknown. Early diagnosis would minimize the
differences between ASD and TD children in AFL of the impairments of social communication development in
arcuate fasciculus (AF). Conclusions: These DTI changes children with early and appropriate intervention. Dawson
in the AF suggest that microstructural anomalies of the AF et al. demonstrated the importance of early detection of
white matter may be associated with language deficits in and early intervention in autism [3].
ASD children exhibiting language regression starting from The underlying pathogenesis of ASD remains
an early age. poorly understood. Most evidence supports the view
that alterations of brain structural and functional
Keywords: autism spectrum disorder; diffusion tensor connectivity, as well as alterations of white matter tracts,
imaging; language regression; language-related white may contribute to the underlying pathogenesis of ASD
matter tract; toddler patients [4–8]. Diffusion tensor imaging (DTI) provides
an unprecedented and quantitative capability to probe
tissue microstructures noninvasively using small water
molecules as a ubiquitous marker.
*Corresponding author:Guangbin Wang, Department of MR, Arcuate fasciculus (AF) is a white matter tract related
Shandong Medical Imaging Research Institute, Shandong to language and can be visualized by 3-dimensional (3D)
University, Jinan 250021, China, E-mail: [email protected] tractography using DTI. The quantitative evaluation of
Lin Zhang, Guangbin Wang, Department of MR, Shandong Medical
this tract using diverse DTI parameters may provide an
Imaging Research Institute, Shandong University, Jinan 250021,
China indicator for early diagnosis of ASD in critical periods
Lin Zhang, Xianlong Qi, Ning Zheng, Department of Radiology, during childhood. Furthermore, social developmental
Jining No.1 People’s Hospital, Jining 272011, China skills can be attained through the early delivered
Kailong Li, Department of Ultrasound Diagnosis, Affiliated Hospital therapeutic intervention.
of Jining Medical University, Jining 272029, China
In this study, we examined the variations of the
Chengqi Zhang, Department of Medical Imaging, Qianfoshan Hospi-
tal, Affiliated to Shandong University, Jinan 250014, China
language-related white matter tract, arcuate fasciculus

Open Access. © 2018 Lin Zhang et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial-
NoDerivs 4.0 License.
91   L. Zhang, et al.

(AF), between ASD toddlers and the typically developing 3-dimensional magnetization-prepared rapid acquisition
(TD) control group using DTI tractography. gradient-echo (T₁3D MP-RAGE) sequence covering the
whole brain with TR = 1900 ms, TE = 2.5 ms, FOV = 250

2 Materials and Methods mm, slices = 176, slice thickness = 1.0 mm, and bandwidth
= 170. For DTI acquisition, a single-shot echo planar (SE-
EPI) sequence was applied with TR = 5500 ms, TE = 92 ms,
2.1 Subjects FOV = 260 mm, matrix = 128 x 128, 20 diffusion encoding
directions, slice thickness = 3.0 mm, and variable b-values
The study was conducted from September 2015 to between 0 and 1000 s/mm2. Note that each subject was
November 2016 at Jining No.1 People’s Hospital (Jining, sedated using chloral hydrate (0.5 g in 10 ml) during MRI
Shangdong Province, China). Participants were 29 scanning with parental consent.
toddlers (15 diagnosed with ASD and 14 TD individuals)
with age, gender and handedness matched. Toddlers with
ASD who presented language regression were recruited 2.3 Data analysis
through the local autism organization. The diagnosis of
ASD was assessed by two experienced neuropsychologists After T₁3D MP-RAGE acquisition, we excluded one toddler
and a clinical psychologist using the Autism Behavior diagnosed with ASD who had gray matter heterotopias.
Checklist (ABC). All of them had a score of at least 30 on We also excluded one toddler with leukomalacia from the
the ABC tests. In addition, all those toddlers had a history control group. In addition, one TD toddler was excluded
of regression, as measured by the operational definition, due to failure in fusing the images between the two
which was previously described by Richler et al [9]. Based acquisitions. T₁3D MP-RAGE also assisted with the precise
on retrospective parental reports, those toddlers had localization of tractography for DTI using image fusion
spontaneously used at least three meaningful words on techniques (Fig. 1 and Fig. 2A).
daily basis for at least 1 month (during the first 1–2 years For tractography, post-processing was performed
after birth), and subsequently stopped using all words using Neuro 3D in the Siemens Syngo Workstation, which
for at least 1 month. The patients with such an abrupt or automatically detects artifacts and corrects eddy current
gradual loss of previously acquired language skills had an deformations. Tractography of the AF was performed by
age range from 1.42 to 3.25 years old. two raters who were blinded to the participants 1-3 [10, 11].
The assessment of language function was based on To access the fiber tracking of AF separately (Fig. 2), a seed
Expressive Language (25 items) and Receptive Language region of interest (ROI) was first defined in the middle
(19 items) subtests of the Chinese Version of the of the posterior limb of the internal capsule (Fig. 1A) by
Psychoeducational Profile (C-PEP). The mean IQs of the tracking the AF on the directionally encoded color maps,
ADS and TD children were obtained using the Wechsler and then a target ROI was defined at the splenium of the
Intelligence Scale for Children (WISC). The age-equivalent corpus callosum (Fig. 1B).
scores assessed with the C-PEP and WISC are presented in Subsequently, fractional anisotropy (FA), average
Table 1. TD children had a score of 90 or above on percentile fiber length (AFL), tract volume (TV) and number of voxels
ranks of expressive language (EL) and receptive language (NV) were calculated to evaluate the tract properties.
(RL) as an inclusion criteria. Participants were excluded Given that language function is hemispherically
if they had neurodevelopmental disorder, history of head lateralized and which is related to handedness, we only
trauma, or other severe medical problems. analyzed parameters of the left AF of these right-handed
The study protocol was approved by the Institutional individuals [12].
Review Board of the Jining No.1 People’s Hospital. Signed Statistical analysis was performed using IBM SPSS v22
informed consent was obtained from parents of the statistical software (IBM SPSS, Chicago, IL). Independent
participating children. sample t-tests and Pearson Chi-squared test were used to
compare age, gender, IQ and C-PEP scores respectively.
Inter-group differences in the properties of AF were
2.2 MRI protocols analyzed with analysis of variance (ANOVA). In addition,
analysis of covariance (ANCOVA) was applied to compare
All magnetic resonance imaging (MRI) scans were acquired the AF measurements between the two groups while
with a 3T Siemens TIM Trio scanner (Siemens AG, Munich, controlling for age. Results were considered significant
Germany). T1-weighted images were acquired using a when P< 0.05.
  AF in ASD with language regression   92

Figure 1. Tracking the AF: (A) The seed ROI is placed in the intense triangular-shaped green structure (circled) in the coronal plane;
(B) A target ROI is placed in the blue structure lateral to the splenium of the corpus callosum in the transverse plane (circled)

Figure 2. (A) 3-dimensional (3D) tract reconstruction of AF is superimposed on a sagittal T1 volume. (B) 3D render tract

3 Results EL and PL. The ANOVA revealed that the ASD group had
a significantly lower FA (F = 6.726, P< 0.05), as well as
As shown in Table 1, there were no significant differences significantly higher TV and NV compared to the TD group,
in age, gender or performance intelligence quotient (PIQ) as shown in Table 2. Covarying for age enhanced the
between the two groups. However, ASD group had lower significance of TV and NV. ANOVA for AFL did not reveal
scores than TD group in verbal intelligence quotient (VIQ), any significant differences between the two groups.
93   L. Zhang, et al.

Table 1. Participant characteristics for autism spectrum disorder (ASD) toddlers and typically developing (TD) controls
Characteristics ASD (n=14) TD (n=12) p value

Gender (male/female) 10/4 8/4 0.56

Age (mean) 2.39±0.32 2.41±0.77 0.93
Age (range) 1.42–3.25 1.25–3.83
C-PEP EL 30.50±10.39 94.00±3.10 <0.001
RL 33.79±15.34 93.83±3.10 <0.001
VIQ 81.21±18.96 104.33±17.53 0.004
PIQ 95.00±14.06 103.25±13.16 0.14

Abbreviations: C-PEP: Chinese Version of the Psychoeducational Profile; EL: expressive language; RL: receptive language; VIQ: Verbal Intelli-
gence Quotient; PIQ: Performance Intelligence Quotient.

Table 2. FA, AFL, TV, and NV of AF comparison between ASD toddlers and typically developing controls
ASD (n=14) TD (n=12) ANOVA ANOVAa

Mean SD Mean SD F p value F p value

Range Range

FA 376.11 18.79 401.01 29.72 6.726 0.016 6.835 0.016

365.25–386.95 382.12–419.89
AFL 87.52 10.77 78.61 15.91 2.867 0.100 3.299 0.082
81.30–93.74 68.50–88.72
TV 5206.69 2172.40 2997.90 1775.38 7.876 0.010 9.512 0.005
3950.38–6460.99 1870.37–4126.42
NV 322.71 134.57 186.33 110.33 7.810 0.010 9.451 0.005
245.01–400.42 116.23–256.44

Abbreviations: FA: fractional anisotropy; AFL: average fiber length; TV: tract volume; NV: number of voxels; AF: arcuate fasciculus; ASD:
autism spectrum disorders; TD: typically developing controls; SD: standard deviation; ANOVA: analysis of variance; ANCOVA: analysis of
covariance. ANOVAa: ANOVA completed using age as a covariate.

4 Discussion White matter (WM) tracts form the structural

foundation for brain connectivity by linking discrete grey
matter regions into integrated neural circuits [20, 21]. In
A recent meta-analysis revealed that the prevalence rate
other words, disruption of the WM tracts that mediate
for regression in ASD was approximately 32% Confidence
connectivity within neural networks could play an
Interval (CI) [30–35] (language regression, 24.9%; language/
important pathogenic role contributing to abnormal brain
social regression, 38.1%), and the average age at which
maturation in ASD.
regression occurred was about 2 years of age; 95% CI [1.7–1.9]
The AF is a WM tract of great importance to language.
[2]. However, there is no consensus regarding the formal
This WM tract connects the frontal expressive language
diagnosis of regression, and not all ASD patients suffer with
area (Broca area) with the posterior temporoparietal
language deficits. Thus, language regression occurring in
receptive language areas (Wernicke area) [22]. We
the early age of ASD can often be overlooked and toddlers
quantitatively evaluated the properties of AF using multi-
with ASD may miss early detection and early intervention
parametric DTI in vivo. DTI is a non-invasive technique
[3,13–17]. We examined the variations of the language-related
capable of delineating WM tracts and providing indirect
white matter tract, arcuate fasciculus (AF), between ASD
quantitative measures of WM integrity by measuring
toddlers and typically developing (TD) group.
water diffusion in the underlying tissue microstructure.
The mechanisms of developmental regression in
Previous DTI studies in ASD generally report higher
ASD remain largely unknown. Thomas et al. reported
diffusivity and lower FA in multiple brain areas [23–25],
that regression is triggered by overaggressive synaptic
such as the corpus callosum [26–28], superior longitudinal
pruning[18]. Similar to the Weak Central Coherence
fasciculus [28–30], internal capsule [31] and frontoparietal
Theory devised by Frith, the symptoms of ASD have
white matter [31, 32]. Our results demonstrated toddlers
been hypothesized to be caused by alterations in brain
with ASD exhibit significantly lower FA in the left AF.
connectivity[19].
Although the nature of AF abnormalities remains to
  AF in ASD with language regression   94

be elucidated, we speculate that the leading causes of 5 Conclusion

the lower FA values are associated with the reduced
myelination in nerve pathways, decreased axonal density The AF of ASD toddlers had significantly lower fractional
(a decrease in the number of axons in the AF with an anisotropy as well as significantly higher tract volume and
increase in intra-axonal space), impaired axonal integrity number of voxels when compared to typically developing
and organization, a decrease in the organization of fibers, controls. Since AF is associated with language, our results
and/or an increase in tortuosity. suggested that language deficits occur at an early age in
A longitudinal DTI study [33] and several post-mortem ASD patients.
studies [34–36] found an early overabundance of thin
axons (representing short-range connections) secondary Ethics approval and consent to participate: The
to suboptimal refinement which can be related to higher TV protocol was approved by the Institutional Review Board
and NV. In addition, a review by Travers et al. (containing of the Jining No.1 People’s Hospital. All subjects agreed to
10 ROI studies and 26 VBA studies investigating AF) participate in this study. Written informed consent was
indicated that the results of these studies have been in obtained from their legal guardians.
consistent [37]. Although most studies suggest that FA is
decreasing, there are still some exceptions reported. For Availability of data and materials: The dataset
example, Lee et al. did not find any significant group FA supporting the results of this article are included within
difference in AF [38]. Additionally, Solso et al. found that the article. Additional clinical data are available from the
there was an increased FA in frontal tracts in ASD toddlers Jining Rehabilitation Center for Autism and neuroimaging
[39]. The discrepancies between these studies may be data are available from the corresponding author.
related to the complexity of fiber geometry and crossing
in AF, or differing participant characteristics such as Competing interests: The authors declare that they have
language ability and/or age. no competing interests.
To minimize the effects of these issues, we tracked
the AF by using the ROI technique for DTI and assessed Authors’ contributions: GBW, CQZ, and LZ were
it separately to eliminate the interference of the crossing responsible for the study’s design. LZ contributed to the
fibers along this WM tract. In addition, it is generally acquisition of neuroimaging sequences, conducted the
known that WM develops dynamically and changes across clinical data collection, and wrote the first draft of the
the lifespan, with myelination continuing from childhood manuscript. LZ, XLQ, NZ analyzed the neuroimaging
in adolescence. To reduce the influence of developmental data. ZL, KOL were responsible for statistical analysis.
trajectories of the brains, our study limited participants to GBW, CHZ assisted in revising the overall manuscript. All
those aged 2–4 years. authors read and approved the final manuscript.
Regarding study limitations, our study lacked
internationally recognized diagnosis criteria for Acknowledgments: The authors thank the patients
regression. Furthermore, our findings were also mixed that contributed to this study. We also thank the Jining
depending on the measurements by parents and Rehabilitation Center for Autism for providing the
clinicians; for example, the disappearance of words that patients’ materials.
the child had previously established is measured by their
parents. In addition, other cognitive functions were not
taken into consideration and were not well controlled.
References
Thus, future studies with larger and more diverse samples [1] Joyce-Beaulieu D., Sulkowski M.L. The Diagnostic and
are needed to verify these findings. The influences of the Statistical Manual of Mental Disorders: Fifth Edition (DSM-5)
toddlers with language retardation may be considered Model of Impairment. In: Goldstein S., Naglieri J. (eds)
another limitation in our study. Assessing Impairment 2016. Springer, Boston, MA.
Future research should investigate the underlying [2] Barger BD, Campbell JM, McDonough JD. Prevalenceand
onset of regression within autism spectrum disorders: a
mechanisms for AF microstructural alterations
meta-analytic review. J Autism and Dev Disord 43, 2013:
associated with delayed language development and ASD. 817–828.
Notwithstanding these limitations, our study adds to the [3] Dawson G, Rogers S, Munson J, et al. Randomized, controlled
few existing studies on the phenomenon of regression trial of an Intervention for toddlers with autism: the early start
during the early ages of children with ASD. denver model. Pediatrics 2010, 125: e17–23.
95   L. Zhang, et al.

[4] Fletcher PT, Whitaker RT, Tao R, et al.Microstructural [24] Groen WB, Buitelaar JK, van der Gaag RJ, et al. Pervasive
Connectivity of the arcuate fasciculus in adolescents with microstructural abnormalities in autism: A DTI study. J
high-functioning autism. Neuroimage 2010, 51: 1117–1125. Psychiatry Neuroscience 2011, 36: 32–40.
[5] Ameis SH, Catani M. Altered white matter connnectivity as a [25] Jou RJ, Jackowski AP, Papademetris X, et al. Diffusion tensor
neural substrate for social impairment in autism spectrum imaging in autism spectrum disorders: Preliminary evidence
disorder. Cortex 2015 62: 158–181. of abnormal neural connectivity. Aust N Z J Psychiatry 2011, 45:
[6] Ha S, Sohn IJ, Kim N, et al. Characteristics of brains in autism 153–162.
spectrum disorder: structure, function and connectivity across [26] Hanaie R, Mohri I, Kagitani-Shimono K, Tachibana M, et al.
the lifespan. Exp Neurobiol 2015, 24: 273–284. Abnormal corpus callosum connectivity, socio-communicative
[7] Shih P, Shen M, Ottl B, et al. Atypical network connectivity for deficits, and motor deficits in children with autism spectrum
imitation in autism spectrum disorder. Neuropsychologia 2015, disorder: a diffusion tensor imaging study. J Autism Dev Disord
48: 2931–2939. 2014, 44: 2209–2220.
[8] Schipul SE, Keller TA, Just MA. Inter-regional brain [27] Kumar A, Sundaram SK, Sivaswamy L, et al. Alterations in
communication and its disturbance in autism. Front Syst frontal lobe tracts and corpus callosum in young children
Neurosci 2011, 5: 1-10. with autism spectrum disorder. Cerebral Cortex 2010, 20:
[9] Richler J, Luyster R, Hsu WL. Is there a “regressive phenotype” 2103–2113.
of Autism Spectrum Disorder associated with the measles- [28] Noriuchi M, Kikuchi Y, Yoshiura T, et al. Altered white matter
mumps-rubella vaccine? A CPEA Study. J Autism Dev Disord fractional anisotropy and social impairment in children with
2006, 36:299–316. autism spectrum disorder. Brain Res 2010, 1362: 141–149.
[10] Wakana S, Caprihan A, Panzenboeck MM, et al. Reproducibility [29] Nagae LM, Zarnow DM, Blaskey L, et al. Elevated mean
of quantitative tractography methods applied to cerebral white diffusivity in the left hemisphere superior longitudinal
matter. Neuroimage 2007, 36: 630–644. fasciculus in autism spectrum disorders increases with more
[11] Smits M, Jiskoot LC, Papma JM. White matter tracts of speech profound language impairment. AJNR Am J Neuroradiol 2012,
and language. Semin ultrasound CT MR 2014, 35: 504–516. 33: 1720–1725.
[12] Lurito J, Dzemidzic M. Determination of cerebral hemisphere [30] Jeong JW, Kumar AK, Sundaram SK, et al. Sharp curvature of
language dominance with functional magnetic resonance frontal lobe white matter pathways in children with autism
imaging. Neuroimaging Clin N Am 2001, 11: 355–363. spectrum disorders: Tract-based morphometry analysis. AJNR.
[13] Lainhart JE, Ozonoff S, Coon H, et al. Autism, regression, and Am J Neuroradiol 2011, 32: 1600–1606.
the broader autism phenotype. Am J Med Genet 2002, 113: [31] Ogur T, Boyunaga OL. Relation of behavior problems with
231–237. findings of cranial diffusion tensor MRI and MR spectroscopy in
[14] Tsai PT. Autism and cerebellar dysfunction: Evidence from autistic children. Int J Clin Exp Med 2015, 8: 5621–5630.
animal models. Semin Fetal Neonatal Med 2016, 21: 349–355. [32] Poustka L, Jennen-Steinmetz C, Henze R, et al. Fronto-temporal
[15] Nelson CA, Zeanah CH, Fox NA, et al. Cognitive recovery disconnectivity and symptom severity in children with autism
in socially deprived young children: the Bucharest Early spectrum disorder. World J Biol Psychiatry 2012, 13: 269–280.
Intervention Project. Science2007, 318: 1937–1940. [33] Wolff JJ, Gu H, Gerig G, et al. Differences in white matter fiber
[16] Howlin P, Magiati I, Charman T. Systematic review of early tract development present from 6 to 24 months in infants with
intensive behavioral interventions for children with autism. Am autism. Am J Psychiatry 2012, 169: 589–600.
J Intellect Dev Disabil 2009, 114: 23–41. [34] Zikopoulos B, Barbas H. Changes in prefrontal axons
[17] Tager-Flusberg H, Kasari C. Minimally verbal school-aged may disrupt the network in autism. J Neurosci 2010, 30:
children with autism spectrumdisorder: the neglected end of 14595–14609.
the spectrum. Autism Res 2013, 6: 468–478. [35] van Kooten IA, Palmen SJ, von Cappeln P, et al. Neurons in the
[18] Thomas MS, Knowland VC, Karmiloff-Smith A. Mechanisms fusiform gyrus are fewer and smaller in autism. Brain 2008,
of developmental regression in autism and the broader 131: 987–999.
phenotype: a neural network modeling approach. Psychol Rev [36] Bauman ML, Kemper TL. Neuroanatomic observations of
2011, 118: 637–654. the brain in autism: a review and future directions. Int J Dev
[19] Frith U, Happe F. Autism: beyond ‘‘theory of mind’’. Cognition Neurosci 2005, 23: 183–187.
1994, 50: 115–132. [37] Travers BG, Adluru N, Ennis C, et a. Diffusion tensor imaging
[20] Fields RD. White matter matters. Sci Am 2008, 298: 42–49 in autismspectrum disorder: a review. Autism Res 2012, 5:
[21] Schmahmann J D, Pandya, DN, Wang R, et al. Association fibre 289–313.
pathways of the brain: parallel observations from diffusion [38] Lee JE, Chung MK, Lazar M, et al. A study of diffusion tensor
spectrum imaging and autoradiography. Brain 2007, 130: imaging by tissue-specific, smoothing-compensated
630–653. voxel-based analysis. Neuroimage 2009, 44: 870–883.
[22] Catani M, Jones DK. Perisylvian language networks of the [39] Solso S, Xu R, Proudfoot J, et al. Diffusion tensor imaging
human brain. Ann Neurol 2005, 57: 8–16. provides evidence of possible axonal overconnectivity in frontal
[23] Shukla DK, Keehn B, Lincoln AJ, et al. White matter compromise lobes in autism spectrum disorder toddlers. Biol Psychiatry
of callosal and subcortical fiber tracts in children with autism 2016, 79: 676–684.
spectrum disorder: a diffusion tensor imaging study. J Am Acad
Child Adolesc Psychiatry 2010, 49: 1269–1278.