Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Autism and Sensory Processing Disorders: Shared White Matter Disruption in Sensory Pathways but Divergent Connectivity in Social-Emotional Pathways

Autism and Sensory Processing Disorders: Shared White Matter Disruption in Sensory Pathways but... Over 90% of children with Autism Spectrum Disorders (ASD) demonstrate atypical sensory behaviors. In fact, hyper- or hyporeactivity to sensory input or unusual interest in sensory aspects of the environment is now included in the DSM-5 diagnostic criteria. However, there are children with sensory processing differences who do not meet an ASD diagnosis but do show atypical sensory behaviors to the same or greater degree as ASD children. We previously demonstrated that children with Sensory Processing Disorders (SPD) have impaired white matter microstructure, and that this white matter microstructural pathology correlates with atypical sensory behavior. In this study, we use diffusion tensor imaging (DTI) fiber tractography to evaluate the structural connectivity of specific white matter tracts in boys with ASD (n = 15) and boys with SPD (n = 16), relative to typically developing children (n = 23). We define white matter tracts using probabilistic streamline tractography and assess the strength of tract connectivity using mean fractional anisotropy. Both the SPD and ASD cohorts demonstrate decreased connectivity relative to controls in parieto-occipital tracts involved in sensory perception and multisensory integration. However, the ASD group alone shows impaired connectivity, relative to controls, in temporal tracts thought to subserve social-emotional processing. In addition to these group difference analyses, we take a dimensional approach to assessing the relationship between white matter connectivity and participant function. These correlational analyses reveal significant associations of white matter connectivity with auditory processing, working memory, social skills, and inattention across our three study groups. These findings help elucidate the roles of specific neural circuits in neurodevelopmental disorders, and begin to explore the dimensional relationship between critical cognitive functions and structural connectivity across affected and unaffected children. Citation: Chang Y-S, Owen JP, Desai SS, Hill SS, Arnett AB, et al. (2014) Autism and Sensory Processing Disorders: Shared White Matter Disruption in Sensory Pathways but Divergent Connectivity in Social-Emotional Pathways. PLoS ONE 9(7): e103038. doi:10.1371/journal.pone.0103038 Editor: Christophe Lenglet, University of Minnesota, United States of America Received December 5, 2013; Accepted June 25, 2014; Published July 30, 2014 Copyright:  2014 Chang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was funded by grants from the Wallace Research Foundation, the Gates Family Foundation and the Holcombe Kawaja Family Foundation. EJM, JPO and PM acknowledge support from the Simons Foundation. PM also acknowledges support from NIH R01 NS060776. EJM has received support from NIH K23 MH083890 and KL2 RR024130. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interest exist. * Email: marcoe@neuropeds.ucsf.edu " These authors are co-first authors on this work. and behavioral flexibility [5]. However, individuals with ASD have Introduction also been shown to have ubiquitous challenges in sensory The human brain is a sensory processor. Its core function is to processing [6] with over 90% of children with autism reported perceive, integrate, interpret, and then facilitate the appropriate to have atypical sensory related behaviors. In fact, hyper- or coordinated response to the visual, tactile, auditory, olfactory, and hyporeactivity to sensory input or unusual interest in sensory proprioceptive information present in the world around us. Thus it aspects of the environment is now included in the current DSM 5 comes as no surprise that inaccurate or imprecise sensory diagnostic criteria for ASD [6]. There are, however, children with processing and multisensory integration (MSI) can lead to sensory processing disorders (SPD) who do not show primary impaired intellectual and social development [1]–[4]. There is a language or social deficits but do exhibit atypical sensory reactivity growing recognition of the crucial importance of sensory and/or sensory interests to the same or greater extent as children processing as it contributes to attention, learning, emotional who meet an ASD diagnosis [1]. Children with SPD remain regulation, and even social function in children affected by a wide critically underserved with regard to their developmental chal- spectrum of neurodevelopmental disorders, including autism. lenges in our society due to the lack of a diagnostic label There is also a growing interest in studying sensory processing recognized in the current DSM 5 manual. Many are instead and cognition as dimensional traits across typically developing attributed labels that better describe the sequelae of SPD, such as children and those with psychiatric labels such as autism. oppositional defiant disorder, than the root of the problem. It is Autism spectrum disorders (ASD) have traditionally been therefore highly relevant to better characterize the biological bases characterized by impaired communication, social interaction, PLOS ONE | www.plosone.org 1 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 1. Cognitive Characterization of TDC, ASD, and SPD Cohorts. TDC Mean 6 Std ASD Mean 6 Std Pval SPD Mean 6 Std Pval PRI 113.5613.5 101.6614.1 0.015 115.8611.5 0.576 VCI 119.2612.7 101.6620.5 0.007 117.4612.8 0.660 WMI 108.4610.9 99.6617.7 0.111 104.4612.8 0.320 PSI 101.3613.6 87.4611.1 0.002 97.1612.9 0.334 PRIs, VCIs, WMIs and PSIs for each cohort, with p values from two-tailed t-tests for differences between TDCs and each patient cohort (statistically significant p values of less than 0,05 are indicated in boldface). doi:10.1371/journal.pone.0103038.t001 of this increasingly recognized neurodevelopmental condition. In working memory, the latter of which has been proposed to be addition, the comparison of children with SPD and ASD may help mediated by stereotypical time-locked spatiotemporal spike timing to illuminate the unique neural mechanisms at the core of the ASD patterns [17]. diagnosis: those facilitating social awareness, interest, and drive. In this study, we examine white matter tracts that we With over 1% of children in the USA carrying an ASD label and hypothesize will be atypical in children with SPD or ASD subjects reports of 5–16% of children in the USA having sensory relative to typically developing children (TDC). Based upon our processing difficulties, it is important to define the neural previous work on white matter microstructure in SPD [16], and underpinnings of these conditions and to delineate the areas of upon previous studies of white matter microstructure in ASD, we overlap and the areas of divergence [1], [2], [7]. The advent of posit that both ASD and SPD subjects will have reduced structural diffusion tensor imaging (DTI) and fiber tractography has enabled connectivity compared to controls in parieto-occipital white matter quantitative, noninvasive evaluation of white matter microstruc- tracts involved with sensory processing and integration, whereas ture and connectivity. There is considerable, albeit contradictory, only ASD subjects will have diminished structural connectivity literature reporting altered structural connectivity in individuals relative to controls in temporal tracts associated with social- with ASD using DTI [8]. There are several studies suggesting emotional processing. Furthermore, we posit that tract connectiv- reduced connectivity via the corpus callosum [9]–[11] as well as ity will correlate with measures reflecting sensory processing, others indicating normal or even elevated fractional anisotropy inattention behavior, social behavior, verbal comprehension, (FA), a measure of white matter tract microstructural integrity processing speed, and working memory across groups. from DTI [12]. Beyond the corpus callosum, there are also reports of other white matter tracts that may show variance from typically Methods developing controls, including the inferior fronto-occipital fascic- The Institutional Review Board (IRB) at the University of ulus (IFOF) and the uncinate fasciculus (UF). A recent meta- California in San Francisco approved this study (UCSF IRB analysis of 25 DTI studies in individuals with autism reports Protocol #: 10-01940). Subjects were recruited from the UCSF decreased FA in the corpus callosum, the left UF, and the left Autism and Neurodevelopment Program clinical sites and superior longitudinal fasciculus (SLF), supporting the theory of research database, and from local online parent board listings. specific underconnectivity in autism focused on tracts supporting Informed consent was obtained from the parents or legal auditory information and language processing [13]. Finally, in guardians, with the assent of all participants. addition to auditory and language related tracts, there is considerable interest in tracts that mediate emotional face recognition, a pervasive deficit in children with autism. DTI 2.1. Demographic, sensory, cognitive and behavioral data studies have specifically investigated the fusiform-hippocampal 2.1.1. General demographics. Sixteen right-handed males and fusiform-amygdala tracts in individuals with autism and have with SPD, fifteen males with ASD (12 right-handed, 1 left-handed, reported variation thought to relate to atypical function [14], [15]. 2 ambidextrous), and 23 right-handed male TDC, all between 8 In comparison to DTI studies of ASD, investigation of structural and 12 years of age, were prospectively enrolled under our IRB connectivity in children with isolated SPD is in its infancy. We protocol. recently reported that, although children with SPD do not exhibit Voxel-based analysis of the DTI data from the 16 SPD subjects morphological abnormalities from structural MR imaging, they and the 23 TDC using tract-based spatial statistics (TBSS) to have strikingly decreased white matter microstructural integrity, investigate white matter microstructure was previously reported in especially in posterior cerebral regions [16]. These regions are [16]. Group differences in the TBSS analysis were determined in a implicated in unimodal sensory processing as well as MSI, and are common atlas space after inter-subject image registration. In the regulated by top-down attention modulation via thalamic projec- present study, we examine white matter connectivity using tions. We further showed that white matter connectivity correlates diffusion fiber tractography in each subject’s native space, with with behavioral measures of unimodal sensory behavior, multi- the addition of an ASD cohort. sensory integration, and inattention. White matter microstructural 2.1.2. General cognition. All subjects were assessed with the integrity is crucial to the speed and bandwidth of information Wechsler Intelligence Scale for Children-Fourth Edition [18] and transmission throughout the brain. Degraded connectivity of were required to have a Perceptual Reasoning Index (PRI) score primary sensory cerebral tracts or of pathways connecting $70. We used PRI as our measure of cognition for inclusion, as multimodal sensory association areas may thereby result in the communication deficits are part of the core diagnosis of ASD. loss of the precise timing of action potential propagation needed Verbal Comprehension Index (VCI), Processing Speed Index for accurate sensory registration and integration. These effects (PSI), and Working Memory Index (WMI) were also obtained may be reflected in assessable metrics such as processing speed and PLOS ONE | www.plosone.org 2 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 2. Sensory Profile Characterization of TDC, ASD, and SPD cohorts. TDC Mean 6Std ASD Mean 6Std SPD Mean 6Std Auditory 33.663.5 *24.465.9 *22.764.9 Tactile 83.365.8 72.468.6 *62.9+8.8 Visual 41.263.0 35.666.3 32.367.1 Inattention 28.763.6 *20.364.4 *17.865.3 Total 172.3611.0 *135.1618.2 *128.5615.8 Multisensory 31.363.1 23.764.5 22.263.7 Asterisks indicate mean scores that fall within the definite difference range. None of the mean scores fell in the probable difference range. doi:10.1371/journal.pone.0103038.t002 from this assessment. These measures are displayed in Table 1 for Diagnostic Inventory-Revised (ADI-R) [21], a parent history each cohort. interview, and the Autism Diagnostic Observation Schedule 2.1.3. Sensory processing evaluation. All subjects were (ADOS) [22], a structured play session. We used current evaluated with the Sensory Profile [19], which is currently the diagnostic scoring for the ADOS and lifetime scoring for the most widely used parent report measure of atypical sensory related ADI-R. None of the TDC cohort had an SCQ score $15. All behavior. The Sensory Profile (SP) is a caregiver report participants in the ASD cohort met criteria on both the ADI-R questionnaire (125 items) which measures behavioral sensory and ADOS; all but one scored $15 on the SCQ. differences, yielding scores within individual sensory domains and Three of the SPD cohort scored above 15 on the SCQ and were factors as well as a total score. A probable difference (PD) in further evaluated with the ADI-R and ADOS. One SPD sensory behavior is defined as a total score between 142 and 154, participant scored above the ASD cutoff on the current diagnosis while a definite difference (DD) is a score of #141. Lower scores scoring of the ADOS but did not meet criteria on the ADI-R. reflect more atypical behavior. We use the auditory, visual, tactile, Another SPD individual met criteria on the ADI-R but not the multisensory integration, and inattention/distractibility scores to ADOS. Neither was considered to meet clinical criteria when explore behavioral correlations based on findings from our prior evaluated by a cognitive behavioral child neurologist with report [16]. expertise in autism and neurodevelopment (EJM). The third Inclusion in the SPD group required a community based SPD participant who scored above 15 on the SCQ met neither the Occupational Therapy diagnosis of Sensory Processing Disorder ADI-R nor ADOS cut-off. A supplementary analysis was plus a score in the definite difference (DD) range, defined as performed, excluding these three SPD subjects from the study greater than two standard deviations from the mean, of either the cohort (Table S2). total or the auditory processing score of the Sensory Profile. Five of 2.1.5. Attention deficits. On the inattention/distractibility the SPD subjects scored in the DD range for total score alone, four factor of the Sensory Profile, eleven of the 16 SPD subjects scored scored in the DD range for the auditory processing score alone, in the definite difference range, four in the probable difference and seven scored in the DD range for both the total and auditory range, and one in the typical range. Of the 15 ASD subjects, seven score. Two ASD subjects scored in the DD range for the total scored in the definite difference range, five scored in the probable score alone, one ASD subject scored in the DD range for the difference range, two scored in the typical range, and one was not auditory score alone, and seven of the ASD subjects scored in the administered the Sensory Profile. Of the 23 TDC, none scored in DD range for both the total and auditory score. The sensory the definite difference range, three in the probable difference profile was not obtained for one ASD individual. All of the range, and twenty in the typical range. Atypical inattention/ controls scored in the normal range (Table 2). distractibility scores on the Sensory Profile do not necessarily 2.1.4. Autism evaluation. All subjects were evaluated with indicate that individuals would meet clinical criteria for an the Social Communication Questionnaire (SCQ), a parent report attention deficit (hyperactivity) disorder (ADHD) diagnosis. ASD screening instrument [20]. All of the ASD cohort (carrying Formal ADHD evaluations were not conducted as part of this community diagnosis of ASD) as well as the SPD individuals with study. a score above threshold ($15) were evaluated with the Autism Table 3. Tractographical approach for temporal tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Fusiform - amygdala Fusiform gyrus Amygdala All other gm regions Fusiform - hippocampus Fusiform gyrus Hippocampus All other gm regions Uncinate fasciculus (UF) Orbitofrontal cortex* Entorhinal cortex + temporal pole All other gm regions Inferior longitudinal fasciculus (ILF) Pericalcarine cortex Inferior temporal cortex Thalamus + all other cortical regions Inferior frontooccipital fasciculus (IFOF) Lingual gyrus Orbitofrontal cortex* Thalamus + all other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined temporal tracts. *Orbitofrontal cortex was created by summing the medial orbitofrontal cortex and lateral orbitofrontal cortex. doi:10.1371/journal.pone.0103038.t003 PLOS ONE | www.plosone.org 3 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 4. Tractographical approach for parieto-occipital tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Optic radiation Pericalcarine cortex Eroded thalamus All other cortical regions Dorsal visual stream Pericalcarine cortex Inferior parietal cortex Thalamus Splenium of the corpus callosum Left lateral occipital cortex Right lateral occipital cortex* All other cortical regions Posterior corona radiata (PCR) (occipital) All occipital regions Cerebral peduncle All other cortical regions Posterior corona radiata (PCR) (parietal) All parietal regions Cerebral peduncle All other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined parieto-occipital tracts. *For the tract through the splenium of the corpus callosum, a callosal waypoint mask was also used. doi:10.1371/journal.pone.0103038.t004 2.1.6. Prematurity. Three of 16 SPD boys were born Image Registration Tool (FLIRT; www.fmrib.ox.ac.uk/fsl/flirt) prematurely, one at 32 weeks gestation and two at 34 weeks with 12-parameter linear image registration [23]. All diffusion- gestation. One of the 23 typically developing children was born weighted volumes were registered to the reference b = 0 s/mm prematurely, at 33 weeks gestation. These four subjects were found volume. To evaluate subject movement, we calculated a scalar to be in the middle of the distribution for global FA and mean FA parameter quantifying the transformation of each diffusion volume extracted from clusters of significantly affected voxels using TBSS to the reference. A heteroscedastic two-sample Student’s t-test for their respective groups, and therefore they were not considered verified that there were no significant differences between SPD, to be outliers [16]. None of the ASD subjects were born ASD, and TDC groups in movement during the DTI scan (p. prematurely. 0.05). The non-brain tissue was removed using the Brain Extraction Tool (BET; http://www.fmrib.ox.ac.uk/analysis/ 2.2. Image acquisition research/bet). FA was calculated using the FMRIB Software MR imaging was performed on a 3T Tim Trio scanner Library (FSL) DTIFIT function. 2.3.2. High angular resolution diffusion imaging (HARDI) (Siemens, Erlangen, Germany) using a 12-channel head coil. Structural MR imaging of the brain was performed with an axial and fiber tractography. The FSL bedpostx tool was used for HARDI reconstruction of the diffusion data, modeling multiple 3D magnetization prepared rapid acquisition gradient-echo (MP- RAGE) T1-weighted sequence (TE = 2.98 ms, TR = 2300 ms, fiber orientations per voxel, and thereby accounting for crossing TI = 900 ms, flip angle of 9u) with a 256 mm field of view fibers [24]. Probabilistic streamline tractography was performed (FOV), and 160 1.0 mm contiguous partitions at a 2566256 using FSL’s probtrackx2 to delineate white matter tracts of matrix. Whole-brain DTI was performed with a multislice 2D interest, using the strategies described in Tables 3–5 and single-shot twice-refocused spin-echo echo-planar sequence with illustrated in Figure S1. Seed, waypoint, termination, and 64 diffusion-encoding directions, diffusion-weighting strength of exclusion masks for tractography were primarily derived from b = 2000 s/mm , iPAT reduction factor of 2, TE/TR = 109/ the gray-white matter boundaries (GWB) of the 82 Freesurfer 8000 ms, averages = 1, interleaved 2.2 mm axial slices with no cortical and subcortical regions, which were automatically gap, and in-plane resolution of 2.262.2 mm with a 1006100 segmented on the T1-weighted MR images using Freesurfer matrix and FOV of 220 mm. An additional volume was acquired 5.1.0 [25] and registered using a linear affine transformation to with no diffusion weighting (b = 0 s/mm ). The total DTI diffusion space using FLIRT. The left and right cerebral peduncles acquisition time was 8.67 min. were manually defined for each subject. 2.3.3 Tract delineation. Subsequent to performance of probabilistic streamline fiber tractography, tract masks for every 2.3. DTI analysis 2.3.1. Pre-processing. The diffusion-weighted images were tract described in Tables 3–5 were separately generated for each corrected for motion and eddy currents using FMRIB’s Linear subject. Each mask was created by taking the intersection of the Table 5. Tractographical approach for frontal tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Anterior thalamic radiation Medial orbitofrontal cortex Eroded thalamus All other gm regions (ATR) (medial orbitofrontal cortex) Anterior thalamic radiation Rostral middle frontal cortex Eroded thalamus All other gm regions (ATR) (rostral middle frontal cortex) Genu of the corpus callosum Left medial orbitofrontal cortex Right medial orbitofrontal cortex All other cortical regions (medial orbitofrontal cortex) Genu of the corpus callosum Left rostral middle frontal cortex Right rostral middle frontal cortex All other cortical regions (rostral middle frontal cortex) Anterior corona radiata (ACR) All frontal regions Cerebral peduncle All other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined frontal tracts. *For the tracts through the genu of the corpus callosum, a callosal waypoint mask was also used. doi:10.1371/journal.pone.0103038.t005 PLOS ONE | www.plosone.org 4 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD PLOS ONE | www.plosone.org 5 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 1. Examples of each delineated tract for a representative subject. Green masks represent frontal tracts, blue masks represent parietal-occipital tracts, and orange masks represent temporal tracts. The tracts are superimposed upon the T1 image, registered to diffusion space and with decreased opacity, of the representative subject. doi:10.1371/journal.pone.0103038.g001 binarized, thresholded, tractography-derived streamline map and tracts were individually assessed to confirm bilateral consistency a binary mask of FA.0.2. The streamline threshold used for and to evaluate hypothesized tract laterality. binarization was separately calculated for each streamline map, and equal to 1% multiplied by the maximum number of 2.4. Statistical analysis of group differences streamlines passing through any voxel in the map. This streamline For each tract, decreases in FA were separately assessed for the threshold was a consistent strategy of removing spurious stream- SPD and ASD cohorts relative to controls using one-tailed lines, while retaining most voxels contained within the desired permutation tests (n = 10,000) (adapted from [28]). Permutation white matter tract. The FA threshold further ensured that the testing was utilized, as it is a nonparametric method and thereby voxels contained within the mask were confined to white matter. does not assume normally distributed data. The true two-sample t Additionally, each tract mask for each subject was visually statistic was calculated for control FA vs. patient FA, and a two- inspected to confirm that the anatomy of each target tract was sample t statistic distribution was generated by permuting the accurately and consistently defined. control and patient labels 10,000 times, calculating a t-statistic White matter connectivity was calculated as the average FA value each time. The one-tailed p value was then calculated as the value within the delineated tract of interest. This measurement has number of permuted t statistic values lying below the true t been shown to be highly reproducible in cross-sectional [26] and statistic, divided by the number of permutations (10,000). Group longitudinal studies [27]. differences were assessed separately for each patient cohort relative Representative examples of each of the 15 delineated tracts are to controls at a false discovery rate (FDR) - corrected p value displayed in Figure 1. All masks used for tractography were the threshold (from p,0.05), with FDR correction applied separately GWBs of Freesurfer regions except for manually-defined cerebral to tracts within each region (separately for the temporal, parietal- peduncles and corpus callosum masks. Eroded thalamus masks occipital, and frontal tracts). Because the perceptual reasoning refer to an eroded version of the Freesurfer thalamus which was index (PRI) scores were significantly lower for the ASD cohort transformed using the fslmaths erode filtering operation with a box compared to the TDC and SPD subjects, a post-hoc group kernel of width 9, a step taken to prevent the thalamic mask from difference analysis was conducted while controlling for PRI. For overlapping the corpus callosum and resulting in spurious each tract, a general linear model (GLM) was fit to the data using interhemispheric streamlines. Except for callosal connections, PRI as a regressor, and permutation tests were performed in the each tract was delineated separately in both the left and right same way as described above, using t statistics for the group hemispheres. Following mask extraction (after thresholding by coefficient estimates from the GLM. Differences were again streamlines and FA), corresponding left and right hemisphere tract assessed using FDR correction within the temporal, parietal- masks were combined for subsequent analysis. The unilateral occipital, and frontal regions. Figure 2. Group differences between TDC, SPD, and ASD subjects in average FA within different temporal tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g002 PLOS ONE | www.plosone.org 6 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 3. Group differences between TDC, SPD, and ASD subjects in average FA within different parietal-occipital tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g003 five subtests of the SP (auditory, visual, tactile, inattention, 2.5. Cognitive associations multisensory integration) were investigated dimensionally across Pearson’s correlations of FA in the 15 examined tracts with the all individuals. Statistical significance was assessed at p,0.05 with VCI, PRI, WMI, PSI, the social component of the SCQ, and the Figure 4. Group differences between TDC, SPD, and ASD subjects in average FA within different frontal tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g004 PLOS ONE | www.plosone.org 7 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 6. Connectivity (FA) in all tracts. Tract TDC mean FA6SD ASD mean FA6SD P-val SPD mean FA6SD P-val Fusiform - amygdala 0.369360.0182 0.354660.0177 0.0112 0.373860.0203 0.2298 Fusiform - hippocampus 0.358760.0146 0.346960.0156 0.0078 0.355860.016 0.282 Uncinate fasciculus 0.342960.0156 0.337560.0137 0.118 0.338860.0099 0.1886 ILF 0.411460.0147 0.400860.0192 0.019 0.402860.0164 0.0468 IFOF 0.395660.0171 0.384460.0156 0.031 0.389360.0128 0.11 PTR (optic radiations) 0.409560.0113 0.402960.0127 0.0516 0.403260.0175 0.0814 Dorsal visual stream 0.415560.0147 0.405260.0179 0.0134 0.400960.0156 0.0028 Splenium of the CC 0.465860.0161 0.458960.0167 0.1012 0.451760.0238 0.0164 (lat occipital) PCR (occipital) 0.419460.0123 0.409360.0161 0.0144 0.408560.019 0.0198 PCR (parietal) 0.418260.0093 0.414260.0168 0.1338 0.412460.0178 0.1018 ACR 0.418260.0091 0.425960.0146 0.0508 0.413660.0156 0.1236 ATR (orbitofrontal) 0.362360.0137 0.362760.0137 0.4984 0.358160.014 0.1694 ATR (rostral middle frontal) 0.353060.0111 0.353260.0143 0.3616 0.345760.0105 0.0238 Genu of the CC (orbitofrontal) 0.436160.0224 0.433860.0179 0.3666 0.429660.0218 0.1916 Genu of the CC 0.410560.0196 0.407860.0208 0.4306 0.400860.0204 0.0674 (rostral middle frontal) The mean and standard deviation of FA within each tract for each group, with associated p values for group differences of the TDC cohort with either the SPD cohort or the ASD cohort. Bolded p values represent significant group differences at p,0.05, FDR corrected. doi:10.1371/journal.pone.0103038.t006 FDR correction across all 15 tracts. For tracts and cognitive/ or parietal PCR relative to TDC; however, there were strong behavioral metrics demonstrating significant associations across trends toward lower connectivity of the optic radiations in both the groups, post-hoc correlational analyses were conducted for the ASD and SPD groups relative to TDC. unilateral tract FA (left and right hemisphere independently) 3.1.3. Group differences of connectivity in frontal across groups, as well as unilateral and bilateral tract FA (left and tracts. Connectivity in the frontal tracts was not significantly right combined) for each cohort (TDC, SPD, and ASD) decreased for either the SPD or ASD cohorts, although the SPD independently. group showed trends towards decreased connectivity for all measured frontal tracts. Results 3.2. Unilateral versus bilateral white matter tracts 3.1. Group differences in white matter connectivity Homologous white matter tracts of the left and right cerebral Figures 2–4 depict group differences of structural connectivity hemispheres were combined for purposes of consolidation and denoted by fractional anisotropy (FA) in tracts predominantly improved statistical power. However, group differences were also involving the temporal, parietal-occipital, or frontal regions. computed unilaterally for each tract. In all cases, the results from Table 6 quantitatively details these group differences. We have bilateral tracts shown here agree with trends or statistically additionally added the results of mean diffusivity, axial diffusivity, significant group differences from the component unilateral tracts, and radial diffusivity for the three groups). with no appreciable hemispheric asymmetries found. 3.1.1. Group differences of connectivity in temporal tracts. Significantly impaired connectivity (lower FA) was 3.3. Accounting for perceptual reasoning index variation detected for the ASD cohort alone relative to the TDC cohort PRI was significantly lower in the ASD cohort compared to the in the fusiform-amygdala and fusiform-hippocampus tracts, the TDC subjects, thus group differences in connectivity were inferior fronto-occipital fasciculi (IFOF), and the inferior longitu- computed while controlling for PRI scores. After controlling for dinal fasciculi (ILF) (p,0.05, FDR corrected). The SPD cohort PRI and including FDR correction, connectivity for the ASD showed no significant differences in these tracts relative to the cohort was no longer significantly lower in the IFOF or ILF, but TDC cohort. There was no significant difference in connectivity of still demonstrated decreased FA in the fusiform -amygdala and the uncinate fasciculi of either the ASD or SPD cohorts relative to fusiform –hippocampus tracts. The results for the SPD subjects the controls. were unchanged when controlling for PRI, as expected since these 3.1.2. Group differences of connectivity in parietal- subjects did not demonstrate differences in PRI relative to TDC. occipital tracts. The SPD group alone showed significantly decreased connectivity in the splenium of the corpus callosum relative to the TDC cohort. Both the SPD and the ASD group 3.4. Cognitive associations Significant combined-group correlations were found between showed reduced connectivity relative to controls in the dorsal visual stream and the posterior corona radiata (occipital portion) WMI and the bilateral optic radiations (r = 0.41, p = 0.003) as well (all results with p,0.05, FDR corrected). as the bilateral PCR (occipital) (r = 0.49, p,0.001) (Figure 5). Neither the SPD nor ASD groups demonstrated significant These tracts both demonstrate left lateralized associations for the differences in the optic radiations (pericalcarine – thalamus PTR) combined groups. The SPD cohort alone demonstrates significant PLOS ONE | www.plosone.org 8 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 5. Combined-group associations between tract connectivity and WMI. The two bilateral tracts demonstrating significant associations between FA and WMI after FDR correction are displayed. Optic radiation: r = 0.41, p = 0.003. PCR (occipital): r = 0.49, p,0.001. Results of unilateral and individual group correlations are displayed in Table 7. doi:10.1371/journal.pone.0103038.g005 individual-group associations between WMI and FA in both of demonstrate right lateralized associations for the combined groups these bilateral tracts, while ASD demonstrates significant or trend- (Table 8). level associations (Table 7). Significant combined-group correlations were found between Significant combined-group correlations were found between the inattention measures of the Sensory Profile and the dorsal the social component of the SCQ and the bilateral fusiform - visual stream (r = 0.38, p = 0.006) as well as the bilateral PCR amygdala (r =20.44, p = 0.001) as well as the bilateral fusiform- (occipital) (r = 0.46, p = 0.001) (Figure 7). There is no strong hippocampus (r =20.39, p = 0.004) (Figure 6). These tracts both evidence of lateralization for the combined groups (Table 9). The association between inattention and FA in the bilateral PCR PLOS ONE | www.plosone.org 9 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 7. Associations between tract connectivity and WMI, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right PTR (optic radiation) All 0.41 0.003 0.36 0.009 0.32 0.020 TDC 20.08 0.715 0.07 0.753 20.14 0.522 ASD 0.54 0.048 0.53 0.054 0.25 0.380 SPD 0.62 0.010 0.35 0.188 0.73 0.001 PCR (occipital) All 0.49 ,0.001 0.50 ,0.001 0.32 0.022 TDC 0.22 0.320 0.39 0.077 20.07 0.746 ASD 0.46 0.101 0.40 0.161 0.30 0.299 SPD 0.67 0.004 0.68 0.004 0.56 0.024 The bilateral tracts demonstrating significant combined-group associations with WMI are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t007 (occipital) is consistent with our previously published findings in to TDC, but relatively unaffected in SPD. While both the ASD the combined SPD and TDC cohorts [16]. and the SPD participants demonstrate white matter pathology in the sensory processing regions of the dorsal visual stream and the Significant combined-group correlations were found between the auditory factor of the Sensory Profile and the bilateral PCR posterior corona radiata, only the SPD cohort demonstrates (occipital) (r = 0.42, p = 0.004) (Figure 8). There is no strong statistically significant differences in the splenium of the corpus callosum relative to the TDC cohort. These findings extend evidence of lateralization for the combined groups (Table 10). The association between this auditory measure and FA in the PCR is previous research using DTI in autism cohorts to include concurrent analysis of children that exhibit sensory processing consistent with our previously published findings in the combined SPD and TDC cohorts [16]. differences, but not the language and social deficits that The combined groups did not demonstrate significant correla- characterize a full ASD diagnosis. While the most extensive white matter alterations in the SPD tions between FA in the 15 investigated tracts and PRI, VCI, or subjects are observed in the parieto-occipital tracts, which subserve the other subscores of the Sensory Profile after correction for multiple comparisons. auditory, tactile, and visual perception and integration, this cohort demonstrates trends towards decreased connectivity compared to TDC in most measured tracts. It is also worth specific comment 3.5. Tract overlap that, while both the SPD and ASD cohorts were affected in these Some tracts demonstrated a significant fraction of shared voxels. fundamental sensory processing tracts, the FA in all but one of Figure 9 depicts the average fraction of each tract’s voxels that are these tracts trended lower for the SPD subjects relative to the ASD shared with every other investigated white matter tract. In the subjects. This difference may reflect the prominence of abnormal most extreme case, a subject average of 77% of the fusiform - sensory related behaviors, which is an inclusion criterion for SPD amygdala tract voxels were contained within the fusiform - group membership, whereas, in general, children with ASD are hippocampus tract. There were also significant spatial overlaps primarily characterized by profound social communication within the delineated parietal-occipital tracts, and between some deficits. In this sample, 65% of children with ASD scored in the parietal-occipital and temporal tracts. These results demonstrate definite difference (DD) range (.2 standard deviations from the that some tracts were not completely independent in the group mean) on the Sensory Profile Total Score and 57% were in the difference results above. Sections of the fusiform-amygdala tract DD range for the Auditory Processing Score. While many children that were independent of the fusiform-hippocampus, and sections with ASD have auditory, tactile and visuomotor processing of the fusiform-hippocampus that were independent of the challenges, these deficits are not as ubiquitous as in our SPD fusiform-amygdala tract, were separately assessed for group cohort. Our findings further suggest that sensory-based behavioral differences. Neither of the independent sections of these tracts deficits in both groups may be predicated on atypical conduction demonstrated statistically significant group differences, suggesting of information from unimodal to multimodal integration regions as that shared voxels drive the observed differences between the ASD well as inefficient transfer of information between hemispheres via cohort and controls. the corpus callosum for the SPD group. Perhaps the most striking finding is that, relative to the control Discussion group, the ASD cohort shows reduced structural connectivity in This study is the first to investigate white matter connectivity of the fusiform gyrus connections to the amygdala and hippocampus, both children with SPD and children with ASD relative to whereas children with SPD do not. These white matter pathways typically developing children. Diffusion MR fiber tracking was are thought to facilitate facial emotional processing, a core feature employed for the hypothesis-driven identification of specific white of autism and the domain of clinical divergence for ASD versus matter tracts. The results suggest both overlapping and divergent SPD [29]. In fact, a recent study reports that infants later white matter microstructural pathology affecting the two clinical diagnosed with autism show reduced attention to essential facial cohorts, with tracts traditionally associated with social emotional information with declining direct eye gaze as early as 2–6 month of processing being significantly affected for the ASD cohort relative age [30]. The neuroanatomy of facial emotion processing has been PLOS ONE | www.plosone.org 10 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 6. Combined-group associations between tract connectivity and SCQ-social. The two bilateral tracts demonstrating significant associations between FA and the social component of the SCQ after FDR correction are displayed. Fusiform-amygdala: r =20.44, p,0.001. Fusiform- hippocampus: r =20.39, p = 0.004. Results of unilateral and individual group correlations are displayed in Table 8. doi:10.1371/journal.pone.0103038.g006 intensively studied with repeated implication of the amygdala- autism, the right hippocampal fusiform tract was suggested to have fusiform system. Individuals with ASD have been reported to have smaller diameter axons corresponding with slower neural trans- less activation of subcortical regions including the amygdala and mission, which was thought to lead to secondary changes in the left fusiform gyrus during subliminal emotional face processing, a lack amygdala-fusiform and hippocampal-fusiform pathways [16]. By of expected volumetric correlation between the amygdala and contrast, a diffusion fiber tractography study of individuals with fusiform gyrus, as well as behavioral deficits in face recognition Williams syndrome (7q11.23 deletion) are reported to show that negatively correlate with left fusiform cortical thickness [31], elevations of FA in fusiform tracts [33]. Individuals with Williams [32]. In a DTI based analysis of adolescents and adults with Syndrome show a social phenotype that is in some ways opposite PLOS ONE | www.plosone.org 11 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 8. Associations between tract connectivity and the SCQ-social, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right Fusiform-amygdala All 20.44 0.001 20.27 0.048 20.37 0.006 TDC 20.39 0.066 20.18 0.423 20.29 0.178 ASD 20.40 0.137 20.34 0.220 20.18 0.515 SPD 20.32 0.233 0.03 0.907 20.50 0.049 Fusiform-hippocampus All 20.39 0.004 20.25 0.065 20.40 0.003 TDC 20.14 0.510 20.11 0.613 20.15 0.508 ASD 20.25 0.370 20.22 0.433 20.12 0.680 SPD 20.27 0.307 20.01 0.979 20.47 0.069 The bilateral tracts demonstrating significant combined-group associations with the social component of the SCQ are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t008 to the autism phenotype with increased attention to faces and behavioral correlates of altered connectivity is warranted. How- abundant social interest and drive. It is thus worthwhile to ever, these findings suggest a role for neuroimaging in under- consider social cognition, or facial emotion recognition specifically, standing the neural mechanisms that differentiate children with a as a continuous trait that might map directly to connectivity of the variety of domain specific deficits, including basic sensory fusiform tract to limbic structures. processing and social emotional processing. The role of develop- There are however additional farther reaching implications for ment and therapeutic interventions on these systems remains an fusiform connectivity disruptions with regard to language devel- open and important question to explore in these clinical cohorts as it is unclear whether these findings are primary or represent a opment. A theoretical model of audiovisual affective speech perception begins with input to primary auditory and visual cortex consequence of aberrant tract remodeling predicated on less [34]. The input module feeds information to the fusiform gyrus as practice of these skills from early infancy. well as the integration module of the superior and middle temporal As can be seen from the group comparison figures, while there cortex. The primary sensory cortices as well as the fusiform gyrus are clear and statistically significant group differences, there is also are reciprocally connected to the amygdala and insula, which considerable overlap in the measurements from tracts across all comprise the emotion module that guides emotional relevance and three groups: ASD, SPD, and TDC. This highlights the may facilitate the rapid recruitment of limbic brain regions by importance of a new direction for cognitive and behavioral visual inputs. Additional contextual information is brought in research based on the investigation of abilities as a continuous through connections with the memory module, including the measure across children rather than split by exceedingly broad hippocampus and parahippocampal gyrus. This framework is and overlapping clinical labels, a concept which has been useful in considering how autism social communication deficits formalized in the Research Domain Criteria (RDoC) Project may map to neuroanatomic networks. [42]. It is in this context that we frame our investigation of In addition to the fusiform connections, our ASD group was associations between cognitive measures and tract connectivity found to have reduced FA in the ILF and the IFOF. This is in line across all three study groups. The fusiform-amygdala and with previous reports, although there is considerable variability in fusiform-hippocampus tracts are the only two (out of 15) tracts to demonstrate significant associations with the social component the literature, likely resulting from group heterogeneity in terms of symptom variability, severity, and age of cohort [8], [35]–[38]. of the SCQ across groups with correction for multiple comparisons The ILF, or inferior longitudinal fasciculus, has been shown to (Figure 6, Table 8). Further investigation of laterality in these two directly connect the occipital cortex to the anterior temporal lobe tracts revealed that these associations are significantly right- and the amygdala. The IFOF originates in the visual cortex runs lateralized (Table 8), an observation which is consistent with the medially to the ILF and directly connects to the inferior frontal Conturo et al. (2008) [16] finding of primary right-lateralized and dorsolateral frontal cortex. In a study involving children with effects in these two tracts for ASD subjects. It is important to note visual perceptional impairment, decreased FA of the ILF that the associations between connectivity in these two tracts and correlated with impaired object recognition [39]. The IFOF likely the SCQ-social measure were not significant on an individual overlaps spatially and functionally with the ILF, and is thought to group basis. However, the individual group correlations all trended in the same (negative) direction as the combined-group be a tract that is relatively new evolutionarily due to its absence in animal brains [40]. In a large lesion-based study population, the correlations (Table 8). Connectivities (FA) in the optic radiation right IFOF in particular is implicated in rapid recognition of and PCR (occipital) were found to be significantly correlated with emotional facial expressions [41]. The finding of significantly WMI after FDR correction in the combined groups. Investigation reduced connectivity in ASD of the ILF and IFOF is in of these associations in the individual cohorts revealed strong associations with WMI for the SPD cohort bilaterally in both of concordance with the reduced connectivity seen specifically in the fusiform connections. What is most revealing is the relative these tracts. The ASD cohort alone also demonstrates significant preservation of these connections in our SPD cohort. Clearly or trend-level associations with WMI and connectivity in these additional investigation to understand the relationship between the tracts. Though the optic radiation and PCR (occipital) were the speed of information transmission in these tracts as well as the only two tracts that demonstrated significant associations with PLOS ONE | www.plosone.org 12 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 7. Combined-group associations between tract connectivity and the inattention measure of the Sensory Profile. The two bilateral tracts demonstrating significant associations between FA and the inattention measure of the Sensory Profile after FDR correction are displayed. Dorsal visual stream: r = 0.38, p = 0.006. PCR (occipital): r = 0.46, p,0.001. Results of unilateral and individual group correlations are displayed in Table 9. doi:10.1371/journal.pone.0103038.g007 WMI after FDR correction, many of the other investigated tracts measures are consistent with our prior findings [16], we did not (including frontal, temporal, and parietal-occipital tracts) demon- find significant associations with the other Sensory Profile strated trend-level associations. This is consistent with reports from measures after correction for multiple comparisons. prior DTI studies of WMI that have found widespread associations There are important limitations to note for this study, which of WMI with white matter connectivity [43], [44]. While our should motivate further investigation. First, we have not deter- findings of significant associations between connectivity in the mined an optimal method for characterizing the sensory subtypes PCR (occipital) and the Sensory Profile auditory and inattention and distinguishing between hypo- or hyper-sensory sensitivity, nor PLOS ONE | www.plosone.org 13 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 9. Associations between tract connectivity and the inattention factor of the Sensory Profile, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right Dorsal Visual Stream All 0.38 0.006 0.34 0.012 0.27 0.052 TDC 20.16 0.479 20.09 0.688 20.19 0.408 ASD 0.37 0.197 0.04 0.885 0.37 0.195 SPD 0.25 0.358 0.05 0.858 0.40 0.121 PCR (occipital) All 0.46 0.001 0.37 0.006 0.41 0.002 TDC 0.00 1.000 0.06 0.772 20.05 0.836 ASD 0.63 0.015 0.44 0.112 0.63 0.017 SPD 0.41 0.117 0.19 0.488 0.54 0.029 The bilateral tracts demonstrating significant combined-group associations with the inattention measure of the Sensory Profile are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t009 do we have sufficient power in this study for sensory subtype group PCR. Despite these spatial overlaps, the group difference results analysis. We and many sensorimotor based researchers are are not identical between overlapping tracts and provide working to develop a phenotyping tool that maps to specific white separately valuable information about structural connectivity in matter tracts, and we hope to identify and characterize separate these subjects. Additional connectivity analysis, both structural and constructs of sensory deficits in larger cohorts going forward. functional, will shed additional light on specific regional contri- Second, tract overlap exists in our results. A significant portion of butions to the neural underpinnings of sensory and emotional the amygdala-fusiform tract is contained within the hippocampal- processing differences. Our investigation is also limited in fusiform tract. In addition, the ILF is partially contained within the generalizability, as all of the subjects were boys between the ages dorsal visual stream and the PTR is partially contained within the of 8 and 12 years in an effort to limit developmental confounds in Figure 8. Combined-group associations between tract connectivity and the auditory measure of the Sensory Profile. The bilateral tract demonstrating significant associations between FA and the auditory measure of the Sensory Profile after FDR correction are displayed. PCR (occipital): r = 0.42, p = 0.002. Results of unilateral and individual group correlations are displayed in Table 10. doi:10.1371/journal.pone.0103038.g008 PLOS ONE | www.plosone.org 14 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 10. Associations between tract connectivity and the auditory factor of the Sensory Profile, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right PCR (occipital) All 0.42 0.004 0.33 0.017 0.37 0.007 TDC 20.33 0.129 20.20 0.353 20.27 0.216 ASD 0.60 0.023 0.38 0.185 0.62 0.021 SPD 0.41 0.114 0.20 0.465 0.49 0.056 The bilateral tracts demonstrating significant combined-group associations with the inattention measure of the Sensory Profile are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t010 this small sample. The PRI scores of the ASD cohort were therefore needed to determine whether these findings generalize to significantly lower than that of the SPD and TDC cohorts; other ages, genders, and intellectual abilities. however, the most important group differences in structural Future research will include investigation of functional connec- connectivity between ASD and controls remained statistically tivity using resting state fMRI and magnetoencephalography significant after regressing out the effect of PRI. Further research is (MEG). The ROIs used to determine structural connectivity in this study can be used to assess differences in functional connectivity Figure 9. Average fraction of tract overlap. Color intensity corresponds to the subject average of the fraction of the voxels of the tracts on the vertical axis that are contained within the tracts on the horizontal axis. Tracts that are more than one-third contained in any other tract are indicated by an asterisk on the vertical axis. doi:10.1371/journal.pone.0103038.g009 PLOS ONE | www.plosone.org 15 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD between these same regions, with functional coupling hypothesized differences based on two-tailed permutation tests with FDR to be reduced where decreased structural connectivity was found correction for 15 comparisons. in this work. While prior studies have found relationships between (TIFF) functional connectivity and white matter volume in ASD [45]– Figure S3 Group differences of RD in all tracts. Units of [47], there have been no reported associations between functional diffusivity are in mm /sec. Asterisks indicate significant connectivity and DTI measures in SPD. differences based on two-tailed permutation tests with FDR We hope that by utilizing larger sample sizes and direct correction for 15 comparisons. assessment of auditory, tactile, visuomotor processing, we will be (TIFF) able to gain a deeper understanding of how neural circuitry differences map to clinically relevant challenges for individual Figure S4 Group differences of AD in all tracts. Units of children. The ultimate goal of this and future work is to guide diffusivity are in mm /sec. Asterisks indicate significant personalized treatments ranging from behavioral interventions and differences based on two-tailed permutation tests with FDR targeted psychopharmacology to cognitive training using child- correction for 15 comparisons. friendly video game platforms. (TIFF) Table S1 Group differences of FA between the ASD and Supporting Information SPD cohorts in ASD-affected temporal tracts. P values are Figure S1 Example ROIs for fiber tracking of the derived from one-tailed permutation tests for SPD FA . ASD FA. homotopic visual tract through the splenium of the Bolded p values indicate significant differences after FDR corpus callosum. Displayed is an axial slice from the FA image correction. of a representative subject with overlaid ROIs for probabilistic (XLSX) fiber tractography. The seed mask is the grey-white matter Table S2 Group differences of FA in all tracts, excluding boundary of the left lateral occipital cortex, and contains voxels SPD subjects with SCQ.15. P values are derived from one- from which 2000 streamlines each are initiated. The termination tailed permutation tests for TDC FA . SPD FA. Bolded p values mask is the grey-white matter boundary of the right lateral indicate significant group differences after FDR correction. occipital cortex, and causes streamlines to terminate upon (XLSX) encountering the mask. The termination mask and the corpus callosum are both used as waypoint masks, indicating that Acknowledgments streamlines need to pass through the corpus callosum and reach the termination mask in order to be retained. The exclusion mask We are grateful to our participants and their families for their time and is the union of the grey-white matter boundaries of all other support. cortical regions, and causes streamlines that encounter these voxels to be excluded. The displayed resultant tract is the result of Author Contributions probabilistic tractography under the previously described con- Conceived and designed the experiments: EJM PM JPO YSC. Performed straints and a subsequent streamline and FA threshold. the experiments: EJM AA SH JH SD. Analyzed the data: YSC JPO. Wrote (TIFF) the paper: YSC JPO EJM PM AA SH JH SD. Figure S2 Group differences of MD in all tracts. Units of diffusivity are in mm /sec. Asterisks indicate significant References 1. Ahn RR, Miller LJ, Milberger S, McIntosh DN (2004) Prevalence of parents’ autism: a study using diffusion spectrum imaging tractography. Psychiatry perceptions of sensory processing disorders among kindergarten children. research 192: 60–66. American Journal of Occupational Therapy 58: 287–293. 12. Thomas C, Humphreys K, Jung KJ, Minshew N, Behrmann M (2011) The anatomy of the callosal and visual-association pathways in high-functioning 2. Ben-Sasson A, Hen L, Fluss R, Cermak SA, Engel-Yeger B, et al. (2009) A meta- autism: a DTI tractography study. Cortex 47(7): 863–73. analysis of sensory modulation symptoms in individuals with autism spectrum 13. Aoki Y, Abe O, Nippashi Y, Yamasue H (2013) Comparison of white matter disorders. Journal of Autism and Developmental Disorders 39: 1–11. integrity between autism spectrum disorder subjects and typically developing 3. Brett-Green BA, Miller LJ, Gavin WJ, Davies PL (2008) Multisensory individuals: a meta-analysis of diffusion tensor imaging tractography studies. integration in children: a preliminary ERP study. Brain Research 1242: 283– Molecular autism 4(1): 25. 14. Conturo TE, Williams DL, Smith CD, Gultepe E, Akbudak E, et al. (2008) 4. Bundy AC, Shia S, Qi L, Miller LJ (2007) How does sensory processing Neuronal fiber pathway abnormalities in autism: an initial MRI diffusion tensor dysfunction affect play?. American Journal of Occupational Therapy 61: 201– tracking study of hippocampo-fusiform and amygdalo-fusiform pathways. J Int Neuropsychol Soc 14: 933–46. 5. American Psychiatric Association (1994) American Psychiatric Association 15. Smith CD, Lori NF, Akbudak E, Sorar E, Gultepe E, et al. (2009) MRI diffusion Diagnostic and statistical manual of mental disorders, Fourth Edition (DSM- tensor tracking of a new amygdalo-fusiform and hippocampo-fusiform pathway IV) Washington, DC. system in humans. J Magn Reson Imaging 29: 1248–61. 6. Marco EJ, Hinkley LB, Hill SS, Nagarajan SS (2011) Sensory processing in 16. Owen JP, Marco EJ, Desai S, Fourie E, Harris J, et al. (2013) Abnormal white autism: a review of neurophysiologic findings. Pediatr Res. 69(5 Pt 2): 48R–54R. matter microstructure in children with sensory processing disorders. Neuro- 7. Perou R, Bitsko RH, Blumberg SJ, Pastor P, Ghandour RM, et al. (2013) Mental Image: Clinical 2: 844–853. health surveillance among children–United States, 2005–2011. Morbidity and 17. Szatmary B, Izhikevich EM (2010) Spike-timing theory of working memory. mortality weekly report. Surveillance summaries (Washington, D.C.: 2002) 62 PLoS Comput. Biol. 6: e1000879.10.1371. Suppl 2: 1–35. 18. Wechsler D (2003) WISC-IV: Administration and Scoring Manual. Psycholog- 8. Travers BG, Adluru NA, Ennis C (2012) Diffusion tensor imaging in autism ical Corporation. spectrum disorder: a review. Autism Research 5: 289–313. 19. Dunn W, Westman K (1997) The sensory profile: the performance of a national 9. Alexander AL, Lee JE, Lazar M, Boudos R, DuBray MB, et al. (2007) Diffusion sample of children without disabilities. American Journal of Occupational tensor imaging of the corpus callosum in Autism. NeuroImage 34: 61–73. Therapy 51: 25–34. 10. Kumar A, Sundaram SK, Sivaswamy L, Behen ME, Makki MI, et al. (2009) 20. Eaves LC, Wingert HD, Ho HH, Mickelson EC (2006) Screening for autism Alterations in Frontal Lobe Tracts and Corpus Callosum in Young Children spectrum disorders with the social communication questionnaire. Journal of with Autism Spectrum Disorder. Cerebral Cortex 20(9): 2103–2113. Developmental and Behavioral Pediatrics 27 (Suppl. 2): S95–S103. 11. Lo YC, Soong WT, Gau SF, Wu YY, Lai MC, et al. (2011) The loss of 21. Lord C, Rutter M, Le Couteur A (1994) Autism diagnostic interview-revised: a asymmetry and reduced interhemispheric connectivity in adolescents with revised version of a diagnostic interview for caregivers of individuals with PLOS ONE | www.plosone.org 16 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD possible pervasive developmental disorders. Journal of Autism and Develop- matter tracts in children with autism spectrum disorder. Proc Natl Acad Sci U S mental Disorders 24: 659–685. A. 111(5): 1981–6. 22. Lord C, Rutter M, Goode S (1989) Autism diagnostic observation schedule: a 36. Jou RJ, Jackowski AP, Papademetris X, Rajeevan N, Staib LH, et al. (2011) standardized observation of communicative and social behavior. Journal of Diffusion tensor imaging in autism spectrum disorders: preliminary evidence of Autism and Developmental Disorders 19: 185–212. abnormal neural connectivity. Aust. N. Z. J. Psychiatry 45: 153–162.10.3109. 23. Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for 37. McGrath J, Johnson K, O’Hanlon E, Garavan H, Gallagher L, et al. (2013) the robust and accurate linear registration and motion correction of brain White matter and visuospatial processing in autism: a constrained spherical images. NeuroImage 17: 825–841. deconvolution tractography study. Autism Res. 6(5): 307–319. 24. Behrens TEJ, Berg HJ, Jbabdi S, Rushworth MF, Woolrich MW (2007). 38. Shukla DK, Keehn B, Muller RA (2011) Tract-specific analyses of diffusion Probabilistic diffusion tractography with multiple fibre orientations: What can tensor imaging show widespread white matter compromise in autism spectrum we gain?. Neuroimage 34: 144–155. disorder. J. Child Psychol. Psychiatry 52: 286–295. 25. Fischl B (2012) FreeSurfer. NeuroImage 62: 774–781. 39. Ortibus E, Verhoeven J, Sunaert S, Casteels I, de Cock P, et al. (2012) Integrity 26. Wakana S, Caprihan A, Panzenboeck MM, Fallon JH, Perry M, et al. (2007) of the inferior longitudinal fasciculus and impaired object recognition in Reproducibility of quantitative tractography methods applied to cerebral white children: a diffusion tensor imaging study. Dev Med Child Neurol 54: 38–43. matter. Neuroimage 36: 630–44. 40. Forkel J, Thiebaut de Schotten M, Kawadler J, dell’Acqua F, Danek A, et al. 27. Danielian LE, Iwata NK, Thomasson DM, Floeter MK (2010) Reliability of (2012) The anatomy of fronto-occipital connections from early blunt dissections fiber tracking measurements in diffusion tensor imaging for longitudinal study. to contemporary tractography. Cortex dx.doi.org/10.1016/j.cor- Neuroimage 49: 1572–1580. tex.2012.09.005. 28. Nichols T, Holmes A (2001) Nonparametric permutation tests for functional 41. Philippi CL, Mehta S, Grabowski T, Adolphs R, Rudrauf D (2009) Damage to neuroimaging: a primer with examples. Human Brain Mapping 15: 1–25. association fiber tracts impairs recognition of the facial expression of emotion. 29. Schultz RT (2005) Developmental deficits in social perception in autism: the role J Neurosci 29: 15089–15099. of the amygdala and fusiform face area. International Journal of Developmental 42. Cuthbert BN, Insel TR (2013) Toward the future of psychiatric diagnosis: the Neuroscience 23 (2): 125–141. seven pillars of RDoC. BMC Med 11: 126. 30. Jones W, Klin A (2013) Attention to eyes is present but in decline in 2-6-month- 43. Tang CY, Eaves EL, Ng JC, Carpenter DC, Mai X, et al. (2010) Brain networks old infants later diagnosed with autism. Nature 504(7480): 427–31. for working memory and factors of intelligence assessed in males and females 31. Dziobek I, Bahnemann M, Convit A, Heekeren HR (2010) The role of the with fMRI and DTI. Intelligence 38: 293–303. fusiform-amygdala system in the pathophysiology of autism. Archives of general 44. Nagy Z, Westerberg H, Klingberg T (2004) Maturation of white matter is psychiatry 67: 397–405. associated with the development of cognitive functions during childhood. J. 32. Kleinhans NM, Richards T, Johnson LC, Weaver KE, Greenson J, et al. (2011) Cogn. Neurosci 16: 1227–1233. fMRI evidence of neural abnormalities in the subcortical face processing system 45. Cherkassky VL, Kana RK, Keller TA, Just MA (2006) Functional connectivity in ASD. NeuroImage 54(1): 697–704. in a baseline resting-state network in autism. Neuroreport 17: 1687–1690. 33. Haas BW, Hoeft F, Barnea-Goraly N, Golarai G, Bellugi U, et al. (2012) 46. Just MA, Cherkassky VL, Keller TA, Kana RK, Minshew NJ (2007) Functional Preliminary evidence of abnormal white matter related to the fusiform gyrus in Williams syndrome: a diffusion tensor imaging tractography study. Genes Brain and anatomical cortical underconnectivity in autism: Evidence from an FMRI Behav 11: 62–68. study of an executive function task and corpus callosum morphometry. Cerebral 34. Jansma H, Roebroeck A, Mu¨nte TF (2013) A network analysis of audiovisual Cortex 17: 951–961. affective speech perception. Neuroscience 256: 230–41. 47. Kana RK, Keller TA, Cherkassky KL, Minshew NJ, Just MA (2006) Sentence 35. Koldewyn K, Yendiki A, Weigelt S, Gweon H, Julian J, et al. (2014) Differences comprehension in autism: Thinking in pictures with decreased functional in the right inferior longitudinal fasciculus but no general disruption of white connectivity. Brain 129(Pt 9): 2484–2493. PLOS ONE | www.plosone.org 17 July 2014 | Volume 9 | Issue 7 | e103038 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png PLoS ONE Pubmed Central

Autism and Sensory Processing Disorders: Shared White Matter Disruption in Sensory Pathways but Divergent Connectivity in Social-Emotional Pathways

Loading next page...
 
/lp/pubmed-central/autism-and-sensory-processing-disorders-shared-white-matter-disruption-5yTIyOug08

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Pubmed Central
Copyright
© 2014 Chang et al
ISSN
1932-6203
eISSN
1932-6203
DOI
10.1371/journal.pone.0103038
Publisher site
See Article on Publisher Site

Abstract

Over 90% of children with Autism Spectrum Disorders (ASD) demonstrate atypical sensory behaviors. In fact, hyper- or hyporeactivity to sensory input or unusual interest in sensory aspects of the environment is now included in the DSM-5 diagnostic criteria. However, there are children with sensory processing differences who do not meet an ASD diagnosis but do show atypical sensory behaviors to the same or greater degree as ASD children. We previously demonstrated that children with Sensory Processing Disorders (SPD) have impaired white matter microstructure, and that this white matter microstructural pathology correlates with atypical sensory behavior. In this study, we use diffusion tensor imaging (DTI) fiber tractography to evaluate the structural connectivity of specific white matter tracts in boys with ASD (n = 15) and boys with SPD (n = 16), relative to typically developing children (n = 23). We define white matter tracts using probabilistic streamline tractography and assess the strength of tract connectivity using mean fractional anisotropy. Both the SPD and ASD cohorts demonstrate decreased connectivity relative to controls in parieto-occipital tracts involved in sensory perception and multisensory integration. However, the ASD group alone shows impaired connectivity, relative to controls, in temporal tracts thought to subserve social-emotional processing. In addition to these group difference analyses, we take a dimensional approach to assessing the relationship between white matter connectivity and participant function. These correlational analyses reveal significant associations of white matter connectivity with auditory processing, working memory, social skills, and inattention across our three study groups. These findings help elucidate the roles of specific neural circuits in neurodevelopmental disorders, and begin to explore the dimensional relationship between critical cognitive functions and structural connectivity across affected and unaffected children. Citation: Chang Y-S, Owen JP, Desai SS, Hill SS, Arnett AB, et al. (2014) Autism and Sensory Processing Disorders: Shared White Matter Disruption in Sensory Pathways but Divergent Connectivity in Social-Emotional Pathways. PLoS ONE 9(7): e103038. doi:10.1371/journal.pone.0103038 Editor: Christophe Lenglet, University of Minnesota, United States of America Received December 5, 2013; Accepted June 25, 2014; Published July 30, 2014 Copyright:  2014 Chang et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was funded by grants from the Wallace Research Foundation, the Gates Family Foundation and the Holcombe Kawaja Family Foundation. EJM, JPO and PM acknowledge support from the Simons Foundation. PM also acknowledges support from NIH R01 NS060776. EJM has received support from NIH K23 MH083890 and KL2 RR024130. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interest exist. * Email: marcoe@neuropeds.ucsf.edu " These authors are co-first authors on this work. and behavioral flexibility [5]. However, individuals with ASD have Introduction also been shown to have ubiquitous challenges in sensory The human brain is a sensory processor. Its core function is to processing [6] with over 90% of children with autism reported perceive, integrate, interpret, and then facilitate the appropriate to have atypical sensory related behaviors. In fact, hyper- or coordinated response to the visual, tactile, auditory, olfactory, and hyporeactivity to sensory input or unusual interest in sensory proprioceptive information present in the world around us. Thus it aspects of the environment is now included in the current DSM 5 comes as no surprise that inaccurate or imprecise sensory diagnostic criteria for ASD [6]. There are, however, children with processing and multisensory integration (MSI) can lead to sensory processing disorders (SPD) who do not show primary impaired intellectual and social development [1]–[4]. There is a language or social deficits but do exhibit atypical sensory reactivity growing recognition of the crucial importance of sensory and/or sensory interests to the same or greater extent as children processing as it contributes to attention, learning, emotional who meet an ASD diagnosis [1]. Children with SPD remain regulation, and even social function in children affected by a wide critically underserved with regard to their developmental chal- spectrum of neurodevelopmental disorders, including autism. lenges in our society due to the lack of a diagnostic label There is also a growing interest in studying sensory processing recognized in the current DSM 5 manual. Many are instead and cognition as dimensional traits across typically developing attributed labels that better describe the sequelae of SPD, such as children and those with psychiatric labels such as autism. oppositional defiant disorder, than the root of the problem. It is Autism spectrum disorders (ASD) have traditionally been therefore highly relevant to better characterize the biological bases characterized by impaired communication, social interaction, PLOS ONE | www.plosone.org 1 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 1. Cognitive Characterization of TDC, ASD, and SPD Cohorts. TDC Mean 6 Std ASD Mean 6 Std Pval SPD Mean 6 Std Pval PRI 113.5613.5 101.6614.1 0.015 115.8611.5 0.576 VCI 119.2612.7 101.6620.5 0.007 117.4612.8 0.660 WMI 108.4610.9 99.6617.7 0.111 104.4612.8 0.320 PSI 101.3613.6 87.4611.1 0.002 97.1612.9 0.334 PRIs, VCIs, WMIs and PSIs for each cohort, with p values from two-tailed t-tests for differences between TDCs and each patient cohort (statistically significant p values of less than 0,05 are indicated in boldface). doi:10.1371/journal.pone.0103038.t001 of this increasingly recognized neurodevelopmental condition. In working memory, the latter of which has been proposed to be addition, the comparison of children with SPD and ASD may help mediated by stereotypical time-locked spatiotemporal spike timing to illuminate the unique neural mechanisms at the core of the ASD patterns [17]. diagnosis: those facilitating social awareness, interest, and drive. In this study, we examine white matter tracts that we With over 1% of children in the USA carrying an ASD label and hypothesize will be atypical in children with SPD or ASD subjects reports of 5–16% of children in the USA having sensory relative to typically developing children (TDC). Based upon our processing difficulties, it is important to define the neural previous work on white matter microstructure in SPD [16], and underpinnings of these conditions and to delineate the areas of upon previous studies of white matter microstructure in ASD, we overlap and the areas of divergence [1], [2], [7]. The advent of posit that both ASD and SPD subjects will have reduced structural diffusion tensor imaging (DTI) and fiber tractography has enabled connectivity compared to controls in parieto-occipital white matter quantitative, noninvasive evaluation of white matter microstruc- tracts involved with sensory processing and integration, whereas ture and connectivity. There is considerable, albeit contradictory, only ASD subjects will have diminished structural connectivity literature reporting altered structural connectivity in individuals relative to controls in temporal tracts associated with social- with ASD using DTI [8]. There are several studies suggesting emotional processing. Furthermore, we posit that tract connectiv- reduced connectivity via the corpus callosum [9]–[11] as well as ity will correlate with measures reflecting sensory processing, others indicating normal or even elevated fractional anisotropy inattention behavior, social behavior, verbal comprehension, (FA), a measure of white matter tract microstructural integrity processing speed, and working memory across groups. from DTI [12]. Beyond the corpus callosum, there are also reports of other white matter tracts that may show variance from typically Methods developing controls, including the inferior fronto-occipital fascic- The Institutional Review Board (IRB) at the University of ulus (IFOF) and the uncinate fasciculus (UF). A recent meta- California in San Francisco approved this study (UCSF IRB analysis of 25 DTI studies in individuals with autism reports Protocol #: 10-01940). Subjects were recruited from the UCSF decreased FA in the corpus callosum, the left UF, and the left Autism and Neurodevelopment Program clinical sites and superior longitudinal fasciculus (SLF), supporting the theory of research database, and from local online parent board listings. specific underconnectivity in autism focused on tracts supporting Informed consent was obtained from the parents or legal auditory information and language processing [13]. Finally, in guardians, with the assent of all participants. addition to auditory and language related tracts, there is considerable interest in tracts that mediate emotional face recognition, a pervasive deficit in children with autism. DTI 2.1. Demographic, sensory, cognitive and behavioral data studies have specifically investigated the fusiform-hippocampal 2.1.1. General demographics. Sixteen right-handed males and fusiform-amygdala tracts in individuals with autism and have with SPD, fifteen males with ASD (12 right-handed, 1 left-handed, reported variation thought to relate to atypical function [14], [15]. 2 ambidextrous), and 23 right-handed male TDC, all between 8 In comparison to DTI studies of ASD, investigation of structural and 12 years of age, were prospectively enrolled under our IRB connectivity in children with isolated SPD is in its infancy. We protocol. recently reported that, although children with SPD do not exhibit Voxel-based analysis of the DTI data from the 16 SPD subjects morphological abnormalities from structural MR imaging, they and the 23 TDC using tract-based spatial statistics (TBSS) to have strikingly decreased white matter microstructural integrity, investigate white matter microstructure was previously reported in especially in posterior cerebral regions [16]. These regions are [16]. Group differences in the TBSS analysis were determined in a implicated in unimodal sensory processing as well as MSI, and are common atlas space after inter-subject image registration. In the regulated by top-down attention modulation via thalamic projec- present study, we examine white matter connectivity using tions. We further showed that white matter connectivity correlates diffusion fiber tractography in each subject’s native space, with with behavioral measures of unimodal sensory behavior, multi- the addition of an ASD cohort. sensory integration, and inattention. White matter microstructural 2.1.2. General cognition. All subjects were assessed with the integrity is crucial to the speed and bandwidth of information Wechsler Intelligence Scale for Children-Fourth Edition [18] and transmission throughout the brain. Degraded connectivity of were required to have a Perceptual Reasoning Index (PRI) score primary sensory cerebral tracts or of pathways connecting $70. We used PRI as our measure of cognition for inclusion, as multimodal sensory association areas may thereby result in the communication deficits are part of the core diagnosis of ASD. loss of the precise timing of action potential propagation needed Verbal Comprehension Index (VCI), Processing Speed Index for accurate sensory registration and integration. These effects (PSI), and Working Memory Index (WMI) were also obtained may be reflected in assessable metrics such as processing speed and PLOS ONE | www.plosone.org 2 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 2. Sensory Profile Characterization of TDC, ASD, and SPD cohorts. TDC Mean 6Std ASD Mean 6Std SPD Mean 6Std Auditory 33.663.5 *24.465.9 *22.764.9 Tactile 83.365.8 72.468.6 *62.9+8.8 Visual 41.263.0 35.666.3 32.367.1 Inattention 28.763.6 *20.364.4 *17.865.3 Total 172.3611.0 *135.1618.2 *128.5615.8 Multisensory 31.363.1 23.764.5 22.263.7 Asterisks indicate mean scores that fall within the definite difference range. None of the mean scores fell in the probable difference range. doi:10.1371/journal.pone.0103038.t002 from this assessment. These measures are displayed in Table 1 for Diagnostic Inventory-Revised (ADI-R) [21], a parent history each cohort. interview, and the Autism Diagnostic Observation Schedule 2.1.3. Sensory processing evaluation. All subjects were (ADOS) [22], a structured play session. We used current evaluated with the Sensory Profile [19], which is currently the diagnostic scoring for the ADOS and lifetime scoring for the most widely used parent report measure of atypical sensory related ADI-R. None of the TDC cohort had an SCQ score $15. All behavior. The Sensory Profile (SP) is a caregiver report participants in the ASD cohort met criteria on both the ADI-R questionnaire (125 items) which measures behavioral sensory and ADOS; all but one scored $15 on the SCQ. differences, yielding scores within individual sensory domains and Three of the SPD cohort scored above 15 on the SCQ and were factors as well as a total score. A probable difference (PD) in further evaluated with the ADI-R and ADOS. One SPD sensory behavior is defined as a total score between 142 and 154, participant scored above the ASD cutoff on the current diagnosis while a definite difference (DD) is a score of #141. Lower scores scoring of the ADOS but did not meet criteria on the ADI-R. reflect more atypical behavior. We use the auditory, visual, tactile, Another SPD individual met criteria on the ADI-R but not the multisensory integration, and inattention/distractibility scores to ADOS. Neither was considered to meet clinical criteria when explore behavioral correlations based on findings from our prior evaluated by a cognitive behavioral child neurologist with report [16]. expertise in autism and neurodevelopment (EJM). The third Inclusion in the SPD group required a community based SPD participant who scored above 15 on the SCQ met neither the Occupational Therapy diagnosis of Sensory Processing Disorder ADI-R nor ADOS cut-off. A supplementary analysis was plus a score in the definite difference (DD) range, defined as performed, excluding these three SPD subjects from the study greater than two standard deviations from the mean, of either the cohort (Table S2). total or the auditory processing score of the Sensory Profile. Five of 2.1.5. Attention deficits. On the inattention/distractibility the SPD subjects scored in the DD range for total score alone, four factor of the Sensory Profile, eleven of the 16 SPD subjects scored scored in the DD range for the auditory processing score alone, in the definite difference range, four in the probable difference and seven scored in the DD range for both the total and auditory range, and one in the typical range. Of the 15 ASD subjects, seven score. Two ASD subjects scored in the DD range for the total scored in the definite difference range, five scored in the probable score alone, one ASD subject scored in the DD range for the difference range, two scored in the typical range, and one was not auditory score alone, and seven of the ASD subjects scored in the administered the Sensory Profile. Of the 23 TDC, none scored in DD range for both the total and auditory score. The sensory the definite difference range, three in the probable difference profile was not obtained for one ASD individual. All of the range, and twenty in the typical range. Atypical inattention/ controls scored in the normal range (Table 2). distractibility scores on the Sensory Profile do not necessarily 2.1.4. Autism evaluation. All subjects were evaluated with indicate that individuals would meet clinical criteria for an the Social Communication Questionnaire (SCQ), a parent report attention deficit (hyperactivity) disorder (ADHD) diagnosis. ASD screening instrument [20]. All of the ASD cohort (carrying Formal ADHD evaluations were not conducted as part of this community diagnosis of ASD) as well as the SPD individuals with study. a score above threshold ($15) were evaluated with the Autism Table 3. Tractographical approach for temporal tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Fusiform - amygdala Fusiform gyrus Amygdala All other gm regions Fusiform - hippocampus Fusiform gyrus Hippocampus All other gm regions Uncinate fasciculus (UF) Orbitofrontal cortex* Entorhinal cortex + temporal pole All other gm regions Inferior longitudinal fasciculus (ILF) Pericalcarine cortex Inferior temporal cortex Thalamus + all other cortical regions Inferior frontooccipital fasciculus (IFOF) Lingual gyrus Orbitofrontal cortex* Thalamus + all other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined temporal tracts. *Orbitofrontal cortex was created by summing the medial orbitofrontal cortex and lateral orbitofrontal cortex. doi:10.1371/journal.pone.0103038.t003 PLOS ONE | www.plosone.org 3 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 4. Tractographical approach for parieto-occipital tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Optic radiation Pericalcarine cortex Eroded thalamus All other cortical regions Dorsal visual stream Pericalcarine cortex Inferior parietal cortex Thalamus Splenium of the corpus callosum Left lateral occipital cortex Right lateral occipital cortex* All other cortical regions Posterior corona radiata (PCR) (occipital) All occipital regions Cerebral peduncle All other cortical regions Posterior corona radiata (PCR) (parietal) All parietal regions Cerebral peduncle All other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined parieto-occipital tracts. *For the tract through the splenium of the corpus callosum, a callosal waypoint mask was also used. doi:10.1371/journal.pone.0103038.t004 2.1.6. Prematurity. Three of 16 SPD boys were born Image Registration Tool (FLIRT; www.fmrib.ox.ac.uk/fsl/flirt) prematurely, one at 32 weeks gestation and two at 34 weeks with 12-parameter linear image registration [23]. All diffusion- gestation. One of the 23 typically developing children was born weighted volumes were registered to the reference b = 0 s/mm prematurely, at 33 weeks gestation. These four subjects were found volume. To evaluate subject movement, we calculated a scalar to be in the middle of the distribution for global FA and mean FA parameter quantifying the transformation of each diffusion volume extracted from clusters of significantly affected voxels using TBSS to the reference. A heteroscedastic two-sample Student’s t-test for their respective groups, and therefore they were not considered verified that there were no significant differences between SPD, to be outliers [16]. None of the ASD subjects were born ASD, and TDC groups in movement during the DTI scan (p. prematurely. 0.05). The non-brain tissue was removed using the Brain Extraction Tool (BET; http://www.fmrib.ox.ac.uk/analysis/ 2.2. Image acquisition research/bet). FA was calculated using the FMRIB Software MR imaging was performed on a 3T Tim Trio scanner Library (FSL) DTIFIT function. 2.3.2. High angular resolution diffusion imaging (HARDI) (Siemens, Erlangen, Germany) using a 12-channel head coil. Structural MR imaging of the brain was performed with an axial and fiber tractography. The FSL bedpostx tool was used for HARDI reconstruction of the diffusion data, modeling multiple 3D magnetization prepared rapid acquisition gradient-echo (MP- RAGE) T1-weighted sequence (TE = 2.98 ms, TR = 2300 ms, fiber orientations per voxel, and thereby accounting for crossing TI = 900 ms, flip angle of 9u) with a 256 mm field of view fibers [24]. Probabilistic streamline tractography was performed (FOV), and 160 1.0 mm contiguous partitions at a 2566256 using FSL’s probtrackx2 to delineate white matter tracts of matrix. Whole-brain DTI was performed with a multislice 2D interest, using the strategies described in Tables 3–5 and single-shot twice-refocused spin-echo echo-planar sequence with illustrated in Figure S1. Seed, waypoint, termination, and 64 diffusion-encoding directions, diffusion-weighting strength of exclusion masks for tractography were primarily derived from b = 2000 s/mm , iPAT reduction factor of 2, TE/TR = 109/ the gray-white matter boundaries (GWB) of the 82 Freesurfer 8000 ms, averages = 1, interleaved 2.2 mm axial slices with no cortical and subcortical regions, which were automatically gap, and in-plane resolution of 2.262.2 mm with a 1006100 segmented on the T1-weighted MR images using Freesurfer matrix and FOV of 220 mm. An additional volume was acquired 5.1.0 [25] and registered using a linear affine transformation to with no diffusion weighting (b = 0 s/mm ). The total DTI diffusion space using FLIRT. The left and right cerebral peduncles acquisition time was 8.67 min. were manually defined for each subject. 2.3.3 Tract delineation. Subsequent to performance of probabilistic streamline fiber tractography, tract masks for every 2.3. DTI analysis 2.3.1. Pre-processing. The diffusion-weighted images were tract described in Tables 3–5 were separately generated for each corrected for motion and eddy currents using FMRIB’s Linear subject. Each mask was created by taking the intersection of the Table 5. Tractographical approach for frontal tracts. White matter tract Seed mask Waypoint and termination mask Exclusion mask Anterior thalamic radiation Medial orbitofrontal cortex Eroded thalamus All other gm regions (ATR) (medial orbitofrontal cortex) Anterior thalamic radiation Rostral middle frontal cortex Eroded thalamus All other gm regions (ATR) (rostral middle frontal cortex) Genu of the corpus callosum Left medial orbitofrontal cortex Right medial orbitofrontal cortex All other cortical regions (medial orbitofrontal cortex) Genu of the corpus callosum Left rostral middle frontal cortex Right rostral middle frontal cortex All other cortical regions (rostral middle frontal cortex) Anterior corona radiata (ACR) All frontal regions Cerebral peduncle All other cortical regions The Freesurfer seed, waypoint, termination, and exclusion masks used in fiber tractography to delineate examined frontal tracts. *For the tracts through the genu of the corpus callosum, a callosal waypoint mask was also used. doi:10.1371/journal.pone.0103038.t005 PLOS ONE | www.plosone.org 4 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD PLOS ONE | www.plosone.org 5 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 1. Examples of each delineated tract for a representative subject. Green masks represent frontal tracts, blue masks represent parietal-occipital tracts, and orange masks represent temporal tracts. The tracts are superimposed upon the T1 image, registered to diffusion space and with decreased opacity, of the representative subject. doi:10.1371/journal.pone.0103038.g001 binarized, thresholded, tractography-derived streamline map and tracts were individually assessed to confirm bilateral consistency a binary mask of FA.0.2. The streamline threshold used for and to evaluate hypothesized tract laterality. binarization was separately calculated for each streamline map, and equal to 1% multiplied by the maximum number of 2.4. Statistical analysis of group differences streamlines passing through any voxel in the map. This streamline For each tract, decreases in FA were separately assessed for the threshold was a consistent strategy of removing spurious stream- SPD and ASD cohorts relative to controls using one-tailed lines, while retaining most voxels contained within the desired permutation tests (n = 10,000) (adapted from [28]). Permutation white matter tract. The FA threshold further ensured that the testing was utilized, as it is a nonparametric method and thereby voxels contained within the mask were confined to white matter. does not assume normally distributed data. The true two-sample t Additionally, each tract mask for each subject was visually statistic was calculated for control FA vs. patient FA, and a two- inspected to confirm that the anatomy of each target tract was sample t statistic distribution was generated by permuting the accurately and consistently defined. control and patient labels 10,000 times, calculating a t-statistic White matter connectivity was calculated as the average FA value each time. The one-tailed p value was then calculated as the value within the delineated tract of interest. This measurement has number of permuted t statistic values lying below the true t been shown to be highly reproducible in cross-sectional [26] and statistic, divided by the number of permutations (10,000). Group longitudinal studies [27]. differences were assessed separately for each patient cohort relative Representative examples of each of the 15 delineated tracts are to controls at a false discovery rate (FDR) - corrected p value displayed in Figure 1. All masks used for tractography were the threshold (from p,0.05), with FDR correction applied separately GWBs of Freesurfer regions except for manually-defined cerebral to tracts within each region (separately for the temporal, parietal- peduncles and corpus callosum masks. Eroded thalamus masks occipital, and frontal tracts). Because the perceptual reasoning refer to an eroded version of the Freesurfer thalamus which was index (PRI) scores were significantly lower for the ASD cohort transformed using the fslmaths erode filtering operation with a box compared to the TDC and SPD subjects, a post-hoc group kernel of width 9, a step taken to prevent the thalamic mask from difference analysis was conducted while controlling for PRI. For overlapping the corpus callosum and resulting in spurious each tract, a general linear model (GLM) was fit to the data using interhemispheric streamlines. Except for callosal connections, PRI as a regressor, and permutation tests were performed in the each tract was delineated separately in both the left and right same way as described above, using t statistics for the group hemispheres. Following mask extraction (after thresholding by coefficient estimates from the GLM. Differences were again streamlines and FA), corresponding left and right hemisphere tract assessed using FDR correction within the temporal, parietal- masks were combined for subsequent analysis. The unilateral occipital, and frontal regions. Figure 2. Group differences between TDC, SPD, and ASD subjects in average FA within different temporal tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g002 PLOS ONE | www.plosone.org 6 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 3. Group differences between TDC, SPD, and ASD subjects in average FA within different parietal-occipital tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g003 five subtests of the SP (auditory, visual, tactile, inattention, 2.5. Cognitive associations multisensory integration) were investigated dimensionally across Pearson’s correlations of FA in the 15 examined tracts with the all individuals. Statistical significance was assessed at p,0.05 with VCI, PRI, WMI, PSI, the social component of the SCQ, and the Figure 4. Group differences between TDC, SPD, and ASD subjects in average FA within different frontal tracts. Crossbars correspond to group averages. Green asterisks depict significant group differences between ASD and TDC subjects, and red asterisks depict significant group differences between SPD and TDC subjects, FDR corrected at p,0.05. doi:10.1371/journal.pone.0103038.g004 PLOS ONE | www.plosone.org 7 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 6. Connectivity (FA) in all tracts. Tract TDC mean FA6SD ASD mean FA6SD P-val SPD mean FA6SD P-val Fusiform - amygdala 0.369360.0182 0.354660.0177 0.0112 0.373860.0203 0.2298 Fusiform - hippocampus 0.358760.0146 0.346960.0156 0.0078 0.355860.016 0.282 Uncinate fasciculus 0.342960.0156 0.337560.0137 0.118 0.338860.0099 0.1886 ILF 0.411460.0147 0.400860.0192 0.019 0.402860.0164 0.0468 IFOF 0.395660.0171 0.384460.0156 0.031 0.389360.0128 0.11 PTR (optic radiations) 0.409560.0113 0.402960.0127 0.0516 0.403260.0175 0.0814 Dorsal visual stream 0.415560.0147 0.405260.0179 0.0134 0.400960.0156 0.0028 Splenium of the CC 0.465860.0161 0.458960.0167 0.1012 0.451760.0238 0.0164 (lat occipital) PCR (occipital) 0.419460.0123 0.409360.0161 0.0144 0.408560.019 0.0198 PCR (parietal) 0.418260.0093 0.414260.0168 0.1338 0.412460.0178 0.1018 ACR 0.418260.0091 0.425960.0146 0.0508 0.413660.0156 0.1236 ATR (orbitofrontal) 0.362360.0137 0.362760.0137 0.4984 0.358160.014 0.1694 ATR (rostral middle frontal) 0.353060.0111 0.353260.0143 0.3616 0.345760.0105 0.0238 Genu of the CC (orbitofrontal) 0.436160.0224 0.433860.0179 0.3666 0.429660.0218 0.1916 Genu of the CC 0.410560.0196 0.407860.0208 0.4306 0.400860.0204 0.0674 (rostral middle frontal) The mean and standard deviation of FA within each tract for each group, with associated p values for group differences of the TDC cohort with either the SPD cohort or the ASD cohort. Bolded p values represent significant group differences at p,0.05, FDR corrected. doi:10.1371/journal.pone.0103038.t006 FDR correction across all 15 tracts. For tracts and cognitive/ or parietal PCR relative to TDC; however, there were strong behavioral metrics demonstrating significant associations across trends toward lower connectivity of the optic radiations in both the groups, post-hoc correlational analyses were conducted for the ASD and SPD groups relative to TDC. unilateral tract FA (left and right hemisphere independently) 3.1.3. Group differences of connectivity in frontal across groups, as well as unilateral and bilateral tract FA (left and tracts. Connectivity in the frontal tracts was not significantly right combined) for each cohort (TDC, SPD, and ASD) decreased for either the SPD or ASD cohorts, although the SPD independently. group showed trends towards decreased connectivity for all measured frontal tracts. Results 3.2. Unilateral versus bilateral white matter tracts 3.1. Group differences in white matter connectivity Homologous white matter tracts of the left and right cerebral Figures 2–4 depict group differences of structural connectivity hemispheres were combined for purposes of consolidation and denoted by fractional anisotropy (FA) in tracts predominantly improved statistical power. However, group differences were also involving the temporal, parietal-occipital, or frontal regions. computed unilaterally for each tract. In all cases, the results from Table 6 quantitatively details these group differences. We have bilateral tracts shown here agree with trends or statistically additionally added the results of mean diffusivity, axial diffusivity, significant group differences from the component unilateral tracts, and radial diffusivity for the three groups). with no appreciable hemispheric asymmetries found. 3.1.1. Group differences of connectivity in temporal tracts. Significantly impaired connectivity (lower FA) was 3.3. Accounting for perceptual reasoning index variation detected for the ASD cohort alone relative to the TDC cohort PRI was significantly lower in the ASD cohort compared to the in the fusiform-amygdala and fusiform-hippocampus tracts, the TDC subjects, thus group differences in connectivity were inferior fronto-occipital fasciculi (IFOF), and the inferior longitu- computed while controlling for PRI scores. After controlling for dinal fasciculi (ILF) (p,0.05, FDR corrected). The SPD cohort PRI and including FDR correction, connectivity for the ASD showed no significant differences in these tracts relative to the cohort was no longer significantly lower in the IFOF or ILF, but TDC cohort. There was no significant difference in connectivity of still demonstrated decreased FA in the fusiform -amygdala and the uncinate fasciculi of either the ASD or SPD cohorts relative to fusiform –hippocampus tracts. The results for the SPD subjects the controls. were unchanged when controlling for PRI, as expected since these 3.1.2. Group differences of connectivity in parietal- subjects did not demonstrate differences in PRI relative to TDC. occipital tracts. The SPD group alone showed significantly decreased connectivity in the splenium of the corpus callosum relative to the TDC cohort. Both the SPD and the ASD group 3.4. Cognitive associations Significant combined-group correlations were found between showed reduced connectivity relative to controls in the dorsal visual stream and the posterior corona radiata (occipital portion) WMI and the bilateral optic radiations (r = 0.41, p = 0.003) as well (all results with p,0.05, FDR corrected). as the bilateral PCR (occipital) (r = 0.49, p,0.001) (Figure 5). Neither the SPD nor ASD groups demonstrated significant These tracts both demonstrate left lateralized associations for the differences in the optic radiations (pericalcarine – thalamus PTR) combined groups. The SPD cohort alone demonstrates significant PLOS ONE | www.plosone.org 8 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 5. Combined-group associations between tract connectivity and WMI. The two bilateral tracts demonstrating significant associations between FA and WMI after FDR correction are displayed. Optic radiation: r = 0.41, p = 0.003. PCR (occipital): r = 0.49, p,0.001. Results of unilateral and individual group correlations are displayed in Table 7. doi:10.1371/journal.pone.0103038.g005 individual-group associations between WMI and FA in both of demonstrate right lateralized associations for the combined groups these bilateral tracts, while ASD demonstrates significant or trend- (Table 8). level associations (Table 7). Significant combined-group correlations were found between Significant combined-group correlations were found between the inattention measures of the Sensory Profile and the dorsal the social component of the SCQ and the bilateral fusiform - visual stream (r = 0.38, p = 0.006) as well as the bilateral PCR amygdala (r =20.44, p = 0.001) as well as the bilateral fusiform- (occipital) (r = 0.46, p = 0.001) (Figure 7). There is no strong hippocampus (r =20.39, p = 0.004) (Figure 6). These tracts both evidence of lateralization for the combined groups (Table 9). The association between inattention and FA in the bilateral PCR PLOS ONE | www.plosone.org 9 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 7. Associations between tract connectivity and WMI, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right PTR (optic radiation) All 0.41 0.003 0.36 0.009 0.32 0.020 TDC 20.08 0.715 0.07 0.753 20.14 0.522 ASD 0.54 0.048 0.53 0.054 0.25 0.380 SPD 0.62 0.010 0.35 0.188 0.73 0.001 PCR (occipital) All 0.49 ,0.001 0.50 ,0.001 0.32 0.022 TDC 0.22 0.320 0.39 0.077 20.07 0.746 ASD 0.46 0.101 0.40 0.161 0.30 0.299 SPD 0.67 0.004 0.68 0.004 0.56 0.024 The bilateral tracts demonstrating significant combined-group associations with WMI are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t007 (occipital) is consistent with our previously published findings in to TDC, but relatively unaffected in SPD. While both the ASD the combined SPD and TDC cohorts [16]. and the SPD participants demonstrate white matter pathology in the sensory processing regions of the dorsal visual stream and the Significant combined-group correlations were found between the auditory factor of the Sensory Profile and the bilateral PCR posterior corona radiata, only the SPD cohort demonstrates (occipital) (r = 0.42, p = 0.004) (Figure 8). There is no strong statistically significant differences in the splenium of the corpus callosum relative to the TDC cohort. These findings extend evidence of lateralization for the combined groups (Table 10). The association between this auditory measure and FA in the PCR is previous research using DTI in autism cohorts to include concurrent analysis of children that exhibit sensory processing consistent with our previously published findings in the combined SPD and TDC cohorts [16]. differences, but not the language and social deficits that The combined groups did not demonstrate significant correla- characterize a full ASD diagnosis. While the most extensive white matter alterations in the SPD tions between FA in the 15 investigated tracts and PRI, VCI, or subjects are observed in the parieto-occipital tracts, which subserve the other subscores of the Sensory Profile after correction for multiple comparisons. auditory, tactile, and visual perception and integration, this cohort demonstrates trends towards decreased connectivity compared to TDC in most measured tracts. It is also worth specific comment 3.5. Tract overlap that, while both the SPD and ASD cohorts were affected in these Some tracts demonstrated a significant fraction of shared voxels. fundamental sensory processing tracts, the FA in all but one of Figure 9 depicts the average fraction of each tract’s voxels that are these tracts trended lower for the SPD subjects relative to the ASD shared with every other investigated white matter tract. In the subjects. This difference may reflect the prominence of abnormal most extreme case, a subject average of 77% of the fusiform - sensory related behaviors, which is an inclusion criterion for SPD amygdala tract voxels were contained within the fusiform - group membership, whereas, in general, children with ASD are hippocampus tract. There were also significant spatial overlaps primarily characterized by profound social communication within the delineated parietal-occipital tracts, and between some deficits. In this sample, 65% of children with ASD scored in the parietal-occipital and temporal tracts. These results demonstrate definite difference (DD) range (.2 standard deviations from the that some tracts were not completely independent in the group mean) on the Sensory Profile Total Score and 57% were in the difference results above. Sections of the fusiform-amygdala tract DD range for the Auditory Processing Score. While many children that were independent of the fusiform-hippocampus, and sections with ASD have auditory, tactile and visuomotor processing of the fusiform-hippocampus that were independent of the challenges, these deficits are not as ubiquitous as in our SPD fusiform-amygdala tract, were separately assessed for group cohort. Our findings further suggest that sensory-based behavioral differences. Neither of the independent sections of these tracts deficits in both groups may be predicated on atypical conduction demonstrated statistically significant group differences, suggesting of information from unimodal to multimodal integration regions as that shared voxels drive the observed differences between the ASD well as inefficient transfer of information between hemispheres via cohort and controls. the corpus callosum for the SPD group. Perhaps the most striking finding is that, relative to the control Discussion group, the ASD cohort shows reduced structural connectivity in This study is the first to investigate white matter connectivity of the fusiform gyrus connections to the amygdala and hippocampus, both children with SPD and children with ASD relative to whereas children with SPD do not. These white matter pathways typically developing children. Diffusion MR fiber tracking was are thought to facilitate facial emotional processing, a core feature employed for the hypothesis-driven identification of specific white of autism and the domain of clinical divergence for ASD versus matter tracts. The results suggest both overlapping and divergent SPD [29]. In fact, a recent study reports that infants later white matter microstructural pathology affecting the two clinical diagnosed with autism show reduced attention to essential facial cohorts, with tracts traditionally associated with social emotional information with declining direct eye gaze as early as 2–6 month of processing being significantly affected for the ASD cohort relative age [30]. The neuroanatomy of facial emotion processing has been PLOS ONE | www.plosone.org 10 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 6. Combined-group associations between tract connectivity and SCQ-social. The two bilateral tracts demonstrating significant associations between FA and the social component of the SCQ after FDR correction are displayed. Fusiform-amygdala: r =20.44, p,0.001. Fusiform- hippocampus: r =20.39, p = 0.004. Results of unilateral and individual group correlations are displayed in Table 8. doi:10.1371/journal.pone.0103038.g006 intensively studied with repeated implication of the amygdala- autism, the right hippocampal fusiform tract was suggested to have fusiform system. Individuals with ASD have been reported to have smaller diameter axons corresponding with slower neural trans- less activation of subcortical regions including the amygdala and mission, which was thought to lead to secondary changes in the left fusiform gyrus during subliminal emotional face processing, a lack amygdala-fusiform and hippocampal-fusiform pathways [16]. By of expected volumetric correlation between the amygdala and contrast, a diffusion fiber tractography study of individuals with fusiform gyrus, as well as behavioral deficits in face recognition Williams syndrome (7q11.23 deletion) are reported to show that negatively correlate with left fusiform cortical thickness [31], elevations of FA in fusiform tracts [33]. Individuals with Williams [32]. In a DTI based analysis of adolescents and adults with Syndrome show a social phenotype that is in some ways opposite PLOS ONE | www.plosone.org 11 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 8. Associations between tract connectivity and the SCQ-social, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right Fusiform-amygdala All 20.44 0.001 20.27 0.048 20.37 0.006 TDC 20.39 0.066 20.18 0.423 20.29 0.178 ASD 20.40 0.137 20.34 0.220 20.18 0.515 SPD 20.32 0.233 0.03 0.907 20.50 0.049 Fusiform-hippocampus All 20.39 0.004 20.25 0.065 20.40 0.003 TDC 20.14 0.510 20.11 0.613 20.15 0.508 ASD 20.25 0.370 20.22 0.433 20.12 0.680 SPD 20.27 0.307 20.01 0.979 20.47 0.069 The bilateral tracts demonstrating significant combined-group associations with the social component of the SCQ are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t008 to the autism phenotype with increased attention to faces and behavioral correlates of altered connectivity is warranted. How- abundant social interest and drive. It is thus worthwhile to ever, these findings suggest a role for neuroimaging in under- consider social cognition, or facial emotion recognition specifically, standing the neural mechanisms that differentiate children with a as a continuous trait that might map directly to connectivity of the variety of domain specific deficits, including basic sensory fusiform tract to limbic structures. processing and social emotional processing. The role of develop- There are however additional farther reaching implications for ment and therapeutic interventions on these systems remains an fusiform connectivity disruptions with regard to language devel- open and important question to explore in these clinical cohorts as it is unclear whether these findings are primary or represent a opment. A theoretical model of audiovisual affective speech perception begins with input to primary auditory and visual cortex consequence of aberrant tract remodeling predicated on less [34]. The input module feeds information to the fusiform gyrus as practice of these skills from early infancy. well as the integration module of the superior and middle temporal As can be seen from the group comparison figures, while there cortex. The primary sensory cortices as well as the fusiform gyrus are clear and statistically significant group differences, there is also are reciprocally connected to the amygdala and insula, which considerable overlap in the measurements from tracts across all comprise the emotion module that guides emotional relevance and three groups: ASD, SPD, and TDC. This highlights the may facilitate the rapid recruitment of limbic brain regions by importance of a new direction for cognitive and behavioral visual inputs. Additional contextual information is brought in research based on the investigation of abilities as a continuous through connections with the memory module, including the measure across children rather than split by exceedingly broad hippocampus and parahippocampal gyrus. This framework is and overlapping clinical labels, a concept which has been useful in considering how autism social communication deficits formalized in the Research Domain Criteria (RDoC) Project may map to neuroanatomic networks. [42]. It is in this context that we frame our investigation of In addition to the fusiform connections, our ASD group was associations between cognitive measures and tract connectivity found to have reduced FA in the ILF and the IFOF. This is in line across all three study groups. The fusiform-amygdala and with previous reports, although there is considerable variability in fusiform-hippocampus tracts are the only two (out of 15) tracts to demonstrate significant associations with the social component the literature, likely resulting from group heterogeneity in terms of symptom variability, severity, and age of cohort [8], [35]–[38]. of the SCQ across groups with correction for multiple comparisons The ILF, or inferior longitudinal fasciculus, has been shown to (Figure 6, Table 8). Further investigation of laterality in these two directly connect the occipital cortex to the anterior temporal lobe tracts revealed that these associations are significantly right- and the amygdala. The IFOF originates in the visual cortex runs lateralized (Table 8), an observation which is consistent with the medially to the ILF and directly connects to the inferior frontal Conturo et al. (2008) [16] finding of primary right-lateralized and dorsolateral frontal cortex. In a study involving children with effects in these two tracts for ASD subjects. It is important to note visual perceptional impairment, decreased FA of the ILF that the associations between connectivity in these two tracts and correlated with impaired object recognition [39]. The IFOF likely the SCQ-social measure were not significant on an individual overlaps spatially and functionally with the ILF, and is thought to group basis. However, the individual group correlations all trended in the same (negative) direction as the combined-group be a tract that is relatively new evolutionarily due to its absence in animal brains [40]. In a large lesion-based study population, the correlations (Table 8). Connectivities (FA) in the optic radiation right IFOF in particular is implicated in rapid recognition of and PCR (occipital) were found to be significantly correlated with emotional facial expressions [41]. The finding of significantly WMI after FDR correction in the combined groups. Investigation reduced connectivity in ASD of the ILF and IFOF is in of these associations in the individual cohorts revealed strong associations with WMI for the SPD cohort bilaterally in both of concordance with the reduced connectivity seen specifically in the fusiform connections. What is most revealing is the relative these tracts. The ASD cohort alone also demonstrates significant preservation of these connections in our SPD cohort. Clearly or trend-level associations with WMI and connectivity in these additional investigation to understand the relationship between the tracts. Though the optic radiation and PCR (occipital) were the speed of information transmission in these tracts as well as the only two tracts that demonstrated significant associations with PLOS ONE | www.plosone.org 12 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Figure 7. Combined-group associations between tract connectivity and the inattention measure of the Sensory Profile. The two bilateral tracts demonstrating significant associations between FA and the inattention measure of the Sensory Profile after FDR correction are displayed. Dorsal visual stream: r = 0.38, p = 0.006. PCR (occipital): r = 0.46, p,0.001. Results of unilateral and individual group correlations are displayed in Table 9. doi:10.1371/journal.pone.0103038.g007 WMI after FDR correction, many of the other investigated tracts measures are consistent with our prior findings [16], we did not (including frontal, temporal, and parietal-occipital tracts) demon- find significant associations with the other Sensory Profile strated trend-level associations. This is consistent with reports from measures after correction for multiple comparisons. prior DTI studies of WMI that have found widespread associations There are important limitations to note for this study, which of WMI with white matter connectivity [43], [44]. While our should motivate further investigation. First, we have not deter- findings of significant associations between connectivity in the mined an optimal method for characterizing the sensory subtypes PCR (occipital) and the Sensory Profile auditory and inattention and distinguishing between hypo- or hyper-sensory sensitivity, nor PLOS ONE | www.plosone.org 13 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 9. Associations between tract connectivity and the inattention factor of the Sensory Profile, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right Dorsal Visual Stream All 0.38 0.006 0.34 0.012 0.27 0.052 TDC 20.16 0.479 20.09 0.688 20.19 0.408 ASD 0.37 0.197 0.04 0.885 0.37 0.195 SPD 0.25 0.358 0.05 0.858 0.40 0.121 PCR (occipital) All 0.46 0.001 0.37 0.006 0.41 0.002 TDC 0.00 1.000 0.06 0.772 20.05 0.836 ASD 0.63 0.015 0.44 0.112 0.63 0.017 SPD 0.41 0.117 0.19 0.488 0.54 0.029 The bilateral tracts demonstrating significant combined-group associations with the inattention measure of the Sensory Profile are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t009 do we have sufficient power in this study for sensory subtype group PCR. Despite these spatial overlaps, the group difference results analysis. We and many sensorimotor based researchers are are not identical between overlapping tracts and provide working to develop a phenotyping tool that maps to specific white separately valuable information about structural connectivity in matter tracts, and we hope to identify and characterize separate these subjects. Additional connectivity analysis, both structural and constructs of sensory deficits in larger cohorts going forward. functional, will shed additional light on specific regional contri- Second, tract overlap exists in our results. A significant portion of butions to the neural underpinnings of sensory and emotional the amygdala-fusiform tract is contained within the hippocampal- processing differences. Our investigation is also limited in fusiform tract. In addition, the ILF is partially contained within the generalizability, as all of the subjects were boys between the ages dorsal visual stream and the PTR is partially contained within the of 8 and 12 years in an effort to limit developmental confounds in Figure 8. Combined-group associations between tract connectivity and the auditory measure of the Sensory Profile. The bilateral tract demonstrating significant associations between FA and the auditory measure of the Sensory Profile after FDR correction are displayed. PCR (occipital): r = 0.42, p = 0.002. Results of unilateral and individual group correlations are displayed in Table 10. doi:10.1371/journal.pone.0103038.g008 PLOS ONE | www.plosone.org 14 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD Table 10. Associations between tract connectivity and the auditory factor of the Sensory Profile, for significantly associated tracts. r - bilateral p - bilateral r - left p - left r - right p - right PCR (occipital) All 0.42 0.004 0.33 0.017 0.37 0.007 TDC 20.33 0.129 20.20 0.353 20.27 0.216 ASD 0.60 0.023 0.38 0.185 0.62 0.021 SPD 0.41 0.114 0.20 0.465 0.49 0.056 The bilateral tracts demonstrating significant combined-group associations with the inattention measure of the Sensory Profile are displayed. Correlation coefficients and p values are displayed for these tracts, along with combined group unilateral associations, individual group bilateral associations, and individual group unilateral associations. doi:10.1371/journal.pone.0103038.t010 this small sample. The PRI scores of the ASD cohort were therefore needed to determine whether these findings generalize to significantly lower than that of the SPD and TDC cohorts; other ages, genders, and intellectual abilities. however, the most important group differences in structural Future research will include investigation of functional connec- connectivity between ASD and controls remained statistically tivity using resting state fMRI and magnetoencephalography significant after regressing out the effect of PRI. Further research is (MEG). The ROIs used to determine structural connectivity in this study can be used to assess differences in functional connectivity Figure 9. Average fraction of tract overlap. Color intensity corresponds to the subject average of the fraction of the voxels of the tracts on the vertical axis that are contained within the tracts on the horizontal axis. Tracts that are more than one-third contained in any other tract are indicated by an asterisk on the vertical axis. doi:10.1371/journal.pone.0103038.g009 PLOS ONE | www.plosone.org 15 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD between these same regions, with functional coupling hypothesized differences based on two-tailed permutation tests with FDR to be reduced where decreased structural connectivity was found correction for 15 comparisons. in this work. While prior studies have found relationships between (TIFF) functional connectivity and white matter volume in ASD [45]– Figure S3 Group differences of RD in all tracts. Units of [47], there have been no reported associations between functional diffusivity are in mm /sec. Asterisks indicate significant connectivity and DTI measures in SPD. differences based on two-tailed permutation tests with FDR We hope that by utilizing larger sample sizes and direct correction for 15 comparisons. assessment of auditory, tactile, visuomotor processing, we will be (TIFF) able to gain a deeper understanding of how neural circuitry differences map to clinically relevant challenges for individual Figure S4 Group differences of AD in all tracts. Units of children. The ultimate goal of this and future work is to guide diffusivity are in mm /sec. Asterisks indicate significant personalized treatments ranging from behavioral interventions and differences based on two-tailed permutation tests with FDR targeted psychopharmacology to cognitive training using child- correction for 15 comparisons. friendly video game platforms. (TIFF) Table S1 Group differences of FA between the ASD and Supporting Information SPD cohorts in ASD-affected temporal tracts. P values are Figure S1 Example ROIs for fiber tracking of the derived from one-tailed permutation tests for SPD FA . ASD FA. homotopic visual tract through the splenium of the Bolded p values indicate significant differences after FDR corpus callosum. Displayed is an axial slice from the FA image correction. of a representative subject with overlaid ROIs for probabilistic (XLSX) fiber tractography. The seed mask is the grey-white matter Table S2 Group differences of FA in all tracts, excluding boundary of the left lateral occipital cortex, and contains voxels SPD subjects with SCQ.15. P values are derived from one- from which 2000 streamlines each are initiated. The termination tailed permutation tests for TDC FA . SPD FA. Bolded p values mask is the grey-white matter boundary of the right lateral indicate significant group differences after FDR correction. occipital cortex, and causes streamlines to terminate upon (XLSX) encountering the mask. The termination mask and the corpus callosum are both used as waypoint masks, indicating that Acknowledgments streamlines need to pass through the corpus callosum and reach the termination mask in order to be retained. The exclusion mask We are grateful to our participants and their families for their time and is the union of the grey-white matter boundaries of all other support. cortical regions, and causes streamlines that encounter these voxels to be excluded. The displayed resultant tract is the result of Author Contributions probabilistic tractography under the previously described con- Conceived and designed the experiments: EJM PM JPO YSC. Performed straints and a subsequent streamline and FA threshold. the experiments: EJM AA SH JH SD. Analyzed the data: YSC JPO. Wrote (TIFF) the paper: YSC JPO EJM PM AA SH JH SD. Figure S2 Group differences of MD in all tracts. Units of diffusivity are in mm /sec. Asterisks indicate significant References 1. Ahn RR, Miller LJ, Milberger S, McIntosh DN (2004) Prevalence of parents’ autism: a study using diffusion spectrum imaging tractography. Psychiatry perceptions of sensory processing disorders among kindergarten children. research 192: 60–66. American Journal of Occupational Therapy 58: 287–293. 12. Thomas C, Humphreys K, Jung KJ, Minshew N, Behrmann M (2011) The anatomy of the callosal and visual-association pathways in high-functioning 2. Ben-Sasson A, Hen L, Fluss R, Cermak SA, Engel-Yeger B, et al. (2009) A meta- autism: a DTI tractography study. Cortex 47(7): 863–73. analysis of sensory modulation symptoms in individuals with autism spectrum 13. Aoki Y, Abe O, Nippashi Y, Yamasue H (2013) Comparison of white matter disorders. Journal of Autism and Developmental Disorders 39: 1–11. integrity between autism spectrum disorder subjects and typically developing 3. Brett-Green BA, Miller LJ, Gavin WJ, Davies PL (2008) Multisensory individuals: a meta-analysis of diffusion tensor imaging tractography studies. integration in children: a preliminary ERP study. Brain Research 1242: 283– Molecular autism 4(1): 25. 14. Conturo TE, Williams DL, Smith CD, Gultepe E, Akbudak E, et al. (2008) 4. Bundy AC, Shia S, Qi L, Miller LJ (2007) How does sensory processing Neuronal fiber pathway abnormalities in autism: an initial MRI diffusion tensor dysfunction affect play?. American Journal of Occupational Therapy 61: 201– tracking study of hippocampo-fusiform and amygdalo-fusiform pathways. J Int Neuropsychol Soc 14: 933–46. 5. American Psychiatric Association (1994) American Psychiatric Association 15. Smith CD, Lori NF, Akbudak E, Sorar E, Gultepe E, et al. (2009) MRI diffusion Diagnostic and statistical manual of mental disorders, Fourth Edition (DSM- tensor tracking of a new amygdalo-fusiform and hippocampo-fusiform pathway IV) Washington, DC. system in humans. J Magn Reson Imaging 29: 1248–61. 6. Marco EJ, Hinkley LB, Hill SS, Nagarajan SS (2011) Sensory processing in 16. Owen JP, Marco EJ, Desai S, Fourie E, Harris J, et al. (2013) Abnormal white autism: a review of neurophysiologic findings. Pediatr Res. 69(5 Pt 2): 48R–54R. matter microstructure in children with sensory processing disorders. Neuro- 7. Perou R, Bitsko RH, Blumberg SJ, Pastor P, Ghandour RM, et al. (2013) Mental Image: Clinical 2: 844–853. health surveillance among children–United States, 2005–2011. Morbidity and 17. Szatmary B, Izhikevich EM (2010) Spike-timing theory of working memory. mortality weekly report. Surveillance summaries (Washington, D.C.: 2002) 62 PLoS Comput. Biol. 6: e1000879.10.1371. Suppl 2: 1–35. 18. Wechsler D (2003) WISC-IV: Administration and Scoring Manual. Psycholog- 8. Travers BG, Adluru NA, Ennis C (2012) Diffusion tensor imaging in autism ical Corporation. spectrum disorder: a review. Autism Research 5: 289–313. 19. Dunn W, Westman K (1997) The sensory profile: the performance of a national 9. Alexander AL, Lee JE, Lazar M, Boudos R, DuBray MB, et al. (2007) Diffusion sample of children without disabilities. American Journal of Occupational tensor imaging of the corpus callosum in Autism. NeuroImage 34: 61–73. Therapy 51: 25–34. 10. Kumar A, Sundaram SK, Sivaswamy L, Behen ME, Makki MI, et al. (2009) 20. Eaves LC, Wingert HD, Ho HH, Mickelson EC (2006) Screening for autism Alterations in Frontal Lobe Tracts and Corpus Callosum in Young Children spectrum disorders with the social communication questionnaire. Journal of with Autism Spectrum Disorder. Cerebral Cortex 20(9): 2103–2113. Developmental and Behavioral Pediatrics 27 (Suppl. 2): S95–S103. 11. Lo YC, Soong WT, Gau SF, Wu YY, Lai MC, et al. (2011) The loss of 21. Lord C, Rutter M, Le Couteur A (1994) Autism diagnostic interview-revised: a asymmetry and reduced interhemispheric connectivity in adolescents with revised version of a diagnostic interview for caregivers of individuals with PLOS ONE | www.plosone.org 16 July 2014 | Volume 9 | Issue 7 | e103038 White Matter Connectivity in ASD vs SPD possible pervasive developmental disorders. Journal of Autism and Develop- matter tracts in children with autism spectrum disorder. Proc Natl Acad Sci U S mental Disorders 24: 659–685. A. 111(5): 1981–6. 22. Lord C, Rutter M, Goode S (1989) Autism diagnostic observation schedule: a 36. Jou RJ, Jackowski AP, Papademetris X, Rajeevan N, Staib LH, et al. (2011) standardized observation of communicative and social behavior. Journal of Diffusion tensor imaging in autism spectrum disorders: preliminary evidence of Autism and Developmental Disorders 19: 185–212. abnormal neural connectivity. Aust. N. Z. J. Psychiatry 45: 153–162.10.3109. 23. Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for 37. McGrath J, Johnson K, O’Hanlon E, Garavan H, Gallagher L, et al. (2013) the robust and accurate linear registration and motion correction of brain White matter and visuospatial processing in autism: a constrained spherical images. NeuroImage 17: 825–841. deconvolution tractography study. Autism Res. 6(5): 307–319. 24. Behrens TEJ, Berg HJ, Jbabdi S, Rushworth MF, Woolrich MW (2007). 38. Shukla DK, Keehn B, Muller RA (2011) Tract-specific analyses of diffusion Probabilistic diffusion tractography with multiple fibre orientations: What can tensor imaging show widespread white matter compromise in autism spectrum we gain?. Neuroimage 34: 144–155. disorder. J. Child Psychol. Psychiatry 52: 286–295. 25. Fischl B (2012) FreeSurfer. NeuroImage 62: 774–781. 39. Ortibus E, Verhoeven J, Sunaert S, Casteels I, de Cock P, et al. (2012) Integrity 26. Wakana S, Caprihan A, Panzenboeck MM, Fallon JH, Perry M, et al. (2007) of the inferior longitudinal fasciculus and impaired object recognition in Reproducibility of quantitative tractography methods applied to cerebral white children: a diffusion tensor imaging study. Dev Med Child Neurol 54: 38–43. matter. Neuroimage 36: 630–44. 40. Forkel J, Thiebaut de Schotten M, Kawadler J, dell’Acqua F, Danek A, et al. 27. Danielian LE, Iwata NK, Thomasson DM, Floeter MK (2010) Reliability of (2012) The anatomy of fronto-occipital connections from early blunt dissections fiber tracking measurements in diffusion tensor imaging for longitudinal study. to contemporary tractography. Cortex dx.doi.org/10.1016/j.cor- Neuroimage 49: 1572–1580. tex.2012.09.005. 28. Nichols T, Holmes A (2001) Nonparametric permutation tests for functional 41. Philippi CL, Mehta S, Grabowski T, Adolphs R, Rudrauf D (2009) Damage to neuroimaging: a primer with examples. Human Brain Mapping 15: 1–25. association fiber tracts impairs recognition of the facial expression of emotion. 29. Schultz RT (2005) Developmental deficits in social perception in autism: the role J Neurosci 29: 15089–15099. of the amygdala and fusiform face area. International Journal of Developmental 42. Cuthbert BN, Insel TR (2013) Toward the future of psychiatric diagnosis: the Neuroscience 23 (2): 125–141. seven pillars of RDoC. BMC Med 11: 126. 30. Jones W, Klin A (2013) Attention to eyes is present but in decline in 2-6-month- 43. Tang CY, Eaves EL, Ng JC, Carpenter DC, Mai X, et al. (2010) Brain networks old infants later diagnosed with autism. Nature 504(7480): 427–31. for working memory and factors of intelligence assessed in males and females 31. Dziobek I, Bahnemann M, Convit A, Heekeren HR (2010) The role of the with fMRI and DTI. Intelligence 38: 293–303. fusiform-amygdala system in the pathophysiology of autism. Archives of general 44. Nagy Z, Westerberg H, Klingberg T (2004) Maturation of white matter is psychiatry 67: 397–405. associated with the development of cognitive functions during childhood. J. 32. Kleinhans NM, Richards T, Johnson LC, Weaver KE, Greenson J, et al. (2011) Cogn. Neurosci 16: 1227–1233. fMRI evidence of neural abnormalities in the subcortical face processing system 45. Cherkassky VL, Kana RK, Keller TA, Just MA (2006) Functional connectivity in ASD. NeuroImage 54(1): 697–704. in a baseline resting-state network in autism. Neuroreport 17: 1687–1690. 33. Haas BW, Hoeft F, Barnea-Goraly N, Golarai G, Bellugi U, et al. (2012) 46. Just MA, Cherkassky VL, Keller TA, Kana RK, Minshew NJ (2007) Functional Preliminary evidence of abnormal white matter related to the fusiform gyrus in Williams syndrome: a diffusion tensor imaging tractography study. Genes Brain and anatomical cortical underconnectivity in autism: Evidence from an FMRI Behav 11: 62–68. study of an executive function task and corpus callosum morphometry. Cerebral 34. Jansma H, Roebroeck A, Mu¨nte TF (2013) A network analysis of audiovisual Cortex 17: 951–961. affective speech perception. Neuroscience 256: 230–41. 47. Kana RK, Keller TA, Cherkassky KL, Minshew NJ, Just MA (2006) Sentence 35. Koldewyn K, Yendiki A, Weigelt S, Gweon H, Julian J, et al. (2014) Differences comprehension in autism: Thinking in pictures with decreased functional in the right inferior longitudinal fasciculus but no general disruption of white connectivity. Brain 129(Pt 9): 2484–2493. PLOS ONE | www.plosone.org 17 July 2014 | Volume 9 | Issue 7 | e103038

Journal

PLoS ONEPubmed Central

Published: Jul 30, 2014

References