Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1007/s11682-017-9678-y
ORIGINAL RESEARCH
93.8
94.2
95.5
95.1
92.3
94.3
92.6
92.7
92.9
93.8
93.8
92.9
scanner and image acquisition parameters varied across
ABIDE acquisition sites (see Supplementary Table 2 for sites
Mean (SD), Range
[7 scores missing]
fcon_1000.projects.nitrc.org/indi/abide/).
ADOS Total
-
-
-
California, San Diego. High-resolution structural images were
0.050 (0.021), 0.0190.098 acquired with a standard FSPGR T1-weighted sequence (TR:
0.045 (0.012), 0.0240.067
t(51) = 0.027, p = 0.979 11.08 ms; TE: 4.3 ms; flip angle: 45; FOV: 256 mm; matrix:
Mean (SD), Range;
3.25 ms; flip angle: 7; FOV: 256 mm; matrix: 256 256; 128
slices; 1.3mm3 resolution). Functional T2-weighted images were
t(139) = 0.048, p = 0.962
obtained in a 6 min scan (TR: 2000 ms; TE: 15 ms; slice thick-
13.93 (2.43), 9.217.7
13.83 (2.26), 8.717.6
13.34 (2.4), 7.2617.5
13.7 (2.67), 8.3917.9
Mean (SD), Range;
ness: 4 mm; flip angle: 90; FOV: 240 mm; matrix: 240 240).
13.4 (2.6), 817.9
t-test, p-value*
Data preprocessing
Age
36 Female
14 Female
11 Female
5 Female
6 Female
6 Female
100
100
100
100
0
0
147
184
110
26
30
25
59
27
42
38
82
N
Participant demographics
ASD
ASD
ASD
ASD
ASD
ASD
TD
TD
TD
TD
TD
TD
(0.008 < f < 0.08 Hz; Power et al. 2014; Satterthwaite et al.
ABIDE-EO
ABIDE-EC
SDSU-EO
Sample
cerebral white matter and lateral ventricles, and their deriva- Analyses were also performed both with and without global
tives. Individual participant masks were created to remove signal regression (GSR), given the ongoing debate highlighting
signal from cerebral white matter and lateral ventricles using both the virtues of GSR in denoising (Power et al. 2014) and its
FSLs FAST automated segmentation (Zhang et al. 2001). potential drawbacks, such as the creation of spurious anti-
Motion was quantified as framewise displacement (FD) correlations (Murphy et al. 2009), which may confound
between two consecutive time points (calculated based on six between-group comparisons (Abbott et al. 2015; Gotts et al.
dimensional rigid-body motion parameters). Any time point 2013). Findings from these additional analyses are presented
with FD > 0.2 mm as well as the following two time points in Supplementary Figs. 1 and 2 for completeness.
were censored, or Bscrubbed^ (Power et al. 2014). Participants
with fewer than 80% of time points remaining after censoring
were excluded from analyses. Average head motion, defined as Results
the root mean square of displacement (RMSD), did not signifi-
cantly differ between groups (ASD vs. TD). Analyses with GSR and ReHo standardization
Regional homogeneity Between-group effects for the analyses with GSR and ReHo
standardization for each subsample are presented in Fig. 1a
Regional homogeneity implements Kendalls coefficient of con- (for cluster listings see Supplementary Table 3). For the ReHo
cordance (KCC) and relies on rank order correlations between pipeline with GSR, group differences in the ABIDE-EO sub-
time series to assess the homogeneity of a given center voxel sample included local overconnectivity (ASD > TD) in right
and its neighboring voxels. ReHo can thus be considered a middle temporal gyrus, and across left calcarine and bilateral
measure of local connectivity (Anderson et al. 2014; Lopez- lingual gyri, and underconnectivity in bilateral thalamus, right
Larson et al. 2011). KCC within a cluster of voxels is equal to insula and inferior frontal gyrus, and left posterior cingulate
the parameter W, which ranges from 0 to 1, with higher values cortex and precuneus. The ABIDE-EC analysis yielded local
indicating greater homogeneity (Zang et al. 2004): overconnectivity in the ASD group across bilateral precuneus,
2 left middle temporal gyrus, left superior frontal gyrus, and
Ri 2 n R right angular/middle temporal gyrus, contrasted by local
W underconnectivity across right paracentral and postcentral gy-
1 2 3
K n n ri, right cuneus and lingual gyrus, right superior and middle
12
occipital gyri, left precentral and postcentral gyri. For the
In a previous study (Maximo et al. 2013), we tested the effect NYU subsample, group differences included local
of varying cluster sizes of 7, 19, and 27 voxels on ReHo in overconnectivity in left calcarine, lingual, and middle tempo-
comparisons between TD and ASD samples, and observed most ral gyri, and local underconnectivity across left superior me-
robust group differences for 27 voxels. For the present study, we dial and mid orbital gyri, left anterior cingulate cortex, right
therefore computed ReHo for a cluster size of 27 voxels. A gray- rolandic operculum, and right superior and middle frontal gy-
matter mask was used to avoid partial-volume effects. Voxel- ri. For the SDSU subsample, local overconnectivity in the
wise ReHo maps were obtained for each participant using ASD groups was detected in left calcarine cortex and cuneus,
AFNIs 3dReHo for a cluster size of 27 voxels. In each partic- right superior and middle temporal gyri, right temporal pole,
ipant, KCC for each voxel was then standardized to a ReHo z- and right amygdala, whereas local underconnectivity occurred
score by subtracting the mean KCC across all brain voxels and in bilateral paracentral as well as anterior and middle cingulate
dividing it by the standard deviation. Between-group differences cortices, right SMA, and left cerebellum. The Grand Total-EO
(ASD vs. TD) were examined with two-sample t-tests. Monte- subsample yielded local overconnectivity in calcarine cortex,
Carlo simulations were applied to correct for multiple compar- lingual gyrus and precuneus bilaterally, right superior tempo-
isons (Forman et al. 1995), using AFNIs 3dClustSim (version ral and supramarginal gyri, and bilateral middle temporal gy-
post May 2015), to obtain a corrected significance level of rus, contrasted by local underconnectivity across bilateral
p < .05 (using a voxelwise threshold of p < .01 uncorrected middle, anterior, and posterior cingulate cortices, bilateral
and a minimum cluster volume of 37 voxels for a corrected thalamus, left cerebellum, and rolandic operculum bilaterally.
threshold at p < .05). The Grand Total-EO + EC ReHo analysis showed local
While KCC standardization, as described above, is a com- overconnectivity in the ASD group across bilateral middle
mon default procedure in the ReHo literature because it reduces temporal gyrus, left calcarine cortex and lingual gyri, right
artifactual effects of individual variability (e.g., differences in cerebellum, left fusiform gyrus, left precuneus, and right
motion across participants), it renders the technique insensitive supramarginal gyrus. Local underconnectivity was detected
to potential global group differences in local connectivity. We across bilateral middle and anterior cingulate cortices, bilateral
therefore also ran parallel analyses without standardization. thalamus, and right rolandic operculum.
Brain Imaging and Behavior
Fig. 1 Surface renderings of regional homogeneity differences between results from pipeline with global signal regression and standardized
ASD and TD groups for each subsample (a), of differences between ReHo are shown (see Supplementary Figure 1 for findings from other
ABIDE eyes open and eyes closed cohorts shown separately for ASD pipelines). The two columns on the left depict the left hemisphere, the two
and TD groups (b), and of cluster overlap between subsamples (c), columns on the right the right hemisphere (p < .05, corrected)
showing only independent subsamples (excluding Grand Total). Only
Brain Imaging and Behavior
Note that Grand Total is included here because it represent- gyri for ABIDE-EC; overconnectivity in bilateral middle tem-
ed the largest available sample, although it should be under- poral gyrus and underconnectivity in thalamus for Grand
stood that this sample was inclusive of other subsamples and Total-EO; overconnectivity in visual regions and
the detected effects thus reflect a summary of those subsam- underconnectivity in thalamus for Grand Total-EO + EC).
ples rather than an independent finding. Analyses comparing ABIDE EO to EC in both ASD and TD
groups yielded similar patterns of local over- and
Analyses without GSR and standardization underconnectivity, albeit to a lesser extent. Nevertheless,
when sites contributing fewer than 10 participants were re-
Results for analyses without GSR and without ReHo stan- moved from multi-site subsamples, largely similar patterns
dardization were mostly similar to those from standardized of effects were retained (Supplementary Fig. 4).
ReHo with global signal regression pipeline, with some ex-
ceptions. For the Grand Total-EO + EC subsample, local
overconnectivity in left middle temporal gyri, left lingual gy- Discussion
rus, and left precuneus, and underconnectivity in bilateral thal-
amus was not seen in the non-GSR non-standardized analysis, Despite relatively large, high-quality subsamples, which were
and overconnectivity in right middle frontal gyrus and matched for head motion and available demographics,
underconnectivity in left rolandic operculum were detected between-group effects differed across datasets, suggesting that
solely in this analysis (see Supplementary Fig. 1a and ReHo may be highly sensitive to site and cohort variability.
Supplementary Table 4 for complete details). Results for anal- Differences in sample size across datasets could explain vari-
yses without GSR but with ReHo standardization were largely ations in overall extent of significant between-group effects,
similar to those without ReHo standardization and are present- which were in fact modest, but not the many differences in
ed in Supplementary Fig. 1b (for cluster listings, see regional patterns of such effects.
Supplementary Table 5). Tw o r e g i o n a l e f f e c t s b o t h r e f l e c t i n g l o c a l
overconnectivity in posterior, visual regions in ASD were
Eye status detected across all samples (except ABIDE-EC) for both the
GSR and the non-GSR pipelines, as well as Grand Total-EO
Two-sample t-tests directly comparing ABIDE eyes open and with additional site and motion covariates. However, the exact
eyes closed subsamples yielded extensive regions of signifi- location of these between-group effects varied between sam-
cantly greater local connectivity for EO compared to EC par- ples (Fig. 1c). Medial clusters peaked around the calcarine
ticipants in occipital regions extending into temporal and pa- fissure (V1) in most samples. Additional posterior clusters
rietal lobes, as well as in mostly ventral frontal and anterior detected in all samples were located in lateral temporo-
cingulate regions. Reduced local connectivity for the same parietal cortex (except for NYU, and for ABIDE-EC with
contrast was seen in mid-cingulate, paracentral, and (mostly added site and motion covariates). A further effect with rela-
medial) temporal regions. These patterns were mostly similar tively high consistency (detected in all samples except NYU)
in ASD and TD groups (see Fig. 1 and Supplementary was local underconnectivity in cingulate cortex. Again, exact
Table 3; for findings from non-GSR and non-standardized locations of this effect varied greatly across samples.
analyses, see Supplementary Fig. 2 and Supplementary Surprisingly, the robust local overconnectivity in right pre-
Tables 4 and 5). frontal cortex reported in the large-sample original ABIDE
study (Di Martino et al. 2014) could not be replicated. A
Analyses with added site and motion regressors few methodological differences may have played a role: Di
Martino and colleagues excluded sites with N < 6, but includ-
To examine effects driven by site-specific sample sizes and ed datasets with < 80% remaining time points after censoring.
motion, two-sample t-tests were calculated with site and head Censoring was only performed on time points of motion peaks
motion (RMSD) as nuisance covariates. These analyses in their study, whereas we additionally censored two subse-
showed similar patterns of effects (Supplementary Fig. 3), quent time points. More importantly, the much wider age
with the exception of ABIDE-EO, which had the lowest ratio range in Di Martino et al. (764 years vs. 718 years in ours)
of participants to number of sites. Specifically, local may account for differences in findings. Related to this, Dajani
overconnectivity (ASD > TD) was found only in right middle and Uddin (2015) reported local overconnectivity effects in
temporal gyrus. For the ABIDE-EC, Grand Total-EO, and right prefrontal cortex that were age-specific. However, their
Grand Total-EO + EC subsamples, results were consistent finding of a small right inferior frontal overconnectivity clus-
with primary analyses with the exception of small clusters of ter only in 711 year-old children with ASD, but neither in
local connectivity that did not survive correction (e.g., local adolescents nor in adults, could not account for the differences
overconnectivity in left middle temporal and superior frontal between our findings and those by Di Martino et al. (2014).
Brain Imaging and Behavior
Note that samples in Dajani and Uddin (2015) were exclusive- we also analyzed all data without any steps that would have
ly drawn from the NYU data in ABIDE, resulting in small resulted in such normalization (i.e., without standardization
subsamples of 20 participants per group in each age bin, and without GSR, which has also been shown to distort group
and that data from eyes open and eyes closed conditions were differences in some instances (Abbott et al. 2015; Gotts et al.
combined in these samples (see below). Itahashi et al. (2015) 2013). Although between-group effects were overall slightly
reported extensive local overconnectivity in right prefrontal less extensive, most regional effects were similar between
cortex in adults with ASD (ages 1950 years), consistent with standardized and non-standardized pipelines. In particular,
the location of findings in Di Martino et al. (2014). Since the even without standardization, there was a mix of local
finding by Itahashi et al. comes from a large and independent under- and overconnectivity effects, indeed with a slight pre-
sample (not included in ABIDE), it may provide a replication dominance of the former. This mixture of connectivity effects
of the original ABIDE findings however, with the potential was also shown in analyses including site and head motion as
caveat that these effects may be found only in adults. If so, this nuisance covariates and excluding sites that contributed fewer
caveat could in turn explain why no corresponding effect was than 10 participants. This corroborates that general local
detected in our study, which was limited to children and ado- overconnectivity in ASD at the coarse spatial scale of fMRI
lescents under the age of 18 years. ReHo does not exist.
Starting with Belmonte et al. (2004) and Just et al. (2004), the Findings of atypically reduced local ReHo in cingulate gyrus
idea of a general principle of atypically reduced long-distance in ASD were relatively consistent across samples, although
connectivity but increased local connectivity in ASD has sur- exact loci differed substantially. Underconnectivity in anterior
vived through repeated cross-citations in many reviews (e.g., cingulate cortex (ACC) was observed in Grand Total as well
Maximo et al. 2014; Minshew and Williams 2007; Vissers as SDSU and NYU eyes open samples (although not in the
et al. 2012; Wass 2011), based on surprisingly slim evidence. ABIDE-EO and EC cohorts). These findings are consistent
The hypothesis of general long-distance underconnectivity in with evidence of reduced gray matter in ACC from voxel-
ASD has been questioned elsewhere (Mller et al. 2011; Nair based morphometry, as reported in a voxelwise meta-
et al. 2014; Rudie and Dapretto 2013; Supekar et al. 2013). analysis (Yang et al. 2016), as well as cytoarchitectonic anom-
Functional MRI ReHo studies, while reporting many often alies in ACC of postmortem brains from children and adults
inconsistent regional findings, have in fact given a clear an- with ASD (Simms et al. 2009; Uppal et al. 2014).
swer to the second part of the claim: There is no general local Reduced distal functional connectivity between medial pre-
overconnectivity in ASD. Instead every single autism ReHo frontal cortex/ACC and posterior cingulate cortex (PCC)/
study published to date (Dajani and Uddin 2015; Di Martino precuneus is one of the best-replicated findings in ASD
et al. 2014; Itahashi et al. 2015; Jiang et al. 2015; Maximo (Abbott et al. 2015; Assaf et al. 2010; Burrows et al. 2016;
et al. 2013; Paakki et al. 2010; Shukla et al. 2010) has detected Doyle-Thomas et al. 2015; Monk et al. 2009; von dem Hagen
region-specific mixtures of atypically increased and reduced et al. 2013; Washington et al. 2013). In light of the findings by
local connectivity in ASD. Burrows et al. (2016), this underconnectivity can be
However, there are a few caveats to be considered. Aside interpreted as reduced interplay between self-referential and
from the coarse spatial scale at which ReHo fMRI assesses other-referential processing (thinking about oneself vs. think-
local connectivity (see below BLimitations and perspec- ing about other people), which may be considered a funda-
tives^), the ReHo pipeline implemented in almost all of the mental component of social cognition. Moreover, in middle
cited studies included a standardization step, which converts cingulate cortex (MCC), hypothesized to be crucial for cogni-
KCC into z-scores, normalizing effects to a distribution tive processes involved during social exchanges and perspec-
around zero in each participant. While this has advantages tive-taking, differential activity has been shown to be dimin-
with respect to differing data quality and the detection of re- ished in ASD (Lu et al. 2015). Our additional finding of
gionally specific effects, any global differences between a reduced local connectivity both in anterior and more posterior
clinical and a control population will be lost. It is therefore sections of the cingulate gyrus further suggests that reduced
theoretically possible that mixed between-group effects re- interplay between the two zones is accompanied by less coor-
ported in all of the above studies were artifacts of ReHo stan- dinated activity within each of them. Aside from other-refe-
dardization. Maximo et al. (2013), however, showed similar rential thought, PCC is considered to be important for arousal
regional patterns of mixed over- and underconnectivity in and awareness, balance between internal and external atten-
ASD even in the absence of standardization. Nonetheless, tion, and environmental change detection (Leech and Sharp
given the relatively small sample size in this earlier study, 2014).
Brain Imaging and Behavior
The ACC also plays a crucial role as a hub of the salience including some non-visual ones (Jao Keehn et al. 2016; Kana
network (SN), which provides a bridge between cerebral cor- et al. 2006; Shen et al. 2012).
tical networks and limbic and autonomic systems that are
crucial for homeostatic, emotional, and visceral functions Eye status has dramatic effect on ReHo patterns
(Dosenbach et al. 2007; Seeley et al. 2007). Furthermore,
the SN acts as a pivot in switching between task-positive Some previous resting state functional connectivity MRI (rs-
and task-negative states (Sridharan et al. 2008), thus being a fcMRI) studies of ASD have included EO and EC data in com-
prime modulator of cognitive state (Dosenbach et al. 2007; bination (e.g., Cerliani et al. 2015; Dajani and Uddin 2015; Di
Seeley et al. 2007). Disturbances of the SN have been found Martino et al. 2014; Itahashi et al. 2015). Our findings, however,
to be associated with behavioral impairment (Greicius 2008; suggest that eye status during rest may have dramatic effects on
Uddin et al. 2013). In ASD, decreased activation of SN re- local synchronization of activity across almost the entire brain.
gions has been observed during inhibition tasks (Agam et al. First, between-group effects in ABIDE-EC diverged greatly
2010; Kana et al. 2007) and in association with skin conduc- from the three EO samples, with bilateral local
tance response (Eilam-Stock et al. 2014), presumably indicat- underconnectivity in visual cortex close to V1 (similar to find-
ing atypical autonomic functioning. Interconnectivity within ings from Itahashi et al. (2015)), and extensive overconnectivity
the SN and with the amygdala has been found to be reduced in in bilateral precuneus and PCC for EC. Direct comparison be-
several studies of ASD (Abbott et al. 2015; Ebisch et al. 2011; tween EO and EC samples showed robust effects of eye status
Kana et al. 2007; von dem Hagen et al. 2013). on ReHo patterns, with much greater local functional connec-
tivity for EO in extensive posterior (visual) and some frontal
regions, contrasted by extensive clusters of greater ReHo for
Visual cortex is locally overconnected EC in mostly limbic and subcortical regions (mid-portions of
the cingulate gyrus, medial temporal lobe, thalamus). Similar
Local overconnectivity in visual cortex was one of the patterns were observed in both TD and ASD groups (Fig. 1b;
two broadly consistent findings across multiple datasets, Supplementary Fig. 3b). Although EO and EC conditions in-
including a Grand Total (EO + EC) with N = 331. cluded different cohorts (no participants scanned under both EO
Although as mentioned, previous ReHo studies have dif- and EC conditions were included), samples were well matched
fered, the finding is in agreement with some reports on and it is unlikely that the robust differences detected by us were
local connectivity in ASD (Keown et al. 2013; Maximo cohort effects. Several previous rs-fcMRI studies have directly
et al. 2013; Washington et al. 2013). The divergent find- compared EO and EC states. The regional patterns of our
ing in Di Martino et al. (2014) may be partly attributed to findings were largely consistent with those reported by Liu
the inclusion of eyes-closed data (see below). However, et al. (2013) and are probably related to similar patterns ob-
since we detected small local overconnectivity clusters in served for the amplitude of low frequency fluctuations (Liu
visual cortex even in the Grand Total sample that included et al. 2013; Wang et al. 2015). Additionally relevant to effects
EC data, other methodological differences mentioned ear- in visual cortex observed by us, Zou et al. (2009) found that
lier (e.g., the much wider age range up to 64 years in Di thalamic connectivity with visual regions was more strongly
Martino et al.) were probable additional factors. negative for EC than for EO.
Overconnectivity in posterior cortices with visual functions In a first approach, the effects of EO status on activity in
may relate to extensive evidence suggesting a potential spe- visual cortex may seem trivial, seemingly indicating that EC
cial status of vision in the uneven neuropsychological profile may be preferable, as it avoids visual stimulation that may
commonly seen in ASD. This includes islands of atypically confound intrinsic functional connectivity (iFC) effects.
enhanced visual function (Simmons et al. 2009), as observed However, aside from low-pass filtering applied here (and
in visual search (Joseph et al. 2009; Kaldy et al. 2013) and in commonly in the iFC literature), which will minimize any
some studies on the embedded figures test (Eussen et al. 2016; effects of visual stimulus processing in higher frequency do-
Horlin et al. 2016; Keehn et al. 2009). Numerous studies of mains, additional considerations speak against such a simple
vision in ASD have suggested a preference for local process- conclusion. Empirical evidence shows that EO states, espe-
ing, possibly at the expense of global gestalt processing cially when eyes are fixated (on a cross), have greater test-
(reviewed in Dakin and Frith 2005), in agreement with the retest reliability for BOLD effect and cerebral blood flow
hypothesis of weak central coherence (Happe 1999). measurements than EC states (Patriat et al. 2013; Zou et al.
Another explanatory proposal has been generally enhanced 2015), suggesting that EO states are better controlled whereas
visual perception in ASD (Mottron et al. 2006). The special EC states are subject to undesirable variability. The most ob-
cognitive-behavioral status of vision in ASD is supported on vious example is the onset of drowsiness and sleep, which is
the neurobiological level by evidence of atypical participation difficult to monitor or prevent in EC conditions. Onset of sleep
of visual cortices across various tasks (Samson et al. 2012), may be associated with substantial changes in BOLD
Brain Imaging and Behavior
correlations (Spoormaker et al. 2010; Tagliazucchi et al. differences in findings may relate to cohort differences.
2012). Even beyond sleep, it is possible that mind wandering Admittedly, the included datasets were not ideally suited
may be more intense in EC than in EO conditions because the for the purposes of the present study. Unfortunately, no
visual link to the outside world that is typically maintained dataset exists that would even approach what would be
throughout the awake state is interrupted. Notably, we found optimal (i.e., very large samples of ASD and TD partici-
that ReHo was higher in the two main nodes of the default pants, each scanned multiple times on different scanners
mode network (PCC, medial prefrontal cortex) in EO than in and with different imaging protocols, both with eyes open
EC datasets, possibly indicating that an expected mental de- and with eyes closed). However, we carefully matched
fault state was maintained more reliably in the EO condition samples group-wise and across sites, covarying for poten-
(for comparison of dynamic changes within the default mode tially confounding factors, and applied strict quality
network between ASD and TD cohorts, see Falahpour et al. control criteria to minimize effects of cohort heterogeneity.
2016). In addition, acquisition of low-motion fMRI data in the
awake state was almost exclusively limited to high-
Limitations and perspectives functioning individuals with ASD across sites. The find-
ings presented here may thus not reflect local functional
As mentioned, fMRI ReHo can assess local connectivity connectivity at the lower end of the spectrum.
only at a relatively coarse spatial scale (9mm3 for each
27-voxel cluster). It can therefore not be directly inferred
what cytoarchitectonic patterns (and anomalies) described Conclusions
in postmortem studies of ASD (reviewed in Palmen et al.
2004) may be reflected in atypically increased or decreased Our study shows that local connectivity between-group effects
ReHo (cf. Schumann and Nordahl 2011). Specifically, it from rs-fMRI are sensitive to site and cohort differences, even
may be tempting to relate local BOLD correlations to hy- when data are selected for high quality and low motion and
potheses of minicolumnar anomalies (Casanova et al. 2006) when groups are tightly matched for demographics and motion.
and excitation/inhibition imbalance (Nelson and Valakh However, local overconnectivity in posterior (mostly visual)
2015; Rubenstein and Merzenich 2003). Reduced regions and underconnectivity in cingulate cortex were detected
intercolumnar inhibition (and thus greater neuronal cross- both in some smaller ASD samples acquired under controlled
excitation) might in principle be associated with greater conditions and large Grand Total samples with greater site var-
synchronization of local neuronal activity. However, iability, lending confidence to the finding. Some inconsistencies
supportive evidence of such a link from combined MR in overall patterns of findings and the surprising variability of
spectroscopy of gamma-aminobutyric acid (GABA) and exact loci of similar findings across samples may be in part
resting state fMRI has, to our knowledge, only come from explained by robust differences for data acquired in eyes open
two studies in motor cortex (Sampaio-Baptista et al. 2015; vs. eyes closed conditions. Our study generates two recommen-
Stagg et al. 2014), and great caution is warranted in the dations: First, careful attention needs to be paid to between-site
absence of more direct evidence. factors of variability in the use of large consortium datasets in
There were trade-offs between data quality (low motion) fcMRI studies, especially those examining local connectivity.
and loss in sample size. To maintain overall large sample Sample size alone will not directly translate into confidence in
sizes, sites contributing small numbers of datasets were in- findings. Second, mixing of data acquired under eyes open and
cluded in some analyses. Differences in sample size and sta- eyes closed conditions can confound findings because between-
tistical power across different analyses may have affected the group differences in some regions may have opposite polarity
findings, yet additional analyses covarying sample size and for the two conditions.
head motion showed that similar patterns of effects were
retained.
Compliance with ethical standards All procedures performed in stud-
Furthermore, since each site contributed different
ies involving human participants were in accordance with the ethical
>cohorts of participants, site-specific differences could standards of the appropriate institutional research boards and with the
not be distinguished from those related to the known 1964 Helsinki declaration and its later amendments or comparable ethical
heterogeneity of ASD. If idiopathic ASD, the population standards.
studied here, is understood as a clinical umbrella label for
Funding This study was supported by the National Institutes of Health
possibly hundreds of etiologically distinct rare disorders
R01 MH081023 (PI: RAM), K01 MH097972 (PI: Inna Fishman), and
(Geschwind and State 2015), complete convergence across IMSD R25GM058906.
cohorts may in fact not be expected. Sites differed in their
exact inclusionary and exclusionary criteria (Supplementary Conflict of interest All authors declare that they have no conflict of
Table 1), and despite our matching procedures, some interest related to the study presented here.
Brain Imaging and Behavior
Just, M. A., Cherkassky, V. L., Keller, T. A., & Minshew, N. J. (2004). principles of autistic perception. Journal of Autism and
Cortical activation and synchronization during sentence comprehen- Developmental Disorders, 36(1), 2743.
sion in high-functioning autism: evidence of underconnectivity. Mller, R.-A. (2007). The study of autism as a distributed disorder.
Brain, 127(Pt 8), 18111821. Mental Retardation and Developmental Disabilities Research
Kaldy, Z., Giserman, I., Carter, A. S., & Blaser, E. (2013). The mecha- Reviews, 13, 8595.
nisms underlying the ASD advantage in visual search. Journal of Mller, R.-A., Shih, P., Keehn, B., Deyoe, J. R., Leyden, K. M., &
Autism and Developmental Disorders, 46, 1513. Shukla, D. K. (2011). Underconnected, but how? A survey of func-
Kana, R. K., Keller, T. A., Cherkassky, V. L., Minshew, N. J., & Just, M. A. tional connectivity MRI studies in autism spectrum disorders.
(2006). Sentence comprehension in autism: thinking in pictures with Cerebral Cortex, 21(10), 22332243.
decreased functional connectivity. Brain, 129(Pt 9), 24842493. Murphy, K., Birn, R. M., Handwerker, D. A., Jones, T. B., & Bandettini,
Kana, R. K., Keller, T. A., Minshew, N. J., & Just, M. A. (2007). P. A. (2009). The impact of global signal regression on resting state
Inhibitory control in high-functioning autism: decreased activation correlations: are anti-correlated networks introduced? NeuroImage,
and underconnectivity in inhibition networks. Biological Psychiatry, 44(3), 893905.
62(3), 198206. Nair, A., Keown, C. L., Datko, M., Shih, P., Keehn, B., & Mller, R. A.
Keehn, B., Brenner, L. A., Ramos, A. I., Lincoln, A. J., Marshall, S. P., & (2014). Impact of methodological variables on functional connec-
Mller, R.-A. (2009). Brief report: eye-movement patterns during an tivity findings in autism spectrum disorders. Human Brain
embedded figures test in children with ASD. Journal of Autism and Mapping, 35(8), 40354048.
Developmental Disorders, 39(2), 383387. Nelson, S. B., & Valakh, V. (2015). Excitatory/inhibitory balance and
Keown, C. L., Shih, P., Nair, A., Peterson, N., & Mller, R.-A. (2013). circuit homeostasis in autism Spectrum disorders. Neuron, 87(4),
Local functional overconnectivity in posterior brain regions is asso- 684698.
ciated with symptom severity in autism spectrum disorders. Cell Nomi, J. S., & Uddin, L. Q. (2015). Developmental changes in large-scale
Reports, 5(3), 567572. network connectivity in autism. NeuroImage Clinical, 7, 732741.
Koldewyn, K., Yendiki, A., Weigelt, S., Gweon, H., Julian, J., Paakki, J. J., Rahko, J., Long, X., Moilanen, I., Tervonen, O., Nikkinen,
Richardson, H., Malloy, C., Saxe, R., Fischl, B., & Kanwisher, N. J., Starck, T., Remes, J., Hurtig, T., Haapsamo, H., Jussila, K.,
(2014). Differences in the right inferior longitudinal fasciculus but Kuusikko-Gauffin, S., Mattila, M. L., Zang, Y., & Kiviniemi, V.
no general disruption of white matter tracts in children with autism (2010). Alterations in regional homogeneity of resting-state brain
spectrum disorder. Proceedings of the National Academy of Sciences activity in autism spectrum disorders. Brain Research, 1321, 169.
of the United States of America, 111(5), 19811986. Palmen, S. J., van Engeland, H., Hof, P. R., & Schmitz, C. (2004).
Leech, R., & Sharp, D. J. (2014). The role of the posterior cingulate cortex Neuropathological findings in autism. Brain, 127(Pt 12), 25722583.
in cognition and disease. Brain, 137(Pt 1), 1232. Patriat, R., Molloy, E. K., Meier, T. B., Kirk, G. R., Nair, V. A.,
Meyerand, M. E., Prabhakaran, V., & Birn, R. M. (2013). The effect
Liu, D., Dong, Z., Zuo, X., Wang, J., & Zang, Y. (2013). Eyes-open/eyes-
of resting condition on resting-state fMRI reliability and consisten-
closed dataset sharing for reproducibility evaluation of resting state
cy: a comparison between resting with eyes open, closed, and fixat-
fMRI data analysis methods. Neuroinformatics, 11(4), 469476.
ed. NeuroImage, 78, 463473.
Lopez-Larson, M. P., Anderson, J. S., Ferguson, M. A., & Yurgelun-
Power, J. D., Mitra, A., Laumann, T. O., Snyder, A. Z., Schlaggar, B. L., &
Todd, D. (2011). Local brain connectivity and associations with
Petersen, S. E. (2014). Methods to detect, characterize, and remove
gender and age. Developmental Cognitive Neuroscience, 1(2),
motion artifact in resting state fMRI. NeuroImage, 84, 320341.
187197.
Rubenstein, J. L., & Merzenich, M. M. (2003). Model of autism: in-
Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism diagnostic inter- creased ratio of excitation/inhibition in key neural systems. Genes,
view-revised: A revised version of a diagnostic interview for care- Brain, and Behavior, 2(5), 255267.
givers of individuals with possible pervasive developmental disor- Rudie, J. D., & Dapretto, M. (2013). Convergent evidence of brain
ders. Journal of Autism and Developmental Disorders, 24, 659685. overconnectivity in children with autism? Cell Reports, 5(3), 565566.
Lord, C., Risi, S., Lambrecht, L., Cook, E. H. Jr., Leventhal, B. L., Sampaio-Baptista, C., Filippini, N., Stagg, C. J., Near, J., Scholz, J., &
Dilavore, P. C., et al. (2000). The autism diagnostic observation Johansen-Berg, H. (2015). Changes in functional connectivity and
schedule-generic: A standard measure of social and communication GABA levels with long-term motor learning. NeuroImage, 106, 1520.
deficits associated with the spectrum of autism. Journal of Autism Samson, F., Mottron, L., Soulieres, I., & Zeffiro, T. A. (2012). Enhanced
and Developmental Disorders, 30, 05223. visual functioning in autism: an ALE meta-analysis. Human Brain
Lu, J. T., Kishida, K. T., De Asis-Cruz, J., Lohrenz, T., Treadwell- Mapping, 33(7), 15531581.
Deering, D., Beauchamp, M., & Montague, P. R. (2015). Single- Satterthwaite, T. D., Elliott, M. A., Gerraty, R. T., Ruparel, K., Loughead,
stimulus functional MRI produces a neural individual difference J., Calkins, M. E., Eickhoff, S. B., Hakonarson, H., Gur, R. C., Gur,
measure for autism spectrum disorder. Clinical Psychological R. E., & Wolf, D. H. (2013). An improved framework for confound
Science, 3(3), 422432. regression and filtering for control of motion artifact in the prepro-
Maximo, J. O., Keown, C. L., Nair, A., & Mller, R.-A. (2013). cessing of resting-state functional connectivity data. Neuroimage,
Approaches to local connectivity in autism using resting state func- 64, 240256.
tional connectivity MRI. Frontiers in Human Neuroscience, 7, 605. Schipul, S. E., Keller, T. A., & Just, M. A. (2011). Inter-regional brain
Maximo, J. O., Cadena, E. J., & Kana, R. K. (2014). The implications of communication and its disturbance in autism. Frontiers in Systems
brain connectivity in the neuropsychology of autism. Neuroscience, 5, 10.
Neuropsychology Review, 24(1), 1631. Schumann, C. M., & Nordahl, C. W. (2011). Bridging the gap between
Minshew, N. J., & Williams, D. L. (2007). The new neurobiology of MRI and postmortem research in autism. Brain Research, 1380,
autism: cortex, connectivity, and neuronal organization. Archives 175186.
of Neurology, 64(7), 945950. Seeley, W. W., Menon, V., Schatzberg, A. F., Keller, J., Glover, G. H.,
Monk, C. S., Peltier, S. J., Wiggins, J. L., Weng, S. J., Carrasco, M., Risi, Kenna, H., Reiss, A. L., & Greicius, M. D. (2007). Dissociable
S., & Lord, C. (2009). Abnormalities of intrinsic functional connec- intrinsic connectivity networks for salience processing and execu-
tivity in autism spectrum disorders. NeuroImage, 47(2), 764772. tive control. The Journal of Neuroscience, 27(9), 23492356.
Mottron, L., Dawson, M., Soulieres, I., Hubert, B., & Burack, J. (2006). Shen, M. D., Shih, P., Ottl, B., Keehn, B., Leyden, K. M., Gaffrey, M. S.,
Enhanced perceptual functioning in autism: an update, and eight & Muller, R. A. (2012). Atypical lexicosemantic function of
Brain Imaging and Behavior
extrastriate cortex in autism spectrum disorder: evidence from func- Uddin, L. Q., Supekar, K., Lynch, C. J., Khouzam, A., Phillips, J.,
tional and effective connectivity. NeuroImage, 62(3), 17801791. Feinstein, C., Ryali, S., & Menon, V. (2013). Salience network-
Shih, P., Keehn, B., Oram, J. K., Leyden, K. M., Keown, C. L., & Mller, based classification and prediction of symptom severity in children
R.-A. (2011). Functional differentiation of posterior superior tempo- with autism. JAMA Psychiatry, 70(8), 111.
ral sulcus in autism: a functional connectivity magnetic resonance Uppal, N., Wicinski, B., Buxbaum, J. D., Heinsen, H., Schmitz, C., &
imaging study. Biological Psychiatry, 70(3), 270277. Hof, P. R. (2014). Neuropathology of the anterior midcingulate cor-
Shukla, D. K., Keehn, B. M., & Mller, R.-A. (2010). Regional homo- tex in young children with autism. Journal of Neuropathology and
geneity of fMRI time series in autism spectrum disorders. Experimental Neurology, 73(9), 891902.
Neuroscience Letters, 476, 4651. Vissers, M. E., Cohen, M. X., & Geurts, H. M. (2012). Brain connectivity
Simmons, D. R., Robertson, A. E., McKay, L. S., Toal, E., McAleer, P., & and high functioning autism: a promising path of research that needs
Pollick, F. E. (2009). Vision in autism spectrum disorders. Vision refined models, methodological convergence, and stronger behavioral
Research, 49(22), 27052739. links. Neuroscience and Biobehavioral Reviews, 36(1), 604625.
Simms, M. L., Kemper, T. L., Timbie, C. M., Bauman, M. L., & Blatt, G. von dem Hagen, E. A., Stoyanova, R. S., Baron-Cohen, S., & Calder, A.
J. (2009). The anterior cingulate cortex in autism: heterogeneity of J. (2013). Reduced functional connectivity within and between 'so-
qualitative and quantitative cytoarchitectonic features suggests pos- cial' resting state networks in autism spectrum conditions. Social
sible subgroups. Acta Neuropathologica, 118(5), 673684. Cognitive and Affective Neuroscience, 8(6), 694701.
Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Wang, X. H., Li, L., Xu, T., & Ding, Z. (2015). Investigating the temporal
Behrens, T. E., Johansen-Berg, H., et al. (2004). Advances in func- patterns within and between intrinsic connectivity networks under
tional and structural MR image analysis and implementation as FSL. eyes-open and eyes-closed resting states: a dynamical functional
NeuroImage, 23(Suppl. 1), S208S219. connectivity study based on phase synchronization. PloS One,
Spoormaker, V. I., Schroter, M. S., Gleiser, P. M., Andrade, K. C., Dresler, 10(10), e0140300.
M., Wehrle, R., Samann, P. G., & Czisch, M. (2010). Development of a
Washington, S. D., Gordon, E. M., Brar, J., Warburton, S., Sawyer, A. T.,
large-scale functional brain network during human non-rapid eye
Wolfe, A., Mease-Ference, E. R., Girton, L., Hailu, A., Mbwana, J.,
movement sleep. The Journal of Neuroscience, 30(34), 1137911387.
Gaillard, W. D., Kalbfleisch, M. L., & Vanmeter, J. W. (2013).
Sridharan, D., Levitin, D. J., & Menon, V. (2008). A critical role for the
Dysmaturation of the default mode network in autism. Human
right fronto-insular cortex in switching between central-executive
Brain Mapping, 35, 1284.
and default-mode networks. Proceedings of the National Academy
Wass, S. (2011). Distortions and disconnections: disrupted brain connec-
of Sciences of the United States of America, 105(34), 1256912574.
tivity in autism. Brain and Cognition, 75(1), 1828.
Stagg, C. J., Bachtiar, V., Amadi, U., Gudberg, C. A., Ilie, A. S., Sampaio-
Baptista, C., O'Shea, J., Woolrich, M., Smith, S. M., Filippini, N., Near, Yang, D. Y., Beam, D., Pelphrey, K. A., Abdullahi, S., & Jou, R. J.
J., & Johansen-Berg, H. (2014). Local GABA concentration is related (2016). Cortical morphological markers in children with autism: a
to network-level resting functional connectivity. eLife, 3, e01465. structural magnetic resonance imaging study of thickness, area, vol-
Supekar, K., Uddin, L. Q., Khouzam, A., Phillips, J., Gaillard, W. D., ume, and gyrification. Molecular autism, 7, 11.
Kenworthy, L. E., Yerys, B. E., Vaidya, C. J., & Menon, V. (2013). Zang, Y., Jiang, T., Lu, Y., He, Y., & Tian, L. (2004). Regional homoge-
Brain hyperconnectivity in children with autism and its links to neity approach to fMRI data analysis. NeuroImage, 22(1), 394400.
social deficits. Cell Reports, 5(3), 738747. Zhang, Y., Brady, M., & Smith, S. (2001). Segmentation of brain MR
Sutton, B. P., Goh, J., Hebrank, A., Welsh, R. C., Chee, M. W. L., & Park, images through a hidden Markov random field model and the
D. C. (2008). Investigation and validation of intersite fMRI studies expectation-maximization algorithm. IEEE Transactions on
using the same imaging hardware. Journal of Magnetic Resonance Medical Imaging, 20, 4557.
Imaging, 28(1), 2128. Zou, Q., Long, X., Zuo, X., Yan, C., Zhu, C., Yang, Y., Liu, D., He, Y., &
Tagliazucchi, E., von Wegner, F., Morzelewski, A., Borisov, S., Jahnke, Zang, Y. (2009). Functional connectivity between the thalamus and
K., & Laufs, H. (2012). Automatic sleep staging using fMRI func- visual cortex under eyes closed and eyes open conditions: a resting-
tional connectivity data. NeuroImage, 63(1), 6372. state fMRI study. Human Brain Mapping, 30(9), 30663078.
Turner, J. A., Damaraju, E., Van Erp, T. G. M., Mathalon, D. H., Ford, J. Zou, Q., Miao, X., Liu, D., Wang, D. J., Zhuo, Y., & Gao, J. H. (2015).
M., Voyvodic, J., et al. (2013). A multi-site resting state fMRI study Reliability comparison of spontaneous brain activities between
on the amplitude of low frequency fluctuations in schizophrenia. BOLD and CBF contrasts in eyes-open and eyes-closed resting
Frontiers in Neuroscience, 7(7 AUG), 137. states. NeuroImage, 121, 91105.