ars.els-cdn.com · web viewthe d-kefs (delis et al., 2001) color-word interference test is based on...
Post on 13-Aug-2020
0 Views
Preview:
TRANSCRIPT
Supplemental Material: Heterogeneity of executive function revealed by a functional random forest approach across ADHD and ASD
1. Methods
1.1 Participant recruitment and diagnostics
ASD participants were recruited via community outreach and mailings, and from OHSU's Child
Development and Rehabilitation Center (CDRC.) ASD participants were phone screened for
basic eligibility criteria prior to their first visit - where both parent and child completed initial
diagnostic testing and questionnaires. After the assessment visit, participants received a
consensus diagnosis from a team of two licensed psychologists and a psychiatrist using research
reliable clinical interviews: Autism Diagnostic Observation Schedule (ADOS-2) (C Lord et al.,
2012) and the Autism Diagnostic Interview (ADI-R) (Catherine Lord et al., 1994), as well as
parent and teacher answered questionnaires including the Social Responsiveness Scale (SRS-2)
(Constantino & Gruber, 2005), Children’s Communication Checklist (CCC-2) (Bishop, 2013),
and a developmental history.
ADHD and typically developing (TD) participants were recruited in a similar fashion through a
parallel study (Karalunas et al., 2018) and were diagnosed with (or without) ADHD and other
comorbidities by a separate clinical team of a similar profile using the Kiddie Schedule of
Affective Disorders and Schizophrenia (K-SADS) (Kaufman et al., 1997) parent interview, as
well as parent and teacher answered questionnaires including the ADHD Rating Scale (ADHD-
RS) (DuPaul et al., 2016), Conners-3 (Conners et al., 2011), Strengths and Difficulties
Questionnaire (SDQ) (Goodman, 1997), Children’s Depression Inventory (CDI) (Kovacs, 1985),
1
123456789
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
and the Multidimensional Anxiety Scale for Children (MASC-2) (March, 2004). All ASD
participants completed the same diagnostic process to assess for ADHD and other comorbidities
and 36 of the ASD participants also met criteria for a comorbid ADHD diagnosis (age mean =
12.3, SD = 2.3). 19 of the ASD participants received a “subthreshold-ADHD” diagnosis -
meaning that they met 5 but not 6 symptoms of inattention or hyperactivity, just shy of the
diagnostic threshold by DSM-IV and DSM-5. Additionally, as a final way to minimize false
positive cases, only participants who met DSM-IV and DSM-5 criteria for ADHD at every
participating study year were included as participants in the ADHD cohort. Similarly, TD
participants were included if they had a stable TD diagnosis at every participating study year.
ASD participants were excluded if they were taking long-acting psychoactive medication, had
genetic abnormalities, neurological impairments, major medical issues or physical disabilities,
significantly impairing Axis I comorbidities (i.e., bipolar disorder, schizophrenia), closed head
injury, seizure disorders, or if they had an estimated IQ <70. ADHD and TD participants
followed the same exclusion criteria with the additional exclusion of an ASD diagnosis, and no-
ADHD for the TD participants. All participants were required to be washed-out of any stimulant
medication for 24-48 hours (depending on the half-life and preparation) for all
neuropsychological and behavioral testing and MRI visits.
1.2 Data Collection Process
After the initial screening visit and diagnostic consensus, ASD participants returned for three
additional visits where the majority of the neuropsychological task data, as described in the
section below, were collected and the participants underwent an MRI. Study data for the single
year of participation were included for ASD participants.
2
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
ADHD and TD participants completed their EF testing and MRI at a companion study of a
longitudinal design (see Karalunas et al). During the first year of the ASD study, companion
study participants were invited to complete a single visit with the ASD study where remaining
data were collected to match the ASD participant data.
1.3 Data matching and further exclusions
ASD participant data were included for a single year of participation. The mean average of time
between the first and last visit was 3 months. For the ADHD and TD participants – their ASD
study visit date was considered the initial time-point and data collected through the companion
study were matched as closely as possible to the ASD data collection dates. As some EF and
ADHD symptoms are known to change or improve over time, any ADHD or TD participant with
ASD study data that exceeded a 9 month cutoff from the companion study were not included in
our participant list.
1.4 Individual task descriptions and review
A total of 43 variables from multiple behavioral tasks and one parent answered questionnaire
were used as inputs in the FRF models. The measures span multiple cognitive domains as
identified in the main manuscript, and are categorized into four broader domains including
cognitive flexibility, response inhibition, working memory, and task control. Tables 1, 2, 3, and 4
summarize the variable inputs, measures they are derived from, and the how the score was
represented.
1.4.1 Verbal Fluency: Cognitive flexibility
The NEPSY verbal fluency task (Korkman et al., 2007)measures a participant’s ability to recall
and list out words in 2 categories (semantic: animals and food) and recall words beginning with a
letter (phonemic: S and F) over a 1 minute span for each item. Including both categorical and
3
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
letter fluency allows for the testing of semantic and phonemic word generation and retrieval. The
verbal fluency task was audio recorded and reviewed by both the administrator and a secondary
tester to verify the accuracy of administration and scoring. Variables included in the model were
the semantic, phonemic, and contrast scores – all of which were scaled.
The majority of prior research indicates that those with ASD tend to perform worse on semantic
word generation than their TD peers (H M Geurts et al., 2004; Spek et al., 2009; Verté et al.,
2006). Still, some researchers have shown equal numbers of total semantic word production for
ASD and TD children and adolescents (ages 6-23) (Begeer et al., 2014). Another group found
delays in performance on letter fluency for ADHD adolescents as compared to TD (Hurks et al.,
2004).
In one study, male ASD participants performed a VF task while undergoing fMRI (Kenworthy et
al., 2013). They found reduced activity in regions associated with executive control compared to
TD, despite no significant differences in word generativity. The ages of the participants in this
study were similar, but had mean of 16(ASD) and 17(TD) years old, putting both groups over the
threshold where one would expect ASD participants to perform more similarly to their TD peers.
1.4.2 Behavior Rating Inventory of Executive Function (BRIEF): Response inhibition,
cognitive flexibility, working memory, task control
The BRIEF (Gioia et al., 2001) is a parent answered questionnaire that covers a broad range of
EF as exhibited by the child in everyday life. For example, an item from the emotional control
module states “Has explosive angry outbursts” and the parent is asked to rate the truthfulness of
the statement, as it pertains to their child, as never (1), sometimes (2), or often (3). The scored
categories include inhibition, shifting, emotional control, working memory, planning and
organizing, organization of materials, monitoring, behavioral regulation, and metacognition.
4
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
Parents answered the questionnaire on a computer or tablet. T-Scores from all domains were
included in the model.
The newness of the BRIEF (2000) may contribute to the limited amount of research. Children
with ADHD have shown impaired ratings on the BRIEF, and the BRIEF itself is highly
correlated with other questionnaire based measures of EF (Mahone et al., 2002). One group
examining ASD and TD participants found that by applying a graph theory metric derived from
task and rest states, they could successfully predict a subject’s BRIEF metacognition index based
on changes within the frontoparietal, salience, and subcortical networks (Lynch et al., 2017). As
previously mentioned in the main manuscript, the BRIEF has also been used to examine the
connection between EF and everyday impairments (Gardiner & Iarocci, 2017).
1.4.3 Color Word Interference: Response inhibition
The D-KEFS (Delis et al., 2001) Color-Word Interference test is based on the Stroop procedure
and involves a participant’s ability to inhibit a learned response. In this case, we examined the
inhibitory condition in which a subject is asked to name the color of the ink that a dissonant color
word name was printed in. For example, the word “blue” might be printed in red ink and
therefore, the correct verbal response is “red.” Variables included in the model were the time to
complete the task, uncorrected errors, self-corrected errors, and total errors – all of which were
scaled.
Inhibiting a response using the stroop task has been found to be largely in-tact for those with
ASD (Adams & Jarrold, 2012). However, as with many other studies, comorbid ADHD is often
unaccounted for when comparing ASD participants to TD. Because some ASD participants may
perform worse than others based on varying levels of ADHD symptoms, we included the stroop
task in the models. Some researchers have found no differences on performance between two
5
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
ADHD “subtypes” (combined and inattentive) and their TD peers, but did find that the ADHD
subtypes were slower than the TD group which may reflect a compensatory mechanism (Nigg et
al., 2002). For this reason, we included both performance and time scores in the models.
1.4.4 Digit Span: Working memory (auditory)
The WISC-IV (Wechsler, 2003)Digit Span task measures a child’s ability to both recall and
manipulate information in short-term storage. A series of numbers are spoken aloud by the tester
and the child is then asked to recall them either in forward or backward order. The amount of
numbers in the sequences increase as the child responds correctly to each administration. For the
model, we included the backward and forward scaled scores, and percentiles for the longest digit
recalled both backward and forward.
Broadly, working memory is known to be impaired in those with ADHD, but there is conflicting
research regarding the role of auditory working memory in ASD. Some researchers found that
ASD subjects show working memory deficits similar to those seen in individuals with EF
impairments (Bennetto et al., 1996). However, another study showed that working memory may
perhaps be in-tact (Ozonoff & Strayer, 2001). It’s possible that differences in performance may
be due to varying numbers of ASD individuals with comorbid ADHD impairments in any given
cohort, which is not accounted for in these studies. Another limitation is the inclusion of only a
single working memory measure, rather than including auditory and spatial working memory in
the same analysis.
1.4.5 Trails: Cognitive flexibility
In the D-KEFS (Delis et al., 2001) trail making task the 4th condition (number-letter switching) is
the primary condition used to assess EF domains of switching, sequencing, and task control. In
condition 4, the participant is asked to trace a line from “a number (1) to a letter (A) to a number
6
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
(2) to a letter (B) and so on, in order, until they reach the end (16).” Variables included in the
model were the total time to completion, sequencing errors, set loss errors, and total errors. All of
the variables were scaled or percentiles.
Two studies showed no significant differences on versions of the trail making task between ASD
and TD (Nakahachi et al., 2006), and high-functioning autism (HFA) and TD groups (Losh et
al., 2009). Another study compared ASD, ADHD, and TD participants and showed that the ASD
group differed from both the ADHD and TD groups (Corbett et al., 2009). However, each of the
aforementioned mentioned studies used either a different version of the trails task (DKEFS or
ATMT) or a different task type (switching vs. sequencing.) These slight differences in effect,
make it harder to compare results across studies as they may be measuring similar, but slightly
different parts of EF. Including the time to completion as well as different types of errors
(sequencing or set-shifting) provides a more comprehensive picture of EF deficits.
1.4.6 Tower Test: Cognitive flexibility, task control
The D-KEFS tower test (Delis et al., 2001) is a table task that requires a participant move disks
of varying sizes across three pegs in order to match their tower to the specified picture using the
fewest possible number of moves. When administered correctly, the complexity of the task
enables the measurement of multiple EF domains including cognitive flexibility and task control.
The tower task was video and audio recorded, and then reviewed by both the initial tester
themselves and a secondary rater to verify the accuracy of both administration and scoring.
Variables included the mean accuracy of the time to first move, overall accuracy, rule violations
per item administered, time per move, total achievement score, and total rule violations. All of
the included variables were either scaled scores or percentiles.
7
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
Researchers using a similarly constructed task showed no significant differences between ASD
and TD individuals on a MANCOVA for Planning (minimum moves, initial thinking, and
subsequent thinking) (Corbett et al., 2009). HFA participants have also shown associations
between planning deficits on the tower task and reduced efficiency in visuospatial short-term
memory (Zinke et al., 2010). Another group showed that ASD youth had the lowest global
performance compared to ADHD and TD groups, but that those with comorbid ASD and ADHD
showed greater improvement trajectories than those that were ASD alone (Unterrainer et al.,
2015). Inclusion of all of the tower variables might enable for a better understanding of how
these impairments relate to other cognitive domains in individuals with ASD and/or ADHD.
1.4.7 Stop Task: Response inhibition, task control
The go/no-go Stop Task (Logan, 1994; Nigg, 1999) measures a participant’s ability to both react
quickly to a stimuli and to inhibit a response. In brief, participants fixate on a white screen and
are presented with a rainbow “X” or “O.” They are asked to either make a key press
corresponding to the X or O, or to inhibit their response at the presentation of an auditory tone.
The variables used in the model included an accuracy measure of X/O key press on go-trials, a
probability measure of inhibition on stop-trials, the stop signal reaction time, mean reaction time
on go-trials, and the standard deviation of reaction times on go-trials. While accuracy of X/O key
press on go-trials serves as a control for other task variables, it may also provide information
about letter anticipation which may be task-control related and thus was included in the models.
All justifications for including data in the FRF should be carefully considered, but another
benefit of the FRF is that it successfully ignores features that show no valuable contribution.
Because the stop task is not scaled, potential age confounds were also examined in the
supplementary analyses.
8
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
It has been widely demonstrated that participants with ADHD perform worse than their TD peers
on the stop task (Senderecka et al., 2012). Yet for one study that covaried for ADHD symptoms
in ASD and ADHD groups, these differences all but disappeared – with the exception of the
ASD group showing increased premature responses (Carter Leno et al., 2017). Meta-analyses on
response inhibition in ASD, including the stop-task, further emphasized that heterogeneity of
ASD may have an effect on the inconsistent results (Hilde M. Geurts et al., 2014). As evidenced
by the inconsistencies across the literature. Similar to other researchers, consideration of
comorbidities such as ADHD is recommended, as well as looking further into ASD subtypes.
1.4.8 Spatial Span: Working memory
The spatial span (Robbins et al., 1994)is a computerized task that measures spatial working
memory. The task presents 10 white boxes in random locations on the screen – a subset of which
change color, one at a time, in a fixed order. In the Forward task, upon completion of the color
change sequence, subjects hear a tone and are asked to click on the boxes in the order in which
they changed color on the screen. The number of squares that change color range from 3 to 9,
with two trials for each span length, and the task discontinues when a child fails both trials of the
same span. The Backward task is presented in the same way, but instead, subjects are asked to
click on the boxes in the opposite order in which they appeared. The forward and backward
spans were counterbalanced and subjects had the opportunity to practice prior to administration
of the task. Because the spatial span task is not scaled, potential age confounds were further
analyzed.
Several recent studies have suggested both impaired (Chen et al., 2016) and non-impaired
(Macizo et al., 2016) spatial working memory among those with ASD. Stronger evidence exists
to support such an impairment among those with ADHD, as they have been shown to exhibit
9
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
deficits in visio-spatial working memory as compared to TD peers across multiple studies
(Kasper et al., 2012). A study comparing participants with ADHD, ASD+ADHD, ASD, and TD
on a spatial working memory task showed that ADHD participants performed worse than both
the TD and ASD+ADHD groups and that both ASD+ADHD and ADHD groups needed longer
to perform the task than TD and ASD (Sinzig et al., 2008).
Response Inhibition Variable importance per model
Measure Variable Type Hyperactive InattentiveBRIEF Emotional Control T-score 0.1928 0.04579
BRIEF Inhibit T-score 0.855 -0.0039
BRIEF Monitor T-score 0.0802 0.0524
Color Word Total errors Scaled -0.0926 0.038
Color Word Total time Scaled -0.001 -0.0656
Color Word Total self-corrected errors Scaled -0.0184 0.0226
Color Word Total Uncorrected errors Scaled -0.0395 -0.0294
Stop Task Mean reaction time Raw 0.0942 -0.0418
Stop Task Standard Deviation on Go trials Raw 0.0959 0.0368
Stop Task Stop stimulus reaction time average Raw 0.053 0.0693
Stop Task Probability of stopping Raw 0.0094 -0.0188
Table 1: Eleven EF variables representing response inhibition included in the FRF models. On the right, the variable importance is shown per model.
Working Memory Variable importance per model
Measure Variable Type Hyperactive InattentiveBRIEF Working Memory T-score 0.0297 0.5102
Digit Span Longest digit recalled backward Percentile 0.0153 -0.0261
Digit Span Longest digit recalled forward Scaled 0.0101 0.0064
Digit Span Total backward score Scaled 0.0291 -0.0416
Digit Span Total forward score Scaled -0.0379 -0.0427
Digit Span Total, forward and backward Scaled 0.0083 -0.0238
Spatial Span Backward, mean reaction time Raw 0.2225 0.0888
Spatial Span Backward, Items correct (total accuracy) Raw 0.1634 0.1141
Spatial Span Backward, items correct/items attempted Raw 0.1359 0.0785
Spatial Span Forward, mean reaction time Raw -0.0124 0.0617
Spatial Span Forward, Items correct (total accuracy) Raw -0.0323 -0.0197
Spatial Span Forward, items correct/items attempted Raw 0.0387 -0.0161
Table 2: Twelve EF variables representing working memory included in the FRF models, with variable importance per model on the right.
10
210
211
212
213
214
215216
217
218219
220
Table 3: Ten EF variables representing cognitive flexibility included in the FRF models, with variable importance per model on the right.
Task Control Variable importance per model
Measure Variable Type Hyperactive InattentiveBRIEF Initiate T-score 0.112 0.3668
BRIEF Org. of materials T-score 0.0143 0.1556
BRIEF Plan and organize T-score 0.1253 0.4186
Tower Accuracy Scaled 0.0259 0.0204
Tower Rule violations per item Scaled -0.0022 -0.0106
Tower Total achievement Scaled 0.0236 0.0231
Tower Total rule violations Percentile -0.0136 -0.0099
Trails Sequencing errors Percentile 0.0765 -0.0485
Trails Total errors Scaled -0.0081 -0.0099
Stop Task Accuracy on Go trials Raw 0.1799 0.1292
Table 4: Ten EF variables representing task control included in the FRF models, with variable importance per model on the right.
1.5 fMRI Data
1.5.1 fMRI data acquisition
Participants were scanned at OHSU’s Advanced Imaging Research Center (AIRC) on a 3.0 T
Siemens Tim Trio Magnetom scanner using a 12 channel head coil. One T1 weighted structural
image (TR = 2300 ms, TE = 3.58 ms, orientation = sagittal, FOV = 256 × 256 matrix, voxel
resolution = 1mmx1mmx1.1 mm slice thickness) was acquired for each participant. fMRI data
(TR=2500ms, TE=30ms, flip angle=90°, voxel resolution = 3.75mm x 3.75mm x 3.80mm) were
collected as three, 5-minute BOLD resting state scans during which the participants were asked
to lie still and look at a white fixation cross on a black display.
11
Cognitive Flexibility Variable importance per model
Measure Variable Type Hyperactive InattentiveBRIEF Behavior regulation index T-score 0.3786 0.0622
BRIEF Metacognition T-score 0.147 0.4029
BRIEF Shift T-score -0.0697 0.1285
Tower Mean 1st move time Scaled 0.109 -0.0355
Tower Time per move Scaled 0.0968 -0.018
Trails Set loss errors Percentile 0.0482 -0.045
Trails Time to completion Scaled 0.004 0.0315
Verbal Fluency Contrast score Scaled -0.0409 -0.0368
Verbal Fluency WGI (Phonemic) Scaled 0.0257 -0.0537
Verbal Fluency WGS (Semantic) Scaled 0.0057 0.013
221222223224225226
227
228229
230
231
232
233
234
235
236
237
238
1.5.2 fMRI data processing
The HCP pipeline consists of five stages: 1) PreFreeSurfer, which corrects for MR gradient and
bias field distortions, aligns T1 images to a template, and registers the structural data to MNI
space; 2) FreeSurfer, which performs segmentation of brain volumes into predetermined regions,
reconstructs the surfaces, and aligns the images to a standardized template; 3) PostFreeSurfer,
which converts the data into standard NIH formats (Li et al., 2016), and applies the surface
registration to a template, and creates a final “brain mask;” 4) fMRIVolume, which removes
spatial distortions, performs motion correction, aligns the fMRI data to the participant’s
anatomical data, normalizes data to a global mean, and masks the data with the final brain mask;
and 5) fMRISurface, which maps the time series to a standard CIFTI format – which represents
the brain in grayordinate space (Glasser et al., 2013). Additional functional connectivity
processing was performed in alignment with current neuroimaging best practices (Mills et al.,
2018; Power et al., 2013). The additional steps included removing spikes in the MR signal, slice
time corrections, accounting and correcting for head motion during and between runs,
normalizing intensity to a whole brain value of 1000, temporal band-pass filtering, and
regressing out nuisance variables of motion estimates, tissue-based signals, and global signal
regression.
1.5.3 Quality control
Prior to performing manual quality control (QC) on data used in analyses, raters completed an
hour long training session followed by scoring a 10 participant reliability set that required an
inter-rater (anchor vs. rater) agreement of .8 to pass reliability. To score the agreement, half-
credit was given for disagreements between a high (1) and medium (2) quality scan, and no
credit was given for a misidentified low quality scan (3). Reliability training and manual QC of
12
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
data for the analyses worked in three steps. In the first step, T1 registration to the brain atlas were
visually assessed and determined to be of high (1), medium (2), or low (3) quality, with a high
quality image showing good registration and little to no warping. The second step involved a
visual inspection of the structural T1 images. A high quality T1 image shows well delineated
gray and white matter whereas a low quality image might show excessive blurring, ringing, or
sawtooth jaggedness in the delineation. Images deemed to be of high or medium quality were
then assessed for good functional registration using the same three tiered metric. High quality
functional registration shows that the data are well mapped to the T1 image across all
dimensions. A poor quality image might show things such as movement artifacts, field of view
cutoffs, distortions, or non-homologous signal or signal dropout. For the analyses, 1’s were
approved, 3’s were rejected, and scans that were a 2 in the functional and/or anatomical stage
were then assessed by a second rater to increase reliability and decrease bias before being
considered for inclusion. We further accounted for the effect of head movement on the MR
signal using motion-targeted “scrubbing”; censoring volumes with framewise displacement
greater than .2mm and eliminating any runs with less than 60 volumes total and/or less than 5
minutes of good resting state data. Only scans that passed all of the above criteria were
considered suitable for analysis.
2. Supplemental Analyses: Methods
13
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
2.1 The effect of age
2.1.1 Examining age differences between ASD and ADHD groups and the effect on the
models
In the main manuscript, we describe our cohort consisting of both ASD and ADHD participants.
Although we were not interested in comparing the ASD and ADHD groups to one another, the
possibility of a batch effect required that we investigate potential group differences. For the first
comparison between the ASD and ADHD participants, we identified a significant age difference
(t(128) = 3.67, p <.001.) We conducted several additional analysis to further examine the effects
of age on the models.
If age were a driving factor, we would expect to see a statistically significant increase in the
accuracy of the models when participant age is included as a 44th variable. Therefore, we ran
additional FRF models including age as an added variable for both the hyperactive and
inattentive models. We then compared the accuracy of the old models to the new models with t-
tests.
2.1.2 Age and unscaled EF variables
Nearly all of the variables were scaled, percentile, or t-scores. However, a handful of variables
were represented as raw scores (see Tables 1, 2, 3, 4.) Because age may be associated with task
performance, we ran a simple correlation between all raw EF measures and age to identify
potential relationships. We then examined variable importance per model to see if any of the
variables identified as highly correlated with age were also frequently used in the models, as any
variables that met both criteria could be considered potential confounds.
2.2 Differences between those with and without usable scan data
Out of the 130 participants, 67 were found to have useable scan data after quality control. We
14
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
wanted to determine if the 67 with useable scan data showed similar demographic profile
compositions to those without imaging data (n = 64). To do this, we ran significance tests
comparing demographic profiles for those without and with usable scan data. This included
diagnosis of ASD vs. ADHD (Fishers exact), gender (Fishers exact), age (t-test), and estimated
IQ (t-test.)
As in the main manuscript, we also wanted to ensure that those with good imaging data in the
identified subgroups (HSG-Mild = 38, HSG-Severe = 29; ISG-Mild = 20, ISG-Severe = 47) did
not represent significant demographic differences between subgroups per model. To do this, we
ran the same significance tests comparing demographic profiles for only those with good
imaging data in identified subgroups for both of the models.
We were also interested to see if those with good imaging data in the identified subgroups
showed similar neurocognitive profiles to the larger subgroups. Therefore, we ran additional t-
tests on the EF variables between subgroup participants with good imaging data and compared
them to the neurocognitive profiles in the main manuscript.
3. Results
3.1 The effects of age
3.1.1 Model accuracy and age as an additional variable
For both the hyperactive and inattentive models, t-tests on accuracy comparing the models with
and without age included as a variable showed no statistically significant difference in the model
performances (Hyperactive: t(58) = .501, p = .61, Inattentive: t(58) = -.14, p = .89.)
3.1.2 Age and EF scores
15
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
Significant results for the simple correlation between age and raw EF scores, corrected for
multiple comparisons, were observed on some of the measures (Table 5). When we examined
variable importance (Tables 1, 2, 3, 4), we found that none of the variables that were correlated
with age were important for the models.
Age correlated to raw EF tasks
Measure Variable Correlation
Stop Task Standard Deviation on Go trials r(127) -.340, p < .001
Stop Task Stop stimulus reaction time average r(121) -.340, p < .001
Spatial Span Backward, items correct (total accuracy) r(125) -.426, p < .001
Spatial Span Backward, mean reaction time r(125) .383, p < .001
Spatial Span Backward, items correct/items attempted r(125) .388, p < .001
Spatial Span Forward, items correct (total accuracy) r(124) -.355, p < .001
Spatial Span Forward, mean reaction time r(124) .333, p < .001Table 5: Significant correlations between age and raw EF tasks are shown in the table with the measure on the left, specific variable in the center column, and the correlation statistic on the right. Degrees of freedom are in the parentheses.
3.2 Demographic comparisons
3.2.1 Demographic profiles for those with and without useable scan data
Significance tests on demographic information (age, gender, IQ, and diagnostic composition)
comparing those with and without useable scan data can be seen in Table 6. Both primary
diagnostic assignment (ASD or ADHD) and age were significantly different between those with
and without useable scan data.
Unusable/no scan data (n = 64) Useable scan data (n = 67) p-valuea
ASD, 38 (59.4%) 26 (38.8%) p = .022
Female gender, n = 16 (25.4%) 15 (22.4) p = .84
Age (7-16y), m = 10.8 (1.8) 12.2 (2.3) t(124.4) = -3.78, p <.001
Est. IQ (Block design) m = 11.1 (3.25) 10.8 (3.2) t(65) = .615, p = .54
Table 6: Demographic comparisons for those with/without usable scan data
3.2.2 Demographic comparisons between identified subgroups for those with good scan
data
16
330
331
332
333
334335336
337
338
339
340
341
342
343
344
345
346
The hyperactive subgroups with good scan data did not show significant differences from each
other on age, gender, IQ, or diagnostic composition (Table 7).
Similarly, the inattentive subgroups with good scan data did not show significant differences
from each other on age, gender, IQ, or diagnostic composition (Table 8).
Useable scan data (n = 67) HSG-Mild (n = 38) HSG-Severe (n = 29) Significance testa
ASD, 26 (39%) 16 (42.1) 10 (34.5) p = .616
Female gender, 15 (22.4%) 7 (18.4) 8 (27.6) p = .393
Age in years (7-16y) m = 12.2 (2.3) 12.4 (2) 11.9 (2.6) t(50.75) = .79, p = .431
Est. IQ (Block design) m = 10.8 (3.2) 10.6 (3.4) 11.1 (2.8) t(65) = -.624, p = .524
Table 7: Demographic comparisons for those with imaging data in the Hyperactive subgroups.
Useable scan data (n = 67) ISG-Mild (n = 20) ISG-Severe (n = 47) Significance testa
ASD, 26 (39%) 9 (45) 17 (36) p = .587
Female gender, 15 (22.4%) 5 (25) 10 (21.3) p = .756
Age in years (7-16y) m = 12.2 (2.3) 12 (2.5) 12.2 (2.2) t(65) = -.301, p = .764
Est. IQ (Block design) m = 10.8 (3.2) 10.6 (2.9) 10.9 (3.3) t(65) = -.4, p = .688
Table 8: Demographic comparisons for those with imaging data in the Inattentive subgroups.
3.3 Neurocognitive profiles for those with useable scan data
The HSG-Mild and HSG-Severe groups mirrored the main manuscript in that the HSG-Mild
subgroup out-performed or had better ratings than the HSG-Severe group across several EF
measures (Figure 1a and Table 9). Four of the five measures were also found to be significantly
different between the HSGs in the main manuscript analysis (BRIEF: Behavioral regulation,
emotional control, inhibit, and Stop Task: go-trial accuracy.) An additional variable, Trails: Total
errors, was also found to be significantly different between the groups.
The ISG-Mild and ISG-Severe groups also showed similarities to the main manuscript analysis
in that the Mild subgroup out-performed or had better ratings than the Severe group across
several EF measures (Figure 1b and Table 10). Six of the eight measures were also observed in
the main manuscript analysis (BRIEF: Initiate, metacognition, monitor, organization of
17
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
materials, plan and organize, working memory.) The ISG-Mild group committed more rule
violations on the Tower of Hanoi, both per item and overall, than the ISG-Severe group.
Figure 1a: Behavioral measures used in the FRF Hyperactive model that significantly differed between subjects with imaging data in the identified subgroups. Normed means (y axis) from Table 9 are plotted per measure on the x axis in a line plot. To better represent the true differences when compared to a normative sample, all measures have been normed to the TD group.1b: Behavioral measures used in the FRF Inattentive model that significantly differed between subjects with imaging data in the identified subgroups. Normed means (y axis) from Table 10 are plotted per measure on the x axis in a line plot. Again, all measures have been normed to the TD group.
Table 9: Normed means compared between HSG's with good imaging data, and plotted in Figure 1a.
Response Inhibition t-test ISG-Severe ISG-Mild
BRIEF: Monitor t(24) = -3.67, p<.001 -2.56 -1.09
Cognitive Flexibility
BRIEF: Metacognition t(65) = -10.25, p<.001 -2.95 -0.90
Working Memory
18
Response Inhibition t-test HSG-Severe HSG-Mild
BRIEF: Inhibit t(59.9) = 11.43, p<.001 -4.55 -1.4
BRIEF: Emotional control t(65) = 3.46, p <.001 -2.85 -1.49
Stop task: Accuracy on Go-trials t(45.1) = 2.15, p = .037 -1.26 -0.33
Cognitive Flexibility
BRIEF: Behavior regulation index t(65) = 6.31, p<.001 -3.95 -1.82
Task Control
Trails: Total errors t(36.5) = 2.16, p = .037 -1.37 0.10
365
366
367
368369370371372373374
375376377378379
380
382383384385386387
388389390
BRIEF: Working memory t(65) = -7.12, p<.001 -3.25 -1.23
Task Control
BRIEF: Initiate t(65) = -7.88, p<.001 -1.67 -0.33
BRIEF: Plan and organize t(31.1) = -8.83, p<.001 -2.72 -0.78
Tower: Rule viol. Percentile t(25.2) = -2.09, p<.001 -1.04 -3.22
Tower: Rule viol. p/item t(128) = -9.70, p<.001 -0.76 -2.05Table 10: Normed means compared between ISG's with good imaging data, and plotted in Figure 1b.
4. Discussion
While it cannot be entirely ruled out, our supplemental analyses suggest that age may not have
impacted model performance. Future studies should carefully consider group matching or other
methods to account for any age discrepancies.
In-scanner motion is a considerable issue among adolescent neuroimaging (Satterthwaite et al.,
2012). It remains a particular problem among pediatric and neurodivergent populations, wherein
“holding still” for prolonged periods of time may be both a function of age as well as a primary
deficit of the disorder/s. Researchers often observer that older participants tend to have better
methods for withstanding lengthy neuroimaging sessions. Therefore, in regards to our study, it is
somewhat expected that those with useable scan data are significantly older than those without.
While we believe we have achieved noteable success in managing participant motion through
mock –MRI sessions and appropriate behavioral interventions, additional techniques to control
for motion should be considered in future studies.
Along with comorbid ADHD features, those with ASD bear the additional burden of sensory
sensitivities, making achieving quality scan data a challenge. We observed a significant
difference in the number of participants with useable scan data when comparing ASD and
ADHD groups, with the number of ASD participants with useable scan data (26/64, 40%)
significantly less than expected as compared to those with ADHD and useable scan data (41/66,
19
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
62%.) Again, we were not comparing groups and used a transdiagnostic model to identify
subgroups. Although we “lost” more ASD participants in the imaging component, we did not
observe any significant demographic differences between both the identified subgroups for the
overall models (see: main manuscript) as well as when comparing subgroups for only those with
good imaging data (Tables 9 and 10).
As previously mentioned, those with greater ADHD and ASD symptoms may subsequently
struggle with things like holding still and withstanding scan duration. Considering only those
with useable neuroimaging data in the overall behavioral models would create a complication by
potentially eliminating those that may show the greatest impairments in EF and ADHD
symptoms. Therefore, we felt it important to consider as many subjects as possible in the main
models to best capture impairments and the relationships between EF and ADHD symptoms.
5. References
Adams, N. C., & Jarrold, C. (2012). Inhibition in autism: Children with autism have difficulty inhibiting irrelevant distractors but not prepotent responses. Journal of Autism and Developmental Disorders, 42(6), 1052–1063. https://doi.org/10.1007/s10803-011-1345-3
Begeer, S., Wierda, M., Scheeren, A. M., Teunisse, J. P., Koot, H. M., & Geurts, H. M. (2014). Verbal fluency in children with autism spectrum disorders: Clustering and switching strategies. Autism, 18(8), 1014–1018. https://doi.org/10.1177/1362361313500381
Bennetto, L., Pennington, B. F., & Rogers, S. J. (1996). Intact and impaired memory functions in autism. Child Development, 67(4), 1816–1835. https://doi.org/10.1111/j.1467-8624.1996.tb01830.x
Bishop, D. (2013). Children’s Communication Checklist (CCC-2). In Encyclopedia of Autism Spectrum Disorders (pp. 614–618). https://doi.org/10.1007/978-1-4419-1698-3_1929
Carter Leno, V., Chandler, S., White, P., Pickles, A., Baird, G., Hobson, C., Smith, A. B., Charman, T., Rubia, K., & Simonoff, E. (2017). Testing the specificity of executive functioning impairments in adolescents with ADHD, ODD/CD and ASD. European Child and Adolescent Psychiatry, 1–10. https://doi.org/10.1007/s00787-017-1089-5
Chen, S.-F. F., Chien, Y.-L. L., Wu, C.-T. T., Shang, C.-Y. Y., Wu, Y.-Y. Y., & Gau, S. S. (2016). Deficits in executive functions among youths with autism spectrum disorders: an age-stratified analysis. Psychological Medicine, 46(8), 1625–1638. https://doi.org/10.1017/S0033291715002238
20
411
412
413
414
415
416
417
418
419
420
421
422
423
424425426427428429430431432433434435436437438439440441442443
Conners, C. K., Pitkanen, J., & Rzepa, S. R. (2011). Conners 3rd Edition (Conners 3; Conners 2008). In Encyclopedia of Clinical Neuropsychology (pp. 675–678). https://doi.org/10.1007/978-0-387-79948-3_1534
Constantino, J., & Gruber, C. (2005). Social responsiveness scale. Western Psychological Services.
Corbett, B. A., Constantine, L. J., Hendren, R., Rocke, D., & Ozonoff, S. (2009). Examining executive functioning in children with autism spectrum disorder, attention deficit hyperactivity disorder and typical development. Psychiatry Research, 166(2–3), 210–222. https://doi.org/S0165-1781(08)00052-8 [pii] 10.1016/j.psychres.2008.02.005
Delis, D. C., Kaplan, E., & Kramer, J. (2001). Delis Kaplan Executive Function System. The Psychological Corporation.
DuPaul, G. J., Reid, R., Anastopoulos, A. D., Lambert, M. C., Watkins, M. W., & Power, T. J. (2016). Parent and teacher ratings of attention-deficit/hyperactivity disorder symptoms: Factor structure and normative data. Psychological Assessment. https://doi.org/10.1037/pas0000166
Gardiner, E., & Iarocci, G. (2017). Everyday executive function predicts adaptive and internalizing behavior among children with and without autism spectrum disorder. Autism Research. https://doi.org/10.1002/aur.1877
Geurts, H M, Verte, S., Oosterlaan, J., Roeyers, H., & Sergeant, J. A. (2004). How specific are executive functioning deficits in attention deficit hyperactivity disorder and autism? Journal of Child Psychology and Psychiatry, 45(4), 836–854. https://doi.org/10.1111/j.1469-7610.2004.00276.xJCPP276 [pii]
Geurts, Hilde M., van den Bergh, S. F. W. M. W. M., & Ruzzano, L. (2014). Prepotent response inhibition and interference control in autism spectrum disorders: Two Meta-Analyses. Autism Research, 7(4), 407–420. https://doi.org/10.1002/aur.1369
Gioia, G. A., Isquith, P. K., Guy, S. C., Kenworthy, L., & Baron, I. S. (2001). Behavior rating inventory of executive function. In Child Neuropsychology. https://doi.org/10.1076/chin.6.3.235.3152
Glasser, M. F., Sotiropoulos, S. N., Wilson, J. A., Coalson, T. S., Fischl, B., Andersson, J. L., Xu, J., Jbabdi, S., Webster, M., Polimeni, J. R., Van Essen, D. C., & Jenkinson, M. (2013). The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage, 80, 105–124. https://doi.org/10.1016/j.neuroimage.2013.04.127
Goodman, R. (1997). The Strengths and Difficulties Questionnaire: a research note. J Child Psychol Psychiatry, 38(5), 581–586. https://doi.org/10.1111/j.1469-7610.1997.tb01545.x
Hurks, P. P. M., Hendriksen, J. G. M., Vles, J. S. H., Kalff, A. C., Feron, F. J. M., Kroes, M., Van Zeben, T. M. C. B., Steyaert, J., & Jolles, J. (2004). Verbal fluency over time as a measure of automatic and controlled processing in children with ADHD. Brain and Cognition. https://doi.org/10.1016/j.bandc.2004.03.003
Karalunas, S. L., Hawkey, E., Gustafsson, H., Miller, M., Langhorst, M., Cordova, M., Fair, D., & Nigg, J. T. (2018). Overlapping and Distinct Cognitive Impairments in Attention-Deficit/Hyperactivity and Autism Spectrum Disorder without Intellectual Disability. Journal of Abnormal Child Psychology. https://doi.org/10.1007/s10802-017-0394-2
Kasper, L. J., Alderson, R. M., & Hudec, K. L. (2012). Moderators of working memory deficits in children with attention-deficit/hyperactivity disorder (ADHD): A meta-analytic review. In Clinical Psychology Review. https://doi.org/10.1016/j.cpr.2012.07.001
Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., Williamson, D., & Ryan,
21
444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489
N. (1997). Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. J Am Acad Child Adolesc Psychiatry, 36(7), 980–988. https://doi.org/10.1097/00004583-199707000-00021
Kenworthy, L., Wallace, G. L., Birn, R., Milleville, S. C., Case, L. K., Bandettini, P. A., & Martin, A. (2013). Aberrant neural mediation of verbal fluency in autism spectrum disorders. Brain and Cognition, 83(2), 218–226. https://doi.org/10.1016/j.bandc.2013.08.003
Korkman, M., Kirk, U., & Kemp, S. (2007). NEPSY-Second Edition (NEPSY-II). Journal of Psychoeducational Assessment. https://doi.org/10.1177/0734282909346716
Kovacs, M. (1985). The Children’s Depression, Inventory (CDI). Psychopharmacology Bulletin, 21(4), 995–998. https://doi.org/10.3724/SP.J.1041.2015.01004
Li, X., Morgan, P. S., Ashburner, J., Smith, J., & Rorden, C. (2016). The first step for neuroimaging data analysis: DICOM to NIfTI conversion. Journal of Neuroscience Methods, 264, 47–56. https://doi.org/10.1016/j.jneumeth.2016.03.001
Logan, G. D. (1994). On the ability to inhibit thought and action: A users guide to the stop-signal paradigm. In Inhibitory processes in attention, memory, and language (pp. 189–239).
Lord, C, Rutter, M., DiLavore, P., Risi, S., & Gotham, K. (2012). Autism diagnostic observation schedule, (ADOS-2). Los Angeles, California.
Lord, Catherine, Rutter, M., & Le Couteur, A. (1994). Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J Autism Dev Disord, 24(5), 659–685. https://doi.org/10.1007/BF02172145
Losh, M., Adolphs, R., Poe, M. D., Couture, S., Penn, D., Baranek, G. T., & Piven, J. (2009). Neuropsychological profile of autism and the broad autism phenotype. Archives of General Psychiatry, 66(5), 518–526. https://doi.org/10.1001/archgenpsychiatry.2009.34
Lynch, C. J., Breeden, A. L., You, X., Ludlum, R., Gaillard, W. D., Kenworthy, L., & Vaidya, C. J. (2017). Executive Dysfunction in Autism Spectrum Disorder Is Associated With a Failure to Modulate Frontoparietal-insular Hub Architecture. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 2(6), 537–545. https://doi.org/10.1016/j.bpsc.2017.03.008
Macizo, P., Soriano, M. F., & Paredes, N. (2016). Phonological and Visuospatial Working Memory in Autism Spectrum Disorders. Journal of Autism and Developmental Disorders, 46(9), 2956–2967. https://doi.org/10.1007/s10803-016-2835-0
Mahone, E. M., Cirino, P. T., Cutting, L. E., Cerrone, P. M., Hagelthorn, K. M., Hiemenz, J. R., Singer, H. S., & Denckla, M. B. (2002). Validity of the behavior rating inventory of executive function in children with ADHD and/or Tourette syndrome. Arch Clin Neuropsychol, 17(7), 643–662. https://doi.org/S0887617701001688 [pii]
March, J. S. (2004). Multidimensional Anxiety Scale for Children (MASCTM). The Use of Psychological Testing for Treatment Planning and Outcomes Assessment: Instruments for Children and Adolescents, 39. https://doi.org/0805843302
Mills, B. D., Miranda-Dominguez, O., Mills, K. L., Earl, E., Cordova, M., Painter, J., Karalunas, S. L., Nigg, J. T., & Fair, D. A. (2018). ADHD and attentional control: Impaired segregation of task positive and task negative brain networks. Network Neuroscience (Cambridge, Mass.), 2(2), 200–217. https://doi.org/10.1162/netn_a_00034
Nakahachi, T., Iwase, M., Takahashi, H., Honaga, E., Sekiyama, R., Ukai, S., Ishii, R., Ishigami, W., Kajimoto, O., Yamashita, K., Hashimoto, R., Tanii, H., Shimizu, A., & Takeda, M.
22
490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535
(2006). Discrepancy of performance among working memory-related tasks in autism spectrum disorders was caused by task characteristics, apart from working memory, which could interfere with task execution. Psychiatry and Clinical Neurosciences, 60(3), 312–318. https://doi.org/10.1111/j.1440-1819.2006.01507.x
Nigg, J. T. (1999). The ADHD response-inhibition deficit as measured by the Stop Task: replication with DSM-IV combined type, extension, and qualification. J Abnorm Child Psychol, 27(5), 393–402. https://doi.org/10.1023/A:1021980002473
Nigg, J. T., Blaskey, L. G., Huang-Pollock, C. L., & Rappley, M. D. (2002). Neuropsychological executive functions and DSM-IV ADHD subtypes. J Am Acad Child Adolesc Psychiatry, 41(1), 59–66. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=11800208
Ozonoff, S., & Strayer, D. L. (2001). Further Evidence of Intact Working Memory in Autism. Journal of Autism and Developmental Disorders, 31(3), 257–263. https://doi.org/10.1023/A:1010794902139
Power, J., Mitra, A., Laumann, T., Snyder, A., Schlaggar, B., & Petersen, S. (2013). Methods to detect, characterize, and remove motion artifact in resting state fMRI. Neuroimage. https://doi.org/10.1016/j.neuroimage.2013.08.048
Robbins, T. W., James, M., Owen, A. M., Sahakian, B. J., McInnes, L., & Rabbitt, P. M. (1994). Cambridge Neuropsychological Test Automated Battery (CANTAB): a factor analytic study of a large sample of normal elderly volunteers. Dementia, 5(5), 266–281.
Satterthwaite, T., Wolf, D., Loughead, J., Ruparel, K., Elliott, M., Hakonarson, H., Gur, R., & Gur, R. (2012). Impact of in-scanner head motion on multiple measures of functional connectivity: relevance for studies of neurodevelopment in youth. Neuroimage, 60(1), 623–632. https://doi.org/10.1016/j.neuroimage.2011.12.063
Senderecka, M., Grabowska, A., Szewczyk, J., Gerc, K., & Chmylak, R. (2012). Response inhibition of children with ADHD in the stop-signal task: An event-related potential study. International Journal of Psychophysiology, 85(1), 93–105. https://doi.org/10.1016/j.ijpsycho.2011.05.007
Sinzig, J., Morsch, D., Bruning, N., Schmidt, M. H., & Lehmkuhl, G. (2008). Inhibition, flexibility, working memory and planning in autism spectrum disorders with and without comorbid ADHD-symptoms. Child and Adolescent Psychiatry and Mental Health, 2(1), 4. https://doi.org/10.1186/1753-2000-2-4
Spek, A., Schatorje, T., & Scholte, E. (2009). Verbal fluency in adults with high functioning autism or Asperger syndrome. Neuropsychologia, 47(3), 652–656. http://ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=emed9&NEWS=N&AN=19084028
Unterrainer, J. M., Rauh, R., Rahm, B., Hardt, J., Kaller, C. P., Klein, C., Paschke-Muller, M., Biscaldi, M., Paschke-M??ller, M., & Biscaldi, M. (2015). Development of Planning in Children with High-Functioning Autism Spectrum Disorders and/or Attention Deficit/Hyperactivity Disorder. Autism Res, 9(7), 739–751. https://doi.org/10.1002/aur.1574
Verté, S., Geurts, H. M., Roeyers, H., Oosterlaan, J., & Sergeant, J. A. (2006). Executive functioning in children with an autism spectrum disorder: Can we differentiate within the spectrum? Journal of Autism and Developmental Disorders, 36(3), 351–372. https://doi.org/10.1007/s10803-006-0074-5
Wechsler, D. (2003). Wechsler Intelligence Scale for Children—4th Edition (WISC-IV®). Harcourt Assessment.
23
536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581
Zinke, K., Fries, E., Altgassen, M., Kirschbaum, C., Dettenborn, L., & Kliegel, M. (2010). Visuospatial short-term memory explains deficits in tower task planning in high-functioning children with autism spectrum disorder. Child Neuropsychology, 16(3), 229–241. https://doi.org/10.1080/09297040903559648
24
582583584585586587
top related