Background

Attention-Deficit/Hyperactivity Disorder (ADHD) is a common neurodevelopmental disorder that is characterized by inattention, hyperactivity, and impulsivity. The fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [1] no longer excludes comorbidity with autism spectrum disorder (ASD) from the diagnosis of ADHD. Numerous pieces of evidence suggest that ADHD and ASD might overlap in genetics, social function and interaction, executive function, and brain imaging changes [2]. Even when a diagnosis of ASD is absent, individuals with ADHD might also exhibit symptoms of autism or autistic traits (ATs) [3].

ATs represent stereotypy and deficits in social interactions, usually defined by positive results on ASD scales. Nevertheless, ATs per se do not meet the clinical threshold for ASD diagnosis [4, 5]. In young people with ADHD, the co-occurrence of ATs can significantly influence their clinical characteristics and their neuropsychological and social functions; this co-occurrence can persist into adulthood [6]. Social deficits might be more associated with oppositional symptoms, while restrictive and repetitive behaviors (RRBs) might predict symptoms of hyperactivity and impulsivity (HI) [7]. In terms of behavioral expression, ADHD probands with ATs (ADHD + ATs) have significantly more problems than those without ATs (ADHD − ATs) [6, 8]. In terms of cognitive functions, impairments in concentration, working memory, nonverbal reasoning, and visuomotor skills are more prominent in ADHD + ATs than in ADHD − ATs [7, 8]. As for social function, patients whose ADHD is comorbid with ATs demonstrate more impairments in terms of problems with spare time, activity impairments, and problems with peers [6].

In addition to the above-mentioned behavioral and cognitive evidence, some studies have tried to investigate ATs in children with ADHD based on brain imaging data to explore the potential underlying neurobiological mechanisms. Most evidence has come from studies comparing ADHD and ASD probands to explore potential overlap** and distinctive brain structural and/or functional alterations. Subcortical-volume changes might be the same between ADHD and ASD, while cortical thickness changes might or might not be the same between the two conditions [9]. Fractional anisotropy of the corpus callosum was found to play an essential role in the overlap** of ADHD and ASD symptoms [10]. ADHD and ASD both feature hypoactivation of the right anterior insula during motor response inhibition tasks [11]. Compared with healthy controls (HCs) and patients who had only ADHD, patients with comorbid ADHD and ASD exhibited decreased functional connectivity (FC) in the local right temporoparietal cortex during theory-of-mind testing [12]. The above-mentioned shared brain features of both conditions, and in particular brain changes in patients with comorbid ADHD and ASD, might help us understand ATs in children with ADHD.

Few studies thus far have directly investigated AT-related changes in brain imaging. A study of structural brain images showed a positive relationship between ASD score and gray-matter volume in adults with ADHD [13]. Although studies have identified several brain region volumes that predict ASD symptoms or affect social cognition, the strong correlations were not unique to ADHD [14, 15]. Cooper et al. explored white-matter (WM) microstructural characteristics of ATs in patients with ADHD. They found that AT-related changes to WM were mainly located in the right posterior limb of the internal capsule/corticospinal tract, right cerebellar peduncle, and midbrain [16], while ADHD severity was correlated with WM microstructure in the left subgenual cingulum [17]. Zhang et al. found that cerebral blood flow in the left middle frontal gyrus was correlated with both ASD and ADHD scores, while that in the left temporal pole was negatively correlated with ASD score, in children and adolescents with comorbid ADHD and ATs. However, their post hoc analysis revealed nonsignificant differences in these clusters between the ADHD groups with and without ATs [18]. One study showed that higher social impairment and more ADHD symptoms in adolescents with ADHD correlated with functional dysconnectivity among the default mode network (DMN), frontoparietal network, and cingulo-opercular network [19]. The above-mentioned studies notwithstanding, evidence from the existing literature is still insufficient to illustrate AT-specific brain changes in ADHD, which might be independent from those correlated with core ADHD symptoms.

Resting-state functional magnetic resonance imaging (rs-fMRI) is a non-invasive tool used to probe the spontaneous neural activity of intrinsic human-brain functional organizations, which likely reflects disorder etiology [20]. Amplitude of low-frequency (0.01–0.08 Hz) fluctuation (ALFF), an index for measuring changes in resting-state blood oxygen level–dependent signals, can relatively and indirectly reflect spontaneous brain activity [66]. Together with the right caudate nucleus, the left precuneus can indicate ADHD in normal groups with 62.52% accuracy [67]. Interventional experiments showed that larger left-precuneus volumes correlated with poor methylphenidate response [68].

In addition to the above-mentioned brain regions, we found a relatively specific correlation between brain functional alteration in the left MOG and ADHD. The MOG is the part of the occipital lobes that contains most of the visual cortex and has been demonstrated to be involved in modulation of unconscious processes by category-selective attention [69]. A recent study showed that Granger causality analysis values from the ventral putamen to the left MOG were significantly negatively correlated with HI symptoms in patients with ADHD [

Conclusions

Among children with ADHD, some subjects also exhibit autistic traits. In this study, subjects with ADHD + ATs showed different behavioral characteristics and potentially specific brain functional alterations. Assessment and exploration of ATs in children with ADHD could help us understand the heterogeneity of ADHD, better explore its pathogenesis, and promote clinical intervention.

Methods

Participants and procedures

For children with ADHD, all subjects were recruited in the clinics of Shenzhen Children’s Hospital (SCH; Shenzhen, China). For HCs, they were recruited from different local elementary schools in Shenzhen by advertisements. A total of 172 medication-naïve patients with clinically diagnosed ADHD based on the DSM-IV from the outpatient department of SCH from June 2017 to December 2020, and 45 HCs were recruited. This cross-sectional study was approved by the Medical Ethics Committee of SCH, and informed consent was obtained from all participants and their parents.

Patients came to the clinic on their first visit. They underwent a semi-structured clinical interview that used the Kiddie Schedule for Affective Disorders and Schizophrenia Present and Lifetime Version (K-SADS-PL) to confirm the diagnosis of ADHD and exclude other psychiatric disorders [71]. Inclusion criteria for children with ADHD were as follows: (1) meeting the diagnosis of ADHD; (2) age 6–16 years; (3) FSIQ assessed using the Wechsler Intelligence Scale for Children, fourth edition ≥ 70; and (4) without any drug, behavioral, or psychological intervention for ADHD. Exclusion criteria were as follows: (1) ASD or sleep disorder diagnosis; (2) tic disorders, intellectual disability, conduct disorder, schizophrenia, affective disorder, or other psychiatric disorder; and (3) physical illnesses and neurological disorders such as epilepsy, short stature, congenital heart disease, enuresis, or immune encephalitis. HCs who were willing to participate our project came to our clinic for inclusion criteria and the exclusion criteria evaluation based on the clinical interview using K-SADS-PL. Any present or lifetime diagnosis of any psychiatric disorder led to exclusion. Other inclusion and exclusion criteria for HCs were the same as those for the ADHD group.

All participants’ parents were asked to complete the SNAP-IV and ASSQ scales for ADHD and ASD symptom evaluation. Subjects with a total score of ASSQ ≥ 12 were defined as having ATs [72]. The ADHD and HC groups was further divided into two subgroups separately, ADHD + ATs, ADHD–ATs and HC + ATs, HC–ATs. Ultimately, only one HC combined with ATs, that cannot be one group, so we excluded this single one for comorbidity with ATs and kept the left three groups (ADHD + ATs, ADHD − ATs and HC − ATs) for analysis.

For the imaging analyses, only subjects who agree to participate and met the following criteria were included for data acquisition: (1) right-handed; (2) do not have metal implants (including non-removable dentures); (3) not suffering from claustrophobia. Any visible abnormalities on MRI images as examined by an experienced radiologist (e.g., cysts) during MRI scans, or excessive head motion with > 3 mm of translation or > 3° of rotation in any direction, led to exclusion for brain analysis. Five children with ADHD were excluded for further imaging analyses due to their excessive head motion.

Finally, 67 ADHD + ATs, 105 ADHD − ATs, and 44 HC − ATs were included for behavioral analyses; and 21 ADHD + ATs, 38 ADHD − ATs, and 43 HC − ATs for imaging analyses.

Measurements

Swanson, Nolan, and Pelham rating scale, version IV

We used the SNAP-IV parent form to evaluate ADHD symptoms. The form includes 18 items on a three-point rating scale (0 = “not at all,” 1 = “just a little,” 2 = “quite a bit,” and 3 = “very much”). The Chinese version of SNAP-IV has been demonstrated to be a reliable and valid instrument, with satisfactory reliability (all Cronbach’s α coefficients of SNAP-IV sub-scales > 0.88) and sufficient sensitivity (0.87) and specificity (0.79) [73].

Autism spectrum screening questionnaire

The ASSQ consists of 27 items on a three-point rating scale (0 = “normal,” 1 = “some abnormality,” and 2 = “definite abnormality”). Nine items were designed for social interaction, 7 for communication problems, and 11 for RRBs [72]. The ASSQ has been proven to have good internal consistency (Cronbach’s α = 0.86) [74]. To define the ADHD + AT group in our study, we used the Mandarin Chinese version, whose cutoff value of 12 can distinguish individuals with ASD from unaffected controls [72].

Resting-state functional MRI

Data acquisition

We acquired rs-fMRI data of children at SCH using a Siemens Skyra scanner (Siemens Healthcare, Forchheim, Germany) with a standard 12-channel head coil. Functional images were acquired using an echo-planar imaging sequence with the following parameters: repetition time = 2000 ms, echo time = 30 ms, flip angle = 90°, thickness/skip = 3.5/0.7 mm, matrix = 64 × 64, field of view = 200 × 200 mm, 33 axial slices, 240 volumes, and 3 × 3 mm in-plane resolution. All participants were asked to close their eyes without falling asleep during the 8 min of fMRI scanning.

Data processing

We acquired resting-state functional images using RESTplus [75] on the MATLAB R2014a platform. Data preprocessing included the following steps: excluding the first 10 time points, slice timing, realignment, segmentation of each participant’s images to the echo planar imaging standard template and then normalization to the Montreal Neurological Institute space (resampled voxel size = 3 × 3 × 3 mm3), spatial smoothing with a 6-mm full-width-at-half-maximum Gaussian filter, linear detrending, and nuisance covariate regression (including the friston 24-parameter model, WM signal, and cerebrospinal-fluid signal) [76].

For higher test–retest reliability in gray matter regions and more sensitive for discerning differences between individuals and groups [26], ALFF, rather than fALFF, was chosen to identify our hypothesis. After preprocessing, we calculated ALFF. Mean ALFF (mALFF; ALFF at each voxel normalized by the mean ALFF of voxels in gray matter) was used for statistical analyses.

Statistical analysis

For behavioral data, analysis of two sample t-test, or variance (ANOVA) and post hoc analysis with the least-significant-difference (LSD) t-test, χ2 test, or Fisher’s exact test were used to compare the demographic and clinical characteristics of the HC − ATs and ADHD two groups or HC − ATs, ADHD − ATs and ADHD + ATs three groups. We used general linear-model analysis with sex, age, and FSIQ as covariates, or analysis of covariance (ANCOVA) and post hoc analysis with the LSD t-test, to compare AT symptoms and ADHD symptoms between subjects with ADHD and HC − ATs or among the three groups. Partial-correlation analysis with sex, age, and FSIQ as covariates was used to separately explore the relationships between AT symptoms and ADHD symptoms in the ADHD and HC − ATs groups.

For neuroimaging data, we used ANCOVA and RESTplus to compare the mALFFs of children in the ADHD + AT, ADHD − AT, and HC − AT groups, with sex, age, and FSIQ as covariates. The multiple-comparison correction was based on Analysis of Functional NeuroImages’s AlphaSim, with a threshold of P < 0.01 and cluster size ≥ 42 [77]. We then extracted mALFF signals from each significant cluster to perform the following post hoc analysis using the LSD t-test: relationships between mALFF values of different brain regions and symptoms as assessed by ASSQ and SNAP-IV were investigated using stepwise multiple linear-regression analyses, after elimination of factors whose variance inflation factor values were > 10.

We performed all behavior and mALFF value statistical analyses using SAS version 9.1 (SAS Institute, Cary, NC, USA) and SPSS version 23 (IBM Corp., Armonk, NY, USA). P < 0.05 was considered significant.