Sensory modulation disorder and its neural circuitry in adults with ADHD: A pilot study

Abstract

Background:

Compared to healthy controls (HCs), individuals with attention-deficit/hyperactivity disorder (ADHD) exhibit more symptoms of sensory processing disorder (SPD), which is associated with difficulties in educational and social activities. Most studies examining comorbid SPD-ADHD have been conducted with children and have not explored relations to brain volumes. In this pilot study, we assessed a subtype of SPD, sensory modulation disorder (SMD), and its relation to select brain volumes in adults with ADHD.

Methods:

We administered part of the Sensory Processing 3-Dimensions Scale (SP3D) to assess subtypes of SMD and collected structural imaging scans from 25 adults with ADHD and 29 healthy controls (HCs).

Results:

Relative to HCs, subjects with ADHD scored higher on sensory craving (SC) and sensory under-responsivity (SUR) subscales. Although sensory over-responsivity (SOR) was marginally higher, this was no longer true when accounting for co-occurring anxiety. In individuals with ADHD, both SC and SUR were positively associated with amygdalar volume, SUR was also positively associated with striatal volume, whereas SOR was negatively associated with posterior ventral diencephalon volume.

Conclusion:

These preliminary findings suggest that SC and SUR may be characteristic of ADHD while SOR may be driven by co-occurring anxiety. Because different modalities were associated with different brain volumes, our findings also suggest that the modalities may involve unique neural circuits, but with a partial overlap between SC and SUR. These pilot data provide support for conducting studies examining SMD in larger samples of adults with ADHD to determine reproducibility, applicability and implications of these findings.

Introduction

Attention Deficit Hyperactivity Disorder (ADHD) is a developmental disorder characterized by inattention, hyperactivity, and impulsivity (Biederman, 2005). Although it begins in childhood, ADHD often persists into adulthood with a worldwide prevalence estimate of approximately 2.5% (Simon, Czobor, Bálint, Mészáros, & Bitter, 2009). Additionally, clinically referred adults with ADHD show high rates of a range of psychiatric comorbidities (Piñeiro-Dieguez, Balanzá-Martínez, García-García, Soler-López, & the CAT Study Group, 2016; Sobanski, 2006; Sobanski et al., 2007).

Sensory processing disorder (SPD) in children with ADHD.

SPD is one comorbidity that has been observed in ADHD, though primarily in children (Miller, Nielsen, & Schoen, 2012; Pfeiffer, Daly, Nicholls, & Gullo, 2015). It is characterized by unusual detection of or response to sensory stimulation, resulting in atypical behavior (Miller, Anzalone, Lane, Cermak, & Osten, 2007). SPD is characterized by difficulties with sensory discrimination, modulation (over- or under-responsivity and craving), and motor control (balance), and can occur in a range of sensory domains (e.g., visual, auditory, tactile).Such sensory processing characteristics may hinder social and cognitive development in children and adolescents (Koenig & Rudney, 2010), and may also negatively impact the everyday life of youth with ADHD. For example, these difficulties have been associated with impairments in the abilities of individuals with ADHD to participate in educational (e.g. study and examinations; Clince, Connolly, & Nolan, 2016), self-perceived leisure (Clince et al., 2016; Engel-Yeger & Ziv-On, 2011), and social activities (Clince et al., 2016). Therefore, understanding SPD in ADHD and its related neural circuitry might help increase the quality of life for such individuals while also lending clues to its underlying neurobiology.

There has been increasing interest in characterizing sensory processing in ADHD. Questionnaire-based measures have shown that children with ADHD exhibit both under- and over-responsivity for a range of auditory, visual, touch, and oral processing domains (Cheung & Siu, 2009; Dunn & Bennett, 2002; Ghanizadeh, 2011; Miller et al., 2012; Pfeiffer et al., 2015; Schoen, Miller, & Sullivan, 2016; Yochman, Alon-Beery, Sribman, & Parush, 2013). For instance, one study using the Sensory Profile found that children were over-responsive to certain fabrics, under-responsive to nociceptive stimuli, and presented with visual perceptive difficulties, such as those required to piece together puzzles (Yochman, Parush, & Ornoy, 2004). Notably, these studies typically relied on parent-report measures, such as the Sensory Profile, which some suggest may fail to accurately capture the child’s own sensory perceptions (Achenbach, McConaughy, & Howell, 1987; De Los Reyes & Kazdin, 2004; Tavassoli et al., 2019; Williams, Failla, Gotham, Woynaroski, & Cascio, 2018).

In contrast to questionnaire-based studies, which show both under- and over-responsivity in children with ADHD, physiological studies have more consistently indicated heightened physiological responsivity, or sensory over-responsivity (SOR), in children with ADHD using measures such as somatosensory evoked potential (Parush, Sohmer, Steinberg, & Kaitz, 1997) and sympathetic markers of nervous system functioning using electrodermal reactivity (EDR) (Mangeot et al., 2001; McIntosh, Miller, Shyu, & Hagerman, 1999; Miller, Reisman, McIntosh, & Simon, 2001). Because physiological studies present more objective measurements of sensory response, the relationship between SOR and ADHD has gained interest (Lane, Reynolds, & Thacker, 2010; Mangeot et al., 2001; Parush, Sohmer, Steinberg, & Kaitz, 2007; Reynolds & Lane, 2008, 2009). Consistent with this specific interest, Lane and colleagues (2010) used a scale, the SensOR (Schoen, Miller, & Green, 2008) to assess sensory over-responsivity across various domains, and found heightened SOR in children with ADHD. Notably, a strength of this scale, in contrast to others like the SP, is that it does not measure behaviors such as attention and hyperactivity that are considered to be derivatives to sensory processing features (Williams et al., 2018; Yochman et al., 2013; Yochman et al., 2004). For example, one item in the SP specifically assesses attention: ‘Has difficulty paying attention.’ In contrast, all items in the SensOR Inventory address bothersome sensations (e.g., These sounds bother me: sirens, alarms, etc.). As behaviors of attention and hyperactivity are core features of ADHD, measures like the SP may falsely attribute core features of ADHD as sensory processing features. Using the SensOR will avoid this potential issue.

SPD in adults with ADHD.

Although the severity of SPD symptoms has been shown to correlate positively with age in children with ADHD (Cheung & Siu, 2009) and symptoms of SPD have been confirmed in both healthy adults (Heller, 2002) and adults with autism (Crane, Goddard, & Pring, 2009), only one study has evaluated SPD in adults with ADHD. Using the Dutch version of the Adult/Adolescent Sensory Profile, Bijlenga and colleagues (2017) reported low registration of stimuli, high sensory sensitivity, low sensory craving and high sensory avoiding behaviors in adults with ADHD. This was found independently of autism spectrum disorder symptoms. Although the authors found both low registration and high sensory sensitivity, individuals with ADHD displayed low sensory craving behavior, which is more consistent with sensory over-responsivity. Overall, these data may suggest that adults with ADHD experience predominantly sensory over-responsivity, which would be consistent with the physiological data observed in children with ADHD. However, as noted above, the SP (used in Bijenga’s study) includes items specific to the core features of ADHD which could reflect features of ADHD rather then sensory processing features themselves. Additionally, Bijlenga and colleagues, compared SPD scores from adults with ADHD only to norms for sensory processing difficulties rather than a control group. As such, it is uncertain whether these sensory difficulties would be evident when compared equivalently to a well-matched group of non-ADHD peers.

Because sensory over-responsivity (SOR) has been best characterized in children with ADHD relative to sensory under-responsivity (SUR) and sensory craving (SC) (Ben-Sasson, Soto, Heberle, Carter, & Briggs-Gowan, 2017; Van Hulle, Schmidt, & Goldsmith, 2012), our primary aim was to determine whether we would find similar findings of SOR in adults with ADHD. As such, we chose to use the SensOR, similar to Lane and colleagues (2010) who used this scale to find heightened SOR in children with ADHD. Currently, the SensOR is embedded in a newer scale, the Sensory Processing 3-Dimensions Scale (SP3D), that also includes scales assessing the remaining SMD subtypes: SUR and SC. As such, we chose to use the SOR as our primary measure, while exploring SUR and SC so that we could better identify a range of sensory processing characteristics in adults with ADHD (Schoen, Miller, & Sullivan, 2016; Schoen, Miller, & Sullivan, 2014).

SOR and its relationship to anxiety and loneliness.

It is unclear whether symptoms of SPD experienced by adults with ADHD are secondary to other variables relevant to both ADHD and SPD, such as anxiety and loneliness. Previous studies have linked symptoms of SOR with anxiety in ADHD (Lane et al., 2010; Reynolds & Lane, 2009; Reynolds, Lane, & Gennings, 2010) and ASD (Neil, Olsson, & Pellicano, 2016). One study utilized the SenSOR Inventory and found a positive association between anxiety levels and SOR in children with ADHD (Lane et al., 2010). We hypothesized that, similar to previous reports in children with ADHD, levels of anxiety would be associated with higher levels of SOR in adults with ADHD as well. Loneliness has also been shown to be elevated in both children and adults with ADHD, and a connection between hypervigilance (state of sensory sensitivity) and loneliness/perceived social isolation has been reported previously (J. T. Cacioppo & Cacioppo, 2014; Grygiel, Humenny, Switaj, Rebisz, & Anczewska, 2014; Langher, Ricci, Reversi, & Krstikj, 2009; Michielsen et al., 2015; Young & Bramham, 2007). Studies have found that loneliness results in an implicit hypervigilance to social threats (J. T. Cacioppo & Cacioppo, 2014; J. T. Cacioppo & Hawkley, 2009; J. T. Cacioppo, Norris, Decety, Monteleone, & Nusbaum, 2009; S. Cacioppo et al., 2016) and is associated with tonic alertness (Layden et al., 2017). Because features of loneliness, namely hypervigilance and tonic alertness, are also features of anxiety, we predicted that loneliness would likewise be associated with higher levels of SOR in individuals with ADHD. Additionally, while anxiety has been linked to SOR in children with ADHD, the relationship between SUR and SC with conditions such as anxiety or loneliness have not yet been assessed to our knowledge. As such, we also explored the relationship of these conditions with these two SMD subtypes.

Neural correlates of sensory processing characteristics.

Understanding the neural substrates of different sensory processing features can be useful in identifying biomarkers for different ADHD groups. However, no study has yet used structural neuroimaging to examine the neural substrate of SOR in either children or adults with ADHD. Therefore, the final aim of this study was to examine relationships between SOR and select brain volumes hypothesized to be associated with sensory processing disorder. Our a priori regions included the basal ganglia, more specifically, the caudate nucleus, putamen, globus pallidus (or pallidum), nucleus accumbens (NAc), subthalamic nucleus (STN) and substantia nigra (SN), hypothesized to play a role in SPD (Koziol, Budding, & Chidekel, 2011), and the amygdala, which has been associated with sensory processing and gating (LeDoux, 2003; Morris et al., 1998; Turner & Herkenham, 1991; Zald, 2003). We also explored the relationships of SUR and SC with these brain volumes as well.

In summary, our aims are to examine/explore: 1) subtypes of SMD in adults with ADHD relative to matched controls; 2) relationships with SMD and both loneliness and anxiety; and 3) relationships between SMD subtypes and volumes of the BG and amygdala.

Methods

Participants

Subjects were 25 adults with ADHD and 29 healthy controls (HC) recruited locally through online advertisements, flyers, and previous participation in ADHD studies at MGH. Subject demographics are listed in Table 1. There were no between-group differences in sex distribution (χ²(1) = .968, p = .325), age (t(52) = .284, p = .777), or IQ (t(52) = .122, p = .903). Subjects provided written informed consent for participation in the study, which was approved by the Partners Human Research Institutional Review Board. For all but the sensory processing and loneliness evaluation of the subjects, exclusion criteria included current use of psychotropic medications (except short-acting psychostimulants for ADHD subjects), IQ < 80, any neurological disorder or major sensorimotor handicaps, a current DSM-IV Axis I mood, psychotic or anxiety disorder (excluding simple phobias), and either a current or chronic history of alcohol or substance abuse/dependence. ADHD subjects currently taking psychostimulants were asked to refrain from taking them 24 hours prior to testing within the lab. The sensory processing subscales and loneliness questionnaires were administered online at a later date for all subjects using REDCap electronic data capture tools hosted at Partners (Harris et al., 2009).

Table 1.

Subject demographics and characteristics.

	ADHD (N = 25)		Controls (N = 29)
	Mean	SD	Mean	SD
Sex (M/F)	11/14		9/20
Age (years)	26.4	8.7	27.0	5.8
Estimated
IQ	119.4	13.5	119.9	13.9

ADHD, Attention Deficit Hyperactivity Disorder; SD, Standard Deviation.

Diagnostic Measures

Psychopathology for exclusion criteria was assessed using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID; (First, Spitzer, Gibbon, & Williams, 2012)). ADHD assessments were conducted with a modified version of the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime version (K-SADS; (Kaufman et al., 1997)). This structured clinical interview assesses the onset of DSM-IV symptoms of ADHD during childhood and its persistence into adulthood. Estimated full-scale IQ was assessed using the vocabulary and matrix reasoning sub-tests from the Wechsler Abbreviated Scale of Intelligence (WASI; (Wechsler, 1999)).

Assessment of sensory processing disorder, anxiety, and loneliness

Sensory processing was assessed via the sensory over-responsivity (SOR), sensory under-responsivity (SUR), and sensory craving (SC) sensory modulation disorders (SMD) subscales within the Sensory Processing 3-Dimensions Scale. (SP3D; S. Mulligan et al., 2018; Schoen et al., 2016; Schoen et al., 2014). The three SMD subscales include 147 items that are divided into additional subscales categorized by sensory domain, resulting in five domains for each subscale: movement, touch, vision, audition, and olfaction. Questions are phrased in a yes/no manner. SOR refers to being overly sensitive to sensory stimulation (e.g., “These visual sensations bother me…”); SUR refers to a reduced sensitivity to sensory stimulation (e.g., “Typically I do not notice…”); SC refers to having a craving for or constantly seeking sensory stimulation (e.g., “I have a constant desire for…”). All three scales were validated with both children and adults using typical cases and individuals with a sensory modulation disorder (SOR, SUR, or SS) determined by a comprehensive exam with an occupational therapist. The SensOR Inventory has high discriminant validity (p’s<0.05-<.001) and reliability (.62-.94) in adults (Schoen et al., 2008). Preliminary validation studies show strong discriminant validity for SUR (p’s<0.02-<.001) and SC (p’s< .001) and adequate internal reliability consistency for SUR (.59-.81) and for SC (.67-.87) (S. Schoen, personal communication, January 13, 2020).

Anxiety was measured through the Anxiety Problems sub scale of the Achenbach System of Empirically Based Assessment-Adult Self-Report (ASEBA ASR; (Achenbach, 2009)). Scales show reliability between 0.71 and 0.99 (Achenbach & Rescorla, 2003) and assesses adaptive functioning, DSM-oriented problems, and substance use. Scores reported are standardized T-scores.

Loneliness was measured using the De Jong Gierveld (DJG; (Gierveld & Tilburg, 2006)) Scale, which assesses emotional and social loneliness. Emotional loneliness results from feeling a lack of intimacy in one’s relationships, while social loneliness results from feeling that one’s social network is too small. The reliability coefficients for the 6-item scale vary between .70 and .76, while the validity coefficients vary between .93 and .95, indicating reasonable to good reliability and validity in an adult population (De Jong Gierveld & Van Tilburg, 2006). It consists of 6 items with each item being rated as a ‘yes’, ‘more or less’, or ‘no’.

MRI acquisition and Image Processing

All participants underwent MRI scanning. T1-weighted multi-echo MEMPRAGE structural MRIs were acquired on a Siemens Tim Trio 3-Tesla scanner from 31 participants (13 ADHD/18 HC) and on a Siemens CONNECTOM 3-Tesla scanner from 23 participants (12 ADHD/11 NC) (Siemens Medical Solutions, Erlangen, Germany). Chi-square analysis revealed no difference in proportion of ADHD and controls run on each of the two scanners (χ²(1)= .338, p = .561), and no difference in sex between the diagnoses for each scanner (Tim trio: χ²(1)= 1.495, p = .221; CONNECTOM: χ²(1)= .164, p = .686). Parameters of the T1-weighted MEMPRAGE scans collected from the Tim trio scanner were: TR = 2.54 s; TE = 1.64/3.5/5.36/7.22 ms; TI: 1.2s; FoV = 256 mm²; voxel size = 1.0 × 1.0 × 1.0 mm, while the parameters of the scans collected from the CONNECTOM scanner included TR = 2.53 s; TE = 1.15/3.03/4.89/6.75 ms; TI: 1.1s; FoV = 256 mm²; voxel size = 1.0 × 1.0 × 1.0 mm.

Structual imaging analyses were conducted using FreeSurfer v.5.3 (http://surfer.nmr.mgh.harvard.edu). which applies an automatic technique to measure brain structure volumes (Fischl et al., 2002). All brain images were manually reviewed and checked for quality control. Regions of interest (ROIs) included the amygdala and basal ganglia structures, specifically the striatum (the ensemble of the caudate, putamen, and nucleus accumbens (NAc)), pallidum, subthalamic nucleus (STN) and substantia nigra (SN). It should be noted that the STN and SN are included in the FreeSurfer segmentation of the ventral diencephalon (VDC). We define the VDC as has been defined previously (Makris et al., 1999) and as is done for the Freesurfer software package (Fischl et al., 2002). The ventral diencephalon (Fischl et al., 2002) is the anatomical region below the thalamus comprising part of the basal forebrain, the entire hypothalamus, and more posteriorly the subthalamic nucleus, red nucleus, and a considerable part of the substantia nigra. It can be divided into two parts: an anterior part, i.e. anterior VDC comprising part of the basal forebrain and the hypothalamus; and a posterior part i.e. posterior VDC comprising the subthalamic nucleus, red nucleus, most of the substantia nigra, and several fiber tracts passing through it. The anterior border of the posterior VDC can be defined by a coronal plane immediately posterior to the mammillary bodies. The posterior border of the posterior VDC is an oblique plane from the posterior commissure to the interpeduncular fossa. To assess the SN and STN, we used the posterior component of the segmented the VDC. Segmentation was completed by an expert neuroanatomist (NM). See Figure 1 that shows an example image of the anterior/posterior division of the VDC as done in this study. In sum, we had 4 a priori ROIs: 1) amygdala, 2) striatum (the ensemble of the caudate, putamen, and Nac), 3) pallidum, and 4) posterior VDC (including the STN and SN).

Figure 1. Manual segmentation of the anterior and posterior VDC.

Left and right anterior VDC are represented in light and dark purple, respectively, while left and right posterior VDC are represented in light and dark red, respectively. The segmentation is presented in 2D sagittal (A) and coronal (B) views. An example tracing of the parcellation of the VDC into its anterior and posterior regions is shown in the sagittal view (C). 3D representation of the segmentation (D). aVDC: anterior Ventral Diencephalon; IPF: Interpeduncular fossa; MB: Mammillary body; PC: Posterior commissure; pVDC: posterior Ventral Diencephalon; VDC, Ventral Diencephalon.

Statistical analyses

Analyses were conducted using R 3.5.0(https://cran.r-proiect.org). Between-group differences in sensory modulation subtypes, anxiety, and loneliness were assessed with two-tailed t-tests. A One-Way Analysis of Covariance (ANCOVA) was performed to assess between-group differences on the SMD subtypes, while adjusting for loneliness and anxiety as covariates. Relationships between brain volumes and SMD subtypes were assessed with partial Pearson’s correlations, correcting for intracranial volume (ICV). A p-value < .05 indicated statistical significance except when analyzing sensory domains. Because there were 5 sensory domains in each of the 3 subscales, 15 tests were conducted to assess sensory domains across all SMD subtypes. A Bonferroni correction with a corrected alpha of 0.003 was used to account for this large number of tests. Outliers, defined as data ± 3 standard deviations away from the mean, were excluded from analyses: One HC from the DJG scales, 1 ADHD and 1 HC from the ASEBA-ASR Anxiety Problems scale, 3 HC subjects from brain volume analyses, and 1 ADHD from amygdala volume analyses. One ADHD subject with incomplete data for all SP3D subscales and 1 HC subject with incomplete data for the SC subscale were excluded from analyses as well.

Results

Behavioral measures

SMD

Participants with ADHD scored significantly higher compared to HC on the SC (t(24) = 4.698, p < .001) and SUR (t(51) = 2.413, p = .019; Fig. 1) subscales, and marginally higher on the SOR subscale (t(51) = 1.947, p = .057; Cohen’s d effect size (ES) = .534; Fig. 2). When examining at the domain level, participants with ADHD scored higher within the visual domain on the SC subscale (SC: t(27.1) = 4.04, p < 0.001; Table 2) and within the auditory domain on the SC and SUR subscales (SC: t(24.51) = 3.94, p < 0.001; SUR: t(26.87) = 4.65, p < 0.001; Table 2).

Figure 2. Adults with ADHD show higher mean scores of SUR and SC, and marginally higher scores of SOR compared to healthy controls.

Error bars denote SEM. Two-tailed t-tests. ADHD, Attention Deficit Hyperactivity Disorder; SC, Sensory Craving; SOR, Sensory Over-Responsivity; SUR, Sensory Under-Responsivity; SP3D, Sensory Processing 3-Dimensions.

Table 2.

Sensory processing scores (Mean ± SD) by domain and modality for each group.

		ADHD (N = 25)		Controls (N = 29)
		Mean	SD	Mean	SD
SS	Tactile	0.67	1.09	0.18	0.39
	Vision*	0.54	0.59	0.04	0.19
	Audition*	1.17	1.34	0.07	0.26
	Olfaction	0.46	0.72	0.18	0.48
	Movement	2.21	0.47	0.61	2.02
SUR	Tactile	1.75	2.11	0.97	1.02
	Vision	0.92	1.44	0.28	0.80
	Audition*	1.63	1.47	0.17	0.47
	Olfaction	0.21	0.41	0.03	0.19
	Movement	0.13	0.34	0.00	0.00
SOR	Tactile	2.71	2.24	1.86	1.79
	Vision	1.88	1.65	0.72	0.92
	Audition	4.67	3.13	3.45	3.32
	Olfaction	0.88	0.99	0.66	0.86
	Movement	0.67	1.01	0.69	1.11

^*denotes significance, with p < 0.003.

ADHD, Attention Deficit Hyperactivity Disorder; SC, Sensory Craving; SD, Standard Deviation; SOR, Sensory Over-Responsivity; SUR, Sensory Under-Responsivity.

Loneliness and anxiety

Adults with ADHD scored similarly to HC on the social loneliness scale (t(51) = .835, p = .408; Fig. 3A), and marginally higher on the emotional loneliness scale (t(39) = 1.941, p = 0.059; Cohen’s d ES = .541; Fig. 3A). Adults with ADHD also showed higher levels of anxiety compared to HC (Mann-Whitney U = 478, HC= 28, ADHD =24, p = .002; Fig. 3B).

Figure 3. Adults with ADHD showed marginally higher levels of (A) emotional loneliness and higher levels of (B) anxiety.

(A) Two-sample t-test. Healthy controls = 28, ADHD = 25. (B) Mann-Whitney U. Healthy controls = 28, ADHD = 24. Scores reported for anxiety are standardized T-scores. Error bars denote SEM. ADHD, Attention Deficit Hyperactivity Disorder.

Given that anxiety and emotional loneliness were significantly or marginally higher in adults with ADHD, we assessed whether between-group differences observed on the SMD subscales were independently associated with diagnosis or accounted for by the effects of these variables. When accounting for anxiety and loneliness, a One-Way Analysis of Covariance (ANCOVA) revealed that diagnosis had a significant effect on SC and a marginally significant effect on SUR, but there was no longer a trend for SOR (SC: F(1,44) = 13.347, p = .001; SUR: F(1,47) = 3.030, p = .088; SOR: F(1,47) = .247, p = .622). Furthermore, there was a significant effect of anxiety on SOR (SOR: F(1,47) = 4.519, p = .039; SC: F(1,44) = 2.313, p = .135; SUR: F(1,47) = .193, p = .663), while emotional loneliness had a significant effect on all three SMD subscales (SC: F(1,44) = 42.096, p <.001; SUR: F(1,47) = 4.769, p = .034; SOR: F(1,47) = 5.810, p = .020).

Relations between Brain volumes and Sensory Processing subscales

Partial Pearson’s correlations, correcting for intracranial volume (ICV), were performed for adults with ADHD between our ROIs and the three SMD subtypes. For adults with ADHD, we found a positive correlation between the amygdala and SC and SUR subscales (SC: r(20) = .461, p = .031; SUR: r(20) = .452, p = .035) and between the striatum and SUR (r(20) = .427, p = .042). We also found a marginally negative correlation between the posterior VDC and SOR subscale (SOR: r(21) = −.385, p = .069). Because anxiety accounted for the between-group difference for SOR, we conducted a partial correlation correcting for both ICV and anxiety to further investigate potential effects of these possible confounds on this relationship. After adjusting for both variables, we found a significant negative correlation between the posterior VDC and SOR (r(18) = −.586, p = .005). (In exploratory analyses for the HC subjects, we found no significant relations between the SMD subscales and ROIs.)

Discussion

In this preliminary study, we show that, compared to controls, adults with ADHD report marginally higher levels of sensory over-responsivity (SOR) as well as significantly higher levels of sensory craving (SC) and under-responsivity (SUR). Sensory processing differences were observed in the visual domain within the SC subscale, and in the auditory domain within the SC and SUR subscales. Furthermore, co-occurring anxiety appears to drive sensory over-responsivity in adults with ADHD. We also show that SC and SUR were positively associated with amygdalar volume and SUR was positively associated with striatal volume, while SOR was negatively associated with volume of the posterior ventral diencephalon, which includes the subthalamic nuclei and substantia nigra. Overall, these preliminary data provide evidence that sensory under-responsivity and craving may be characteristic of ADHD and to provide clues to their underlying neural substrates.

Sensory Modulation (SMD) in Adults with ADHD

Our data, showing elevated levels of SC, SUR, and (marginally) SOR, are in accordance with the previous report by Bijlenga and colleagues (2017), that show relatively higher levels of hypo- and hyper-sensitivities in adults with ADHD. We expand on these data by specifying that these sensitivities are specific to sensory under-responsivity, while over-responsivity appears to be secondary to anxiety. Our data also indicate increased sensory craving in adults with ADHD. This finding may seem contradictory to previous findings of Bijlenga and colleagues who found lower sensory craving behavior in adults with ADHD relative to controls. However, this contradiction may be due to differences between the scale we used and the scale they used, the Sensory Profile. The Sensory Profile does not solely map onto SPD subscales but also includes emotional and fine motor items as well as a number of items that fail to group in a components factor analysis (Schoen, Miller, & Green, 2008). With regards to sensory domains, our results of auditory differences in sensory craving and under-responsivity are consistent with their findings of hypo-sensitivity in the auditory modality (Bijlenga, Tjon-Ka-Jie, Schuijers, & Kooij, 2017), although we were also able to identify differences within the visual domain for sensory craving.

Influence of Anxiety and Loneliness on SMD subtypes

While Lane et al. (2010) suggest an overlap between SOR, anxiety, and ADHD in children, we show that anxiety specifically moderates the link between SOR and ADHD in adults (Lane et al., 2010). This finding suggests that only SC and potentially SUR, which was marginally significant after adjusting for emotional loneliness and anxiety, may be inherent to ADHD as opposed to SOR, which appears to be driven by co-occurring anxiety. These results may further illuminate past findings of lower sensory gating in adults with ADHD by suggesting that lower sensory gating, which presents as sensory over-load, may be accounted for by anxiety (Davies, Chang, & Gavin, 2009; Feifel, Minassian, & Perry, 2009; Holstein et al., 2013; Micoulaud-Franchi et al., 2015; Olincy et al., 2000). Though preliminary, these findings are consistent with the primary anxiety model of Ben-Sasson and Green (2010), who proposed three theories on the link between SOR and anxiety: the primary anxiety model, primary SOR model, and a non-causal model linking the two conditions (Green & Ben-Sasson, 2010). In the primary anxiety model, the authors propose that symptoms of anxiety, such as hypervigilance and attentional biases, cause patients to attend to sensory stimuli. This heightened attention then leads to over-responsivity, which is exacerbated through conditioning. These data provide some support for larger studies designed to test this model.

Finally, we also found that emotional, but not social, loneliness is associated with ADHD in adults. It appears that while adults with ADHD do not perceive themselves as having a smaller social network than their peers, they still feel more emotionally isolated, and thus lonelier than controls. This finding is consistent with previous data indicating that adults with ADHD do not only feel lonely, but often have difficulty maintaining interpersonal relationships (Michielsen et al., 2015; Young & Bramham, 2007). With regards to SMD, we found an association between emotional loneliness and all SMD subscales. However, this loneliness did not appear to account for SMD behaviors.

Associations between Brain Volumes and SMD Patterns

When relating brain volumes to sensory processing in adults with ADHD, we found a positive association between amygdalar volume and both sensory craving and under-responsivity, and between striatal volume and sensory under-responsivity. We also found a marginally negative association between the posterior VDC volume and over-responsivity. The correlation between the posterior VDC and SOR became significant once the potentially confounding influence of anxiety was accounted for. As we observed a correlation between SOR and the posterior VDC and between SUR and the striatum, our results support a role of the basal ganglia, localized in the substantia nigra and subthalamic nuclei in SOR and in the striatum in SUR. Because dopamine-related genes have been found in the substantia nigra, which is included in the posterior VDC, our results are consistent with previous studies that suggests SOR may be linked to dysfunctional dopaminergic receptors in the substantia nigra/ventral tegmental area (Schneider et al., 2019) and in the nigrostriatal dopaminergic path (Schneider et al., 2008; Schneider et al., 2007). Overall, our data suggest that SUR and SOR involve different regions of the basal ganglia circuit, potentially involving the dopaminergic system, while SC and SUR may involve the limbic circuit. These data could serve to provide hypotheses for larger studies specifically designed to test these associations.

Limitations

Our study is limited by its sample size. Although it was sufficiently powered to detect several hypothesized effects, it lacked power to conduct more exploratory analyses or detect smaller effects. Indeed, we found a few trends in our data that, with a larger sample size, may have been statistically significant. As such, our results need to be considered preliminary and cautiously until larger samples can replicate the findings. Another potential limitation is our use of two scanners for MRI acquisition. However, we balanced our groups by both diagnosis and gender across the two scanners, so it is rather unlikely that there was a systematic difference accounting for our effects. Nonetheless, future studies with larger samples should be conducted on the same scanner. Also, we did not control for symptoms of autism spectrum disorder (ASD), which often co-occur with ADHD and have been hypothesized to account for SPD symptomatology observed in children with ADHD (Davis & Kollins, 2012; A. Mulligan et al., 2009). However, a recent study of adults with ADHD showed that symptoms of SPD were independent of autism spectrum disorder symptoms, (Bijlenga et al., 2017), supporting the assertion that SPD – at least in adults cannot be fully accounted for by autism spectrum disorder in ADHD. Nonetheless, future studies addressing this question should account for ASD in the study design. Finally, although we selected only a few a prior regions of interest for the brain imaging analyses, we did not correct for multiple comparisons for these regions thus raising the possibility of Type I errors. Relatedly, while many regions have been implicated in the pathophysiology of ADHD, we focused on a few key structures that are involved in both ADHD and SPD, such as the striatum and pallidum (Seidman et al., 2011; Seidman, Valera, & Makris, 2005), in order to maximize our power due to our small sample size. For a more comprehensive understanding of SPD and its neural circuitry in individuals with ADHD, future studies should also examine other regions such as cortical somatosensory regions.

Future Directions

Our preliminary finding of elevated symptoms of SPD in adults with ADHD relative to controls is the first to suggest that sensory under-responsivity and sensory craving are characteristic of ADHD and provide additional support that the co-morbidity of SPD and ADHD is not exclusive to children. Because SPD has been shown to be associated with functional activities, including social participation and school-work (Clince et al., 2016), more research is needed to assess the applicability and implications of our findings in an adult population with ADHD (Koenig & Rudney, 2010).

Additionally, as previous findings have shown altered white matter structure in children with SPD, future examination of white matter in relation to measures of SPD might help us better understand the clinical significance of SPD-ADHD comorbidity. (Chang et al., 2014; Owen et al., 2013). Likewise, analyzing additional aspects of SPD could further characterize SPD in individuals with ADHD. For example, in addition to sensory modulation disorder, which was the focus of this study, it might also be fruitful to examine sensory-based motor disorder, characterized by difficulty with posture, balance, motor coordination and/or motor tasks, in individuals with ADHD. As difficulties with motor coordination and balance have been observed in both children (Fliers et al., 2008) and adults (Hove et al., 2015) with ADHD, assessing motor problems in the context of SPD could reveal new insights into the motor difficulties often observed in ADHD.

Abbreviations:

ADHD

Attention-Deficit/Hyperactivity Disorder

BG

Basal Ganglia

SS

Sensory Craving

SDD

Sensory Discrimination Disorder

SMD

Sensory Modulation Disorder

SOR

Sensory Over-Responsivity

SPD

Sensory Processing Disorder

SUR

Sensory Under-Responsivity

VDC

Ventral Diencephalon