Increased Sensory Processing Atypicalities in Parents of Multiplex ASD Families Versus Typically Developing and Simplex ASD Families

Considering that ASD has moderate to strong genetic influences (Hallmayer et al. 2011; Ritvo et al. 1985), and that sensory processing abnormalities are shown to be moderately heritable (Goldsmith et al. 2006) and highly prevalent in individuals with ASD (Baranek et al. 2006), it has been suggested that sensory processing abnormalities may share genetic influences with the defining characteristics of ASD (DeLorey et al. 2011; Peñagarikano et al. 2011; Tavassoli et al. 2012). Rodent models of ASD have found that several genetic and epigenetic insults known to produce ASD-like symptoms of social impairments and repetitive/stereotypic behaviors in rodents have also resulted in sensory processing abnormalities (for a review, see Argyropoulos et al. 2013). For instance, Peñagarikano et al. (2011) found that a knockout of CNTNAP2, a well-known ASD candidate gene, resulted in increased thermal and olfactory hypersensitivity in the affected mice. Similarly, DeLorey et al. (2011) found that heterozygosity for the ASD candidate gene GABRB3 in rodents was significantly associated with increased thermal and tactile hypersensitivity. Following these studies, Tavassoli et al. (2012) investigated whether hypersensitivity was also related to GABRB3 variations in humans. They found that behavioral and parent-report measurements of tactile hyperresponsivity in typically developing children were associated with common variations in the GABRB3 candidate gene, thus further supporting genetic implications of sensory processing abnormalities in ASD.

In addition to molecular genetics methods, an alternative method of examining whether particular symptoms are related to genetic influences of ASD is through the assessment of these symptoms in parents and relatives of ASD probands. Relatives of individuals with ASD have an increased chance of displaying mild autistic traits (Bailey et al. 1998; Piven et al. 1997; Szatmari et al. 2008; Taylor et al. 2013), described as the “broader autism phenotype” (BAP; Bolton et al. 1994). BAP traits may represent mild phenotypic expressions of the same genetic influences responsible for ASD (Bailey et al. 1998; Piven 2001). Thus, investigating ASD-like traits in relatives can aid in the search for intermediate phenotypes that may contribute to the genetic liability for ASD (Piven 2001).

The likelihood of BAP traits in relatives is higher in families in which multiple members are diagnosed with ASD (multiple-incidence/multiplex families; MPX) compared to families in which only one relative has ASD (single-incidence/simplex families; SPX), which is likely due to differing genetic mechanisms in these families. While some cases of ASD are heavily influenced by spontaneous de novo mutations (rare genetic mutations present in the child but absent in the parents) of large effect producing sporadic cases of ASD (i.e., SPX) (O’Roak et al. 2012; Sebat et al. 2007), others might be due to the inheritance of ASD-related genes producing familial cases of ASD (i.e., MPX) (Virkud et al. 2009). Supporting the former hypothesis, Sebat et al. (2007) found that the rate of de novo copy number variants was significantly higher in ASD probands from SPX (10 %) compared to those from MPX (3 %) families and control families with no diagnoses of autism (1 %).

Additional evidence supporting the genetic distinction between MPX and SPX families derives from several studies showing that ASD-related traits were more common in relatives from MPX families than those from SPX families (e.g., Bernier et al. 2012; Losh et al. 2008), implying that ASD in the former group is strongly influenced by inheritance of familial ASD traits. Szatmari et al. (2000) evaluated ASD traits in 1362 biological relatives of 78 ASD probands and found that social impairments were significantly more prominent in MPX than in SPX relatives (p < .001), and MPX relatives were also more likely than SPX to have impairments in two or more areas of the three primary symptom categories of ASD (p = .05). Subsequent studies found similar results, showing that, compared to SPX relatives, MPX relatives (usually parents and/or unaffected siblings) had worse pragmatic language and lower quality friendships (Losh et al. 2008), less social interest and less non-verbal communication expression (Gerdts et al. 2013), less social motivation (Bernier et al. 2012), and scored higher on the Social Responsiveness Scale (SRS), signifying more ASD-related deficits (Constantino et al. 2010). A recent study by Oerlemans et al. (2015) also found significant differences in the number of autistic traits between SPX and MPX unaffected siblings, but not between unaffected SPX/MPX parents. Only one study failed to find differences in autistic traits between MPX and SPX families, although they did find marginally significant differences between MPX fathers and control fathers on the SRS (De la Marche et al. 2011). To date, no study has evaluated sensory processing atypicalities in MPX versus SPX families, and therefore it is currently unclear if this common ASD symptom may contribute to the genetic liability for familial cases of ASD.

Two recent studies have evaluated sensory processing in relatives of ASD individuals (De la Marche et al. 2012; Uljarević et al. 2014), although neither used the MPX/SPX distinction outlined above. Both studies used the adolescent/adult sensory profile (AASP) self-report questionnaire, which consists of four quadrants (subscales) corresponding to different types of sensory reactions: hyposensitivity, hypersensitivity, sensation seeking and sensation avoidance (Brown and Dunn 2002). De la Marche et al. (2012) compared 56 non-affected adolescent siblings of ASD individuals with 33 adolescent control participants, and found similar AASP scores between groups except for the Sensation Seeking quadrant, in which siblings of ASD individuals scored significantly lower than adolescent controls (p < .001). The authors concluded that decreased sensation seeking behaviors might be an endophenotypic trait of ASD.

Only one study to our knowledge has investigated sensory processing atypicalities in parents of children with ASD. Uljarević et al. (2014) found that 49 of the 50 mothers (98 %) of children with ASD scored one or more standard deviations (SD) outside the AASP normative means on at least one quadrant of the AASP, and 22 (44 %) scored two or more SDs outside the norms. While Uljarević et al.’s study produced valuable initial discoveries, they only included descriptive findings that compared their sample to the normative means and abnormal classifications found in the AASP manual. As such, a carefully controlled comparison group is needed in order to clearly understand the extent to which sensory processing in parents of ASD children differs from parents of typically developing (TD) children.

In addition, Uljarević et al. (2014) did not control for factors previously associated with scores on self-report sensory processing questionnaires, such as gender and presence of mental disorders, and thus it is unclear if these factors could partially explain their results. Females are known to have a higher sensitivity than males in several modalities (for a review, see Velle 1987). Three recent studies using self-report questionnaires have further supported gender differences in sensory processing, finding that females reported significantly more sensory processing atypicalities than males (Engel-Yeger 2012; Horder et al. 2014; Tavassoli et al. 2014).

Sensory processing abnormalities have been associated with various mental disorders¹ and disorder symptoms, including ADHD (Mangeot et al. 2001; Ashburner et al. 2008), bipolar disorder (Brown et al. 2002), obsessive–compulsive disorder (OCD; Dar et al. 2012; Rieke and Anderson 2009), post-traumatic stress disorder (PTSD; Engel-Yeger et al. 2013), schizophrenia (Brown et al. 2002), anxiety (Engel-Yeger and Dunn 2011; Horder et al. 2014; Liss et al. 2005) and depression (Liss et al. 2005). In addition, relatives of ASD probands are known to have increased rates of several mental disorders that are associated with sensory symptoms, including OCD (Bolton et al. 1998), affective disorders (Bolton et al. 1998; Piven and Palmer 1999), bipolar disorders (DeLong and Nohna 1994), and schizophrenia (Daniels et al. 2008). Therefore, in order to disentangle influences specific to ASD, it is crucial to control for presence of mental disorders when investigating sensory processing in relatives of individuals with ASD.

The purpose of the present study was to examine how sensory processing differs in parents of children with ASD from SPX and MPX families compared to parents of TD children, in order to gain further insight into the heritable links between sensory symptoms and ASD. This study sought to expand Uljarević et al.’s (2014) initial findings by comparing parents of ASD children with a control group, including fathers in both groups, and examining confounding variables such as gender and presence of mental disorders. In addition, this study aimed to further differentiate the effects of genetic liability for ASD by separating the parents of ASD children into pre-defined SPX and MPX parent groups.

In the present study, the AASP was used to measure sensory processing in parents to allow for direct comparison of this study’s results with the two past studies exploring atypical sensory processing in relatives of individuals with ASD (De la Marche et al. 2012; Uljarević et al. 2014). Based on past research demonstrating that sensory processing abnormalities are heritable (Goldsmith et al. 2006) and related to ASD candidate genes (DeLorey et al. 2011; Peñagarikano et al. 2011; Tavassoli et al. 2012), it was hypothesized that parents of ASD children who presumably have the highest genetic liability for ASD (MPX) would differ from parents with lower ASD genetic liability (SPX) and parents with no such liability (parents of typically developing children; P-TD) in sensory processing scores on all four quadrants of the AASP.

Three groups of participants were recruited: parents from multiplex ASD families, parents from simplex ASD families, and parents who had a TD child with no biological ASD relatives. Participants were excluded from the study if they had their own diagnosis of ASD (n = 8), and one participant was excluded due to extreme outlier scores on two of the outcome measures (through use of the outlier labeling rule of Hoaglin and Iglewicz 1987), which suggested rushing or response bias.

Participants were included in the P-TD group if they were the biological parent of at least one typically developing child 4+ years old. Participants were excluded from the P-TD group if any of their biological children had a developmental disorder (n = 16) or if there were suspicions that their child might have ASD (n = 12). As this study focused on traits in ASD-affected families compared to TD families, participants in the P-TD group who had a biologically-related first-, second-, or third-degree family member with ASD were also excluded (n = 7), thus leaving 85 participants in the P-TD group.

Participants were included in the ASD-affected parent groups (MPX or SPX) if they reported they were the biological parent of at least one child 4+ years old who was diagnosed with ASD by a licensed professional (or a previous DSM classification name, such as Asperger’s syndrome, autism disorder, or pervasive developmental disorder-not otherwise specified). In order to include a wide range of severity, it was not made prerequisite that the ASD child was free of comorbid disorders or intellectual disability.

Participants from the ASD-affected group were designated to the SPX group if they (1) had only one biological child with ASD, (2) had at least one biological child without ASD, and (3) had no biological first-, second- or third-degree relatives with ASD. Therefore, singleton ASD families were excluded from this group (n = 48), which left 121 parents in the SPX group. While previous ASD studies included only families with two or more biological ASD siblings in their MPX group (e.g., Losh et al. 2008), we used a more liberal definition of MPX and, in addition to previous criteria, we also included parents who had one biological child with ASD and additionally had one or more first-, second-, or third-degree family members diagnosed with ASD who was biologically related to both the participant and to his or her child with ASD. Although the present study is the first to our knowledge to apply this expanded MPX definition to ASD families, various studies in other fields have used a similar definition of MPX, in which MPX families consisted of two or more first- or second-degree related family members (e.g., Blouin et al. 1998; Michel et al. 2001). As the purpose of the SPX/MPX separation is to represent the underlying genetic distinction between sporadic and familial cases of ASD, we chose this broader definition of MPX because the chances are extremely low that two or more cases of ASD in one biological family are both due to independently occurring rare de novo mutations that similarly resulted in ASD expression. Therefore, by broadening the definition to include any biological relative with ASD, not just a sibling, we maximize the chances of including all familial cases of ASD on the presumption that two cases in one biological family are most likely due to commonly inherited ASD traits. Nevertheless, in order to ensure that this method of distinction was valid, preliminary analyses were run to investigate whether those in the traditional definition of MPX (two or more ASD siblings in one nuclear family; n = 30) differed from those in the broader definition (one ASD child + one ASD family member biologically related to the ASD child and the parent; n = 24). No significant differences were found between these two groups on any outcome measures (.45 ≤ p ≤ .79), and therefore participants were confidently pooled together to form one MPX parent group (n = 54).

Table 1 shows the demographics of the P-TD (n = 85), SPX (n = 121), and MPX (n = 54) parent groups. Chi square tests of independence for each demographic variable showed that the three groups did not significantly differ in regard to gender, country of residence, education level, or amount of parents in each group currently diagnosed with a mental disorder (all ps > .05). The three groups differed in the amount of parents in the youngest age range (p < .05), with the P-TD group reporting more than expected in this youngest range and the MPX group reporting less than expected. However, as age was found to be unrelated to sensory processing scores in young- and middle-aged adult populations in three past studies all using different sensory questionnaires including the AASP (Crane et al. 2009; Robertson and Simmons 2013; Tavassoli et al. 2013), this significant difference in age between the P-TD and MPX groups was not problematic for the present study.

Participants were recruited through various methods of asking third party administrators to share the study’s website with potential participants. Recruitment for the majority of the two ASD-affected groups consisted of calling and/or emailing the head of 419 ASD centers, societies, parent support groups, and schools across the United States to request advertisement of the study’s website. Fifty-five groups agreed to participate through either: hanging the study’s flyer in their center; posting the study on their website, social media pages, or online discussion groups; or sending a mass email to all families involved with the organization. The study was also posted on several websites geared toward ASD research or increasing ASD awareness.

The AASP is one of the most widely used self-report questionnaires to evaluate sensory processing in adults (Brown and Dunn 2002). This questionnaire is based on Dunn’s (1997) model of sensory processing, which explains reactions to sensory input through a four-quadrant model. One dimension of this model is dedicated to neurological threshold (high threshold: hyposensitive; low threshold: hypersensitive), and the other to behavioral response/self-regulation (active or passive response to the given threshold). There is one subscale for each of the four quadrants, which consist of low registration (LR; high threshold/passive behavioral response), Sensation Seeking (S. Seeking; high threshold/active behavioral response), Sensory Sensitivity (S. Sensitivity; low threshold/passive behavioral response), and Sensation Avoidance (S. Avoidance; low threshold/active behavioral response).

The AASP has 60 items (15 items for each quadrant subscale), which are dispersed throughout six sensory modalities (Taste/Smell, Auditory, Visual, Tactile, Movement, and Activity). Participants respond on an increasing five-point scale how often they behave in the way described by the item (Almost Never, Seldom, Occasionally, Frequently, or Almost Always). Scores for each quadrant range from 15 to 75. An example item is: “I stay away from noisy settings.”

Standard errors of measurement of the AASP range from 3.58 to 4.51 (Brown and Dunn 2002), and internal consistency was found to be good for LR (α = .78), adequate for S. Seeking (α = .60), and good for S. Sensitivity (α = .78) and S. Avoidance (α = .77) (Brown et al. 2001). In the present sample, internal consistencies of the AASP quadrants were similar to Brown et al.’s (2001) results, with good reliability for LR (α = .78), S. Seeking (α = .70), S. Sensitivity (α = .82), and S. Avoidance (α = .84). Supporting concurrent validity of the AASP, scores in the low threshold quadrants (S. Sensitivity and S. Avoidance) were shown to have strong correlations with another self-report questionnaire of hypersensitivity/over-responsivity in adults, the Sensory Over-Responsivity (SensOR) Inventory: SensOR versus S. Sensitivity: r = .74; SensOR versus S. Avoidance: r = .64; both ps < .001 (Schoen et al. 2008).

Using a sample of 496 participants without disabilities aged 18–64 years old, Brown and Dunn (2002) have produced normative means with five classification groups of scores corresponding to how much an individual score differs from the normative mean for each quadrant. The five classification groups follow a normal distribution for each quadrant, for which a score below the 2nd percentile is considered “Much Less than Most People,” a score between the 2nd and 16th percentile is “Less than Most People,” between the 16th and 84th percentile is “Similar to Most People,” between the 84th and 98th percentile is “More than Most People,” and a score above the 98^th percentile is “Much More than Most People.”

The demographics questionnaire inquired about general characteristics of the participant, such as gender, age group, and education level, but also characteristics specific to this study, such as number of biological children and family history of ASD. As it was not possible to conduct clinical diagnostic methods due to the online format of this study, participants were asked to personally report if they had ever been diagnosed with any mental disorder, and if so, which disorder(s). They were also asked which disorders, if any, they were currently diagnosed with.

In order to compare our results with past studies and the AASP normative means, we first looked at the amount of participants who scored in the extreme outer percentiles of the AASP normative distribution, which referred to scores that were below the 2nd percentile (AASP classification: “Much Less than Most People”) or above the 98th percentile (AASP classification: “Much More than Most People”) for each quadrant. Comparing percentages of extreme scores to those found in the AASP normative sample is common in most studies using the AASP (e.g., Rieke and Anderson 2009; Uljarević et al. 2014), and is a good complementary method in addition to statistically comparing group means.

To statistically evaluate if sensory processing differed between SPX and MPX parents of ASD children and parents of TD children, we used a one-way multivariate analysis of covariance (MANCOVA) with diagnostic group (P-TD/SPX/MPX) as fixed factor, gender and presence of mental disorders as covariates, and the four AASP quadrant scores as dependent variables. Univariate analyses of covariance (ANCOVAs) were then performed for each quadrant, followed by post hoc analyses using Bonferroni’s correction method for multiple comparisons to determine group differences. Atypical sensory processing was defined as significantly differing from the “typical” group (i.e., P-TD) in any quadrant score.

Figure 1 shows the percentage of each group that scored in the extreme outer percentiles in the four quadrants of the AASP. These percentages are compared to those found in the AASP normative sample (i.e., by definition, 4 % of the normative sample scored in the outer two percentiles in each quadrant). In all four quadrants, there was a visible linear trend across families (P-TD < SPX < MPX) in the amount of extreme scores.

After assessing the total amount of extreme sensory scores in each group, it was found that 53 % of MPX parents scored in the outer two percentiles in at least one AASP quadrant (28 % had extreme scores in only one quadrant, 15 % in two quadrants, 6 % in three quadrants, and 4 % in all four quadrants). In comparison, 27 % of SPX parents scored in the outer two percentiles in at least one quadrant (10 % in one quadrant, 12 % in two quadrants, and 5 % in three quadrants), and 20 % of P-TD parents scored in this range in at least one quadrant (14 % in one quadrant, 2 % in two quadrants, 4 % in three quadrants).

The preliminary t tests revealed that females scored significantly higher than males in two of the four AASP quadrants: S. Seeking, t(258) = 3.221, p = .001, and S. Sensitivity, t(258) = 3.382, p < .001; while S. Avoidance showed a trend: t(258) = 1.777, p = .07. In addition, parents currently diagnosed with a mental disorder scored significantly higher than parents without such diagnoses in three quadrants: LR, t(258) = 3.688, p < .0001; S. Sensitivity, t(258) = 4.185, p < .0001; and S. Avoidance t(258) = 4.129, p < .0001. Therefore, both gender and presence of a mental disorder were used as covariates in subsequent analyses.

All dependent measures for the MANCOVA and the modality-specific ANCOVAs met assumptions for homogeneity (p > .05 for all Levene’s homogeneity of variance tests), thus permitting parametric analyses. Raw scores and standard deviations of each group are presented in Table 2.

The MANCOVA revealed significant differences in AASP scores among the P-TD, SPX, and MPX parent groups, F(8504) = 2.107, p = .034, Wilks’ Lambda = .936; η _p ²  = .32. Subsequent ANCOVAs found that scores significantly differed among parent groups in all four sensory quadrants: LR, F(2255) = 3.796, p = .024, η _p ²  = .029; S. Seeking, F(2255) = 3.246, p = .041, η _p ²  = .025; S. Sensitivity, F(2255) = 5.649, p = .004, η _p ²  = .042; and S. Avoidance, F(2255) = 5.825, p = .003, η _p ²  = .044. Figure 2 illustrates group differences after correcting for multiple comparisons using Bonferroni’s method.

As shown in Table 3, results of the modality-specific ANCOVAs and Bonferroni corrected post hoc analyses revealed that MPX parents scored significantly higher than both SPX and P-TD parents in the auditory total score, auditory S. Sensitivity, auditory S. Avoidance, visual total score, visual LR, and visual S. Avoidance (all ps < .005), and MPX parents also scored higher than P-TD parents in auditory LR, activity LR, touch S. Sensitivity, and visual S. Sensitivity (all ps < .05). In addition, it was found that SPX parents scored higher than P-TD in auditory LR (p < .01), while P-TD parents scored higher than SPX in the movement total score (p < .05).