Sensorimotor Behavior in Individuals with Autism Spectrum Disorder and Their Unaffected Biological Parents

Background: Sensorimotor impairments are common in autism spectrum disorder (ASD) and evident in unaffected first-degree relatives, suggesting that they may serve as important endophenotypes associated with inherited risk. We tested the familiality of sensorimotor impairments in ASD across multiple motor behaviors and effector systems and in relation to parental broader autism phenotypic (BAP) characteristics. Methods: Fifty-eight autistic individuals (probands), 109 parents, and 89 control participants completed tests of manual motor and oculomotor control. Sensorimotor tests varied in their involvement of rapid, feedforward control and sustained, sensory feedback control processes. Subgroup analyses compared families with at least one parent showing BAP traits (BAP+) and those in which neither parent showed BAP traits (BAP-). Results: Probands with BAP- parents (BAP- probands) showed rapid manual motor and oculomotor deficits, while BAP+ probands showed sustained motor impairments compared to controls. BAP- parents showed impaired rapid oculomotor and sustained manual motor abilities relative to BAP+ parents and controls. Atypical rapid oculomotor impairments also were familial. Limitations: Larger samples of ASD families including greater samples of probands with BAP+ parents are needed. Genetic studies also are needed to link sensorimotor endophenotype findings directly to genes. Conclusions: Results indicate rapid sensorimotor behaviors are selectively impacted in BAP- probands and their parents and may reflect familial liabilities for ASD that are independent of familial autistic traits. Sustained sensorimotor behaviors were affected in BAP+ probands and BAP- parents re ecting familial traits that may only confer risk when combined with parental autistic trait liabilities. These findings provide new evidence that rapid and sustained sensorimotor alterations represent strong but separate familial pathways of ASD risk that demonstrate unique interactions with mechanisms related to parental autistic traits.


Introduction
Autism spectrum disorder (ASD) is etiologically diverse. Highly penetrant but rare gene variants have been implicated in up to 20% of cases [1,2], and several environmental factors also confer risk [3]. Still, ASD is highly heritable, and a plurality of cases involve multiple common allelic variants that individually contribute only incremental risk [4][5][6]. These common variants are di cult to detect given their small individual effects and variability across affected individuals. Understanding of pathogenic processes associated with majority of cases thus remains limited but could be advanced by identifying downstream traits more strongly associated with causal mechanisms than diagnostic status.
Establishing endophenotypes, or quantitative trait dimensions that are not directly observable, associated with affectation status, and related to genetic risk for a condition, can offer a powerful method for clarifying pathogenicity [7,8]. Few endophenotypes have been identi ed in ASD, though several trait dimensions have shown promise for assessing familial risk. For example, "broader autism phenotypic" (BAP) traits, including social aloofness, communication di culties, and rigid personality traits, are more prevalent in unaffected family members of autistic individuals relative to the non-autistic population [9][10][11][12]. Cognitive processing de cits, including reduced behavioral exibility and inhibitory control, also are associated with core autism traits and disrupted in unaffected parents compared to age-matched controls [13]. Importantly, behavioral in exibility is more severe in parents with BAP features and their autistic children relative to parents without BAP features, suggesting that these neurocognitive phenotypes and BAP traits overlap in families to confer risk for ASD [13]. The relationships between separate candidate endophenotypes associated with ASD seldom have been studied but may be critical to clarifying mechanisms of heritable risk for ASD.
Our group previously demonstrated, in a separate sample, that rapid sensorimotor behaviors supported by feedforward planning and sustained behaviors guided by sensory feedback each are impaired in autism across manual motor and oculomotor systems [17,[20][21][22]. We also documented similar oculomotor differences in unaffected rst-degree relatives of autistic individuals [19]. These ndings suggest that rapid and sustained sensorimotor behaviors may represent endophenotypes associated with familial risk for ASD. In the present study, we aimed to replicate and extend these ndings in three key ways. First, we examined sensorimotor behaviors in a large sample of autistic families across manual motor and oculomotor systems. Second, to determine the extent to which familial sensorimotor and BAP traits interact to confer risk for ASD, we compared sensorimotor behaviors in families with and without parental BAP traits. Third, we tested rapid and sustained sensorimotor behaviors in family trios allowing us to assess inter-relationships across autistic individuals and their biological parents.

Methods and Materials
Participants Fifty-eight autistic participants (probands; aged 5-22 years), 109 of their unaffected biological parents (proband parents; aged 29-54 years), and 89 control participants were examined (Table 1). This sample included 43 family trios (proband, biological mother and father), 15 family dyads (i.e., proband and one parent), three probands whose parents did not complete testing, and three proband parents whose child did not complete testing. Forty-two typically developing (TD) controls were matched on age, handedness, and nonverbal IQ to the probands, and 47 neurotypical adults were matched with proband parents on age, sex, handedness, and nonverbal IQ (i.e., parent controls). Due to scheduling, technical, or compliance issues, multiple individuals did not complete all sensorimotor tests (see Tables S1 and S2 for details). IQ was estimated using the Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition (4 probands, 8 TD controls; 23), Differential Ability Scale, Second Edition (3 probands; 24), or Wechsler Abbreviated Scale of Intelligence, Second Edition (51 probands, 34 TD controls, 108 proband parents, and 44 parent controls; 25). Three parent controls and one proband parent did not complete IQ testing due to scheduling di culties or limited English pro ciency. Note. Data are reported as mean and standard deviation in parentheses. a Biological sex was compared across groups using a chi-square test. Biological sex is not reported for one parent control. PIQ = performance IQ; VIQ = verbal IQ. ***p < 0.001, **p < 0.01.
ASD diagnoses were con rmed using the Autism Diagnostic Inventory-Revised (ADI-R; 26), the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2; 27), and expert clinical opinion based on DSM-5 criteria [28]. Autistic participants were excluded if they had any known genetic condition associated with ASD (e.g., Fragile X Syndrome, etc.). Control participants had a score ≤ 8 on the Social Communication Questionnaire (SCQ; 29) and no known history of psychiatric or neurological disorders, rst-or second-degree relatives with ASD, or rst-degree relatives with a major psychiatric disorder. Two proband parents scored higher on the SCQ (≥ 15); one completed an ADOS-2 and did not meet classi cation criteria for ASD, and one did not complete the ADOS-2 due to scheduling di culties. There were no prior concerns about an ASD diagnosis, so they are included in nal analyses. No participants were taking any medications known to affect sensorimotor function at the time of testing (e.g., stimulants, antipsychotics, anticonvulsants, benzodiazepines; 30) or had a history of head injury, birth injury, or seizure disorder. Adult participants provided written consent. For minors, a legal guardian's written consent and the minor's written assent were obtained. The study was conducted according to procedures approved by the University of Illinois at Chicago and the University of Texas Southwestern Medical Center Institutional Review Boards.

Procedures Precision Grip Tasks
During precision grip testing, participants were seated in a darkened room 52 cm from a 102 cm Samsung LCD monitor (resolution: 1366 x 768; refresh rate: 120 Hz). Participants used their thumb and index nger to press against two opposing precision load cells (ELFF-B4-100N; Entran). A Colbourn (V72-25) resistive bridge strain ampli ed load cell analog signals. Data were sampled at 120 Hz with a 16-bit analog-to-digital converter (DI-720; DATAQ Instruments) and converted to Newtons.
Before testing, participants' maximum voluntary contraction (MVC) was calculated for each hand using the average of their maximum force output across three, three second trials. During testing, participants viewed a white horizontal force bar that moved upwards with increased force and a static target bar (i.e., red during rest and green to cue the start of the trial; Fig. 1). Participants were instructed to press the load cells when the red target bar turned green and continue pressing until the bar turned red. Participants completed "rapid" (2 seconds) and "sustained" (8 seconds) trials at 15%, 45%, and 85% of their MVC based on evidence that relative group differences vary as a function of force load [17]. The rapid test included two blocks of ve trials for each hand at each force level (60 trials). The sustained test included two blocks of three trials for each hand at each force level (36 trials).
Force output for each trial was low-pass ltered via a double-pass 4th -order Butterworth lter with a cutoff of 15 Hz. Data were analyzed using custom MATLAB scripts [22]. During the rapid test, we examined the accuracy of participants' initial force output, de ned as the force at the rise phase offset divided by the target force. The rise phase offset was the timepoint when the rate of force increase rst fell below 5% of the peak rate of force increase, and the force level was within 90-110% of the mean force of the sustained phase [22].
During the sustained test, we measured force output for six seconds after the rise phase offset. Force variability was examined using coe cient of variation (CoV) calculated as the standard deviation of the force time series divided by participants' mean force. Trials in which participants stopped pressing for > 1 second or completed < 4 seconds of sustained force were excluded.

Oculomotor Tasks
Participants completed oculomotor testing in a darkened black room with a chinrest positioned 63 cm from a 102 cm anti-glare LCD monitor (resolution: 1920 x 1080; refresh rate: 60 Hz). Participants' eye movements were recorded using an infrared, binocular camera-based eye tracking system with a 500 Hz sampling rate and a gaze-position error of < 0.5 degrees of visual angle (EyeLink II, SR Research Ltd., Canada).
Participants completed visually guided saccade (VGS) and smooth pursuit tasks (Fig. 2). During VGS, participants xated on a central crosshair, then looked toward peripheral targets at ± 12 or 24 degrees of visual angle given prior ndings that de cits vary across saccade amplitudes [20]. Twelve-and 24-degree targets were presented in two blocks of 30 trials. During the smooth pursuit task, participants xated on a central target and then tracked the target from 0 to ± 15 degrees moving at 2.5, 7.4, 14.9, 22.2, or 30 degrees/second. Participants completed 40 interleaved trials (4 trials per direction per velocity). A vepoint calibration was completed prior to oculomotor testing and a 3-point drift correction was completed before each subsequent block of trials.
Digital nite impulse response lters with non-linear transition bands were applied to the eye movement data prior to scoring. For the VGS test, saccade onset and offset were identi ed as the timepoints when eye velocity exceeded and fell below 30 degrees/second, respectively. The accuracy, peak velocity, and duration of the primary saccade (i.e., the rst saccade that moved at least 20% of the distance to the target) were examined. Accuracy was the absolute value of the horizontal distance in degrees of visual angle between the nal saccade location and the target location [20]. Saccades with latencies ≤ 70 ms were excluded from analyses. To control for differences in saccade amplitude, we divided peak velocity and duration by saccade amplitude. For the smooth pursuit task, pursuit onset was de ned as the timepoint when eye gaze exceeded 2 degrees/second for 20 ms. Non-pursuit eye movements (e.g., saccades, blinks) and artifacts (e.g., large head movements) were removed. The ratio of pursuit velocity to the target velocity (i.e., pursuit gain) was examined.

Clinical Measures
Symptoms related to ASD were assessed using the ADOS-2 calibrated severity scale (CSS; 31,32) and ADI-R algorithm scores. The Repetitive Behavior Scale-Revised (RBS-R; 33) was used to measure repetitive behaviors. The Conners Parent Rating Scale (Conners-3; 34) was used to measure inattention and hyperactivity/impulsivity. Subclinical ASD features were assessed in parents using the Broad Autism Phenotype-Questionnaire (BAP-Q; 101 self-report, 3 spouse report; 35). Parents were identi ed as "BAP+" if they exceeded the threshold for any BAP-Q subscale based on published cutoffs (9; see Table 2 for details). Probands with at least one BAP + parent were identi ed as "BAP+" (N = 22; including four probands with two BAP + parents, 17 probands with one BAP + and one BAP-parent, and one proband with one BAP + parent and one parent who did not complete the BAP-Q). Probands with two BAP-parents were categorized as "BAP-" (N = 23). Probands with missing BAP-Q data from at least one parent were excluded from BAP+/-subgroup analyses (N = 11) unless their parent completing the assessment met BAP + criteria.  [17,37], age (converted to z-scores) was included in the MLMs for proband but not parent analyses. Sex was included in parent but not proband analyses due to the small number of female probands. All dichotomous variables (e.g., hand, group) were centered using contrast coding. Random variance components for the intercept for force level and all two-and three-way interactions were included. Grip force variability (CoV) was transformed using natural log transformations due to a non-normal distribution. We compared nested models using likelihood ratio tests to determine whether each predictor improved model t. Variables that did not improve model t (p > 0.05) were removed. Models with signi cant interaction effects included all lower-level interactions and main effects of variables included in the interaction in the nal models [38].
MLMs for oculomotor variables were identical to those for precision gripping variables except level 1 predictors included target direction (left vs right), target amplitude (12 vs 24 degrees), and target velocity (for smooth pursuit gain only) as well as a random slope for target velocity in the smooth pursuit MLM. Target velocity was linearly transformed with 2.5 degrees/second set to zero. To compare performance across BAP groups, similar MLMs were conducted with group de ned as a factor including BAP + and BAP-probands or BAP + and BAP-parents, with each control group as the reference group. Cohen's d values were calculated for each signi cant group effect or group-related interaction.
Associations between precision gripping, oculomotor behaviors, and autistic traits were examined using Spearman correlations separately for probands and proband parents. Only correlations with |r| >0.50 and p < 0.01 are reported. We examined differences in autistic traits between BAP + and BAP-probands using t-tests. The familiality of sensorimotor behaviors within family trios and duos (i.e., proband and one proband parent) was determined using the Sequential Oligogenic Linkage Analysis Routines (SOLAR; Southwest Foundation for Biomedical Research; 39) as done previously [13].
Female parent controls showed greater force overshoot than proband mothers, though this was a small effect (d = 0.117). Proband fathers and male controls showed similar performance (group x sex: β=-0.020, SE = 0.008, p = 0.008; Figure S2). No differences between BAP + or BAP-parents and controls were seen for rapid force accuracy.

Relationship Between Manual Motor and Oculomotor Behavior
For TD controls, increased rapid force overshooting at 15% MVC was related to increased saccade velocity (r = 0.579, p = 0.002; Table S3). For autistic individuals, increased pursuit gain was related to reduced grip force variability (r=-0.541, p = 0.002) and increased rapid grip force accuracy at 85% MVC (r = 0.522, p = 0.003). No other relationships between precision gripping and oculomotor outcomes were signi cant.

Discussion
We replicated our previous ndings of feedforward and feedback sensorimotor impairments in autistic individuals [17,[20][21][22] and found that probands and their unaffected biological parents show similar patterns of sensorimotor control di culties relative to neurotypical individuals which extends prior family studies [13,19] in four key ways. First, we show that sensorimotor behaviors are impacted in unaffected parents across oculomotor and skeletomotor systems for rapid and sustained sensorimotor behaviors. Second, we found that sensorimotor di culties were more severe in BAP-parents suggesting that sensorimotor impairments and BAP traits are familial characteristics that each confer unique risk for ASD. Third, rapid sensorimotor behaviors were more severely disrupted in BAP-probands across effectors while sustained behaviors were selectively impacted in BAP + probands. Finally, atypical saccade dynamics were highly familial, particularly between BAP-parents and their children, indicating that they may represent inherited liabilities in a subgroup of affected individuals without parental BAP traits. These results provide new evidence for multiple sensorimotor endophenotypes in ASD and highlight the complex interactions with liabilities re ected by parental autistic traits.

Rapid and Sustained Sensorimotor Behavior in Probands and Parents
Probands and their biological parents showed similar impairments in rapid sensorimotor behaviors relative to controls implicating di culties utilizing internal action representations to execute precise motor behaviors before sensory feedback information is available. This feedforward motor control dysfunction is more severely impacted in ASD families without parental BAP features. These novel ndings suggest that rapid sensorimotor di culties may represent inherited pathogenic risk processes that are independent from the risk accounted for by the presence of parental BAP traits [40,41].
The familiality of these traits is re ected in our ndings that saccade velocity and duration were highly inter-correlated among probands and their biological parents. Importantly, saccade dynamics were more strongly familial among BAP-parents and their children than in BAP + parents and their children, suggesting that they may be useful for separating the in uence of different risk mechanisms for ASD. These results are consistent with research suggesting that inherited liability for ASD can be separated into independent components, including motor behaviors, attention, and social di culties [18, 42,43], and implicate rapid sensorimotor impairments as a new, distinct endophenotype useful for indexing familial liability with distinct ASD risk genes.
Our sustained sensorimotor outcomes indicated that utilizing sensory feedback for optimizing motor function is selectively impaired in BAP + probands and BAP-parents, suggesting overlapping or interacting risk with parental autistic traits. Further, BAP + probands showed more severe hyperactive/impulsive symptoms and repetitive behaviors compared to BAP-probands implicating a common inherited mechanism underlying clinical symptoms and impairments during sustained motor tasks which require higher attentional control compared to rapid motor tasks. Alternatively, as the majority of BAP + probands in our sample (77%) had one BAP + parent and one BAP-parent, sustained motor impairments in BAP + probands may re ect a combination of additive risk for ASD including familial autistic traits and sensorimotor feedback impairment, inherited separately from each parent. Studies assessing additive risk for ASD in offspring as a function of parental traits are needed to determine how sustained sensorimotor and autistic trait endophenotypes may interact to confer separable or joint risk.

Limitations
While this study provides new insights into multiple familial endophenotypes associated with inherited risk for ASD, several limitations should be noted. First, many of our conclusions are drawn from relatively small samples of autistic individuals and their families. We emphasize the need for replication of these ndings with larger samples of ASD families, including more autistic females, to address critical questions regarding independent or interactive familial risk factors conferred by separate sensorimotor and other endophenotypes. This also includes larger family trio samples with a greater number of probands with BAP + parents. Further, while a replication sample was not explicitly included in this study, these results extend previous ndings from our group using separate samples examining manual motor [22,44] and oculomotor control [20,21]. Additionally, while family studies can offer evidence supporting inherited liabilities, genetic studies are needed to link endophenotype ndings directly to genes. Finally, further examination of sensorimotor impairments as endophenotypes in early childhood may support earlier identi cation of ASD risk and targeted early intervention for motor impairments.

Conclusions
Our ndings identify rapid and sustained sensorimotor behaviors as discrete processes that are impaired in probands and their unaffected parents, affect multiple effector systems, and differentially relate to parental BAP traits. Findings that rapid motor di culties are selectively impacted in BAP-probands and parents and are familial suggest that they may confer unique risk of ASD, independent of autistic trait inheritance, while sustained sensorimotor behaviors may jointly confer risk with BAP traits. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests
Dr. Mosconi is leading an investigator-initiated study of behavioral in exibility in ASD funded by Acadia.  Figure 1 Precision gripping tasks. A. Precision gripping stimuli. Participants viewed a red target bar and white force bar. When the red target bar turned green, participants were instructed to press the load cells as quickly as possible so that the white force bar reached the level of the green target bar and continue pressing so that the white force bar maintained the level of the green target bar. B. Load cell apparatus.  Grip force accuracy during rapid precision gripping for probands and TD controls across force levels as a function of age and broader autism phenotype (BAP) classi cation. During the rapid grip task, Accuracy of visually guided saccades across target step amplitudes for probands (A) and proband parents (B) as a function of broader autism phenotype (BAP) classi cation. BAP+ (d=0.278) and BAP-   Duration (after controlling for saccade amplitude) of visually guided saccades across target step amplitudes for probands (A) and proband parents (B) as a function of broader autism phenotype (BAP) Force variability for proband parents and parent controls across force levels as a function of broader autism phenotype (BAP) classi cation and sex. BAP-fathers showed higher sustained force variability relative to BAP+ fathers (45% MVC: d=0.452; 85% MVC: d=0.380) and parent control males (45% MVC: d= 0.264; 85% MVC: d=0.356) at 45% and 85% MVC (A) and BAP-mothers showed higher sustained force variability relative to parent control females (d=0.297) and BAP+ mothers (d=0.280) at 15% MVC [B; group (BAP-vs Control) x MVC x sex: β=-0.028, SE=0.010, p=0.004]. *p<0.05; **p < 0.01; ***p < 0.001. Figure 9