Although repetitive behavior is a cardinal feature of the autism phenotype, it is not a unitary construct, nor is its occurrence unique to autism. There is substantial variability in the form, function, and biological basis of behaviors constituting the repetitive behavior domain. Having previously found that children with FXS+Aut differ from children with iAut in both patterns of repetitive behavior  and caudate volumes [10, 12, 22], we hypothesized that brain-behavior associations would likewise differentiate idiopathic autism from autism associated with an etiologically defined disorder, that is FXS. For boys with both FXS and iAut, we found that overall severity and number of topographies of SIB were positively associated with CN volumes. This pattern was particularly strong among the subgroup of boys with FXS+Aut, who, by definition, are characterized by higher rates of repetitive motor behavior when compared with boys with FXS who do not meet criteria for autism.
Unique to boys with iAut, we found that compulsive and ritual behaviors were significantly correlated with CN volumes consistent with similar studies of older children and adults with the disorder [16, 17, 19]. While ritual behavior was not significantly associated with caudate volumes in FXS or the FXS+Aut subgroup, these relationships were near the level of significance, suggesting the possibility of a shared mechanism for this specific type of behavior. Interestingly, the same was not true for compulsive behavior, despite presumed overlap in form and function. In general, associations between CN and forms of repetitive behavior were congruent between all boys with FXS regardless of an autism diagnosis, but were in part dissociable from patterns observed in boys with iAut (Figure 1). To ensure specificity of our primary analyses, we also examined the relationship between amygdala volumes and repetitive behaviors. We found no significant correlations between left and right amygdala volumes and any measure of repetitive behavior for boys with FXS, FXS+Aut, or iAut.
Our hypothesis that repetitive motor behaviors would be linked to CN volume in FXS was confirmed with regard to SIB but not stereotypical motor behavior. Although rates of motor stereotypy are significantly elevated in FXS [23, 33], we found no relationship between the production of this behavior and CN volume in our sample. The specificity of the self-injury finding lends support to a taxometric rather than a unitary approach to disentangling brain-behavior relationships. That is, while stereotypy and SIB may share common neurobehavioral features, these behaviors might not merely be dimensional expressions of the same phenomenon. This argument may extend to distinctions among neural mechanisms underlying similar behavior across genetic disorders. For example, though SIB is common to both FX and Lesch-Nyhan syndromes, the putative neural mechanisms underlying the emergence of functionally similar behaviors might qualitatively differ between these syndromes, as evidenced by inverse patterns of caudate pathology [9–11, 20, 21]. In this study, SIB was associated with CN volume in boys with iAut in similar fashion to boys with FXS, suggesting some degree of shared pathophysiology. For young children with either FXS or iAut, a larger caudate may indicate an increased risk of developing SIB. This correlation was modest among boys with iAut, however, and statistically significant only for the left CN. The stronger link evident among boys with FXS may be owed to etiological homogeneity. A larger sample of children with iAut might afford the opportunity to define subgroups based on specific neurobehavioral features.
The neural phenotype of FXS, and to a lesser extent iAut, includes significant caudate enlargement irrespective of self-injurious or repetitive behavior [10–12, 20–22], In FXS, this morphological feature probably stems from early overgrowth followed by dampened dendritic elimination . Understanding the mechanisms and course of early caudate development in FXS may inform the pathogenesis of SIB. For example, a foundational aberration in caudate structure may confer risk for the development of SIB through altered function or atypical response to feedback . It may be telling that the average age of onset for SIB in children with FXS  coincides with the peak and gradual decline of synaptic density in typically developing children . While we can only speculate based on the present findings, it is further possible that both initial and subsequent risk for SIB associated with FXS is tightly linked to atypical developmental processes associated with FMRP (fragile X mental retardation protein), such as reduced synaptic plasticity generally or dysregulated activity involving striatal circuitry specifically [14, 34, 35]. In this developmental framework, the emergence of SIB may be explained by two possible phenomena. First, SIB risk may be concomitant with extent of striatal malformation stemming from atypical developmental elimination of dendritic spines, particularly in the CN. Second, further increases in striatal volume may result from an additive process associated with altered function stemming from early overgrowth and reciprocally tied to the behavioral performance of SIB.
With regard to this second possibility, findings from preclinical models suggest that significant caudate enlargement can result from chronic dopamine accumulation secondary to D2 receptor antagonism . Persistent D2 receptor antagonism in the dorsal striatum, which includes the caudate, has been associated with chronically elevated nociception (pain sensitivity) , a phenomenon that has been observed in persons with SIB using a variety of biobehavioral approaches [39, 40]. Similarly, atypical dopaminergic function in the dorsal striatum has been associated with significant elevations in basal stress response , evidence of which is seen in numerous studies of human and nonhuman primate studies of SIB [42–44]. Although altered dopaminergic function has been tied to both repetitive self-injury and FXS [7, 8, 13, 14] it remains but one of a number of possible striatal mechanisms underlying motor dysregulation . Volumetric and correlational findings alone cannot speak directly to these issues, and a multimodal, developmental approach involving clinical and preclinical models is necessary to elucidate the complex neurobehavioral processes underlying SIB associated with FXS and iAut .
There are several limitations to the present findings. Given our sample size, we were unable to break out subdomains of repetitive behavior by specific form. While the total number of topographies was strongly associated with CN volume, a larger sample is required, to understand whether discrete forms of repetitive and SIBs (for example, self-biting) are differentially linked to brain measures. This study did not account for the influence of environmental variables on behavior, and thus the relationship between brain, behavior, and environment is unknown. Because SIB both shapes, and is shaped by, the environment in which it occurs [46, 47], properties such as behavioral function could further inform the brain-behavior associations reported here. Similarly, it would be illuminating to measure to what extent, if any, associated features such as arousal or anxiety moderate the relationship between subtypes of repetitive behavior and brain structure or function . Overall caudate volume is a rather blunt measure of brain anatomy, and the extent to which it informs underlying pathology is limited. It is plausible, for instance, that more nuanced qualities of caudate morphology, beyond overall volume, drive associations with repetitive behavior.
Potential next steps include leveraging multimodal imaging to further examine the striatal neurobiology of repetitive behaviors associated with FXS and iAut. Fruitful approaches might include alternative measures of morphometry, such as shape or surface area, as well as diffusion tensor MRI, which offers the potential to measure structural properties of cortico-striato-thalamo-cortical connectivity. Given that the relationship between the striatum and repetitive behavior is a dynamic one, longitudinal data would provide the opportunity to chart the developmental course of reciprocal brain-behavior change. Such data would likewise allow investigators to explore the utility of striatal neuroimaging measures as a clinically relevant risk marker for SIB. The finding that caudate volume is linked to SIB in both FXS and iAut may provide a common referent from which clinical treatment studies might build. Such work could more closely examine underlying mechanisms with an eye toward well-defined targets, bearing in mind the need to carefully measure discrete classes of behavior rather than blunt outcomes, for example, ‘stereotypy’ and ‘self-injury’ versus ‘repetitive behavior’ or ‘irritability’. Recently, there is exciting promise that an mGlu5 inhibitor could target repetitive behaviors generally and self-injury specifically given demonstrable effects on analog behaviors and dendritic architecture in preclinical studies of FXS and autism [48, 49]. In human beings, this promising treatment approach could be used in tandem with behavior analytic strategies aimed at bolstering adaptive replacements for SIB . Finally, replication with larger samples or among individuals with other genetic neurodevelopmental disorders associated with SIB is necessary to confirm and possibly extend the present findings.