Here, we present examinations of putative genotype–phenotype associations between three SNPs (5-HTTLPR, MAOA and COMT), selected a priori, and a range of clinically relevant behaviours in males with FXS to identify additional genetic risk for these behavioural characteristics. Males with FXS are at high risk of displaying each of the behavioural characteristics investigated in the current study, namely, challenging behaviour (defined here as physical aggression, property destruction, stereotyped behaviour, and self-injury), autism symptomatology, ADHD characteristics, repetitive behaviour and low mood. However, the frequency and severity of each of these behavioural characteristics is variable within the FXS population. Our results indicate that common genetic variation in the dopamine-related COMT genotype contributes meaningfully to clinical variability in challenging and repetitive behaviours and interest and pleasure characteristics in this population. Specifically, compared to the AG and GG genotypes, the AA genotype was associated with reduced risk for property destruction, stereotyped and compulsive behaviour, and with greater interest and pleasure in the environment. No association was observed between behavioural characteristics and either serotonin-related (5-HTTLPR) or monoamine oxidase related (MAOA) genotypes. The results of the current study contribute to the emerging field on personalised treatments and, if replicated, highlight that variation in COMT may inform tailored interventions. Understanding the genetic basis for variable behavioural expression across people with FXS has implications for early detection of the heightened risk for these and other clinically significant behaviours, which can ultimately optimise outcomes for these individuals through biologically informed and preventative interventions.
The lack of association between the 5-HTTLPR genotype and challenging behaviour is inconsistent with the one existing study investigating this relationship in males with FXS . This previous study reported that individuals with the L/L genotype displayed significantly higher levels of aggressive and destructive behaviour than those with the S/S genotype, and higher levels of stereotypic behaviour than those with the S/L genotype. A potential reason for this discrepancy is differences in population characteristics. Specifically, 75 and 98% of the sample reported in Hessl et al.  demonstrated aggressive and stereotyped behaviour, respectively, within a two-month period. In contrast, only 40 and 60% of the sample included in the current study demonstrated aggressive and stereotyped behaviour, respectively, within the previous month. Thus, the overall severity of challenging behaviour was substantially higher in the study sample reported by Hessl et al.  than in the current sample. In addition, the measures used to capture information on property destruction, self-injurious and aggressive behaviour differed between the two studies. The study by Hessl et al.  used a five-point frequency scale and a four-point severity scale to assess self-injurious behaviour, stereotyped behaviour, and aggression/destruction over the past two months. The current study focussed on the presence of aggressive behaviour and property destruction over the past one month, and the presence and severity of self-injurious behaviour with the latter capturing length of episode, restraint, and frequency on a five-point scale. In the present study, stereotyped behaviour was captured using two instruments, one assessing presence and another assessing frequency on a five-point scale. All these measurement differences may have also contributed to the discrepant findings.
The association between the L/L genotype and aggressive behaviour has also been reported in individuals with ASD , but only when using a single parent-interview item and not when this same behaviour was measured via observation. The same relationship has also been reported in individuals with intellectual disability who display aggressive behaviour . Again, the severity of aggressive behaviour may be higher in individuals living in residential homes than with their parents or primary caregivers, as was the case with most participants in the current study. Therefore, one possible explanation for the discrepancy between the findings reported in the current study and those of Hessl et al. is an interaction among FXS, 5-HTTLPR genotypes and environments, which increase overall rates of challenging behaviour in individuals with FXS. Future studies should, therefore, include measures of participant and environmental factors related to risks for challenging behaviour. In addition, challenging behaviour encapsulates a broad range of behaviours and, therefore, specificity of the construct being measured is important in future studies of genotype–phenotype associations [see 20].
The results reported here do not support existing research conducted in the other populations indicating a relationship between 5-HTTLPR and other behavioural characteristics including ADHD [21,22,23], ASD symptom severity  and depression [24, 25]. Many of these studies were conducted in the general population, which highlights the importance of conducting genotype–phenotype studies in unique, well-defined populations, such as FXS.
The current study revealed no significant differences in behavioural characteristics between individuals with three versus four MAOA repeats. This is consistent with previous findings by Hessl et al. . However, limited statistical power may account for the results in both the existing and current study. Research has generally reported mixed results for MAOA genotypes and aggressive behaviour within the general population, particularly when a direct association between genotype and phenotype is investigated, as effects of MAOA appear more likely to be mediated by environmental interactions [26, 27]. Contradictory results have also been reported in neurodevelopmental disorders with one study reporting a twofold higher risk of autism in individuals with four versus three MAOA repeats , and another study reporting that the three repeat allele was associated with increased severity of autism . Individuals with FXS show an atypical profile of ASD-related impairments, which has previously accounted for subtle differences in FXS and idiopathic ASD populations , and may explain the lack of association reported here.
The results reported here revealed that the AA genotype of COMT is associated with greater interest and pleasure, and with reduced risk for property destruction, compulsive behaviour and stereotyped behaviour. To our knowledge, this is the first study to explore the relationship between genotypes of this dopamine-related SNP and behavioural characteristics in individuals with FXS, and indicates that variation in genetically-mediated dopamine levels in the brain may go some way toward explaining the variability in the presence and severity of a number of behavioural characteristics. Interestingly, the AA genotype has been associated with increased depression in the general population [30, 31], suggesting that the mechanisms contributing to this genotype–phenotype relationship may be different in those with and without FXS. Mixed results have been reported in the general population with regard to other behavioural characteristics and, therefore, the current study expands on this by highlighting a relationship between the AA genotype and compulsive and stereotyped behaviour in individuals with a well-defined genetic syndrome. The AA genotype results in higher dopamine in the prefrontal cortex. Importantly, the effect of these increases in dopaminergic function may be dependent on the level of pre-existing dopamine in the system . For example, increased levels of dopamine, or larger doses of dopamine administered as treatment in clinical populations such as Parkinson’s Disease, have been linked to impulsivity and compulsive behaviour [33, 34], whereas they decrease impulsivity in individuals with ADHD . These findings point to a likely important relationship between dopaminergic function and compulsive behaviour, in which COMT genotypes play a role. However, the exact mechanisms linking COMT, dopaminergic function and compulsive behaviour for individuals with neurodevelopmental disorders requires further investigation.
There are several strengths to this investigation into the role of three distinct SNPs on specific clinically relevant behavioural characteristics, including the a priori selection of SNPs based on existing theoretical bases. In addition, the current study includes the largest FXS sample to date to examine SNPs as a genetic basis for variability in the behavioural phenotype. Finally, the current study used a number of standardised measures, which have been designed specifically for people with intellectual disability, to assess a wide range of behavioural characteristics, covering challenging behaviour, autism symptomatology, ADHD characteristics, repetitive behaviour, and mood/interest and pleasure. A limitation of the current study is the lack of IQ measures to characterise the ability level of participants. However, this was a product of conducting a postal survey in order to maximise the response rate for the largest possible sample size, and the Wessex served as a proxy for intellectual and adaptive functioning. In addition, a number of statistical comparisons were employed to fully investigate the role of 5-HTTLPR, MAOA and COMT variation in behavioural characteristics. The number of comparisons, as well as the modest sample size, particularly in the COMT A/A group, increases the chance of Type 1 error. However, such issues are inevitable when collecting data from individuals with rare genetic syndromes. Applying stringent statistical methods to counteract this may eliminate the reporting of a true association. Here, measures were not taken to account for multiple comparisons. Rather, it is suggested that the results are interpreted with caution and replication is encouraged. In addition, given the challenges associated with collecting saliva samples from individuals with FXS, especially those displaying self-injurious or aggressive behaviour, the sample of this study, as well as other studies utilising similar methodologies, may be biased in such a way that excludes those with the most severe presentations of these behaviours.