The present study employed single-voxel MRS to examine in vivo neurometabolite concentrations in humans with FXS and provides direct evidence of altered metabolite concentration in the caudate nucleus. We demonstrate significantly reduced levels of choline/creatine and Glx/creatine in a group of males and females with FXS, relative to a group of individuals without FXS who were matched for age, sex and general intellectual functioning. These results are in line with the only previously published human FXS MRS study  and they corroborate previous reports of altered neurometabolic functioning in animal models of FXS . Aberrant neurometabolite levels may underlie some of the clinical symptoms seen in FXS and they may be related to aberrant receptor signaling seen in animal models [17, 18].
FXS has previously been associated with greatly enlarged caudate size [10–12] and aberrant frontostriatal executive functioning networks [14, 24]. We provide evidence for altered metabolite concentrations, further elucidating atypical caudate neurobiology in FXS. Given the caudate’s role in learning, memory and executive functions , aberrant metabolite levels in this region may mediate some of the behavioral and cognitive deficits associated with FXS. Although the precise effects of FMRP on neurometabolism are not fully understood, recent findings indicate that lack of FMRP results in aberrant functioning of specific GPCRs, mAChRs and mGluRs [17, 18], which are highly expressed in striatal circuits . Therefore, the altered neurometabolite levels reported here may be related to hypersensitive mAChR and mGluR signaling. Additionally, FMRP plays a role in regulating calcium-dependent potassium (BK) channels, which are highly expressed in striatal circuits and may also contribute to altered metabolite levels . The direct causal pathway between hypersensitive receptor functioning, BK channel dysregulation and decreased metabolite levels revealed by MRS has yet to be determined, but our results provide an important, although indirect, link.
Glutamate, glutamine and GABA contribute to the Glx peak at 3.78 ppm, although the contribution of GABA is extremely small . Glutamate and glutamine levels are indirectly related to glutamatergic signaling, which is critical for synaptic plasticity and learning ; thus, decreased Glx may be a biomarker for learning deficits associated with FXS. A pilot study examining premutation carriers of the FMR1 gene did not find glutamatergic abnormalities in this condition , which is associated with between 55 to 200 CGG repeats and generally normal, though potentially variable overall FMRP production . However, decreased levels of MRS visible Glx have been reported for individuals with autism spectrum disorders (ASD) , a set of behaviorally defined disorders in which cognitive and behavioral symptoms overlap with those observed in FXS . As with FXS, animal models of ASD have revealed functional abnormalities in both excitatory (glutamate) and inhibitory (GABA) systems [43–45]. These findings suggest some degree of common neurobiological alteration despite differential origin for cognitive and behavioral symptoms in FXS (reduced FMRP) and idiopathic ASD (variable unknown causes). MRS examinations of ASD have reported decreased levels of NAA [46, 47], which has not been previously shown in FXS, although we did report a trend for lower NAA/creatine ratios and lower absolute NAA in FXS. Future studies comparing ASD to FXS directly may be needed to understand common and divergent neurobiological underpinnings.
The MRS visible choline peak at 3.22 ppm includes phosphocholine and glycerophosphocholine, phospholipids involved in membrane synthesis and integrity, which are markers of cellular density . Decreased choline within the FXS group may be indicative of decreased overall cellular density in the caudate. Free choline, the precursor for acetylcholine, represents a relatively small portion of the MRS visible choline peak, yet this peak correlates with in vivo acetylcholine measured in rat brain . This animal research suggests reduced choline may indicate altered acetylcholine levels in humans, but more evidence is needed to determine the reliability of MRS signal as a marker of acetylcholine level. Such a non-invasive marker would be extremely useful for the study of FXS given the evidence for altered acetylcholine receptor signaling in Fmr1-KO mice .
Our primary results suggest that choline and Glx differences are present in both males and females with FXS. Analysis for females only confirms that females with FXS have significantly reduced choline and Glx which, in context with previous research demonstrating altered metabolite levels in males , indicates that these neurometabolic systems may be viable candidates for pharmacological treatment endpoints in both sexes. We did not have a large enough sample of male participants to draw conclusions regarding males, but similar effect sizes for male and female participants indicate that similar altered metabolite concentration may exist in both sexes. Future studies with larger sample sizes in each sex are essential for expanding knowledge in this area.
We explored the relationship between neurometabolite concentration and cognitive/behavioral functioning within each group, but found no significant correlations. The measures of cognitive/behavioral functioning we utilized may not have been sensitive enough to detect such relationships and we did not include specific measures of learning or memory, which may be related to choline  and glutamine  metabolism. Furthermore, sex differences or medication usage may have obscured the relationship between cognitive/behavioral functioning and metabolite concentration. Larger sample sizes, wider age ranges and longitudinal data points are required to clearly elucidate such complex brain/behavior relationships.
The nature of our study population dictated inclusion of participants taking medication and, although there was no within group relationship between metabolite concentration and medication usage, we cannot rule out the possibility that medication has some effect on metabolite concentration. Our post hoc analysis including only medication-free individuals showed a trend for lower choline/creatine and Glx/creatine for the FXS group, but differences did not reach significance. Including only medication-free individuals biased our sample toward higher functioning individuals in each group and reduced the statistical power. Larger-scale investigations are required to adequately address the relationships among metabolite concentration, medication usage and phenotypes associated with FXS.
We present metabolite data referenced to creatine, a metabolite widely used as a reference in human MRS, because its concentration remains stable regardless of changes in energy metabolism or disease progression , although research suggests creatine levels may be altered in the Fmr1-KO mouse . Therefore, we conducted a secondary analysis using absolute water referenced values for each metabolite and noted significant group differences in choline and Glx, as well as in NAA. We interpret the difference in NAA with caution, since we were not able to account for group level covariates in the analysis of absolute concentration and we noted only a trend for lower values in the FXS group on the NAA/creatine ratio. Importantly, we did not find a significant group difference in creatine, supporting the use of that metabolite as a reference in our analysis. We were unable to quantify GABA or glutamine concentrations individually, or to examine more than one region of interest, given our limited time frame for MRI data acquisition. Future investigations employing higher magnet strength, spectral editing and multi-voxel imaging may further elucidate the neurometabolic alterations in FXS.