Updated report on tools to measure outcomes of clinical trials in fragile X syndrome

Aberrant Behavior Checklist-Community (ABC-C), adapted for FXS (ABC-C_FX). The process of refactoring the ABC-C for FXS, through a multi-site Fragile X Clinical and Research Consortium (FXCRC) collaboration [111], led to the elimination of three items and, more importantly, to a different factor structure that added novel subscales that seem to better represent autistic and social anxiety behaviors. The ABC-C_FX factor structure was further supported by a subsequent study conducted by Novartis and RTI/Our Fragile X World [103]. The ABC-C_FX has been used in every drug trial performed in FXS since its publication in late 2011, including a pediatric arbaclofen trial that showed some significant improvements [112, 113]. Based on these controlled trials and their open-label extensions, it has been concluded that the ABC-C_FX could be used as either a shorter- or longer-term outcome measure (see Table 3). Considering its initial psychometric evaluation [111], re-assessment [103], and application to FXS trials, in which several subscales have demonstrated sensitivity, the ABC-C_FX falls into the moderate-strong category of instruments (two positive randomized placebo-controlled trials: phase 2 mavoglurant [9], phase 3 pediatric arbaclofen [46, 113]). This labeling corresponds mainly to the ABC-C_FX Irritability subscale, pertinent to the Disruptive Behavior domain. However, two other ABC-C_FX subscales, specifically Social Avoidance and Hyperactivity, demonstrate many of the tool quality features of the Irritability subscale in a phase 3 pediatric arbaclofen study [113]. The main criticism continues to be its limited coverage of some key areas (e.g., non-social anxiety), which is discussed below, as well as the susceptibility to placebo effects. Nonetheless, rating variability and placebo effects are not specific to the ABC-C_FX but are associated with any parent/caregiver report measure.

Fragile X Syndrome Rating Scale (FXSRS). As concluded by the Working Group, development of a new behavior rating scale covering the FXS behavioral phenotype and associated symptoms is a worthwhile effort. An initial attempt has been completed in the context of a trofinetide phase II trial conducted in adolescents and adults with FXS by Neuren Pharmaceuticals [63, 114]. The FXSRS is a 34-item rating scale that covers a wide range of behavioral symptoms, including a FXS Core Phenotype subscale (i.e., 10 symptoms more prominent or prevalent in FXS than in other neurodevelopmental disorders), a FXS with Autism Phenotype subscale (i.e., six features shared to a similar degree by FXS and idiopathic ASD), and an Associated Phenotypic Features subscale (i.e., 18 features shared to a similar degree by FXS and general ID) subscales [114]. Items are answered using a 4-point severity or frequency scoring system. Publication of the FXSRS is pending; however, description of features in a group of 70 males with FXS, aged 12–45 years, presented at a meeting in 2014 [114], suggests it meets the requirements stated in the 2013 Report, including its potential use as shorter- and longer-term outcome measure (Table 3). Although grading of the FXSRS at this point is not possible or limited, it was employed in the aforementioned trofinetide study. This trial was reported as positive, based on an analytical strategy using multiple endpoints including the FXSRS (Berry-Kravis et al. 2016; www.​neurenpharma.​com/​IRM/​PDF/​1557/​Trofinetidesucce​ssfulinPhase2tri​alinFragileX).

Behavior Problems Inventory-Short Form (BPI-S; [115]). The BPI-S is a 30-item parent report measure that includes aggressive/destructive, self-injury, and stereotypic behaviors subscales rated over the previous 2-month period. Items are answered using a 5-point frequency and a 4-point severity scoring system. Studies employing an earlier version of the scale (BPI-01; [116, 117]) reported that in 50 males with FXS, aged 8–24 years, 79% engaged in self-injurious behavior (SIB), 98% in stereotypic behavior, and 75% in aggressive/destructive behavior. Moreover, 33% of the sample demonstrated aggression and 80% stereotypic behavior, both on a daily basis. The BPI-S could be used as either shorter- or longer-term outcome measure. However, it has not been subjected to reproducibility analyses in FXS and, at present, it is a limited quality tool.

The ADHD Test (ADHDT; [123]). A study employing the ADHDT reported that 68 males and females with FXS, aged 15–25 years, displayed a good range of scores [125]. The ADHDT contains 36 items divided into 3 subscales: Hyperactivity (13 items), Impulsivity (10 items), and Inattention (12 items). The ADHDT was normed to a group of 3- to 23-year-old individuals who met the DSM-IV criteria for ADHD. The psychometric properties of the ADHDT have been reported to be good, but its only application to FXS is as a measure of ADHD symptoms in conjunction with the ABC-C_FX Hyperactivity subscale.

Conners Rating Scales-Revised [124]. Another recent study involving 46 boys with FXS, aged 4–11 years, showed “stable and striking impairments in inattention” on the Conners Teacher Rating Scale (CTRS) and the Conners Parent Rating Scale (CPRS) [126]. Both scales have four subscales: ADHD Index, Hyperactivity, Cognitive, and Oppositional, scored on a 4-point Likert scale, and an age-normed T score above 70 on the ADHD index is indicative of an ADHD diagnosis. The CPRS has reported good to excellent internal consistency across the four scales [124, 126], with scores on the CPRS generally higher than on the CTRS.

Child Behavior Checklist (CBCL)’s ADHD subscale [127]. Several observational studies have employed the CBCL (ages 1.5–5 years, ages 6–18 years; [127, 128]) as a measure of ADHD symptoms in FXS [reviewed in [3]. The CBCL is a 118-item parent rating scale used to assess internalizing and externalizing symptoms in children without ID. Items are answered using a 3-point Likert scale. The DSM-ADHD scale is one of several DSM scales of the CBCL based on criteria from the DSM-IV, with T scores above 70 being indicative of an ADHD diagnosis. The CBCL was normed on a large, representative sample of children and adolescents and, in this population as well as in idiopathic ASD, has overall good psychometric properties [11, 129]. In a recent longitudinal study employing this measure, Grefer and colleagues reported that the mean DSM-ADHD raw scores increased over a 2-year period in a sample of 33 preschool-aged boys with FXS [130]. The percentage of boys who obtained T scores in the clinical range also increased from 9 to 12% over the same period. However, considering that the norms of the CBCL were not developed for children with ID, the reliability of this scale for individuals with FXS is unknown, and it is unclear whether this scale can be useful in clinical trials given its narrow range of scores.

The SNAP-IV [40]. This is a revision of the Swanson, Nolan and Pelham (SNAP) Questionnaire [131]. The SNAP-IV is an 18-item questionnaire addressing symptoms of ADHD (72 additional items cover ADHD-related behaviors such as those of Oppositional Defiant Disorder). It has been used in a completed (ampakine CX516) study [40] and an ongoing (ganaxolone) trial in FXS, showing very good test-retest reproducibility but no sensitivity to change in the CX516 trial [40].

Overall, all these potential instruments for evaluating ADHD symptoms in FXS are promising shorter-term outcome measures, but their tool quality fall in the limited to no evidence available quality range.

Pediatric Anxiety Rating Scale (PARS). A pilot observational study of the PARS in FXS, involving 49 subjects, aged 5–35 years, confirmed its potential for clinical trials. It demonstrated good psychometric properties, including feasibility, reliability, and convergent validity [138]. Responsiveness to change (i.e., sensitivity) is still unknown, although this information may become available soon since the PARS has been already implemented in three clinical trials.

The Anxiety, Depression and Mood Scale (ADAMS). A recent observational study of the ADAMS in Rett syndrome concluded that, among measures of anxiety-like behaviors, it has the best psychometric properties [17]. Although sensitivity was not examined in this investigation, it included an assessment of the ADAMS’ functional relevance. ADAMS scores correlated inversely with adaptive behavior skills and quality of life scores [17] indicating that this is a clinically meaningful measure for interventions (as defined by the FDA) for Rett syndrome and probably other neurodevelopmental disorders such as FXS. Moreover, a pilot open-label trial with mecasermin (recombinant human IGF-1) in Rett syndrome demonstrated that the ADAMS Social Avoidance subscale’s was mildly sensitive to response to treatment [139]. A recently completed but not yet reported trial of ganaxolone in FXS employed, in addition to the PARS, the ADAMS.

Autism Diagnostic Observation Schedule-Generic (ADOS-G). The ADOS-G is a widely used and well-established semi-structured, interactive instrument designed to assess aspects of social reciprocal interaction, communication, stereotyped behaviors, and restricted interests and play [147]. It is included in this review because the ADOS-G Severity Score is an overall measure of autistic behavior severity [148] and it has been administered pre- and post-treatment in the Early Start Denver Model (i.e., early intervention) randomized controlled trials in idiopathic ASD [149, 150]. There have been some recent refinements of the ADOS-G algorithm, such as the development of the ADOS-Calibrated Severity Score (ADOS-CSS) [148] that is independent of age, IQ, and language level [148, 151, 152]. The ADOS-G and, in particular, the ADOS-CSS seem to be suited for shorter-term trials [149, 150]. Together, based on data on reliability [147, 153‐155] and validity [147, 154‐158], the tool quality of the ADOS measures is in the limited to moderate range in idiopathic ASD. However, because of the lack of studies in individuals with the disorder, ADOS measures should be considered as having limited tool quality for FXS.

Social Responsiveness Scale (SRS). The features of the SRS were reviewed in the preceding section on “Social cognition”. A single open-label 10-week trial of acamprosate involving 12 subjects with FXS, aged 6–17 years, demonstrated a significant improvement in the scores of the SRS and several other cognitive and behavioral measures [159]. As noted in the “Social cognition” subsection under the “Cognitive measures” section, the SRS is suited for evaluating either shorter-term or longer-term effects, and its tool quality should be considered in the limited-moderate range in FXS.

Childhood Autism Rating Scale (CARS). This is a widely used 15-item observation and parent/caregiver interview that quantifies the severity of behaviors associated with ASD, with total scores ≥30 strongly suggesting the presence of the disorder [160, 161]. The CARS, Second edition (CARS-2) is a more recently published clinician-completed rating also used to determine ASD symptom severity [162]. The CARS-2 is suitable for evaluating shorter-term effects; its reliability [160, 163, 164], validity [165, 166], and lack of application to trials in FXS place it at the limited tool quality level.

Social Communication Questionnaire (SCQ; originally called the Autism Screening Questionnaire; [167]). The SCQ is a 40-item parents/caregivers rating scale based on the ADI-R [168] that screens for current core ASD behaviors and at age 4–5 years (lifetime). Language items not suitable for non-verbal children can be omitted; it is scored according to language level (maximum score 32 or 39), with higher scores indicating more severe symptoms. Five studies have used the SCQ to characterize autistic behavior and its severity in FXS [141, 169‐172]. However, none has examined measurement properties including sensitivity. The SCQ is adequate for evaluating shorter-term outcomes. SCQ’s reliability [173, 174] and validity ([163]; reviewed in [12]) support the limited tool quality level.

Repetitive Behavior Scale-Revised (RBS-R). RRBs are a broad range of behaviors that are subdivided into two conceptual categories [175]: “lower-order” motor actions associated with lower developmental levels and characterized by repetition of movement (e.g., dyskinesias, stereotyped and repetitive manipulation of objects, and repetitive forms of self-injurious behavior (SIB)), and more complex or “higher-order” behaviors associated with higher cognitive abilities (e.g., object attachments, insistence on sameness, repetitive language, and circumscribed interests) [176, 177]. Both categories of behavior appear to be a function of an overall cognitive/behavioral rigidity/lack of flexibility [178]. Although no single RRB appears to be specific to ASD, an elevated pattern of RRB occurrence, co-occurrence, and severity characterizes the disorder [179]. The RBS-R is the most widely applied instrument for evaluating RRBs. The RBS-R is a 43-item parent/caregiver rating scale that was normed on individuals with ID and revised to capture some of the complex RRBs observed in ASD [179]. The items have been grouped into six subscales: Stereotyped, Self-injurious, Compulsive, Ritualistic, Sameness, and Restricted interests. The RBS-R was designed for use in populations with ASD or related neurodevelopmental disorders [179, 180]. RBS-R’s factor structure has been empirically evaluated in multiple studies [181‐184]; recently, Bishop and colleagues concluded that a five-factor structure (Sensory Motor, Restricted Interests, Self-injury, Compulsive, and Ritualistic/Sameness and Sensory Motor behaviors) was the best solution for the RBS-R [183]. The scale’s psychometric properties are variable, ranging from poor (test-retest reliability) to strong (internal consistency) [180]. Thus, its overall quality as tool in idiopathic ASD should be considered moderate. The RBS-R was applied in a study that found different profiles of RRBs in young boys, aged 3–5 years, with either FXS and ASD or idiopathic ASD [185]. Both groups were similar with respect to lower-order (motoric) RRBs, but differed in terms of more complex forms that were less severe in FXS with ASD [186]. Furthermore, Wolff and colleagues found positive correlations between Self-injurious total scores on the RBS-R and number of SIB topographies and bilateral caudate nuclei volumes in the FXS group comorbid for ASD [187]. This study also reported that the profile of RRBs remains stable in FXS from preschool through at least the middle school years. This is in contradiction with the analyses of the ABC-C_FX Stereotypy, which demonstrates a decline in scores over time [103]. To date, the only FXS-specific factor analyses conducted on any measure of RRBs are those performed for the development of the ABC-C_FX . Although the RBS-R has been used in some FXS trials, information on its sensitivity to change is not yet available. Thus, the RBS-R seems suited for evaluating shorter-term and possibly also long-term effects. In the review of measures of RRBs for trials in idiopathic ASD by Scahill and colleagues, the authors concluded that the RBS-R, as well as the ABC-C Stereotypy subscale, Stereotyped Behavior Scale (SBS), and the Repetitive Behavior Questionnaire (RBQ) (see the next section), are appropriate with conditions [15]. Based on the latter and the lack of data in FXS, the RBS-R’s tool quality should be considered limited. However, because of the relevance of RRBs to the FXS behavioral phenotype, the RBS-R and similar instruments should be further studied in terms of suitability for clinical trials for this genetic disorder.

Repetitive Behavior Scale-Revised (RBS-R). Although RRB categories have been reported in FXS at a global level, there is limited research characterizing the relative frequency of different types of RRBs or association with other factors. The RBS-R, discussed in more detail in the previous section on “Autistic behavior,” was employed by Oakes and colleagues to examine the profile of RRBs in 39 boys with FXS, aged 6–10 years, without specifying their ASD status [194]. Restricted Interests and Sensory Motor behaviors were reported as most problematic, in contrast to SIB. Non-verbal IQ was negatively related to RRBs in general, whereas anxiety and social affective symptoms of ASD were positively correlated with scores on Restricted Interests. Anxiety was also positively correlated with scores of Compulsive behaviors and Ritualistic Sameness behaviors. Despite the relative specificity of the reported profile [194], it is important to point out that all subscales of the RBS-R were significantly inter-correlated. No application of the RBS-R for measuring changes in non-ASD-related RRBs has been reported. Therefore, this particular use of the RBS-R cannot be evaluated or a tool quality grading of no evidence available is the most appropriate.

ABC-C Stereotypy and Inappropriate Speech subscales. These are two of the five original subscales of the ABC-C that cover RRBs. The content of both subscales was relatively preserved after the factor analysis that originated the ABC-C_FX (i.e., one Stereotypy item was incorporated into the new Socially Unresponsive/Lethargic subscale; [111]), demonstrating their appropriateness for FXS. Two independent studies demonstrated the usefulness of these subscales for delineating trajectories of RRBs in FXS [103, 111]. While these cross-sectional analyses showed that behaviors under Stereotypy are in the moderate range of severity and tend to slightly decrease over time, those under Inappropriate Speech are on the severe range and remain relatively stable between 6 and 25 years. Since the ABC-C_FX autistic behavior-oriented Socially Unresponsive/Lethargic subscale did not incorporate many items of Stereotypy and Inappropriate Speech, it is assumed that these subscales reflect predominantly non-autistic RRBs. However, no direct evidence supporting this is currently available. As the ABC-C_FX Inappropriate Speech subscale includes only four items; this may limit its sensitivity in a similar fashion to the Social Avoidance subscale (see the “Anxiety” subsection, under the “Behavior and Emotion measures” section). We can conclude that, as other ABC-C_FX subscales, Stereotypy and Inappropriate Speech are adequate for assessing either shorter-term or longer-term outcomes. Psychometric data, including their application to several recent FXS trials, support a moderate general tool quality label. Nonetheless, no data are available on the specific use of these instruments for measuring non-autistic RRBs in FXS. Therefore, their tool quality for this specific indication would correspond to no evidence available.

Repetitive Behavior Questionnaire (RBQ). The RBQ is a 19-item scale that shares some individual items with the RBS-R. Moss and colleagues examined the presence of RRBs across six groups of individuals with ID, including males with FXS (n = 191), aged 6–47 years. It was reported that the FXS group had higher scores on all five subscales (Stereotyped Behavior, Compulsive Behavior, Insistence on Sameness, Restricted Preferences, and Repetitive Speech); moreover, they also had more severe RRBs than two other ID groups in the categories of Compulsive Behavior, Insistence on Sameness, and Repetitive Speech [189].

Behavior Problems Inventory (BPI). Employing an earlier version of the BPI scale [195], an instrument described in the “Problem behaviors: focus on disruptive behavior domain” subsection, and the RBS, Bodfish and colleagues found that the difference in RRBs between ID and ASD was mainly quantitative [179]. Namely, they reported that the occurrence of specific topographies of RRBs as well as their severity in 34 adults (23 males, 11 females) with non-ASD ID (CARS <30) matched to that of an ASD group. This finding supports the notion that the RRBs in ASD are similar to those described in ID [196]. While both groups, ID and ASD, had significant patterns of RRBs co-occurrence, the exception was the frequency of dyskinesias that was higher in the ASD group.

Mitogen-activated protein kinases/extracellular signal-regulated kinases (MAPK/ERKs). The MAPK/ERKs are a group of proteins that regulate nodal points for several signaling cascades. MAPK/ERK regulation mediated by phosphorylation is typically termed activation. Delayed early-phase phosphorylation of ERK has been noted in lymphocytes of individuals with FXS and in neurons and thymocytes of Fmr1 knockout mice. This delay is described as an enhanced time to half maximum ERK activation following phorbol ester stimulation [199]. ERK peripheral lymphocytic activation has been assessed pre- and post-drug treatment in studies of lithium and riluzole in FXS. ERK activation was significantly reduced (i.e., normalized) following lithium [55] or riluzole [200] administration. In an open-label trial involving 16 individuals with FXS, aged 6–23 years, lithium (a mood stabilizer that inhibits the phospholipase C signaling pathway and GSK-3B [201]) use was associated with normalization of ERK activation kinetics but no improvement in maladaptive behaviors evaluated by the ABC-C [54]. These findings can be explained, at least in part, because of the high variability of the biochemical assay (Berry-Kravis, personal communication). Riluzole is FDA approved for treatment of amyotrophic lateral sclerosis and is postulated to reduce glutamate release, block NMDA receptors, and enhance GABA activation [202]. In this pilot 6-week trial including six males with FXS, aged 19–24 years, Erickson and colleagues found that despite normalization of ERK activation kinetics, no significant clinical improvements were noted [200]. A recent evaluation of ERK and Akt (another major kinase pathway) phosphorylation in individuals with FXS, in relation to lovastatin treatment, demonstrated normalization of baseline increased ERK activity after drug administration. Of note, these changes in ERK phosphorylation were correlated with clinical response to lovastatin [203].

Brain-derived neurotrophic factor (BDNF). BDNF is a growth factor and synaptic modulator that supports survival, growth, and differentiation of neurons. BDNF has been shown to impact FMRP expression [204]; its application to hippocampal slices from Fmr1 knockout mice leads to rescue of long-term potentiation deficits [205]. BDNF expression has also been noted to be reduced in Fmr1 knockout mice [206]. In a study of acamprosate in 12 individuals with FXS, aged 5–17 years, BDNF plasma levels were shown to increase significantly after 10 weeks of treatment with the drug [159]. Nonetheless, BDNF levels in the nine subjects deemed to be responders did not correlate with their clinical response. Acamprosate is drug approved for the treatment of alcoholism, which is postulated to attenuate glutamatergic activity, particularly NMDA receptor-dependent, and potentiate GABA A receptor activity [207]. Additional relevant data on BDNF were obtained in a small open-label fluoxetine trial in children and adolescents with idiopathic ASD. Using the Autism Treatment Evaluation Checklist, the cohort showed improvements in several aspects of communication, socialization, and cognitive awareness, changes that were correlated to decreases in serum levels of BDNF [208]. Since the direction of the BDNF changes was unexpected, this study emphasizes the need for further exploration of BDNF and related molecules as biomarkers of response to treatment in neurodevelopmental disorders including FXS.

Amyloid precursor protein (APP). APP is a key neural modulator, which is processed by two pathways. The best known is the so-called amyloidogenic or β pathway that leads to the production of the neurotoxic Aβ40 and Aβ42 fragments, which have been implicated in the pathophysiology of Alzheimer disease. The non-amyloidogenic or α pathway results in the production of the soluble neurotrophic APP alpha (sAPPα) [209]. FMRP is known to regulate APP mRNA expression [210, 211], with baseline APP levels being elevated in Fmr1 knockout mice [211]. APP and its cleavage products were measured in a subset of nine participants (mean age 10.9 years) in the acamprosate trial described above [159]. Total APP and sAPPα were reduced after treatment, while Aβ40 and Aβ42 remained unchanged. These APP-related decreases correlated modestly with improvements on the ABC-C Lethargy/Social Withdrawal subscale.

Prepulse inhibition (PPI). The potential of PPI as a biomarker in FXS was confirmed in an early trial (n = 12) with the mGluR5 antagonist fenobam that showed a ≥ 20% improvement in half of the subjects (although there was no clear clinical improvement) [226]. However, subsequent work with PPI reveals that the measure is very sensitive to differences in equipment and environment, which has led to lack of inter-site reproducibility. Consequently, PPI may be most useful in early proof-of-concept trials as opposed to large confirmatory multi-site trials. PPI seems suited for evaluating shorter-term effects, and its tool quality is limited-moderate for the aforementioned reasons.

Eye tracking and pupillometry. A standardized protocol quantifying pupillary diameter and looking time and number of fixations to the eye region for a series of happy, calm, and fearful human faces has been shown to be suitable for individuals of varying ages with FXS, with a 7–51 years age range, with strong feasibility and test-retest reliability for all outputs [85]. This protocol has been applied in several clinical trials, although a full analysis of sensitivity to treatment has not yet been published. However, preliminary reports indicate normalization of parameters in response to treatment with minocycline [87] or the mGluR5 antagonist mavoglurant [88]. The eye-tracking protocol described above is the only one employed in clinical trials in FXS; nonetheless, several others have been developed and are potentially useful (see the “Social cognition” section). In conclusion, although reports on ongoing studies may change our conclusions, at present, eye tracking and pupillometry should be considered moderate level quality tools suited for evaluating shorter-term effects in FXS trials.

Event-related potentials (ERPs). Several observational and treatment studies, mainly pilot type, have utilized EEG to characterize ERPs in FXS and to assess treatment response and associated changes in cortical activity. Van der Molen et al. showed that the N1 and P2 components of auditory ERPs from 16 males with FXS, aged 18–42 years, were significantly enhanced. Schneider and colleagues observed changes in electrocortical habituation to auditory stimuli (changes between the first and last of 45 presented stimuli), specifically improved N1 and P2, associated with minocycline treatment (double-blind, placebo-controlled crossover, n = 12, mean age 10.5 years), suggesting the feasibility and sensitivity of ERPs as a biomarker in FXS treatment trials [227]. In another study, Yang et al. employed auditory ERPs to evaluate the effects of chronic memantine treatment on verbal memory in individuals with fragile X-associated tremor/ataxia syndrome (FXTAS) [228, 229]. The authors reported treatment-associated benefit on cued-recall memory as well as corresponding changes in N400 ERP response [228, 229]. Recently, Ethridge and colleagues also examined ERPs during a somewhat different passive auditory habituation task in individuals with FXS (n = 14, 14–57 years) and found “giant” N1 amplitudes to all four repeated stimuli in the FXS group more frequently than in a matched control subjects [230]. Of relevance to behavioral measures in FXS, the N1 amplitude enhancement was strongly correlated with auditory processing abnormalities on the Sensory Profile, a caregiver questionnaire employed in the ASD field that measures children’s sensory processing abilities [231], and with increased ASD features on the SCQ [230]. N2 ERP amplitudes were also decreased in FXS and correlated with scores on the ABC-C Irritability subscale. Overall, this study [230] exemplifies the type of information provided by auditory ERPs in FXS that links biomarkers and behavioral outcome measures. No study so far has examined age- or gender-related differences in ERP profiles in FXS, and no data are available on test-retest reliability or other measurement properties. However, studies are currently in progress to examine resting state and MMN in FXS. The use of ERP as a biomarker of multiple aspects of neural activity and processing in FXS is currently being explored at several centers. This includes the introduction of auditory ERP as a biomarker in the upcoming mavoglurant and language learning trial in children with FXS (U01 NS096767). Thus, at present, auditory ERP seems to be a biomarker suited to evaluating medication shorter- or longer-term effects, with current evidence supporting a limited-moderate tool quality. Other potentially valuable neurophysiological measures, in particular EEG spectral analyses, are just beginning to be examined in FXS, as described in the following section.

Structural MRI. Research in other disorders indicates that significant changes in gray and white matter volume, white matter microstructure, and brain activation can occur in association with intervention within relatively short periods of time. For example, Roberto and colleagues showed that short-term (mean 50 days) weight restoration in under-weighed adult patients with anorexia nervosa was associated with significant increases in gray and white matter volume relative to healthy controls [237]. Eijk et al. demonstrated that, in individuals with alcohol abuse, 2 weeks of supervised abstinence was associated with significant recovery of gray matter volume in several brain regions including the cerebellum and parietal lobe [238]. There are many other examples of relatively rapid changes in neuroimaging metrics in clinical groups receiving disorder-focused intervention. Administration of risperidone and ziprasidone to individuals with schizophrenia led to increased cerebral gray matter volume after 28 days of treatment [239] and administration of citalopram to depressed patients increased hippocampal gray matter within 8 weeks [240]. Changes have also been observed in white matter microstructure by DWI; patients with multiple sclerosis showed increases in fractional anisotropy within 2 months of facilitation of physiotherapy [241], depressed patients showed increase in fractional anisotropy with 4 weeks of psychotherapy [242], and patients with obsessive-compulsive disorder showed a reversal of abnormal white matter microstructure after a 12-week course of antidepressant (SSRI) treatment [243].

Functional MRI. The feasibility and utility of fMRI as a putative biomarker of intervention has been demonstrated in hundreds of treatment studies, as described in recent reviews of pharmacology, psychotherapy, pain management, and rehabilitation [244‐250]. A recently completed study by Reiss and colleagues utilized fMRI as an outcome measure in a clinical trial of donepezil for FXS; these functional imaging data are currently being analyzed to determine if drug-related effects are observable and associated with clinical change.