Aberrant Neural Response During Face Processing in Girls with Fragile X Syndrome: Defining Potential Brain Biomarkers for Treatment Studies

Abstract

Background:

Children and adolescents with fragile X syndrome (FXS) manifest significant symptoms of anxiety, particularly in response to face-to-face social interaction. In this study we used functional near-infrared spectroscopy (fNIRS) to reveal a specific pattern of brain activation and habituation in response to face stimuli in young girls with FXS, an important but understudied clinical population.

Methods:

Participants were 32 girls with FXS (age: 11.8 ± 2.9 years) and a control group of 28 girls without FXS (age: 10.5 ± 2.3 years) matched for age, general cognitive function and autism symptoms. Functional NIRS was used to assess brain activation during a face habituation task including repeated upright/inverted faces and greeble (nonface) objects.

Results:

Compared to the control group, girls with FXS showed significant hyper-activation in the frontopolar and dorsal lateral prefrontal cortices in response to all face stimuli (upright + inverted). Lack of neural habituation (and significant sensitization) was also observed in the FXS group in the frontopolar cortex in response to upright face stimuli. Finally, aberrant frontopolar sensitization in response to upright faces in girls with FXS was significantly correlated with notable cognitive-behavioral and social-emotional outcomes relevant to this condition including executive function, autism symptoms, depression and anxiety.

Conclusions:

These findings strongly support a hypothesis of neural hyper-activation and accentuated sensitization during face processing in FXS, a phenomenon that could be developed as a biomarker endpoint for improving treatment trial evaluation in girls with this condition.

Introduction

Fragile X syndrome (FXS), a common neurodevelopmental disorder caused by disrupted expression of the fragile X mental retardation-1 (FMR1) gene, is the most commonly known heritable cause of intellectual disability and autism spectrum disorder (1). The FMR1 full mutation results in hypermethylation, transcriptional silencing, and failure to express the Fragile X Mental Retardation Protein (FMRP), which plays a critical role in the regulation of synaptic plasticity and brain development (2). Because FXS is an X chromosome-linked condition, males are usually more severely affected than females who benefit from the moderating effects of a second (unaffected) X chromosome (3). Though less affected with respect to overall intellectual function, many females with FXS still experience poor functional outcomes due to high levels of anxiety, executive function and social deficits (4).

Social-emotional problems, such as social withdrawal, shyness, and social avoidance, are among the earliest and most maladaptive symptoms of FXS (3, 5), wherein eye-gaze avoidance during social interaction is a hallmark of the disorder (6). Developing effective interventions to address the neural underpinnings of social avoidance in FXS holds promise for improving long-term outcomes. However, at present, it remains a challenge to precisely monitor and evaluate the neural systems that are directly related to the development of social dysfunction. A more complete understanding of the association between observable behaviors and neural function is required to identify promising biomarkers for advancing a precision medicine approach to clinical care for individuals with FXS.

A large body of research utilizing structural magnetic resonance imaging (MRI) and functional MRI (fMRI) has linked social anxiety in FXS to abnormalities in neural systems underlying face and emotion processing, as suggested by aberrant morphology and activation in several key regions including the fusiform gyrus, superior temporal gyrus, anterior cingulate, insula, thalamus, prefrontal cortex and amygdala (7–9). These regions have been reported to contribute to neural systems underlying social cognition and emotion processing, with associated dysfunction hypothesized to contribute to social anxiety and avoidance observed in FXS. Functional MRI studies of FXS from our group have also demonstrated a deficit in habituation (significant sensitization) of neural responses to face stimuli (direct and averted gaze) in widespread areas including the frontal cortex, anterior cingulate and fusiform gyrus (9). Thus, an overarching aim of the current study was to focus on presumptive aberrant profiles of habituation in FXS relative to controls.

Despite the convergence of evidence showing abnormal neuroanatomy and neural responses in FXS, there remain important challenges in bringing widely used structural MRI and fMRI into clinical practice. Particularly, fMRI is costly to perform, highly sensitive to body-motion artifacts and has rigorous measurement requirements, making it difficult to use in routine examination of children, particularly those with developmental and behavioral problems. To bridge this gap, functional near-infrared spectroscopy (fNIRS), a noninvasive optical imaging technique that measures changes in cortical oxy-hemoglobin (HbO) and deoxy-hemoglobin (HbR) concentrations that are coupled with neuronal metabolic activity (10, 11), has been used as an alternative to fMRI for studying a variety of brain disorders, including Alzheimer’s disease (12–14) and autism spectrum disorder (15). Functional NIRS offers several advantages over fMRI, including portability, higher temporal resolution, and greater resilience to motion artifacts, which permits assessment of real-time brain dynamics in naturalistic settings (16, 17).

In this study, we sought to characterize the neural response in young girls with FXS and a control group without FXS during a face habituation task. We only included females for several reasons. First, FXS occurs in up to 1 in 5000 females. However, likely due to a milder phenotype relative to males, females with FXS are underrepresented in research studies. Yet, their more diverse and broader range of symptoms and overall higher IQ than boys allow females to play a particularly important role in understanding the complexities of the FXS phenotype. In particular, the study of females can potentially help elucidate neural mechanisms underlying hallmark symptoms of FXS such as anxiety, avoidance, and hyperarousal. Exploring the biopsychosocial bases of these social impairments in females with FXS, a critical step in developing more effective approaches to treatment, provides an opportunity to examine the impact of symptoms from the perspective of the individual as well as those who provide care. We hypothesized that, based on our previous fMRI study (9), girls with FXS would display aberrant neural habituation in the frontal cortex relative to an age, sex, and cognitively matched control group. Additionally, we sought to investigate associations between fNIRS-derived neural habituation measures and cognitive-behavioral and social-emotional symptoms, with the overarching goal of identifying potential neuro-biomarkers that could be used in future treatment trials in girls with FXS.

Methods and Materials

Participants

In this study, 32 girls (age: 11.76 ± 2.93 years) with FXS and a control group of 28 girls (age: 10.52±2.32 years) without FXS were recruited. The diagnosis of FXS was confirmed by molecular genetic testing obtained during initial screening/upon study enrollment. Girls in the control group were recruited through parent organizations, regional centers, school districts, social media, and flyer services throughout California. The research was approved by the Stanford University Institutional Review Board and performed in accordance with the Declaration of Helsinki. Participants and their parents/guardian were fully informed about the purpose of the research and provided written, informed consent prior to the start of the experiment.

The groups were matched on verbal ability using the Differential Ability Scales, Second Edition (DAS-II) (18), executive function using Behavior Rating Inventory of Executive Function-Global Executive Composite, Second Edition (BRIEF-2-GEC) (19), and autism symptoms/social behavior based on parent ratings from the Social Responsiveness Scale-2 (SRS-2) (20). We also assessed the participants’ depression, anxiety, and avoidance symptoms using the Anxiety, Depression, and Mood Scale (ADAMS-General anxiety/Depressed Mood/Social avoidance) (21) and Child Behavior Checklist (Parents report form for ages 6–18, CBCL-Anxious/Depressed) (22). All behavioral assessments were administered and scored based on standardized procedures according to each respective testing manual. All behavioral assessments, except the DAS-II, were completed by parent/caregiver. The DAS-II score was generated from direct testing of the participant.

Face Habituation Task and FNIRS Data Acquisition

A face habituation paradigm was employed to investigate our primary hypothesis. The face habituation task was presented in a pseudorandom order of 15 blocks including either upright faces, inverted faces or nonface ‘greeble’ objects interspersed with rest periods (Figure 1A, more details in Supplement). Each block only included one stimulus type with 12 exemplars per block (3 seconds duration for each exemplar), resulting in a total block duration of 36 seconds. Five blocks of each stimulus type were presented in a pseudo-random order interspersed with 10 seconds resting periods. To maximize attention to the stimuli, participants were asked to respond via button press to indicate the gender of each stimulus (including the greeble stimuli).

Figure 1.

Experimental design of this study. (A): The face habituation task; (B): An example of the measurement setup (a signed consent form has been obtained from the participant for the publication of this figure); (C): The location of fNIRS channels (projected from scalp to cortical surface, channels within the same ROI were assigned with the same color).

Hemodynamic activity of each participant was recorded using the NIRScout system (NIRX, Germany) with a sampling frequency of 7.8 Hz. A total of 48 fNIRS channels were placed bilaterally over the prefrontal and temporal cortices according to the international 10–20 EEG placement system to form 48 measurement channels (Figure 1B–C).

FNIRS Data Analysis

After the preprocessing of the raw fNIRS data (see Supplement), the General Linear Model (GLM) was employed to perform first-level analysis to identify activation in response to different conditions (upright face/inverted face/greeble objects) for each participant. Canonical hemodynamic response function was used to construct the design matrix. The group-level analysis was performed based on individual channel’ beta values obtained from the first-level models to identify group-wise activation patterns in response to each condition using a linear mixed effect model. Condition was treated as a fixed effect and participants were treated as random variables in the linear mixed effect model. Finally, any distinct cortical activation between conditions and between groups were assessed using two-tailed t-test.

To characterize the temporal dynamics of each participant’s task-evoked neural response, HbO signal was first normalized to a z-score, and single block HbO signal was segmented according to task timing, creating 5 blocks for each condition (upright face/inverted face/greeble), respectively. The participant’s neural activation at each block was computed by averaging the numerical values of the normalized HbO within each corresponding block. Between-block activation difference was tested within each group and between the groups using repeated analysis of variance (ANOVA) and post hoc two sample t-tests (two-tailed). For each condition, we defined the participant’s neural habituation as the difference between the mean HbO change at block 5 and block 1 (HbO_{Block5-Block1}). Finally, neural habituation values for the regions that were highly activated by the task, as tested by the group-level GLM analysis in the FXS group, were compared between the FXS group and the control group to characterize the pattern of neural response unique to the FXS group using two sample t-tests (two-tailed). Post-hoc Pearson’s correlation analysis was conducted to examine relationships between neural habituation measure and individual cognitive-behavioral measures. All multiple comparisons were controlled by Benjamini–Hochberg procedure (FDR = 0.05).

Results

Demographic, Behavior and Clinical Rating Scores

Demographic characteristics and assessment scores for both groups are summarized in Table 1. Overall, the groups did not differ significantly with regard to age, verbal IQ (DAS-II Verbal Ability), overall executive functioning (BRIEF-2 Global Executive Composite), depression (ADAMS Depressed Mood), anxiety (CBCL Anxious/Depressed), and autism symptoms/social behavior (SRS-2 Total Score). As anticipated, significant differences in ADAMS-General Anxiety and ADAMS-Social avoidance scores were observed between two groups. Task performance results, including accuracy and reaction time, were missing for five participants due to a technical issue (three girls with FXS and two girls in the control group). Comparisons of task accuracy and reaction time indicated no significant difference between the two groups.

Table 1.

Descriptive Statistics and Clinical Characteristics for the Fragile X Syndrome and control group (all p values were obtained via two sample t-test)

Characteristic	Fragile X Syndrome (n = 32)		Control (n = 28)		Analysis
	Mean	SD	Mean	SD	t	df	P
Age	11.76	2.93	10.52	2.32	1.799	58	0.077
DAS-II (Verbal Ability)a	81.97	17.75	89.54	19.57	−1.571	58	0.122
BRIEF-2 (GEC)b	61.96	11.31	64.50	10.63	−0.892	58	0.376
ADAMS (Depressed Mood)c	4.19	5.60	2.00	2.05	2.058	58	0.046
ADAMS (General Anxiety)d	6.88	4.29	4.11	3.53	2.707	58	0.009
ADAMS (Avoidance)e	6.00	5.09	2.79	2.90	2.947	58	0.005
CBCL (Anxious/Depressed)f	62.00	10.96	58.32	8.49	1.438	58	0.156
SRS-2 (Total Score)g	66.31	13.17	64.32	12.43	0.599	58	0.551
Task accuracy (%correct)	84.41	12.31	82.76	15.93	0.432	53	0.667
Reaction time (ms)	1135.31	201.69	1045.99	223.99	1.556	53	0.126

^aStandard score with mean=100, SD=15

^bT-scores

^cRaw scores

^dRaw scores

^eRaw scores

^fT-scores

^gT-scores.

FNIRS Regional Activation

Region-wise group analyses of the task-evoked activation revealed significant within as well as between-group findings. Specifically, girls in the FXS group demonstrated significant activation in frontopolar cortex (BA 10) and dorsal lateral prefrontal (DLPFC, BA 46) cortex in response to upright face stimuli (Figure 2A, upright face vs. baseline). They demonstrated significant activation at similar functional areas including frontopolar cortex (BA 10) and DLPFC (BA 9) cortex in response to inverted face stimuli (Figure 2B, inverted face vs. baseline). No significant brain activation was induced by greeble objects in the FXS group except for a region corresponding to right Wernicke’s area (BA 40, Figure 2C, greeble vs. baseline). For between-group analyses, compared to the control group, the FXS group showed significantly stronger activation in frontopolar cortex and DLPFC in response to the upright and inverted face conditions, respectively (Figure 3A–B). When combining all face stimuli (upright face + inverted face), stronger activation in frontopolar cortex (BA 10) and DLPFC (BA 9) remained significant in the FXS group compared to the control group (Figure 3C). Finally, girls with FXS did not show significantly greater brain activation in response to greeble objects compared to controls.

Figure 2.

Significant brain activation in response to different conditions in the FXS group. A: Upright face; B: Inverted face (this fNIRS channel covered both BA 9 and BA 10); C: Greeble object. Colored regions indicate p < 0.05 (FXS > baseline, false discovery rate corrected, see coordinates and t values in Table S1). L: left; R: right; BA: Brodmann Area; DLPFC: dorsal lateral prefrontal cortex.

Figure 3.

Significant brain activation differences between the FXS and control groups in response to each condition. A: Upright face; B: Inverted face (this fNIRS channel covered both BA 9 and BA 10); C: All faces (upright + inverted faces). Colored regions indicate p < 0.05 (FXS > Control, false discovery rate corrected, see coordinates and t values in Table S2). L: left; R: right; BA: Brodmann Area; DLPFC: dorsal lateral prefrontal cortex.

Region-of-Interest Analysis of Habituation

Based on the results of cortical response to the face and inverted-face conditions, we selected the frontopolar cortex and DLPFC as target regions for analysis of potential group differences in neural habituation. Repeated ANOVA indicated that there was a significant group-block interaction on the brain response to upright face stimuli at frontopolar cortex (df = 4, F = 3.637, p = 0.007). Girls with FXS showed increased activation in response to all facial stimuli (upright and inverted faces) in the frontopolar cortex across the experiment (Figure 4A–C). Particularly, activation induced by upright face stimuli in block 5 was significantly stronger than that in block 1 (t = 3.148, df = 31, p = 004) and block 3 (t = 3.007, df = 31, p = 005). Conversely, the control group demonstrated slightly reduced activation during the experiment. Compared to the control group, the FXS group presented significantly stronger response to upright face stimuli in the frontopolar cortex at block 5 (t = 2.707, df = 58, p = 0.009, Figure 4A). Finally, the FXS group showed significantly increased neural activation (sensitization) in response to upright face stimuli from block 1 to block 5 (HbO_{Block5-Block1}) in the frontopolar cortex compared to the control group (t = 2.689, df = 58, p = 0.009) (Figure 4D).

Figure 4.

(A-C): Temporal fluctuation of HbO in response to different conditions (A: Upright face; B: Inverted face; C: Greeble) at the frontopolar cortex. (D-F): Group-averaged habituation (HbO_{Block5-Block1}) analysis at the frontopolar cortex and DLPFC in response to different conditions. Note that the HbO values of channels within the same cortical regions were averaged. Asterisk indicates significant difference between two groups (p < 0.05). Note that sensitization in Y-axis equals change of activation between block 5 and block 1 (HbO_{Block5-Block1}), where value of 0 = no habituation, negative value = habituation, positive value = sensitization. A.U.: Arbitrary Unit.

Correlation between Neural Sensitization and Cognitive-Behavioral Measures

We examined the correlations between cognitive-behavioral and neural sensitization measures in the frontopolar cortex and DLPFC regions that demonstrated abnormal activation patterns in response to all face stimuli in the FXS group. The cognitive-behavioral measures that were utilized included the DAS-II Verbal Ability, BRIEF-2-GEC, CBCL Anxious/Depressed, ADAMS General Anxiety, ADAMS Depressed Mood, ADAMS-Social avoidance, and SRS-2 Total score. Frontopolar sensitization induced by upright face stimuli in the FXS group was significantly correlated with BRIEF-2 GEC (r = 0.559, p = 0.001), CBCL Anxious/Depressed (r = 0.458, p = 0.008), ADAMS Depressed Mood (r = 0.446, p = 0.011), ADAMS General Anxiety score (r = 0.411, p = 0.019), ADAMS-Social avoidance and SRS-2 Total score (r = 0.434, p = 0.013) (Figure 5). A marginally significant correlation was also found between the frontopolar sensitization measure and the ADAMS-Social avoidance score (r = 0.322, p = 0.070). These significant correlations indicated that stronger neural sensitization in the FXS group was associated with greater cognitive-behavioral impairment. No significant correlation was found between the frontopolar sensitization measure and the DAS-II Verbal Ability score (r = −0.216, p = 0.236). Similar correlation analyses conducted in the control group were not significant, and no other correlations reached statistical significance. We also assessed the between-group differences on the correlation coefficients using a series of z-tests. Compared to the control group, the correlation strength in the FXS group was significantly greater for the relationship between neural sensitization levels in frontopolar cortex and BRIEF-2 GEC (p = 0.025), CBCL Anxious/Depressed scores (p = 0.014), and ADAMS Depressed Mood (p = 0.027). No other between-group differences in correlation strength reached statistical significance. We also performed correlation analyses between the significant frontopolar/DLPFC activation in response to different face stimuli (upright face / inverted face) and the above behavioral measures in the FXS group. However, none of these regional activations was significantly correlated with the FXS-linked behavioral measures.

Figure 5.

Correlation between frontopolar sensitization (HbO_{Block5-Block1}) induced by upright face stimuli and various cognitive-behavioral measures, including BRIEF-2-Global executive composite, CBCL-Anxiety, ADAMS-Anxiety, ADAMS-Depression and SRS-2. Solid and dash regression lines were plotted for FXS group and control group, respectively. Correlation coefficients and p values were only shown in the plots for significant correlations in FXS group (Control group: correlation coefficients ranged from −0.264 to 0.079, p values ranged from 0.174 to 0.964. Multiple correlations were controlled by FDR). Note that sensitization in X-axis equals the change of activation between block 5 and block 1 (HbO_{Block5-Block1}), where value of 0 = no habituation, negative value = habituation, positive value = sensitization.

Discussion

This study represents the first investigation using fNIRS, a portable optical imaging technique, to characterize the pattern of neural response to face stimuli in girls with FXS compared with individuals matched for intellectual and cognitive-behavioral features. In terms of task evoked activation, girls with FXS displayed consistent activation in the frontopolar cortex and DLPFC in response to all facial stimuli (upright face + inverted face). Aberrant prefrontal hyper-activation in FXS was also present when compared to the control group. In addition, the temporal habituation analysis revealed stronger sensitization in the frontopolar cortex in FXS in response to upright face stimuli compared to the control group. Finally, post-hoc correlation analyses between neural response to upright face stimuli and multiple cognitive-behavioral measures suggest that aberrant neural sensitization in girls with FXS is related to individual variation in cognitive and behavioral symptoms.

Habituation to repeated stimuli is thought to be an essential component of neural plasticity that reflects more efficient neural processing in typically developing individuals (23). This premise is partially supported by the findings obtained from the control group in our study – a trend of decreased activation in response to repeated face stimuli. Conversely, the lack of a typical habituation response observed in the FXS group indicates aberrant modulation of neural responses to repeated face stimuli. Specifically, significant neural sensitization was seen in the frontopolar cortex (BA 10), the most anterior part of the prefrontal cortex. This finding aligns with our previous fMRI study that identified a significant sensitization phenomenon in the frontal cortex in individuals with FXS compared to the control group (9). Previous studies have also shown that frontopolar cortex serves as an integrative center for higher-order emotional, cognitive, and social processes (24, 25). Therefore, the aberrant neural sensitization observed in the frontopolar cortex of girls with FXS here provides evidence that the neural response to face processing in FXS may be related to atypical activity in frontopolar cortex that is involved in higher-order emotion regulation and social functioning.

The FXS group demonstrated hyper-activation in the frontopolar cortex and DLPFC in response to repeated face stimuli, brain regions that are directly implicated in executive function and social cognition. This finding is consistent with our previous fMRI study that individuals with FXS displayed greater activation in response to direct and averted eye gaze in a prefrontal region known to be involved in executive processes, such as perceptual decision making and modulation of emotional response (7, 26, 27). On the other hand, recent studies have pointed out the essential roles of frontopolar cortex and DLPFC in conflict adaption and resolution during successive trials (28, 29). That is, performance on a trial is flexibly modulated by preceding trial congruency in tasks that require cognitive control. In the present study, all task blocks were presented to the participants in a pseudorandom order and required a relatively quick response. It is possible that the switch of successive but incongruent blocks, such as from “upright face” block to “inverted face” block, may evoke the hyper-activation at the frontopolar cortex and DLPFC in individuals to regulate and adapt the conflict between the preceding and current block. Whether conflict adaptation is distinctly associated with the known emotional and social problems in girls with FXS remains to be further elucidated.

Brain activation and aberrant neural sensitization in regions noted for involvement in social and emotion processing in the FXS group (frontopolar cortex and DLPFC) were observed reliably in response to face stimuli, especially upright faces, but not to greeble objects. In fact, this finding is expected since it has been suggested that recognizing greeble objects is not essentially related to face processing (30). In particular, greebles are designed to match for all low-level properties of faces but exclude any social information (31). Lack of activation for greebles indicates that habituation deficits do not extend to non-face objects. A previous study also demonstrated that children with autism devoted greater attention to the face stimuli than greeble objects (32). Additionally, inverted faces are an intermediary condition with all the visual information of a upright face but the inverted nature doesn’t necessarily activate the social processing networks in the same way (31). This premise is supported by the findings in our study – non-significant sensitization in response to inverted faces in FXS compared to the control group, though significant brain activation in prefrontal cortex and a trend of temporally increased activation in response to inverted faces were all clearly observed in FXS group. Taken together, it is possible that social-cognitive dysfunction in girls with FXS is specifically sensitive to face processing, and the neural mechanism underlying this processing might be different than the recognition of other common objects.

The pathophysiology of FXS has been well described as arising from mutations of the FMR1 gene. In the past decades, a number of placebo-controlled clinical treatment trials have been carried out to test the pharmacologic effectiveness of targeted interventions in FXS (33–35). Although a recent exploratory Phase 2 clinical trial demonstrated the efficacy of a phosphodiesterase-4D (PDE4D) allosteric inhibitor (BPN14770) in improving cognitive function and behavioral outcomes in 30 male adults with FXS (36), all these trials have failed to show meaningful improvement in primary cognitive or behavioral endpoints in a large FXS cohort. This is likely due, in part, to difficulty in choosing a sensitive primary endpoint for evaluating treatment effects (37). To date, cognitive and behavioral endpoints have been typically chosen as primary outcomes. However, cognitive and behavioral symptoms in individuals with FXS are complicated and dynamic as they develop progressively in association with education, work, and life experiences. Such measures may not be effective in readily observing pharmacological effects on neural function or plasticity.

A recent clinical trial study from our group showed that abnormal functional activation induced by gaze contact could be ameliorated by pharmacological intervention before cognitive or behavioral changes were observed (38). This finding indicates that neural biomarkers associated with cognition and behavior might represent a more sensitive and intermediate outcome measure for evaluating pharmacological treatments in FXS. In the present study, when correlated with multiple social-behavioral measures, the neural sensitization measure at the frontopolar cortex in response to upright face stimuli revealed significant correlations with executive function (BRIEF-2-GEC), autistic symptoms (SRS-2), depression (ADAMS) and anxiety (CBCL, ADAMS) in the FXS group. This finding suggests that aberrant neural processing of faces in girls with FXS may be associated with multiple aspects of the variable cognitive and behavioral phenotype (e.g., autistic symptoms, anxiety, and executive function). Together with previous fMRI studies from our group, neural biomarkers derived from neuroimaging may hold promise for use as intermediate phenotypic endpoints for use in future clinical trials. However, additional studies will be required to more accurately illustrate the relationships between brain imaging measures and various FXS-linked symptoms since improving cognitive and behavioral symptoms associated with this important condition remains the primary goal of clinical trials.

Apart from the potential of being used as a syndrome-specific outcome measure for clinical trials, we can also infer that aberrant neural sensitization observed in FXS is likely not solely due to general social-cognitive ability or autistic symptoms because our control group was matched to the FXS group on these parameters (Table 1). These findings strongly support the premise that neurobiological mechanisms underlying face processing in girls with FXS are distinguishable from those underlying similar symptoms in autism and social communication disorders. Therefore, aberrant neural sensitization specifically related to FXS may serve as an important reference when designing disease-specific treatments for FXS. More importantly, compared to the rigorous scan environment of fMRI, the accessibility and low cost of fNIRS makes this neural-linked measure more applicable for use in routine examination of children, particularly those with developmental and behavioral problems. However, more studies should be conducted to validate the test-retest reliability of aberrant neural sensitization in FXS before it can be adapted to advance a precision medicine approach or be considered as an intermediate phenotypic endpoint to evaluate pharmacological efficacy in clinical trials.

Several limitations of this study should be acknowledged. Although we found consistent functional activation and aberrant sensitization at certain brain regions, and neural responses were significantly correlated with various cognitive-behavioral rating scores, our findings may not generalize to males with FXS and participants from different diversity profiles. Future studies with a larger sample size and enhanced ethnic diversity for each sex will be important to clarify the brain-behavior-social deficits association in FXS. Additionally, it should be noted that behavioral assessments in girls with FXS can be complicated by limitations in accuracy of self-report, and atypical symptom manifestation due to developmental disability. Particularly, limited expression ability and social impairment in patients with FXS can prevent reliable communication of symptoms, necessitating reliance on parent/caregiver reporting, which may be biased or limited by recall inaccuracy. Moreover, it remains an open question as to what analysis strategy will be more appropriate to define neural sensitization and/or identify brain-imaging biomarkers for assessing specific symptoms in FXS. Exploring different analysis strategies, such as functional connectivity analysis, may be a future research direction. As well, in this study we did not test the age- and IQ-matched girls in the control group for the presence of the genetic mutation associated with FXS. However, it is very unlikely that any girls in the control group had (unidentified) FXS as a previous study from our group showed that the prevalence of the FXS mutation in girls with mean IQ and range similar to the control group was extremely low (39). Finally, we were able to employ only partial brain coverage for fNIRS measurement rather than a whole-head coverage, which may have hindered a more comprehensive understanding of neural activation in response to face stimuli in FXS. With the availability of new equipment (NIH S10OD026925), we plan to enlarge fNIRS brain coverage in future studies. It should also be acknowledged that the inherent depth limitation of fNIRS imposes constraints for measurement of activity at subcortical and inferior cortical areas, such as fusiform and amygdala; this prevented us from replicating findings from previous fMRI studies.