The postnatal stage is a significantly dynamic brain growth and development period characterized by both macrostructural and microstructural changes. Macrostructural changes include global/regional brain volume and weight modifications as well as changes in the regional gray matter (GM)/white matter (WM) ratio, and microstructural changes mainly involve cell differentiation, myelination, synaptogenesis, and cortical layering (1, 2). As a severe neurological disease, Rett syndrome (RTT) leads to neurodevelopmental abnormalities in this crucial period, primarily arising from de novo mutations in the X-chromosomal methyl-CpG binding protein gene 2 (MECP2) (3–5). RTT is manifested as a severe cognitive, social, and physical impairment. Neuroimaging has been used extensively to assess the brain structure, connectivity, and function, providing a valuable tool to link neuronal activity, anatomic structure, cerebral function, and several complex clinical events (6). The application of neuroimaging to RTT has been explored for decades. Various imaging techniques have been used to answer the fundamental questions about the biological basis of RTT and to characterize in vivo disease pathology. Neuroimaging studies on RTT provide essential insights into anatomical, functional, metabolic, and dynamic changes in the brain reflecting the biological effects of MECP2 mutations and allow to monitor future therapeutic outcomes. The present article reviews the current applications of multimodal neuroimaging to RTT and provides potential future directions in this field.

A formal literature review was conducted on PubMed (https://pubmed.ncbi.nlm.nih.gov/) with the following search terms: (“Rett syndrome” OR “MECP2”) AND (“magnetic resonance imaging” OR “positron emission tomography” OR “single-photon emission tomography” OR “diffusion MRI” OR “diffusion tensor imaging” OR “diffusion kurtosis imaging” OR “neurite orientation dispersion and density imaging” OR “magnetic resonance spectroscopy” OR “cerebral blood flow” OR “arterial spin labeling” OR “MRI” OR “PET” OR “SPECT” OR “MRS” OR “DTI” OR “DKI” OR “NODDI” OR “ASL”). All searches were updated in October 2021. The retrieved results were then filtered according to the inclusion criteria reported as follows.

The study inclusion criteria required articles of full-text publications in English or translated into English. The research subjects of eligible articles were RTT subjects with MECP2 mutations. We set the primary focus of our review on the studies applying multimodal neuroimaging to RTT with MECP2 mutations, so both human and animal studies were included. Human studies are summarized in Supplementary Table S1, and animal studies are summarized in Supplementary Table S2.

Case reports and conference communications were excluded. Also, non-English articles and studies on MECP2 duplication syndrome or RTT caused by the mutations of genes other than MECP2 were not considered.

Rett syndrome is a severe neurodevelopmental disorder named after the Austrian pediatrician Andreas Rett, who first described this disorder in 1966 (7). RTT almost exclusively affects women, and ~1 in 10,000–15,000 girls have this disease (8). In RTT, normal development is usually observed up to 7–18 months, followed by the developmental regression in which previously acquired skills are lost (4, 9). Most patients progress through four clinical stages and show various clinical features. The diagnostic criteria and clinical stages for RTT are presented in Table 1 (10–13).

More than 90% of classical RTT arises from de novo mutations in MECP2 (3–5). As a transcriptional regulator, MECP2 is expressed widely throughout the whole body and particularly in the mature neurons of the brain (14, 15). It plays a crucial role in neuronal development, differentiation, and synaptic plasticity (16–18). It is becoming clear that MECP2 is almost unexpressed in early embryonic stages in mice and humans, and its expression gradually increases in the postnatal stage and childhood (19, 20). Researchers have identified hundreds of different MECP2 mutations contributing to distinct clinical phenotypes and disease severities (18, 21, 22). There has been a consensus in eight “hotspot” MECP2 mutations in RTT (R106W, R133C, T158M, R168X, R255X, R270X, R294X, and R306C), which affect >60% of the documented cases (23). Extensive sample statistics found that patients with R133C, R294X, R306C, or T158M manifest with milder phenotypes, while the cases with R106W, R168X, R255X, or R270X, or large deletions show more severe disease forms (24, 25). In addition, several studies on MECP2 deletions have found that specific neuronal populations are involved in distinct pathophysiological mechanisms leading to different clinical phenotypes (26–35). These details are described in Supplementary Table S3.

Although some of the disease mechanisms have been unraveled, genetic heterogeneity, genotype-phenotype interplays, and epigenetic factors in RTT are still not fully understood. A few previous studies have reported that a broad spectrum of disabilities in girls with RTT reflects the pervasive abnormalities of brain growth (especially the developmental phase of intense synaptogenesis) and connectivity (the formation of neural signaling pathways) (36, 37). However, the aforementioned findings were almost based on postmortem brain tissue analyses of patients with RTT, while in vivo monitoring of the underlying pathophysiological processes is more helpful to understand the development of RTT disease. Although there is still no cure for RTT and the available treatments are mainly symptomatic (Supplementary Table S3), alleviating symptoms, reducing pain and discomfort, and increasing the quality of life are essential for both patients and caregivers (38). Early identification is a prerequisite for the implementation of targeted and timely therapeutic approaches. A few studies have shown that early intervention can delay the developmental regression in girls with RTT, and the treatment of symptoms can alleviate the associated pain. Thus, searching for neurofunctional markers for early diagnosis, discovering potential targets for effective therapies, and exploring efficacy evaluation indicators are pivotal.

The addition of MRI to the diagnostic armamentarium of neurodevelopmental disorders may be considered as a revolution in this field. Of all the three-dimensional imaging techniques, MRI has the best soft-tissue contrast. Moreover, different MRI sequences can reflect corresponding tissue contrasts, thus providing a wealth of information about the brain structure and tissue microstructure (39, 40). MRI has been identified as an effective morphological imaging method for detecting structural abnormalities in RTT due to its high resolution and its optimized GM/WM delineation (41).

Rett syndrome is characterized by acquired microcephaly (42). Accordingly, brain MRI studies on patients with RTT have shown global brain atrophy and specific regional GM/WM reductions in the frontal and temporal lobe, hippocampus, caudate nucleus, striatum, thalamus, midbrain, and WM tracts (43–45). Animal studies have been further conducted to explore the relationship between macroscopic brain structure changes and RTT phenotype. Saywell et al. investigated MECP2-null mice, a widely recognized experimental model of RTT, using conventional brain MRI and found a global reduction in its brain size, a feature constantly observed in patients with RTT (46). Reduced cerebellum size may account for some of the neurological signs observed in RTT, including cognition and motor coordination impairments. Moreover, the authors found significant thinning of some specific structures, such as the motor cortex and the corpus callosum (46). Elsewhere, Allemang-Grand et al. used a high-resolution MRI with deformation-based morphometric approaches to examine the brain structure of different mouse models carrying MECP2 mutations, demonstrating the severity of the mutation and the stage of behavioral impairment were associated with the degree of neuroanatomical changes (47).

Although qualitative analysis is helpful for clinical decision-making, it does not provide quantitative values to monitor the developmental status (48). In this context, it is noteworthy that only a few quantitative studies on the brain morphology related to RTT have been carried out. In the field of animal studies, Patrick et al. created an MRI atlas for detailed cerebellar volume analysis and quantitatively investigated genetic effects on this structure (49). Using this approach, the authors were able to reveal a complex interplay between MECP2 mutations, cerebellar volumetric changes, repetitive behaviors, and learning (49). In the field of human studies, Carter et al. used complementary semiautomated Talairach- and voxel-based approaches to study MRI scans acquired in female cases carrying MECP2 mutations. The authors provided novel pieces of evidence on selective reductions of dorsal parietal GM and the preservation of the occipital cortex in RTT (50), and further reported a correlation between anterior frontal lobe reduction and clinical severity. As regards cortical WM, mild and diffuse reductions have been previously reported and linked to axonal pathology (50). Notably, studies in this field have selectively shown decreased volumes of the cerebrum, cerebellum, and caudate nucleus (43, 44, 50–53). Surface-based morphological approaches, including cortical gyrification and regional cortical thickness evaluation, have not been fully explored. Previous studies using quantitative analyses (including surface- and voxel-based measurements) observed no significant differences in global cortical gyrification, thickness, and volumes, as well as in regional cortical thickness between patients with RTT/Rett-like (RTT-l) (cases carrying the MECP2 mutation but not fulfilling the diagnostic criteria for RTT) and normal controls (54), but evidenced a significant volumetric reduction of the cerebellum. Given that the patients with RTT/RTT-l included in this study were younger than those described in other reports, it is plausible to hypothesize that cerebellar volume reductions may precede regional cortical atrophy, providing a potential early diagnostic marker in patients with RTT/RTT-l (54).

In addition to morphological studies, MRI sequences can provide other information about brain changes in patients with RTT; for example, the underlying tissue microstructure can be examined using diffusion MRI (dMRI) (55–57), functional information can be gathered using functional MRI (fMRI) (57), metabolic differences can be identified using magnetic resonance spectroscopy (MRS) (46), and cerebral blood flow (CBF) can be assessed using arterial spin labeling (ASL) (58, 59).

Complex microstructural changes are increasingly recognized as significant contributors to neurodevelopmental disease even in the absence of gross brain morphologic changes (2). dMRI can noninvasively monitor microstructural changes and identify the implicated neural networks by mapping the distribution and movement of water molecules in the brain tissue. Recent studies have shown that it is also sensitive to cortical microstructure properties, such as radial and tangential fiber populations and neuropil volume fraction (60–65). Previous morphologic studies have suggested that the decreased cortical WM volume observed in RTT may be secondary to neuronal somata changes and primary axonal disturbances (6, 66). However, confirming this hypothesis requires the application of complementary imaging approaches such as dMRI. This technique also provides a new perspective for understanding the underlying pathological mechanism, assisting in diagnosis, and evaluating the neurobiological bases of the observed symptoms.

Diffusion tensor imaging (DTI) is an MRI technique based on the physical principle of water molecule diffusion restrictions across WM tracts (67). The tensor can derive several parameters, including fractional anisotropy (FA) and mean diffusivity (MD). FA represents the degree of anisotropy of the diffusion, which is a sensitive imaging biomarker for axonal organization and myelin integrity (2). MD represents the magnitude of diffusion, which is a commonly used scalar measurement.

In patients with RTT, a reduced FA has been identified in left peripheral WM areas (including middle temporal, middle occipital, precuneus, and postcentral regions), left major WM tracts (such as the superior longitudinal fasciculus, sagittal stratum, and corpus callosum), and the cingulum bilaterally (48). By studying the correlation between DTI-derived FA measurements and specific clinical features in patients with RTT, Mahmood et al. obtained the following findings: firstly, FA in the superior longitudinal fasciculus was significantly associated with speaking abilities; secondly, FA reductions in the anterior cingulate gyrus were associated with the characteristic mood and behavioral changes often observed in patients with RTT; and thirdly, the common observation of intact visual capabilities might be in accordance with the normal to increased FA values identified in the posterior corona radiate (68). Taken together, these preliminary studies suggest that DTI may represent a valuable noninvasive technique to assess WM tract pathological processes and add specificity to the assessment of RTT clinical severity.

Diffusion tensor imaging obtains the two types of information: quantitative water diffusion parameters described above and global brain WM organization metrics. The latter includes maps of fiber bundle orientation using color-coded DTI maps and a more precise delineation of specific fiber pathways by using tractography, which is based on identifying tracts using the color maps (67, 69). Fiber bundles are delineated using the software that traces the trajectory of the vectors representing water diffusion. Fiber delineation/reconstruction studies usually focus on specific pathways, which have been postulated to be involved in the disease. To date, tractography technology has been the most widely applied technology in animal studies. Wang et al. reported abnormal brain WM developmental dynamics and network topological organizations in RTT monkey models across different clinical stages via longitudinal DTI (70). They revealed that the MECP2 mutation could lead to early protracted WM myelinization affecting later synaptic pruning and inducing abnormal functional segregation of the brain in RTT (70). Early abnormal WM development may be the underlying neural mechanism for some of the significant abnormal clinical neurobehavioral phenotypes, and it may also serve as an early predictor of RTT (70).

However, the major limitation of DTI quantitative parameters is that FA and MD provide nonspecific measures of pathology. The assumption of a single compartment with Gaussian diffusion does not adequately model the involved biological systems, like the WM, with its complex fiber architectures, or the GM, where diffusion is relatively isotropic, and this model fits poorly (71). A variety of more advanced models of tissue diffusion providing alternative parameters are now available, such as diffusion kurtosis imaging (DKI) or neurite orientation dispersion and density imaging (NODDI). These have the potential to identify previously unseen structural abnormalities and improve our understanding of underlying microstructural changes. However, no DKI or NODDI studies on RTT have been reported so far.

The most widely used imaging methods of cerebral blood flow evaluation are single-photon emission CT (SPECT) or PET imaging (72–74), and the most consistently reported finding in this field is the presence of frontal hypoperfusion. Nielsen et al. studied seven patients with RTT with ¹³³Xe SPECT scans and found that global CBF was decreased and the hypoperfusion foci were located mainly in prefrontal and temporoparietal areas (72). Burroni et al. performed ^99mTc-ECD brain SPECT imaging on 12 girls with classical RTT and a control group of normal children, also attaining similar findings (74). These observations align with the neuropathological evidence of a global reduction in brain size and an alteration in the dendritic and synaptic trees in RTT. Moreover, no significant right-to-left asymmetry was found in any regions of interest of the cortex, which confirms that RTT is a diffuse and nonfocal neurological disorder (74). Brain perfusion abnormalities were seen more often in stage IV patients with RTT rather than in stage III patients with RTT, supporting the notion that CBF reductions probably reflect clinical disease progression (74, 75). Researchers have also reported a reduced CBF in patients with normal brain MRI scans, suggesting that SPECT may be able to reveal functional alterations before the occurrence of appreciable morphological or structural damage (74). Lappalainen et al. performed perfusion SPECT and electroencephalography (EEG) on 13 patients with RTT and found that frontal hypoperfusion was consistent with frontal paroxysmal activity on EEG, and these two alterations were associated with disease progression (73). Finally, Naidu et al. performed PET studies using ¹⁵O-labeled water and observed a decreased blood flow in the frontal regions, similar to the observations made by Yoshikawa et al. in patients with RTT (6, 76).

Although SPECT and PET are excellent methods for measuring CBF, it is difficult to justify their use in young children who were given the radioactivity. However, recent developments in novel MRI approaches provide an excellent alternative to PET for measuring CBF. Naidu et al. used a transfer-insensitive labeling technique to measure CBF based on the concept of ASL (58), revealing greater hypoperfusion in the frontal lobe compared to other brain regions (6).

PET is helpful not only for brain blood perfusion but also for brain functional imaging. Due to its associated radiation exposure, the use of PET to assess pediatric neurodevelopmental disorders has been greatly limited. Therefore, PET has a vast unexplored potential to improve our understanding of the pathophysiology of neurodevelopmental disorders in children (77). PET biomarkers can be applied to disease diagnosis, clinical progression monitoring, and treatment response evaluation.

While there is still no cure for RTT, the fundamental research discoveries achieved over the past few decades have enabled to set the basis for the development of new potential therapies. Therapeutic approaches for RTT are divided into the following three categories: symptom treatment, pharmacological modulators of downstream MECP2 targets, and genetic interventions (95). Therefore, it is necessary to search for neurofunctional markers to track drug safety and treatment response. A longitudinal brain MRI study of RTT has shown that MRI may reveal the efficacy of treatment interventions on the neuroanatomy, particularly across the critical neural networks that govern classical RTT symptoms (96).

Furthermore, early diagnosis and concerted rehabilitation efforts will be essential for improving the efficacy of therapies for RTT. Meanwhile, as RTT is characterized by complex clinical symptoms progressing through the different stages over time and varying from one individual to another, clinicians need novel measures that can reflect multilevel changes at several levels (95). Being at the interception between etiology, clinical diagnosis, and treatment, neuroimaging applications to RTT need to be further developed. In this review, we have summarized a few imaging literature studies on RTT with MECP2 mutations and compared various imaging modalities to clarify their strengths and weaknesses (Table 2). As only a few studies have been conducted in this series of patients, this field of research should still be considered in its early stages. We, therefore, believe that there is still ample space for further neuroimaging studies on RTT, taking into account the following research priorities.

First, we must improve research on MRI morphologic imaging of patients with RTT by comparing the identified characteristics to those of the normal development population and summarizing the imaging characteristics of abnormal brain morphologic development in patients with RTT with different disease stages and phenotypes (97). Second, NODDI technology will be used to explore the imaging characteristics of brain microstructural changes in patients with RTT at different stages and phenotypes. It is expected that NODDI technology will play a significant role in the early diagnosis of this disease and the evaluation of therapeutic efficacy. Third, basic studies have discovered that MECP2 gene mutations lead to abnormalities in many downstream neurons and related nerve signaling pathways, but the detailed mechanism is still not fully clarified, and it is necessary to develop specific neuroimaging methods/sequences or multimodal imaging to dynamically observe the abovementioned changes in the brain in vivo. Pharmacological modulators of downstream MECP2 targets are being developed, and neuroimaging will play an essential role in future patient-specific drug selection and drug efficacy evaluation. Fourth, multiple MRI modalities (multimodal MRI/MRS approach, fusion imaging with ASL and dMRI, or fusion imaging with MRI and PET) and various learning algorithms, like the combination of NODDI and surface-based analyses, have been designed to provide personalized data. Machine-learning methods, such as deep learning-based segmentation of brain tissues from dMRI, have been proposed and achieved a high degree of accuracy (98), which will further apply to RTT. Lastly, exploring combinations with nonimaging biomarkers and further identifying those biomarkers' longitudinal trajectories and orders will point to the most potential combinations (99). Calabrese et al. used diffusion tensor magnetic resonance histology to track microstructural changes in the rat brain throughout normal postnatal neurodevelopment and then correlated these changes with the changes in the cytoarchitecture. They also provided a comprehensive database of image sets as a foundation for future studies (2). Consequently, a combination of gene-neuroimaging-pathophysiology and clinical phenotype analyses can effectively characterize disease states in the RTT population (Figure 1).

In this review, we have attempted to summarize the findings from the conducted MRI to PET studies over the past few decades on RTT with MECP2 mutations. MRI morphologic imaging is particularly sensitive to brain structural abnormalities in patients with RTT, while dMRI provides valuable information on brain microstructural changes and fMRI enables elucidating the underlying dysfunctional mechanisms. Finally, MRS and PET modalities play a crucial role in the exploration of metabolic alterations in this complex neurodevelopmental disease. We emphasize that the field has not flourished in the area of pediatric disorders compared to adult neurodegenerative disorders. There also remains an enormous opportunity to improve our understanding of RTT through molecular imaging with MRI and PET technology. These advances will be of great significance for the clinical diagnosis of RTT and the formulation of individualized treatment plans.

YK: guarantor of integrity of the entire study, study concepts and design, and manuscript preparation. NC, ND, Q-bL, and X-fJ: literature research. G-qZ and Z-hY: manuscript editing. All authors contributed to the article and approved the submitted version.

This work was supported by the PhD Research Foundation of the Affiliated Hospital of Jining Medical University (Grant No. 2016-BS-016).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.