Brief Report: Regression Timing and Associated Features in MECP2 Duplication Syndrome

Abstract

The aim of this study was to determine the frequency, timing, and associated features of developmental regression in MECP2 duplication syndrome. We also examined whether duplication size was associated with regression. Comprehensive psychological evaluations were used to assess 17 boys with MECP2 duplication syndrome. Information about regression was gathered via parent report. Eight of 17 boys exhibited regression in language skills, while seven of 17 exhibited regression in other skill areas. Regression in “other skill” areas coincided with seizure onset and with a prior autism diagnosis in six of seven participants. Regression was not associated with duplication size. Questions remain as to why some boys regress, and future work is necessary to understand the underlying mechanism(s) that causes regression.

Introduction

Mutations in the methyl CpG-binding protein 2 gene (MECP2) were first identified as a primary cause of Rett syndrome (RTT) in females (Amir 1999). The more recently described MECP2 duplication syndrome is the most common subtelomeric duplication identified in a clinical population surveyed by high resolution human genome analyses using array Comparative Genomic Hybridization (aCGH). This duplication syndrome primarily affects males, and is an X-linked genomic disorder (Lupski 2009) characterized by infantile hypotonia, absent to limited speech, severe intellectual disability, progressive neurological difficulties (including spasticity and seizures), recurrent respiratory infections, and gastrointestinal difficulties (Friez et al. 2006; Ramocki et al. 2009; Van Esch 2012). Phenotypic features are due to a MECP2 gene dosage effect as further evidenced by a more severe phenotype conveyed by triplication of this locus (Carvalho et al. 2011). Features of autism spectrum disorder (ASD), including repetitive behaviors, sensory processing difficulties, and difficulties with social affect are present in a large proportion of affected individuals (Peters et al. 2013; Ramocki et al. 2009). Females are less often affected, and the spectrum of phenotypes ranges from mild to the same severity as affected boys depending on the underlying genetic factors (Bijlsma et al. 2012; Grasshoff et al. 2011). Developmental regression, inclusive of a partial or complete loss of acquired highest spoken language skills (including babbling), and a loss of motor skills are major diagnostic criteria for classic Rett Syndrome (RTT) (Neul et al. 2010). Because RTT is rarely diagnosed prior to the regression phase, studies of regression are based primarily on retrospective parental report (Neul et al. 2010). The phenomenon of regression is less clear within MECP2 duplication syndrome. Prior studies based on retrospective parental report reveal that regression can occur within MECP2 duplication syndrome (del Gaudio et al. 2006; Ramocki et al. 2009), but the frequency, timing, and associated features of regression have not been formally characterized.

Regression is typically defined as a partial or complete loss of skills related to language, socialization, self-help skills, and/or motor skills. In ASD and RTT, the loss of skills occurs at an early age (usually between 15 and 30 months), and may or may not be associated with seizure onset or medical illness (Barger et al. 2012; Neul et al. 2010; Ozonoff et al. 2008). In some situations, a child may exhibit developmental delays prior to a loss of skills (early onset pattern), or no delays may be apparent before the loss (Ozonoff et al. 2008). A recent study based on video analysis found that approximately 38.5 % of children with ASD exhibited a regressive course (Ozonoff et al. 2011). In ASD, studies suggest that regression occurs within the context of intact motor abilities and even advanced motor milestones, suggesting a different pattern and course from that which occurs in RTT and MECP2 duplication syndrome (Bernabei et al. 2007; Jones and Campbell 2010). In MECP2 duplication syndrome, all boys have early onset developmental delays and there are some reports of seizures and associated regression, occurring at varying ages (i.e. late childhood to early adolescence) and with varying percentages (Vignoli et al. 2012). A prior study of MECP2 duplication syndrome revealed that 3 out of 9 boys exhibited regression/loss of skills at the time of evaluation (Ramocki et al. 2009). In one of these cases, the onset of regression occurred at a younger age (36 months). In the two other cases, however, a loss of skills occurred during late childhood and coincided with the onset of seizures.

It is unclear whether regression is a core phenotypic feature of MECP2 duplication syndrome that occurs secondary to overexpression of MECP2 directly and with variable phenotypic expression, or whether regression only occurs in a small proportion of individuals due to factors indirectly related to MECP2 overexpression. We have not been able to reliably predict when or in which children regression will occur, and it is unknown as to whether duplication size predicts regression. A better understanding of regression in this patient population is essential in order to facilitate prognosis and to help families and physicians make the best long-term treatment plans and choices for affected individuals. It is also important to inform future preclinical and clinical therapeutic trials. In this study, we examined a larger cohort of 17 boys with MECP2 duplication syndrome. Our goal was to determine the average age of regression, the skills that were lost (language, social, self-help, or motor skills), whether regression coincides with seizures and/or medical illness, whether losses are gradual or sudden, and whether a prior diagnosis of ASD predicts a later decline in skills. We also examined whether or not duplication size corresponded to a loss of skills.

Methods

Participants

The participants (n = 17) were all male and ranged between the ages of 3–10 years. Females were not included given that their phenotype is more variable (Grasshoff et al. 2011). All participants with a confirmed genetic diagnosis were offered enrollment and were ascertained as part of their attendance at the first ever worldwide conference on MECP2 duplication syndrome. All were additionally evaluated by a pediatric clinical neurologist (M.B.R.) prior to study enrollment. This study was approved by the Institutional Review Boards at Vanderbilt University and Baylor College of Medicine, and all families gave informed consent to participate.

Genetic Testing

All participants had an established clinical diagnosis. To independently confirm the diagnosis and to determine the size, genomic extent and gene content for each rearrangement, we designed a tiling-path oligonucleotide microarray spanning 4.6 MB across the MECP2 region on Xq28. The custom 4 × 44 k Agilent Technologies (Santa Clara, CA, USA) microarray was designed using the Agilent e-array website (http://earray.chem.agilent.com/earray/). We selected 22,000 probes covering Chromosome X: 150,000,000–154,600,000 (NCBI build 36), including the MECP2 gene, which represents an average distribution of 1 probe per 250 bp. Probe labeling and hybridization were performed as previously described (Carvalho et al. 2011). To estimate duplication size, we calculated the distance between breakpoints/coordinates (i.e. where probe signals traversed from no gain to gain, signaling increased dosage of the genomic interval complementary to the interrogating oligonucleotide probe).

Instruments

To assess regression, we utilized the Autism Diagnostic Interview—Revised (ADI-R) questions regarding regression. A portion of the ADI-R contains questions about early milestones, symptom onset, and regression. Since the ADI-R questions regarding language-regression require the minimum use of at least five different words (other than mama and dada) on a daily basis for at least 3 months, a research-based regression interview was designed as an extension of the ADI-R to capture sub-threshold losses of skills as separated into language (e.g. loss of cooing, babbling, vocal imitation, use of 3 or more words, communicative intent), and non-language skills (hand use, motor skills, self-help skills, social interest, and curiosity) (Luyster et al. 2005; Parr et al. 2011). This interview utilizes the same coding and similar questions to the ADI-R, and is being utilized as part of the (ASD) Simons Simplex Collection (SSC). Information regarding the age of the loss, duration of the loss, as well as any association with physical illness is captured. There are also questions regarding any recovery of skills after the loss (or whether the skills loss persists).

Participants also received a comprehensive developmental evaluation. The Vineland Adaptive Behavior Scales—Second Edition (VABS-II) was completed as a semi-structured interview with parents to assess overall adaptive behavior. Visual reception/nonverbal cognitive abilities and fine motor skills (nonverbal cognitive abilities), gross motor skills, and expressive/receptive language abilities were assessed using the Bayley Scales of Infant and Toddler Development-Third Edition (BSID-III). This instrument was selected given the developmental level of our participants. For our participants who were mobile (n = 9), we also completed the Autism Diagnostic Observation Schedule, Module 1 to assess Social Affect (inclusive of communication and reciprocal social interaction) and Restricted/Repetitive behaviours related to ASD. Our assessments were based on the recently revised ADOS algorithms (Gotham et al. 2007) which include the domains of Social Affect and Restricted-Repetitive Behaviors. All participants (regardless of their chronological age) received Module 1, which is appropriate for individuals who are nonverbal. In order to assess skills such as joint attention, it is important that the participant have some mobility. Even in cases where we could not perform the ADOS due to loss of mobility (n = 8), we conducted detailed interviews (i.e. the ADI-R, clinical interview) to determine whether or not there was a prior diagnosis of ASD. We detail the results for part of our sample (n = 9) on these measures elsewhere (Peters et al. 2013).

Statistical Analysis

For the purposes of all analyses, language regression was analyzed separately from regression in other skill areas. We utilized descriptive statistics to examine the average age (mean, median) and age range of language and non-language regression. Because of the small sample size, non-parametric statistics (Mann–Whitney U test) were used to compare any differences between regressors versus non-regressors for chronological age. To check whether an autism diagnosis was related to regression status (language and non-language), we used contingency analysis. To determine the relationship between duplication size and regression (separated into language vs. non-language), we used bivariate correlations.

Results

Table 1 lists the number of boys who exhibited language regression versus no language regression, and then regression in “other skill” areas versus no loss in other skill areas. Six boys exhibited losses in both language and non-language areas. Two boys exhibited language losses alone (without any losses to date in other skill areas). One boy exhibited losses in other skill areas, but his parents noted that he never attained enough language (e.g. even cooing or babbling skills) to exhibit a loss in any language skills. Six of eight boys who exhibited a language regression had a prior ASD diagnosis, but five of nine boys without language regression also had a prior ASD diagnosis and hence there was no significant association (Contingency Coefficient = .20; p = .40). Six of seven boys with regression in “other skill” areas had a prior ASD diagnosis, while five of ten without regression in other skill areas had a prior ASD diagnosis (Contingency Coefficient = .34; p = .12). There are no significant differences in chronological age between language regressors and non-regressors (p = .37), or between boys who exhibit regression in other skills versus no regression in other skills (p = .23).

Table 1.

Language regression, regression in other skills, and prior autism diagnosis

	Number (n)	Prior autism diagnosis
No Language regression	9	5/9
Language regression	8	6/8
No regression in “other skills”	10	5/10
Regression in “other skills”	7	6/7

Table 2 characterizes the skills that were lost amongst the boys who did exhibit regression. For language skills, it was most common for boys to exhibit losses in lower level skills such as babbling, cooing, and the meaning/intent of communication. Two of the eight boys with language regression alone (without losses in motor skills) were able, with intensive therapy, to recover language skills at the level that they functioned at prior to the loss. The most common losses in “other skill” areas included loss of motor skills (related to posture, gait, and coordination), curiosity, and hand use. One boy with a regression in “other skill” areas re-learned to walk (but he still maintained a loss of language skills), but all others had not recovered skills that they lost.

Table 2.

Number of participants showing a loss of skills by domain

Cooing	Babbling	Vocal imitation	Use of 3 or more words	Communicative intent	Hand use	Motor skills (posture, gait, coordination)	Self-help skills	Social engagement	Curiosity
5/8	7/8	4/8	2/8	6/8	4/7	7/7	4/7	3/7	5/7

Table 3 shows the average age of skills loss, the age range of regression onset, and whether or not losses were associated with seizures. In the majority of cases (6/7), regression in “other skill” areas was associated with seizure onset. In five of these six cases, the participants had been given a formal diagnosis of “intractable epilepsy” by a pediatric neurologist (based on EEG findings and clinical evaluation). In the sixth case, the parent described several “staring spells,” although a formal diagnosis of epilepsy has not been given. Confirmed seizure onset in our patients ranged between 36 and 84 months of age; all but one had a seizure onset of age 5 (60 months) and later. In five of six cases, the age of onset for language regression coincided with the age of onset for regression in other skill areas. In one case, the age of language regression occurred at 24 months, but he did not begin to lose motor skills until 84 months of age, which coincided with the onset of intractable seizures (absence, myoclonic, and atonic seizures). Not surprisingly, boys who exhibited regression in “other skill” areas (n = 7) were significantly lower functioning on all of our developmental parameters (p range = .003–.023) as compared to boys who had not regressed (See Table 4). This was primarily attributable to a loss of mobility (i.e. walking) and hand use, which hindered their ability to interact with manipulatives. Most of these participants previously had some level of language use (at least babbling vowel-consonant sounds), were able to walk, engage in functional play, and had developed some self-help skills (e.g. feeding themselves) prior to regression and the onset of intractable epilepsy.

Table 3.

Average age of regression

	Average age of loss (mean, SD)	Range of ages for losses (months)	Loss associated with seizures
Language regression	39.12 months; 13.58 months	24–60	4/8
Regression in other skills	54.57 months; 20.82 months	30–84	6/7

Table 4.

Comparison of developmental scores between regression and non-regression groups

	Cognitive skills	Receptive language	Expressive language	Fine motor	Gross motor
Regression (n = 7)	8.00	4.50	7.00	10.00	11.00
Non-regression (n = 10)	15.00	10.00	10.00	15.00	18.00

Data are depicted in age equivalents (months) and are shown as the Median for each group

The results of genetic analyses reveal that all subjects duplicate the MECP2 gene; three boys (BAB 2769, BAB3255, BAB3259) have partial genomic triplications. Duplications (and partial triplications) ranged in size between .31 and 4.7 MB (See Fig. 1). There was no relationship between duplication size and language regression (p = .26) or between duplication size and regression in “other skill” areas (p = .41). We also did not find any significant relationships between duplication size and refractory epilepsy, and one participant with refractory epilepsy and significant regression actually had the smallest duplication size of all of our participants (BAB2625).

Fig. 1.

Genomic region harboring duplications in our cohort of 17 patients analyzed by oligonucleotide array CGH. Solid red bars represent duplications and blue bars represent triplicated regions. Genomic coordinates are given relative to NCBI Build 36 for the X chromosome. *Duplication is larger than the resolution of our custom aCGH (>4.7 MB)

Discussion

In this study, we examined in greater detail regression timing and related features in MECP2 duplication syndrome. We found that 9 of 17 boys exhibited a loss of language and/or motor skills, with six boys exhibiting a loss across both areas. The majority of boys who lost language skills lost subthreshold skills such as cooing and babbling. It is notable that on the original ADI-R, most of our participants would not have met the threshold criteria for a loss of language skills (i.e. use of at least five different words other than mama or dada for a period of at least 3 months). This extension of the ADI-R may therefore have some utility in capturing subthreshold losses in lower functioning populations (Parr et al. 2011; Richler et al. 2006). When examining the timing and course of regression in MECP2 duplication syndrome, it appears that in comparison to ASD and to RTT, the onset of skills loss is later. In addition, the regression in MECP2 duplication syndrome may be distinct from that which occurs within ASD. When within the context of refractory epilepsy in particular, the losses of skills in our participants are dramatic (e.g. loss of mobility, head control, self-help skills, and/or language, and hand use), and, in some cases, with little recovery of functioning. The size of the duplications in our subset of participants with refractory epilepsy was quite variable and in fact, our participant with the smallest duplication size exhibited a significant regression within the context of early onset seizures at 36 months. More broadly, our results demonstrate that loss of skills in MECP2 duplication syndrome is not related to duplication size/gene content. It is unclear whether regression associated with MECP2 duplication syndrome therefore represents: (1) consequences of an epileptic encephalopathy (interestingly 3/5 of the boys with intractable epilepsy were clinically diagnosed with Lennox-Gastaut syndrome), (2) whether regression phenotypes and epilepsy simply represent common downstream effects of MeCP2 overproduction on neurons or neural networks in associated brain regions, (3) whether a secondary cause such as genetic background or CNS infection and/or CNS inflammation contributes to both regression and epilepsy, or (4) whether the clinical outcome represents a complex combination of these possibilities. The timing of seizure onset and correlation with behavioral phenotypes has not been studied in mouse models, but may offer further clues as to etiology/mechanisms and treatment.

While over 50 % of our sample exhibited a loss of skills at some point during development (for a period of at least 6 months), it is notable that two boys with language regression alone were able to recover functioning with the assistance of intensive speech/language therapy. Further studies will be required in order to examine whether the majority of boys with MECP2 duplication syndrome will/will not recover any function post-regression. At a minimum, our results suggest the need for year-round therapeutic services, constant repetition of skills and concepts, and within the context of regression, retraining of lost skills. In cases where there has been loss of hand-use and mobility, there may be utility in using alternative methods of communication, such as eye-gaze technology as is being utilized in RTT (Djukic et al. 2012). In comparison to previous studies (Ramocki et al. 2009), our results reflect a higher percentage of boys with MECP2 duplication syndrome exhibiting a loss of skills. This may still reflect an underestimate of true level of regression since our sample only included boys between the ages of 3–10. More specifically, future studies should include older individuals as there are reports of refractory epilepsy, neurological declines, and sudden death in boys who are in later adolescence-early adulthood (Van Esch 2012). Regardless of age, given our findings and those of another recent study (Vignoli et al. 2012), we do suggest that if a patient without epilepsy presents with regression, then a careful history to assess for subtle seizures and an EEG study is indicated.

A limitation of our study is that we relied on parental reports of regression as opposed to retrospective or even prospective video studies. As has been demonstrated in studies of individuals with ASD, there may be some benefit to using video studies, since retrospective parent reports do not always correspond to what is noted on video analysis (particularly as related to social communication skills) (Ozonoff et al. 2011). More specifically, this study of ASD participants demonstrated that less than half of the participants whose home videos displayed clear evidence of a decline in social-communication were reported to have had a regression by parents. Similarly, less than half of the participants were reported by their parents as having early onset delays in spite of clear evidence of social communication delays from video analysis. As applied to MECP2 duplication syndrome, future studies should incorporate retrospective video analysis in cases where a genetic diagnosis occurs after a regression. In cases where genetic diagnosis occurs prior to any loss of skills, we propose longitudinal, prospective developmental assessments that incorporate objective video analysis across settings (clinic and home) to document the earliest detection of skill loss and timing and may offer further clues for intervention and treatment.

A recent study demonstrated that duplication of the MECP2 gene causes immune dysfunction due in part to the suppression of IFN-γ production from T helper cells in both mice and humans (Yang et al. 2012). Although not examined within this study, future studies should examine whether insidious infection or neuro-inflammation contribute to regression and epilepsy in affected patients. The seizures in the subset of boys reported here are exceptionally refractory to medical treatment, and in fact 2/5 boys have also failed the ketogenic diet, and 2/5 have failed vagus nerve stimulation. These data suggest that the mechanism of epileptogenesis in this syndrome may be distinct from the pathways currently targeted by available epilepsy medications; and the question remains whether or not excellent seizure control would prevent or reverse regression in affected patients. Furthermore, if underlying infection or inflammation has a role, then could targeted treatment potentially prevent epilepsy, regression, and perhaps partially improve other neurological phenotypes and overall quality of life? Further studies in mouse models and in human patients are needed to dissect the underlying pathogenesis of disease in affected patients and to develop treatments that decrease both morbidity and mortality. Therapeutics that specifically decrease gene expression or levels of the MeCP2 protein should theoretically improve or potentially reverse some clinical phenotypes.