Fragile X syndrome (FXS) is caused by the expansion of a CGG repeat in the 5′ untranslated region of the FMR1 gene located at Xq27.3 (for review see 1). In its normal form, FMR1 contains 5 to 55 CGG repeats. In the presence of over 200 repeats there is extensive methylation of the CpG island in the gene’s promoter region, resulting in silencing of FMR1 expression. Premutation alleles with 55 to 200 repeats are unstable and may expand to full mutation alleles when transmitted maternally. Fragile X mental retardation protein (FMRP), encoded by FMR1, is an RNA-binding protein that travels from the nucleus to the cytoplasm, where it regulates synaptic and cytoskeleton-associated proteins 2. In the absence of FMRP, excessive and dysregulated mRNA translation leads to altered synaptic function and loss of protein synthesis-dependent plasticity 1. FMRP can also regulate mRNA transport in dendrites and is important for axonal guidance and formation of neural circuits.

The cognitive, behavioral and morphological manifestations of FXS are highly variable 3. Individuals with FXS often display mild dysmorphic features, with long face, prominent ears, arched palate and macroorchidism. They can also exhibit a large spectrum of neuropsychiatric phenotypes, spanning from severe intellectual disability (ID) to mild learning problems or emotional dysregulation, especially in heterozygous women 3. Autistic behaviors, including impairments in social interaction, social anxiety, gaze avoidance, sensory hypersensitivity, stereotypic movements and behaviors, poor motor coordination, delayed speech development and echolalia, are also observed frequently in individuals with FXS 4. FXS is the most common monogenic cause of autism spectrum disorders (ASD), identified in approximately 2% of cases 56. Around 40% to 60% of male and 20% of female patients with FXS meet criteria for ASD 789. Men with FXS often display both ID and autistic features, with those exhibiting autistic traits being more impaired 89101112131415. In contrast, women are protected by the normal allele on the second X chromosome and usually exhibit a less severe phenotype. Although approximately 25% to 30% have ID (intelligence quotient (IQ) < 70), women with FXS usually have an IQ in the borderline to low normal range (75 to 90); most, however, exhibit neurocognitive dysfunctions and anxiety as well as emotional and attention deficits 1617. The variability of FXS-related phenotype in women is influenced by the level of residual FMRP expression and X inactivation skewing. Skewing that favors the mutant allele increases the clinical impairment, whereas skewing favoring the normal allele decreases the likelihood of symptoms 1819. In contrast with the frequent claim that cognitive impairment accounts for the co-morbidity between FXS and autism 132021, here we describe the case of two sisters, one with high-functioning autism and the other with pervasive developmental disorder not otherwise specified (PDD-NOS), both with low normal IQ. Both carried a full mutation in the FMR1 gene with random X inactivation.

Women who are heterozygous for the full FMR1 mutation are less cognitively impaired than men, with about one-third presenting with ID and the rest exhibiting IQs in the borderline to low normal range 111617. Although the majority of women with FXS are not intellectually disabled, they often manifest a wide range of psychiatric and cognitive impairments, including anxious disorders, attention deficit-hyperactivity disorders, ASD and executive deficits 2425. In women as well as in men with FXS, the literature reports a strong association between ID and autistic features, with individuals with a lower IQ being more likely to be diagnosed with autism 89101214152026. A negative correlation between IQ scores and the intensity of autistic symptoms, as measured with the ADOS, was observed in men and women 1315. The level of cognitive impairment is a major predictor of autistic behavior in FXS, stronger than FMRP levels in lymphocytes 1327. These findings would seem to support the suggestion that the occurrence of autistic features in individuals with genetic syndromes and ID is largely the consequence of the cognitive deficit, which decreases their compensatory capacity 28. Here, we describe the case of two sisters carrying a full mutation of FMR1 with high-functioning ASD, suggesting that autistic features can emerge independently from ID, at least in women. It should be noted, however, that both our patients have an IQ that is low relative to estimates based on familial intelligence, as shown previously in women with FXS 29. In agreement with our findings, a previous study found a lack of correlation between autistic behaviors and IQ among girls with FXS 30. Thus, fragile X mutations can manifest in a wide variety of clinical diagnoses crossing categorical boundaries, specifically ID and ASD. The mechanisms involved in the determinism of specific trajectories remain to be elucidated and could be influenced by genetic, epigenetic and environmental processes 26.

The prevalence of women with high-functioning ASD and FXS is probably underestimated because FXS testing is not systematically included in the etiological screening of patients with high-functioning ASD. In the absence of a family history of ID, premature ovarian failure or Parkinsonism, the search for FMR1 abnormalities is usually not recommended in patients with high-functioning ASD (American Academy of Pediatrics, 31). Shyness or social anxiety, commonly mentioned in studies of women with FXS, could be, at least in some cases, misinterpretations of symptoms belonging to the autism spectrum. Also, mild manifestations of ASD are not evaluated or are misevaluated in most reports by the use of inadequate clinical instruments to explore social and communicative deficits in patients with FXS. Thus, if ASD were indeed underestimated in high-functioning patients with FXS, one could suggest that ID and ASD could be phenotypic variants, spanning from ID without ASD to ASD without ID. Specifically in women with FXS, the significant heterogeneity reported for cognitive impairment ranging from severely impaired to normal IQ, or even high IQ in rare cases 32, could also be true for autistic symptoms, with some individuals presenting independence between both dimensions, as shown for many other genetic disorders implicated in the etiology of ASD 33. In particular, many X-linked ID genes have been associated with a profound phenotypic heterogeneity and can be disrupted in individuals without ID. For example, NLGN4X mutations are involved in a wide spectrum of phenotypes, ranging from mild isolated ID without communication deficits to Asperger syndrome with normal intelligence (for review see 34).

The phenotypic severity of full mutations in women is widely assumed to be determined by X inactivation 1819. However, in our patients, the proportion of active normal alleles was approximately at equilibrium (0.55) in both cases. This contrasts with previous findings, in which women who are phenotypically normal have a skewed X inactivation pattern in favor of the inactive X carrying the mutation 1819. Additional mechanisms may underlie the variable expressivity observed in FXS. First, the individual genetic background may modulate the FXS phenotype. In our proband, however, the search for rare CNVs did not reveal mutations in other genes that could explain the development of autistic features independently of ID. Second, variations in the FMRP level resulting from mosaicism of the FMR1 repeat expansion or incomplete FMR1 promoter methylation 1335 can also impact the phenotypic variability observed in patients. For example, there are several reports of male patients with FXS with learning disability but not ID, with partial FMRP expression arising from transcription of unmethylated alleles 35, including a male with Asperger syndrome 36. In the sisters reported here, the mutated alleles were fully methylated. Finally, the partial block in 21-dehydrogenase reported in Patient 1 could potentially explain why she was more severely affected than her younger sister. Girls with congenital adrenal hyperplasia have more autistic traits than their unaffected sisters, suggesting that prenatal exposure to high levels of testosterone increases the vulnerability to ASD 37.

Our report highlights the necessity for further investigation of the association between ASD and FXS in larger samples of high-functioning female individuals with ASD, which could contribute to the understanding of the complexity of the relationship between ID and ASD. In FXS as well as in many other genetic disorders, patients have been shown to exhibit autistic symptoms spanning from mild to severe, associated or not with ID 33. This report stresses the importance of clinicians being aware of the association between a full mutation of FMR1 and high-functioning ASD in women and, vice versa, the importance of a careful psychiatric examination of women with FXS.

ADI-R: Autism Diagnostic Interview-Revised; ADOS: Autism Diagnostic Observation Schedule; AR: Androgen receptor; ASD: Autism spectrum disorder; CNV: copy number variation; FMRP: Fragile X mental retardation protein; FXS: Fragile X syndrome; FXTAS: Fragile X-associated tremor/ataxia syndrome; ID: Intellectual disability; IQ: Intellectual quotient; PCR: Polymerase chain reaction; PDD-NOS: Pervasive developmental disorder not otherwise specified; WAIS: Wechsler Adult Intelligence Scale.

The authors declare they have no competing interests.

PC, MGB, AB and ML helped to collect the data. CB and RD interpreted the results, drafted the manuscript and were, with ML and TB, the main investigators. SK and SD performed the molecular experiments. All authors read and approved the final manuscript.