A Brazilian cohort of individuals with Phelan-McDermid syndrome: genotype-phenotype correlation and identification of an atypical case

Thirty-four individuals with clinical and genetic diagnosis of PMS, members of the “Associação dos Amigos e Familiares da Síndrome de Phelan-McDermid - Brasil” (https://​www.​phelanmcdermidbr​asil.​com/​), were included in this study. This cohort consisted of 16 males (47%) and 18 females (53%), with age ranging from 1.8 to 19.6 years old (mean = 7.54, SD = 3.86). Parents or guardians provided genetic exams and completed a standardized medical history questionnaire. The questionnaire included queries about the individuals’ development, comorbidities, autism diagnosis, and genetic tests. In addition to that, 27 families also provided pictures of their affected children.

Two other independent cohorts were included in this study in order to estimate the frequency of 22q13.3 deletions in Brazilian ASD or ID individuals. An ASD cohort of 829 individuals [656 (79.1%) males and 173 (20.9%) females] was tested by multiplex ligation-dependent probe amplification (MLPA) and ascertained at CEGH-CEL (Centro de Pesquisas sobre o Genoma Humano e Células Tronco, Instituto de Biociências (IB), USP) between 2009 and 2017, with a diagnosis or diagnostic hypothesis of ASD. The other cohort consisted of 2297 individuals, mostly under 18 years old and were referred to chromosomal microarray testing for presenting ID, either associated or not with other clinical signs. These samples were tested by the Laboratório de Pesquisa em Genética Humana from the Departamento de Genética e Biologia Evolutiva, IB, USP, from 2004 to 2018.

Parents or guardians of individuals from both cohorts signed a consent form. The study was approved by the Brazilian National Research Ethics Commission.

Every copy number variant (CNV) was characterized by type (deletion or duplication), size, gene content (based on RefSeq gene data available on the UCSC genome browser, using the hg19), allele population frequency (based on the Database of Genomic Variation Database - DGV), literature, and clinical information about disease-associated genes and inheritance (whenever possible). This information together with the International Standard Cytogenomic Array (ISCA) and the American College of Medical Genetic (ACMG) standards [14] were taken into account to classify the CNVs into pathogenic, benign, or variant of uncertain clinical significance (VOUS).

In total, 18 out of 34 individuals were evaluated by a specialist (neurologist, psychiatrist, or psychologist) for ASD. In addition to medical diagnosis, parents were asked to answer about autistic-like behavior or ASD, which considered the following autism symptoms: no-eye contact, repetitive behavior, lack of social interaction, difficulty in changing routines. The questionnaire presented four possible answers: (1) No, he/she has no autism symptoms; (2) No, a detailed assessment was made and autism was ruled out; (3) Yes, autism was diagnosed in an evaluation with a neurologist/psychologist/psychiatrist; (4) Yes, there has been no diagnosis but we know from his/her behavior.

All individuals carry deletions affecting SHANK3 (Table 1). Deletion size was estimated in 29 individuals tested by CMA; they carried terminal deletions with a mean size of 4.1 Mb (ranging from 0.049 to 9.1 Mb) (Fig. 1 and Table 1). Two of them showed mosaicism: one with an 8.3 Mb deletion (P4) and the other with an 8.8 Mb deletion (P25). Parental genetic testing was available for nine individuals and all deletions were de novo.

We also identified additional rare CNVs (n = 10) located in genomic regions other than 22q13.3 in six individuals (Table 2). Three out of six individuals carried terminal duplications varying in size from 0.4 to 14.9 Mb, thus suggesting that the genomic imbalances resulted from the presence of a translocation derivative chromosome. The remaining three individuals had interstitial CNVs varying in size from 0.03 to 14.8 Mb. All these additional CNVs encompass OMIM genes. Of these, two are associated with genomic alterations that clinically overlap PMS: 12q24 duplication (P4) and chromosome 16p13.3 duplication syndrome (P31); therefore, they were classified as pathogenic. Moreover, one CNV (P28) affects a gene associated with autosomal dominant non-insulin-dependent diabetes mellitus. It was not possible to establish whether the remaining CNVs added to the clinical variability of these individuals, since they encompass regions with genes that were either not previously associated with disorders or associated with recessive disorders; therefore, they were classified as variants of unknown clinical significance (VOUS).

Several dysmorphisms were evaluated in 26 individuals (Additional file 1: Table S1). The most common morphological features (frequency > 70%) were ear abnormalities, sparse eyebrows, large/fleshy hands, and fine fingers. Regarding comorbidities, the most commonly reported were increased pain tolerance and hypotonia, occurring in 80% and 85% of the individuals, respectively (Table 3, Additional file 1: Table S2). In addition, 60% of individuals had a history of recurrent upper respiratory tract infections (Table 3). Other common conditions (frequency > 50%) were gastroesophageal reflux disease and chewing difficulties (Table 3).

All except one individual included in this study have neurodevelopmental delay. We evaluated language and motor development based on the information the parents reported. For language status, we considered only individuals above 3 years old, who were classified into four different classes: speech absence, unintelligible speech (babbling), speaks a few words, and speak sentences. Eighty-four percent (21/25) of children showed unintelligible or absent speech, and only three children (3/25; 12%) speak sentences (P14, P30, and P32) and were respectively 9, 11, and 19 years old at the time of evaluation. Regarding motor development, 75% (25/33) of them walk, with a mean age of onset at 2.27 years old (SD = 0.84). The remaining eight individuals did not walk at the time of evaluation (aged from 1.8 to 10).

Furthermore, individual P32 has a phenotype that differs from that of the other PMS individuals and is described in detail below.

Individual P32 is a 20-year-old male diagnosed with PMS at the age of 18, when he failed a psychometric test when applying for a driver’s license. His karyotype was normal. Array-CGH analysis revealed the deletion of 110 kb in one allele in 22q13.3, that partially encompasses SHANK3 (from exon 10 toward its 3′ end) and two other genes (ACR and RABL2B). The MLPA test in his parents confirmed that the CNV in P32 individual is a de novo mutation (Additional file 1: Figure S1). He is the only child of healthy non-consanguineous parents; her mother gave birth at the age of 26 by vaginal delivery at 40 weeks of gestation (Apgar scores 8/9, weight 3650 g, length 52 cm). Early development was characterized by neonatal hypotonia, and he had head control at 8 months of age. He started speaking at the age of 1 and walked when he was 1 year and 8 months old. Literacy occurred on a regular basis and he attended regular elementary through high school programs, without the need for additional assistance. At the age of 20, Wechsler Adult Intelligence Scale showed verbal, performance, and full scale IQ at 93, 99, and 96, respectively. The individual presents deficits in attention and in processing speed, in addition to difficulty in adjusting to social norms. He also presents mild phonemic impairment and some autistic-like features, such as difficulties related to making and keeping friendships, social and emotional responsiveness, processing figured speech, sensory sensitivity, and the insistence on sticking to routine. In turn, executive functions, such as logical reasoning, abstraction, and operational memory, are preserved. Currently, he is in the seventh semester of studies toward getting his bachelor’s degree in Computer Sciences at a private university. Clinical evaluation at the age of 20 showed dysmorphic features, such as elongated face, hypertelorism, prognathism, small and slightly bulbous nose, and arched palate. Doppler echocardiography, electroneurography, magnetic resonance imaging (RM) and electroencephalogram (EEG), abdominal and renal ultrasound, and audiological evaluation were performed between 2015 and 2018 and showed normal results.

Furthermore, 73.5% (25/34) of individuals presented autism. ASD was diagnosed by a specialist in 18 of them (ASD frequency: 13/18 or 72.2%), while the remaining 16 had the ASD diagnosis based on the questionnaire (ASD frequency: 12/16 or 75%)

Twenty-nine individuals with CMA results were included for genotype-phenotype correlation analysis. A cluster hierarchical analysis was performed in order to determine similarities among groups of PMS individuals using comorbidity data and resulted in four groups that mainly differ in frequency and type of comorbidities (Fig. 2; Additional file 1: Table S3). In this analysis, the degree of severity of each comorbidity was not considered. A positive relationship between the groups and the median of deletion size of their members was observed, but with a borderline P value (P = 0.05, Kruskal–Wallis test). Groups 2 and 4 presented the largest and smallest median deletion size, respectively. Group 4 consisted of individuals who were characterized by the absence of some comorbidities, such as lymphedema, renal abnormalities, gastroesophageal reflux, strabismus and sleep disturbance, which are clinical features that are present in some individuals from group 2. Once we tested the correlation between each comorbidity individually with the deletion size (Mann–Whitney test), only renal abnormalities (P value = 0.011) and lymphedema (P value = 0.0480) had a significant correlation (Additional file 1: Table S4). The two individuals with additional pathogenic CNVs (P4 and P31) were clustered in the expected group according to their 22q13 deletion size, that is, P4 (deletion of 8.3 Mb) clustered to group 2 (largest deletion) and P31 (0.9 Mb) to group 4 (smallest median deletion).

We also tested the correlation of deletion size with language status and autistic features. The only three individuals that were able to speak full sentences (P14, P30, and P32) were in the group with the smallest mean deletion size (group 4), thus suggesting a correlation with deletion size, which was statistically confirmed (Kruskal–Wallis, P = 0.04). Finally, we did not find positive correlation between ASD and deletion size (Mann Whitney, P = 0.38), nor between groups from the clustering analysis and ASD frequency (Chi-square = 0.2651).

In order to verify the smallest regions in 22q13.3 that were more likely associated with renal abnormalities, we analyzed 14 individuals from the literature [9, 13, 14] with deletions in the 22q13.3 region detected by CMA who presented renal abnormalities (Additional file 1: Figure S2 and Table S5). The smallest deletion found was 1.3 Mb, as described by Tabet et al. (2017). We also determined the penetrance of renal abnormalities in individuals with deletions encompassing four candidate genes for renal abnormalities: ZBED4, CELSR1, FBLN1, and UPK3A, by using data from 101 individuals (72 individuals from the literature in addition to our 29 individuals; Additional file 1: Table S6). The penetrance observed for alterations encompassing only ZBED4, the closest gene to SHANK3, was 14.7%, while for the alterations encompassing ZBED4 and CELSR1 was 36.4%, and for alterations encompassing all four genes was 40.0%.