Article Text

Download PDFPDF

Original article
Association between deletion size and important phenotypes expands the genomic region of interest in Phelan–McDermid syndrome (22q13 deletion syndrome)
  1. Sara M Sarasua1,2,
  2. Alka Dwivedi1,
  3. Luigi Boccuto1,
  4. Jonathan D Rollins1,
  5. Chin-Fu Chen1,2,
  6. R Curtis Rogers1,2,
  7. Katy Phelan3,
  8. Barbara R DuPont1,2,
  9. Julianne S Collins1,2
  1. 1Greenwood Genetic Center, Greenwood, South Carolina, USA
  2. 2Department of Genetics and Biochemistry, Clemson University, Clemson, South Carolina, USA
  3. 3Hayward Genetics Center, Department of Pediatrics, Tulane University School of Medicine, New Orleans, Louisiana, USA
  1. Correspondence to Sara M Sarasua, Office of Bioinformatics and Epidemiology, Greenwood Genetic Center, 101 Gregor Mendel Circle, Greenwood, South Carolina, 29646, USA; smsaras{at}


Background The clinical features of Phelan–McDermid syndrome (also known as 22q13 deletion syndrome) are highly variable and include hypotonia, speech and other developmental delays, autistic traits and mildly dysmorphic features. Patient deletion sizes are also highly variable, prompting this genotype–phenotype association study.

Methods Terminal deletion breakpoints were identified for 71 individuals in a patient cohort using a custom-designed high-resolution oligonucleotide array comparative genomic hybridisation platform with a resolution of 100 bp.

Results Patient deletion sizes were highly variable, ranging from 0.22 to 9.22 Mb, and no common breakpoint was observed. SHANK3, the major candidate gene for the neurologic features of the syndrome, was deleted in all cases. Sixteen features (neonatal hypotonia, neonatal hyporeflexia, neonatal feeding problems, speech/language delay, delayed age at crawling, delayed age at walking, severity of developmental delay, male genital anomalies, dysplastic toenails, large or fleshy hands, macrocephaly, tall stature, facial asymmetry, full brow, atypical reflexes and dolichocephaly) were found to be significantly associated with larger deletion sizes, suggesting the role of additional genes or regulatory regions proximal to SHANK3. Individuals with autism spectrum disorders (ASDs) were found to have smaller deletion sizes (median deletion size of 3.39 Mb) than those without ASDs (median deletion size 6.03 Mb, p=0.0144). This may reflect the difficulty in diagnosing ASDs in individuals with severe developmental delay.

Conclusions This genotype–phenotype analysis explains some of the phenotypic variability in the syndrome and identifies new genomic regions with a high likelihood for causing important developmental phenotypes such as speech delay.

  • Phelan–McDermid syndrome
  • 22q13 deletion syndrome
  • genotype–phenotype
  • autism
  • speech and language
  • neurosciences
  • copy-number
  • cytogenetics
  • developmental
  • genetic epidemiology
  • cytogenetics
  • immunology (including allergy)
  • other endocrinology
  • drugs: endocrine system
  • molecular genetics
  • metabolic disorders
  • diabetes
  • obesity, genetics
  • microarray
  • neurosciences
View Full Text

Statistics from


Phelan–McDermid syndrome (PMS (MIM 606232)), also known as 22q13 deletion syndrome, is a rare syndrome characterised by developmental delay, absent or impaired speech, neonatal hypotonia, autistic traits and mild dysmorphic features.1–8 Affected individuals have deletions ranging in size from 100 kb to over 9 Mb.5 Because PMS is considered to be underdiagnosed, the true prevalence is unknown.1 In an evaluation of more than 11 000 cases with developmental disabilities, 22q was the second most frequent subtelomeric rearrangement, identified in 0.2% of those evaluated.9 Simple terminal deletions account for approximately 75% of PMS cases.10 The other cases have been comprised of translocations in the 22q13 region,2 11–13 ring chromosome 221 2 14–16 and mosaics.17 18

Until recently, diagnostic technologies relied upon cytogenetic banding and fluorescence in situ hybridisation or bacterial artificial chromosome comparative genomic hybridisation (CGH), which are not always able to detect smaller deletions or to accurately measure deletion size or breakpoints. Oligo-array CGH allows for a much higher resolution of the chromosomal breakpoints.

Most of the published work suggests that the loss of one copy of SHANK3 (SH3 and multiple ankyrin repeat domains 3, also referred to as ProSAP2 or proline-rich synapse-associated protein 2) is responsible for the neurological features of the PMS phenotype, since SHANK3 maps within the region of common deletion observed in patients.5 6 A patient with t(12;22)(q24.1;q13.3), which disrupted SHANK3, gave evidence implicating this gene as a major candidate for the neurological features of the syndrome.11 Further, a de novo interstitial deletion disrupting only SHANK3 was observed in an individual with developmental delay, speech delay and mildly dysmorphic features including ptosis, epicanthal folds and cupped ears.19 The authors concluded that haploinsufficiency of SHANK3 alone, and not genes telomeric to it, was responsible for PMS.

SHANK3 is predominantly expressed in brain tissue, and the expression appears to be regulated by tissue-specific methylation.20 21 SHANK3, a structural protein, is considered to be critical in the assembly, maintenance and plasticity of the postsynaptic density (PSD) at excitatory synapses in the brain.22 SHANK3, along with other PSD components, including cell adhesion molecules such as neurexins23–26 and neuroligins,27 28 as well as scaffolding proteins, has been found to be associated with autism spectrum disorders (ASDs).29–31 However, a recent finding of two individuals with PMS phenotypes and interstitial deletions outside of the SHANK3 region has expanded the search for additional genetic causes of the 22q13 deletion phenotype.32

Another gene located near SHANK3, and deleted in most cases of PMS, is IB2. IB2 (islet brain 2), also known as MAPK8IP2 (mitogen-activated protein kinase 8 interacting protein 2), is 70 kb proximal to SHANK3 and performs critical neurologic functions.33 Giza et al demonstrated that the IB2 protein is located in the PSD and throughout the brain and is important in synaptic transmission and neural morphology. A full knockout mouse model of IB2 demonstrated reduced cognitive ability, learning and social interaction.33

This analysis, the first genotype–phenotype comparison to use high-resolution deletion breakpoint mapping on a large sample size, addressed the hypothesis that additional genes or regions of chromosome 22q13 besides SHANK3 contribute to the PMS phenotype.

Subjects and methods


Study subjects were previously diagnosed as having PMS, and most attended one or more PMS Family Support Conferences held in 2001, 2004, 2006 and 2008 in Greenville, South Carolina, and in 2006 in Melbourne, Australia. The majority of blood specimens were collected at the 2006 and 2008 conferences in Greenville, South Carolina. The study was approved by the Institutional Review Board of Self Regional Healthcare (Greenwood, South Carolina), and all participants' parents or guardians provided signed informed consent forms. All participants have a terminal deletion encompassing the SHANK3 gene. Individuals with known or self-reported chromosomal anomalies other than terminal deletions were not included in the analysis in order to focus the study on 22q13 deletion effects. Some participants in the present study may have participated in other studies published elsewhere.

Information on physical features was obtained from physical examinations conducted by experienced clinical geneticists following standardised assessment checklists or, for two patients, abstracted from a medical record. Stature and head circumference were measured at the Family Conferences, and the remaining physical features were evaluated based upon clinical judgement. These physician-confirmed features are listed in supplemental table 1. Medical history was obtained from standardised medical history questionnaires administered during an in-person interview at the Family Conferences or completed independently by parents and mailed or emailed to the investigators. When available, this parent-provided information was supplemented with information obtained from the physical examinations or from the abstracted medical record. These features are presented in supplemental table 2. For 72% of participants, more than one record source was available (physical examination, medical history questionnaire or evaluation of the same individual across several years). In instances where a discrepancy between records was identified, physician-provided answers were used in place of parent-provided information. In cases where health information differed between different years of participation for an individual, positive responses were used such that the information represents ‘ever’ reporting a feature. This information was entered in a Microsoft Access (Redmond, Washington, USA) database and checked for accuracy.

Genetic analysis

Genetic deletions were measured using a custom 4×44K 60-mer oligo-array designed to cover chromosome 22q12.3-qter by Oxford Gene Technology (Oxford, UK). In brief, genomic DNA was isolated and subsequently purified using the Zymo DNA Clean & Concentrator kit (ZymoResearch, Orange, California, USA) according to manufacturer's instructions. Reference DNA used in the comparative hybridisation was obtained from Promega (Madison, Wisconsin, USA). DNA concentration and purity were determined with an ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, Delaware, USA). Two μg of DNA from patients' peripheral blood and reference samples were digested at 37°C for 2 h with 5 U of RsaI and AluI (Promega). After heat inactivation of the enzymes, the samples were labelled with either Cy3- or Cy5-dUTP using the Agilent Genomic DNA labelling kit PLUS (Santa Clara, California, USA). Samples were purified with YM-30 Microcon filters from Millipore (Bedford, Massachusetts, USA). Hybridisation and washes were conducted employing Oxford Gene Technology's (OGT) CytoSure Chromosome 22q specific array protocol (Oxford, UK). Arrays were scanned with the GenePix 4000B scanner (Molecular Devices, Sunnyvale, California, USA). Array feature extraction was performed with GenePix Pro 6.1. Copy number/data analysis was performed with OGT's CytoSure/Oligome viewer software package. Deletion breakpoints are accurate to 100 bp resolution, and the array CGH genomic coordinates were established according to the 2006 human genome build 18 (GRCh 36/NCBI build 36.1).34 Deletion sizes were plotted on the genome browser using the University of California at Santa Cruz Genome Browser.35

Statistical analysis

Statistical analyses were performed using SAS V.9.2 (SAS Institute Inc.) to examine the association between deletion size and clinical features. Because the distribution of deletion sizes was non-normal, non-parametric statistical tests including the Spearman rank correlation coefficient and two-sided Wilcoxon rank-sum test were used when the dependent variable was deletion size. In cases where expected cell sizes were small (<5), exact methods were used. Linear regression was used to examine the effect of the deletion size on continuous outcome measures such as speech and developmental features.


Phenotype and deletion breakpoint data from 71 individuals were used in this analysis. Physical examination information was available for 54 (76%), and parent-provided medical history was available for 61 (86%) individuals. A total of 84 clinical phenotypic features were assessed. The age of participants ranged from 0.4 to 40 years, with a mean of 7.6 years (SD of 2.5 years). The cohort was composed of 42 females and 29 males for a female to male ratio of 1.45:1. Median deletion size was similar for males (6.03 Mb) and females (5.24 Mb, p=0.2678). Deletion sizes ranged from 0.22 to 9.22 Mb (see figure 1), with a mean of 5.08 Mb, a median of 5.25 Mb and an SD of 2.56 Mb. SHANK3 was deleted in all individuals, and IB2 was deleted in all but two individuals. Individuals with physical examination data tended to have smaller deletions (mean=4.7 Mb) than those with only medical history information (mean=6.5 Mb, p=0.0102), although the full range of deletion sizes was observed for both groups. The most common features observed in this cohort include developmental delay, expressive speech/language delay, long eyelashes, increased pain tolerance, dysplastic toenails and hypotonia.

Figure 1

Distribution of deletion sizes among 71 patients with Phelan–McDermid syndrome. (A) Ideogram of chromosome 22q13.2-qter. (B) Horizontal bars represent deleted regions, sorted by deletion size. Terminal deletions ranged from 0.2 to 9.2 Mb in size and covered chromosome regions 22q13.2–22q13.3. Figures produced using the UCSC genome browser.35

Sixteen of 84 characteristics were found to be significantly associated with larger deletion sizes (tables 1 and 2; a list of all assessed features is provided in supplemental tables 1 and 2). These characteristics tended to be related to physical features including dolichocephaly, facial asymmetry, full (prominent) brow, large or fleshy hands, dysplastic toenails, tall stature, macrocephaly, atypical reflexes and male genital anomalies; neonatal features including hypotonia, feeding problems and hyporeflexia; and developmental delays including speech delay, developmental delay, later age to crawl and later age to walk. Among those with atypical reflexes, median deletion sizes were similar among those with strong reflexes (n=7, median deletion size 6.46 Mb, range of 1.85–8.96 Mb) and those with weak reflexes (n=9, median deletion size 6.57, range of 1.98–8.20 Mb). Both were significantly larger than those with typical reflexes (median deletion size 4.19 Mb, range 0.34–8.62 Mb). Behavioural features were not associated with increased deletion size. However, two features, ASD and aggressive behaviour, were associated with smaller median deletion sizes.

Table 1

Physical examination phenotypes showing significant differences in 22q13 deletion sizes*

Table 2

Medical history features showing significant differences in 22q13 deletion sizes*

Prenatal and neonatal features

Features present in the neonatal period, including hypotonia, feeding problems and hyporeflexia, were all associated with larger deletion sizes (table 2). No association was found between deletion size and low birth weight (reported for 23% of the sample) or preterm birth (reported for 25% of the sample; supplemental table 2).

Autism spectrum disorders

ASDs were reported in 26% of the patients over age 3 years (table 2). The median deletion size for those with ASDs was smaller (3.39 Mb, range 0.22–7.19 Mb) than the median deletion size for those without ASDs (6.03 Mb, range 0.57–9.22, p=0.0144). Similar to ASDs, aggressive behaviour was also associated with smaller median deletion size (table 2).

Developmental delay

All patients had some degree of developmental delay. Severity of developmental delay, as rated by parents on an ordinal scale from mild, moderate, severe and profound, was significantly and positively associated with deletion size when using linear regression (p=0.009). Similarly, later age to crawl (p=0.0032) and later age to walk (p<0.0001) were significantly associated with larger deletion sizes.

Speech and language delay

Speech was absent or delayed for 100% of individuals. On the questionnaires administered in 2004 and 2006, parents were asked whether speech was absent or severely impaired and to note how many words the patient used. In the questionnaire administered in 2008, parents were asked whether speech was absent or severely impaired and how many words were used in a sentence and to provide additional comments describing speech. Among those over 3 years of age for whom speech information was provided (n=50), half reported no speech. Another 28% had 40 or fewer words and did not report speaking in phrases or sentences. The final 22% reported having sentences or phrases, talking as the primary means of communication or using more than 40 words and were coded as having ‘sentences’ (table 2). Median deletion size was higher for those with absent speech (6.72 Mb) and smaller for those with sentences (3.27 Mb). When deletion size was examined in detail, none of the 22 individuals with deletion sizes greater than 5.3 Mb was reported to speak in sentences, whereas 39% of the 28 individuals with deletion sizes smaller than 5.3 Mb use sentences (p=0.001). Deletion size showed a significant negative correlation with the number of words spoken (p=0.0102).


Growth was found to be non-linearly associated with deletion size (table 1); those with normal stature had the smallest median deletion size at 4.80 Mb, while those with tall stature (>95th percentile) had a median deletion size of 6.18 Mb, and those with short stature (<5th percentile) had the largest median deletion size at 8.06 Mb. Macrocephaly was associated with increased deletion size (median deletion size 6.99 Mb), whereas those with microcephaly had similar deletion sizes compared to those with normocephaly (median deletion size of 3.32 Mb compared to 3.34 Mb, respectively).


This genotype–phenotype study is the largest to date to include high-resolution deletion breakpoint genotype information along with clinical features to identify associations between features of PMS and deletion sizes. Prior smaller studies found mixed indications of association between deletion size and the severity of the phenotype.5 In particular, our findings are consistent with previous reports of correlations between increased deletion size and the severity of selected phenotypes.6 15 36 A case series of eight patients, which included neuroimaging, found that individuals with small deletions (∼0.15 Mb) had decreased brain anomalies when compared to individuals with larger deletions.7 In another study of 12 subjects with oligo-array CGH breakpoint data, patients with large hands tended to have larger deletions,36 in agreement with our findings. In a study of 30 individuals with 22q13 deletions manifesting as ring chromosomes, Jeffries et al found that increased deletion size positively correlated with dysmorphic features related to ears, toenails and philtrum as well as the developmental features of increased severity of developmental delay and speech delay.15 Wilson et al found a positive correlation between increased deletion size and developmental delay, hypotonia, head circumference, ear infections, pointed chin, dental anomalies and several measures of independent behaviour and daily living abilities.6 Our study also found an association of increased deletion size with neonatal hypotonia, head circumference and facial features (tables 1 and 2).

Supporting our finding of larger deletions being associated with more severe phenotypes are reports of three individuals with interstitial deletions overlapping the larger deletion regions in our patients.32 37 Wilson et al report two cases with an intact SHANK3 gene, but presenting with speech delay (two words each and no sentences), macrocephaly, tall stature, hypotonia, delayed walking and developmental delay.32 Fujita et al described an 18-month-old Japanese girl with a del(22)(q13.1q13.2) with hypotonia, psychomotor delays, minor dysmorphic features (including dolichocephaly and full brow) and hearing loss due to inner ear anomalies.37 These cases suggest that there are clinically important genes located proximal to SHANK3.

Speech and language

Most prior studies reporting on the effect of deletion size on speech and language delay do not distinguish between absent speech and delayed or impaired speech. The present study found genomic differences by severity of language delay, particularly the dramatic difference in ability to speak in sentences being present in nearly 40% of those with smaller deletions but absent in those with larger deletions. Previous findings of severe speech impairment among two cases with interstitial deletions32 support the presence of genes affecting speech in the regions proximal to SHANK3. None of the genes in the deletion region was identified as being a transcriptional target of FOXP2, a transcription factor known to be associated with speech.38 Future work is critically needed to distinguish degrees of speech impairment with the genomic regions deleted.

Autism spectrum disorders

In this study, individuals reported by parents to have an ASD had smaller deletions than those without an ASD, but all are missing a copy of SHANK3, a gene implicated in autism.29–31 Approximately 26% of those over age 3 were reported to have an ASD; all were reported to have some degree of developmental delay. The prevalence of ASD in the larger population of patients with PMS is unknown, with published reports ranging from 0% to 94% (0/8 (0%),7 1/6 (17%),39 6/11 (55%),40 3/5 (60%),36 12/27 (44%) or 23/27 (85%) depending on the definition,15 and 17/18 (94%)5). Complicating a comparison are the different diagnostic and reporting criteria used, as well as ages and degree of developmental delay of patients assessed. Our analysis of autism is subject to several limitations. We relied upon parent report of an autism or ASD diagnosis rather than a standardised instrument such as the Autism Diagnostic Interview, Revised,41 which would have improved the validity of this assessment. We found that those with larger deletions tended to be more severely developmentally delayed. It is possible that more severely impaired individuals may have been less likely to have been assessed for autism, that a diagnosis of PMS may have ‘replaced’ or precluded an autism diagnosis, that more severe physical and intellectual disabilities may obscure autistic features or that the assessments may be difficult to administer to or be inappropriate for the more severely impaired.


One of the commonly associated phenotypes of PMS is ‘normal to accelerated growth.’4 5 A recent analysis of the same cohort noted that tall stature (>95th percentile) and short stature (<5th percentile), as well as microcephaly (<3rd percentile) and macrocephaly (>97th percentile), are common in individuals with 22q13 deletion.42 Additionally, the present analysis provides support for the presence of distinct deletion regions associated with short stature, tall stature and macrocephaly.


There are several limitations to the present analyses. A large number of phenotypes were examined with statistical tests and, as these analyses were considered exploratory, were not corrected for multiple testing. Given that 84 phenotypes were examined and using a p value cut-off of <0.05, one might expect four to be identified due to chance, whereas 18 phenotypes were found to be statistically significant (16 associated with larger deletion sizes and 2 with smaller deletion sizes). Additionally, some phenotypes may be correlated with each other. Further, high-resolution karyotype information was not available for all participants. Individuals with known or self-reported chromosomal anomalies other than simple deletions were removed from the analysis. It is still possible that some individuals in the present cohort include patients with r(22), translocations or other anomalies. Nonetheless, prior studies of r(22) noted similar phenotypes to 22q13 deletion,2 15 40 indicating that the ring structure does not alter the phenotype. The proportion of our study group reporting chromosomal anomalies (37/108 or 34%) is similar to those reported elsewhere.1 10 Thus, it is unlikely that our results are significantly confounded by the presence of unaccounted-for chromosomal rearrangements. It should be noted that the use of chromosomal microarrays to detect copy number variants is now recommended as the primary test, over G-banding or fluorescence in situ hybridisation, for individuals with developmental delay.43 Finally, analysis of physical features was restricted to those who had medical records or physical examinations at the family conferences. This restriction was used to most accurately characterise phenotypes. However, it was found that individuals with physical examinations tended to have smaller deletion sizes. It is possible that individuals with larger deletions, and possibly a more severe phenotype, were less likely to travel to participate in the family conferences, and thus, those included in the present analysis group may represent a somewhat less severely affected population. However, the full range of deletion sizes was observed in this group.


This study implicates genomic regions proximal to SHANK3 in language, movement and developmental delay and some dysmorphic features in individuals with PMS. This is the first study of PMS to distinguish different degrees of speech delay with deletion size. These findings are critical as they provide further regions of interest to help elucidate clinical implications for affected individuals. Current investigations of SHANK3, IB2 and other telomeric genes should be supplemented to determine the independent and additive impacts of the additional loss of genes, micro-RNAs or regulatory elements in this region of chromosome 22q13.

Web resources


The authors sadly announce the passing of Dr Julianne Collins and honour her for her significant contributions to the fields of birth defects and developmental disabilities research and genetic epidemiology. The authors gratefully acknowledge the assistance of Kristi Atkinson, Kaitlyn O'Gorman, Alyssa Sattler, Keri Sprouse and Taylor Sartain for data entry and verification. We also appreciate the review and helpful comments provided on the manuscript by Drs Charles E Schwartz and Roger E Stevenson.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Funding Support for this study was provided by the Greenwood Genetic Center, the Phelan–McDermid Syndrome Foundation (SMS), the Genetics Endowment of South Carolina, the South Carolina Birth Defects Foundation and the South Carolina Department of Disabilities and Special Needs.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval The study was conducted with the approval of the Institutional Review Board of Self Regional Healthcare (Greenwood, South Carolina).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.