Article Text

17q21.31 microduplication patients are characterised by behavioural problems and poor social interaction
  1. B Grisart1,
  2. L Willatt2,
  3. A Destrée1,
  4. J-P Fryns3,
  5. K Rack1,
  6. T de Ravel3,
  7. J Rosenfeld4,
  8. J R Vermeesch3,
  9. C Verellen-Dumoulin1,
  10. R Sandford2
  1. 1
    Centre de Génétique Humaine, Institut de Pathologie et de Génétique, Charleroi, Belgium
  2. 2
    East Anglian Medical Genetics Service, Cytogenetics Laboratory, Addenbrooke’s Treatment Centre, Addenbrooke’s Hospital, Cambridge, UK
  3. 3
    Center for Human Genetics, University Hospitals Leuven, Leuven, Belgium
  4. 4
    Signature Genomic Laboratories, LLC, Spokane, Washington, USA
  1. Dr B Grisart, Centre de Génétique Humaine, Institut de Pathologie et de Génétique, 25 Avenue Georges Lemaìtre, B-6041 Charleroi, Belgium; bernard.grisart{at}


Background: Microdeletions at 17q21.31 have recently been shown to cause a novel syndrome. Here we identify the reciprocal 17q21.31 duplication syndrome in 4 patients.

Method: Patients with the 17q21.31 duplication were identified by screening a large cohort of patients (n = 13 070) with mental retardation and congenital malformation by comparative genomic hybridisation microarray. Parental origin was investigated in 3 patients by quantitative polymerase chain reaction and microsatellite genotyping.

Results: In three cases it was possible to show that duplication arose de novo. Intellectual skills range from normal to mild mental retardation. Patients are characterised by poor social interaction, with relationship difficulties, reminiscent of autistic spectrum disorders. Other features are rather variable with no striking common phenotypic features. Parental origin was investigated for 3 patients. In all cases duplication was of maternal origin either through interchromosomal (2 cases) or interchromatid (1 case) rearrangement. The 3 mothers are all carriers of the inverted H2 haplotype, emphasising the role of local genomic architecture alteration as a predisposing factor for this duplication.

Conclusion: Autistic features observed in our patients suggest that genes in the duplicated interval should be considered as candidates for disorders in the autistic spectrum. Other phenotypic observations are rather variable or aspecific. This adds 17q21.31 duplications to a growing group of recently identified genomic disorders with variable penetrance and expressivity.

Statistics from

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.

The widespread use of comparative genomic hybridisation (CGH) microarray has resulted in the identification of many new genomic disorders in patients with mental retardation with, or without, multiple congenital anomalies (MR/MCA) due to recurrent microdeletions. Many of these new syndromes including the 17q21.31 microdeletion were recently reviewed by Slavotinek et al.1 The 17q21.31 microdeletion, with a critical region of 424 kb in size, was the first to be identified by screening large cohorts of patients by CGH microarrays even before the detailed characterisation of the associated clinical features. The clinical picture was recently documented for 22 patients. It included developmental delay, hypotonia, facial dysmorphism and usually a friendly and amiable disposition.2 The microdeletion encompasses six genes: two ORF (open reading frame) and four known genes (MAPT, IMP5, CRHR1 and STH). Three of these are good candidates to explain the neurological disturbance observed in these patients: MAPT (OMIM 157140), IMP5 (OMIM 608284) and CRHR1 (OMIM 122561). The microdeletion breakpoints were recurrently mapped within a low copy repeat.26 Even if there is no formal proof that the underlying mechanism is non-allelic homologous recombination (NAHR), it seems reasonable to postulate that NAHR between two directly oriented repeats is actually involved. Moreover, an inversion within 17q21.31, present in the Western European population at a frequency of 20%,7 was shown to be closely associated with this rearrangement. Inverted haplotypes (H2) differ notably from normal haplotypes (H1) by the orientation of critical repeat sequences. Such an orientation allows NAHR to occur either between H1 and H2 or between two H2 haplotypes.

The frequency of this microdeletion syndrome was estimated to be around 1 in 16 000 live births.2 Given the mechanisms involved, the reciprocal duplication is expected to occur in a ratio of roughly one duplication for two deletions.8 However, to our knowledge only one patient with a 17q21.31 microduplication has been reported to date.9

Here we report four additional patients with a 17q21.31 microduplication with extensive molecular and clinical data.


Patient clinical data

Patient 1

This girl is the first child of unrelated parents (fig 1, table 1). She was born at term after a pregnancy complicated by placental detachment in the first month of gestation. The birth weight was 3.570 kg (P50). She started to walk at 12 months but with a tendency to tiptoe-walking resulting in tendinous retraction. Language onset was normal but she developed difficulties in sentence formation and pronunciation which persist at 6 years of age. She was referred for genetic consultation for behavioural problems, hyperactivity and difficulties with social interaction. The intellectual quotient (IQ) was evaluated to be 105 (WPPSI-R: Wechsler Preschool and Primary Scale of Intelligence-Revised) with a verbal IQ of 108 and performance IQ of 100. The developmental quotient was 91 for fine mobility and 73 for global mobility. Brain magnetic resonance imaging (MRI) was performed with normal results.

Figure 1

Pictures of patient 1 (a) and patient 3 (b). Consents to publish these pictures were obtained from the parents of the patients.

Table 1 List of clinical observations reported for five patients with 17q21.31 duplication

At 6 years, growth parameters were in the normal range: height was at the 90th centile (126 cm) and weight at the 50th centile (23.3 kg). Microcephaly was observed with head circumference at –2SD (49 cm). Some facial dysmorphic features were noticed with synophrys, puffy eyelids, short philtrum, thin upper lip, retrognathia and dysplastic ears. Slight hirsutism was observed on the back. Clinodactyly of the fifth finger, single palmar crease on right hand, and partial bilateral syndactyly between the second and third, and third and fourth, toes was also noted. Hormonal blood measurements (follicle stimulating hormone (FSH), luteinising hormone (LH), dehydroepiandrosterone (DHEA) sulfate, 17-hydroxyprogesterone) were in the normal range. These patient data were submitted in DECIPHER database ( with identification number LEM249002.

Patient 2

This girl is the only child of unrelated parents (table 1). The pregnancy was uncomplicated. Birth weight was 3.5 kg (P50), length 53 cm (P90) and head circumference 34.2 cm (P25–50). Meconium aspiration occurred during delivery which resulted in no apparent sequelae. Feeding and early milestones were normal. She started to walk at the age of 2 years when hypotonia was noticed. At age 3 she demonstrated obsessive behaviour characterised by an adherence to ritual in performing various common tasks. She had difficulty with toilet training. She was very passive, often played by herself and did not demand affection. First clinical evaluation was carried out at age of 6. She had significant behavioural problems at school and home, poor social interaction with peers, disruptive behaviour and occasional aggressive outbursts. Her IQ was evaluated at 78 (WPPSI-R). Verbal skills and vocabulary were normal. She has normal reasoning and comprehension but she has some difficulty following instructional language. Fine motor skills, spatial awareness and non-verbal communication were evaluated as poor. She was very rule driven, preferring routine tasks. At the age of 11 some dysmorphic features were noted including epicanthal folds, large posteriorly rotated ears, short upturned nose, short philtrum, flat midface, high arched palate, prominent upper incisors, low posterior hairline, thick hair, mild truncal obesity and tapering fingers.

The behavioural problems persisted with difficulties maintaining friendships and occasional inappropriate responses to other people. Fine motor skills remain poor and her obsessional behaviour is still problematic. Growth parameters are in the normal range with height and weight at the 91st centile and head circumference between the 50th and 75th centile.

The common 17p11.2 deletion responsible for hereditary neuropathy with liability to pressure palsies (HNPP, OMIM 162500) was also identified but she is asymptomatic for this syndrome. This rearrangement was inherited from her mother. Therefore patient 2 has both an HNPP deletion and a 17q21.31 duplication with a maternal origin for both (see FISH (fluorescence in situ hybridisation) analysis in results section). DECIPHER identification number for this patient is CAM249012.

Patient 3

This boy is the second child of non-consanguineous healthy parents (fig 1, table 1). His sister is healthy. The pregnancy and delivery occurred without complications and he was born at term with a birth weight of 2.890 kg (P3), length of 50 cm (P50) and head circumference of 34 cm (P10). He underwent bilateral inguinal herniorraphy between 10–12 months of age. He also had an orchidectomy due to torsion. His early milestones were considered normal: he could walk independently at 14–16 months and said his first words at 12–14 months of age. He received speech therapy at the age of 3.5 years as he was socially anxious and his verbal communication was poor at nursery school. He also began tiptoe walking and physiotherapy was initiated. Assessment with the WPPSI-R at the age of 4 years 9 months indicated a total IQ of 81 (verbal IQ = 80, performance IQ = 87). His auditory memory, visuomotor interaction and word finding capacities were poor. Due to his socio-communicative difficulties, certain elements of the autism spectrum disorder (ASD) were recognised in him. Investigations showed a normal hearing test, normal MRI of the brain and normal electromyogram (EMG). Clinical examination at the age of 7 years 4 months revealed a height of 127 cm (50th centile), weight of 25.3 kg (50th centile) and head circumference of 50.7 cm (10th centile). He has brachycephaly, an open square facies with a pointed chin. A congenital hypopigmented 3×3 cm mark was found on his left flank and there was no hirsutism. He has normal symmetric reflexes and shortened gastrocnemius muscles (right more than left). DECIPHER identification number for this patient is CHG249083.

Patient 4

This man was born after an uncomplicated pregnancy by caesarean section (table 1). He is the second child of unrelated healthy parents. Birth weight was 5.1 kg (>97th centile) and length 58 cm (>97th centile). Mild hypotonia and nasal congestion was noted but feeding was normal. An inguinal hernia was repaired at age 6 months. All major developmental milestones were delayed. He walked at 27 months and speech was significantly delayed. A muscle biopsy performed for hypotonia was reported as normal. A diagnosis of Asperger’s syndrome was made by a clinical psychologist at age 11 years. The history also included a diagnosis of depression and attention deficit hyperactivity disorder (ADHD) for which he receives long term medication. He required educational support in all subjects and a formal IQ was in the low-normal range. He was also described as a loner and established no close friendships. Eye contact and social interaction were poor. His behaviour was occasionally complicated by rages but these were not associated with self-harm. His height and weight were noted to be <3rd centile but no cause was identified. Puberty was delayed. LH/FSH and testosterone values were all reported as low. A brain MRI was also reported as normal. At age 18, height and head circumference were between 10th–25th centiles and weight remained <3rd centile. He was dysmorphic with thin arched eyebrows, deep set eyes and down slanting palpebral fissures. Mild malar hypoplasia was also noted. The ears were simple and elongated (>+2SD). Mild pectus excavatum, fifth finger clinodactyly, 2–3 toe syndactyly and a sandal gap were also present. Secondary sexual characteristics were absent and he had a slender body habitus with a generalised lack of body fat. The remainder of the physical examination was normal. DECIPHER identification number for this patient is LEM249014.

Microarray analysis

Patients’ DNA were analysed with four different CGH technologies. For patient 1, CGH microarray analysis was performed using a 44 K oligonucleotide microarray (Agilent, Santa Clara, California, USA). Patient 2 was analysed using a 250 K Affymetrix Nsp single nucleotide polymorphism (SNP) array and Nexus software. Patient 3 was investigated using a 1 Mb bacterial artificial chromosome (BAC) array as previously reported10 and confirmed using a whole genome tiling path array. Patient 4 was studied using the Signature Genomics BAC/P1-derived artificial chromosome (PAC) array.

Parental origin of the duplication

Quantitative PCR analysis

Quantitative polymerase chain reaction (PCR) was used to confirm the rearrangement. Primer design and real-time PCR amplification were performed as described by Hoebeeck et al.11 ABgene SYBRgreen mastermix was used according to the manufacturer’s (ABgene, Epsom, UK) recommendations with 250 nM final primers concentration and 10 ng of DNA in a volume of 25 μl. Ct values obtained from PCR amplification of exon 1 of MAPT (GCCGTCTTCCGCCAAGAG, GGAGCCGATCTTGGACTTGACATT) were normalised by two reference genes (primers for GPR15: GGTCCCTGGTGGCCTTAATT and TTGCTGGTAATGGGCACACA and primers for P53: CCCAAGCAATGGATGATTTGA and GAGCTTCATCTGGACCTGGGT).

Microsatellite and haplotype analysis

Microsatellites were used to determine the parental origin of the duplication. Primers are reported in supplemental table 1. Two microsatellites (D17S810 and DG17S920) located outside the duplication were genotyped as well as three microsatellites (VNTR17_1, VNTR17_4 and D17GS142) inside the duplication. PCR were performed using AmpliTaq Gold (Applied Biosystems, Foster City, California, USA) in semi-quantitative conditions by running only 25 cycles of amplification to allow semi-quantitative analysis of the number of inherited alleles, based on the analysis of relative height of alleles observed on the electrophoretic data.

The H2 specific 238 bp deletion12 was genotyped using primers GGAAGACGTTCTCACTGATCTG and AGGAGTCTGGCTTCAGTCTCTC, followed by electrophoresis on a 1% agarose gel.

FISH analysis

BAC probes were used to confirm the rearrangement observed in patient 2. BAC RP11-64B12 at PMP22 locus (17p12) was used in conjunction with RP11-100C5 and RP11-413P22 at the 17q21.31 microduplication locus.


Microarray results

The size of the duplication in the three patients investigated with non-targeted array (patient 1–3) ranged from 585–763 kb. An overview of the breakpoints is given in fig 2 where the duplications are aligned with the recurrent deletion2 and copy number variation (CNV) of region 17q21.31. Exact breakpoint positions are listed in supplemental table 2. The difference in the breakpoints likely reflects the different resolution of the four array types used.

Figure 2

Schematic representation of genomic duplication in Kirschhoff’s patient9 and in the 4 patients presented here. Array used in each case is noted on the right. Duplicated genomic regions are showed as white rectangles. A thin horizontal line on both sides illustrate the break point regions. The exact distal position of the duplication was not available in Kirschhoff’s paper (thin dotted line). The regions of overlap of normal and duplicated bacterial artificial chromosome (BAC) in patient 3 are shown as a thin horizontal line even if these regions do not actually correspond to breakpoints regions. The minimal recurrent deletion reported by Koolen et al2 is depicted as a shaded rectangle. Dotted rectangles illustrates known copy number variations (CNV). Positions of the four genes mentioned in the discussion are depicted as black rectangles. The thin vertical line localise the position of the recurrent proximal breakpoint.2

Parental origin of the duplication

Quantitative PCR analysis

Quantitative PCR was performed on patient and parental DNA to determine the origin of the duplication using a PCR system developed in exon 11 of MAPT. In patients 1 to 3 the duplication appeared de novo (data not shown). De novo occurrence in patient 4 could not be determined due to lack of parental DNA.

Microsatellite and haplotype analysis

To investigate the origin of the duplication, the probands and their parents were genotyped for five microsatellites which were phased in order to identify transmitted haplotypes (supplemental table 1). In all cases the duplication occurred on the maternal chromosome. Detailed analysis of genotypic data shows that the duplication occurred by interchromosomal rearrangement (probably NAHR) between H1 and H2 haplotypes in patient 1 and 2 whereas it occurs by interchromatid rearrangement (probably NAHR) between two H2 haplotypes in patient 3. An illustration of electrophoretic data obtained from a capillary sequencer, as well as detailed interpretation of these microsatellite data, is shown in supplemental fig 1. An H2 specific deletion of 238 bp was also genotyped by PCR (supplemental table 2 and supplemental fig 2). Patient 4 could not be tested for parental origin because parental DNAs were unavailable.

FISH analysis

Patient 2 was investigated by FISH given the presence of two rearrangements on chromosome 17. Co-hybridisation of HNPP locus specific probes (RP11-64B12) and 17q21.31 specific probes (RP11-100C5 and RP11-413P22) demonstrated that the 17q21.31 duplication occurs on the same chromosome as the HNPP microdeletion (data not shown).


The 17q21.31 microduplication described here is a novel syndrome characterised by features of the autism spectrum disorder. All patients have behavioural problems with social interaction and communication difficulties and mild psychomotor retardation, suggesting that some of the genes within the duplication interval may be candidates for the autistic spectrum, even if at this stage this hypothesis requires formal testing. A linkage study of autistic families13 has identified strong linkage within 17q21. However, this association was not confirmed by a larger linkage study reported by the Autism Consortium14 nor in the CNV analysis of autistic families reported by Sebat et al.15 The duplicated interval contains six genes (two ORFs and four known genes). Some of these are good candidates for the impaired cognitive function or motor skills observed in our patients. Some mutations of MAPT (microtubule associated protein TAU) altering the biochemistry of tau protein have been reported to cause a greater abundance of unbound tau species and formation of insoluble tau inclusions in frontotemporal dementia and parkinsonism (Pittman et al16), which could therefore be considered as a “toxic gain of function mutation”. Increased accumulation could be expected from dosage sensitive overexpression of the duplicated MAPT gene observed in our patients. However, MAPT gene duplications have not been identified in cases of frontotemporal lobar degeneration.17 Patients 1, 2 and 4 have had a brain scan with no sign of neurodegeneration. However, it should be mentioned that our patients are between 6–20 years of age whereas the first signs of degeneration in frontotemporal dementia usually become apparent around 45–55 years of age.16 Then a follow-up of these patients for this particular aspect is certainly advisable to determine their evolution with age. CRHR1 (corticotropin releasing hormone receptor 1) is expressed in the embryonic and postnatal cerebellum and could play a role in central nervous system development by affecting neurone survival through a depolarisation mechanism.18 Duplication of the gene coding for CRHR1 could explain the impaired motor skills reported for at least two patients (patient 1 and 2). Moreover experimental evidence suggests that CRHR1 is involved in stress response and anxiety related behaviour.19 CRHR1 is one of the preferred targets for depression, anxiety and stress related therapy.20 Again duplication of this gene could contribute to the behavioural troubles and poor social interaction observed in four of the five patients reported with 17q21.31 microduplication. Finally, IMP5 encodes a gene highly homologous to the presenilin 1 and 2 genes, which are involved in APP (amyloid precursor protein) cleavage. Interaction of IMP5, or other presenilin analogues, with APP has not been formally established but experimental evidence suggests that protease others than presenilin 1 and 2 are involved in APP cleavage,21 22 since APP cleavage still occurs in presenilin 1 and 2 double knock-out mouse fibroblasts.

The other phenotypic aspects are very variable and much milder. Other common features displayed by these four cases and by the patient reported by Kirchhoff et al14 include: dysplastic ears (4/5), hypotonia (3/5), hyperactivity (3/4), poor verbal skills (3/5), prominent incisors (3/5), poor motor skills (2/3), tiptoe walking (2/4) and short nose (2/5). More cases are required to define accurately the clinical spectrum associated with this microduplication as many features described are mild and non-specific. Patient 2 was noted to share a number of dysmorphic features with the case described by Kirchhoff, including hypotonia, poor motor skills, dysplastic ears, short nose, short philtrum, high arched palate, prominent incisors and hirsutism. The role of the duplication in the aetiology of the hypogonadism in case 4 is unclear and will await identification of further cases. However, it suggests that all cases of 17q21.31 microduplication should be assessed for this potential complication. A patient identified by Hunter Genetics in Australia has been recently reported in the DECIPHER database ( This patient was reported to have mental retardation, abnormal shaped teeth and tapering fingers.

No phenotypic similarities or differences between deletion and duplication carriers are obvious. Deletion carriers seem to be more affected (moderate to severe mental retardation) than duplication carriers (mild mental retardation in our four patients). It is also interesting to note that, whereas deletion carriers have a friendly and amiable behaviour, duplication carriers have some traits reminiscent of the autistic spectrum. This opposite phenotypic situation should certainly be kept in mind in the future when additional patients are clinically assessed.

The phenotypic variability is striking and adds this syndrome to a growing group of chromosomal imbalances with variable expressivity. Such variability is observed, especially in association with microduplication genomic disorders: duplications of the SMS,23 Williams–Beuren,24 22q1125 and NF126 loci result in a heterogeneous phenotype ranging from normal to severe. Distal deletion and duplication of the 22q11 locus has been reported in patients with highly variable symptoms, especially for the duplication carriers.25 27 Another recently identified microdeletion/duplication genomic disorder of 1.35 Mb at 1q21.1 also escapes syndromic classification due to large phenotypic variability such that identification of carriers probably precludes a phenotypic based evaluation.28

The broad variability of phenotypes associated with specific chromosomal rearrangements suggests that many mildly affected individuals remain undiagnosed. This generates an ascertainment bias in the evaluation of the frequency of such rearrangements. The cohort investigated here was made of 13 070 patients. A total of 24 deletion carriers were observed. Based on a frequency of 2% for individuals presenting with mental retardation in the human population, the prevalence of 17q21.31 deletion could be evaluated to 1 in 55 000, as explained in Koolen et al,2 which is lower than the value of 1 in 16 000 reported by Koolen et al.2 This certainly reflects a bias in the selection of patients. Similarly, the prevalence of duplication can be evaluated to 1 in 327 000 which is less than half the prevalence of deletion which would be mechanistically expected from the model proposed by Turner et al8 for deletion and duplication resulting from NAHR.

Koolen et al2 were able to map the proximal breakpoint to a 500 bp interval (position 41046729–41047168 bp on chromosome 17) corresponding to a L2 LINE element. This is comparable to the proximal breakpoints in the four patients with the microduplication. However, as mentioned by Koolen et al,2 the distal breakpoint falls within a CNV region, known to be polymorphic in normal individuals. Interestingly, each patient with the duplication investigated with non-targeted arrays (patients 1–3 and Kirchhoff’s patient) show an amplified copy number for this CNV, indicating either that one of the parents is a carrier of a multi-copy allele for this CNV or that the duplication involves a more distal breakpoint.

In three of four cases reported here, the microduplication could be demonstrated to be of maternal origin whereas a paternal origin was established for the first patient described by Kirchhoff et al.9 In the three cases tested, the parent of origin was either heterozygote or homozygote for the H2 haplotype. In patient 3 the duplication arose through interchromatid rearrangement whereas, for the two others, an interchromosomal mechanism was discovered. This again emphasises the role of the inverted H2 haplotype in this microdeletion/microduplication syndrome as recently investigated in detail in the human and great apes lineage.29

In conclusion, we report four additional cases with 17q21.31 microduplication (for three of which de novo occurrence could be assessed) which contribute to documenting the molecular and clinical features associated with this recently discovered syndrome. While the microdeletion associated phenotype is now clearly established, the microduplication is associated with a very variable phenotype, of which behavioural problems and poor social interaction seem to be the most consistent.


We would like to thank Mélanie D’Amico and Dr Pascale Hilbert for their support with microsatellites analysis. We are grateful to the MicroArray Facility, Flanders Interuniversity Institute for Biotechnology (VIB) for their help in the spotting of the arrays and the Mapping Core and Map Finishing groups of the Wellcome Trust Sanger Institute for the initial clone supply and verification.


Supplementary materials


  • ▸ Additional tables and figures are published online only at

  • Funding: This work was funded by a grant from the Institute de Recherche Scientifique en Pathologie et en Génétique and by grants to JV from the Catholic University of Leuven (GOA/2006/12 and Centre of Excellence SymBioSys, Research Council K.U.Leuven EF/05/007) and the IWT (IWT 60848). This work was also funded by a Biomedical Research grant from the National Institute for Health Research (NIHR), UK.

  • Competing interests: J Rosenfeld is an employee of Signature Genomic.

  • Patient consent: Obtained.