American Journal of Medical Genetics Part C (Seminars in Medical Genetics) 166C:276–289 (2014)
A R T I C L E
The ARID1B Phenotype: What We Have Learned so Far GIJS W.E. SANTEN*, JILL CLAYTON‐SMITH,
AND
THE ARID1B‐CSS CONSORTIUM
Evidence is now accumulating from a number of sequencing studies that ARID1B not only appears to be one of the most frequently mutated intellectual disability (ID) genes, but that the range of phenotypes caused by ARID1B mutations seems to be extremely wide. Thus, it is one of the most interesting ID genes identified so far in the exome sequencing era. In this article, we review the literature surrounding ARID1B and attempt to delineate the ARID1B phenotype. The vast majority of published ARID1B patients have been ascertained through studies of Coffin–Siris syndrome (CSS), which leads to bias when documenting the frequencies of phenotypic features. Additional observations of those individuals ascertained through exome sequencing studies helps in delineation of the broader clinical phenotype. We are currently establishing an ARID1B consortium, aimed at collecting ARID1B patients identified through genome‐wide sequencing strategies. We hope that this endeavor will eventually lead to a more comprehensive view of the ARID1B phenotype. © 2014 Wiley Periodicals, Inc. KEY WORDS: Coffin–Siris syndrome; ARID1B; review
How to cite this article: Santen GW, Clayton‐Smith J the ARID1B‐CSS consortium. 2014. The ARID1B phenotype: What we have learned so far. Am J Med Genet Part C 166C:276–289.
INTRODUCTION Over the past few years ARID1B has been recognized as a major intellectual disability (ID) gene. Indeed, it appears to be the gene most frequently mutated in large‐scale sequencing studies in patients with intellectual disability (Hurles M et al., ESHG 2013 Conference). In this review we summarize the available literature and clinical information on ARID1B thus far. We begin with the literature on patients with cytogenetic imbalances, followed by a description of patients with ARID1B point mutations identified in both non‐syndromic ID and Coffin–Siris syndrome (CSS) (OMIM 135900). Through incorporation of all this information, we attempt to delineate the ARID1B phenotype and discuss the importance of such work for clinicians, patients, and their parents. We close by making several recommen-
dations for diagnosis and management of these patients.
METHODS Literature Search To review published reports of how ARID1B has been linked to ID, we performed a literature search in PUBMED using the keywords ‘ARID1B’ and ‘6q25 deletion/translocation’. No relevant reports were identified by searching for ‘ARID1B duplication’ or ‘6q25 duplication’. Clinical Data To complement the clinical information on ARID1B patients with sequence variants gathered by ourselves, we obtained clinical data of 60 ARID1B patients from several CSS cohort studies.
[Wieczorek et al., 2013; Tsurusaki et al., 2014; Santen et al., 2013]. Although, when analyzing clinical features, selection bias will influence the frequency of some CSS‐specific features observed, the same does not necessarily apply to other features such as myopia and epilepsy, as these do not help a clinician in triggering the CSS diagnosis.
RESULTS AND DISCUSSION We describe firstly those cases with chromosomal deletions and rearrangements involving ARID1B and then those with sequence variants. Cytogenetic Abnormalities Involving ARID1B We reviewed reports of individuals with deletions or rearrangements involving
The authors declare no conflicts of interest. Gijs Santen is a Clinical Geneticist in training at the Leiden University Medical Center striving to improve clinical delineation of the Coffin‐Siris syndrome. As postdoctoral researcher, he works on elucidation of the pathophysiological mechanisms behind the mutations identified in Coffin‐Siris syndrome. Jill Clayton‐Smith is a Consultant Clinical Geneticist and Honorary Professor working within a large regional genetic service in the North West of England. Her main interest is in paediatric dysmorphology, especially syndromic learning disability. *Correspondence to: Gijs W.E. Santen, M.D., Ph.D, Department of Clinical Genetics, Leiden University Medical Center, Postbus 9600, 2300 RC Leiden. E‐mail: [email protected] DOI 10.1002/ajmg.c.31414 Article first published online in Wiley Online Library (wileyonlinelibrary.com): 28 August 2014
ß 2014 Wiley Periodicals, Inc.
ARTICLE
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
the ARID1B locus at chromosome 6q25.3. Reports of terminal 6q25 deletions were excluded, as the large size of most of these deletions is such that it is unlikely that ARID1B haploinsufficiency is the sole cause of the 6q25 terminal deletion phenotype. Furthermore, we only included papers which included molecular data (e.g., SNP‐ array or STR (Short Tandem Repeat)‐ data) to define deletion size. A summary of the included cases is given in Table I and Figure 1. The deletions range in size from 58 kb to 14.5 Mb. In earlier reports especially, it is difficult to be sure that ARID1B is truly deleted. The first report of a patient with a phenotype likely to have been caused by ARID1B haploinsufficiency was published in 1998 [Pirola et al., 1998]. The patient, who had ID and agenesis of the corpus callosum (ACC), had a deletion
of 5.5–7.6 Mb which included ARID1B in addition to 10 other genes. Interestingly, the patient was noted to have long
The first report of a patient with a phenotype likely to have been caused by ARID1B haploinsufficiency was published in 1998. The patient, who had ID and agenesis of the corpus callosum (ACC), had a deletion of 5.5–7.6 Mb which included ARID1B in addition to 10 other genes.
277
distal phalanges, rather than short phalanges as often observed in CSS [Coffin and Siris, 1970; Schrier et al., 2012]. One year later, three patients were published with large and overlapping deletions of varying sizes of 6q24.2‐ q25.1, two of these included ARID1B [Sukumar et al., 1999]. The patients all had ID, and feeding problems were described in patient 2, whereas patient 3 was reported to have hypertrichosis and ACC. The first review of micro deletions of 6q25 appeared in the European Journal of Human Genetics in 2009 [Nagamani et al., 2009]. Common features in the four patients described in this paper are ID, microcephaly, hearing loss (in all), and ACC (in 2/3). One of the patients had epileptic seizures. The sizes of the deletions ranged from 3.77 to 13.81 Mb, and the smallest region of overlap (SRO)
TABLE I. Summary of Literature of Reported Patients With a Deletion of ARID1B, See Also Figure 1 Reference
Patient no.
Deletion
Deletion size
Phenotype
6q25.1q25.3 6q25.1q25.3 6q25.1q26
Hoyer et al. [2012] Santen et al. [2012] Tsurusaki et al. [2012] Wieczorek et al. [2013] Wieczorek et al. [2013] Santen et al. [2013]
1 2 3 1 2 3 4 1 1 1 2 3 4 5 6 7 8 1 6 12 K2428 K2438 24
5.5–7.6 Mb ? ? 3.77 Mb 6.7 Mb 10.3 Mb 13.81 Mb NA 0.2 Mb NA 0.2 Mb 0.6 Mb 1.9 Mb 2.7 Mb 4.6 Mb 8.2 Mb 14.5 Mb 2.5 Mb 0.88 Mb 9.2 Mb (total) 58 Kb 2 Mb 7 Mb
Sim et al. [2014] Vengoechea et al. [2014]
1 1
ID, ACC ID, feeding difficulties ID, ACC, hypertrichosis ID, microcephaly, hearing loss ID, microcephaly, hearing loss, seizures ID, microcephaly, hearing loss, ACC ID, microcephaly, hearing loss, ACC ID, ACC, speech delay Autism ID, speech impairment, ACC ID, speech impairment, hypertrichosis ID, speech impairment ID, speech impairment ID, speech impairment ID, speech impairment, pACC ID, speech impairment, pACC, hypertrichosis ID, speech impairment, hypertrichosis ID, speech impairments, no ACC ID, speech delay, ACC ID, absent speech, seizure, no ACC ID, hypotonia, no ACC ID, hypotonia, no ACC ID, feeding problems, seizures, absent speech, no ACC ID, absent speech, plantar fat pads ID, hearing loss, myopia, thyroid cancer
Pirola et al. [1998] Sukumar et al. [1999] Nagamani et al. [2009]
Backx et al. [2011] Nord et al. [2011] Halgren et al. [2012] Santen et al. [2012a], pt 5 Santen et al. [2012], pt 4
t(6;14)(q25.3;q13.2) t(1;6)(p31;q25)
t(4;6) (q35;q25.3),
1.2 Mb 4.2 Mb
ID, intellectual disability; (p)ACC, (partial) agenesis of the corpus callosum.
278
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
ARTICLE
Figure 1. Depiction of the published deletions including ARID1B (top) and zoomed in on ARID1B itself (bottom).
was 3.52 Mb and contained 12 protein coding genes. With the benefit of hindsight it is interesting to note that although ARID1B is right in the middle of the SRO, it was not considered a candidate to explain the phenotype in these patients. Rather, three other genes, TIMA2, NOX3, and SYNJ2 were considered the primary candidates based on their function. The first paper to make a direct link between ARID1B and ID was a report in 2011 of a translocation between chromosomes 6 and 14, t(6;14)(q25.3;q13.2) [Backx et al., 2011]. The authors performed an extensive genetic workup
and showed that two genes were disrupted by the translocation (ARID1B and MRPP3) and that several fusion transcripts were present for MRPP3, which might explain part of the phenotype. However, they rightly concluded that ARID1B was a good candidate to explain the neurodevelopmental phenotype in their patient who had ID, ACC, and speech delay. A further clue for the pathogenicity of ARID1B came from a study which reported a patient with autism and a deletion of 3 exons of ARID1B, leading to a frame shift and a premature stop [Nord et al., 2011].
Four additional patients with isolated ARID1B disruption/mutations were published on‐line in 2011 [Halgren et al., 2012]. One patient had a translocation disrupting ARID1B (t(1;6) (p31;q25)), and the phenotype was compared to seven patients with ARID1B deletions (three of which were limited to ARID1B) [Halgren et al., 2012]. They conclude that ID, speech impairment, autism, and ACC are key features of ARID1B haploinsufficiency. Hoyer et al. reported a patient within a 2.5 Mb microdeletion of 6q25, including ARID1B, which led
ARTICLE
them to sequence the ARID1B gene in a large cohort (see below) [Hoyer et al., 2012]. This patient had severe ID, hypotonia, no speech at the age of 3 year, but not ACC. In our paper reporting ARID1B mutations in CSS patients [Santen et al., 2012a], we described three patients with ARID1B deletions which were detected with routine SNP‐array investigation (deletion sizes 0.73–2.72 Mb, two of which were previously published [Halgren et al., 2012]). One of these three patients had some features suggestive of CSS, as she had relatively small fingernails. The other patients did not have classical CSS, although some facial resemblances were present. Tsurusaki et al, while reporting mutations in other subunits of the BAF complex in CSS patients, reported on one patient with two nearly adjacent microdeletions of 6q25, 9.2 Mb in total, which included ARID1B [Tsurusaki et al., 2012]. In two CSS follow‐up studies, two additional patients were described with deletions of 2 Mb and 7 Mb, both encompassing several other genes [Santen et al., 2013; Wieczorek et al., 2013]. Sim et al. reported a patient with an unbalanced translocation (46,XY,t(4;6) (q35;q25.3), resulting in a deletion of 1.2 Mb encompassing four genes including ARID1B. This patient had moderate ID, absent speech, and several dysmorphic features [Sim et al., 2014]. Lastly, a patient was reported with an ARID1B phenotype and papillary thyroid cancer who had a 4.2 Mb deletion encompassing 14 genes including ARID1B [Vengoechea et al., 2014]. Unfortunately, the report does not contain additional molecular data which might link the germline ARID1B mutation to the thyroid cancer, for example by demonstrating a second hit in the tumor. We have previously discussed whether ARID1B mutations might lead to an increased risk of cancer, and concluded there is no clinical evidence for this at the present time [Santen et al., 2012b, 2013]. It remains crucial to continue to follow‐up ARID1B patients into adulthood and report any incidences of malignancies.
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
Phenotypes of Patient with Cytogenetic Abnormalities Involving ARID1B The first review of 6q25.3 deletions [Nagamani et al., 2009], concluded that ID, ACC, microcephaly, and deafness were key features of ARID1B haploinsufficiency. A second review [Halgren et al., 2012] largely concurred with these findings, with speech delay and autistic features being shared by many of the patients. Hoyer et al. [2012] concluded that autistic features, structural brain abnormalities (such as ACC) and deafness are rare and they suggested that this latter feature was more common in patients with larger deletions. Thus far, two patients have had translocations through ARID1B, five have had intragenic deletions and 10 patients have had larger deletions. Summarizing the information from all these patients it seems that intellectual disability is present in a variable degree that ACC is a common feature, and that deafness is relatively uncommon.
Point Mutations in ARID1B The final proof of the implication of ARID1B in ID came in 2012, when three published papers firmly established its relevance as an ID gene [Hoyer et al., 2012; Santen et al., 2012a; Tsurusaki et al., 2012]. Firstly, a study reported the screening of five candidate genes from within a 2.5 Mb microdeletion of 6q25, including ARID1B, in 900 patients with intellectual disability. Mutations were identified in eight additional patients (0.9%), one of which was a small intragenic duplication [Hoyer et al., 2012]. Although this number may seem relatively low, it must be recognized that ID is a very heterogeneous disease and a mutation rate of 1% in a relatively unselected cohort is considered very high. Shortly after this first report, two papers linked ARID1B, and other subunits of the BAF (mSWI/SNF) complex, to the CSS phenotype [Santen et al., 2012a; Tsurusaki et al., 2012]. We described three patients with CSS and an ARID1B mutation
279
[Santen et al., 2012a]. Tsurusaki and coworkers detected ARID1B point mutations in 5/22 patients with CSS and identified mutations in other subunits of the BAF complex in several patients, linking them to CSS [Tsurusaki et al., 2012]. For a review on these genes see the article elsewhere in this issue. Subsequent follow‐up papers in cohorts with CSS confirmed that the relative frequency of ARID1B mutations in CSS is high (68%, 83%, and 76% respectively) [Santen et al., 2013; Tsurusaki et al., 2014; Wieczorek et al., 2013]. Several recent reports have further broadened the ARID1B phenotype. One Coffin–Siris patient with an ARID1B mutation and obesity, macrocephaly, hepatomegaly, and hyperinsulinism was reported [Vals et al., 2014] though it is not yet possible to conclude if these features are truly associated with ARID1B or not. Interestingly, triggered by a patient with a deletion, ARID1B mutations have also been identified in four patients with a phenotype reminiscent of Pierpont syndrome, characterized by the presence of plantar fat pads [Sim et al., 2014]. Phenotype of CSS Patients with ARID1B Mutations We have aggregated the data on 60 Coffin–Siris patients (30 female, 50%) with ARID1B mutations, combining the available clinical information from previous studies [Santen et al., 2013; Tsurusaki et al., 2014; Wieczorek et al., 2013]. Of these 60 patients, three had deletions also including other genes, and 57 had mutations restricted to ARID1B. Although it is well known that in some cases multiple genes in a microdeletion can cause part of the phenotype [Santen et al., 2012c], we have chosen to include the patients with an ARID1B deletion in the comparison because of their prior CSS diagnosis. For about 50% of those the clinical information has been updated by the referring clinicians. The unedited clinical data per patient is included as Supplementary Table SI. In Table II, aggregated data of the features occurring in >25% of the patients are
280
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
displayed, ordered by frequency. Table III contains the aggregated data ordered by organ system. Below, we discuss only those features in detail of which we find the frequencies particularly relevant. As the current data is derived from patients with a prior CSS diagnosis, we cannot make definitive conclusions on the phenotypes of nonsyndromic ARID1B patients. However, we do expect significant overlap for many features. Previous reviews on the ARID1B microdeletion syndrome have been contradictory with respect to ACC, hearing problems and autistic features being part of the syndrome [Nagamani et al., 2009; Halgren et al., 2012; Hoyer et al., 2012]. In the current study 17/48 (35%) patients have some degree of hypoplasia of the corpus callosum, confirming this feature as a core part of the ARID1B phenotype. Hearing problems have been noted in 8/ 56 (14%) patients with an ARID1B mutation, although most of those are mild. We find it difficult to assess autistic features in patients with this degree of intellectual disability, and have not attempted to investigate this part of the phenotype in our patients.
In the current study 17/48 (35%) patients have some degree of hypoplasia of the corpus callosum, confirming this feature as a core part of the ARID1B phenotype.
no feeding difficulties whatsoever. Indeed, when focusing on the degree of ID for example, this ranges from low‐normal to severe (Fig. 2A), although most patients are reported to have moderate ID. Figure 2B shows the distribution of the development quotient. The reported quotients are on different scales and different ages, thus limiting the value of this data (n ¼ 14). With respect to speech impairment, while we were convinced in our first ARID1B paper that severe speech impairment (not more than a few words) was a consistent feature in the patients [Santen et al., 2012a], we have now clinically assessed several ARID1B patients who were able to communicate well. Indeed, one of the three patients of our first report has developed speech in the past years. This means that we are now able to confer to parents of patients that although speech development may be severely delayed, it is well possible that it may still develop, even beyond the age of 3–4 years. To illustrate this we have included a histogram of the age at which patients spoke their first words (Fig. 3). We have data on 25 patients, three of which were reported not to speak (at ages 15 months, 15 years and 19 years respectively). From the histogram it is clear that most patients can speak some words at the age of 3.5, but that the variation is, again, quite large. Motor development is delayed in CSS. The age at which patients were able to walk without support was reported for 47 patients (Fig. 4). Although most patients are able to walk by their third birthday, this may be delayed until 4–6 years.
Childhood Development
Remarkable Features
As previously observed [Santen et al., 2013], the ARID1B phenotype is enormously variable. We have seen patients with ARID1B mutations who hardly speak, or who have severe feeding difficulties necessitating PEG tube feeding for years, as well as patients who are able to communicate well, are able to start a PC or tablet to view YouTube movies, taught themselves how to read, and have
In our latest follow‐up study we reported on 28 patients with ARID1B mutations and a CSS diagnosis [Santen et al., 2013]. We identified some rarer features which were thus far not described, the most striking one being the severe myopia (up to 18D). In the current study we see myopia in 9/46 (20%) patients. The severity of the myopia ranged from 2 to 22D, five patients were reported to
ARTICLE
have
A R T I C L E
The ARID1B Phenotype: What We Have Learned so Far GIJS W.E. SANTEN*, JILL CLAYTON‐SMITH,
AND
THE ARID1B‐CSS CONSORTIUM
Evidence is now accumulating from a number of sequencing studies that ARID1B not only appears to be one of the most frequently mutated intellectual disability (ID) genes, but that the range of phenotypes caused by ARID1B mutations seems to be extremely wide. Thus, it is one of the most interesting ID genes identified so far in the exome sequencing era. In this article, we review the literature surrounding ARID1B and attempt to delineate the ARID1B phenotype. The vast majority of published ARID1B patients have been ascertained through studies of Coffin–Siris syndrome (CSS), which leads to bias when documenting the frequencies of phenotypic features. Additional observations of those individuals ascertained through exome sequencing studies helps in delineation of the broader clinical phenotype. We are currently establishing an ARID1B consortium, aimed at collecting ARID1B patients identified through genome‐wide sequencing strategies. We hope that this endeavor will eventually lead to a more comprehensive view of the ARID1B phenotype. © 2014 Wiley Periodicals, Inc. KEY WORDS: Coffin–Siris syndrome; ARID1B; review
How to cite this article: Santen GW, Clayton‐Smith J the ARID1B‐CSS consortium. 2014. The ARID1B phenotype: What we have learned so far. Am J Med Genet Part C 166C:276–289.
INTRODUCTION Over the past few years ARID1B has been recognized as a major intellectual disability (ID) gene. Indeed, it appears to be the gene most frequently mutated in large‐scale sequencing studies in patients with intellectual disability (Hurles M et al., ESHG 2013 Conference). In this review we summarize the available literature and clinical information on ARID1B thus far. We begin with the literature on patients with cytogenetic imbalances, followed by a description of patients with ARID1B point mutations identified in both non‐syndromic ID and Coffin–Siris syndrome (CSS) (OMIM 135900). Through incorporation of all this information, we attempt to delineate the ARID1B phenotype and discuss the importance of such work for clinicians, patients, and their parents. We close by making several recommen-
dations for diagnosis and management of these patients.
METHODS Literature Search To review published reports of how ARID1B has been linked to ID, we performed a literature search in PUBMED using the keywords ‘ARID1B’ and ‘6q25 deletion/translocation’. No relevant reports were identified by searching for ‘ARID1B duplication’ or ‘6q25 duplication’. Clinical Data To complement the clinical information on ARID1B patients with sequence variants gathered by ourselves, we obtained clinical data of 60 ARID1B patients from several CSS cohort studies.
[Wieczorek et al., 2013; Tsurusaki et al., 2014; Santen et al., 2013]. Although, when analyzing clinical features, selection bias will influence the frequency of some CSS‐specific features observed, the same does not necessarily apply to other features such as myopia and epilepsy, as these do not help a clinician in triggering the CSS diagnosis.
RESULTS AND DISCUSSION We describe firstly those cases with chromosomal deletions and rearrangements involving ARID1B and then those with sequence variants. Cytogenetic Abnormalities Involving ARID1B We reviewed reports of individuals with deletions or rearrangements involving
The authors declare no conflicts of interest. Gijs Santen is a Clinical Geneticist in training at the Leiden University Medical Center striving to improve clinical delineation of the Coffin‐Siris syndrome. As postdoctoral researcher, he works on elucidation of the pathophysiological mechanisms behind the mutations identified in Coffin‐Siris syndrome. Jill Clayton‐Smith is a Consultant Clinical Geneticist and Honorary Professor working within a large regional genetic service in the North West of England. Her main interest is in paediatric dysmorphology, especially syndromic learning disability. *Correspondence to: Gijs W.E. Santen, M.D., Ph.D, Department of Clinical Genetics, Leiden University Medical Center, Postbus 9600, 2300 RC Leiden. E‐mail: [email protected] DOI 10.1002/ajmg.c.31414 Article first published online in Wiley Online Library (wileyonlinelibrary.com): 28 August 2014
ß 2014 Wiley Periodicals, Inc.
ARTICLE
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
the ARID1B locus at chromosome 6q25.3. Reports of terminal 6q25 deletions were excluded, as the large size of most of these deletions is such that it is unlikely that ARID1B haploinsufficiency is the sole cause of the 6q25 terminal deletion phenotype. Furthermore, we only included papers which included molecular data (e.g., SNP‐ array or STR (Short Tandem Repeat)‐ data) to define deletion size. A summary of the included cases is given in Table I and Figure 1. The deletions range in size from 58 kb to 14.5 Mb. In earlier reports especially, it is difficult to be sure that ARID1B is truly deleted. The first report of a patient with a phenotype likely to have been caused by ARID1B haploinsufficiency was published in 1998 [Pirola et al., 1998]. The patient, who had ID and agenesis of the corpus callosum (ACC), had a deletion
of 5.5–7.6 Mb which included ARID1B in addition to 10 other genes. Interestingly, the patient was noted to have long
The first report of a patient with a phenotype likely to have been caused by ARID1B haploinsufficiency was published in 1998. The patient, who had ID and agenesis of the corpus callosum (ACC), had a deletion of 5.5–7.6 Mb which included ARID1B in addition to 10 other genes.
277
distal phalanges, rather than short phalanges as often observed in CSS [Coffin and Siris, 1970; Schrier et al., 2012]. One year later, three patients were published with large and overlapping deletions of varying sizes of 6q24.2‐ q25.1, two of these included ARID1B [Sukumar et al., 1999]. The patients all had ID, and feeding problems were described in patient 2, whereas patient 3 was reported to have hypertrichosis and ACC. The first review of micro deletions of 6q25 appeared in the European Journal of Human Genetics in 2009 [Nagamani et al., 2009]. Common features in the four patients described in this paper are ID, microcephaly, hearing loss (in all), and ACC (in 2/3). One of the patients had epileptic seizures. The sizes of the deletions ranged from 3.77 to 13.81 Mb, and the smallest region of overlap (SRO)
TABLE I. Summary of Literature of Reported Patients With a Deletion of ARID1B, See Also Figure 1 Reference
Patient no.
Deletion
Deletion size
Phenotype
6q25.1q25.3 6q25.1q25.3 6q25.1q26
Hoyer et al. [2012] Santen et al. [2012] Tsurusaki et al. [2012] Wieczorek et al. [2013] Wieczorek et al. [2013] Santen et al. [2013]
1 2 3 1 2 3 4 1 1 1 2 3 4 5 6 7 8 1 6 12 K2428 K2438 24
5.5–7.6 Mb ? ? 3.77 Mb 6.7 Mb 10.3 Mb 13.81 Mb NA 0.2 Mb NA 0.2 Mb 0.6 Mb 1.9 Mb 2.7 Mb 4.6 Mb 8.2 Mb 14.5 Mb 2.5 Mb 0.88 Mb 9.2 Mb (total) 58 Kb 2 Mb 7 Mb
Sim et al. [2014] Vengoechea et al. [2014]
1 1
ID, ACC ID, feeding difficulties ID, ACC, hypertrichosis ID, microcephaly, hearing loss ID, microcephaly, hearing loss, seizures ID, microcephaly, hearing loss, ACC ID, microcephaly, hearing loss, ACC ID, ACC, speech delay Autism ID, speech impairment, ACC ID, speech impairment, hypertrichosis ID, speech impairment ID, speech impairment ID, speech impairment ID, speech impairment, pACC ID, speech impairment, pACC, hypertrichosis ID, speech impairment, hypertrichosis ID, speech impairments, no ACC ID, speech delay, ACC ID, absent speech, seizure, no ACC ID, hypotonia, no ACC ID, hypotonia, no ACC ID, feeding problems, seizures, absent speech, no ACC ID, absent speech, plantar fat pads ID, hearing loss, myopia, thyroid cancer
Pirola et al. [1998] Sukumar et al. [1999] Nagamani et al. [2009]
Backx et al. [2011] Nord et al. [2011] Halgren et al. [2012] Santen et al. [2012a], pt 5 Santen et al. [2012], pt 4
t(6;14)(q25.3;q13.2) t(1;6)(p31;q25)
t(4;6) (q35;q25.3),
1.2 Mb 4.2 Mb
ID, intellectual disability; (p)ACC, (partial) agenesis of the corpus callosum.
278
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
ARTICLE
Figure 1. Depiction of the published deletions including ARID1B (top) and zoomed in on ARID1B itself (bottom).
was 3.52 Mb and contained 12 protein coding genes. With the benefit of hindsight it is interesting to note that although ARID1B is right in the middle of the SRO, it was not considered a candidate to explain the phenotype in these patients. Rather, three other genes, TIMA2, NOX3, and SYNJ2 were considered the primary candidates based on their function. The first paper to make a direct link between ARID1B and ID was a report in 2011 of a translocation between chromosomes 6 and 14, t(6;14)(q25.3;q13.2) [Backx et al., 2011]. The authors performed an extensive genetic workup
and showed that two genes were disrupted by the translocation (ARID1B and MRPP3) and that several fusion transcripts were present for MRPP3, which might explain part of the phenotype. However, they rightly concluded that ARID1B was a good candidate to explain the neurodevelopmental phenotype in their patient who had ID, ACC, and speech delay. A further clue for the pathogenicity of ARID1B came from a study which reported a patient with autism and a deletion of 3 exons of ARID1B, leading to a frame shift and a premature stop [Nord et al., 2011].
Four additional patients with isolated ARID1B disruption/mutations were published on‐line in 2011 [Halgren et al., 2012]. One patient had a translocation disrupting ARID1B (t(1;6) (p31;q25)), and the phenotype was compared to seven patients with ARID1B deletions (three of which were limited to ARID1B) [Halgren et al., 2012]. They conclude that ID, speech impairment, autism, and ACC are key features of ARID1B haploinsufficiency. Hoyer et al. reported a patient within a 2.5 Mb microdeletion of 6q25, including ARID1B, which led
ARTICLE
them to sequence the ARID1B gene in a large cohort (see below) [Hoyer et al., 2012]. This patient had severe ID, hypotonia, no speech at the age of 3 year, but not ACC. In our paper reporting ARID1B mutations in CSS patients [Santen et al., 2012a], we described three patients with ARID1B deletions which were detected with routine SNP‐array investigation (deletion sizes 0.73–2.72 Mb, two of which were previously published [Halgren et al., 2012]). One of these three patients had some features suggestive of CSS, as she had relatively small fingernails. The other patients did not have classical CSS, although some facial resemblances were present. Tsurusaki et al, while reporting mutations in other subunits of the BAF complex in CSS patients, reported on one patient with two nearly adjacent microdeletions of 6q25, 9.2 Mb in total, which included ARID1B [Tsurusaki et al., 2012]. In two CSS follow‐up studies, two additional patients were described with deletions of 2 Mb and 7 Mb, both encompassing several other genes [Santen et al., 2013; Wieczorek et al., 2013]. Sim et al. reported a patient with an unbalanced translocation (46,XY,t(4;6) (q35;q25.3), resulting in a deletion of 1.2 Mb encompassing four genes including ARID1B. This patient had moderate ID, absent speech, and several dysmorphic features [Sim et al., 2014]. Lastly, a patient was reported with an ARID1B phenotype and papillary thyroid cancer who had a 4.2 Mb deletion encompassing 14 genes including ARID1B [Vengoechea et al., 2014]. Unfortunately, the report does not contain additional molecular data which might link the germline ARID1B mutation to the thyroid cancer, for example by demonstrating a second hit in the tumor. We have previously discussed whether ARID1B mutations might lead to an increased risk of cancer, and concluded there is no clinical evidence for this at the present time [Santen et al., 2012b, 2013]. It remains crucial to continue to follow‐up ARID1B patients into adulthood and report any incidences of malignancies.
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
Phenotypes of Patient with Cytogenetic Abnormalities Involving ARID1B The first review of 6q25.3 deletions [Nagamani et al., 2009], concluded that ID, ACC, microcephaly, and deafness were key features of ARID1B haploinsufficiency. A second review [Halgren et al., 2012] largely concurred with these findings, with speech delay and autistic features being shared by many of the patients. Hoyer et al. [2012] concluded that autistic features, structural brain abnormalities (such as ACC) and deafness are rare and they suggested that this latter feature was more common in patients with larger deletions. Thus far, two patients have had translocations through ARID1B, five have had intragenic deletions and 10 patients have had larger deletions. Summarizing the information from all these patients it seems that intellectual disability is present in a variable degree that ACC is a common feature, and that deafness is relatively uncommon.
Point Mutations in ARID1B The final proof of the implication of ARID1B in ID came in 2012, when three published papers firmly established its relevance as an ID gene [Hoyer et al., 2012; Santen et al., 2012a; Tsurusaki et al., 2012]. Firstly, a study reported the screening of five candidate genes from within a 2.5 Mb microdeletion of 6q25, including ARID1B, in 900 patients with intellectual disability. Mutations were identified in eight additional patients (0.9%), one of which was a small intragenic duplication [Hoyer et al., 2012]. Although this number may seem relatively low, it must be recognized that ID is a very heterogeneous disease and a mutation rate of 1% in a relatively unselected cohort is considered very high. Shortly after this first report, two papers linked ARID1B, and other subunits of the BAF (mSWI/SNF) complex, to the CSS phenotype [Santen et al., 2012a; Tsurusaki et al., 2012]. We described three patients with CSS and an ARID1B mutation
279
[Santen et al., 2012a]. Tsurusaki and coworkers detected ARID1B point mutations in 5/22 patients with CSS and identified mutations in other subunits of the BAF complex in several patients, linking them to CSS [Tsurusaki et al., 2012]. For a review on these genes see the article elsewhere in this issue. Subsequent follow‐up papers in cohorts with CSS confirmed that the relative frequency of ARID1B mutations in CSS is high (68%, 83%, and 76% respectively) [Santen et al., 2013; Tsurusaki et al., 2014; Wieczorek et al., 2013]. Several recent reports have further broadened the ARID1B phenotype. One Coffin–Siris patient with an ARID1B mutation and obesity, macrocephaly, hepatomegaly, and hyperinsulinism was reported [Vals et al., 2014] though it is not yet possible to conclude if these features are truly associated with ARID1B or not. Interestingly, triggered by a patient with a deletion, ARID1B mutations have also been identified in four patients with a phenotype reminiscent of Pierpont syndrome, characterized by the presence of plantar fat pads [Sim et al., 2014]. Phenotype of CSS Patients with ARID1B Mutations We have aggregated the data on 60 Coffin–Siris patients (30 female, 50%) with ARID1B mutations, combining the available clinical information from previous studies [Santen et al., 2013; Tsurusaki et al., 2014; Wieczorek et al., 2013]. Of these 60 patients, three had deletions also including other genes, and 57 had mutations restricted to ARID1B. Although it is well known that in some cases multiple genes in a microdeletion can cause part of the phenotype [Santen et al., 2012c], we have chosen to include the patients with an ARID1B deletion in the comparison because of their prior CSS diagnosis. For about 50% of those the clinical information has been updated by the referring clinicians. The unedited clinical data per patient is included as Supplementary Table SI. In Table II, aggregated data of the features occurring in >25% of the patients are
280
AMERICAN JOURNAL OF MEDICAL GENETICS PART C (SEMINARS IN MEDICAL GENETICS)
displayed, ordered by frequency. Table III contains the aggregated data ordered by organ system. Below, we discuss only those features in detail of which we find the frequencies particularly relevant. As the current data is derived from patients with a prior CSS diagnosis, we cannot make definitive conclusions on the phenotypes of nonsyndromic ARID1B patients. However, we do expect significant overlap for many features. Previous reviews on the ARID1B microdeletion syndrome have been contradictory with respect to ACC, hearing problems and autistic features being part of the syndrome [Nagamani et al., 2009; Halgren et al., 2012; Hoyer et al., 2012]. In the current study 17/48 (35%) patients have some degree of hypoplasia of the corpus callosum, confirming this feature as a core part of the ARID1B phenotype. Hearing problems have been noted in 8/ 56 (14%) patients with an ARID1B mutation, although most of those are mild. We find it difficult to assess autistic features in patients with this degree of intellectual disability, and have not attempted to investigate this part of the phenotype in our patients.
In the current study 17/48 (35%) patients have some degree of hypoplasia of the corpus callosum, confirming this feature as a core part of the ARID1B phenotype.
no feeding difficulties whatsoever. Indeed, when focusing on the degree of ID for example, this ranges from low‐normal to severe (Fig. 2A), although most patients are reported to have moderate ID. Figure 2B shows the distribution of the development quotient. The reported quotients are on different scales and different ages, thus limiting the value of this data (n ¼ 14). With respect to speech impairment, while we were convinced in our first ARID1B paper that severe speech impairment (not more than a few words) was a consistent feature in the patients [Santen et al., 2012a], we have now clinically assessed several ARID1B patients who were able to communicate well. Indeed, one of the three patients of our first report has developed speech in the past years. This means that we are now able to confer to parents of patients that although speech development may be severely delayed, it is well possible that it may still develop, even beyond the age of 3–4 years. To illustrate this we have included a histogram of the age at which patients spoke their first words (Fig. 3). We have data on 25 patients, three of which were reported not to speak (at ages 15 months, 15 years and 19 years respectively). From the histogram it is clear that most patients can speak some words at the age of 3.5, but that the variation is, again, quite large. Motor development is delayed in CSS. The age at which patients were able to walk without support was reported for 47 patients (Fig. 4). Although most patients are able to walk by their third birthday, this may be delayed until 4–6 years.
Childhood Development
Remarkable Features
As previously observed [Santen et al., 2013], the ARID1B phenotype is enormously variable. We have seen patients with ARID1B mutations who hardly speak, or who have severe feeding difficulties necessitating PEG tube feeding for years, as well as patients who are able to communicate well, are able to start a PC or tablet to view YouTube movies, taught themselves how to read, and have
In our latest follow‐up study we reported on 28 patients with ARID1B mutations and a CSS diagnosis [Santen et al., 2013]. We identified some rarer features which were thus far not described, the most striking one being the severe myopia (up to 18D). In the current study we see myopia in 9/46 (20%) patients. The severity of the myopia ranged from 2 to 22D, five patients were reported to
ARTICLE
have
