Copy number variation plays an important role in clinical epilepsy

Abstract

Objective

To evaluate the role of copy number abnormalities detectable by chromosomal microarray (CMA) testing in patients with epilepsy at a tertiary care center.

Methods

We identified patients with ICD-9 codes for epilepsy or seizures and clinical CMA testing performed between October 2006 and February 2011 at Boston Children’s Hospital. We reviewed medical records and included patients meeting criteria for epilepsy. We phenotypically characterized patients with epilepsy-associated abnormalities on CMA.

Results

Of 973 patients who had CMA and ICD-9 codes for epilepsy or seizures, 805 patients satisfied criteria for epilepsy. We observed 437 copy number variants (CNVs) in 323 patients (1–4 per patient), including 185 (42%) deletions and 252 (58%) duplications. Forty (9%) were confirmed de novo, 186 (43%) were inherited, and parental data were unavailable for 211 (48%). Excluding full chromosome trisomies, CNV size ranged from 18 kb to 142 Mb, and 34% were over 500 kb. In at least 40 cases (5%), the epilepsy phenotype was explained by a CNV, including 29 patients with epilepsy-associated syndromes and 11 with likely disease-associated CNVs involving epilepsy genes or “hotspots.” We observed numerous recurrent CNVs including 10 involving loss or gain of Xp22.31, a region described in patients with and without epilepsy.

Interpretation

Copy number abnormalities play an important role in patients with epilepsy. Given that the diagnostic yield of CMA for epilepsy patients is similar to the yield in autism spectrum disorders and in prenatal diagnosis, for which published guidelines recommend testing with CMA, we recommend the implementation of CMA in the evaluation of unexplained epilepsy.

Introduction

Copy number variants (CNVs) play an increasingly recognized role in the genetics of epilepsy. Several large studies have identified recurrent genomic “hotspots” that predispose to idiopathic epilepsy, including 1q21.1, 15q11.2, 15q13.3, 15q11-q13, 16p11.2, and 16p13.11.^1–4 CNVs occur in these regions due to non-allelic homologous recombination between flanking segmental duplications. The specific genes responsible for the susceptibility to epilepsy in these CNVs have not been clearly identified in many instances, though CNVs involving known epilepsy genes (e.g., SCN1A or KCNQ2) have also been found.^5–11 Rare, non-recurrent CNVs have also been associated with epilepsy.³ “Large” CNVs, greater than 500 kb, are more often pathogenic than smaller CNVs.¹² CNVs in patients with epilepsy, especially patients who also have intellectual disabilities (ID) and other neuropsychiatric symptoms, are often larger and more gene-rich than those in controls.¹³ One study of 102 patients with epilepsy, with or without additional neurodevelopmental abnormalities, identified a likely disease-causing CNV in 10 cases (almost 10%).¹⁴

Chromosomal microarrays (CMAs) have been increasingly included in the evaluation of epilepsy when a genetic etiology is suspected. Such suspicions may be raised by the presence of dysmorphic features, developmental delay (DD), autism spectrum disorder (ASD), or family history of epilepsy. Nonetheless, there are no formal guidelines yet that include CMA in the evaluation of epilepsy patients. We sought to describe our experience with diagnostic CMA in a cohort of 805 patients with epilepsy in order to demonstrate the yield of CMA in a series of patients with epilepsy ascertained clinically and to identify novel CNVs as possible candidate regions for future study.

Methods

Patient ascertainment

This study was approved by the Boston Children’s Hospital Institutional Review Board. We identified patients with an ICD-9 code for epilepsy or seizures who had a clinical CMA performed at Boston Children’s Hospital between October 2006 and February 2011. Two pediatric neurologists trained in epilepsy (HEO and JA) reviewed medical records for each patient to confirm the diagnosis of epilepsy, defined as recurrent unprovoked seizures or a single unprovoked seizure and abnormal EEG or MRI warranting treatment with anti-epileptic medication. For patients meeting inclusion criteria, we collected information about their CNVs (location, coordinates, size, genes encompassed, inheritance). For patients with CNVs consistent with known epilepsy syndromes, containing known epilepsy genes, in known epilepsy “hotspots,” or in areas with recurrent, overlapping CNVs, we analyzed phenotypic data and classified each patient with focal vs. generalized epilepsy, epilepsy syndrome, other diagnoses, imaging abnormalities, other genetic or metabolic testing results, risk factors for epilepsy, neuropsychiatric co-morbidities, and family history of epilepsy or febrile seizures. Pathogenicity of CNVs was determined based on CMA report, consistency with clinical phenotype, and updated information on known epilepsy syndromes, genes, and “hotspots.”

Array-CGH

Whole genome oligonucleotide array comparative genomic hybridization (aCGH) arrays (Agilent 244 K) were used to detect CNVs greater than 35–50 kb. CMA was performed when possible in the biological parents of individuals with CNVs of unknown or possible significance. Genomic coordinates provided refer to the human genome build hg18 except where specified.

Pathway analysis

Ingenuity Pathway Analysis (IPA) was used to evaluate gene content and identify potential candidate genes for epilepsy.

Results

Diagnostic yield of CMA in epilepsy patients

Of 6564 CMAs performed for all indications from October 2006 to February 2011, 973 patients had an ICD-9 code for epilepsy or seizures, and 168 of these were excluded because they did not have epilepsy based on record review (Figure 1). Of the 805 patients who met inclusion criteria, we identified 437 CNVs in 323 patients (40%), range 1–4 CNV per patient. This included 184 (42%) deletions and 253 (58%) duplications. Forty (9%) were confirmed de novo, 186 (43%) inherited, and 211 (48%) of unknown inheritance. Of the inherited CNVs, 101 were maternally inherited, 78 were paternally inherited, one was an inherited homozygous deletion (one copy from each parent), and six were reported inherited but the parent of origin was not specified. The size of the CNVs ranged from 18 kb to 142 Mb, not including three whole chromosomal trisomies (X, 21, and 9), and 147 (34%) were larger than 500 kb in size (Figure 2). CNVs provided an explanation for the epilepsy in at least 40 cases (5%), including 29 patients with known syndromes involving epilepsy and 11 patients with deletions in known epilepsy genes or hotspots (Tables 1–3).

Figure 1.

Flowchart of patient selection and chromosomal microarray genotype results

Figure 2.

Pie chart of CNV sizes in our cohort.

Table 1.

CNVs in our cohort in regions consistent with genetic syndromes associated with epilepsy.

Known syndrome associated with epilepsy	Number of cases	Coordinates
22q11 duplication syndrome	4	Chr22: 17,086,001 – 17,390,449 Chr22: 17,274,135 – 19,835,414 Chr22: 17,274,835 – 19,835,417 Chr22: 17,270,271 – 19,891,514
1p36 deletion syndrome	3	Chr1: 4,010,525 – 15,294,730 Chr1: 836,543 – 2,623,228 Chr1: 1,239,023 – 2,033,572
Mowat-Wilson syndrome (ZEB2 deletion, 2q22.3)	3	Chr2: 144,551,411 – 144898926 Chr2: 143,268,368 – 147,688,255 Chr2: 144,874,055 – 145,796,638
Wolf-Hirschhorn syndrome (4p16.3 deletion)	3	Chr4: 0 – 8,759,904 Chr4: 41,413 – 2,762,699 Chr4: 62,447 – 13,267,439
Dravet syndrome (SCN1A deletion, 2q24.3)	2	Chr2: 166,040,826 – 167,790,520 Chr2: 165,476,423 – 171,820,243
Williams-Beuren region reciprocal syndrome (duplication 7q11.23)	2	Chr7: 72,039,022 – 73,780,204 Chr7: 72,338,350 – 74,993,703
Kleefstra syndrome (9q34.3 deletion)	2	Chr9: 138,863,456 – 140,241,935 Chr9: 138,622,852 – 140,145,683
Angelman syndrome (maternal 15q11-q13 deletion)	2	Chr15: 21,219,452 – 26,199,055 Chr15: 20,307,869 – 26,208,602
Phelan-McDermid syndrome (22q13.3 deletion)	2	Chr22: 44,181,296 – 49,524,226 Chr22: 48,041,489 – 49,565,875
MECP2 duplication syndrome (Xq28)	2	ChrX: 152,806,398 – 153,266,347 ChrX: 152,765,830 – 153,266,347
1q43-q44 deletion syndrome	1	Chr1: 235,963,770 – 247,190,718
Terminal 6q deletion syndrome	1	Chr6: 161,333,762 – 170,748,662
Benign familial neonatal convulsions (KCNQ2 deletion, 20q13.33)	1	Chr20: 61,040,936 – 62,379,059
22q11.2 deletion syndrome	1	Chr22: 17,403,824 – 19,835,417
Total # of patients with identified genetic syndromes involving epilepsy as a result of chromosomal microarray testing	29/805 (3.6%)

Table 3.

CNVs in cohort in regions of known epilepsy hotspots.

Epilepsy “hotspot”	# cases	Coordinates (hg18)	Epilepsy Phenotype	Neuropsychiatric co- morbidities
1q21.1	1 dup	chr1: 144,984,952–146,317,921	Focal epilepsy with status epilepticus.	DD, learning problems, ADHD
15q11.2	1 dup, 1 del	Duplication: chr15: 20,307,869–20,773,190 Deletion: chr15: 19,121,757–20,851,679	Duplication case - focal seizures Deletion case – neonatal seizures	Duplication case - congenital hydrocephalus, DD, dysmorphic Deletion case – GDD
15q13.3	1 dup, 1 del	chr15: 29,803,656 – 30,298,096 chr15: 28,719,136 – 30,298,096	Duplication case - myoclonic astatic epilepsy Deletion case - primary generalized epilepsy	Duplication case - speech delay Deletion case - GDD, ID, autistic features
15q11-q13 duplication	4 dup	1. chr15: 18,362,555 – 30,232,544 2. chr15: 21,195,208 – 26,214,052 3. chr15: 21,208,377 – 26,208,602 4. chr15: 18,362,555 – 28,160,686 (4 copies) and 28,517,805 – 30686791 (3 copies)	Generalized, focal or mixed epilepsy	All with ID or autism
16p11.2	5 dup, 4 del	Duplications: 1. chr16: 29,572,030 −30,106,852 2. chr16: 29,258,288 – 30,106,808 3. chr16: 28,740,997 – 29,030,068 4. chr16: 21,476,688 – 29,238,851 5. chr16: 29,581,455 – 30,240,082 Deletions: 1. chr16: 2,958,202 – 30,106,252 2. chr16: 29,528,190 – 30,106,852 3. chr16: 28,394,123 – 29,238,851 4. chr16: 30308460 – 30,342,665	Duplication cases – focal, generalized or both with mixture of seizure types including spasms/drop attacks Deletion cases - focal or generalized seizures	Duplication cases– one with polymicrogyria; all with DD/ID; one with autism Deletion cases – all with GDD and/or learning problems; one with macrocephaly
16p13.11	2 dup, 3 del	Duplications: 1. chr16: 15,034,210 – 1619,973 2. chr16: 14,687,636 – 16,199,736 Deletions: 1. chr16: 14,876,356 – 16,174,751 2. chr16: 14,876,356 – 16,199,682 3. chr16: 15,399,028 – 16,199,682	Duplication cases – one with neonatal seizures; one with focal seizures onset at 2 years Deletion cases – generalized or focal epilepsy	Duplication cases – one with normal development; one with cortical dysplasia and DD/ID Deletion cases – all with DD/ID; one with autism spectrum disorder

Because association with epilepsy is most established with deletions, only the deletions were considered in this report as pathogenic CNVs explaining the epilepsy syndromes. ADHD = attention deficit hyperactivity disorder, DD = developmental delay, GDD = global developmental delay, ID = intellectual disability.

CNVs associated with known epilepsy-associated syndromes

Twenty-nine patients (3.6%) had CNVs with well-established associations with syndromic epilepsy (Table 1). These included 22q11.2 duplication syndrome (n=4), 1p36 deletion syndrome (n=3), Mowat-Wilson syndrome (n=3), Wolf-Hirschhorn syndrome (n=3), Dravet syndrome (n=2), Williams-Beuren region reciprocal duplication syndrome (n=2), Kleefstra syndrome (2), Angelman syndrome (n=2), Phelan-McDermid syndrome (n=2), MECP2 duplication syndrome (n=2), 1q43-q44 deletion syndrome (n=1), terminal 6q deletion syndrome (n=1), benign familial neonatal convulsions (n=1), and 22q11.2 deletion syndrome (n=1). All of these patients’ signs and symptoms were consistent with the reported phenotypes associated with these CNVs, and the syndromes were considered responsible for their epilepsy. The duplication 17p11.2 was not considered the cause of the epilepsy in three patients with Potocki-Lupski syndrome because patients with this syndrome are described as having abnormal EEGs without clinical epilepsy.¹⁵

CNVs involving other known epilepsy genes

Twenty-two patients (2.7%) had CNVs containing known epilepsy genes not included in the known syndromes above, of which seven (0.9%) are likely disease-associated (Table 2). These included one homozygous deletion of PLCB1¹⁶ and six heterozygous deletions involving CACNB4, CHRNA7, GABRA1, GABRG2, or PRRT2. Three of these seven (including CHRNA7 or PRRT2) overlap with the hotspots discussed below.

Table 2.

CNVs in cohort containing known epilepsy genes.

Epilepsy gene	Pathogenic CNV(#)	Coordinates (hg18)
PLCB1 (20p12.3)	1 deletion, homozygous	chr20: 8,047,741 – 8,523,461	Malignant migrating partial epilepsy of infancy (this case reported in Poduri et al. 2012)
CACNB4(2q23.3)	2 deletions, heterozygous	chr2: 152,254,443 – 156,539,284 chr2: 152,062,510 – 15,285,180	Case 1 - GTCs and asymmetric tonic seizures, EEG generalized Case 2 - GTCs, normal EEG
CHRNA7(15q13.3) *overlaps with hotspot	1 deletion, heterozygous	chr15: 28,719,136 – 30,298,096	Idiopathic generalized epilepsy (likely juvenile absence epilepsy) with moderate ID and hypotonia
GABRA1 and GABRG2(5q34)	1 deletion, heterozygous	chr5: 161,048,671 – 161,558,604	Mixed focal and generalized epilepsy with GTCs and focal seizures, generalized and multifocal epileptiform activity
PRRT2(16p11.2) *overlaps hotspot	2 deletions, heterozygous	chr16: 29,582,020 – 30,106,252 chr16: 29,528,190 – 30,106,852	Case 1 - GTCs onset within 1st years of life and controlled in setting of GDD/ID Case 2 - GTCs in first year of life in setting of GDD, hypotonia, macrocephaly
Epilepsy gene	Possibly pathogenic CNV		Epilepsy phenotype
NRXN1 (2p16.3)	2 deletions, heterozygous	chr2: 51,079,674 – 51,236,317 chr2: 50,392,247 – 50,782,724	Case 1 - Slumping and vomiting then generalized convulsive seizures with normal EEGs, in setting of autism Case 2 - Myoclonic seizures, GTCs, and focal seizures with mixed generalized and multifocal epileptiform activity in the setting of autism
Epilepsy gene	Not likely pathogenic, (duplication or heterozygous deletion in AR condition)		Epilepsy phenotype
CNTNAP2 (7q35- q36.1)	1 deletion, heterozygous	chr7: 145,919,587 – 146,373,456	Focal seizures in the setting of GDD and tic disorder, right central spikes on EEG
CLN3 (16p11.2) *overlaps with hotspot	1 deletion, heterozygous	chr16: 28,394,123 – 29,238,851	GTCs, focal onset in setting of GDD
GLDC(9p24.1)	1 deletion, 1 duplication	chr9: 6,410,789 – 6,631,818 chr9: 6,516,152 – 7,003,793	Case 1 (deletion) – Refractory generalized epilepsy with tonic seizures onset at 2 years in setting of PVNH and autism Case 2 (duplication) – Infantile spasms then focal seizures with GDD, hypotonia
KCNQ2 and CHRNA4(20q13.33)	1 duplication	chr20: 59,721,605 – 61,720,694	Landau-Kleffner variant
CHRNA7(15q13.3) *overlaps with hotspot	1 duplication	chr15: 29,803,656 – 30,298,096	Myoclonic-astatic epilepsy
CLN5(13q22.3)	1 duplication	chr13: 43,127,591 – 114,123,698	Nocturnal frontal lobe epilepsy in the setting of hypomelanosis of Ito, hypotonia and DD
GABRA1(5q34)	1 duplication	chr5: 161,130,609 – 161,317,552	Focal seizures (eye deviation and limp) in the setting of fever, normal EEG, + microcephaly
GAMT(19p13.3)	1 duplication	chr19: 848,315 – 1,942,330	Benign rolandic epilepsy (+ learning disability and hypotonia). Bilateral central-temporal spikes
NRXN1(2p16.3)	1 duplication	chr2: 50,412,268 – 50,729,054	GTCs, easily controlled, in the setting of autism
UBE3A(15q11.2) *overlaps with hotspot	4 duplications	chr15: 21,195,208 – 26,214,052 chr15: 21,208,377 – 26,208,602 chr15: 18,362,555 – 28,160,686 (4 copies) and 28,517,805 – 30,686,791 (3 copies) chr15: 18,362,555 – 30,232,544	Case 1 - Focal seizures + autism Case 2 - Generalized seizures + autism Case 3 - Intractable mixed epilepsy (drop attacks, focal seizures, myoclonic seizures, GTC seizures) with ID Case 4 - Generalized seizures + autism

Section 1 above are those that are likely causal of the epilepsy phenotype. Section 2 is a CNV possibly causal of the epilepsy phenotype. Those classified as possibly pathogenic are heterozygous deletions in NRXN1. Section 3 lists CNVs that are not likely to be pathogenic because they are 1) duplications in genes without known disease caused by duplications or 2) single deletions in genes associated with autosomal recessively inherited conditions. AR = autosomal recessive. DD = developmental delay, GDD = global developmental delay, GTC = generalized tonic clonic seizure, ID = intellectual disability, PVNH = periventricular nodular heterotopia.

Suspected pathogenic CNVs included one patient with migrating partial epilepsy of infancy and a homozygous deletion of PLCB1,¹⁶ a deletion previously associated with early onset epilepsy.^{16, 17} Two deletions included CACNB4, associated with susceptibility to idiopathic generalized epilepsy,¹⁸ in one patient with generalized epilepsy and another with generalized tonic-clonic seizures (GTCs) and normal EEG. One patient with deletion of chromosome 15q13.3, including CHRNA7 and located within the epilepsy “hotspot” 15q13.3,^{3, 19} presented with generalized epilepsy, global developmental delay (GDD) and autistic features, all described in patients with the typical 15q13.3 deletion.^19–23 One deletion of chromosome 5q34 involved the GABA-A receptor subunit-encoding genes GABRA1 and GABRG2. Both genes are associated with susceptibility for genetic generalized epilepsies (GGE), and GABRG2 has been identified in patients with febrile seizures, GEFS+ and Dravet syndrome.^{24, 25} The patient had focal and generalized seizures and generalized and multifocal epileptiform activity on EEG. Two deletions involved the gene PRRT2, included in the hotspot region 16p11.2 associated with ASD and DD.²⁶ This gene has been associated with benign familial infantile epilepsy and paroxysmal kinesigenic dyskinesia;²⁷ our two cases had GTCs within the first year of life and GDD, but no movement disorder at last follow-up (2 and 8 years).

Two heterozygous deletions involving NRXN1 (2p16.3) were considered possibly associated with epilepsy. Homozygous loss of function of NRXN1 is associated with Pitt Hopkins-like syndrome with epilepsy; heterozygous loss of function of NRXN1 has been associated with susceptibility to GGE.^28–33 Both of our patients had GTCs, and one also had myoclonic and focal seizures; both had ASD, also previously associated with NRXN1.^{32, 34}

Some CNVs involving epilepsy genes were not thought to be sufficient to cause disease, including three heterozygous deletions of genes known to cause autosomal recessive conditions (CNTNAP2, CLN3, GLDC) and ten duplications of genes associated with epilepsy only in the setting of loss of function (KCNQ2 and CHRNA4, GABRA1, GAMT, NRXN1, UBE3A, GLDC) (Table 2). Sequencing of genes associated with autosomal recessive conditions was not done.

Known epilepsy hotspots identified by CMA

Twenty-three patients (2.9%) had CNVs in the epilepsy hotspots 1q21.1, 15q11.2, 15q13.3, 15q11-q13, 16p11.2, and 16p13.11. The epilepsy phenotypes and neuropsychiatric comorbidities of these patients are summarized in Table 3. In some instances, the phenotypes are similar to those previously reported, especially in the patients with generalized epilepsy.^{1–3, 21, 35}

Our series contained one duplication of 1q21.1 in a patient with focal epilepsy as well as DD and learning difficulties. Previous reports have described neurodevelopmental disabilities and epilepsy in association with deletions of 1q21.1, but the association with duplications is less established.^{1, 3, 36, 37}

One patient with a deletion of 15q11.2, in the BP1-BP2 region, had neonatal seizures and GDD. One case with a duplication of 15q11.2 had congenital hydrocephalus, and the relative role of the CNV versus the hydrocephalus in this patient’s epilepsy can-not be determined. Duplications at this locus are seen both in patients with epilepsy and in the general population.³

The epilepsy candidate gene CHRNA7 lies on chromosome 15q13.3. One patient with a duplication in this region had myoclonic astatic epilepsy (Doose syndrome), not previously reported in association with this CNV. One patient with a deletion is described above;^{3, 19, 21} duplications have not previously been associated with epilepsy.^{3, 19, 21}

Four duplications in the region 15q11-q13 were detected in patients with generalized, focal, or mixed epilepsy and ID or ASD.

In the 16p11.2 region, we identified five duplications and four deletions in our series (Table 3). The first two duplications and the first two deletions include the gene PRRT2, described above. Patients with duplications exhibited a mixture of seizures types notably including spasms and drop attacks (not previously reported). One patient had polymicrogyria. The patients with deletions had focal or generalized seizures. One patient had macrocephaly. All patients with CNVs in this region had DD or learning problems.

In the 16p13.11 region, we identified two patients with duplications and three cases with deletions. One patient with a duplication had neonatal seizures and normal development, and the other had focal epilepsy, ID, and a cortical dysplasia. The patients with deletions had a mixture of generalized and focal epilepsy and ID; one had ASD. Deletions of 16p13.11 are a known risk factor for both focal and generalized epilepsies; the role of duplications as a risk factor for epilepsy is not established.^{1, 2, 38}

Additional regions with overlapping CNVs in our epilepsy series

Overlapping CNVs were found in numerous additional regions (Figure 3). One of these regions, Xp22.31, was identified as an area of interest because of multiple overlapping CNVs in our series and reports of patients with epilepsy. Extending the time frame from October 2006 to August 2012, we identified ten unrelated patients with epilepsy and overlapping CNVs in the Xp22.3 region (4 deletions, 6 duplications), ranging in size from 311 kb to 9.6 Mb (Table 4, Figure 4). Four had additional CNVs; the patients were a combination of males and females (Supplemental Tables 1 and 2). One of the deletions was de novo; the rest of the CNVs were inherited or of unknown inheritance. One of the duplication cases had another etiology for her epilepsy, a pathogenic missense mutation in the gene PCDH19. The other nine cases had no other identified etiology for epilepsy. The minimal region of overlap for the patients with deletions and epilepsy in this series was chrX: 7,061,905–7,920,059.

Figure 3.

Regions of overlapping CNVs in our series, not including known epilepsy genes, syndromes involving epilepsy, or epilepsy “hot spots.”

Table 4.

Epilepsy phenotypes in cases with CNVs at Xp22.31.

Case	Xp22.31 deletions, loci and if de novo	Other CNVs	Gender	Age at sz onset	Focal or generalized epilepsy	Seizure types	EEG	Epilepsy syndrome
Deletion 1	chrX:7,061,905- 7,920,059 (859 kb). De novo FISH confirmed	Xq22.1 deletion (452 kb, de novo), 2p11.2 deletion (33kb, inherited)	F	2 wks	Focal	Tonic, focal +/− 2ndary generalized	1–2 month – Intermittent slowing right or left temporal, sharp waves right frontal and temporal, seizures right frontotemporal onset 9–10 months - normal	None, neonatal/infantile seizures
Deletion 2	chrX:6,477,006- 8,091,810 (1.6 Mb), unknown inheritance FISH confirmed	None	M	5 yrs	Unknown	1 GTC and abnormal EEG	“epileptiform activity”	None
Deletion 3	chrX:6,467,403- 8,091,751 (1.6 Mb), unknown inheritance	None	M	4 yrs	Focal	Focal tonic or clonic +/− 2ndary generalization	Bilateral central- temporal spikes	Rolandic epilepsy
Deletion 4	Xp22.33-p22.2 deletion chrX:2,710,316- 12,342,188 (9.6 Mb), presumed de novo	Xp22.2- q28 duplication (142 Mb)	F	11 mo	Focal	focal, GTC	Left posterior quadrant slowing, generalized slowing, epileptiform activity left temporal and occipital or bi- posterior quadrants	None
	Xp22.31 duplications, loci and if de novo
Duplication 1	chrX: 8,413,301- 8,724,726 (311 kb, maternal inheritance)	4p16.1 deletion (202 kb), de novo	F	21 mo	Focal and Generalized	GTC, Absence, Focal (hemi- clonic)	Multifocal or generalized epileptiform activity	Dravet-like onset and consistent with known mutation in PCDH19
Duplication 2	chrX:7,788,165- 8,385,637 (597 kb), paternal inheritance	None	F	3 1/2 yrs	Focal	GTC, focal	Intermittent bi- occipital slowing left more than right, slow PDR Normal at 6 yrs	None
Duplication 3	chrX:6,455,262- 8,091,810 (1.6 Mb), unknown inheritance	None	F	2 mo	Generalized	Tonic, absence, myoclonic, drop attacks	Multifocal and generalized epileptiform activity, poorly sustained PDR, lack of normal sleep features; Frequent electrographic seizures (staring and myoclonic jerks) with generalized spike and waves followed by decrements associated with jerks	Not well defined, refractory myoclonic epilepsy
Duplication 4	chrX:6,463,313- 8,091,751 (1.6 Mb), paternal inheritance FISH confirmed	Paternally inherited 4q32.3 duplication (609 kb)	F	1st week of life	Focal	Clonic (facial twitching, never captured on EEG)	Intermittent slowing left tempo-parietal at 8 yrs Normal as a neonate and at 9 yrs	Neonatal seizures
Duplication 5	chrX:6,467,403- 8,091,751 (1.6 Mb), maternal inheritance	None	F	5 mo	Focal	Febrile SE, focal with 2ndary generalization, myoclonic	Most normal, one with background slowing	None
Duplication 6	chrX:6,461,822- 8,075,193 (1.7 Mb) Unknown inheritance	None	M	1 mo	Focal	Clonic	Normal	None

GDD = global developmental delay, GTC = generalized tonic clonic seizure, mo = month, SE=status epilepticus, Yrs = years, Wks = weeks.

Figure 4.

Cases with epilepsy and deletions involving Xp22.31 in our October 2006 - August 2012 series, in the literature, and in the Decipher database.

In Table 4, we describe the epilepsy phenotypes: focal epilepsy in seven patients (3 deletions, 4 duplications), generalized epilepsy in one patient (duplication), focal plus generalized epilepsy in one patient (duplication), and not clearly defined as focal or generalized in one patient (deletion). The predominant seizure types were focal or GTCs. Age at seizure onset ranged from one week to five years. MRIs of the brain were normal except in two cases. One patient with a deletion of Xp22.3 had isodicentric X chromosome with duplication Xp22.2-q28 and deletion Xp22.33-p22.2. This patient had pons and cerebellar vermis hypoplasia, generalized volume loss, white matter T2 signal change, and malrotation of the hippocampi. One patient with a duplication had cerebellar hypoplasia. All of the patients had DD and ID; one had ASD. Other associated abnormalities included dysmorphic features (three deletions, one duplication) and hypotonia (three deletions, three duplications). There was no clear phenotypic distinction between patients with deletions and duplications or between males and females, and no strong evidence of epilepsy in relatives carrying the CNV (Supplemental Tables 1 and 2).

CNVs in this region were identified in 31 patients without epilepsy in the same time period. Of these 31 patients, seven had deletions (four males, three females). Fifty-eight percent (18/31) of the patients without epilepsy had DD and/or ASD. Phenotypic features of all patients with Xp22.31 deletions or duplications, with and without epilepsy, are reviewed (Supplemental Tables 1 and 2).

Alternate genetic etiologies identified in some patients with CNVs

In some patients, genetic etiologies other than the identified CNVs were identified for the epilepsy. For example, in one patient with a maternally inherited 17q12 duplication, a pathogenic mutation was identified in the gene CDKL5, and her phenotype was consistent with this gene mutation.¹¹ One patient with four CNVs was found to have neuronal ceroid lipofuscinosis, identified by low PPT1 enzyme activity on skin biopsy. One patient with an inherited 17p13.3 deletion (not involving LIS1) was later found to have a likely pathogenic KCNQ2 mutation, and as above one patient with an Xp22.31 duplication had a pathogenic PCDH19 mutation. No alternate genetic etiology was identified in any of the patients whose CNVs were classified as likely pathogenic.

CNVs identify potential novel candidate genes for epilepsy

By evaluating the gene content of the CNVs in this series and performing pathway analysis using Ingenuity Pathway Analysis (IPA), we identified numerous potential candidate genes for epilepsy involved in neurodevelopment, synaptic transmission, ion channels, or pathways overlapping with known epilepsy genes. A summary of candidate genes identified in de novo CNVs is provided (Table 5). Additional candidate genes in CNVs of unknown inheritance >500 kb in size include EPHB1, KY, CTNNA, TDRDN, and NKAIN2 in duplications and HOMER2 in a deletion.

Table 5.

Additional candidate genes for epilepsy in confirmedde novo CNVs.

Deletions
AGRN	GALR1	MAPK8IP2	PARK2	RTN4R
AKT3	GRIN1	MBD5	PI4KA	SNAP29
ARID1B	JAKMIP1	MBP	PI4KAP1	THBS2
CACNA1B	KCNJ3	NIPA1	PMP22	WNT7B
DVL1	KIF5C	PAFAH1B1	PRKCZ
Duplications
APBA3	EFNB3	MYH10	PCDH gene cluster
ARHGEF7	EGR3	NLGN2	PPP3CC
COL4A1	FGF14	NSTR1	PURA
DLG4	GABARAP	P2RX1	SOX6
DOC2A	MAPK3	P2RX5	SH3GL1

A) deletions and B) duplications. These include genes coding for ion channels, proteins involved in structure or organization of the synapse or in synaptic transmission of neurotransmitters, gene regulatory proteins, and proteins involved in neurodevelopment. The PCDH gene cluster on chromosome 5q31.2-q22 includes the following genes: PCDHA1, PCDHA2, PCDHA3, PCDHA4, PCDHA7, PCDHA8, PCDHA9, PCDHA10, PCDHA11, PCDHA12, PCDHA13, PCDHAC1, PCDHAC2, PCDHB1, PCDHB2, PCDHB3, PCDHB4, PCDHB5, PCDHB6, PCDHB7, PCDHB8, PCDHB9, PCDHB10, PCDHB11, PCDHB12, PCDHB13, PCDHB14, PCDHB15, PCDHB16, PCDHB17, PCDHB18, PCDHB19P, PCDHGA1, PCDHGA2, PCDHGA3, PCDHGA4, PCDHGA5, PCDHGA6, PCDHGA7, PCDHGA8, PCDHGA9, PCDHGA10, PCDHGA11, PCDHGA12, PCDHGB1, PCDHGB2, PCDHGB3, PCDHGB4, PCDHGB5, PCDHGB6, PCDHGB7, PCDHB8P, PCDHGC3, PCDHGC4, PCDHGC5, PCDH1.

Discussion

At least 5% of pediatric epilepsy patients in our series harbor pathogenic CNVs, and additional patients harbor possibly pathogenic CNVs. While more directed testing for specific epilepsy genes or genetic syndromes is indicated for some patients based on clinical phenotype, CMA is an excellent diagnostic tool for unexplained epilepsy as it can identify structural genomic rearrangements that contribute to epileptogenesis. CMA may identify genetic causes or risk factors for epilepsy among patients for whom the phenotype does not suggest a more focused diagnostic evaluation or patients for whom the full phenotype has not yet manifested at the time when the child presents with epilepsy. We advocate for caution in attributing the etiology of epilepsy to novel CNVs. It is possible that some CNVs serve as risk factors or modifiers of disease in combination with other factors, genetic and non-genetic. More evidence is needed to determine the pathogenicity of more rare CNVs with epilepsy.

We identified patients with well-defined deletion/duplication syndromes with epilepsy included in the phenotypes. Our patients with CNVs involving known epilepsy genes had epilepsy phenotypes consistent with the literature. Two patients with heterozygous deletions of the gene NRXN1 provide further support for a gene dose effect with increased risk for epilepsy even with single copy disruptions.^28–33 Further study of other CNVs identified in patients with epilepsy compared to control populations may help to further elucidate the role of chromosomal regions and individual genes in epileptogenesis and to clarify the range of associated epilepsy phenotypes.

Our data show that the epilepsy “hotspots” 1q21.1, 15q11.2, 15q13.3, 15q11-q13, 16p11.2, and 16p13.11 have relevance in a clinical setting. Prior studies identifying these hotspots focused on GGE or idiopathic focal epilepsies, but some of our patients harboring these CNVs had other epilepsy phenotypes, expanding the clinical spectrum of these CNVs. One patient with a 15q13.3 duplication involving the gene CHRNA7 had myoclonic astatic epilepsy, a novel association that needs to be replicated in additional patients to be considered robust. Two patients had neonatal seizures, one with a 15q11.2 deletion and one with a 16p13.11 duplication. It is likely, in fact, that there is considerable phenotypic variability associated with these and other CNVs, as is proving to be the case for mutations in many of the known epilepsy genes. These data thus provide support for the use of CMA as a diagnostic test in a wide range of epilepsy phenotypes if the etiology is unknown.

We reviewed the Xp22.31 region in depth. Contiguous gene syndromes caused by deletions in the region of Xp22.3 are well established.^{39, 40} While DD and ID are commonly reported,^39–42 epilepsy is reported in only a small number of patients with CNVs involving this region.^42–51 In our series, we observed epilepsy in 24% of the patients with CNVs in this region, and the CNVs overlap with those of patients with epilepsy reported in the literature and in the DECIPHER database. The critical overlapping region in patients with seizures in our series (chrX: 7,061,905–7,920,059, overlapping with most but not all cases in the literature) is not an identified hotspot for deletions in the general population, based on the Database of Genomic Variants (DGV, http://dgv.tcag.ca/dgv/app/home, accessed 9/30/13). This study provides additional evidence that CNVs in the Xp22.31 region may predispose to epilepsy, albeit with variable penetrance. Further evidence from additional cases and controls is required to determine the strength of this relationship.

Because this is an observational study, there may have been a bias toward patients with epilepsy who also had multiple congenital anomalies and/or neurodevelopmental disorders in addition to epilepsy, increasing their likelihood of having pathogenic CNVs.¹³ Thus, the yield of CMA in epilepsy overall may be overestimated by this series. However, the group that we studied is representative of the patient population seen in referral centers for pediatric neurology, where much of the genetic testing for epilepsy occurs.

The diagnostic yield we observed is comparable to the yield in other clinical indications for CMA, including ID, ASD, and prenatal diagnosis.^{4, 38, 52–55} Based on available evidence, a consensus statement in 2010 estimated that the diagnostic yield of CMA for patients with unexplained DD/ID, ASD, or multiple congenital anomalies is about 15–20%.⁵² Prospective studies of patients with GGE, idiopathic focal epilepsies or epileptic encephalopathies have provided similar yields, and there is evidence that up to 10% of patients with epilepsy and ID have disease associated CNVs.^{1–4, 56}

The advantages to the patient and clinician of having an identified CNV include the following: 1) ending the often long diagnostic process, 2) providing families with prognostic information and additional features of the syndrome that may be medically relevant, and 3) in some cases providing families access to syndrome-specific research or support resources and organizations. It less often directly impacts epilepsy treatment, but as chromosomal disorders are better defined, additional prospective data on efficacy of anti-epileptic medications in each syndrome can be collected.

In conclusion, CNVs play a causative role in some cases of pediatric epilepsy. Coupled, when appropriate, with evaluation of one or more epilepsy genes, or in some cases whole exome sequencing, CMA should be included in the diagnostic evaluation of patients with unexplained epilepsy. Specifically, we recommend CMA as part of the initial evaluation of unexplained epilepsy, particularly among patients with associated dysmorphic features, DD/ID, ASD, malformations, or a history of parental consanguinity. Furthermore, identifying regions of overlapping CNVs between patients with epilepsy is a starting point toward identifying candidate genes and additional genetic causes of epilepsy.