Genetic Basis of Brain Malformations

The development of the human cerebral cortex is a complex dynamic process that occurs during several gestational weeks [Gleeson and Walsh, 2000]. During the first stage, stem cells proliferate and differentiate into young neurons or glial cells deep in the forebrain, in the ventricular and subventricular zones lining the cerebral cavity. During the second stage, cortical neurons migrate away from their place of origin: most cells migrate along the radial glial fibers from the periventricular region towards the pial surface, where each successive generation passes one another and settles in an inside-out pattern within the cortical plate. When neurons reach their final destination, they stop migrating and order themselves into specific ‘architectonic' patterns guiding cells to the correct location in the cerebral cortex. This third phase involves the final organization within the typical 6 layers of the cortex, associated with synaptogenesis and apoptosis [Barkovich et al., 2012].

Abnormal cortical development is increasingly recognized as a cause of developmental disabilities and epilepsy. This recognition is largely due to improved magnetic resonance imaging (MRI) resolution, which makes it possible to assess the distribution and depth of cortical sulci, cortical thickness, the boundaries between grey and white matter, and variations in signal intensity. Abnormalities of any or all of these features may be observed in different malformations of cortical development (MCD), which may be restricted to discrete cortical areas or may, alternatively, be diffuse [Guerrini et al., 2008; Guerrini and Dobyns, 2014].

A classification scheme has been developed categorizing MCD into 3 major groups that recapitulate the main developmental steps as malformations of cell proliferation, neuronal migration, or postmigrational cortical organization and connectivity. Although the classification of MCD has advanced substantially during the past decade, in practice only a few categories are used, including lissencephaly (LIS), polymicrogyria, schizencephaly, focal cortical dysplasia (FCD), and periventricular nodular heterotopia (PNH). However, emerging evidence suggests that MCD are far more heterogeneous than this classification suggests [Guerrini and Dobyns, 2014].

So far, more than 100 genes have been associated with one or more types of MCD. The biological pathways include cell-cycle regulation at many steps (especially mitosis and cell division), apoptosis, cell-fate specification, cytoskeletal structure and function, neuronal migration and basement-membrane function, and many inborn errors of metabolism. A subset of MCD genes - especially those associated with megalencephaly - are associated with postzygotic (i.e., mosaic) mutations [Lee et al., 2012].

Genetic testing needs accurate assessment of imaging features and familial distribution, if any, and can be straightforward in some disorders but requires a complex diagnostic algorithm in others. Because of substantial genotypic and phenotypic heterogeneity for most of these genes, a comprehensive analysis of clinical, imaging, and genetic data is needed to properly define these disorders. Exome sequencing and ultra-high field MRI are rapidly modifying the classification of these disorders [Guerrini and Dobyns, 2014].

In this themed issue dedicated to the genetic of epilepsy, we limited our contribution to those malformations that have more frequently been associated with epilepsy.

LIS or smooth brain and the associated malformation known as subcortical band heterotopia (SBH) are the classic malformations associated with deficient neuronal migration [Dobyns et al., 2012].

LIS (from the Greek words ‘lissos' meaning smooth and ‘enkephalos' meaning brain) is a neuronal migration disorder characterized by absent (agyria) or decreased (pachygyria) convolutions, cortical thickening, and a smooth cerebral surface (fig. 1A) [Barkovich et al., 2012; Guerrini and Dobyns, 2014]. Different subtypes of LIS are readily distinguished based on the number of cortical layers affected and include 2-layered, 3-layered, and 4-layered forms [Forman et al., 2005]. The most common, classical LIS (4-layered form), features a very thick cortex (10-20 mm vs. the normal 4 mm) and no other major brain malformations. The cytoarchitecture consists of 4 primitive layers, including an outer marginal layer, which contains Cajal-Retzius neurons (layer 1); a superficial cellular layer, which contains numerous large and disorganized pyramidal neurons (layer 2) corresponding to the true cortex, a variable cell-sparse layer (layer 3), and a deep cellular layer (composed of medium and small neurons), which extends more than half the width of the mantle (layer 4) [Golden and Harding, 2004]. SBH is a related disorder in which bands of grey matter are interposed in the white matter between the cortex and the lateral ventricles (fig. 1B) [Guerrini and Parrini, 2010]. Histopathology demonstrates that heterotopic neurons settle close to the cortex in a pattern suggestive of laminar organization.

The severity of LIS and SBH varies from complete or nearly complete agyria (grades 1 and 2), to mixed agyria-pachygyria (grade 3), to pachygyria only (grade 4), to mixed pachygyria-SBH (grade 5), and finally to SBH only (grade 6) [Dobyns et al., 2012]. Based on genetic findings, the full range of LIS now extends from severe LIS with cerebellar hypoplasia to classic LIS to SBH, and also includes a polymicrogyria-like cortical malformation that can be distinguished from both LIS and typical polymicrogyria by high-resolution brain imaging.

Children with the most common types of LIS (or SBH) typically appear normal as newborns. Most affected children come to medical attention during the first year of life due to neurological deficits in the first weeks or months [Barkovich et al., 2012]. The major medical problems encountered are ongoing feeding problems and epilepsy of many different types that are often intractable.

Classical LIS is rare, with a prevalence of about 12 per million births. Patients with severe LIS have early developmental delay, early diffuse hypotonia, later spastic quadriplegia, and eventual severe or profound mental retardation. Seizures occur in over 90% of LIS children, with onset before 6 months in about 75% of the cases [Guerrini and Filippi, 2005]. Between 35 and 85% of children with classic LIS develop infantile spasms, often without classic hypsarrhythmia. Most LIS children will subsequently continue to have epileptic spasms in association with other seizure types. The EEGs show diffuse, high-amplitude, fast rhythms, which are considered to be highly specific for this malformation.

The main clinical manifestations of SBH are mental retardation and epilepsy [Barkovich et al., 1994]. Epilepsy is present in almost all patients and is intractable in about 65% of the cases. About 50% of these epilepsy patients have focal seizures, and the remaining 50% have generalized epilepsy, often within the spectrum of Lennox-Gastaut syndrome. Those with more severe MRI abnormalities have significantly earlier seizure onset and are more likely to develop Lennox-Gastaut syndrome [Guerrini and Filippi, 2005].

Children with some LIS syndromes (especially Miller-Dieker syndrome, and severe forms of LIS with cerebellar hypoplasia or the X-linked syndrome of LIS with abnormal genitalia) have a severe course, and the most severe forms have high mortality rates. However, these data do not apply to children with less severe forms of LIS, SBH, or LIS with cerebellar hypoplasia, as all of these disorders are associated with better motor and cognitive function and longer survival [Dobyns et al., 2012].

LIS, SBH, and LIS with cerebellar hypoplasia always have a genetic cause. Studies to date have identified 14 LIS genes (table 1), which account for roughly 90% of patients. However, 2 major genes have been associated with classic LIS and SBH. The LIS1 gene causes the autosomal form of LIS [Reiner et al., 1993], while the DCX gene is X linked [des Portes et al., 1998; Gleeson et al., 1998]. Although each gene can result in either LIS or SBH, most cases of classic LIS are due to deletions or mutations in the LIS1 gene [Mei et al., 2008], whereas most cases of SBH are due to mutations in the DCX gene [Matsumoto et al., 2001]. LIS1 encodes a 45-kDa protein (PAFAH1B1), which functions as a regulatory subunit of platelet-activating factor acetylhydrolase (PAF-AH) [Hirotsune et al., 1998]. PAFAH1B1 heterozygous mutant mice show a dose-dependent histopathological disorganization of cortical lamination as well as hippocampal and cerebellar cortical defects [Hirotsune et al., 1998]. DCX encodes a 40-kDa microtubule-associated protein (DCX) that is expressed in migrating neuroblasts [Gleeson et al., 2000a]. The DCX protein contains 2 tandem-conserved repeats. Each repeat binds to tubulin, and both repeats are necessary for microtubule polymerization and stabilization.

LIS1-related LIS is more severe in the posterior brain regions (posterior-to-anterior gradient), whereas DCX-related LIS is more severe in the anterior brain (anterior-to-posterior gradient). About 60% of patients with posterior-to-anterior isolated LIS carry genomic alterations or mutations involving LIS1 [Mei et al., 2008]. A simplified gyral pattern in the posterior brain, with underlying SBH, has been associated with mosaic mutations of LIS1 [Sicca et al., 2003]. Miller-Dieker syndrome is characterized by severe 4-layered LIS with diffuse agyria and no clear gradient, a typical facial appearance (prominent forehead, bitemporal hollowing, short nose with upturned nares, protuberant upper lip and a small jaw), and other birth defects (e.g., heart malformations) [Cardoso et al., 2003]. Miller-Dieker syndrome is caused by a deletion in LIS1 and contiguous genes in the 17p13.3 region. Deletion of 2 additional genes, CRK and YWHAE, telomeric to LIS1, may contribute to the most severe LIS grade and dysmorphic features [Cardoso et al., 2003].

Most DCX mutations cause anterior-to-posterior SBH/pachygyria. Mutations in DCX have been found in all reported pedigrees and in 80% of sporadic females and 25% of sporadic males with SBH [Matsumoto et al., 2001]. Genomic deletions in the DCX gene have been identified in females with sporadic SBH and in males with X-linked LIS [Mei et al., 2007]. Maternal germline or mosaic DCX mutations may occur in about 10% of the cases of either SBH or X-linked LIS [Gleeson et al., 2000b]. Hemizygous males with DCX mutations have classical LIS, but rare boys with missense mutations and anteriorly predominant SBH and rare females with DCX mutations and normal brain MRI have been described [Guerrini et al., 2003]. Rarer forms of LIS have been identified.

X-linked LIS with absent corpus callosum and ambiguous genitalia (XLAG) results from mutations in the ARX gene [Kato et al., 2004]. Affected patients show LIS with a posterior-to-anterior gradient, absent corpus callosum, a moderate increase of cortical thickness (only 6-7 mm), atrophic striatal and thalamic nuclei, postnatal microcephaly, neonatal-onset epilepsy, hypothalamic dysfunction including deficient temperature regulation, chronic diarrhea, and ambiguous genitalia with micropenis and cryptorchidism. Most XLAG patients die within one year after birth [Okazaki et al., 2008].

Autosomal recessive LIS with cerebellar hypoplasia consists of mild frontal predominant LIS, plus severe hippocampal and cerebellar hypoplasia and dysplasia. This pattern has been associated with mutations in RELN [Hong et al., 2000] or VLDLR (an essential cell-surface receptor for reelin) (fig. 1C, D) [Boycott et al., 2005].

Mutations in the ACTB and ACTG1 genes are associated with Baraitser-Winter cerebrofrontofacial syndrome which is characterized by the combination of hypertelorism, a broad nose with a large tip and prominent root, congenital nonmyopathic ptosis, ridged metopic suture, arched eyebrows, iris or retinal coloboma, sensorineural deafness, shoulder girdle muscle bulk, and progressive joint stiffness [Rivière et al., 2012a; Di Donato et al., 2014; Verloes et al., 2015]. Many patients with ACTB and ACTG1 mutations have pachygyria with an anterior-to-posterior severity gradient, similar to that observed in males with DCX mutations [Rivière et al., 2012a; Verloes et al., 2015].

A mutation in the CDK5 gene has been identified in 4 individuals with LIS and cerebellar hypoplasia born from a highly consanguineous family [Magen et al., 2015].

In patients with classic LIS, cytogenetic and molecular investigations are part of the diagnostic process. When Miller-Dieker syndrome is suspected, array-CGH analysis is indicated. When isolated LIS is diagnosed, careful assessment of the antero-posterior gradient of the abnormal cortical pattern will suggest whether to investigate LIS1 or DCX. When LIS is more severe posteriorly, it is worth performing MLPA first to rule out LIS1 deletions/duplications. If a deletion/duplication is not found, LIS1 sequencing should then be performed. In boys whose MRI shows more severe pachygyria in the frontal lobes, sequencing of the DCX gene is indicated. In patients with SBH, direct sequencing of DCX should be performed. If a DCX mutation is not found, MLPA analysis then should be performed. Direct sequencing is also indicated in the mothers of patients harboring a DCX mutation or other female relatives. Analysis of the ACTB and ACTG1 genes should be performed in patients with frontal predominant pachygyria who are not harboring DCX mutations.

All reported LIS1 alterations are de novo. Given the theoretical risk of germline mosaicism in either parent (which has never been demonstrated for LIS1), a couple with a child with LIS is usually given a 1% recurrence risk. When a DCX mutation is found in a boy with LIS, mutation analysis of DCX should be extended to the proband's mother, even if her brain MRI is normal. If the mother is a mutation carrier, the mutation will be transmitted according to Mendelian inheritance. If the mother is not a carrier, there still is a risk of germline mosaicism, which is roughly estimated ∼5%.

Mutations of tubulin genes were first reported as causing LIS (for TUBA1A) [Keays et al., 2007] or polymicrogyria (for TUBB2B) [Jaglin et al., 2009]. However, the cortical malformations seen in most individuals with mutations of tubulin or tubulin motor genes comprise a wide spectrum of morphological abnormalities whose characteristics are at times distinctive but can also overlap with those of LIS and polymicrogyria by brain imaging and neuropathology [Cushion et al., 2013; Poirier et al., 2013]. The full range of these malformations vary from extreme LIS with completely absent gyri, total agenesis of the corpus callosum, and severe cerebellar hypoplasia; to less severe LIS with moderate-to-severe cerebellar hypoplasia; to classic LIS, and to an atypical polymicrogyria-like cortical malformation with cerebellar hypoplasia [Cushion et al., 2013; Poirier et al., 2013]. Cortical thickness varies and can be mildly thin, normal, or mildly thick. Most children with mutations of tubulin genes have severe intellectual disability and intractable seizures.

Nine genes (KIF2A, KIF5C, TUBA1A, TUBA8, TUBB, TUBB2B, TUBB3, TUBG1, and DYNC1H1; table 1) associated with tubulinopathies have been identified so far [Bahi-Buisson et al., 2014]. Findings from functional studies suggest that abnormal brain development in tubulinopathies results from a dominant negative effect of heterozygous missense mutations (in the absence of loss-of-function mutations) on the regulation of microtubule-dependent mitotic processes in progenitor cells, and on the trafficking activities of the microtubule-dependent molecular motors KIF2A, KIF5C, and DYNC1H1 in postmitotic neuronal cells [Poirier et al., 2013].

The most severe phenotype of tubulinopathies consists of LIS with cerebellar and brainstem hypoplasia. These children all have profound deficits, intractable epilepsy and short survival. Heterozygous missense mutations in the TUBA1A gene have been found in this syndrome, with TUBA1A accounting for ∼30% of the patients (fig. 1E-G) [Kumar et al., 2010; Cushion et al., 2013]. A less severe phenotype consists of moderate LIS with cerebellar hypoplasia with pachygyria, with callosal defects and moderate hypoplasia of the brainstem and cerebellum [Morris-Rosendahl et al., 2008; Kumar et al., 2010]. A third phenotype has isolated LIS sequence with posterior predominant LIS matching the LIS1 pattern [Kumar et al., 2010]. Finally, some patients exhibit polymicrogyria plus variable agenesis of the corpus callosum as well as constant hypoplasia of the brainstem and cerebellum [Kumar et al., 2010].

Testing is available for all of the tubulin and tubulin motor genes so far identified. For all but TUBA8, only de novo heterozygous missense mutations have been found. Several families with 2 affected sibs have been reported, explained by low-level gonadal mosaicism in one parent [Zillhardt et al., 2016]. The phenotype (if any) associated with truncation or deletion mutations of most of the tubulin genes is unknown. Probably, these genes are intolerant to loss-of-function mutations, as also suggested by the observation in the ExAC Database (http://exac.broadinstitute.org/); the tubulin genes, with the exception of TUBA8, have a probability of loss-of-function intolerance near to 1.00, a value consistent with an almost complete intolerance. A single family with a tubulinopathy phenotype has been reported with a homozygous mutation of TUBA8 [Abdollahi et al., 2009].

There are 3 main groups of heterotopia: periventricular (usually nodular: PNH), subcortical and leptomeningeal (glioneuronal heterotopia found over the surface of the brain), of which only the first 2 can be detected by imaging. PNH is by far the most frequent. SBH is a mild form of LIS and has been dealt with in a previous section.

PNH consists of nodules of grey matter located along the lateral ventricles with a total failure of migration of some neurons [Barkovich et al., 2012; Guerrini and Dobyns, 2014]; it ranges from isolated, single, to confluent bilateral nodules (fig. 1H). The overlying cortex may show an abnormal organization. When the nodules are bilateral and numerous, a genetic basis is probable, and associated brain malformations are often reported [Parrini et al., 2006].

Brain Imaging

Patients with the classic X-linked PNH typically have bilateral contiguous nodules that spare the temporal horns and mild cerebellar vermis hypoplasia with mega-cisterna magna [Parrini et al., 2006]. Patients with rare autosomal recessive bilateral PNH can have severe congenital microcephaly and thin overlying cortex with abnormal gyri [Sheen et al., 2004]. In the fairly common posterior predominant syndromes, PNH is limited to the trigones, temporal and occipital horns, and can be associated with overlying polymicrogyria, hippocampal and cerebellar hypoplasia, or hydrocephalus [Pisano et al., 2012].

Clinical Features

Although most patients with PNH come to medical attention because they have focal epilepsy of variable severity, there is a wide spectrum of clinical presentations, including several syndromes with intellectual disability and dysmorphic facial features. There is some correlation between the size of PNH and the likelihood of concomitant structural abnormality of the cortex and clinical severity [Parrini et al., 2006], but there seems to be no correlation between the size and number of heterotopic nodules and cognitive outcome or epilepsy severity. Most females with PNH due to FLNA mutations have epilepsy, with normal or borderline cognitive level. Age at seizure onset is variable. Most patients have focal seizures, which can be easily controlled or refractory [Parrini et al., 2006]. There is no clear relationship between epilepsy severity and the extent of nodular heterotopia.

Genetic Basis and Diagnosis

The most common syndromic form of PNH (X-linked PNH) consists of bilateral contiguous or near contiguous nodules and is most often caused by FLNA mutations. All other PNH syndromes are rare.

X-linked PNH is a clinically and genetically heterogeneous disorder occurring most frequently in women, associated with high rates of prenatal lethality in male foetuses, and a 50% recurrence risk in the female offspring. Almost 100% of the families and 26% of sporadic patients harbor mutations in the FLNA gene [Parrini et al., 2006], which also causes cardiovascular abnormalities in some patients of both sexes and gut malformations in boys. Only a few male patients with PNH owing to FLNA mutations have been reported [Guerrini et al., 2004]. FLNA encodes a large actin-binding phosphoprotein that stabilizes the cytoskeleton and contributes to the formation of focal adhesions along the ventricular epithelium [Fox et al., 1998]. FLNA may be required for the initial attachment of neurons onto the radial glial scaffolding before migration from the ventricular zone [Lu et al., 2006]. Failure of migrating neurons to attach to radial glia is one likely mechanism leading to the formation of heterotopia.

A rare recessive form of PNH owing to mutations of the ARFGEF2 gene was described in 2 consanguineous pedigrees [Sheen et al., 2004] in which affected children had microcephaly, severe cognitive delay, and early-onset seizures. The ARFGEF2 gene encodes a protein designated BIG2 that converts guanine diphosphate to guanine triphosphate. This activates ADP-ribosylation factors, which regulate vesicle trafficking and transport of molecules from the cell interior to the cell surface, where they can bind to other molecules or be secreted by the cell [Sheen et al., 2004]. Thus, BIG2 may assist in the transport of FLNA to the cell surface.

Other genetic forms of PNH have been associated with chromosomal rearrangements, in particular 6q27 deletion (fig. 1I, J) including C6orf70 (also known as ERMARD) mapping in 6q27, and a few additional putative causal genes (EML1, FAT4 and DCHS1; table 1) [Cappello et al., 2013; Conti et al., 2013; Kielar et al., 2014].

Laboratory Investigations and Genetic Counseling

FLNA mutation analysis should be performed in patients with ‘classical' bilateral PNH. When PNH is associated with microcephaly, the autosomal recessive form caused by ARFGEF2 mutations should be ruled out. In patients with PNH associated with other brain malformations or extraneurological defects, array CGH should be considered.

Classical PNH is much more frequent in women and likely to be caused by FLNA mutations. Among carrier women, about half have de novo FLNA mutations, whereas the remaining have inherited mutations. Although maternal transmission is much more likely, father-to-daughter transmission is possible. Given that germline mosaicism of FLNA has never been reported, the recurrence risk (for other children) seems to be very low when a mutation is found in the proband but neither parent is a carrier. Counseling is very difficult when PNH is not related to either FLNA or ARFGEF2, and array-CGH study is advised. Familial PNH unrelated to these genes is exceptionally low.

The term polymicrogyria defines an excessive number of abnormally small gyri that produce an irregular cortical surface with a lumpy aspect [Bielschowsky, 1916]. Polymicrogyria can be limited to a single gyrus, involving a portion of one hemisphere, be bilateral and asymmetrical, bilateral and symmetrical, or diffuse. Sometimes, it is associated with deep clefts that may extend through the entire cerebral mantle to communicate with the lateral ventricle (schizencephaly) [Barkovich and Kjos, 1992]. Macroscopically, polymicrogyria appears as an irregular or pebbled cortical surface. The perisylvian cortex is the most frequently affected. The cortex often appears thickened to 8-12 mm, but when viewed microscopically, it is overfolded and not necessarily thick. While polymicrogyria most often occurs as an isolated malformation, it can co-occur with several other brain malformations, including microcephaly, megalencephaly, grey matter heterotopia, ventriculomegaly as well as abnormalities of the septum pellucidum, corpus callosum, brainstem and cerebellum.

When schizencephaly is present, the cortex edges can seem to fuse (closed lips) or stay at distance (open lips). The clefts of schizencephaly can be unilateral or bilateral. An area of polymicrogyria can occur in the cortex contralateral to a unilateral cleft [Yakovlev and Wadsworth, 1946].

Using CT and low-field strength MRI, polymicrogyria is difficult to discern and may only appear as a mildly thickened, irregular cortex. For this reason, polymicrogyria is frequently misdiagnosed as pachygyria or LIS. With use of high-field or ultra-high-field strength MRI with appropriate age-specific protocols, polymicrogyria can be reliably differentiated - especially the classic form of polymicrogyria with perisylvian location, schizencephaly, and several other forms - from other cortical malformations and its extent fully recognized [De Ciantis et al., 2015]. The imaging appearance of polymicrogyria varies with the patient's age. In newborns and young infants, the malformed cortex is very thin with multiple, very small undulations. After myelination, polymicrogyria appears as thickened cortex (usually 6-10 mm) with irregular cortex-white matter junction. In schizencephaly, the grey matter lining the cleft has the imaging appearance of polymicrogyria with an irregular surface, deep infolding (the cleft), mildly thick cortex, and stippling of the interface between grey and white matter. Schizencephaly is often bilateral but frequently asymmetrical; the contralateral hemisphere should be closely assessed for milder clefts or polymicrogyria without cleft [Barkovich and Kjos, 1992].

Polymicrogyria has been described in a number of topographic patterns. Most of these are bilateral and symmetrical, the most common of which is bilateral perisylvian polymicrogyria, although the perisylvian form may be asymmetric or unilateral. Other bilateral symmetric forms are generalized, bilateral frontal, and parasagittal parieto-occipital polymicrogyria [Guerrini et al., 1997, 2000; Barkovich et al., 1999; Leventer et al., 2010], although little or no neuropathologic data is available to support the classification.

The clinical manifestations of polymicrogyria vary widely and depend on several factors. The most severe outcomes occur in children with severe microcephaly (-3 SD or smaller), abnormal neurological examination, widespread anatomic abnormality, and additional brain malformations such as heterotopia or cerebellar hypoplasia. The best outcomes are in individuals who have localized unilateral polymicrogyria without other malformations. Polymicrogyria can affect eloquent cortical areas representing language or primary motor functions, yet these functions can be retained with little or no disability [Guerrini and Barba, 2010].

Bilateral perisylvian polymicrogyria is associated with mild to moderate intellectual disability, epilepsy, and impaired oromotor skills. More severely affected patients have minimal or no expressive speech. The frequency of epilepsy in these patients is 60-85%, although seizure onset may not occur until the second decade, usually between 4 and 12 years of age [Kuzniecky et al., 1993; Barkovich et al., 1999]. Seizure types include atypical absences (62%), atonic and tonic drop attacks (73%), generalized tonic-clonic (35%), and focal seizures (26%).

Patients with closed-lip schizencephaly typically present with hemiparesis or motor delay, whereas patients with open-lip schizencephaly typically present with hydrocephalus, seizures and intellectual disability, which can be severe [Packard et al., 1997]. Seizure types include infantile spasms, complex partial seizures as well as tonic, atonic, and tonic-clonic seizures, although these are less common.

Numerous causes, both genetic and nongenetic, have been associated with polymicrogyria. Nongenetic causes other than hypoxia or hypoperfusion relate mainly to congenital infections, primarily cytomegalovirus [Evrard et al., 1989; Barkovich and Lindan, 1994]. Although the number of cases of polymicrogyria secondary to cytomegalovirus is not known, the real magnitude of the problem is probably underestimated, and an analysis for cytomegalovirus infections of dry blood spot should be performed in newborns with this malformation.

Polymicrogyria is associated with a wide number of patterns and syndromes as well as mutations in several genes (table 1). Various polymicrogyria syndromes have been described, which have been designated according to their lobar topography [Barkovich et al., 2012].

Various types of single-gene inheritance have been hypothesized for polymicrogyria, based on observations of families with X-linked [Guerreiro et al., 2000; Villard et al., 2002; Santos et al., 2008], autosomal dominant [Guerreiro et al., 2000; Chang et al., 2006], and autosomal recessive forms [Hilburger et al., 1993; Guerreiro et al., 2000]. Bilateral frontoparietal polymicrogyria, (fig. 1K) has been associated with mutations in the GPR56 gene in families with recessive pedigrees [Piao et al., 2004]. However, the imaging characteristics of bilateral frontoparietal polymicrogyria resemble those of the cobblestone malformative spectrum (muscle-eye-brain disease and Fukuyama congenital muscular dystrophy) [Guerrini and Dobyns, 2014].

Polymicrogyria occurs with a few types of severe congenital microcephaly, such as autosomal recessive syndromes associated with mutations in the WDR62 [Bilgüvar et al., 2010], NDE1 [Alkuraya et al., 2011], or KATNB1 [Mishra-Gorur et al., 2014] genes. Polymicrogyria with microcephaly or normal head size has been reported with several tubulin and tubulin motor genes, especially TUBB2B (fig. 1G) [Jaglin et al., 2009; Guerrini et al., 2012] and DYNC1H1 (fig. 1L, M) [Poirier et al., 2013]. Autosomal recessive forms of polymicrogyria have been linked to mutations of RTTN [Kheradmand Kia et al., 2012]. Also, polymicrogyria has now been reported in several megalencephaly syndromes. Recently, it has been demonstrated that mutations in PIK3R2, a pivotal gene of the PI3K-AKT-mTOR pathway, may account for up to 15% of all patients with bilateral perisylvian polymicrogyria with or without megalencephaly (fig. 1N) [Mirzaa et al., 2015].

Some copy-number variants have been associated with polymicrogyria (table 1), but only deletions in 1p36.3 and 22q11.2 are common [Robin et al., 2006; Dobyns et al., 2008]. Indeed, when these 2 loci are excluded, copy number variants seem to be rare. A causal gene has not been identified for any of these loci.

The term megalencephaly refers to an abnormally large brain that exceeds the mean for age and gender by 2 SD [DeMyer, 1986]. Megalencephaly has most often been classified simply as a disorder of brain size, but recent studies have shown that megalencephaly with normal cortex by imaging, megalencephaly with polymicrogyria, and dysplastic megalencephaly (including classic hemimegalencephaly) as well as FCD can all result from mutations in genes in the PI3K-AKT-mTOR pathway [Lee et al., 2012; Poduri et al., 2012; Rivière et al., 2012b].

Hemimegalencephaly and FCD constitute a spectrum of malformations of cortical development with shared neuropathological features. The former is primarily defined by macroscopic enlargement of (more or less) one hemisphere, while FCD is primarily defined by histopathology. As currently classified [Blümcke et al., 2011], FCD encompasses a wide spectrum of cortical malformations with variable features, including microscopic neuronal heterotopia, dyslamination, and abnormal cell types. FCD has been divided into 3 major types and 9 subtypes based on histopathological features [Blümcke and Spreafico, 2011; Blümcke et al., 2011]. Type 1 FCD is characterized by abnormal cortical lamination, type 2 FCD includes cortical dyslamination with dysmorphic neurons (2a) and balloon cells (2b), type 3 FCD occurs in combination with other brain lesions (e.g., tumors) [Blümcke et al., 2011].

The histological changes in hemimegalencephaly, which can be considered as an extreme hemispheric form of FCD, are similar, if not identical, to FCD type 2, with cortical dyslamination and dysmorphic neurons, without (type 2a) or with (type 2b) balloon cells, blurred junctions between grey and white matter, and increased heterotopic neurons in white matter [De Rosa et al., 1992; Blümcke et al., 2011; Guerrini et al., 2015].

The most common cortical malformation in megalencephaly is perisylvian polymicrogyria that looks very similar to perisylvian polymicrogyria in patients with normal or small head size. The cortical changes in hemimegalencephaly are severe and consist of an enlargement of part or a complete hemisphere (or less often bilateral asymmetrical involvement) with no consistent preference for which lobes of the brain are enlarged [Salamon et al., 2006]. Patients with FCD usually have enlarged gyri with either smooth or irregular cortical surface and increased subcortical signal intensity. Some individuals with FCD type 2b have migration tracts underlined by hyperintense radially oriented white matter bands that extend from the dysplastic cortex to the periventricular region, an imaging feature that has been termed transmantle dysplasia [Barkovich et al., 1997].

The developmental and health complications of megalencephaly differ widely. The most common problems include developmental delay, intellectual disability, and seizures that can start early in life and be intractable. Children with diffuse symmetrical or mildly asymmetrical megalencephaly, with or without associated polymicrogyria, have a large head size at birth that soon exceeds +3 SD [Mirzaa et al., 2012a]. Their early development is delayed, and later cognitive development varies from normal to severe intellectual disability. Seizures can begin at any time in childhood.

Individuals with hemimegalencephaly have early developmental delay and more severe intellectual disability than those with diffuse symmetrical or mildly asymmetrical megalencephaly. Epilepsy usually begins in the first weeks or months of life and can include infantile spasms and other epileptic encephalopathies.

The most common clinical sequelae of FCD are seizures. Developmental delay, cognitive disability, and focal neurological deficits are only observed with extensive dysplasias. Early seizure onset has been associated with infantile spasms with asymmetrical or focal features [Guerrini and Filippi, 2005]. FCD type 2 is a frequent cause of focal status epilepticus and, together with FCD type 1, is the most common pathological substrate in surgical series of epilepsy.

Megalencephaly without cortical malformations occurs in benign autosomal dominant macrocephaly, a poorly defined disorder. Several syndromes have been associated with megalencephaly and include neurofibromatosis type 1 due to NF1 microdeletions that also involve the RNF135 gene and Sotos syndrome (with NSD1 mutations), Weaver syndrome (with EZH2 mutations), in addition to Bannayan-Riley-Ruvalcaba syndrome, Cowden syndrome, and severe megalencephaly with autism (all 3 with PTEN mutations) [Mirzaa et al., 2012b]. Megalencephaly with polymicrogyria occurs in megalencephaly-capillary malformation syndrome (with mutations of PIK3CA) and megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome (with mutations of PIK3R2 or AKT3) [Mirzaa et al., 2012b, 2015]. Hemimegalencephaly most often occurs without syndromic features, and has recently been associated with mosaic mutations of PIK3CA, AKT3, and MTOR (fig. 1O, P) [Lee et al., 2012]. Hemimegalencephaly has also been associated with the linear nevus sebaceous syndrome (also known as Schimmelpenning syndrome) and, rarely, with CLOVES syndrome (congenital lipomatous overgrowth with vascular, epidermal, and skeletal anomalies), tuberous sclerosis, hemihypertrophy, and hypomelanosis of Ito [D'Agostino et al., 2004; Tinkle et al., 2005].

The highly focal and variable nature of FCD type 2b, and the pathological resemblance to tubers in tuberous sclerosis, led to the hypothesis that somatic mosaic mutations of genes that encode proteins in the PI3K-AKT-mTOR pathway, which includes the tuberous sclerosis associated genes TSC1 and TSC2, were implicated in FCD [Crino, 2009]. This hypothesis has been in part confirmed by studies documenting pathogenic germline and mosaic mutations in the mTOR gene or in other genes belonging to the PI3K-AKT-mTOR pathway (i.e., AKT3, DEPDC5, and NPRL3) in the dysplasic tissue of FCD type 2a and 2b [D'Gama et al., 2015; Lim et al., 2015; Sim et al., 2015]. In addition, somatic duplications in the 1q chromosomal region encompassing the AKT3 gene have been associated with megalencephaly, hemimegalencephaly, and FCD type 1b [Poduri et al., 2012; Wang et al., 2013; Conti et al., 2015].

When performing genetic testing for disorders possibly caused by mosaic mutations, the screening of multiple tissues (e.g., blood, saliva, and skin fibroblasts) is advisable. Indeed, although testing DNA extracted from blood is the gold standard for identifying de novo constitutive mutations, the analysis of different tissues may help in identifying mutations that are only present in a subset of somatic cells [Nellist et al., 2015].

The clinical spectrum of epilepsy associated with MCD is broad. Understanding the genetic and molecular basis of these malformations is advancing rapidly with new genes being reported frequently, providing evidence that different cortical malformations are probably secondary to abnormalities disrupting different developmental stages. The recent advances in next-generation sequencing have opened new perspectives in the study of the genetic causes of MCD. In particular, massive parallel sequencing of panels targeting key genes involved in cortical development represents a fast and cost-effective tool for the genetic diagnosis of these malformations. The screening of multiple genes in the same experiment also has the invaluable advantage of boosting new genotype-phenotype correlations and possibly identifying genetic causes of MCD for which molecular diagnosis is still missing. However, depending on the number of genes screened in each experiment, next-generation sequencing analysis may provide large numbers of variants whose interpretation is at times difficult. To overcome misleading interpretations of variants, it is always necessary to evaluate their potential pathogenic effect in the context of each patient's phenotype. As a consequence, a tight interaction between referring physicians and molecular biologists is mandatory.

This study was funded by EU Seventh Framework Programme (FP7) under the project DESIRE grant agreement N602531 (to R.G.).

The authors declare to have no conflict of interest related to the contents of this article.