Cortical Brain Malformations: Title and subTitle BreakEffect of Clinical, Neuroradiological, and Modern Genetic Classification FREE

Malformations of cortical development (MCDs) are a major source of mental retardation, motor dysfunction, and epilepsy in children. Up to 10% of children with epilepsy are estimated to have an MCD, and the percentage is greater in those with intractable epilepsy.¹ Not all patients with MCD develop epilepsy, however, and the level of motor and mental handicap varies. In recent years, much progress has been made in understanding the genetic and molecular basis of cortical development. Several essential genes have been identified (Table 1). Also, MCDs have been described in patients with known syndromes such as the Adams-Oliver, Delleman, MICRO, and Kabuki make-up syndromes.^{2 - 6} Illustratively, the London Medical Dysmorphology Database includes 46 syndromes with MCD as a requirement and 232 with MCD as a possible feature.

Accurate classification is important for patient care and for improving insight into different phenotypes and causes. The 2005 classification system for MCD distinguishes disorders of cell proliferation, migration, and cortical organization and is useful for guiding the diagnostic approach in individual patients.⁷

This approach starts with a detailed brain magnetic resonance imaging (MRI) study, combined with medical and family history and a pediatric and neurological examination. Affected relatives, parental consanguinity, dysmorphic features, and typical MCD patterns can suggest genetic causes. Complications during pregnancy or congenital infections can be compatible with a fetal insult. Internal organ dysfunction and a progressive course may suggest a metabolic disease.⁸

Apart from monogenetic mutations, chromosomal anomalies should be looked for. Some microdeletions can cause distinctive MCD, particularly interstitial deletions of chromosome 17p in lissencephaly, duplications of 5p in nodular heterotopia, and microdeletions of 22q11 in polymicrogyria (PMG).^{9 - 11}

The aims of the present study were to evaluate the effect of clinical, radiological, and genetic test results on etiological diagnosis in a series of consecutive cases of MCD in a university children's hospital; to describe the relationship between diagnosis and clinical features; and to evaluate the consequences of an etiological diagnosis for disease management and genetic counseling.

The medical records of children with MCD referred to the Department of Pediatric Neurology or the Department of Clinical Genetics at the Erasmus Medical Center were retrospectively collected from January 1, 1992, through December 31, 2001, and prospectively from January 1, 2002, to January 1, 2006. We stopped inclusion when we started to recruit referrals actively. Information on the retrospective group was collected using the radiology reports as documented in the medical correspondence database of the Erasmus Medical Center–Sophia Children's Hospital using the terms simplified gyral pattern, lissencephaly, pachygyria, polymicrogyria, heterotopia, cortical dysplasia, and migration disorder. All patients with available neuroimaging studies were included. All neuroimaging findings (MRI and computed tomography) were reevaluated by a pediatric neurologist (I.F.M.d.C.), a clinical geneticist (G.M.S.M.), and a pediatric neuroradiologist (M.H.L.). From 1992 to 2005, we included the most recent MRI study if a diagnosis of MCD was made when the patient was younger than 16 years. Consensus was required and patients were classified into 4 groups on the basis of MRI criteria.^{7 ,12} None of the patients had a molecularly confirmed diagnosis.

We excluded patients with neoplasms, tuberous sclerosis, and neurofibromatosis 1. We excluded those with hydranencephaly and holoprosencephaly, unless associated with obvious MCD such as periventricular nodular heterotopia. We excluded patients with infratentorial abnormalities only and patients with a normal cortex and microcephaly or macrocephaly.

If possible, patients were examined at the multidisciplinary outpatient clinic. Epilepsy was diagnosed when seizures were present and electroencephalographic findings were abnormal. The level of mental retardation was defined as mild (IQ, 70-80), moderate (IQ, 50-69), or severe (IQ, < 50), based on age-appropriate neurodevelopmental testing, performed by the school or by our clinic.

Diagnosis was confirmed retrospectively in 48 patients, with 65 patients added prospectively, resulting in a case series of 113 patients (55 boys and 58 girls). They were classified into 4 groups on the basis of MRI findings. Further classification was based on results of the physical examination and genetic and metabolic tests (Table 2 and Figure 1).

In the retrospective search, 76 children were found to have a documented diagnosis of MCD. Thirteen were excluded because of missing brain imaging studies. In 15, the diagnosis was reversed. In the remaining 48 cases, an MCD was confirmed, although the classification was revised in 11 patients. In 6, a diagnosis of pachygyria was changed to PMG. The opposite change was made in 1 case. In 2 cases, a diagnosis of lissencephaly was revised to simplified gyral pattern (SGP). A presumed cortical dysplasia was found to be PMG in 2 cases.

In the 10 years of retrospective analysis, we found 48 cases; in the 4 years of prospective analysis, 65 cases. This change may be owing to increased alertness and/or the quantity and quality of brain imaging. In addition, the number of first-patient visits increased greatly. At the time of the study, patients were not referred to us preferentially. Per year, 1700 to 2000 new patients were referred to the pediatric neurology outpatient clinic, making the incidence of MCD in our population approximately 8 cases for every 1000 new patients.

Before or during follow-up, 68 patients (60%) developed epilepsy. Significantly fewer patients with PMG had epilepsy than in the other groups (Table 3). They may develop epilepsy at an older age. Mental retardation was present in 106 of 113 patients, although the degree was mild in 27. The severity of mental retardation was correlated with the presence of epilepsy and its age at onset (P = .001). If a first seizure happened before 9 months of age (median age at onset), most of these patients (31 of 35 [91%]) showed moderate to severe mental retardation, whereas this was found in 17 (57%) of the 30 patients who developed seizures later in life. This probably reflects the severity of the underlying brain malformation. All 22 children who presented with neonatal seizures or infantile spasms developed severe mental retardation, and 8 died during infancy. Overall, 17 children died during follow-up. Neurological motor problems were common, including a spastic tetraparesis in 26 children, a spastic hemiparesis in 19, and severe hypotonia in 18. A range of congenital major and minor anomalies was seen.

The standard karyotype was normal in all patients who underwent testing unless otherwise specified. Further genetic analysis is described per group (Figure 1 and Table 4).

We found 11 cases of congenital microcephalies with MCD (group 1), including 7 with SGP and 4 with agyria or pachygyria. Two unrelated patients had SGP and diabetes mellitus. In patient 1, the cause was Wolcott-Rallison syndrome; in the other, this cause was excluded. Three of the remaining patients with SGP had consanguineous parents. Two patients with SGP had mild or no mental retardation and normal motor skills (eg, Figure 2A). None had evidence of a gestational insult.

Patients with SGP are candidates for testing of the MCPH loci (Table 1), but tests are currently not available. The patient with Wolcott-Rallison syndrome had a homozygote missense mutation in EIF2AK3 and was included in a previously published report.¹³ In 3 of the 4 patients with agyria or pachygyria, interstitial deletions in 17p13.3 and LIS1 mutations were excluded; for the remaining patient, no DNA was available.

Twenty-four patients were diagnosed as having syndromes or diseases that fell within the spectrum of lissencephaly, pachygyria, and subcortical band heterotopia (group 2A) (Table 4). In 16 of these patients, a genetic syndrome was diagnosed. Five had Miller-Dieker syndrome with typical facial features, severe mental retardation, and epilepsy (Figure 3A). Four patients with pachygyria and moderate or severe mental retardation had a LIS1 mutation (Figure 3B). Two patients with milder retardation and MRI abnormalities had a somatic mosaic LIS1 mutation (Figure 3C).¹⁴ Two girls with subcortical band heterotopia and epilepsy had a DCX mutation (Figure 3D). One girl was diagnosed as having Baraitser-Winter syndrome (patient 25) (Figure 2B).¹⁵ Patient 26 had a muscle-eye-brain–type disease with muscular dystrophy, elevated levels of creatine kinase, and visual impairment. His MRI did not show a cobblestone-type lissencephaly, but rather occipital pachygyria with nonprogressive leukodystrophy (Figure 2C). Arguably, this patient could be classified as belonging to group 2B. One boy was diagnosed as having FG syndrome on the basis of dysmorphic features and urogenital abnormalities.¹⁶ He is undergoing testing for the R916W mutation in the HOPA gene (OMIM 300188).¹⁷

Of the 8 patients with unknown etiology, 1 pedigree shows an affected mother and son, suggesting an autosomal dominant or X-linked inheritance, but the mutation analysis results of LIS1, DCX, and ARX were normal. One patient had a brother with microlissencephaly in group 1. There was no consanguinity in group 2A. Most of the patients had severe psychomotor retardation with or without epilepsy (16 of 24), whereas only 5 patients had normal development or a mild mental deficit.

The 5 patients with Miller-Dieker syndrome had an interstitial deletion of 17p13.3 confirmed by fluorescence in situ hybridization analysis. Sequencing of LIS1 was performed in 12 patients and showed pathological mutations in 4, including 2 nonsense mutations (W261X and G147X), 1 frameshift mutation (162delA), and 1 intron mutation (900 + 1G > C; IVS8 + 1G > C). In 2 patients, a somatic mosaicism was found (mosaic 162delA and R113X). Analysis of DCX was performed in 8 patients. In 2, a nonsense mutation was found (R303X and R272X). No mutations were found in 2 patients in whom ARX and RELN were tested. None of the patients fit the typical phenotype of X-linked lissencephaly with abnormal genitalia for ARX or of pachygyria, cerebellar anomalies, and lymphedema for RELN (Table 1). In patient 25, results of microscopic analysis of a muscle biopsy were normal, and mutations in FCMD (OMIM 607440) and FKRP (OMIM 606596) were excluded.

Two patients with cobblestone-type lissencephalies (group 2B) had Walker-Warburg syndrome (Table 4). Both died in infancy (Figure 2D). No muscle biopsy material was preserved for histochemical analysis of α-dystroglycans.¹⁸ Mutations in POMT1 and POMGnT1 were excluded in both patients. Further genetic analysis will be performed when methods become available.

We found 14 patients with nodular heterotopia without cortical abnormalities (group 2C) (Table 4). One patient with a history of maternal cocaine abuse during pregnancy, mild retardation, epilepsy, and diffuse bilateral periventricular nodular heterotopia (BPNH) had an FLNA mutation (patient 38, Figure 3E).¹⁹ We assumed there was an autosomal recessive syndrome in 3 siblings with microcephaly, BPNH, and cerebellar atrophy without FLNA or ARFGEF2 mutations. One patient had a syndrome with an occipital meningocele and BPNH, a combination reported previously.²⁰ Patient 41 died in infancy of a severe cardiac anomaly and BPNH with enlarged ventricles. In another boy, maternal alcohol abuse during pregnancy was documented, but alcohol has not been clearly linked to BPNH in humans. Overall, symptoms ranged from drug-responsive epilepsy with normal or near-normal mental and motor development to epilepsy with severe handicap.

Testing of FLNA was performed in 10 patients in group 2C.¹⁹ A pathogenic frameshift mutation (7104delG) was found in patient 40. In patient 41, a previously unreported missense change in exon 20 was found, but it was not possible to perform additional tests to prove pathogenicity. For 2 patients, no DNA was available. Testing of ARFGEF2 was performed in 3 patients, with normal results.²¹

In 11 patients with heterotopia combined with overlying PMG (group 2D), 7 presented with severe psychomotor retardation, whereas 4 showed normal or near-normal mental development independent of epilepsy or the extent of MRI abnormalities (Table 4). We diagnosed frontonasal dysplasia in a patient with a frontal encephalocele and facial asymmetry. Considering the overlap with craniofrontonasal dysplasia, mutations in EFNB1 were excluded (OMIM 300035). One girl had a large Xq chromosome deletion (patient 53). In 2 patients, there was parental consanguinity.

A standard karyotype of patient 53 showed a deletion of Xq21qter, including the FLNA and DCX loci. This finding was excluded in her parents. It was not possible to conduct X chromosome inactivation studies to analyze the relevance.

We established a causative diagnosis in 20 of 48 cases with PMG and schizencephaly (group 3A) (Table 4). Four patients belonged to a pedigree with Goldberg-Shprintzen syndrome.²² Ten cases were diagnosed as having a syndrome or a monogenetic disorder associated with PMG. One patient fulfilled criteria for Adams-Oliver syndrome (Figure 2E), another for Sturge-Weber syndrome, another for Aicardi syndrome, and another for Joubert syndrome.^{2 ,7 ,23 - 24} Patient 67 had a muscle-eye-brain–type disease with schizencephaly, congenital cataracts, and muscular dystrophy. Patient 71 was diagnosed as having a merosin-negative muscular dystrophy (Figure 2F). Three patients, 2 of whom are from the same pedigree, were found to have a syndrome consisting of megalencephaly, perisylvian PMG, postaxial polydactyly, and hydrocephalus.²⁵ In an additional 7 patients, there was documented evidence of a gestational insult that could have caused PMG. Congenital cytomegalovirus infection was excluded in 9 patients with unexplained PMG; in 2 additional patients, this infection could not be excluded. We suspected a genetic cause in 6 patients with multiple congenital abnormalities and in 4 with consanguineous parents or multiple affected family members. In 11 of 48 patients, we found no clues to an underlying cause.

In all patients examined after 2001, microdeletions at the 22q11 locus were excluded. Analysis of KIAA1279 in patients with Goldberg-Shprintzen syndrome was performed as described by Brooks et al.²² Testing for mutations in GPR56 was performed in 13 cases of bilateral perisylvian PMG, with negative results. Mutations in AHI1 (OMIM 608894), associated with Joubert syndrome, were excluded in patient 72. In patient 67, analysis of α-dystroglycans in muscle biopsy specimens was ongoing at last follow-up. In patient 71, LAMA2 (OMIM 156225) analysis was not available at the time of diagnosis.

Metabolic causes were found in 2 patients (group 4) diagnosed as having Zellweger syndrome and malonyl-coenzyme A decarboxylase deficiency, an inborn metabolic disease (Table 4). Polymicrogyria is associated with Zellweger syndrome, but diffuse severe PMG with cerebellar abnormalities is rare (Figure 3F).²⁶ To our knowledge, pachygyria and nodular heterotopia have not been described in previous reports of malonyl-coenzyme A decarboxylase deficiency.²⁷

Peroxisomal fatty acid oxidation in fibroblasts and a mutation in PEX10 (OMIM 602859) confirmed the diagnosis in patient 113. Malonyl-coenzyme A decarboxylase in fibroblasts of patient 112 was deficient, and MLYCD (OMIM 606761) analysis showed absence of a transcript.²⁷

This study of a large consecutive group of patients with MCD shows the effect of combining radiological, clinical, and molecular analysis. After radiological classification, analysis for associated congenital anomalies and dysmorphic features, and genetic tests, a definite diagnosis of the underlying cause was found in 38 cases (34%). For 21 patients (19%), this included molecular and/or genetic confirmation, and in 17 (15%), this was a syndrome diagnosis with an unknown genetic defect. In another 7 patients, we found evidence of a gestational insult, for a total of 45 etiological diagnoses (40%). These results improved patient care and gave parents opportunities for reproductive choices and prenatal diagnosis.

Some smaller studies have been published, most of which focus on a specific MCD subgroup.^{12 ,19 ,28 - 30} One cohort study of similar size reviewed radiological and clinical findings, without syndrome diagnoses or molecular diagnostic testing.³¹

Our retrospective revision of all neuroimaging studies showed that 26 of 63 patients (41%) had received previous misdiagnoses. A quality MRI of the brain and a skilled neuroradiologist are essential for a correct classification and the choice of diagnostic tests. The 2005 MCD classification is useful, although some patients remain hard to classify, especially those with combinations of heterotopia and abnormal cortex.^{7 ,12}

For some MCD phenotypes such as Walker-Warburg syndrome, genetic heterogeneity is wide, making molecular confirmation difficult.¹⁸ In others such as microcephaly with SGP, tests for known genes are not routinely available. Telomere multiplex ligation-dependent probe amplification is also becoming available to assist in diagnosis. More extensive testing can further increase the number of molecularly confirmed diagnoses.

As expected from previous observations, genetic causes were most often confirmed in our patients with lissencephaly or pachygyria.^{29 ,32} Mutations in LIS1 or DCX explained the lissencephaly and pachygyria in more than half of our patients. The genetic causes of PMG are still largely elusive. Mutations in GPR56 and KIAA1279 only explain a small percentage of PMG cases.^{22 ,33} We tested GPR56 in 13 patients with bilateral frontoparietal PMG, but found no mutations. None of these patients had the typical pons hypoplasia and white matter abnormalities, further confirming the specificity of the phenotype associated with GPR56 mutations.³⁰

In our cohort, most patients were diagnosed as having MCDs when they developed seizures. Patients with PMG have a lower incidence of epilepsy, and motor dysfunction often prompted medical attention. The overall risk of psychomotor retardation is high, but 34 of our 113 patients had an IQ greater than 70. Epilepsy is a negative prognostic factor, particularly infantile spasms or neonatal seizures. In 1 patient with BPNH, we found a pathogenic FLNA mutation, whereas her developmental delay previously was accredited to prenatal exposure to cocaine. This supports the predictive power of a specific radiological pattern for genetic testing. We did not find mutations in patients whose MRI pattern was not typical for the gene tested. We recommend careful classification before ordering genetic tests.

Nine patients in our cohort fit a known syndrome with an unknown genetic cause. Another 15 patients presented with an apparently unique syndrome, meaning a combination of multiple anomalies likely to result from an underlying genetic cause. Furthermore, a genetic cause is likely in patients with affected family members with or without consanguinity (12 patients) or in patients with consanguineous parents only (7 patients). These 3 groups are candidates for whole-genome analysis by means of new techniques designed to discover other genes involved in brain development. This approach has been successful in cohorts of patients with congenital malformations and mental retardation.³⁴ In addition, consanguineous families with multiple affected members can be explored by improved linkage techniques.

Our cohort represents a heterogeneous group but closely follows clinical practice, where most patients are referred with an unclassified MCD. We show that classification based on radiological, clinical genetic, and neurological examinations combined with genetic testing can yield important information about monogenetic, syndromal, and metabolic causes and can lead to improvement of patient care and genetic counseling. This requires a multidisciplinary team specialized in neuroradiology, pediatric neurology, and genetics. Even then, the underlying cause remains elusive in more than 50% of patients, and the suspicion of an underlying genetic cause remains in many of our unclassified cases. This encourages exploitation of new genome-wide techniques.

Correspondence: Grazia M. S. Mancini, MD, PhD, Department of Clinical Genetics, Erasmus Medical Center, PO Box 1738, 3000 DR Rotterdam, the Netherlands (g.mancini@erasmusmc.nl).

Accepted for Publication: September 20, 2007.

Author Contributions: Drs Lequin and de Coo contributed equally to this study. Study concept and design: de Wit, Lequin, de Coo, Brusse, and Mancini. Acquisition of data: de Wit, Lequin, de Coo, and Mancini. Analysis and interpretation of data: de Wit, Lequin, Halley, van de Graaf, Schot, Verheijen, and Mancini. Drafting of the manuscript: de Wit and Mancini. Critical revision of the manuscript for important intellectual content: de Wit, Lequin, de Coo, Brusse, Halley, van de Graaf, Schot, Verheijen, and Mancini. Statistical analysis: de Wit and de Coo. Obtained funding: Mancini. Administrative, technical, and material support: Halley, van de Graaf, Schot, and Verheijen. Study supervision: Mancini.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Revolving Fund of the Erasmus Medical Center (Dr Mancini).

Additional Contributions: At the Erasmus Medical Center, Cokkie H. Wouters, PhD, performed standard G banding karyotype and fluorescence in situ hybridization analysis; Dr Halley performed the DNA analysis of LIS1, DCX, ARX, FLNA, ARFGEF2, RELN, KIAA1279, and GPR56; and G. Cees Schoonderwoerd, PhD, Jan G. M. Huijmans, PhD, and Dr Verheijen performed the enzyme analysis of malonyl-coenzyme A decarboxylase and the routine urine and plasma metabolic screening. Analysis of EIF2AK3 was performed at the Institut Pasteur, Paris, France, by Cécile Julier, PhD; analysis of POMT1 and POMGnT1 was performed at the Radboud University Nijmegen, Nijmegen, the Netherlands, by Erik A. Sistermans, PhD; analysis of EFNB1 was performed at the University Of Oxford, Oxford, England, by Andrew O. M. Wilkie, PhD; and analysis of AHI1 was performed at the University Medical Center Utrecht, Utrecht, the Netherlands, by J. K. (Hans) Ploos van Amstel, PhD. Peroxisomal fatty acid oxidation was analyzed at the University of Amsterdam, Amsterdam, the Netherlands, by Hans R. Waterham, PhD, and M. (Ries) Duran, PhD. A. J. (Jim) Barkovich, MD, PhD, and William B. Dobyns, MD, PhD, provided revision of the MRI data. Willem F. M. Arts, MD, PhD, Jaap P. Braakhekke, MD, Alice S. Brooks, MD, PhD, Piotr Carbaat, MD, Coriene E. Catsman-Berrevoets, MD, PhD, Jojanneke F. de Rijk-van Andel, MD, PhD, Paul Govaert, MD, PhD, Rudy van Koster, MD, PhD, J. B. C. (Hans) de Klerk, MD, M. J. C. (Aik) Rovers, MD, Liesbeth S. Smit, MD, Frans Visscher, MD, Patrick J. Willems, and Michel Willemsen, MD, PhD, referred patients. Martinus F. Niermeijer, MD, PhD, provided a critical reading of the manuscript.