This is the largest study of patients with polymicrogyria performed to date. The previous largest study was a retrospective MRI analysis of 71 patients, which confirmed a number of bilateral polymicrogyria patterns and showed their relative frequencies (Hayashi et al.
). A potential criticism of imaging-based studies such as ours is the detection of a malformation defined historically by pathological features through the use of imaging criteria. This is a valid criticism, but there is no other way to study a large number of patients with polymicrogyria as it is rarely life-threatening and is seldom resected during epilepsy surgery, primarily due to the frequent involvement of eloquent cortex. Therefore, a deficiency exists in the literature correlating pathological and imaging findings. Where data do exist, they confirm that the criteria used in this study to identify polymicrogyria by MRI correlate with the pathological finding of polymicrogyria (Thompson et al.
This study confirms previous findings by identifying certain common topographical patterns of bilateral and unilateral polymicrogyria, a predilection for polymicrogyria to involve the perisylvian cortex and the high frequency of additional non-cortical abnormalities (Hayashi et al.
). We confirmed previous reports of non-perisylvian phenotypes such as bilateral frontal (Guerrini et al.
, bilateral frontoparietal (Chang et al.
), bilateral generalized (Chang et al.
) and bilateral mesial occipital (Guerrini et al.
) forms of polymicrogyria. In addition, we identified nine rare and mostly novel patterns of polymicrogyria including multifocal polymicrogyria, polymicrogyria associated with Sturge–Weber syndrome and polymicrogyria in association with deep transmantle clefts not fulfilling criteria for schizencephaly.
Polymicrogyria has a predilection for the perisylvian cortex, with the perisylvian region being the region of maximal severity in 214 patients (65%), including fourteen patients with PNH/perisylvian polymicrogyria. Whilst the other 35% of patients had polymicrogyria that may have involved the perisylvian cortex, they either showed no region of maximal severity (generalized polymicrogyria) or another region of maximal severity. This may explain the difference between this study and the previously largest imaging study which showed perisylvian involvement in 80% of their patients, including those with maximal involvement in other cortical regions (Hayashi et al.
). We defined perisylvian polymicrogyria as showing a perisylvian gradient, i.e. maximal severity in the perisylvian cortex, either limited to this region, or extending beyond it in one or more directions. We found that the spectrum of perisylvian polymicrogyria is greater than reported in the existing literature, from mild partial perisylvian forms to forms extending well beyond the immediate perisylvian region. In fact, the typical patient with perisylvian polymicrogyria in our study had extension of the malformation well beyond the immediate perisylvian region.
The second most common pattern of polymicrogyria was generalized. Two forms of generalized polymicrogyria were identified; one with normal white matter and the other with diffuse high T2 signal and thinning of the white matter. This latter form has not been described as a distinct entity previously and may reflect widespread dysmyelination and abnormal development of both grey and white matter. It is likely that a number of these patients had congenital cytomegalovirus infection or peroxisomal disorders, especially those with microcephaly.
The third most common pattern of polymicrogyria was that in association with periventricular grey matter heterotopia, divided into PNH associated with perisylvian polymicrogyria and PNH associated with posterior polymicrogyria. PNH/polymicrogyria was classified separately from other patterns based on an assumption of the timing and potential aetiology of the aberrant cortical development leading to the malformation. PNH is thought to arise as an early defect of neuronal migration, at the stage of initiation of migration from the periventricular zone (Fox et al.
). Polymicrogyria on the other hand, is generally considered to be a defect of later neuronal migration or early cortical organization (Barkovich et al.
). Therefore, it was assumed that in cases of PNH/polymicrogyria the first abnormal step in cortical development leads to the PNH, with the polymicrogyria occurring subsequently as a consequence. This is supported by the finding that in most cases, the polymicrogyria appeared in the cortical region overlying the PNH. PNH/polymicrogyria has been described in detail in a related paper that included some of the patients in this study (Wieck et al.
), as well as in a subsequent paper by Parrini et al.
). The posterior form of PNH/polymicrogyria has additional frequent abnormalities of the hippocampi, cerebellum or corpus callosum, which is atypical for most other types of polymicrogyria.
The fourth most common pattern of polymicrogyria was frontal polymicrogyria, which was subdivided into the bilateral, exclusively frontal form (‘frontal only polymicrogyria’) and the frontoparietal form which extends posteriorly beyond the Rolandic fissures. Other than this difference, the patterns and types of polymicrogyria appear similar, with no other imaging features to reliably distinguish them. Chang and colleagues found frequent abnormalities in the white matter, brainstem and cerebellum in their series of 19 patients with frontoparietal polymicrogyria (Chang et al.
). We did not identify such changes to help distinguish between the two frontal polymicrogyria phenotypes, yet our series only had three patients with frontoparietal polymicrogyria. It is yet to be seen whether the rare frontoparietal polymicrogyria pattern represents a more severe form of frontal only polymicrogyria, or whether the two are separate malformations of cortical development. This will require further genotype–phenotype correlation of patients with frontal polymicrogyria and GPR56
mutations, or the finding of new genes for frontal polymicrogyria. Imaging data from humans (Dobyns et al.
) and pathological data from the mouse with loss of the Gpr56
gene (Li et al.
) suggest that the brain malformations seen in patients with GPR56
mutations may have more in common with those seen in patients with congenital muscular dystrophies and cobblestone lissencephaly, than with those seen in patients with other forms of polymicrogyria.
A number of rare polymicrogyria patterns were identified. Parasagittal parieto-occipital polymicrogyria has been described previously (Guerrini et al.
). Other patterns deserve specific mention as they may shed light on the aetiology of polymicrogyria. There were two patients with Sturge–Weber syndrome and polymicrogyria in the region underlying the pial angiomatosis. Polymicrogyria associated with Sturge–Weber syndrome may be under-represented in this series as it is often not diagnosed in the absence of pathological data (Simonati et al.
; Maton et al.
) as the cortical calcifications of Sturge–Weber syndrome can appear similar to polymicrogyria on MRI. The association of polymicrogyria and Sturge–Weber syndrome may suggest that some forms of polymicrogyria are indeed related to hypoperfusion or microvascular malformations. Multifocal forms of polymicrogyria were seen in two patients with Aicardi syndrome. Aicardi syndrome is a disorder occurring almost exclusively in females, manifest by multiple congenital anomalies including complete agenesis of the corpus callosum and often polymicrogyria. It is presumed to be due to a mutation of a gene on the X-chromosome, although to date no causative gene has been identified (Aicardi, 2005
Three patients in this series were included that may shed light on the association between schizencephaly and other forms of polymicrogyria. One of these had a cleft directed towards the lateral ventricle, but not communicating with it, which was lined by polymicrogyria. It is reasonable to suggest that such clefts may be incomplete forms of schizencephaly. Two patients had typical perisylvian polymicrogyria in one hemisphere and schizencephaly in the other. These patients suggest there is a severity spectrum of cortical clefting that spans from bilateral schizencephaly, to patterns of deep clefts lined by polymicrogyria but not communicating with the lateral ventricles, to perisylvian polymicrogyria. The features in common in all these disorders are deep abnormal fissures lined by polymicrogyria and it is likely that these three entities have a shared pathogenesis. Current classification systems now include schizencephaly as a form of polymicrogyria (Barkovich et al.
The imaging findings of polymicrogyria suggest that it is a disorder of fissures and sulcation. Perisylvian polymicrogyria affects the region around the Sylvian fissures; frontal polymicrogyria is limited posteriorly by the Rolandic fissure, parasagittal parieto-occiptal polymicrogyria is centred around the parieto-occipital and calcarine sulci. In schizencephaly, polymicrogyria is centred around a deep cleft which is essentially an abnormally oriented and deep sulcus. In many cases of polymicrogyria the fissures are malformed, being deep or abnormally orientated. No other malformation of cortical development affects the fissuring and sulcation of the cortex in such a pattern, which is the opposite of lissencephaly where there is either an absence or simplification of sulcation. Thus, elucidating the molecular and developmental basis of polymicrogyria may provide insight into the processes of gyrification and sulcation in addition to microscopic cortical development, especially the development of the Sylvian fissures and perisylvian cortex.
A strength of our study is its wide ascertainment through sources including general paediatric neurologists, developmental paediatricians and clinical geneticists ensuring a more accurate representation of the spectrum and types of polymicrogyria than studies with ascertainment through specialist epilepsy centres. Certain important clinical patterns emerged to aid in identifying clinical/imaging correlations, despite the incomplete clinical data set. Some of these findings confirmed what could be predicted intuitively. For example, generalized polymicrogyria was more likely to present with global developmental delay and at an earlier age than other polymicrogyria patterns. Other findings confirmed those previously been reported in smaller studies (Guerrini et al.
, 1992a, b
; Kuzniecky et al.
). For example, patients with bilateral perisylvian polymicrogyria were likely to have pseudobulbar palsy and isolated language delay as prominent clinical sequelae. These and other findings confirm that the clinical data are reliable, and have provided meaningful and statistically-significant information.
Even though epilepsy was the most common clinical problem, a significant number of patients presented with hemiplegia, microcephaly, global developmental delay, an abnormal antenatal ultrasound or with multiple congenital anomalies well before the onset of seizures. In addition, the age at presentation is considerably younger than that reported in previous studies, with over 50% patients presenting within the first year. The differences between this and some previous studies is likely to reflect both the large numbers of patients and our wider ascertainment base.
Polymicrogyria is a highly epileptogenic lesion with approximately 80% of patients eventually developing seizures, the majority within the first five years. The frequency of epilepsy did not differ significantly between any of the major patterns of polymicrogyria, or between subtypes of polymicrogyria within the same main pattern. This may suggest that the epileptogenicity of polymicrogyric cortex is relatively consistent regardless of the topography, extent or laterality, although generalized polymicrogyria had a significantly lower age at seizure onset than other patterns, and bilateral perisylvian polymicrogyria had a significantly lower age at seizure onset than unilateral perisylvian polymicrogyria. In 23 patients (7%), the onset of seizures did not occur until after the first decade and in one patient did not occur until after 30 years. This does not appear clearly due to differences in the pattern of polymicrogyria and further studies focussing on patients without seizures or with a delayed onset of seizures may provide insight into mechanisms protective of seizure generation in individuals otherwise predisposed to epilepsy by the presence of polymicrogyria.
The data regarding sex prevalence with significant skewing towards males were highly suggestive of X-linked inheritance in patients with these polymicrogyria patterns. The identification of skewing towards males in this cohort previously led to linkage studies of five multiplex families with perisylvian polymicrogyria confirming a locus at Xq28 (Villard et al.
), although thus far no causative genes at this locus have been identified. Two additional loci for perisylvian polymicrogyria have subsequently been identified on the X-chromosome (Santos et al.
; Roll et al.
), and it is likely that genes for polymicrogyria will be identified from the X-chromosome in the future.
Most malformations of cortical development do not involve the entire cortex equally, but show regions of maximal severity. For example, lissencephaly shows two main gradient patterns, one with an anterior > posterior severity gradient (with maximal severity in the frontal lobes) and the other with a posterior > anterior severity gradient (with maximal severity in the occipital lobes). Whilst these patterns had been noted for some time, their significance was not appreciated until the genetic basis of lissencephaly was elucidated; the anterior > posterior pattern being associated with mutations of the DCX
gene and the posterior > anterior pattern being associated with mutations of the LIS1
gene (Pilz et al.
; Dobyns et al.
). Therefore, the decision to divide polymicrogyria subtypes according to severity gradients was deliberate in the hope that such a division of imaging phenotypes may correlate with the underlying molecular basis. A major aim of our study was to advance the understanding of polymicrogyria from imaging phenotypes to polymicrogyria syndromes.
This will require the incorporation of multiple components, including imaging features, clinical features, patterns of inheritance and eventually aetiology including gene identification. A proposal outlining the common polymicrogyria syndromes is shown in Supplementary Table 1
. Polymicrogyria is a heterogeneous malformation of cortical development and it is likely to represent the common endpoint of multiple different aberrations occurring during cortical development. Delineation of the different polymicrogyria syndromes and their aetiologies will ultimately provide better diagnostic, prognostic and genetic counselling, improved prenatal and carrier testing, and will progress our understanding of normal human cortical developmental pathways.