Clinical, functional, and neurophysiologic assessment of dysplastic cortical networks: Implications for cortical functioning and surgical management

Authors

  • Michael Duchowny

    1. Comprehensive Epilepsy Program and Brain Institute, Miami Children’s Hospital and the Department of Neurology, University of Miami, Miller School of Medicine, Miami, Florida, U.S.A.
    Search for more papers by this author

Address correspondence to Michael Duchowny, MD, Department of Neurology, Miami Children’s Hospital, 3200 S.W. 60th Court, Suite 302, Miami, FL 33155, U.S.A. E-mail: michael.duchowny@mch.com

Summary

Cortical malformations are highly epileptogenic lesions associated with complex, unanticipated, and often aberrant electrophysiologic and functional relationships. These relationships are inextricably linked to widespread cortical networks subserving eloquent functions, particularly language and motor ability. Cytomegalic neurons but not balloon cells in Palmini type 2 dysplastic cortex are intrinsically hyperexcitable and contribute to local epileptogenesis and functional responsiveness. However, there is much evidence that focal cortical dysplasia is rarely a localized or even regional process, and is a functionally, electrophysiologically, and ultimately clinically integrated neural network disorder. Not surprisingly, malformed cortex is implicated in cognitive dysfunction, particularly disturbances of linguistic processing. An understanding of these relationships is critical for successful epilepsy surgery. Gains in surgical prognosis rely on multiple diagnostic modalities to delineate complex anatomic, electrophysiologic, and functional relationships in magnetic resonance imaging (MRI)–negative patients with rates of seizure-freedom roughly comparable to lesional patients

The embryologic development of the cerebral cortex in higher primates is a stunningly complex process that unfolds in a highly organized and regulated sequence. Neuroblasts differentiate and outwardly migrate from the subventricular zone to their final destination in the cortical mantle. This intricate process is responsible for both the elaborate biologic complexity of the neocortex and the framework for higher cortical functions that define our species.

With an appreciation of the complexity of neuroembryologic development of the cerebral cortex comes the recognition that this process is far from infallible. Although previously underappreciated in clinical practice, cortical malformations are now detected with increasing frequency in the modern magnetic resonance imaging (MRI) era. The protean consequences of disordered neuronal differentiation and migration cause severe neurologic morbidity and mortality. In one survey of the clinical and imaging features of cortical malformations in a pediatric cohort, three-fourths had seizures, more than two-thirds had developmental delay or intellectual disability, and one-half had an abnormal neurologic examination (Leventer et al., 1999). Eighteen percent had associated congenital abnormalities, attesting to the link between cortical and somatic pathoembryologic processes. The presentation of cortical malformation is more severe in childhood, but adults may also exhibit significant neurologic disability (Bast et al., 2006).

Most neurologic deficits in patients with cortical malformations are fixed, with the notable exception of epileptic seizures. Seizures are a frequent and dynamic problem and typically begin in early postnatal life (Fauser et al., 2006). Moreover, a high proportion of seizure disorders are pharmacoresistant and ultimately intractable. Not surprisingly, these high-risk patients are overrepresented at specialized epilepsy centers. Given the alarmingly high rate of medical unresponsiveness, surgical therapy is a high priority for arresting seizures and preventing neurodevelopmental deterioration. Surgical therapy is a particularly important option at centers caring for extremely young patients.

The outcome of early surgery for focal cortical dysplasia (FCD) was disappointing. The first series to describe the histopathologic features of FCD (Taylor et al., 1971) reported seizure-freedom in only 20% of cases. Subsequent results were similar (Palmini et al., 1991; Hirabayashi et al., 1993), and a 43% rate of seizure-freedom reported in another study (Raymond et al., 1995) included a high proportion of dysembryoplastic neuroepithelial cell tumors. Apart from this subgroup, the success rate for remaining patients, as evidenced by complete abolition of seizures, was considerably diminished. These low rates of seizure-freedom stood in stark contrast to surgical outcome for seizures caused by other pathologic substrates including hippocampal sclerosis, porencephaly, vascular malformations, and tumors (Zentner et al., 1997; Jeong et al., 1999; Moritake et al., 2008).

Initial surgeries for FCD-induced seizures were performed as one-stage excisional procedures guided by electrocorticography (ECoG). Surgical targets were constructed from noninvasive data that were fine-tuned by intraoperative interictal recordings and functional cortical mapping. When necessary, language cortex could be delineated in the operating room.

Whether the early disappointing surgical experience reflected inexact electrophysiologic or functional localization, or both, is difficult to determine. In this review, evidence is presented showing that both information sets are critical to surgical success. It is argued further that cortical malformations including FCD are highly epileptogenic and associated with complex, unanticipated, and often aberrant electrophysiologic and functional relationships, and that these relationships are inextricably linked to widespread cortical networks subserving eloquent functions, particularly language, motor ability, and vision. Furthermore, characterizing FCD as a network dysfunction disorder with multifocal features rather than a localized or even regional process is required for surgical success. Evidence to support this hypothesis comes from many different sources with studies in surgical patients having an important role.

Dysplastic Cortex Is Intrinsically Hyperexcitable

The heightened electrical excitability and epileptogenic potential of dysplastic cortex were early recognizable features of surgical patients. Palmini et al. (1995) and Gambardella et al. (1996) first recorded continuous epileptiform discharges (CEDs) in patients with FCD and showed that resection was associated with greater surgical success. Notably, CEDs occurred in normal-appearing as well as visibly dysplastic cortex, thus serving as important physiologic markers of the epileptogenic region.

Intrinsic hyperexcitability of dysplastic cortex has been confirmed repeatedly. Application of the convulsant drug 4-aminopyridine induces spontaneous seizure-like discharges in neocortical slices from patients with malformed cortex (Mattia et al., 1995), reflecting activation of N-methyl-d-aspartate (NMDA) receptors potentiated by γ-aminobutyric acid (GABA)A receptor–mediated conductance (Avoli et al., 1999). Intralesional recordings reveal interictal repetitive and rhythmic spike and polyspike–wave discharges associated with fusiform micro-polyspikes and ictal mid-amplitude 14–18 Hz rhythmic activity followed by low-voltage recruiting fast frequencies (Francione et al., 2003).

Electroencephalography (EEG) and magnetoencephalography (MEG) source analyses also reveal intrinsic epileptogenicity in MRI-positive dysplastic cortex (Morioka et al., 1999; Bast et al., 2004; Otsubo et al., 2005). EEG–functional MRI (fMRI) (blood oxygenation–level dependent) BOLD responses confirm discrete areas of intrinsic hyperexcitability in polymicrogyric cortex (Kobayashi et al., 2005). The spectroscopic features of focally dysplastic cortex suggest developmentally immature characteristics, with overrepresentation of inhibitory and excitatory neurotransmitters (Simister et al., 2007).

Intrinsic hyperexcitability of dysplastic cells contributes to heightened ictogenesis and rapid propagation. Paired-pulse stimulations of the epileptogenic focus in dysplastic cortex reveal increased cortical excitability and decreased cortical inhibition consistent with a net increase in intrinsic excitability during seizure generation (Matsumoto et al., 2005). Regional polyspikes characterize FCD in extratemporal cortex (Noachter et al., 2008), but fast-frequency discharges predominate at seizure onset, and the interval between onset and propagation is rapid (Turkdogan et al., 2005). Propagation time to the temporal neocortex is shorter for dysplastic tissue than in hippocampal sclerosis and is independent of histopathologic features (Fauser et al., 2006; Fauser & Schulze-Bonhage, 2006).

Postnatal maturation influences the expression of dysplastic hyperexcitability, as afterdischarge thresholds show a steep decline with age (Jayakar et al., 1992; Chitoku et al., 2003). Younger patients evidence remote afterdischarges distant to the stimulation site (Chitoku et al., 2003).

Physiologic impairment of intrinsic synaptic transmission characterizes animal models of cortical dysplasia (Zhu & Roper, 2000). Cortical GABAergic neurons show reduced excitatory drive (Xiang et al., 2006), whereas GABA-mediated synaptic inhibition reduces inhibitory postsynaptic current (IPSC) frequency (Calcagnotto et al., 2005). Gap junctions may implement epileptiform activity and synchronize propagating pathways (Gigout et al., 2006).

Epileptogenesis and Functionality in Dysplastic Cortex Are Cell-Specific

The cortical dysplasias are classified clinically according to histopathologic features (Palmini et al., 2004). Less severe forms of dysplasia include ectopic neurons in subcortical white matter or immediately adjacent to layer 1 (mMCD), and architectural disturbances characterized by a microcolumnar arrangement of cortical neurons (type 1). In addition to dyslamination, more severe forms of dysplasia evidence more prominent disturbances, including cytomegalic neurons (type 2A) and balloon cells (type 2B).

Are both cell types in type 2 dysplasia epileptogenic? Initial recordings in Palmini type 2 tissue were either inconclusive (Lurton et al., 2002) or suggested balloon cell hyperexcitability (Rosenow et al., 1998). However, balloon cells are unlikely candidates, as they originate from pluripotential progenitors cells. ECoG recordings of cytomegalic neurons reveal continuous spiking correlating with the duration of epilepsy and number of seizures, whereas balloon cell regions exhibit no ECoG abnormalities. Cytomegalic neurons are linked to continuous spiking, bursting, and recruiting discharges in patients with glioneuronal tumors (Ferrier et al., 2006). Balloon cells do not respond functionally to direct stimulation in the precentral gyrus (Marucic et al., 2002; Boonyapisit et al., 2003). Balloon cell–containing cortical regions are not ictally active, implicating adjacent non–balloon cell regions in epileptogenesis (Rosenow et al., 1998; Chassoux et al., 2000).

In vitro investigations in Palmini type 2 FCD confirm cytomegalic neurons as primary epileptogenic cells and balloon cells being electrophysiologically and functionally inactive (Mathern et al., 2000). Balloon cells lack active voltage-gated or ligand-gated currents and synaptic input. Patch-clamp recordings that reduce K+ conductance in cytomegalic neurons generate large Ca2+ currents (Cepeda et al., 2003). Cytomegalic neurons also display decreased spontaneous glutamatergic synaptic activity in areas of neuroimaging abnormalities (Cepeda et al., 2005). Therefore, in vitro electrophysiologic differences between these two morphologically distinct cellular elements implicate cytomegalic neurons as epileptogenic.

Palmini type 1 FCD lack these cell-specific features. Therefore, although unique cellular properties of Palmini type 2 may influence specific epileptogenic and functional relationships, noncellular factors must play an important role in epileptogenesis.

Focal Cortical Dysplasia Is a Neural Network Disorder

Even MRI-positive dysplastic lesions may be associated with epileptogenesis and frank seizure onset at remote sites. Electrophysiologic recordings reveal distant secondary zones of ictal activation. These “intraictally” activated regions are capable of self-sustaining epileptogenesis and may arise at a considerable distance from the primary focus, and contribute to surgical failure if not fully excised along with the primary focus (Jayakar et al., 1994) Fig. 1.

Figure 1.


Intraictal activation in a 9-year-old right-handed girl with medically resistant partial epilepsy of right frontoparietal lobe origin. Subdural implantation of a 64-contact subdural grid over the convexity and an 8-contact subdural strip covering the interhemispheric surface. (A) Seizure onset: A restricted area of low-voltage fast-frequency discharges is localized to three adjacent electrodes in the top row of the convexity grid. (B) Twenty seconds after seizure onset: A secondary area of independent activation is noted at two electrodes in the interhemispheric strip. (C) Ninety seconds after seizure onset: The region of secondary activation persists as independent intraictal focus that outlasts the primary epileptogenic region on the convexity.

Intraictal activation is characteristic of dysplastic tissue and often but not invariably associated with interictal spikes on scalp EEG (Raymond & Fish, 1996). Intra-ictally activated zones may be contiguous or adjacent to the primary epileptogenic zone, extralobar, or contralateral. Their anatomic location is rarely predictable as, propagation pathways may not coincide with major fascicular pathways (Thiebaut de Schotten et al., 2008). Rather, intraictal activation propagates via corticocortical pathways or aberrant neural networks (Duchowny et al., 2000). Therefore, MRI may identify dysplastic cortex but rarely specifies its full functional extent.

Complex extralesional interactions are characteristic of cortical malformations. A stereo-EEG study of the epileptogenic zone in four patients with polymicrogyria revealed widespread epileptogenic networks linked to high-frequency spiking in polymicrogyric cortex (Chassoux et al., 2008). EEG-fMRI recorded ictal and interictal events in patients with FCD; nodular heterotopia and subcortical band heterotopia reveal activation of the overlying cortex and band heterotopia during ictal and interictal events (Tyvaert et al., 2008). Similarly, EEG-fMRI studies in tuberous sclerosis complex reveal epileptogenesis at distant sites (Jacobs et al., 2008).

Network complexity is illustrated in a stereo-EEG study of bilateral periventricular nodular heterotopias (Valton et al., 2008). Unprovoked seizures were associated with interdependent activation of normal and heterotopic cortex. Stimulation along a depth electrode in heterotopic cortex activated distinct remote cortical structures.

Widespread anatomic abnormalities occur in children with pachygyria (Barkovich, 1994), polymicrogyria (Ferrer et al., 1994), and periventricular and subcortical heterotopia (Dubeau et al., 1995). Diffusion tensor imaging studies reveal abnormalities of white matter in FCD, even regions of normal T2-signal intensity underlying dysplastic cortex (Lee et al., 2004; Widjaja et al., 2007) indicating microstructural degradation of subjacent white matter (Gross et al., 2005).

Dysplastic Neural Networks Are Functionally Integrated

Mapping studies of epilepsy surgical patients reveals a close association between dysplastic tissue, the epileptogenic zone, and eloquent cortical function. Seizure onset commonly occurs within or near cortical areas for language, motor function, or vision (Leblanc et al., 1991, 1995; Duchowny et al., 1996). Eloquent cortical regions are defined in prenatal life and do not relocate postnatally despite repeated epileptic bombardment (Oki et al., 1999).

Structural and functional overlap is frequent in FCD (Sisodiya et al., 1995). Electrical stimulation of the schizencephalic cleft produces functional motor responses (Leblanc et al., 1991), and [18F] deoxyglucose-PET (positron emission tomography) and single photon emission computed tomography (SPECT) studies reveal resting metabolic activity in heterotopic nodules similar to that of overlying cortex (Miura et al., 1993; Iannetti et al., 2001). Activation paradigms induce regional cerebral blood flow (rCBF) changes not only in malformed brain regions, but in cortex overlying band and subependymal heterotopias and normal appearing cortical regions (Richardson et al., 1998). Similar findings were reported for event-related potentials in cortical malformations and glial-neuronal tumors (Kirschstein et al., 2002), and for finger tapping, somatosensory, visual, and language-related fMRI tasks (Vitali et al., 2008).

A particularly close functional relationship exists between subcortical heterotopic nodules and overlying cortex. Finger tapping in a 12-year-old boy with subcortical band heterotopia (SBH) activated a restricted region of frontal cortex in both the SBH and overlying cortex (Pinard et al., 2000). A 19-year-old woman with SBH demonstrated simultaneous activation of the SBH and overlying cortex on fMRI and seizure onset from both outer and heterotopic cortex (Mai et al., 2003). Activation of the ventricular wall has also been observed (Spreer et al., 2001). Discrete nodular heterotopic nodules in three patients demonstrated task-related increases in rCBF on PET (Muller et al., 1998). There were no abnormalities of neuropsychologic function, suggesting that heterotopic neurons may network with normal cortical and subcortical regions. Similarly, abnormal somatosensory evoked potentials have been recorded in dysplastic sensorimotor cortex showing minimal or absent changes in sensorimotor function (Raymond et al., 1997).

Dysplastic Neural Networks Are Atypically Organized

Cortical stimulation mapping in dysplastic motor cortex reveal anomalously organized homunculi in some patients (Duchowny & Jayakar, 1993; Preul et al., 1997). The hand region may lie superior to shoulder region or shoulder region superior and inferior to hand and finger cortex. A 13-year-old girl with cortical malformation and contralateral hemiparesis exhibited paretic hand sensation in the ipsilateral unaffected hemisphere anterior to contralateral cortical representation of sensation (Maegaki et al., 1995). Similarly, Little et al. (2007) reported a 14-year-old boy with frontal lobe heterotopia and motor function that localized to the posterior heterotopic margin well anterior to primary motor cortex. Gondo et al. (2000) observed anomalous sensory representation within an epileptogenic zone adjacent to dysplastic cortex in a 5-year-old boy. Abnormal glial or neuronal proliferation of polymicrogyric cortex does not lead to functional reorganization (Burneo et al., 2004).

The relationship of language competence to cortical malformation is more complex (Duchowny, 2007). A close relationship between FCD, language cortex, and the epileptogenic region has been demonstrated repeatedly (Leblanc et al., 1995Duchowny et al., 1996; Janszky et al., 2003). Language cortex is generally conserved in patients with FCD, suggesting that malformations alter but do not physically destroy eloquent cortex. Dysplastic networks are thus “congenitally inefficient” rather than imperfectly restored.

Language cortex may not reside in classical language areas in patients with FCD (Devinsky et al., 1993), and its distribution is often atypical (Bell et al., 2002; Yuan et al., 2006). Language regions rarely conformed to classical language sites in children and adolescents with focal cortical malformations (Liegeois et al., 2004). In contrast to destructive lesions that promote contralateral relocation of language sites, dominant hemisphere dysplasia may produce intrahemispheric relocation. Utilizing coregistered MRI and extraoperative stimulation mapping in patients with chronic partial epilepsy, Kadis et al. (2007) showed that expressive language sites often reside in cortical regions anterior and superior to classical Broca’s area.

Is anomalous representation of eloquent cortical areas always associated with inefficiencies of motor and cognitive processing? Extensive unilateral cortical malformations produce mild contralateral hemiparesis (Maegaki et al., 1995; Preul et al., 1997). In a study of 54 children with refractory focal epilepsy due to developmental lesions, Klein et al. (2000) noted global reductions in cognition. Dominant hemisphere lesions were linked to greater language deficits, whereas nondominant hemisphere lesions compromised nonverbal function. Dominant hemisphere lesions were associated with greater contralateral deficits, but there was no significant relationship between severity of histopathology and intellectual function, suggesting that network dysfunction rather than cell type was the critical variable.

Utilizing neuroimaging, Gaillard et al. (2007) analyzed clinical and pathologic correlates of atypical language representation in a cohort of 102 patients with left hemisphere epileptogenic zones. Atypical handedness, specific structural lesions, and age at seizure onset were most important. Atypical language dominance and atypical handedness were unexpectedly prevalent in MRI-negative patients, a group consisting almost uniformly of low-grade cortical dysplasia (Jayakar et al., 2008).

These clinical observations collectively confirm that many patients with cortical malformations, even discrete dysplastic lesions, evidence both atypical anatomic representation of functional regions and widespread inefficiencies of cortical networks for cognitive and motor activities. The resulting patterns are outside the limits of normal variation for the classical sensorimotor homunculus (Farrell et al., 2007). With rare exceptions (Lado et al., 2002), the anomalous representation and functional inefficiency results from prenatal malformative errors. Although the resultant networks may share a close relationship with pathways for epileptic spread, it is unclear how well propagation pathways are integrated into the abnormal functional networks or if persistent epileptic discharges contribute to further postnatal reorganization.

Not unexpectedly, cortical malformations are also implicated in developmental disorders of linguistic processing. Careful postmortem dissection reveals low grade dysplastic changes in the dominant hemisphere planum temporale (Galaburda et al., 1985). Developmental language disturbance and posterior parietal polymicrogyria has also been described (Guerreiro et al., 2002). Patients with periventricular nodular heterotopia have normal intelligence but exhibit specific reading impairments mimicking developmental dyslexia (Chang et al., 2005). Diffusion tensor imaging in this population reveals focal disruptions of white matter microstructure and organization in the vicinity of the gray matter nodules (Chang et al., 2007).

Postoperative Seizure Freedom Requires Complete Excision of the Dysplastic Network

The complex functional relationships of cortical dysplasia, its intrinsic epileptogenicity, and participation in widespread, atypical neural networks all constitute significant barriers to surgical success, but several series report significantly improved rates of seizure-freedom (Chassoux et al., 2000; Edwards et al., 2000; Hong et al., 2000; Paolicchi et al., 2000; Kloss et al., 2002; Tassi et al., 2002; Bautista et al., 2003; Kral et al., 2003; Srikijvilaikul et al., 2003; Krsek et al., 2008).

Several modifications of the presurgical evaluation improve seizure prognosis. Advances in neuroimaging including higher field strength magnets, optimized imaging protocols, quantification of volumes, MR spectroscopy, and functional imaging allow detection of subtle dysplastic lesions (Briellmann et al., 2003). Delineating a discrete lesion helps define the epileptogenic region, whereas a negative MRI is associated with greater seizure persistence (Chapman et al., 2005).

Although not discounting the value of anatomically localizing a surgical target (Kral et al., 2003), lesionectomy of dysplastic cortex does not guarantee seizure-freedom (Widdess-Walsh et al., 2007). It is axiomatic that no epilepsy surgery procedure, or for that matter no surgical procedure, is uniformly successful, but FCD is especially problematic. Dysplastic tissue margins are rarely well delimited and thus more difficult to define surgically than other pathologies. Therefore, FCD, whether MRI-positive or MRI-negative, bears some similarity to an infiltrative lesion and is thus more difficult to excise entirely.

Diverse and unpredictable functional relationships of FCD also contribute to surgical failure. Given the close relationship of FCD to eloquent cortical regions, it may not be possible to completely excise epileptogenic dysplastic cortex without neurologic deficit. Aberrant epileptogenic neural substrates within the dysplastic network must also be identified and excised.

The recently improved surgical seizure prognosis following surgery for cortical malformations directly follows the application of multiple modalities to delineate existing anatomic, electrophysiologic, and functional relationships. Even MRI-negative patients achieve rates of seizure-freedom roughly comparable to lesional patients (Widdess-Walsh et al., 2007; Jayakar et al., 2008). Extraoperative recordings utilizing subdural electrode arrays and stereo-EEG are especially useful for achieving complete resections (Paolicchi et al., 2000; Chassoux et al., 2000). These results demonstrate that although the presence of a visible lesion on MRI is a useful marker of the epileptogenic zone, complete excision of the functionally abnormal tissue is a necessary prerequisite for seizure-freedom. In this regard, complete delineation of electrophysiologically abnormal cortex is mandatory.

Incomplete resection and lower grade dysplastic tissue histopathology, bilateral EEG abnormalities, dysplastic cortex in the contralateral hemisphere, and multiple ictal semiologies diminish seizure prognosis (Chassoux et al., 2000; Edwards et al., 2000; Cohen-Gadol et al., 2004; Widdess-Walsh et al., 2007; Krsek et al., 2008; Kim et al., 2009; Krsek et al., 2009). Focal spikes on scalp EEG also positively predict outcome (Jayakar et al., 2008).

The contribution of Palmini subtype to surgical prognosis has been actively debated. Although Fauser et al. (2004) reported better outcome for Palmini class 1 patients, others have found improved prognosis associated with Palmini class 2 features (Hudgins et al., 2005; Widdess-Walsh et al., 2005; Krsek et al., 2008).

Conclusion

All available evidence to date, largely derived from clinical and translational studies of patients with cortical dysplasia, suggest that for epilepsy surgery to be successful, the boundaries between dysplastic and normal cortex must be well characterized. This task is challenging, as dysplastic cortex is specified prenatally and thus integrated into developing cortical networks. In particular, to achieve seizure-freedom, the epileptogenic networks and atypical organization of eloquent cortical regions must be clearly defined. FCD thus differs from pathologic entities acquired in postnatal life such as tumors, gliosis, and vascular malformations, which often destroy existing networks. Invasive studies are often necessary in MRI-negative cases, but may contribute to success in MRI-positive cases with divergent data or unclear seizure patterns. The mechanisms for the establishment of aberrant neural networks and atypical brain organization await elucidation. Data derived from future studies should prove immensely useful in the quest for seizure-freedom in patients with intractable epilepsy due to cortical dysplasia.

Acknowledgment

I confirm that I have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: The author has no conflicts of interest to disclose.

Ancillary