Frontal Focal Cortical Dysplasias: Too Thin Here, Too Thick There, and the Folding Just Isn't Right!

Commentary

Whole-brain MRI Phenotyping in Dysplasia-Related Frontal Lobe Epilepsy.

Hong SJ, Bernhardt BC, Schrader DS, Bernasconi N, Bernasconi A. Neurology 2016;86:643–650.

OBJECTIVE: To perform whole-brain morphometry in patients with frontal lobe epilepsy and evaluate the utility of group-level patterns for individualized diagnosis and prognosis. METHODS: We compared MRI-based cortical thickness and folding complexity between 2 frontal lobe epilepsy cohorts with histologically verified focal cortical dysplasia (FCD) (13 type I; 28 type II) and 41 closely matched controls. Pattern learning algorithms evaluated the utility of group-level findings to predict histologic FCD subtype, the side of the seizure focus, and postsurgical seizure outcome in single individuals. RESULTS: Relative to controls, FCD type I displayed multilobar cortical thinning that was most marked in ipsilateral frontal cortices. Conversely, type II showed thickening in temporal and postcentral cortices. Cortical folding also diverged, with increased complexity in prefrontal cortices in type I and decreases in type II. Group-level findings successfully guided automated FCD subtype classification (type I: 100%; type II: 96%), seizure focus lateralization (type I: 92%; type II: 86%), and outcome prediction (type I: 92%; type II: 82%). CONCLUSION: FCD subtypes relate to diverse whole-brain structural phenotypes. While cortical thickening in type II may indicate delayed pruning, a thin cortex in type I likely results from combined effects of seizure excitotoxicity and the primary malformation. Group-level patterns have a high translational value in guiding individualized diagnostics

Seizures, as a consequence of cortical dysplasias, are some of the most challenging to treat, as the semiology is disabling and highly refractory to medications. Surgical treatment provides a chance for seizure freedom or good seizure control for many of these patients. There is a marked improvement in surgical outcome if a resectable lesion can be found on imaging (1). As the search of these lesions has become extremely important, much effort has been spent on careful examination of neuroimaging of various modalities.

The rare visually gifted neuroradiologist exists who, by a seemingly unteachable process more akin to sorcery, finds the small critical lesion on the MRI. For the rest of us, much effort has been spent on a variety of technologies to make these lesions more easily appreciable over the past 20 years, during which neuroimaging has become widely available. General strategies have included using increasingly powerful MRI magnets with specially designed head coils to increase signal-to-noise ratio, post processing of images to highlight barely visible lesions, or imaging with radio nucleotide tracers. All of these processes are expensive, time consuming, and technically challenging. PET tracers such as alpha-methyl-tryptophan, have shown promise (2) but have not found widespread adoption; FDG imaging shows high sensitivity in temporal lobe epilepsies but are less sensitive in extratemporal epilepsies (3). Although MRI has had the greatest impact, findings on visual assessment that have included cortical thickening, gray-white matter junction blurring, abnormal gyral patterns, and abnormal signal intensities on MRI, can be small, subtle, indistinct, and difficult to interpret. The jump from 1.5T to 3T MRI scanners have made previously difficult to see lesions more visible, but the number of de novo new lesions detected is limited to an addition 5 to 8 percent (4, 5), which is critical on a per patient basis, but disappointing on a group-wise assessment. Early reports from 7T magnets, which may cost up to $7 million per scanner, suggest similar magnitude of advances (6). With each iteration, the number of lesions found is increasing but appears to be reaching an asymptotic limit. It is clear that a deeper understanding of the biological underpinnings of these lesions, particularly in relationship to imaging, is needed for further advances to be made.

The latest paper from the epilepsy neuroimaging laboratory of the Montreal Neurological Institute, Whole-brain MRI phenotyping in dysplasia-related frontal lobe epilepsy, is a step in that direction. By applying highly sophisticated post processing imaging techniques, previous works from this group have resulted in impressive advances in the detection of small dysplasias (7). In this work, they retrospectively examine two groups of focal cortical dysplasias (13 patients with types I, and 28 patients with type II) classified based on histological examination of resected tissue and compare them with a group of 41 matched controls. They report widespread cortical thinning in FCD type I and more focal cortical thickening in type II. By measuring curvature, a geometric index capturing the angulation of cortical gyri and sulci, they report increased folding complexity in type I and decreased complexity in type II.

The imaging/statistical technique in this study is a technical and computational tour-de-force, encompassing a decade of sophisticated post processing tool development, though it does raise concern, like any other study employing heavy filtering and smoothing, as to how much parameter selection can affect results. Another strength of this study is the careful separation of the two different types of FCDs. Further subdivision would have been ideal but likely unrealistic due to the number of patients required. The authors are also careful in eliminating visible lesions to avoid confounding group-wise results.

As the authors point out, results of this imaging study confirm previously observed findings of abnormal thinning in FCD type I, thickened cortical structures in FCD type II, and abnormal gyrational patterns (8). What is novel about this study is their usage of automated machine-learning techniques to accurately classify these patients into proper tissue classes, focus lateralization, and outcome in single patients. However, the validity of these results can only be achieved by an independent, preferably external dataset, or a prospective consecutive dataset. Until that is performed, the results must be interpreted with caution. This has presumably not been performed due to the small number of patients that necessarily exists at any one center, but would be amenable to a multi-institutional collaboration. This difficulty is not unique to this paper. Many of the MRI studies in FCDs have not been replicated outside the originating institutions, possibly due to the differences in imaging and clinical protocols (9), and the necessity of elaborate technical/statistical skills as well as hard to use imaging tools, many of which require an army of computer-savvy support staff–do not expect to perform these analyses on your home PC!

Their results do allow some observations made regarding the biology of the underlying disease. Progressive cortical thinning is a phenomenon that is seen in other epilepsies and may also be confounded by seizure duration and medication use. That the cortex is thickened is more difficult to explain and likely represent processes that are not fully understood; the authors postulate several possibilities, including delayed pruning and abnormal correlated growth, which remain to be verified.

All of these have implications for individual patient care in addition to adding to our understanding of the biology of this disease. It follows a pattern seen in temporal lobe epilepsies that also reveal that diffuse morphological changes with cortical thinning are associated with poorer postsurgical outcomes (10), and such findings can help counsel the patient regarding prognosis. It also re-emphasizes the fact that despite its name, this is a nonfocal disease, and that many patients with lesional excision have persistent seizures. Recent advances in the molecular mechanism of FCDs implicating the mTOR pathway have opened up exciting new areas of investigation (11), but although these may have been drivers of the disease, patients presenting with epilepsy already likely have abnormal functional networks (12) that may not necessarily be remediable by a “magic bullet.”

As such, sophisticated imaging and computational research remains an urgent priority in determining optimal treatment of these patients. Abnormal functional networks, their relation to morphometry, and surgical strategy, need to be clarified. It is notable, though, that all the images were obtained on 1.5T MRI scanners, currently two generations behind the latest scanners. One may be cautiously optimistic that studies with modern scanners and ever-increasing machine learning techniques may eventually lead to improved understanding of both the biology as well as provide individualized treatment options.