Categorizing cortical dysplasia lesions for surgical outcome using network functional connectivity

View More View Less
  • 1 Department of Physics, University of Cincinnati, Cincinnati;
  • | 2 Division of Neurology, Cincinnati Children’s Hospital Medical Center, Cincinnati;
  • | 3 Department of Radiology, Cincinnati Children’s Hospital Medical Center, Cincinnati;
  • | 4 Division of Pediatric Neurosurgery, Cincinnati Children’s Hospital Medical Center, Cincinnati;
  • | 5 Imaging Research Center, Department of Radiology, Cincinnati Children’s Hospital Medical Center, Cincinnati; and
  • | 6 Department of Radiology, University of Cincinnati College of Medicine, Cincinnati, Ohio
Free access

OBJECTIVE

Focal cortical dysplasia (FCD) is often associated with drug-resistant epilepsy, leading to a recommendation to surgically remove the seizure focus. Predicting outcome for resection of FCD is challenging, requiring a new approach. Lesion-symptom mapping is a powerful and broadly applicable method for linking neurological symptoms or outcomes to damage to particular brain regions. In this work, the authors applied lesion network mapping, an expansion of the traditional approach, to search for the association of lesion network connectivity with surgical outcomes. They hypothesized that connectivity of lesion volumes, preoperatively identified by MRI, would associate with seizure outcomes after surgery in a pediatric cohort with FCD.

METHODS

This retrospective study included 21 patients spanning the ages of 3 months to 17.7 years with FCD lesions who underwent surgery for drug-resistant epilepsy. The mean brain-wide functional connectivity map of each lesion volume was assessed across a database of resting-state functional MRI data from healthy children (spanning approximately 2.9 to 18.9 years old) compiled at the authors’ institution. Lesion connectivity maps were averaged across age and sex groupings from the database and matched to each patient. The authors sought to associate voxel-wise differences in these maps with subject-specific surgical outcome (seizure free vs persistent seizures).

RESULTS

Lesion volumes with persistent seizures after surgery tended to have stronger connectivity to attention and motor networks and weaker connectivity to the default mode network compared with lesion volumes with seizure-free surgical outcome.

CONCLUSIONS

Network connectivity–based lesion-outcome mapping may offer new insight for determining the impact of lesion volumes discerned according to both size and specific location. The results of this pilot study could be validated with a larger set of data, with the ultimate goal of allowing examination of lesions in patients with FCD and predicting their surgical outcomes.

ABBREVIATIONS

BOLD = blood oxygenation level–dependent; C-MIND = Cincinnati MR Imaging of Neuro-Development; DMN = default mode network; EEG = electroencephalography; FCD = functional cortical dysplasia; ILAE = International League Against Epilepsy; MNI = Montreal Neurological Institute; ROI = region of interest; rsfMRI = resting-state functional MRI.

OBJECTIVE

Focal cortical dysplasia (FCD) is often associated with drug-resistant epilepsy, leading to a recommendation to surgically remove the seizure focus. Predicting outcome for resection of FCD is challenging, requiring a new approach. Lesion-symptom mapping is a powerful and broadly applicable method for linking neurological symptoms or outcomes to damage to particular brain regions. In this work, the authors applied lesion network mapping, an expansion of the traditional approach, to search for the association of lesion network connectivity with surgical outcomes. They hypothesized that connectivity of lesion volumes, preoperatively identified by MRI, would associate with seizure outcomes after surgery in a pediatric cohort with FCD.

METHODS

This retrospective study included 21 patients spanning the ages of 3 months to 17.7 years with FCD lesions who underwent surgery for drug-resistant epilepsy. The mean brain-wide functional connectivity map of each lesion volume was assessed across a database of resting-state functional MRI data from healthy children (spanning approximately 2.9 to 18.9 years old) compiled at the authors’ institution. Lesion connectivity maps were averaged across age and sex groupings from the database and matched to each patient. The authors sought to associate voxel-wise differences in these maps with subject-specific surgical outcome (seizure free vs persistent seizures).

RESULTS

Lesion volumes with persistent seizures after surgery tended to have stronger connectivity to attention and motor networks and weaker connectivity to the default mode network compared with lesion volumes with seizure-free surgical outcome.

CONCLUSIONS

Network connectivity–based lesion-outcome mapping may offer new insight for determining the impact of lesion volumes discerned according to both size and specific location. The results of this pilot study could be validated with a larger set of data, with the ultimate goal of allowing examination of lesions in patients with FCD and predicting their surgical outcomes.

In Brief

Investigators applied lesion network mapping to children with focal cortical dysplasia who underwent surgery for drug-resistant epilepsy. Lesion volumes with persistent seizures after surgery tended to have stronger connectivity to attention and motor networks and weaker connectivity to the default mode network, compared with lesion volumes with seizure-free surgical outcomes. Network connectivity–based lesion-outcomes mapping may offer new insight for determining the impact of lesion volumes discerned according to both size and specific location.

Focal cortical dysplasia (FCD) occurs as a result of aberrant cell migration and differentiation during neurodevelopment that can cause malfunction of nerve cell activity, including seizures.1 FCD is a common finding among cases of pediatric epilepsy requiring surgery.2,3 Patients with FCD and drug-resistant epilepsy benefit from surgery to remove, ablate, or disconnect the seizure focus, but the long-term seizure-free outcome in most series is 40%–50%.4 An imaging-based approach could improve decision-making in pediatric epilepsy surgery if it can be used to predict outcome.

Lesion-symptom mapping studies investigate relationships between damage to different brain regions and behavioral deficits to make inferences about the functional neuroanatomy of linguistic or cognitive processes. The traditional approach to lesion-symptom mapping has the limitation that patients must be grouped by lesion or behavior, meaning that sometimes symptoms do not localize to a single region or lesions do not tend to spatially overlap in patients with common behavioral deficits. A more recent method, voxel-based lesion symptom mapping, does not require patients to be classified by either lesion or behavior. This method is used to study lesion-behavior relationships on a voxel-by-voxel basis.5

Resting-state functional MRI (rsfMRI) has proven useful for assessing the brain as a collection of networks of interregional connectivity. Low-frequency ( < 0.1 Hz) spontaneous fluctuations in blood oxygenation level–dependent (BOLD) signal have been observed when patients are not performing a task. Distinct brain regions can have synchronous resting fluctuations with spatial distributions that coincide with recognized functional networks from task experiments.6 By using this resting-state approach, one can explore the brain’s functional organization. A number of networks are consistently found in healthy subjects, at different stages of consciousness, and across species using resting-state functional connectivity.7,8

When comparing connectivity between patients, a particular region or “seed” might be identified as central to a network of interest, as determined by some a priori hypotheses. This approach extracts a representative BOLD time course from the voxels of the seed region and computes its correlation with all other voxels in the brain, generating a connectivity map. We can represent the seed time course as the mean of the voxel time courses in the seed or as the first eigenvariate of the signals from the seed region.9

In this work, we employed an expansion of lesion-symptom mapping that seeks an association between functional connectivity of regions affected by lesions and outcomes or behavior.10 Specifically, this approach generates connectivity maps corresponding to each lesion, used as a seed, in rsfMRI data from a large database of healthy individuals. The rationale behind this approach is that brain network connectivity disrupted by a lesion, rather than just the functionality of the lesioned region, may be more salient for prediction of symptoms, behavior, and outcomes.

The lesion network mapping method was first demonstrated by Boes et al.11 to study network connectivity of heterogeneously distributed subcortical lesions associated with peduncular hallucinosis and other syndromes. Subsequently, the approach has been used in studies of wide-ranging conditions in adults, including depression,12 amnesia,13 cervical dystonia,14 Parkinson’s disease,15 freezing of gait,16 and epilepsy,17 among others. Mithani et al.18 recently applied lesion network mapping to a cohort comprising both pediatric and adult patients who underwent a laser ablation procedure on epileptogenic lesions. Using a large collection of rsfMRI data from healthy adults, the authors discerned distinct lesion connectivity patterns between patients who were and those who were not seizure free.

This study applied a lesion network connectivity mapping method to lesions preoperatively identified by MRI in a group of children with FCD and drug-resistant epilepsy who underwent epilepsy surgery. We sought connectivity patterns related to the outcome of surgery for these patients using lesion volumes as seeds imposed on brains in a large database of healthy children. Our hypothesis was that there would be regional distinctions between lesion connectivity patterns according to surgical outcome.

Methods

Participants

We retrospectively collected data on 21 children (8 boys and 13 girls), ranging in age from 1.4 to 17.7 years (mean 7.9 years, SD 5.2 years). Children were identified from a larger, previously published data set.19 Study inclusion required presence of an FCD lesion on preoperative MRI, subsequent epilepsy surgery, and pathological diagnosis of type IIa or IIb FCD based on International League Against Epilepsy (ILAE) criteria.20 After surgery, resection status was assessed as complete or partial by either MRI or CT. At postsurgical follow-up, after a minimum of 12 months, outcome was classified in a binary manner as either seizure free (SF) with no auras, or persistent seizures (PS), based on the patient’s last office visit. Surgical outcome and other characteristics of the participants are summarized in Table 1. The study was approved by the IRB of Cincinnati Children’s Hospital Medical Center.

TABLE 1.

Participant and FCD lesion characteristics

Pt No.Age

(yrs)
SexPathology ILAE SyndromeScalp EEG LocalizationResection Status*ILAE Op OutcomeNo. of Matched Reference Subjects
15.8FIIBOFRt FC114
215.5FIIAFPRt FC115
36.3FIIASMARt parP114
46.6FIIBParRt parC114
517.1MIIAParRt parP315
610.0MIIAParRt parC113
74.9FIIBCingLt parC114
88.5MIIBFPRt FC113
94.7FIIBOFLt FC113
104.7FIIBCingRt parC313
112.5FIIAAmygLt TC113
1217.7FIIAAmygLt TC115
1316.3FIIASMABilat FC315
147.3MIIAFPRt FC113
157.7FIIAParRt FC514
168.5MIIBSMARt FC113
1710.9FIIBCingRt FC113
184.1MIIBMotLt FC110
193.3MIIBSMALt FC313
201.4MIIAMotLt FC113
212.4FIIBSMALt OC113

Amyg = amygdala; cing = cingulate; F = frontal; FP = frontoparietal; mot = motor; O = occipital; OF = orbitofrontal; par = parietal; pt = patient; SMA = supplementary motor area; T = temporal.

C = complete resection; P = partial resection with >2/3 resected, assessed by MRI and CT imaging.

1, seizure free, no auras; 3, 1–3 seizure days per year, with or without auras; and 5, less than 50% reduction of baseline seizure days to 100% increase of baseline seizure days, with or without auras.

Number of unique individuals from the C-MIND database comprising the 15 data sets matched to each participant.

Lesion and Seizure Characterizations

Electroencephalography (EEG) data before surgery were acquired using standard international 10–20 placement. Results were reviewed using a standard timescale by a neurologist after applying a 1-Hz high-pass filter, a 70-Hz low-pass filter, and a notch filter to remove 60-Hz artifact. Digital signals were reviewed using a minimum of two montages: anteroposterior bipolar and either average reference or transverse montages.

Each seizure type, as determined by detailed history in office notes, was reviewed independently for video EEG localization. Up to 3 seizures were reviewed per seizure type. Seizure onset was carefully reviewed prior to time of first clinical change. Localization by EEG was determined based on a consistent early ictal change. Semiology was determined by review of office notes for aura, if any, and video of reviewed seizures, fitting the patient’s semiology to ILAE syndrome.20 Details of seizure characteristics per participant are included in Table 1.

Clinical Imaging and Lesion Identification

MRI was acquired at 3T field strength prior to surgery as part of routine clinical protocol for patients with seizures. The clinical protocol included sequences with a variety of contrasts, including T1-weighted, T2-weighted, and susceptibility-weighted images together with diffusion tensor imaging and cerebral blood flow via arterial spin labeling.

The 3D T1-weighted images were originally acquired at narrowly varying resolution. T1-weighted images were subjected to isotropic transformation in Brain Voyager QX (Brain Innovation) as necessary to result in 1-mm3 isotropic resolution. Lesions were traced in 3 dimensions on the resulting anatomical images by a board-certified neuroradiologist (J.L.), with 23 years of experience in epilepsy imaging, using MRIcron (version 1, 2015, https://www.nitrc.org/projects/mricron). Representations of each lesion and associated anatomy in all available imaging sequences, viewed simultaneously on a clinical PACS system (MergePACS, Merge Healthcare), aided the tracing procedure. For each lesion, a 3D region-of-interest (ROI) mask was generated by manual tracing.

Neuroimaging Database of Healthy Children

Data used for connectivity analysis were obtained from a database known as Cincinnati MR Imaging of Neuro-Development (C-MIND, https://research.cchmc.org/c-mind/manual-project-overview), collected by the Pediatric Functional Neuroimaging Research Network under National Institutes of Health contract HHSN275200900018C.21 We extracted structural and rsfMRI data from C-MIND for 129 healthy children, 46 of whom underwent scanning twice, of whom 25 underwent scanning a third time at intervals of approximately 1 year, resulting in a total of 200 imaging data sets. Among the 200 data sets, the age at acquisition was evenly distributed in the range from 2.9 to 18.9 years (86 males and 114 females). All rsfMRI data used for our analysis were acquired in children without sedation and while awake, via gradient-echo echo-planar imaging (EPI), with the following parameters: TR 2000 msec, TE 35 msec, in-plane FOV 240 × 240 mm, matrix 80 × 80, and 36 slices with 4-mm thickness. Each resting-state session collected 150 image volumes over a period of 5 minutes while the subjects were instructed to relax with their eyes fixated on a white cross at the center of a black screen. High-resolution T1-weighted structural images were collected using a 3D MPRAGE sequence with the following parameters:22 TR 8.1 msec, TE 3.7 msec, TI 939 msec, flip angle 8°, FOV 256 × 224 × 160 mm, 1 mm isotropic resolution. Imaging in young children was accomplished via a behavioral protocol, including tangible reinforcement, introduction to the scanner environment, and practice.23

Data Analysis

Preprocessing

The anatomical (T1-weighted) volumes and lesion masks for the FCD patients were preprocessed using Statistical Parametric Mapping software (SPM12, http://www.fil.ion.ucl.ac.uk/spm/software/spm12/) in the MATLAB computing environment (The MathWorks). The T1-weighted anatomical clinical image of each participant was first transformed into the Montreal Neurological Institute (MNI) standard space using an iterative approach24 that also segmented the volume into gray matter, white matter, and CSF tissue classes. Once the transformation was established for the anatomical image, it was applied to the lesion mask to localize the lesion in standard space.

Each rsfMRI data set from the C-MIND database was also preprocessed in SPM12 beginning with rigid-body realignment of all functional images in each rsfMRI data set to the first image of the series using 3 translational and 3 rotational adjustments per image. After coregistration of the corresponding structural image to the mean realigned functional image, the unified segmentation and normalization process described above was carried out on the structural image. The resulting normalization transformation was applied to the functional series and to the tissue segmentation maps. Normalized rsfMRI data were then fed into the CONN Toolbox, v17a; MATLAB-based software that works with SPM12 for the computation and display of functional connectivity (https://www.nitrc.org/projects/conn).25 Functional connectivity, defined as seed-to-voxel or ROI-to-ROI correlation of BOLD signal from the rsfMRI series, was calculated in CONN after accounting for potential confounds from motion and other physiological sources. This was achieved by applying the aCompCor26 approach of regressing out the first 16 eigenvariates and their derivatives extracted from the brain regions masked as white matter and CSF with the rationale that these signals would be unrelated to neuronal activity. The 6 motion parameters from realignment and their derivatives were also regressed out as confounds. Masks corresponding to the normalized lesions were then applied in CONN as seed regions. Brain-wide voxel-wise BOLD signal correlation to the mean signal within each lesion mask was assessed to create a connectivity map for each lesion for each of the 200 imaging data sets extracted from C-MIND.

Statistical Analysis

The focus of our statistical analysis was to examine how the surgical outcome (seizure free vs persistent seizures) might be related to specific functional connectivity patterns (network connectivity) in healthy brains corresponding to regions occupied by FCD lesions. The representative connectivity map for each FCD participant’s lesion was generated as the mean seed-to-voxel connectivity map using the lesion mask as seed among the 15 C-MIND data sets matched most closely to the participant according to sex and age. For the participants in the age range of 3–15 years, the closest 15 data sets from C-MIND were within 1 year of the participant’s age. Those participants in the extreme age ranges of 0–3 years and 15–18 years could be matched to 15 C-MIND data sets within 3 years of age. Matching by age and sex was optimized for participants by including longitudinal data from some of the C-MIND subjects. Nevertheless, a median of 13 independent C-MIND subjects were used to match the study participants, with a minimum of 10 and a maximum of 15, as detailed in Table 1.

Our strategy of using matched healthy references for each FCD participant is based on evidence that connectivity in healthy children depends on age and sex, with age dependence steepening as age decreases.27

The mean connectivity map for each FCD lesion was generated in SPM12 across the age- and sex-matched lesion seed-to-voxel connectivity maps from C-MIND. Following this, we compared the mean connectivity maps between FCD participant groups according to surgical outcome, with adjustment for lesion type. Group differences were assessed via a two-sample t-test in SPM12 including lesion type as covariate. Lesion type was included as a remaining distinction among participants that plausibly impacts surgical outcome.28,29 After applying a nominal voxel-wise threshold to the T-maps, we discerned clusters of voxels with significant differences in connectivity to lesions associated with different surgical outcomes (seizure free vs persistent seizures). We report outcomes reaching cluster p values < 0.05, both family-wise error corrected and uncorrected for multiple comparisons.

Results

The Lesions

FCD lesions varied in location and extent with only limited spatial overlap. Figure 1 displays the overlap of all lesions and of the persistent seizure and seizure-free cases separately. The maximum spatial overlap is 3 lesions in all groupings, limited to the right postcentral gyrus. The delineated lesions for all subjects individually are shown in Supplementary Fig. 1. Surgery resulted in complete resection of 19 of the 21 lesions. Partial resections (> two-thirds of the lesion) were achieved for 2 lesions, 1 in the seizure-free group and 1 in the persistent seizure group (Table 1). There was not a significant difference in the frequency of partial resections between groups (χ2 = 0.83, p = 0.36). In consideration of the low and balanced frequency of partial resections and the limited sample size, we did not exclude or adjust according to resection status in subsequent analyses.

FIG. 1.
FIG. 1.

Overlay of all lesions, persistent seizures (PS) lesions only, and seizure-free (SF) lesions only. Colors indicate how many distinct lesions occupy the same space. MNI slice coordinates are shown at the top of each slice. Neurological orientation convention used. Figure is available in color online only.

Dependence of Surgical Outcome on Lesion-Specific Connectivity

Average connectivity maps, generated for each lesion as seed among healthy brains matched for age and sex, were collected for the persistent seizure and seizure-free groups of FCD patients. Voxel-wise comparisons of the mean connectivity, across lesions for the persistent seizure and seizure-free groups of surgical outcome, showed that lesions for which seizures persisted after surgery tended to have more positive connectivity to frontoparietal, anterior cingulate, and insular regions (“hot” clusters in Fig. 2A and Table 2). The mean connectivity maps across lesions are shown for the persistent seizure and seizure-free groups in Fig. 2B and C, respectively, for reference. Inspection of the mean connectivity maps reveals that for the frontoparietal regions and more inferior parts of the insulae, positive correlations for both categories of lesion were compared, whereas the anterior cingulate and more superior parts of the insulae underwent a change in sign from negative (seizure free) to positive (persistent seizures) correlation in the comparison. The “cool” clusters in Fig. 2A, on the other hand, illustrate that more negative correlation was found for lesions with persistent seizures to the posterior cingulate, medial prefrontal cortex, and angular gyri. Referring again to Fig. 2B and C, the posterior cingulate tended to be negatively correlated to lesions corresponding to both surgical outcomes, while the medial frontal and angular gyri tended to be positively correlated to seizure-free lesions but changing to negative correlation for lesions associated with persistent seizures after surgery.

FIG. 2.
FIG. 2.

Regions that distinguish connectivity strength to lesions associated with seizure-free surgical outcome and lesions associated with persistent seizures after surgery. A: Persistent seizures versus seizure free contrast. Hot colors indicate seizure free > persistent seizures. Cool colors indicate persistent seizures > seizure free. B: Mean connectivity map for lesions. C: Mean connectivity map for seizure-free sessions. Color bars for positive and negative connectivity are shown at the bottom. Neurological orientation convention used. Figure is available in color online only.

TABLE 2.

Voxel clusters reflecting connectivity contrast to FCD lesions with persistent seizures versus seizure-free surgical outcomes

ContrastAnatomical Regions HemispherePeak MNI coordinatesp Valuek
XYZ
PS > SFMid cingulum

SMA
Lt, rt

Lt, rt
08320.007*2582
Supramarginal gyrus

Insula

Sup temporal pole

Rolandic operc

Inf frontal operc

Sup temporal gyrus

Precentral gyrus

Postcentral gyrus
Rt

Rt

Rt

Rt

Rt

Rt

Rt

Rt
70−26200.002*3239
Supramarginal gyrus

Sup temporal gyrus

Postcentral gyrus
Lt

Lt

Lt
−66−32260.04694
Sup temporal pole

Insula

Rolandic operc

Inf frontal operc

Precentral gyrus
Lt

Lt

Lt

Lt

Lt
−568−20.03787
SF > PSMed orbital frontal

Sup orbital frontal
Lt, rt

Lt
−1268−80.011508
Vermis

Precuneus

Cuneus

Post cingulum

Mid cingulum
Lt, rt

Lt, rt

Lt, rt

Lt, rt

Lt, rt
−8−50420.011688
Angular gyrus

Mid occipital
Lt

Lt
−42−82340.07746
Mid temporal

Angular gyrus

Mid occipital
Rt

Rt

Rt
40−62260.09640

Inf = inferior; mid = middle; operc = operculum; PS = persistent seizures; SF = seizure free; SMA = supplementary motor area; sup = superior.

Family-wise error corrected, k = cluster size (voxels), with 2 × 2 × 2–mm voxel size.

Cluster-level p value, uncorrected.

Discussion

This work suggests that the outcomes of surgery to resect FCD lesions associated with seizures in children may depend on the functional connectivity of the regions occupied by the lesions in healthy brains. We utilized a lesion network mapping technique seeking voxels with the largest differences in their connectivity with lesions whose resection resulted in persistent seizures compared with lesions with a seizure-free outcome of resection.

The surgical outcome analysis indicated that lesions associated with persistent seizures after surgery tend to have more positive correlation to supplementary motor area (SMA), precentral, postcentral, and midcingulate brain regions attributed to sensorimotor processing and to elements of the salience network, including opercular and insular regions. Parts of the cingulate and insulae, in fact, underwent a change in sign, from negative correlation with lesions that had SF surgical outcomes to positive correlation with lesions resulting in PS. This may suggest that patients with lesions occupying and disrupting components that play a strong role in key functional networks associated with sensory and active task-related processing will be more likely to maintain seizures after resection. In this case, the lesions may appear to be surgically amenable to resection in standard practice because they are not located in areas known to be critical for function. However, they may reside in nodes of distributed networks important for engagement, such as those known as attention and salience networks. Other investigators have demonstrated that seizure generation and propagation, including in FCD patients, is a network phenomenon,30 wherein seizure initiation may depend on aberrant connectivity to regions distant from visualized lesions by virtue of being components of an epileptic network. Extending this idea, our results suggest that when an FCD lesion occupies a region that normally serves as a node positively connected within a network used for sensory, attentional, engagement, that network is more likely to continue generating seizures even after removal of the lesioned node.

Another commonly recognized functional network, the default mode network (DMN), was also associated with different surgical outcomes. Comprising the posterior cingulate, medial prefrontal cortex, the angular gyri, and other regions, the DMN is understood to be more active during introspection and disengaged when the brain is actively performing an attention-demanding task.31 Our surgical outcome analysis of lesion connectivity resulted in more negative correlation between DMN regions and lesions for which surgery resulted in persistent seizures. In medial prefrontal regions and angular gyri, the sign of correlation went from positive for seizure-free lesions to negative for lesions associated with persistent seizures. Thus, lesions that occupied nodes most anticorrelated to the DMN tended to have poorer surgical outcomes. This result aligns with the idea, expressed above, that persistent seizures after surgery are associated with lesion membership in task-relevant networks that tend to be anticorrelated to the DMN.

The results of this study are consistent with those of Mithani et al.18 for outcomes of ablative removal of epileptogenic lesions if we consider differences in approach to lesion network mapping. That study found that being seizure free after ablation was associated with connectivity to the orbitofrontal cortex and opercular inferior frontal cortex. Lack of seizure freedom was associated with connectivity to elements of the DMN, hippocampus, amygdala, cerebellum, and temporal regions. Their approach, however, binarizes lesion connectivity maps without consideration of the sign of connectivity. Thus, in this light, there is alignment with our findings of increased positive connectivity to orbitofrontal regions in seizure-free patients and deeper negative connectivity to the DMN for patients with persistent seizures.

Functional neuroimaging of brains with FCD consistently finds alterations of global network architecture.32,33 Investigating the connectivity of the lesion location itself, however, is challenging to interpret because the tissue has been damaged. The technique we employed in this study, which investigates connectivity of the lesion location based on a cohort of healthy subjects, does not require functional imaging of the patient. We thereby associate lesions with surgical outcome not only according to the location and the size of lesions but also by the way they connect to canonical task-related networks and the DMN.

There are some limitations that need to be considered when interpreting the outcomes of this study. The lesions encountered in our cohort of patients varied considerably in size, even though they did not spatially overlap to a high degree. For the larger lesions, localizing results to specific areas in the brain is difficult. The approach we took is most advantageous under circumstances when lesions are small with minimal spatial overlap. The limited size of our retrospective cohort of patients constrained statistical power to the extent that some of the reported voxel clusters did not survive correction for multiple comparisons. In addition, surgical outcomes among the patients were imbalanced (5/21 with persistent seizures). In light of this, the outcomes serve as a basis for additional studies testing this novel method on larger cohorts of FCD patients.

We investigated a focused analysis using age- and sex-based matching of our healthy cohort to each FCD patient, given the potential for age- and sex-related variation in healthy brain connectivity patterns. Our healthy brain database, C-MIND, provided only about 15 data sets reasonably matched in age and sex to each FCD patient. Future development of our analysis approach would require larger age- and sex-specific databases of functional connectivity data for healthy subjects.

Conclusions

Consideration of the functional connectivity of regions occupied by FCD lesions in healthy brains can provide insights for outcomes of surgery for lesion resection. Outcomes suggest that lesions more strongly anticorrelated to the DMN and more strongly correlated to other, task-related networks are less likely to result in freedom from seizures after resection. FCD lesion mapping based on healthy brain network connectivity is a potentially important technique for developing a biomarker of seizure outcome after lesion resection.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: DiFrancesco, Greiner, Leach. Acquisition of data: Greiner, Leach. Analysis and interpretation of data: all authors. Drafting the article: Bdaiwi. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: DiFrancesco. Statistical analysis: DiFrancesco, Bdaiwi. Study supervision: DiFrancesco, Greiner.

Supplemental Information

Online-Only Content

Supplemental material is available with the online version of the article.

Previous Presentations

A related abstract of this work was presented at the Organization of Human Brain Mapping meeting, Rome, Italy, June 3–13, 2019.

References

  • 1

    Tahta A, Turgut M. Focal cortical dysplasia: etiology, epileptogenesis, classification, clinical presentation, imaging, and management. Childs Nerv Syst. 2020;36(12):29392947.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Hauptman JS, Mathern GW. Surgical treatment of epilepsy associated with cortical dysplasia: 2012 update. Epilepsia. 2012;53(suppl 4):98104.

  • 3

    Lerner JT, Salamon N, Hauptman JS, Velasco TR, Hemb M, et al. Assessment and surgical outcomes for mild type I and severe type II cortical dysplasia: a critical review and the UCLA experience. Epilepsia. 2009;50(6):13101335.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Guerrini R, Duchowny M, Jayakar P, Krsek P, Kahane P, et al. Diagnostic methods and treatment options for focal cortical dysplasia. Epilepsia. 2015;56(11):16691686.

  • 5

    Bates E, Wilson SM, Saygin AP, Dick F, Sereno MI, et al. Voxel-based lesion–symptom mapping. Nat Neurosci. 2003;6(5):448450.

  • 6

    Lee MH, Smyser CD, Shimony JS. Resting-state fMRI: a review of methods and clinical applications. AJNR Am J Neuroradiol. 2013;34(10):18661872.

  • 7

    Cole DM, Smith SM, Beckmann CF. Advances and pitfalls in the analysis and interpretation of resting-state FMRI data. Front Syst Neurosci. 2010;4:8.

  • 8

    Rosazza C, Minati L. Resting-state brain networks: literature review and clinical applications. Neurol Sci. 2011;32(5):773785.

  • 9

    Poldrack RA, Mumford JA, Nichols TE. Handbook of Functional MRI Data Analysis. Cambridge University Press;2011.

  • 10

    Fox MD. Mapping symptoms to brain networks with the human connectome. N Engl J Med. 2018;379(23):22372245.

  • 11

    Boes AD, Prasad S, Liu H, Liu Q, Pascual-Leone A, et al. Network localization of neurological symptoms from focal brain lesions. Brain. 2015;138(Pt 10):30613075.

  • 12

    Padmanabhan JL, Cooke D, Joutsa J, Siddiqi SH, Ferguson M, et al. A human depression circuit derived from focal brain lesions. Biol Psychiatry. 2019;86(10):749758.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Ferguson MA, Lim C, Cooke D, Darby RR, Wu O, et al. A human memory circuit derived from brain lesions causing amnesia. Nat Commun. 2019;10(1):3497.

  • 14

    Corp DT, Joutsa J, Darby RR, Delnooz CCS, van de Warrenburg BPC, et al. Network localization of cervical dystonia based on causal brain lesions. Brain. 2019;142(6):16601674.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Weil RS, Hsu JK, Darby RR, Soussand L, Fox MD. Neuroimaging in Parkinson’s disease dementia: connecting the dots. Brain Commun. 2019;1(1):fcz006.

  • 16

    Fasano A, Laganiere SE, Lam S, Fox MD. Lesions causing freezing of gait localize to a cerebellar functional network. Ann Neurol. 2017;81(1):129141.

  • 17

    Mansouri AM, Germann J, Boutet A, Elias GJB, Mithani K, et al. Identification of neural networks preferentially engaged by epileptogenic mass lesions through lesion network mapping analysis. Sci Rep. 2020;10(1):10989.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Mithani K, Boutet A, Germann J, Elias GJB, Weil AG, et al. Lesion network localization of seizure freedom following MR-guided laser interstitial thermal ablation. Sci Rep. 2019;9(1):18598.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Maynard LM, Leach JL, Horn PS, Spaeth CG, Mangano FT, et al. Epilepsy prevalence and severity predictors in MRI-identified focal cortical dysplasia. Epilepsy Res. 2017;132:4149.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Commission on Classification and Terminology of the International League Against Epilepsy.Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia. 1989;30(4):389399.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Vannest J, Rajagopal A, Cicchino ND, Franks-Henry J, Simpson SM, et al. Factors determining success of awake and asleep magnetic resonance imaging scans in nonsedated children. Neuropediatrics. 2014;45(6):370377.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Mugler JP III, Brookeman JR. Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE). Magn Reson Med. 1990;15(1):152157.

  • 23

    Vannest J, Rajagopal A, Cicchino ND, Franks-Henry J, Simpson SM, et al. Factors determining success of awake and asleep magnetic resonance imaging scans in nonsedated children. Neuropediatrics. 2014;45(6):370377.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Ashburner J, Friston KJ. Unified segmentation. Neuroimage. 2005;26(3):839851.

  • 25

    Whitfield-Gabrieli S, Nieto-Castanon A. Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect. 2012;2(3):125141.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Behzadi Y, Restom K, Liau J, Liu TT. A component based noise correction method (CompCor) for BOLD and perfusion based fMRI. Neuroimage. 2007;37(1):90101.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Stiles J, Jernigan TL. The basics of brain development. Neuropsychol Rev. 2010;20(4):327348.

  • 28

    Leach JL, Greiner HM, Miles L, Mangano FT. Imaging spectrum of cortical dysplasia in children. Semin Roentgenol. 2014;49(1):99111.

  • 29

    Leach JL, Miles L, Henkel DM, Greiner HM, Kukreja MK, et al. Magnetic resonance imaging abnormalities in the resection region correlate with histopathological type, gliosis extent, and postoperative outcome in pediatric cortical dysplasia. J Neurosurg Pediatr. 2014;14(1):6880.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Kramer MA, Cash SS. Epilepsy as a disorder of cortical network organization. Neuroscientist. 2012;18(4):360372.

  • 31

    Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A. 2001;98(2):676682.

  • 32

    DeMarco AT, Turkeltaub PE. Functional anomaly mapping reveals local and distant dysfunction caused by brain lesions. Neuroimage. 2020;215:116806.

  • 33

    Eldaief MC, McMains S, Hutchison RM, Halko MA, Pascual-Leone A. Reconfiguration of intrinsic functional coupling patterns following circumscribed network lesions. Cereb Cortex. 2017;27(5):28942910.

    • PubMed
    • Search Google Scholar
    • Export Citation

Supplementary Materials

  • View in gallery

    Overlay of all lesions, persistent seizures (PS) lesions only, and seizure-free (SF) lesions only. Colors indicate how many distinct lesions occupy the same space. MNI slice coordinates are shown at the top of each slice. Neurological orientation convention used. Figure is available in color online only.

  • View in gallery

    Regions that distinguish connectivity strength to lesions associated with seizure-free surgical outcome and lesions associated with persistent seizures after surgery. A: Persistent seizures versus seizure free contrast. Hot colors indicate seizure free > persistent seizures. Cool colors indicate persistent seizures > seizure free. B: Mean connectivity map for lesions. C: Mean connectivity map for seizure-free sessions. Color bars for positive and negative connectivity are shown at the bottom. Neurological orientation convention used. Figure is available in color online only.

  • 1

    Tahta A, Turgut M. Focal cortical dysplasia: etiology, epileptogenesis, classification, clinical presentation, imaging, and management. Childs Nerv Syst. 2020;36(12):29392947.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Hauptman JS, Mathern GW. Surgical treatment of epilepsy associated with cortical dysplasia: 2012 update. Epilepsia. 2012;53(suppl 4):98104.

  • 3

    Lerner JT, Salamon N, Hauptman JS, Velasco TR, Hemb M, et al. Assessment and surgical outcomes for mild type I and severe type II cortical dysplasia: a critical review and the UCLA experience. Epilepsia. 2009;50(6):13101335.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Guerrini R, Duchowny M, Jayakar P, Krsek P, Kahane P, et al. Diagnostic methods and treatment options for focal cortical dysplasia. Epilepsia. 2015;56(11):16691686.

  • 5

    Bates E, Wilson SM, Saygin AP, Dick F, Sereno MI, et al. Voxel-based lesion–symptom mapping. Nat Neurosci. 2003;6(5):448450.

  • 6

    Lee MH, Smyser CD, Shimony JS. Resting-state fMRI: a review of methods and clinical applications. AJNR Am J Neuroradiol. 2013;34(10):18661872.

  • 7

    Cole DM, Smith SM, Beckmann CF. Advances and pitfalls in the analysis and interpretation of resting-state FMRI data. Front Syst Neurosci. 2010;4:8.

  • 8

    Rosazza C, Minati L. Resting-state brain networks: literature review and clinical applications. Neurol Sci. 2011;32(5):773785.

  • 9

    Poldrack RA, Mumford JA, Nichols TE. Handbook of Functional MRI Data Analysis. Cambridge University Press;2011.

  • 10

    Fox MD. Mapping symptoms to brain networks with the human connectome. N Engl J Med. 2018;379(23):22372245.

  • 11

    Boes AD, Prasad S, Liu H, Liu Q, Pascual-Leone A, et al. Network localization of neurological symptoms from focal brain lesions. Brain. 2015;138(Pt 10):30613075.

  • 12

    Padmanabhan JL, Cooke D, Joutsa J, Siddiqi SH, Ferguson M, et al. A human depression circuit derived from focal brain lesions. Biol Psychiatry. 2019;86(10):749758.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Ferguson MA, Lim C, Cooke D, Darby RR, Wu O, et al. A human memory circuit derived from brain lesions causing amnesia. Nat Commun. 2019;10(1):3497.

  • 14

    Corp DT, Joutsa J, Darby RR, Delnooz CCS, van de Warrenburg BPC, et al. Network localization of cervical dystonia based on causal brain lesions. Brain. 2019;142(6):16601674.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Weil RS, Hsu JK, Darby RR, Soussand L, Fox MD. Neuroimaging in Parkinson’s disease dementia: connecting the dots. Brain Commun. 2019;1(1):fcz006.

  • 16

    Fasano A, Laganiere SE, Lam S, Fox MD. Lesions causing freezing of gait localize to a cerebellar functional network. Ann Neurol. 2017;81(1):129141.

  • 17

    Mansouri AM, Germann J, Boutet A, Elias GJB, Mithani K, et al. Identification of neural networks preferentially engaged by epileptogenic mass lesions through lesion network mapping analysis. Sci Rep. 2020;10(1):10989.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Mithani K, Boutet A, Germann J, Elias GJB, Weil AG, et al. Lesion network localization of seizure freedom following MR-guided laser interstitial thermal ablation. Sci Rep. 2019;9(1):18598.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Maynard LM, Leach JL, Horn PS, Spaeth CG, Mangano FT, et al. Epilepsy prevalence and severity predictors in MRI-identified focal cortical dysplasia. Epilepsy Res. 2017;132:4149.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Commission on Classification and Terminology of the International League Against Epilepsy.Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia. 1989;30(4):389399.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Vannest J, Rajagopal A, Cicchino ND, Franks-Henry J, Simpson SM, et al. Factors determining success of awake and asleep magnetic resonance imaging scans in nonsedated children. Neuropediatrics. 2014;45(6):370377.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Mugler JP III, Brookeman JR. Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE). Magn Reson Med. 1990;15(1):152157.

  • 23

    Vannest J, Rajagopal A, Cicchino ND, Franks-Henry J, Simpson SM, et al. Factors determining success of awake and asleep magnetic resonance imaging scans in nonsedated children. Neuropediatrics. 2014;45(6):370377.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Ashburner J, Friston KJ. Unified segmentation. Neuroimage. 2005;26(3):839851.

  • 25

    Whitfield-Gabrieli S, Nieto-Castanon A. Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect. 2012;2(3):125141.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Behzadi Y, Restom K, Liau J, Liu TT. A component based noise correction method (CompCor) for BOLD and perfusion based fMRI. Neuroimage. 2007;37(1):90101.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Stiles J, Jernigan TL. The basics of brain development. Neuropsychol Rev. 2010;20(4):327348.

  • 28

    Leach JL, Greiner HM, Miles L, Mangano FT. Imaging spectrum of cortical dysplasia in children. Semin Roentgenol. 2014;49(1):99111.

  • 29

    Leach JL, Miles L, Henkel DM, Greiner HM, Kukreja MK, et al. Magnetic resonance imaging abnormalities in the resection region correlate with histopathological type, gliosis extent, and postoperative outcome in pediatric cortical dysplasia. J Neurosurg Pediatr. 2014;14(1):6880.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Kramer MA, Cash SS. Epilepsy as a disorder of cortical network organization. Neuroscientist. 2012;18(4):360372.

  • 31

    Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A. 2001;98(2):676682.

  • 32

    DeMarco AT, Turkeltaub PE. Functional anomaly mapping reveals local and distant dysfunction caused by brain lesions. Neuroimage. 2020;215:116806.

  • 33

    Eldaief MC, McMains S, Hutchison RM, Halko MA, Pascual-Leone A. Reconfiguration of intrinsic functional coupling patterns following circumscribed network lesions. Cereb Cortex. 2017;27(5):28942910.

    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 2230 1942 0
Full Text Views 565 511 390
PDF Downloads 223 154 27
EPUB Downloads 0 0 0