Skip to main content
NCBI home page
Cerebral Cortex (New York, NY) logoLink to Cerebral Cortex (New York, NY)
. 2017 Nov 23;28(10):3665–3672. doi: 10.1093/cercor/bhx247

Corpus Callosum White Matter Diffusivity Reflects Cumulative Neurological Comorbidity in Tuberous Sclerosis Complex

Fiona M Baumer 1,#,2, Jurriaan M Peters 1,2,3,#, Sean Clancy 2,4, Anna K Prohl 2,4, Sanjay P Prabhu 2,4, Benoit Scherrer 2,4, Floor E Jansen 3, Kees P J Braun 3, Mustafa Sahin 1, Aymeric Stamm 2,4,5, Simon K Warfield 2,4,✉
PMCID: PMC6132277  PMID: 29939236

Abstract

Introduction: Neurological manifestations in Tuberous Sclerosis Complex (TSC) are highly variable. Diffusion tensor imaging (DTI) may reflect the neurological disease burden. We analyzed the association of autism spectrum disorder (ASD), intellectual disability (ID) and epilepsy with callosal DTI metrics in subjects with and without TSC. Methods: 186 children underwent 3T MRI DTI: 51 with TSC (19 with concurrent ASD), 46 with non-syndromic ASD and 89 healthy controls (HC). Subgroups were based on presence of TSC, ASD, ID, and epilepsy. Density-weighted DTI metrics obtained from tractography of the corpus callosum were fitted using a 2-parameter growth model. We estimated distributions using bootstrapping and calculated half-life and asymptote of the fitted curves. Results: TSC was associated with a lower callosal fractional anisotropy (FA) than ASD, and ASD with a lower FA than HC. ID, epilepsy and ASD diagnosis were each associated with lower FA values, demonstrating additive effects. In TSC, the largest change in FA was related to a comorbid diagnosis of ASD. Mean diffusivity (MD) showed an inverse relationship to FA. Some subgroups were too small for reliable data fitting. Conclusions: Using a cross-disorder approach, this study demonstrates cumulative abnormality of callosal white matter diffusion with increasing neurological comorbidity.

Keywords: autism spectrum disorder, diffusion tensor imaging, epilepsy, intellectual disability, tuberous sclerosis complex

Introduction

Tuberous sclerosis complex (TSC) is a genetic multisystem disorder causing formation of hamartomatous lesions in various organs including the brain, where they are referred to as tubers. TSC is associated with severe neurological manifestations including epilepsy (90%), intellectual disability (ID, 40%), and autism spectrum disorder (ASD, 50%) (Curatolo et al. 2008). With a prevalence of 1 in 6000, it is one of the most common genetic causes of epilepsy and ASD (Davis et al. 2015).

TSC is a model disease for studying both epilepsy and ASD, and their co-occurrence. Early onset epilepsy and early cognitive delays (Jansen et al. 2008; Numis et al. 2011; Jeste et al. 2014) are associated with ASD in TSC. Moreover, improved cognitive outcome with early and aggressive treatment of infantile spasms has been reported (Bombardieri et al. 2010; Jozwiak et al. 2011), suggesting an epileptic encephalopathy contributes to poor neurodevelopmental outcome. In addition, the autism phenotype in TSC is comparable to nonsyndromic ASD (Bruining et al. 2014; Jeste et al. 2016). Understanding the development of these conditions in TSC will not only offer opportunities for targeted interventions in TSC, but will also increase insight in other populations with epilepsy and ASD.

While the identification of early clinical risk factors in a patient can guide therapeutic decisions, effects of even earlier, pre-emptive treatments with mTOR inhibitors or vigabatrin on neurocognitive outcomes are now being studied (Julich and Sahin 2014). Such research, however, would significantly benefit from a robust imaging biomarker. An ideal marker should be reliably reproducible, reflect the underlying neurobiology, correlate with symptom severity and change in response to treatment.

Diffusion tensor imaging (DTI) bares such promise, as it noninvasively probes microstructural properties of the brain by characterizing the average ability of water molecules to diffuse at the tissue microscopic level. DTI measures in TSC are abnormal even in structurally normal appearing white matter and were reported to change with exposure to mTOR inhibitors (Tillema et al. 2012). Decreased fractional anisotropy (FA) and increased mean diffusivity (MD) have been reported in 2 large white matter tracts—the corpus callosum and the arcuate fasciculus—in subjects with TSC and ASD when compared to subjects with TSC but without ASD (Peters et al. 2012; Lewis et al. 2013). These studies, however, did not account for the presence of ID and epilepsy. Altered diffusion of the callosal white matter has also been reported in nonsyndromic ASD (Alexander et al. 2007), with greater abnormalities in those subjects with ID. It remains unclear, therefore, if in TSC the altered diffusion is associated specifically with ASD or is also reflective of other neurologic comorbidities.

The aim of this study was to examine the association of callosal white matter microstructural integrity with concurrent neurological manifestations in both TSC and ASD. To this end, we compared DTI values among 3 groups: TSC, nonsyndromic ASD, and healthy controls (HC), and assessed impact of co-morbid ASD, ID and epilepsy.

Materials and Methods

Subjects

Fifty-one patients with TSC, none treated with mTOR inhibitors, who had undergone 3T MRI were identified through the Boston Children’s Hospital Multidisciplinary Tuberous Sclerosis Program. All patients were diagnosed with definite TSC based on clinical or genetic criteria (Northrup et al. 2013); their medical records were reviewed. Forty-six subjects with nonsyndromic ASD (thus without TSC) who had undergone 3T MRI as part of their clinical care were identified via chart review of the Boston Children’s Hospital clinic and neuroradiology records. Patients with prior brain surgery or exposure to mTOR inhibitors were excluded. In all patients, the ASD diagnoses were based on evaluation by a board-certified pediatric neurologist (in case of TSC by JMP or MS) or developmental medicine specialist, using the Diagnostic and Statistical Manual (DSM-IV-TR and DSM-V), supplemented with a formal neuropsychological evaluation including the Autism Diagnostic Observation Schedule (ADOS) in ASD (N = 10) and TSC (N = 9) (Lord et al. 2000).

Eighty-nine healthy controls (HC) had a normal neurological exam and a normal MRI on review by a pediatric neuroradiologist (S.P.P.), performed as part of routine clinical care (e.g., for soft indications like tension type headache, new onset simple tic disorder) or as part of this research project. Controls did not undergo formal neuropsychological evaluation as part of this study, but histories were negative for neurodevelopmental problems.

Medical Ethics Approval

Subject identification and data acquisition were conducted using a protocol approved by the institutional review board of Boston Children’s Hospital.

Clinical Data

Electronic medical records including available neuropsychological test results, school performance, and clinic notes were reviewed to gather information about neurodevelopment and seizure history. Due to the retrospective nature of the study, a continuous measure of cognition was not feasible. Intelligence was evaluated based on formal neuropsychological assessment or estimated clinically based on development and level of education. To prevent undersampled subgroups and subsequent underpowering of the study, intelligence was categorized as either ID (IQ < 70; severely impaired language or judgment; or ADL dependence) or no ID (IQ > 70; could have learning disability, mild language delays or minor limitations in adaptive function).

The epilepsy variable was categorized as present or absent, as more sophisticated classification schemes based on the E-Chess (Humphrey et al. 2008) resulted in undersampling of subgroups.

Image Acquisition

On a 3T MRI, the routine clinical protocol in our institution includes both structural imaging and diffusion-weighted imaging. Procedural sedation (most often with propofol) was used for clinical imaging studies only if necessary to prevent excessive motion. The imaging protocol included (1) a T1-weighted high-resolution magnetization-prepared rapid acquisition gradient-echo (MPRAGE) sequence, (2) a T2-weighted turbo spin-echo (TSE) sequence, (3) a sagittal 3D isotropic T2 fluid-attenuated inversion recovery (FLAIR), and (4) axial diffusion imaging using a single shell HARDI with 30 slices with b = 1000 s/mm2 and 5 b = 0 images. The intracranial cavity was segmented following the structural MRI. Diffusion images were aligned to the T1-weighted MPRAGE to compensate for distortion and patient motion. We estimated the tensor fit with robust least squares, as previously described (Peters et al. 2015).

Tractography

We used a stochastic algorithm for tractography, combining the speed and accuracy of deterministic decision-making at each voxel with probabilistic sampling to better explore the space of all possible streamlines (Peters et al. 2012). The corpus callosum tractography was created from automated generation of region of interest (ROI), based on a collection of manually delineated template brains and the STAPLE algorithm, as detailed prior (Suarez et al. 2012). After visual review for anatomical accuracy, streamline density-weighted corrected scalar measures of FA and MD were obtained to correct for partial voluming effects and spurious tracts (Peters et al. 2012).

Statistical Analysis

Statistical analysis was done using the R statistical software package (R_Core_Team,2015). Group differences in clinical variables (age, gender, epilepsy, ID) were calculated using chi-square and student t-tests. Adjustment for volumetric differences of the corpus callosum was done by incorporating age into the model, and by using density-weighted statistics in the calculation of mean values (Peters et al. 2012). Further correction for actual callosal volumes did not change the analysis in large subgroups and resulted in underpowering of smaller subgroups. Univariate analysis demonstrated gender not to be a confounder, and it was omitted from the model.

Each subject belonged to 1 of 3 main groups: TSC (regardless of presence of ASD), ASD (patients with nonsyndromic ASD), and HC (healthy controls). Further subgroups were made depending on the presence of comorbid conditions. Thus, for each subject, the final data was composed of 4 “categorical” random variables: one (1) group status variable (TSC, ASD, and HC), and 3 clinical variables (2) epilepsy, (3) ID and (4) ASD. Each subject also had 3 “continuous” random variables: (1) age, and DTI metrics (2) FA and (3) MD, at corresponding age.

Using available data, we modeled the developmental trajectory of FA and MD by age. The sample size of the subgroups when including all the clinical variables ranged from 0 (certain combinations do not exist, e.g., TSC with epilepsy and ID but no ASD) to 89 (healthy controls). Subgroups with a sample size lower than 5 were omitted as they could not be fitted reliably, but the sample size of the smallest subgroup was still only 8 patients. The statistical analysis was designed to accommodate fitting of such low sample size datasets as follows:

(1) We first selected the best performing 2-parameter growth model for fitting with non-negative nonlinear least squares. Our prior work informed the model choice given the assumed complexity of the developmental trajectory of the DTI metrics over time (Peters et al. 2012, 2015; Lewis et al. 2013). To test which of 2 known 2-parameter growth models (Michaelis-Menten and exponential model) had the best predictive ability of new data (i.e., data not previously used for its modeling), we carried out both leave-1-out and leave-2-out cross-validation schemes. The exponential model was superior, and is parametrized by its asymptote A and its half-life H (the time required to reach half the asymptote).

(2) Next, we applied 3 levels of data fitting to assess consistency of findings in incrementally smaller samples. The first level involved only the 3 main subgroups TSC, ASD and HC, with large sample sizes. At the second level, subgroups were made based on the 3 clinical variables ID, epilepsy and ASD. At the third level, all sufficiently large possible subgroup combinations were fitted.

Subgroup differences of the estimated means of the asymptote and half-life of the fitted DTI metrics were assessed for consistency using a split-half analysis. As estimators of the coefficients of a nonlinear regression are typically not normally distributed (even when residuals are), we applied a bootstrap strategy to estimate the distribution of the estimator of the DTI metrics curve per subgroup. Using 10 000 bootstrap replicates of curve estimates, we calculated the mean of the distribution, and the confidence intervals for any given significance level, using the adjusted bootstrap percentile method (DiCiccio and Efron 1996). An example of the fitting procedure is provided in Figure 1.

Figure 1.

Figure 1.

Open in a new tab

Illustration of the model fitting procedure. (A) The original sample is shown, and the exponential model is fitted on the bootstrap (a subset of the original data, top middle) using a nonlinear least squares technique. A single fit is obtained. (B) This bootstrapping procedure is repeated 10 000 times, to obtain the mean and median distributions, with confidence intervals. Derived from these fitted curves, the half-life H (time needed to reach half the asymptote value) and the asymptote A are shown, with confidence intervals.

Results

Demographic data and clinical variables are presented in Table 1. HC were older than patients with TSC alone, TSC and ASD, and nonsyndromic ASD. There were more males in the nonsyndromic ASD group compared with HC, but no group differences in gender distribution otherwise.

Table 1.

Demographic data and clinical variables

TSC alone (n=32) TSC & ASD (n=19) ASD (n=46) HC (n=89)
TSC alone vs. TSC & ASD ASD vs. TSC alone/TSC & ASD HC vs. TSC alone/TSC & ASD/ASD
Age (range) 8.2 (1.0-21.0) ns 10.3 (1.1-27.2) ns/ns 8.8 (2.7-19.9) ***/*/ *** 13.3 (1.1-25.3)
Gender M (%) 20 (62.5) ns 11 (57.9) ns/ns 33 (71.7) ns/ns/ns 44 (49.4)
ID (%) 4 (12.5) *** 14 (73.7) */ns 20 (43.5) n/a —
Epilepsy (%) 18 (56.2) * 18 (94.7) ns/*** 18 (29.1) n/a —
Open in a new tab

ASD, autism spectrum disorder; F, Female; M, Male; TSC, tuberous sclerosis complex.

*P < 0.05, **P < 0.005, ***P < 0.0005.

By selection, all HC had normal intelligence and none had epilepsy. Patients with TSC and ASD had more frequent ID compared to those with TSC alone and to those with nonsyndromic ASD. Prevalence of epilepsy was also higher in patients with both TSC and ASD compared to TSC alone and to nonsyndromic ASD. Patients without epilepsy were equally common in the TSC alone and nonsyndromic ASD groups.

FA values of callosal white matter are shown for each of the subgroups in Figures 2–4. MD metrics were complimentary, and are summarized in corresponding supplementary Figures e1-3. FA mean differences and confidence intervals of the half-life and asymptote of the fitted curves are shown in supplementary Figures e4-6.

Figure 2.

Figure 2.

Open in a new tab

Predicted FA values with age for the 3 main populations. Predicted FA values of the model fit are displayed for age, for each of the 3 main groups: Healthy controls (HC, in red), TSC (in green), and nonsyndromic autism spectrum disorder (ASD, in blue). The developmental trajectory of white matter maturation has been described previously (Peters et al. 2012, 2015), and was used to inform the model.

Figure 4.

Figure 4.

Open in a new tab

Predicted FA values with age for all possible subgroups. Predicted FA values for the model fit for subgroups with ≥8 subjects are displayed. The healthy control group (HC, red) is not split up into subgroups. The TSC subgroup is split into 3 groups. TSC patients either had both ID and ASD or neither. The group without ID and ASD could be further divided into those with and without epilepsy. There were not enough TSC subjects with ID and ASD without epilepsy to create a separate subgroup, so subjects in this subgroup had all 3 conditions. The ASD group was split into 4 subgroups: without ID (with and without epilepsy) and with ID (with and without epilepsy). Consistent and cumulative changes in the modeled FA trajectories are shown from the co-occurrence of TSC, ID, epilepsy and ASD. (Note that the age range is 3 years and up as patients in subgroups too small for stable fitting were omitted).

Step-wise, 3 levels of analysis were done as follows:

(1) At the first level, the FA for age was predicted for the 3 main groups (HC, TSC, and ASD) (Fig. 2). Not adjusted for neurodevelopmental and epilepsy variables, patients with TSC had lower FA (higher MD) compared with those with ASD, and compared to the HC group. The ASD group also had a lower FA (higher MD) compared with the HC group.

(2) At the second level, 3 clinical variables (epilepsy, ID, and ASD) were used to divide the main groups into subgroups (Fig. 3). For the presence of epilepsy: TSC with and without epilepsy, ASD with and without epilepsy (Fig. 3A). For the presence of ID: TSC with and without ID, ASD with and without ID (Fig. 3B). For the presence of ASD only 2 subgroups could be formed: TSC with and without ASD (Fig. 3C).

Figure 3.

Figure 3.

Open in a new tab

Predicted FA values with age, corrected for cognition, epilepsy, and ASD. Predicted FA values of the model fit of the subgroups of TSC (green) and ASD (blue) are shown. For comparison, the HC group is shown in red. In (A), the changes of the model fit related to epilepsy are shown in the TSC population and in the ASD population. TSC with epilepsy has lower FA values than TSC without epilepsy. In (B), a decrease of the FA values is seen related to the presence of ID in both TSC and in ASD groups. In (C) the model shows TSC with ASD is associated with decreased FA values compared to TSC without ASD.

For epilepsy, the predicted fit for FA is shown in Figure 3A. TSC patients with recurrent seizures had lower callosal FA values than TSC patients without epilepsy. Again, this same finding was present in the ASD subgroup with and without epilepsy. The ASD group without epilepsy had higher FA values than healthy controls.

For ID, the predicted fit for FA is shown in Figure 3B. The TSC with ID subgroup had lower FA values than the TSC without ID subgroup. Similarly, the ASD with ID subgroup had lower FA values than the ASD without ID subgroup.

For ASD, the predicted fit for FA is shown in Figure 3C. Subjects with both TSC and ASD had significantly lower FA values than TSC alone, and ASD alone.

(3) At the third level, the possible iterations of the main groups (HC, TSC, ASD) with all 3 clinical variables (ID, epilepsy and ASD) were fitted (Fig. 4). Small subgroups with 5 subjects or less (e.g., patients with TSC, no ID, no epilepsy but with ASD) were omitted as the model would not be stable.

Seven subgroups (3 for TSC and 4 for ASD) were large enough for a stable model fit of FA prediction (Fig. 4). TSC patients without ID could be split into subgroups with and without epilepsy; the presence of epilepsy was again associated with lower FA values. A subgroup of TSC patients with all 3 neurological comorbidities (ASD, ID, and epilepsy) demonstrated the lowest white matter FA values.

For ASD, patients without ID were split into those with and those without epilepsy. The ASD patients with ID were also split into those with and those without epilepsy. In all of these ASD subgroups, ID and epilepsy were each associated with a decreased callosal FA. The subgroup with TSC and all 3 neurological conditions (ID, epilepsy and ASD) had the lowest FA values.

Leave-one-out cross-validation (LOOCV) was applied to groups of sufficient size for a stable model fit, and revealed a pattern that was consistent with above findings using the full data-set (Appendix 1).

Discussion

This study demonstrates that both TSC and ASD are associated with reduced callosal white matter microstructural integrity. We find that DTI metrics of the corpus callosum are cumulatively associated with comorbid neurological symptoms in TSC and in nonsyndromic ASD. These findings reflect the complex relationship between biological disease burden, and subsequent ID, ASD and epilepsy in TSC (van Eeghen et al. 2013).

In our sample set, the presence of ID was associated with lower FA values of the white matter in both the TSC and ASD subgroups. In a previous DTI study of 20 patients with TSC, aberrant network connectivity was associated with ID (Im et al. 2015), but in that study and in another, an association with ASD was not found (Baumer et al. 2015). Another study in a nonsyndromic ASD population reported a similar inverse association between IQ and FA values of the corpus callosum (Alexander et al. 2007). Our study suggests that ID is independently associated with decreased microstructural integrity across different neurologic conditions.

Epilepsy was also associated with decreased microstructural integrity of callosal white matter in both the TSC and ASD subgroups. In TSC, the association of recurrent seizures with altered diffusion of white matter tracts has been reported in a cross-sectional study of 20 patients (Moavero et al. 2016) and in a longitudinal study of 17 children (Baumer et al. 2015), but these studies did not account for all comorbid conditions. Seizures can arise from areas with relatively high diffusivity, and can in turn contribute to both local and remote diffusion via inflammation and gliosis (Jansen et al. 2003; Otte et al. 2012; Yogi et al. 2015). The decreased callosal FA associated with epilepsy seen in our study population can therefore reflect both a higher disease burden and superimposed effects from refractory seizures.

In this study, most imaging was obtained in an elective setting and not for medical emergencies. Thus, while all patients labeled as “epilepsy” had an active pharmacoresistent seizure disorder, immediate but transient effects of prolonged seizures on callosal white matter diffusion are likely negligible here.

In the TSC population, we previously reported that reduced white matter microstructural integrity in the corpus callosum correlates with a diagnosis of ASD (Peters et al. 2012) and may be a potential imaging biomarker for this condition. Our current study, with a larger sample size, is in line with this finding. As a group, children with TSC have lower callosal FA values than controls (Fig. 2); the subgroup with ASD drives this difference (Fig. 3). TSC patients without ASD have a lower FA than controls, but this difference is smaller. In our third level analysis (Fig. 4), in our large sample we were still unable to isolate a sufficiently large group with TSC and ASD but without comorbid ID and epilepsy. This confirms previous reports of a high degree of ID and ASD in TSC (Jeste et al. 2016). Other white matter pathways with more relevance to ASD pathophysiology may be more specific and will be investigated in future. In nonsyndromic ASD, we also found aberrant callosal DTI metrics, in agreement with previous studies (Alexander et al. 2007; Cheng et al. 2010; Travers et al. 2015). The differences between nonsyndromic ASD subgroups and control subjects were much smaller, however, as compared to corresponding TSC subgroups.

The exact mechanisms underlying corpus callosum deficits in TSC remain to be identified; however, we can speculate about a number of possibilities that play a role in the pathogenesis. First, mutations in TSC1 and TSC2 are known to affect neuronal migration. It is possible that defects in migration lead to aberrant lamination such that callosal neurons may not be properly located in the superficial layers of the neocortex. The radial migration lines and cortical tubers with dyslamination in both TSC patient brains (Feliciano et al. 2013; Lim and Crino 2013) as well as in animal models of TSC (Uhlmann et al. 2002; Meikle et al. 2007) display migration abnormalities consistent with such a possibility. Second, development of the corpus callosum involves production of pioneer neurons that send out contralateral axons. These pioneer neurons and their axon migration has not specifically been studied in TSC. It is however clear that axon guidance molecules, such as ephrins, lead to aberrant signaling in TSC-deficient neurons, resulting in axon guidance defects (Nie et al. 2010); this may affect migration of callosal pioneer neurons. Third, the initial phase of axonal overproduction is followed by a period of axon elimination in the early postnatal period in primates (LaMantia and Rakic 1990). It is not clear what the effect of TSC-loss is on axon overproduction and elimination. Fourth, myelination can be affected by TSC-loss in either neurons or in oligodendrocytes and result in hypomyelination in the white matter, including in the corpus callosum (Meikle et al. 2007; Carson et al. 2015). Finally, increased heterotopic cells, and small satellite lesions referred to as “microtubers” may all contribute to alterations in the corpus callosum and normal appearing white matter (Marcotte et al. 2012; Ruppe et al. 2014; Peters et al. 2015). The neuropathological substrate of diffusion abnormality in TSC may be investigated through ex vivo imaging of resection specimens obtained at the time of epilepsy surgery as well as neuropathology of postmortem tissues.

In TSC, patients with ASD had significantly different callosal FA values compared to those without ASD, supporting its potential use as a predictive biomarker if confirmed in prospective studies in presymptomatic patients. Confounding effects, however, from ID and epilepsy could not be completely removed. Thus, the DTI changes in this study are neither sufficiently specific nor large enough early on, limiting the predictive value. Currently, a large multi-center study is collecting the clinical phenotype and longitudinal imaging from a young age (clinicaltrials.gov NCT01780441); such prospective data is needed to use DTI in the prediction of specific neurological comorbidities.

The scope of this study was limited to the examination of a single major white matter pathway, and other pathways are under investigation. While the corpus callosum structure has a role in the pathophysiology of ASD (Travers et al. 2012), it does not directly subserve the language or neurobehavioral functions affected in ASD. Examination of tracts involved in ASD-specific deficits in language, stimulus reward systems, and social cognition may yield more specific markers (Fletcher et al. 2010; Abrams et al. 2013; Von Der Heide et al. 2013). Finally, as DTI techniques advance, novel acquisition schemes may enhance the ability for group-wise comparisons of white matter tracts and their role in ASD pathophysiology (Scherrer and Warfield 2012; Taquet et al. 2014).

Though our fitting procedure to obtain the half-life and asymptote for all possible subgroups was sophisticated, it was applied to complex, sparsely sampled data with many potential confounders. As TSC is a rare disease, a large enough sample for conventional null-hypothesis significance probability testing of all variables (age, group status, epilepsy, ID, ASD, FA/MD) was not possible in this imaging study. Rather, we applied a nonlinear regression analysis and demonstrate group differences between incrementally smaller subgroups on the basis of “confidence intervals” of the 2 main coefficients that describe the fitted curves. In addition, the “consistent direction” of the change in FA (MD) with each clinical variable (ID, epilepsy, ASD) across the different subgroups adds validity to our findings.

Other potential variables that could affect callosal white matter diffusion metrics include anti-epileptic and psychoactive medications, and quantity and location of structural lesions (e.g., radial migration lines, tubers). Inclusion of all such variables in the study was not feasible as it would further diminish power through small subgroup sizes.

It is unclear why the 18 patients with ASD without epilepsy have higher FA values than the HC group; we suspect this is a sampling bias or resulting from fitting sparse data. Similarly, at a young age, TSC with epilepsy had higher FA values, likely reflecting more TSC subjects with epilepsy than without, as seizure onset occurs in 80% before age 3 (Curatolo et al. 2015).

Conclusion

Using a cross-disorder approach, this study demonstrates cumulative abnormality of callosal white matter diffusion metrics with increasing neurological comorbidity in both TSC and ASD. In TSC, the lack of specificity of callosal DTI metrics limits its current use to a marker of overall neurological outcome. Other white matter pathways should be studied, and longitudinal, prospective data are needed for validation and increased specificity.

Supplementary Material

Supplementary Data
Click here for additional data file. (1.1MB, zip)

Footnotes

We are indebted to our patients and heathy controls for participation, and to Boston Children’s Hospital MRI technical staff for their diligent assistance with data acquisition. Conflict of Interest: None declared.

Funding

F.M.B. is supported by a KL2 Mentored Career Development Award of the Stanford Clinical and Translational Science Award to Spectrum (NIH KL2 TR 001083) and (UL1 TR 001085). J.M.P., S.C., B.S., M.S., and S.W. are supported by NIH R01 NS079788 and U01 NS082320 grants. A. Prohl is supported by Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Research Resources and the National Center for Advancing Translational Sciences, NIH award UL1 TR001102). S. Prabhu is supported by the Department of Defense W81XWH-11-1-0365 and NIH U01 NS082320 grants. F.E.J. is supported by the Framework Program FP7/2007-2013 under the project acronym EPISTOP (grant agreement no. 602 391). K.P.J.B. reports no disclosures relevant to the manuscript. M.S. is additionally supported by an NIH U54 HD090255 grant and the Boston Children’s Hospital Translational Research Program. The Developmental Synaptopathies Consortium (U54 NS092090) is part of the NCATS Rare Diseases Clinical Research Network (RDCRN). RDCRN is an initiative of the Office of Rare Diseases Research (ORDR), NCATS, funded through collaboration between NCATS, NIMH, NINDS, and NICHD. A.S. is supported by an NIH R01 EB013248 grant.

References

  1. Abrams DA, Lynch CJ, Cheng KM, Phillips J, Supekar K, Ryali S, Uddin LQ, Menon V. 2013. Underconnectivity between voice-selective cortex and reward circuitry in children with autism. Proc Natl Acad Sci U S A. 110:12060–12065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Alexander AL, Lee JE, Lazar M, Boudos R, DuBray MB, Oakes TR, Miller JN, Lu J, Jeong EK, McMahon WM, et al. 2007. Diffusion tensor imaging of the corpus callosum in Autism. Neuroimage. 34:61–73. [DOI] [PubMed] [Google Scholar]
  3. Baumer FM, Song JW, Mitchell PD, Pienaar R, Sahin M, Grant PE, Takahashi E. 2015. Longitudinal changes in diffusion properties in white matter pathways of children with tuberous sclerosis complex. Pediatr Neurol. 52:615–623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bombardieri R, Pinci M, Moavero R, Cerminara C, Curatolo P. 2010. Early control of seizures improves long-term outcome in children with tuberous sclerosis complex. Eur J Paediatr Neurol. 14:146–149. [DOI] [PubMed] [Google Scholar]
  5. Bruining H, Eijkemans MJ, Kas MJ, Curran SR, Vorstman JA, Bolton PF. 2014. Behavioral signatures related to genetic disorders in autism. Mol Autism. 5:11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Carson RP, Kelm ND, West KL, Does MD, Fu C, Weaver G, McBrier E, Parker B, Grier MD, Ess KC. 2015. Hypomyelination following deletion of Tsc2 in oligodendrocyte precursors. Ann Clin Transl Neurol. 2:1041–1054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cheng Y, Chou KH, Chen IY, Fan YT, Decety J, Lin CP. 2010. Atypical development of white matter microstructure in adolescents with autism spectrum disorders. Neuroimage. 50:873–882. [DOI] [PubMed] [Google Scholar]
  8. Curatolo P, Bombardieri R, Jozwiak S. 2008. Tuberous sclerosis. Lancet. 372:657–668. [DOI] [PubMed] [Google Scholar]
  9. Curatolo P, Moavero R, de Vries PJ. 2015. Neurological and neuropsychiatric aspects of tuberous sclerosis complex. Lancet Neurol. 14:733–745. [DOI] [PubMed] [Google Scholar]
  10. Davis PE, Peters JM, Krueger DA, Sahin M. 2015. Tuberous sclerosis: a new frontier in targeted treatment of autism. Neurotherapeutics. 12:572–583. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. DiCiccio TJ, Efron B. 1996. Bootstrap confidence intervals. Statist Sci. 11:189–228. [Google Scholar]
  12. Feliciano DM, Lin TV, Hartman NW, Bartley CM, Kubera C, Hsieh L, Lafourcade C, O’Keefe RA, Bordey A. 2013. A circuitry and biochemical basis for tuberous sclerosis symptoms: from epilepsy to neurocognitive deficits. Int J Dev Neurosci. 31:667–678. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Fletcher PT, Whitaker RT, Tao R, DuBray MB, Froehlich A, Ravichandran C, Alexander AL, Bigler ED, Lange N, Lainhart JE. 2010. Microstructural connectivity of the arcuate fasciculus in adolescents with high-functioning autism. Neuroimage. 51:1117–1125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Humphrey A, Ploubidis GB, Yates JR, Steinberg T, Bolton PF. 2008. The early childhood epilepsy severity scale (E-Chess). Epilepsy Res. 79:139–145. [DOI] [PubMed] [Google Scholar]
  15. Im K, Ahtam B, Haehn D, Peters JM, Warfield SK, Sahin M, Ellen Grant P. 2015. Altered structural brain networks in tuberous sclerosis complex. Cereb Cortex. 26:2046–2058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Jansen FE, Braun KP, van Nieuwenhuizen O, Huiskamp G, Vincken KL, van Huffelen AC, van der Grond J. 2003. Diffusion-weighted magnetic resonance imaging and identification of the epileptogenic tuber in patients with tuberous sclerosis. Arch Neurol. 60:1580–1584. [DOI] [PubMed] [Google Scholar]
  17. Jansen FE, Vincken KL, Algra A, Anbeek P, Braams O, Nellist M, Zonnenberg BA, Jennekens-Schinkel A, van den Ouweland A, Halley D, et al. 2008. Cognitive impairment in tuberous sclerosis complex is a multifactorial condition. Neurology. 70:916–923. [DOI] [PubMed] [Google Scholar]
  18. Jeste SS, Varcin KJ, Hellemann GS, Gulsrud AC, Bhatt R, Kasari C, Wu JY, Sahin M, Nelson CA 3rd. 2016. Symptom profiles of autism spectrum disorder in tuberous sclerosis complex. Neurology. 87:766–772. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Jeste SS, Wu JY, Senturk D, Varcin K, Ko J, McCarthy B, Shimizu C, Dies K, Vogel-Farley V, Sahin M, et al. 2014. Early developmental trajectories associated with ASD in infants with tuberous sclerosis complex. Neurology. 83:160–168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jozwiak S, Kotulska K, Domanska-Pakiela D, Lojszczyk B, Syczewska M, Chmielewski D, Dunin-Wasowicz D, Kmiec T, Szymkiewicz-Dangel J, Kornacka M, et al. 2011. Antiepileptic treatment before the onset of seizures reduces epilepsy severity and risk of mental retardation in infants with tuberous sclerosis complex. Eur J Paediatr Neurol. 15:424–431. [DOI] [PubMed] [Google Scholar]
  21. Julich K, Sahin M. 2014. Mechanism-based treatment in tuberous sclerosis complex. Pediatr Neurol. 50:290–296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. LaMantia AS, Rakic P. 1990. Axon overproduction and elimination in the corpus callosum of the developing rhesus monkey. J Neurosci. 10:2156–2175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lewis WW, Sahin M, Scherrer B, Peters JM, Suarez RO, Vogel-Farley VK, Jeste SS, Gregas MC, Prabhu SP, Nelson CA 3rd, et al. 2013. Impaired language pathways in tuberous sclerosis complex patients with autism spectrum disorders. Cereb Cortex. 23:1526–1532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lim KC, Crino PB. 2013. Focal malformations of cortical development: new vistas for molecular pathogenesis. Neuroscience. 252:262–276. [DOI] [PubMed] [Google Scholar]
  25. Lord C, Risi S, Lambrecht L, Cook EH Jr., Leventhal BL, DiLavore PC, Pickles A, Rutter M. 2000. The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. J Autism Dev Disord. 30:205–223. [PubMed] [Google Scholar]
  26. Marcotte L, Aronica E, Baybis M, Crino PB. 2012. Cytoarchitectural alterations are widespread in cerebral cortex in tuberous sclerosis complex. Acta Neuropathol. 123:685–693. [DOI] [PubMed] [Google Scholar]
  27. Meikle L, Talos DM, Onda H, Pollizzi K, Rotenberg A, Sahin M, Jensen FE, Kwiatkowski DJ. 2007. A mouse model of tuberous sclerosis: neuronal loss of Tsc1 causes dysplastic and ectopic neurons, reduced myelination, seizure activity, and limited survival. J Neurosci. 27:5546–5558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Moavero R, Napolitano A, Cusmai R, Vigevano F, Figa-Talamanca L, Calbi G, Curatolo P, Bernardi B. 2016. White matter disruption is associated with persistent seizures in tuberous sclerosis complex. Epilepsy Behav. 60:63–67. [DOI] [PubMed] [Google Scholar]
  29. Nie D, Di Nardo A, Han JM, Baharanyi H, Kramvis I, Huynh T, Dabora S, Codeluppi S, Pandolfi PP, Pasquale EB, et al. 2010. Tsc2-Rheb signaling regulates EphA-mediated axon guidance. Nat Neurosci. 13:163–172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Northrup H, Krueger DA, Group ITSCC . 2013. Tuberous sclerosis complex diagnostic criteria update: recommendations of the 2012 Iinternational Tuberous Sclerosis Complex Consensus Conference. Pediatr Neurol. 49:243–254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Numis AL, Major P, Montenegro MA, Muzykewicz DA, Pulsifer MB, Thiele EA. 2011. Identification of risk factors for autism spectrum disorders in tuberous sclerosis complex. Neurology. 76:981–987. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Otte WM, van Eijsden P, Sander JW, Duncan JS, Dijkhuizen RM, Braun KP. 2012. A meta-analysis of white matter changes in temporal lobe epilepsy as studied with diffusion tensor imaging. Epilepsia. 53:659–667. [DOI] [PubMed] [Google Scholar]
  33. Peters JM, Prohl AK, Tomas-Fernandez XK, Taquet M, Scherrer B, Prabhu SP, Lidov HG, Singh JM, Jansen FE, Braun KP, et al. 2015. Tubers are neither static nor discrete: Evidence from serial diffusion tensor imaging. Neurology. 85:1536–1545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Peters JM, Sahin M, Vogel-Farley VK, Jeste SS, Nelson CA 3rd, Gregas MC, Prabhu SP, Scherrer B, Warfield SK. 2012. Loss of white matter microstructural integrity is associated with adverse neurological outcome in tuberous sclerosis complex. Acad Radiol. 19:17–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. R_Core_Team 2015. R: A language and environment for statistical computing. Vienna, Australia: R Foundation for Statistical Computing. [Google Scholar]
  36. Ruppe V, Dilsiz P, Reiss CS, Carlson C, Devinsky O, Zagzag D, Weiner HL, Talos DM. 2014. Developmental brain abnormalities in tuberous sclerosis complex: a comparative tissue analysis of cortical tubers and perituberal cortex. Epilepsia. 55:539–550. [DOI] [PubMed] [Google Scholar]
  37. Scherrer B, Warfield SK. 2012. Parametric representation of multiple white matter fascicles from cube and sphere diffusion MRI. PLoS ONE. 7:e48232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Suarez RO, Commowick O, Prabhu SP, Warfield SK. 2012. Automated delineation of white matter fiber tracts with a multiple region-of-interest approach. Neuroimage. 59:3690–3700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Taquet M, Scherrer B, Commowick O, Peters JM, Sahin M, Macq B, Warfield SK. 2014. A mathematical framework for the registration and analysis of multi-fascicle models for population studies of the brain microstructure. IEEE Trans Med Imaging. 33:504–517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Tillema JM, Leach JL, Krueger DA, Franz DN. 2012. Everolimus alters white matter diffusion in tuberous sclerosis complex. Neurology. 78:526–531. [DOI] [PubMed] [Google Scholar]
  41. Travers BG, Adluru N, Ennis C, Tromp DP, Destiche D, Doran S, Bigler ED, Lange N, Lainhart JE, Alexander AL. 2012. Diffusion tensor imaging in autism spectrum disorder: a review. Autism Res. 5:289–313. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Travers BG, Tromp do PM, Adluru N, Lange N, Destiche D, Ennis C, Nielsen JA, Froehlich AL, Prigge MB, Fletcher PT, et al. 2015. Atypical development of white matter microstructure of the corpus callosum in males with autism: a longitudinal investigation. Mol Autism. 6:15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Uhlmann EJ, Wong M, Baldwin RL, Bajenaru ML, Onda H, Kwiatkowski DJ, Yamada K, Gutmann DH. 2002. Astrocyte-specific TSC1 conditional knockout mice exhibit abnormal neuronal organization and seizures. Ann Neurol. 52:285–296. [DOI] [PubMed] [Google Scholar]
  44. van Eeghen AM, Pulsifer MB, Merker VL, Neumeyer AM, van Eeghen EE, Thibert RL, Cole AJ, Leigh FA, Plotkin SR, Thiele EA. 2013. Understanding relationships between autism, intelligence, and epilepsy: a cross-disorder approach. Dev Med Child Neurol. 55:146–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Von Der Heide RJ, Skipper LM, Klobusicky E, Olson IR. 2013. Dissecting the uncinate fasciculus: disorders, controversies and a hypothesis. Brain. 136:1692–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Yogi A, Hirata Y, Karavaeva E, Harris RJ, Wu JY, Yudovin SL, Linetsky M, Mathern GW, Ellingson BM, Salamon N. 2015. DTI of tuber and perituberal tissue can predict epileptogenicity in tuberous sclerosis complex. Neurology. 85:2011–2015. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
Click here for additional data file. (1.1MB, zip)

Articles from Cerebral Cortex (New York, NY) are provided here courtesy of Oxford University Press

RESOURCES