Potential for treatment of severe autism in tuberous sclerosis complex

Abstract

The Food and Drug Administration (FDA) has approved two mechanism-based treatments for tuberous sclerosis complex (TSC)-everolimus and vigabatrin. However, these treatments have not been systematically studied in individuals with TSC and severe autism. The aim of this review is to identify the clinical features of severe autism in TSC, applicable preclinical models, and potential barriers that may warrant strategic planning in the design phase of clinical trial development. A comprehensive search strategy was formed and searched across PubMed, Embase and SCOPUS from their inception to 2/21/12, 3/16/12, and 3/12/12 respectively. After the final search date, relevant, updated articles were selected from PubMed abstracts generated electronically and emailed daily from PubMed. The references of selected articles were searched, and relevant articles were selected. A search of clinicaltrials.gov was completed using the search term “TSC” and “tuberous sclerosis complex”. Autism has been reported in as many as 60% of individuals with TSC; however, review of the literature revealed few data to support clear classification of the severity of autism in TSC. Variability was identified in the diagnostic approach, assessment of cognition, and functional outcome among the reviewed studies and case reports. Objective outcome measures were not used in many early studies; however, diffusion tensor imaging of white matter, neurophysiologic variability in infantile spasms, and cortical tuber subcategories were examined in recent studies and may be useful for objective classification of TSC in future studies. Mechanism-based treatments for TSC are currently available. However, this literature review revealed two potential barriers to successful design and implementation of clinical trials in individuals with severe autism-an unclear definition of the population and lack of validated outcome measures. Recent studies of objective outcome measures in TSC and further study of applicable preclinical models present an opportunity to overcome these barriers.

Core tip: Children with severe behaviors and cognitive impairment may benefit from newly available mechanism-based treatments; however, several factors warrant consideration in clinical trial design and implementation.

INTRODUCTION

Tuberous sclerosis complex (TSC) is a leading cause of syndromic autism characterized by multi-system hamartomas. Clinically, the diagnosis is classified as definite, probable, or possible TSC based on the presence of a specific combination of major and minor features (Table 1)[1] or genetically by mutations in TSC1 or TSC2. Although autism has been reported in as many as 61% of individuals with TSC[2,3], severity is not frequently reported. Existing neurologic comorbidities include epilepsy in 60%-90%[4-6], intellectual disability in 45%[7], self-injury in 10%[8], and severe aggression in 13%[9]. The contribution of these neurologic comorbidities or the features of TSC to the severity of autism in this population is unknown.

Table 1.

Criteria for clinical diagnosis of tuberous sclerosis complex

Major features	Minor features
Cortical tubers	Dental enamel pits
Subependymal nodules	Hamartomatous rectal polyps
Subependymal giant cell astrocytoma	Bone cysts
Hypomelanotic macules (3 or more)	Cerebral white matter radial migration lines
Shagreen patch	Gingival fibromas
Facial angiofibromas or forehead plaque	Nonrenal hamartoma
Multiple renal nodular hamartomas	Retinal achromatic patches
Nontraumatic ungual or periungual fibromas	“Confetti” skin lesions
Cardiac rhabdomyoma, 1 or >	Multiple renal cysts
Pulmonary Lymphangiomyomatosis and/or renal angiomyolipomas

Tuberous sclerosis complex (TSC) can be clinically diagnosed as definite, probable, or possible based on the presence or absence of a specific combination of major and minor features. Definite TSC: two major features or one major and two minor features; Probable TSC: one major and one minor feature; Possible TSC: one major feature or two or more minor features.

The neurobiology underlying this condition has been established. TSC1 and TSC2 encode hamartin and tuberin respectively, which indirectly inhibit mammalian target of rapamycin (mTOR). This enzyme is an essential component of two complexes, mTORC1 and mTORC2, which have distinct, wide-ranging effects on gene transcription, protein translation and cell proliferation. Excessive mTOR activity results from a mutation in either TSC1 or TSC2. Disrupted synaptic plasticity, characterized by excessive glutamate activity, may occur through downstream effects of this excessive mTOR activity on ribosomal s6 and EIF4E. Targeted drug development based on this neurobiology has resulted in FDA approval of the mTOR inhibitor, everolimus, and the irreversible inhibitor of GABA transaminase, vigabatrin, for individuals with TSC. While everolimus targets the underlying mechanism of TSC and may have wide-ranging effects, vigabatrin, in addition to anti-seizure effects, may also have an impact on glutamatergic mechanisms important in brain development, synaptic plasticity and learning.

TSC is considered a cause of syndromic autism; however, causality and determinants of severity are unknown. Additionally, the risk/benefit profile of approved treatments in this population has not been tested in a clinical trial. The aim of this review is to identify the clinical features of severe autism in TSC, applicable preclinical models, and potential barriers that may warrant strategic planning in the design phase of clinical trial development.

In order to identify key research, a comprehensive search strategy was formed and then searched across PubMed, Embase and SCOPUS from their inception to 2/21/12, 3/16/12, and 3/12/12 respectively. Database-specific controlled vocabulary terms were combined with keyword terms and phrases for each concept. These terms and phrases were then combined and translated for use in each database. The results were limited to studies conducted on humans and to those written in English. References from key papers were also reviewed and key studies, including preclinical studies were included.

AUTISM

Autism is a syndrome clinically defined by the presence of stereotyped, repetitive behavior and impairments in language and social interaction with onset prior to the age of 3[10]. TSC, the first identified cause of autism, is now considered a leading cause of syndromic autism[2,3,11,12]. The prevalence of autism in TSC is 26%-61% with an average prevalence of 32%[2,3,9,13-23]. Variability in operational definitions of autism and study design, in particular the approach to individuals with intellectual disability, may contribute to variation in these estimates. We did not discover any studies of the severity of autism in TSC. However, self-injury was observed in 10% of a large clinical population. Self-injury, one of the main parental concerns in TSC, may signal a severe form of autism in this subgroup[24].

Operational definitions of autism in TSC

The Diagnostic Statistical Manual and the International Classification of Disease have provided checklist criteria for defining autism. The study with the highest reported prevalence of autism and 3 additional studies used these checklists[3,10,19-21,25]. Other instruments have been developed to assist professionals in identifying individuals with autism, and combined use of the Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview (ADI, ADI-R) has been widely accepted as the gold standard. However, the original sample used in developing these instruments did not include individuals with profound intellectual disability[26,27]. The Social Responsiveness Scale (SRS) and the Social Communication Questionnaire (SCQ) are screening instruments for autism; however, the presence of intellectual disability decreases the specificity of SCQ from 80% to 67%[28,29]. The Hunt and Dennis Questionnaire is a screening instrument designed for use in individuals with TSC, behavioral difficulties, and profound intellectual disability. Autism is evaluated by 13 out of 321 questions; however, psychometric properties have not been reported to date[9].

Prevalence study designs

The diagnostic approach used in each of the 19 reviewed studies is indicated in Table 2. Multiple sequenced evaluations were used in 3 studies. The highest reported prevalence of autism in TSC was determined using 3 structured interviews, Childhood Autism Rating Scale (CARS), Autistic Behavior Checklist, and review of DSM III-R criteria[3]. In addition to variation in diagnostic approach, inclusion of children with severe or profound intellectual disability varied among these studies and may contribute to variability in prevalence estimates.

Table 2.

Prevalence studies for autism in tuberous sclerosis complex

Instrument	Administration method(time)	Children with severe or profound ID included (Y/N)	Measured prevalence of autism in TSC (N)	Ref.
ADOS	Observation schedule (20-30 min)	N	29% (28) +ADI	[18]
			15-33% (4 age-based groups of 12-15)
ADI	Structured Interview (2-3 h)	Y-severe	54% (13)	[15,16]
			20% (20)
SRS	65-item Screening Questionnaire (15-20 min)	N (37 w/IQ data)	52% (21)	[22]
SCQ	40-item Screening Questionnaire (15-20 min)	Y-severe	43% (21)	[22]
Hunt and Dennis Questionnaire	321 item interview/13-item subset for autism (1-4 h)	Y-estimates only	50% (90)	[2,9,14,23]
			26% (23)
			24% (21)
			5% (131)
DSM III-R	Checklist (10-20 min)	Y-severe	61% (28)	[3]

These are reviewed studies reporting epidemiology of autism in tuberous sclerosis complex (TSC). ADI: Autism diagnostic interview; DSM: Diagnostic and statistical manual of mental disorders; ADOS: Autism diagnostic observation schedule; SRS: Social responsiveness scale; SCQ: Social communication questionnaire; Y: Yes; N: No.

Risk factors

Potential risk factors and clinical associations of autism in TSC include epilepsy, structural or neurophysiological abnormalities, and/or joint effects of these categories, but causality has not been established. Intellectual disability and autism were not uniformly characterized among the studies, making it difficult to compare studies focused on risk factors associated with autism or ASD to those focused on risk factors associated with cognitive impairment. Among studies using standard measures to assess cognition, the choice of measure was often based on the assessed functional level or age of the individual. However, this approach was not routinely applied to assessments for autism or ASD. Therefore, we included studies of risk factors for autism and/or intellectual disability in TSC.

Epilepsy

Epilepsy affects 60%-90% of individuals with TSC[2,6,30-38]. Age of epilepsy onset, seizure type and/or severity have been evaluated as potential risk factors for autism and/or intellectual disability in children with TSC. Early age of seizure onset[2,33,34,36-40] and infantile spasms were most commonly identified as risk factors[2,14,17,34,34-42,41-46].

Epilepsy intractability and/or severity was characterized using measures such as development of multiple seizure types after infantile spasms, time to epilepsy control, duration of seizure control, and/or EEG features and was reported as a potential risk factor in 6 of 8 studies that characterized severity[14,37,38,41,46,47].

Neuropathology

Early studies focused solely on lesion location[34,39,40,43,45,48] as a potential risk factor for autism in TSC. Temporal lobe location was linked to poor neurodevelopmental outcomes, including severe language impairment, autism and intellectual disability in these studies; however, this approach did not fully explain the neurodevelopmental phenotype of TSC. Newer studies, illuminating the neurobiology of classic brain lesions in TSC through detailed structural and functional analysis, have expanded knowledge regarding the potential impact of these lesions on the neurodevelopmental phenotype in TSC[21,49-51].

Different approaches to analysis of brain lesions in TSC have identified features associated with autism. Cortical tuber and/or white matter lesion load has been identified as a potential risk factor for poor neurodevelopmental outcomes[14,33,36,42,50]. Cortical lesions seem to worsen the neurological phenotype. The mechanism by which this occurs is unclear, but may be associated with epilepsy and/or circuit disruption or reorganization. T1, T2 and FLAIR imaging have been used to identify specific characteristics of white matter that are associated with autism[51]. White matter abnormalities have also been detected using diffusion tensor imaging fractional anisotropy (FA), a measure of white matter integrity. Lower FA values indicate loss of the typically restricted diffusion found in normal white matter, and are thought to represent deficits in white matter. Autism spectrum disorder in 12 individuals with TSC was associated with lower average FA when compared to 28 individuals with TSC without autism spectrum disorder and 29 age-matched controls[52]. A subsequent study compared the FA of the arcuate fasciculus, which interconnects language areas in the temporal and frontal lobes, among individuals with TSC, with or without autism, to typical controls. FA in this key pathway for language was lowest in those individuals with TSC and autism[53].

Functional analysis has been carried out using EEG, PET, and diffusion tensor imaging (DTI). Across all of these studies, temporal lobe abnormalities were most commonly identified as a risk factor for poor neurodevelopmental outcome. In a retrospective study of 19 individuals, temporal lobe epileptiform activity and seizure onset within the first 36 mo of life were independently associated with autism and PDD[39]. Interictal temporal lobe spikes increased the likelihood of autism spectrum disorder by a factor of 15 in another group[20]. Poor neurodevelopmental outcome was also associated with a subtype of hypsarrhythmia[47]. PET studies revealed reduced glucose metabolism (a measure of neuronal activity) in temporal lobe of individuals with autism and TSC, but increased glucose metabolism in deep cerebellar nuclei that was associated with intellectual disability, stereotypical behavior, communication impairments, and impaired social interaction[43]. Subsequent PET studies examined tubers in cerebellar cortex and found decreased glucose metabolism associated with symptoms of autism[54]. Output from cerebellar cortex normally inhibits the deep cerebellar nuclei, and decreased activity in cerebellar cortex would be expected to result in increased activity in the deep cerebellar nuclei, as reported in the earlier study. The effect of tuber location within the cerebellum has also been examined, and children with right cerebellar tubers showed increased social isolation and deficits in communication and development, compared to children with left cerebellar tubers[54]. The output of the right cerebellum influences activity in the left cerebral cortex, where language areas are located in most individuals, and taken together these findings support a role for cerebellar deficits in TSC-associated autism.

Joint effects: neuropathology and epilepsy

Almost half of the reviewed studies simultaneously investigated features of both epilepsy and neuropathology as potential risk factors. In the majority of those studies, autism was associated with joint effects of these two risk factors[33,34,36,39,40,42,43,45,47,48]. Thus, multiple factors should be considered in a risk assessment for autism in individuals with TSC. Development of a standard, generalizable approach to assessing these complex risk factors would be informative for clinical trial design in this population.

NEUROBIOLOGY OF TSC RELATED TO DEVELOPMENTAL DISABILITIES

An excellent and comprehensive review recently outlined effects of TSC proteins and the mTOR pathway in the nervous system[55]. A working knowledge of basic neuroscience research as it pertains to TSC is critical in considering therapeutic targets for humans. Highlighted here are discoveries most relevant to patients with severe manifestations of TSC: central nervous system effects of TSC, animal models with phenotypes that mimic particular manifestations of TSC, and effects of rapamycin in these models. Autism, epilepsy and cognitive impairment have all been modeled preclinically, and treatment with inhibitors of mammalian target of rapamycin, such as rapamycin and everolimus, has demonstrated potentially beneficial effects in some models (Table 3).

Table 3.

Selected preclinical models of tuberous sclerosis complex

Model	Genetic manipulation	Behavioral effects	Seizures	Autism	Synaptic plasticity	Neuropathology	Rapamycin	Ref.
Tsc1+/- mice	Heterozygous Tsc1 deletion, exons 6-8	↓Hippocampal dependent learning	None	↓Social interaction	-	None	-	[58]
Tsc2+/- Eker rat	Spontaneous autosomal dominant (Heterozygous)	-	-	-	↓LTP, ↓LTD, ↑PPF	-	-	[61]
Tsc2+/- Eker rat	Spontaneous autosomal dominant (Heterozygous)	No learning and memory deficits	No spontaneous	↓Social interaction, ↓↓ after seizure induction	-	-	-	[59]
Tsc2+/-	Heterozygous disruption in second exon	↓Hippocampal dependent learning	-	-	↓Threshold for L-LTP	-	Reversed all	[93]
Tsc2+/- or WT pups from Tsc2+/- or WT dams	Heterozygous Tsc2 deletion	↑Maternal care by Tsc2+/- dams	-	↑Vocalization in WT and Tsc2+/- pups of Tsc2+/- dams	-	-	-	[60]
Tsc2+/-	Heterozygous Tsc2 deletion	-	-	-	↓mGluR-LTD,	-	Reverses deficits in protein-synthesis-dependent mGluR-LTD	[62]
					↓Arc synthesis
Tsc1fl/fl Nestin-rtTA(+)TetOp-cre(+)	Mosaic homozygous Tsc1 deletion in cortical neural progenitors	-	Yes+	-	-	Heterotopias with enlarged, pS6+ neurons	↑Survival, ↓seizures, ↓neuropath	[94]
						White matter nodules
Tsc1fl/-Syn1-Cre	Homozygous Tsc1 deletion in neurons from mid-gestation; (heterozygous in all other cells)	-	Yes+	-	-	Dysplastic neurons	↑Survival, ↑myelination, ↑body weight, ↓neurologic impairment	[95,96]
						↓Cortical organization
						↓Myelin
Tsc1fl/fl Emx1-Cre	Homozygous Tsc1 deletion in early embryonic neural progenitors	-	Yes+	-	-	↓Cortical organization	↓Seizures, ↑survival, ↓glial abnl, ↑weight, ↓brain size	[97]
						↑Brain size
						↓Myelin
						tHick astrocyte processes
Tsc1fl/fl Nestin-Cre	Homozygous Tsc1 deletion in differentiating neurons	-	-	-	↑EPSCs	↑Cell size	↓Spine width ↑spine length	[98]
					↑AMPA	↑Spine width
						↑Spine length
						↓Spine density
Tsc1fl/fl Syn1-Cre	Homozygous Tsc1 deletion in neurons	-	No spontaneous	-	↑EPSCs	No gross abnormalities	-	[99]
			Epileptiform discharges		↑AMPA
Tsc1fl/fl GFAP-Cre	Homozygous Tsc1 inactivation in glial-fibrillary acidic protein (GFAP)-+ cells		Yes+, ↓Glt1			↑Astrocytes	↑Glt1, ↑survival ↓neuropath Early: prevented epilepsy, Late: decreased seizure frequency	[100,101]
						↑Brain size
						↓Hippocampal organization
Tsc2fl/fl GFAP-Cre	Homozygous Tsc2 inactivation in GFAP + cells	-	Yes++, ↓Glt1	-	-	↑Astrocytes	↑Survival, ↓seizures, ↓neuropath	[102]
						↑Brain size
						↓Hippocampal organization
Tsc1fl/+ L7-Cre, or Tsc1fl/fl L7-Cre	Heterozygous or homozygous Tsc1 deletion limited to cerebellar Purkinje cells	Normal acquisition, ↓reversal of spatial learning in homozygous mutants	-	↓social interaction in both genotypes ↑grooming, vocalization	PC ↓excitability in heterozygous, ↓↓ in homozygous	PC loss in homozygous, ↑PC dendritic spine density in both heterozygous and homozygous	Reversed pathological and behavioral abnormalities	[64]
Tsc2fl/- , or Tsc2fl/-Pcp2-Cre	Heterozygous Tsc2 deletion (global), or homozygous Tsc2 deletion in cerebellar Purkinje cells, heterozygous in other cells	-	-	↑repetitive behavior in homozygous, ↓social interaction in both genotypes	-	PC loss in homozygous,	Reversed social deficits	[63]

These are existing preclinical models designed to re-capitulate neurologic features of tuberous sclerosis complex (TSC). - Not examined; +: Mild seizure severity; ++: Moderate seizure severity.

TSC is an autosomal dominant condition that occurs when there is a mutation in either TSC1, which is located on chromosome 9q34 and encodes the protein hamartin, or TSC2, which is located on chromosome 16 and encodes the protein tuberin. Normally, TSC1 (hamartin) and TSC2 (tuberin) form a complex in which TSC1 stabilizes TSC2 by blocking its ubiquitination and degradation. The TSC1/TSC2 complex inactivates Rheb, which otherwise stimulates mammalian target of rapamycin (mTOR) activity. Mutations in TSC1/TSC2 lead to increased mTOR activity and dysregulation of gene transcription, metabolism, and cell proliferation. Although most cells in individuals with TSC have a single germline mutation in TSC1 or TSC2, recent studies discovered a second somatic mutation limited to giant cells within cortical tubers or to subependymal giant cell astrocytomas (SEGAs). These cells with mutations in both alleles of TSC1 or TSC2 were characterized by hyperactivation of mTOR[56,57]. These studies suggest that a two-hit mechanism may underlie the formation of cortical tubers or SEGAs, but the contribution of this mechanism to development of autism/ASD remains unclear.

Animal models have been developed to investigate the effects of heterozygous mutation of TSC1 or TSC2, or of conditional deletion of one or both alleles in different populations of neurons or in astrocytes. The characteristics and major findings obtained with selected models are summarized in Table 3.

Heterozygous Tsc1 and Tsc2 models

Mouse lines that are Tsc1^+/- or Tsc2^+/- and a rat strain with a spontaneous mutation in Tsc2 (Eker rat) have been characterized in a number of studies. Although there are no neuropathologic findings and no spontaneous seizures reported in these heterozygous mutants, behavioral abnormalities include social interaction deficits[58,59], abnormal vocalization[60] and learning deficits that may be related to altered synaptic plasticity[58,61]. Both synaptic and behavioral learning deficits were reversed by acute rapamycin treatment. One study found that social interaction deficits in the Eker rat were exacerbated by seizure induction in the postnatal period[59]. Thus, in animal models with heterozygous mutations in Tsc1 or Tsc2, social interaction deficits and altered maternal-pup interactions that model some aspects of autism occur in the absence of cortical lesions or epilepsy, but can be exacerbated by seizures induced early in postnatal brain development. The latter findings support intervention to control seizures in infants with TSC.

TSC and Fragile X syndrome may share neurobiology related to intellectual disability and autism, since these conditions have been found to result in opposite effects on metabotropic glutamate receptor (mGluR)-mediated protein synthesis in a recent preclinical study. In Tsc2^+/- mice, mGluR-mediated long term depression was abnormally decreased, but in Fmr^-/y mice it was increased. These alterations were rescued by up or down modulation of mGluR5 activity, respectively, and by rapamycin treatment in Tsc2^+/- mice. Investigators also crossed the two mouse lines and observed reversal of memory deficits in offspring expressing both mutations[62].

Conditional homozygous deletion of Tsc1 or Tsc2

Mice with conditional deletion of both alleles of Tsc1 or Tsc2 in cortical neurons exhibit a variety of structural abnormalities such as increased brain and cell size, dysplastic neurons and deficits in cortical organization; mosaic homozygous inactivation of Tsc1 produced heterotopias and white matter nodules that resemble some cortical lesions in individuals with TSC. Spontaneous epileptiform discharges, seizures and increased excitatory synaptic activity have also been reported in these mice. Conditional homozygous deletion of Tsc1 in GFAP-expressing cells, which include astrocytes and possibly adult neural progenitors in neurogenic niches such as the hippocampal subgranular zone, decreased the expression of the astrocyte transporter for glutamate, Glt1. These mice exhibited spontaneous seizures, altered hippocampal organization, as well as increased astrocyte number and brain size. Thus, homozygous deletion of Tsc1 or Tsc2 produced more severe effects on brain structure and function and also reduced survival; rapamycin treatment reversed or ameliorated glial and neuronal abnormalities and increased survival. In the mice with homozygous deletion of Tsc1 in astrocytes and GFAP-positive neural progenitors, early rapamycin treatment prevented seizures and later treatment reduced seizure frequency. None of these studies of mice with homozygous deletion in cortical neurons or astrocytes examined social behavior.

Two recent studies examined the behavioral effects of conditional loss of one or both alleles of Tsc1 or Tsc2 that was limited to cerebellar Purkinje cells (PCs)[63,64]. Homozygous deletion caused a progressive loss of PCs, and both heterozygous and homozygous deletion mutants exhibited a striking increase in PC dendritic spine density and reduced PC excitability. Autistic-like behaviors were observed in both heterozygous and homozygous mutants, including deficits in social approach, response to social novelty, altered vocalization during a limited postnatal period, and increased grooming. Chronic rapamycin treatment initiated at about 1 wk of age reversed the neuropathologic and behavioral effects in these mice. These animal models provide important insights into the role of the cerebellum in autism caused by TSC, which was initially proposed based on human neuropathologic and PET imaging studies.

These animal models of TSC exhibit the neuropathological, neurophysiological and behavioral abnormalities characteristic of TSC. They have been used to evaluate therapies based on mTOR inhibition, antiepileptic medications, and positive modulation of mGluR5 activity, and provide essential platforms for developing and preclinical testing of new therapies.

IMPLICATIONS FOR TREATMENT OF AUTISM

Currently available behavioral and pharmacological treatment options for children with TSC and autism do not differ from treatments available to children without TSC[65,66]. Mutations in TSC1 or TSC2 lead to increased activity of mTOR. Two mechanism-based treatments have been FDA-approved for use in individuals with TSC, everolimus for tumors (SEGAs and renal angiomyolipomas) and vigabatrin for infantile spasms. It is unknown whether either treatment would be effective for severe autism.

The protein kinase mTOR is incorporated into two protein complexes, mTORC1 and mTORC2, which have distinct downstream effects on gene transcription, metabolism, cell proliferation, and synaptic plasticity. Everolimus targets excess mTOR activity in the protein complex mTORC1 (mTORC2 is unaffected). After 12-18 mo of treatment, everolimus has been shown to increase abnormally low FA in an open-label study of 20 individuals with TSC and SEGAs[67], indicating an improvement in white matter structure. A clinical trial of everolimus for neurocognition in high-functioning individuals with TSC is ongoing and may include individuals with autism.

Vigabatrin, an irreversible inhibitor of γ-aminobutyric acid (GABA) transaminase, targets excessive levels of glutamate (the brain’s most common excitatory neurotransmitter) by increasing levels of GABA (the brain’s inhibitory neurotransmitter). To date, there are no active clinical trials of vigabatrin for neurocognition or autism. Directly targeting the neurobiology of TSC may be therapeutic for children with severe autism; however, definitive clinical trials are warranted. Successful design and implementation of these trials will require careful planning. In this review, we identified 4 areas of need: (1) validation of severe autism as an endophenotype in TSC; (2) natural history of severe autism in TSC; (3) selection of reliable, valid outcome measures in this population; and (4) assessment of recruitment and protocol compliance (feasibility factors relevant to trial implementation).

DISCUSSION

To date, discovery of mechanism-based treatments for TSC has not impacted the clinical and educational intervention offered to children with autism and TSC. A severe form of autism characterized by self-injurious behavior and/or aggression may exist in some individuals with TSC; however, few data are available about this potential endophenotype. Level of parental concern, danger to self and others, decreased quality of life, and lack of full participation in the community provide a rationale for prioritizing inclusion of this population in clinical trials. Clearly, improved treatment options for individuals with extreme behavioral manifestations of TSC are needed. Targeting the neurobiology directly may result in greater improvement than non-specific treatments.

This structured review of TSC literature revealed important clinical and preclinical features of TSC, including autism, intellectual disability, self-injury, and/or aggression in individuals with TSC. Several clinical associations were without clear causation, such as location of cortical tubers and autism. However, clinical trials demonstrating the efficacy of vigabatrin in individuals with TSC and infantile spasms[68] and of everolimus for non-surgically resectable SEGAs[69] have led to FDA approval for these indications. A comprehensive phenotypic analysis of patients with TSC can be achieved using measures that provide an assessment of neural activity, such as EEG and PET scans, newer MRI methods that detect subtle structural abnormalities, and appropriate behavioral assessments. This approach may also prove useful for evaluating the efficacy of mechanism-based treatments.

Review of the basic science literature highlighted the biological pathways and downstream effects of increased mTOR activity in TSC, alterations in neurochemistry in TSC, effects of pharmacological agents that directly impact the mTOR pathway, and described several animal models that recapitulate important features of TSC in humans. Although an animal model that exhibits all of the severe manifestations of this disorder does not currently exist, further study of the current models will continue to lead to important discoveries.

FUTURE DIRECTIONS

Shared deficits in neurobiology related to neurodevelopmental disabilities may represent an opportunity to generalize findings from research in TSC to individuals affected by related conditions. Abnormalities in mTOR signaling and associated pathways, as seen in TSC, have been identified in focal cortical dysplasia type II, hemimegalencephaly[70], phosphatase and tensin homologue hamartoma syndromes, such as autism associated with mutations in PTEN[71-73], neurofibromatosis-1[74-76], and autism associated with mutations in EIF4E (eukaryotic translation initiation factor 4E)[77-79]. Targeting pathophysiological mechanisms has resulted in phenotypic rescue in animal models of several other single gene disorders, such as Lhermitte-Duclos[80], Fragile X syndrome[81], Angelman syndrome[82], Rett syndrome[83], Neurofibromatosis I[84,85], Down syndrome[86,87], and Rubenstein-Taybi[88-90].

Although usually considered to be chronic, life-long conditions, there is hope for children and adults affected by TSC and other diseases that are associated with severe neurological manifestations, such as self-injurious behavior, aggression, intellectual disability, autism and seizures. Rescue of the synaptic plasticity deficits that underlie these manifestations may eventually be achieved with comprehensive treatment by pairing available mechanism-based treatments with evidence-based educational and behavioral interventions[91,92].

ACKNOWLEDGEMENTS

The authors thank clinical librarian Victoria Goode, MLIS, for assistance with development of search strategies, research students Raechel Mattison and Nassira Bougrab for assistance, and colleagues Louis Hagopian, PhD, Lee Wachtel, MD, and Dejan Budimirovic for helpful discussions.