Sotos Syndrome: Deep Neuroimaging Phenotyping Reveals a High Prevalence of Malformations of Cortical Development

Imaging Findings

The most intriguing finding in this study is the 95% prevalence of cortical development malformations in patients with Sotos syndrome. This high prevalence has not been described in previous studies, most published before the term “dysgyria” was introduced. This finding may link the role of the NSD1 gene in brain development with the appearance on MR imaging.

Several previous studies have investigated brain development in Sotos syndrome, with only a few cases describing gray matter heterotopia¹¹ and a single case report of polymicrogyria.¹² There has been one additional report of an individual with polymicrogyria and an unbalanced translocation, including a deletion encompassing the NSD1 gene.¹⁰ The contribution of ≥1 of the numerous other deleted or duplicated genes to the imaging phenotype of this individual cannot be excluded, though our current findings are in support of the NSD1 deletion as a possible cause for polymicrogyria in that patient.

Incomplete hippocampal rotation can be a normal variant, especially when it occurs unilaterally on the left, or alternatively, the incomplete hippocampal rotation can be a developmental abnormality.¹⁷ In our cohort, incomplete hippocampal rotation tended to be bilateral and more common than in healthy children (50.6% of cases), suggesting that it may be secondary to regional temporal dysgyria from abnormal cortical folding. Normal hippocampal folding is theorized to be secondary to adjacent cortical development, thus potentially explaining concurrent cortical malformations.¹⁸ A statistically significant association was found between incomplete hippocampal rotation and dysgyria, suggesting that it may represent an imaging stigmata of Sotos syndrome.

Whitehead et al¹⁹ showed abnormalities involving the massa intermedia and anterior commissure to be more common in patients with midline abnormalities than in those without them. We found, in concordance, massa intermedia absence in 23.4% and a thin anterior commissure in 49.4% of patients with Sotos syndrome. These commonly affected midline structures should be evaluated as part of routine MR imaging reporting in patients with Sotos syndrome.

Slight dysmorphism of the brainstem, mainly a shallow pontomedullary sulcus, was observed in 23.4% of patients. It is most likely a feature secondary to the described cranial abnormalities; hence, no true malformation was observed, but it possibly represents another aspect of brain maldevelopment.

Other imaging findings corroborate previous reports, with macrocephaly, ventriculomegaly, and midline abnormalities observed in most MR imaging scans inspected and in all age groups.

Fetal Imaging

Published data regarding the prenatal presentation and imaging features of Sotos syndrome are sparse. Zhang et al²⁰ described a case series of 7 patients with subtle nonspecific ultrasound findings such as ventriculomegaly. Lu et al²¹ published a case with fetal MR imaging showing ventriculomegaly and polyhydramnios. The findings on 1 fetal MR imaging included in our study are similar to those found postnatally for this child (Figs 2 and 3). However, the follow-up postnatal MR imaging revealed maturation of dysgyria, which was merely hinted at in the fetal MR imaging as asymmetry only. Together, cortical gyral abnormalities or asymmetries, abnormal hippocampal rotation, ventriculomegaly, and enlargement of the cavum septum pellucidum should raise the possibility of Sotos syndrome and the possible need for further genetic testing.

Neoplasia

Sotos syndrome, like other overgrowth syndromes, is known to entail an increased risk of malignancies.²² The prevalence of 5.2% in our cohort supports the evidence of an elevated neoplasm risk in these patients with a variance in tumor type and location.

Clinical Findings

The clinical findings in our cohort are consistent with the previous literature, with core features of overgrowth, psycho-cognitive impairment, seizures, scoliosis, and other neurologic or ophthalmologic abnormalities. The high prevalence of developmental delay and intellectual disability, autism spectrum disorder, and attention deficit/hyperactivity disorder have been described in previous studies.^23,24 The somewhat lower prevalence of autism spectrum disorder and attention deficit/hyperactivity disorder may be explained by the young age in our cohort, patients not yet evaluated for such conditions. This understanding could enable early assessment and proper educational adjustments.

Genetic and Biochemical Literature Review

Several findings support the important role of the NSD1 gene in brain development and provide possible correlates for the observed malformations of cortical development in our cohort. Human expression data (Human Protein Atlas, Genotype-Tissue Expression Project [GTEx; https://www.genome.gov/Funded-Programs-Projects/Genotype-Tissue-Expression-Project]) demonstrate that NSD1 is expressed in the brain and in the cerebral cortex. It binds upstream of the promoter regions, thereby regulating transcription through interactions with methylated lysine 36 of histone H3 (H3K36) and RNA polymerase II.²⁵ Oishi et al²⁶ have shown that there is a predominantly neuronal expression of NSD1 in the cerebral cortex of mice. Complete knockout of NSD1 impaired migration and laminar positioning of cortical neurons²⁷ and caused high apoptosis and an inability to complete gastrulation, ultimately resulting in lethality.⁵

A key downstream effector of NSD1 is the adenomatous polyposis coli 2 (APC2) gene. The APC2 gene has a crucial role in neuronal migration,²⁸ and biallelic pathogenic variants in APC2 are known to lead to a neurodevelopmental disorder characterized by brain malformations, including cortical dysplasia and lissencephaly among others.²⁹ Most interesting, Almuriekhi et al²⁷ and Mastrangelo et al³⁰ have described 3 individuals from 2 families with biallelic pathogenic variants in the APC2 gene, presenting with features resembling Sotos syndrome, including macrocephaly. Additionally, APC2 is a regulator of the Wnt signaling pathway, and there is substantial evidence of a connection between NSD1 and the Wnt signaling pathway,³¹ which could possibly contribute to the increased incidence of neoplasms in Sotos syndrome.

The findings of Quintero-Rivera et al³² suggest that NSD1 overexpression influences the PI3K/AKT/mTOR signaling pathway. Because the PI3K/AKT/mTOR pathway has a well-established role in cell growth, they propose that the undergrowth seen in NSD1 overexpression is related to the effect it has on this pathway. It would, therefore, be feasible that NSD1 loss of function may likewise cause overgrowth through the PI3K/AKT/mTOR pathway. Pathogenic variants in the genes of the PI3K/AKT/mTOR pathway are a well described cause of megalencephaly,³³ once again linking NSD1 with malformations of cortical development.

Implications

The unexpected high prevalence of cortical malformations in our study carries important clinical-practical implications. When classic Sotos syndrome–related neurologic signs and symptoms are present, brain MR imaging should be performed and scrutinized for typical midline abnormalities and cortical malformations that may be subtle. The cortical malformations in Sotos syndrome often manifest as dysgyria or nonspecific cortical asymmetries on fetal MR imaging and, therefore, may be easily overlooked, especially without an optimal MR technique or when there are motion artifacts.

Limitations

Variability in imaging was an unavoidable consequence in this multicenter, retrospective study that included examinations acquired with various MR imaging hardware, magnet field strengths, and imaging protocols. Our inclusion criteria aimed to eliminate or at least diminish the effect of these differences. The clinical data were available in most patients and did not require patient re-examination. The multicenter structure does pose a strength in the reduction of selection bias. Furthermore, the wide consensus with an expert from each center reduces observer bias.

Brain malformation classification and terminology have had many revisions across the years, which may explain the previous underdiagnosis and under-reporting. Dysgyria, a rather recent term, was introduced in association with the Tubulin gene family pathogenic variants and has since been identified in other conditions. A qualitative assessment of dysgyria rather than a measurement-based grading may present a limitation of our study, and further studies should focus on such measurements of the cortical structure as well as better classifying dysgyria.

The assessment of the anterior commissure was only possible in patients with a high-resolution sagittal T1WI. Due to the differences in imaging protocols, we did not perform a measurement and used a binary qualitative criterion based on visual inspection. Further research regarding dysgyria subtypes, quantitative brain measurements, as well as genetic, histologic, and clinical correlations will be necessary for a better, repeatable, and precise classification.