Spontaneous spinal CSF leaks: a rare variant exome sequencing study and functional analysis.

backgroundSpontaneous spinal CSF leaks are associated with connective tissue diseases including Marfan syndrome and Loeys-Dietz syndrome, which are caused by mutations in genes that influence the content and integrity of the extracellular matrix. Patients with spontaneous spinal CSF leaks without a defined connective tissue disease diagnosis can show subtle or non-specific connective tissue disease manifestations, suggesting that mutations in extracellular matrix proteins might contribute to more common presentations of this condition. By doing a whole-exome sequencing study, we aimed to elucidate the genetic basis of spontaneous spinal CSF leaks.

methodsThrough retrospective medical record review at Cedars-Sinai Hospital (Los Angeles, USA), we identified 42 individuals who had lateral spontaneous (ie, type 1b) spinal CSF leaks. We did a whole-exome sequencing study in these individuals and compared these data with the results of whole-exome sequencing for three independent control cohorts (2244 unrelated and unaffected adults recruited from various sites in the USA and 714 and 913 individuals recruited to separate sequencing initiatives at the University of Antwerp, Belgium). We used an in-silico prediction tool to establish the location of variants in the tertiary structure of the protein encoded by the top candidate gene. We also tested wild-type and mutant protein fragments for integrin-mediated binding to human dural fibroblasts in vitro. With CRISPR-Cas9 gene editing, we generated three mouse models harbouring different variants in the top candidate gene equivalent to variants found in individuals with type 1b spontaneous spinal CSF leaks and compared them with an established mouse model of Marfan syndrome. Intrathecal infusion testing was used to establish dural integrity and CSF leak properties in these mice.

findingsWhole exome sequencing on 42 unrelated individuals with type 1b spontaneous spinal CSF leaks who had been seen between Jan 1, 2006 and Dec 31, 2019 (35 [83%] were women, seven [17%] were men, 38 [90%] were White, two [5%] were Black, and two [5%] were Hispanic) identified potential causative genes. Nine (21%) of 42 individuals with type 1b spontaneous spinal CSF leak had rare functional variants in FBN2. Significant enrichment in rare functional FBN2 variants was observed on comparison of the patient cohort with the Mendel discovery cohort (177 [8%] of 2244, p=0·041, odds ratio (OR) 3·18 [95% CI 1·50-6·76]) and the Belgian whole-exome (51 [7%] of 714, p=0·004, OR 3·55 [1·61-7·81]) and Belgian thoracic aortic aneurysm and dissection (45 [5%] of 913, p=0·0003, OR 5·26 [2·38-11·66]) validation cohorts. FBN2 variants in individuals with type 1b spontaneous spinal CSF leaks showed a non-random domain distribution in fibrillin-2, with enrichment in TGF-β binding protein-like (TB) domains. Two out of three tested variants reduced fibrillin-2 fragment adhesion to human dural fibroblast in vitro. Mice carrying variants equivalent to the three FBN2 variants in TB domains found in individuals with type 1b spontaneous spinal CSF leak (Fbn2

interpretationRare deleterious variants in FBN2 might cause type 1b spontaneous spinal CSF leak, supporting the integration of FBN2 genetic testing into clinical practice. As in other connective tissue diseases, the disruption of cell adhesion to extracellular matrix proteins might participate in the pathophysiology of spontaneous spinal CSF leaks. The generation of mouse models of spontaneous spinal CSF leaks will help the development of pharmacological therapeutic strategies.

fundingThe Howard Hughes Medical Institute, the Marfan Foundation, the Pease-Scheeler Fund, and the Biotechnology and Biological Sciences Research Council.