Stroke-Inducing Cavernous Angiomas Linked to Gut Microbiome

July 3, 2020

An NIH-funded study provided supporting evidence of an association between high levels of gram-negative bacteria and a stroke-, seizure-, and headache-inducing vascular abnormality.

Sean Polster, MD

Findings from a National Institutes of Health (NIH)-funded nationwide study suggest that composition of gut bacteria is linked to a stroke-related abnormality: cavernous angiomas (CA), also known as cerebral cavernous malformations, irregular bundles of brittle blood vessels in the brain or spinal cord.1,2

Using advanced genomic analysis methods, the assessment compared stool samples from 122 individuals with ≥1 CA confirmed by brain scan, to samples from an age- and sex-matched, non-CA control cohort. On average, those with CAs had more gram-negative bacteria while though without had more gram-positive bacteria. These lesions, which contain slow-moving or stagnant blood, can cause hemorrhagic strokes, seizures, or headaches. The current treatment for them involves the surgical removal of lesions when it is safe to do so.2

Ultimately, the abundance of 3 bacterial species—the gram-negative Odoribacter splanchnicus, which was significantly increased in CA patients (false discovery rate corrected P value [pFDR] ≤.05); and the gram-positive Faecalibacterium prausnitzii and Bifidobacterium adolescentis, which were significantly decreased (pFDR ≤.05) in CA patients—allowed the authors to distinguish those with CAs from controls.

This work, conducted by Sean Polster, MD, neurosurgery resident, University of Chicago, and colleagues, builds on prior studies in mice and a small number of patients which suggested there may be a link between CA and gut bacteria. Polster et al. noted that the results provide the first demonstration in humans of a “permissive microbiome” associated with the formation of neurovascular lesions in the brain.

READ MORE: Stroke Care Evolves During These Uncertain Times

Polster et al. noted that using receiver operating characteristic (ROC) curves, the optimal weighted combination of the 3 bacterial species provided 92% sensitivity and 67% specificity in association with CA diagnosis. Similar bacterial species respectively distinguished sporadic/solitary and familial/multifocal CA cases from non-CA cases.

“We have demonstrated that application of machine learning methods can identify weighted combinations of microbiome signatures and plasma biomarkers that enhance disease severity associations,” the authors noted.

Their analysis showed that some gut bacteria compositions could identify aggressive versus non-aggressive forms of the disease as well as those with recent symptomatic hemorrhages. Multi-variate differential abundance analyses and random forest classifier identified 5 taxa which contributed significantly, Bifidobacterium adolescentis, Bacteroides eggerthii, Bacteroides dorei, Dorea, and Escherichia coli—by which the differences could moderately distinguish aggressive versus non-aggressive CA patients. Usinig 16S rRNA amplicon sequencing data, the group observed that α- and β-diversity significantly differ between aggressive and non-aggressive CA patients, with a significant decrease (pFDR ≤.05) in the proportion of 2 exact sequence variations belonging to genus Bacteroides in aggressive disease.

Those with CA who have clinically defined symptomatic hemorrhage (CASH) showed that a significant network could be constructed by co-occurrence network analysis for non-CASH patients (R2 >0.6; P <.05), but not CASH patients. Multi-variate differential abundance analyses and random forest analyses identified 6 taxa—Faecalibacterium prausnitzii, Oscillobacter, Lactobacillus rhamnosus, Enterobacter cloacae, Odoribacter laneus, and Bacteroides cellulosilyticus—with which ROC curve could moderately distinguish the 2 groups.

“Microbiome associations of disease severity and hemorrhage appear different and more complex than lipopolysaccharide mechanisms reflecting CA diagnosis (i.e., lesion development), and these require further investigations,” Polster et al. wrote.


1. Polster SP, Sharma A, Tanes C, et al. Permissive microbiome characterizes human subjects with a neurovascular disease cavernous angioma. Nature Communications. 2020;11:2659. doi: 10.1038/s41467-020-16436-w

2. Study ties stroke-related brain blood vessel abnormality to gut bacteria. NIH website. Published June 3, 2020. Accessed July 1, 2020.