- In this study, a combination of metabolomics and machine learning was used to investigate the prognostic value of cerebrospinal fluid (CSF) samples from patients with severe aneurysmal subarachnoid hemorrhage (aSAH)
- CSF concentrations of ornithine, dimethylguanidine valeric acid and symmetric dimethylarginine were associated with poor functional outcome (rated on the modified Rankin Scale at discharge and 90 days after aSAH)
- These three metabolites are related and help regulate synthesis of nitric oxide
- It's not yet known whether these metabolites are a marker of damage or are causative agents that might be therapeutic targets
Aneurysmal subarachnoid hemorrhage (aSAH) is often fatal or results in a poor neurologic outcome, which is difficult to predict based on clinical and demographic factors. In a previous study, researchers at Massachusetts General Hospital used machine learning to analyze more than 160 metabolites in blood samples, and they identified taurine as a potential biomarker of functional outcome after aSAH.
Subscribe to the latest updates from Neuroscience Advances in Motion
Studies of serum are limited to metabolites in the peripheral circulation, whereas cerebrospinal fluid (CSF) contains metabolites more proximate to the site of aSAH. Therefore, Matthew Koch, MD, neurosurgery resident, and Aman Patel, MD, director of Cerebrovascular and Endovascular Neurosurgery at Mass General; W. Taylor Kimberly, MD, PhD, chief of the Division of Neurocritical Care at Mass General; and colleagues applied a similar approach to the study of CSF metabolites. In Neurosurgery, they report that several CSF metabolites may have a role in prognostication for patients with severe aSAH.
The study material was 140 CSF samples collected from external ventricular drains placed in 81 adults with aSAH. Three samples were collected from each patient: on days 0–5 after aSAH, days 6–10 and days 11–15. The researchers also collected CSF samples from 16 patients undergoing elective surgery for nonruptured aneurysms, who served as controls.
Functional outcome was rated on the modified Rankin Scale at hospital discharge and on day 90.
CSF Metabolites Related to Outcomes
The team used machine learning to rank 138 metabolites according to their ability to predict functional outcomes. Metabolites in this analysis that predicted poor outcomes at both discharge and 90 days were:
- Ornithine (P=0.002)
- Dimethylguanidine valeric acid (DMGV, P=NS)
- Symmetric dimethylarginine (SDMA, P=0.001)
These three metabolites are related and help regulate the synthesis of nitric oxide.
Confirmation of Results
The researchers then applied a second type of machine learning, which prioritizes features that best distinguish between two classes (in this case, good versus poor outcome). Only two metabolites were identified in both analyses: ornithine and SDMA.
SDMA concentrations were significantly lower in controls than in aSAH patients with poor outcomes (P=0.009) and did not differ significantly between controls and patients with good outcomes.
Effect of Time
Ornithine, DMGV and SDMA were significantly elevated across CSF collection times in patients who had poor outcomes.
Poor neurologic outcomes from aSAH are a consequence of both primary brain injury (the rupture) and secondary injury (the sequelae of the hemorrhage). Secondary injury is an important therapeutic target.
It's not yet known whether the three metabolites identified in this study are a marker of damage or are causative agents that might be targeted for intervention. The researchers are planning additional work to explore the origin of elevated SDMA, DMGV and ornithine in aSAH.
Learn more about the Cerebrovascular Surgery Program at Mass General
Refer a patient to the Department of Neurosurgery