A Methylation Signature in Brain Fluid Predicts Who Survives a Deadly Stroke

When a brain aneurysm ruptures, blood floods the narrow spaces surrounding the brain, triggering one of the most devastating forms of stroke known to medicine. Aneurysmal subarachnoid hemorrhage, or aSAH, kills between 25 and 50 percent of its victims within a month, and roughly 60 percent of survivors are left with disabilities that erode their ability to work, live independently, and maintain relationships. For decades, neurologists have struggled to explain why two patients with seemingly similar injuries can follow radically different recovery paths. Now, a team of researchers at the University of Pittsburgh has uncovered a striking clue written not in the genome itself, but in its chemical packaging — and the discovery points to an unexpected player in brain injury: an iron-processing enzyme called STEAP3.

The study, published in Epigenetics Communications, took advantage of a rare and precious resource: cerebrospinal fluid, the clear liquid that bathes the brain and spinal cord, collected day after day from patients fighting for their lives in a neurovascular intensive care unit. Because cerebrospinal fluid is drained as part of routine clinical management to relieve pressure inside the skull after a hemorrhage, the researchers could sample a tissue that sits in intimate contact with the injured brain — something that blood tests simply cannot offer. From 260 patients, they generated longitudinal, genome-wide DNA methylation data spanning 13 days after the bleed, then validated their findings in an independent group of 100 patients.

DNA methylation is a chemical modification in which small methyl groups are attached to specific positions in the DNA sequence, typically dampening the activity of nearby genes. It is a master regulator of how the same genetic blueprint produces neurons, immune cells, and everything else in the body, and it is known to influence brain function, plasticity, and recovery after injury. The Pittsburgh team focused on 36 candidate genes involved in iron homeostasis, the body’s system for managing iron — a logical target because a ruptured aneurysm dumps hemoglobin-rich blood into the cerebrospinal fluid, where heme is broken down to release free, unbound ferrous iron. That free iron is dangerous: it fuels a distinctive form of iron-dependent cell death called ferroptosis, which has been implicated in secondary brain injury after hemorrhagic stroke in animal models.

To make sense of the longitudinal data, the researchers employed group-based trajectory analysis, an unsupervised clustering method that sorts patients into groups sharing similar patterns of change over time. Rather than estimating a single average trajectory for the whole sample, the technique asks whether the population is actually composed of distinct subgroups, each with its own methylation course. Crucially, the team automated the modeling procedure with a predefined decision tree, removing the subjective, iterative choices that traditionally plague this kind of analysis. Of 637 methylation sites examined across the candidate genes, 183 sites in 33 genes yielded reliable trajectory groups and were carried forward for association testing against patient outcomes.

One site towered above the rest: cg25713625, a methylation position within the STEAP3 gene, which encodes a metalloreductase responsible for converting stable, insoluble ferric iron into its more reactive, soluble ferrous form. In the discovery sample, patients whose methylation at this site fell into the highest trajectory group faced dramatically worse odds than those in the lowest group. The odds of an unfavorable outcome on the Glasgow Outcome Scale — a measure of functional recovery — were 11.7 times higher at three months and 15.6 times higher at twelve months. The odds of death were 19.1 times higher at three months and 12.8 times higher at twelve months, all meeting the study’s empirical thresholds for statistical significance.

What makes these findings compelling is that they replicated. In the independent sample of 100 patients, the same trajectory structure emerged, and the associations held: an 8.2-fold increase in odds of unfavorable recovery at three months, a 6.3-fold increase at twelve months, and a 2.3-fold increase in the odds of death at three months, with a suggestive association for death at twelve months. The effect was additive — with each step up in methylation trajectory group, the proportion of patients with poor outcomes climbed steeply. In the replication sample, not a single patient in the lowest methylation group had an unfavorable Glasgow Outcome Scale score at three months, compared with 61.5 percent of those in the highest group.

Biologically, the result makes unsettling sense. STEAP3 is essential for normal iron handling — its deficiency causes microcytic anemia with iron overload — but in the aftermath of a hemorrhage, its activity may become a liability. By converting ferric iron to the ferrous form, STEAP3 potentially increases the pool of reactive iron available to drive lipid peroxidation, the oxidative destruction of cell membranes that lies at the heart of ferroptosis. Animal studies support this logic: elevated iron after experimental subarachnoid hemorrhage raises lipid peroxide levels, and ferrostatin-1, a lipophilic antioxidant that shields membranes from lipid oxidation, reduces free iron and prevents both ferroptosis and early brain injury in those models. STEAP3 has also been linked to p53-mediated apoptosis and to immune function, and it is the only member of the STEAP family highly expressed in macrophages. Intriguingly, previous work found STEAP3 to be more than 2.5-fold overexpressed in the walls of unruptured intracranial aneurysms compared with control artery tissue.

Two other candidate genes told a more cautionary tale. A methylation site in the amyloid precursor protein gene, APP, which participates in iron efflux by stabilizing ferritin and ferroportin, showed promising associations with favorable outcomes and survival in the discovery phase, but the trajectory structure failed to replicate. A dynamic site in the inflammatory cytokine gene TNF, which sits within a hotspot of outcome-associated methylation positions, also failed replication. Only the STEAP3 signal survived independent testing, underscoring how easily early findings in epigenetic studies can dissolve under scrutiny — and how valuable embedded replication is for distinguishing robust signals from statistical noise.

Perhaps the most clinically surprising observation was that methylation at the STEAP3 site barely changed over the 13-day monitoring window. The trajectories were largely flat, suggesting that a single early measurement might capture the same predictive information as repeated sampling. Post hoc cross-sectional analyses confirmed persistent associations between continuous methylation levels and long-term outcomes at multiple time points from days 3 through 11 after the hemorrhage. The association also survived adjustment for cell type heterogeneity — the shifting mixture of blood, brain, and vessel cells that contaminates cerebrospinal fluid after a bleed — which strengthens the case that the signal reflects genuine biology rather than changing cellular composition. One caveat: methylation at the site showed only a small to moderate correlation between blood and cerebrospinal fluid, indicating that a simple blood test would likely not substitute for the brain-proximal fluid.

The researchers are careful to frame this as a first step rather than a finished clinical tool. The study population was predominantly composed of participants who self-identified as White, limiting generalizability, and the candidate gene approach, though rigorously corrected for multiple testing, may have missed signals in genes that could not be reliably modeled, including two iron homeostasis genes on the X chromosome that were excluded during quality control. Functional studies are needed to determine whether methylation at cg25713625 actually alters STEAP3 activity and ferroptosis in the injured brain, or whether it merely marks some deeper vulnerability. Still, the prospect is tantalizing: a methylation marker measurable in fluid already being drained from patients could one day help clinicians identify, within days of a devastating hemorrhage, who needs the most aggressive monitoring — and it could spotlight iron metabolism, and ferroptosis in particular, as a target for therapies that might finally move the needle on outcomes in this brutal disease.

Subject of Research: DNA methylation of the iron homeostasis gene STEAP3 in cerebrospinal fluid as a predictor of outcomes after aneurysmal subarachnoid hemorrhage

Article Title: Iron homeostasis pathway DNA methylation trajectories reveal a role for STEAP3 metalloreductase in patient outcomes after aneurysmal subarachnoid hemorrhage

Article References: Heinsberg, L. W., Weeks, D. E., Alexander, S. A., Minster, R. L., Sherwood, P. R., Poloyac, S. M., Deslouches, S., Crago, E. A., & Conley, Y. P. (2021). Iron homeostasis pathway DNA methylation trajectories reveal a role for STEAP3 metalloreductase in patient outcomes after aneurysmal subarachnoid hemorrhage. Epigenetics Communications, 1(1), Article 4. https://doi.org/10.1186/s43682-021-00003-5

Image Credits: AI Generated

DOI: 10.1186/s43682-021-00003-5

Keywords: aneurysmal subarachnoid hemorrhage, DNA methylation, STEAP3, iron homeostasis, ferroptosis, cerebrospinal fluid, epigenetics, biomarker, stroke, group-based trajectory analysis, lipid peroxidation, Glasgow Outcome Scale