Neurofilament light chain predicts cognitive decline after cardiac arrest where the standard test fails

European Society of Cardiology

When a person's heart stops and is restarted, the brain sits at the center of every clinical decision that follows. How much damage occurred during the minutes without blood flow? Will the patient recover cognitive function, and to what degree? Clinicians need answers to these questions quickly, often within the first days of hospitalization, because those answers shape treatment intensity, rehabilitation planning, and the conversations families dread most.

The current standard tool for estimating brain injury after cardiac arrest is a blood test for neuron-specific enolase, an enzyme released when neurons are damaged. It is widely used but widely criticized. Factors unrelated to brain damage - hemolysis from blood draws, certain cancers, even the handling of the blood sample - can elevate neuron-specific enolase levels and produce misleading readings.

A study presented at ESC Acute CardioVascular Care 2026 in Lisbon suggests there may be a better option. Neurofilament light chain, a structural protein released when nerve cell axons are injured, outperformed neuron-specific enolase in predicting long-term cognitive outcomes after out-of-hospital cardiac arrest.

Two biomarkers, one time point, different answers

Martin Meyer of Rigshospitalet - Copenhagen University and his colleagues analyzed blood samples from participants in the BOX trial (Blood Pressure and Oxygenation Targets after Cardiac Arrest), a major randomized study of resuscitated patients who were comatose upon hospital admission. Both neurofilament light chain and neuron-specific enolase were measured in samples taken at 48 hours after cardiac arrest - a clinically practical time point when patients are still in intensive care and management decisions are being made.

Cognitive function was assessed months later using the Montreal Cognitive Assessment (MoCA), a validated screening tool that measures attention, memory, language, visuospatial skills, and executive function. The analysis focused on survivors who had both biomarker measurements and follow-up MoCA scores available.

The central finding was clear: neurofilament light chain levels at 48 hours were inversely correlated with MoCA scores. Higher blood levels of the protein - indicating more severe axonal injury - predicted worse cognitive function months down the line. Neuron-specific enolase levels at the same time point showed no such correlation.

Why one protein tracks brain damage better than another

The biological explanation makes intuitive sense. Neuron-specific enolase is a glycolytic enzyme found in neurons but also present in red blood cells, platelets, and neuroendocrine cells throughout the body. Its levels in blood can rise for reasons that have nothing to do with brain injury. A slightly hemolyzed blood sample - something that happens routinely in busy emergency departments - can produce a falsely elevated reading.

Neurofilament light chain, by contrast, is a structural component of neuronal axons. It enters the bloodstream primarily when axons are damaged. While not perfectly brain-specific - peripheral nerve injury can also release it - it is far more tightly linked to central nervous system damage than neuron-specific enolase. Its blood levels tend to rise and fall in proportion to the actual extent of neuronal injury.

This specificity is exactly what clinicians need when trying to distinguish patients with severe brain injury from those with milder damage who may recover cognitive function with rehabilitation.

What early detection could change in clinical practice

Meyer outlined several potential clinical applications if the findings hold up in larger studies. Early neurofilament light chain measurement could help optimize the decision-making process for additional tests and brain imaging. Rather than ordering expensive MRI scans or EEG monitoring on every cardiac arrest survivor, clinicians could use the blood test to identify which patients most need detailed neurological workup.

Rehabilitation targeting stands to benefit as well. Patients identified early as having significant brain injury could be enrolled in cognitive rehabilitation programs sooner, potentially improving outcomes. Conversely, patients with low neurofilament light chain levels - suggesting limited brain damage - could be reassured earlier, reducing the period of uncertainty that families find so distressing.

Perhaps most importantly, the biomarker could improve prognostic conversations. Telling a family what to expect after a cardiac arrest is one of the most difficult tasks in critical care medicine. A biomarker that reliably correlates with long-term cognitive function gives clinicians a more objective basis for those conversations than the clinical judgment and imperfect tests currently available.

The sample size and validation problem

The study has limitations that must be stated directly. The analysis involved a subset of BOX trial participants who had both biomarker measurements and MoCA follow-up data. The exact sample size of this subset, drawn from a trial originally designed for different primary endpoints, may not be large enough to establish definitive cutoff values for clinical use.

Neurofilament light chain assays are not yet standardized across laboratories. Different testing platforms may produce different absolute values, making it difficult to establish universal thresholds that clinicians could apply consistently. Standardization efforts are underway but not complete.

The BOX trial enrolled patients who were comatose on hospital admission - the most severely affected cardiac arrest survivors. Whether neurofilament light chain performs equally well in patients who regain consciousness quickly, or in those resuscitated from in-hospital cardiac arrest (where the circumstances and duration of brain ischemia may differ), is unknown.

The MoCA, while validated and widely used, is a screening tool rather than a comprehensive neuropsychological assessment. Subtle cognitive deficits in specific domains might be missed, and the relationship between biomarker levels and more granular cognitive outcomes remains to be explored.

From conference presentation to clinical guideline

This study adds to a growing body of evidence favoring neurofilament light chain over neuron-specific enolase for neuroprognostication after cardiac arrest. Previous studies have shown better diagnostic performance for predicting mortality and severe disability. The current work extends this to a different outcome - long-term cognitive function as measured by a standardized assessment - and finds the same advantage.

The path from a conference presentation to a change in clinical guidelines involves validation in larger, prospective studies; standardization of assay methods; establishment of clinically actionable cutoff values; and integration into existing prognostication algorithms that incorporate multiple data sources. None of these steps is trivial.

But the direction of the evidence is consistent. Neuron-specific enolase, despite its long history and widespread use, may be measuring the wrong thing - or measuring the right thing too imprecisely to guide the decisions that matter most. A structural protein that reflects actual axonal damage, measured at a time when clinical decisions are being made, represents a more rational approach to a problem that affects thousands of families every year.