The Brain's Heartbeat Signal Is Not a Biomarker Yet: A Cautionary Meta-Analysis
- Random-effects multilevel meta-analysis and systematic parameter review from the University of South Australia, with the Laureate Institute for Brain Research (Tulsa) and UCLA; the target was the heartbeat evoked potential (HEP), an EEG index of cortical processing of the heartbeat and the leading candidate neural marker of cardiac interoception; preregistered on OSF.
- Across clinical subgroups, HEP amplitude showed no difference between clinical conditions and healthy comparisons.
- Large negative effect estimates appeared for neurological and cardiovascular disease, but these rested on a small number of studies and were flagged for caution.
- Clinical category only partially explained the wide variability in HEP parameters (latency windows, reference electrodes, scalp topographies); the authors conclude no firm claim about HEP as a clinical biomarker can currently be made, given substantial heterogeneity and widespread methodological inconsistency.
Interoception has a favourite neural marker. The heartbeat evoked potential, the small cortical response time-locked to each beat, gets cited as the objective handle on how the brain processes the body, and increasingly as a candidate biomarker for psychiatric and neurological conditions. This meta-analysis is the field checking its own enthusiasm, and the answer is a useful no-not-yet.
What the data shows
Pooling HEP studies across clinical populations, the headline comparison, patients versus healthy controls, came out flat. No reliable amplitude difference. Where large effects did appear, in neurological and cardiovascular disease, they came from a handful of studies and the authors explicitly warn against reading them as established. The more instructive finding is methodological. HEP is not measured the same way twice: latency windows, reference electrodes and scalp topographies vary widely across labs, and clinical category explained only part of that variance. In other words, much of what looks like signal in the individual-study literature may be measurement choices, not biology. The authors' conclusion is deliberately deflationary: standardisation has to come before the HEP can be called a biomarker of anything.
This is a negative result done well, preregistered, multilevel, transparent about the thinness of some subgroups.
For your practice
Most clinicians will never record an EEG, so the value here is epistemic hygiene. Interoception is a real and clinically useful construct, but the market around it, apps, devices and training programmes, increasingly leans on "objective" neural and physiological markers to sell precision it does not have. When a product or a paper claims to measure your patient's interoceptive brain signal, this meta-analysis is the reference for scepticism: the flagship marker does not yet separate clinical from healthy groups and is not measured consistently enough to compare across settings. Keep using interoceptive interventions on the strength of behavioural and clinical outcomes. Do not anchor case formulation or progress claims to a single neurophysiological number.
The heartbeat evoked potential is the field's favourite objective marker of interoception, and it cannot yet tell clinical brains from healthy ones.
As a meta-analysis it inherits the heterogeneity and small subgroup sizes of the underlying literature, which is part of its own argument; several disease-specific estimates rest on very few studies.