Same Cells, Opposite Effects
You might expect that if an immune marker is associated with poorer heart regulation at rest, it would be even worse during the stress of exercise. But a study of 83 older adults in Chile found the opposite. Researchers led by Matías Castillo-Aguilar at the Universidad de Magallanes measured immune cell levels from blood samples and then tracked heart rate variability (HRV) while participants rested, exercised, and recovered [1].
What they found was unexpected: several immune cell types that correlated with worse heart regulation at rest showed attenuated or even reversed associations during exercise. Cells that looked like liabilities in a resting blood panel appeared to behave differently — sometimes more favourably — once the body was under physical stress. This inversion pattern appeared across multiple cell types and heart rate measures, suggesting it's a real feature of how the immune and nervous systems interact.
What the Study Actually Did
The idea that the immune and nervous systems talk to each other isn't new. The nervous system can modulate immune activity through pathways like the vagus nerve's inflammatory reflex [2], and immune cells can in turn influence neural signalling through cytokines and direct contact with nerve endings [3]. HRV itself — the beat-to-beat variation in heart rate that reflects autonomic flexibility — is increasingly recognised as a functional marker of how well that regulatory system holds up with age [4]. What's been missing is evidence of how this crosstalk behaves dynamically during physical stress, rather than just at rest.
To test that, the researchers recruited 83 older adults (mean age ~71, roughly two-thirds women) from Punta Arenas, Chile. Each participant had their baseline immune profile measured via blood draw and flow cytometry, capturing counts of total lymphocytes, B-cells, T-cells, the CD4+/CD8+ T-cell ratio, and Natural Killer (NK) cells — including the CD56-dim and CD56-bright NK subpopulations.
Participants then wore a Polar H10 heart rate monitor while sitting at rest, performing a standardised Two-Minute Step Test (marching in place at a set knee height), and resting again for five minutes of recovery. The researchers used Bayesian statistical models to test whether the relationship between baseline immune cell counts and HRV changed across the three phases — rest, exercise, and recovery — while accounting for age, sex, and body composition.
The Inversion Pattern
The study's central finding is a recurring paradox: immune cells that looked like bad news at rest often flipped their association during exercise.
Take total lymphocyte counts. At rest, participants with higher counts tended to have a faster heart rate — typically a sign of less parasympathetic ("rest and digest") control. You'd expect that pattern to persist or worsen under the stress of exercise. Instead, higher lymphocyte counts were associated with a relatively slower heart rate during exertion than the resting data would predict. The same kind of reversal showed up in frequency-domain HRV measures, which capture parasympathetic activity from a different angle.
B-cells told the story in reverse. Higher B-cell counts were associated with stronger parasympathetic tone at rest — on paper, a healthier-looking profile. But during exercise, that favourable association flipped: participants with more B-cells showed a sharper withdrawal of parasympathetic control. A good resting profile didn't translate into a smoother response under stress.
Then there's the CD4+/CD8+ T-cell ratio, a marker clinicians sometimes use as a rough gauge of immune aging. A higher ratio was linked to lower HRV at rest across several measures — less variability, less parasympathetic activity. During exercise, those negative associations weakened or reversed. In other words, the ratio that looked worst at rest didn't necessarily predict the worst exercise response.
In each case, the pattern is the same: what the immune system appears to be doing at rest doesn't reliably tell you what it will do when the body is under physical stress.
NK Cells Tell the Clearest Story
Natural Killer cells gave the sharpest version of this pattern, partly because the researchers could separate them into two functionally distinct subtypes.
CD56-dim NK cells are the more abundant type in older adults. They're primarily cytotoxic — built to kill infected or abnormal cells. CD56-bright NK cells are rarer and play more of a regulatory, signalling role. At rest, these two subtypes pointed in opposite directions: higher CD56-dim counts were linked to a slower heart rate (more parasympathetic influence), while higher CD56-bright counts were linked to a faster one.
During exercise, the divergence widened. Participants with more CD56-dim cells maintained a higher parasympathetic index during exertion — as though these cells helped buffer the body's sympathetic "fight or flight" surge. Participants with more CD56-bright cells showed the opposite: a lower parasympathetic index, suggesting less buffering and a more pronounced sympathetic takeover.
The picture isn't as simple as "dim = good, bright = bad." CD56-dim cells were also associated with higher sympathetic and stress indices during exercise, meaning they correlated with greater overall autonomic activation — not just parasympathetic protection. The authors suggest this may reflect CD56-dim cells supporting a more responsive, flexible autonomic system rather than simply a calmer one. But the mechanisms remain speculative — this study measured associations, not the biological pathways behind them.
What This Means — and What It Doesn't
The practical implication is straightforward: if your baseline immune profile shapes how your autonomic nervous system responds to exercise, then two people of the same age and fitness level might experience the same workout in genuinely different ways at a physiological level. That could help explain the wide variability in how older adults respond to exercise — variability that fitness level and age alone don't fully account for.
The authors suggest this could eventually inform more personalised exercise approaches — for example, using clinically accessible immune markers to anticipate how someone's heart might handle physical stress. That's a reasonable direction for future research, but it's worth flagging what the study doesn't establish. This was a cross-sectional observation — immune profiles and HRV were measured at the same time, so causation can't be inferred. The exercise challenge was a Two-Minute Step Test, not a high-intensity or prolonged session. And the sample, while reasonable for this kind of exploratory work, was 83 people from a single region. The team also didn't measure cytokines or signalling molecules directly, so the biological mechanisms behind these associations remain speculative.
Still, the consistency of the inversion pattern across cell types and HRV domains is striking. Castillo-Aguilar et al. propose several plausible mechanisms, and the study makes a strong case that the immune-autonomic interface deserves more attention in aging and exercise research. For now, the takeaway isn't a new training protocol. It's a shift in perspective: the immune system isn't just a background player in how your body handles physical stress. It's actively shaping the response — and not always in the direction you'd expect from resting data alone.
References
- Castillo-Aguilar, M. et al. "The Immune-Autonomic Interface in Aging: Baseline Immune Profile Shapes Cardiac Autonomic Response to Exercise," Aging Cell 25 (2026) e70428. https://doi.org/10.1111/acel.70428
- Jin, H. et al. "A Body–Brain Circuit That Regulates Body Inflammatory Responses," Nature 630 (2024) 695–703. https://doi.org/10.1038/s41586-024-07469-y
- Wheeler, M. A. and F. J. Quintana. "The Neuroimmune Connectome in Health and Disease," Nature 638 (2025) 333–342. https://doi.org/10.1038/s41586-024-08474-x
- Olivieri, F. et al. "Heart Rate Variability and Autonomic Nervous System Imbalance: Potential Biomarkers and Detectable Hallmarks of Aging and Inflammaging," Ageing Research Reviews 101 (2024) 102521. https://doi.org/10.1016/j.arr.2024.102521