Four Weeks of Aerobic Exercise Fails to Lift BDNF Levels After Stroke, Rigorous Trial Finds

One of the most compelling ideas in modern stroke rehabilitation is that the weeks immediately following a stroke represent a window of heightened neuroplasticity, a period when the injured brain is unusually receptive to interventions that might steer its recovery. Central to this idea is brain-derived neurotrophic factor, or BDNF, a molecule that supports neuronal survival, the formation of new synapses, dendritic spine growth, and long-term potentiation, the cellular substrate of learning. In animal models, boosting BDNF reliably enhances motor recovery after stroke. In healthy humans, a single bout of aerobic exercise raises circulating BDNF, and much of that increase is thought to originate in the brain itself. It seemed a short logical step to assume that prescribing aerobic exercise to stroke patients would raise their BDNF levels and, through that mechanism, promote biological repair.

A new analysis of one of the most rigorously designed trials in this field now complicates that assumption. In a pre-specified secondary biomarker analysis of the PHYS-STROKE trial, published in Frontiers in Stroke, researchers led by Torsten Rackoll of Charité—University Medical Center Berlin examined whether four weeks of moderate-intensity treadmill training, started in the early subacute phase after stroke, could raise serum BDNF compared with an active control. The answer, measured carefully across six months of follow-up, was no.

PHYS-STROKE was a multicentre, randomized, assessor-blinded trial conducted across seven rehabilitation clinics and early-rehabilitation wards in the Berlin area between 2013 and 2017. Two hundred participants, enrolled between 5 and 45 days after an imaging-confirmed ischemic or hemorrhagic stroke, were allocated equally to two interventions. The experimental group completed up to 20 supervised sessions of body-weight-supported treadmill walking, with intensity prescribed at 55 to 65 percent of age-predicted maximal heart rate and monitored continuously through heart-rate sensors. The control group received progressive muscle relaxation based on the Jacobson method, matched in session frequency and duration but deliberately designed to avoid cardiovascular activation. Both groups continued their standard rehabilitation, which in Germany includes daily physiotherapy and, as needed, occupational, speech, and neuropsychological therapy.

The participants were genuinely impaired. The median National Institutes of Health Stroke Scale score was 8, and the median Barthel Index, a measure of functional independence, was 50, indicating moderate to severe disability. Mean serum BDNF at baseline was 17.2 nanograms per milliliter, and mild depressive symptoms were common. Fasting blood samples were drawn at baseline, immediately after the four-week intervention, and at three and six months after stroke. Serum mature BDNF was quantified using a Luminex-based multiplex immunoassay targeting the mature BDNF region, with samples stored at minus 80 degrees Celsius and analyzed in three calibrated batches.

The statistical approach was appropriately conservative. The primary analysis used linear mixed-effects models for repeated measures, with treatment group, time, and their interaction as fixed effects, adjusted for baseline BDNF, age, sex, stroke severity, platelet count, depressive symptoms measured with the CES-D scale, pre-stroke smoking status, and study center. Because platelets store and release BDNF during coagulation, platelet count was included as a covariate. Missing data, which were substantial—156 of 200 participants had at least one follow-up measurement, and only 118 had complete data across all time points—were handled with multiple imputation by chained equations across 1,000 imputed datasets, with complete-case and change-score analyses run as sensitivity checks.

The headline result was unambiguous. Serum BDNF rose modestly across the whole cohort, with most of the increase occurring between baseline and three months after stroke, and then remained stable. But at six months, the estimated mean concentration was virtually identical between groups: 24.6 nanograms per milliliter in the aerobic training group versus 24.3 in the relaxation group, an adjusted difference of minus 1.15 nanograms per milliliter with a p-value of 0.74. Crucially, the same null result appeared immediately after the intervention period, when a biological effect of four weeks of training would have been most detectable. The convergence of initially divergent baseline values—BDNF was higher in the training group at the start—may partly reflect regression to the mean rather than any treatment effect, a caveat the authors acknowledge openly.

Exploratory analyses probing whether depressive symptoms, intervention dose, or sex modified the treatment effect found no robust signals. The one nominally significant subgroup finding—higher six-month BDNF among women than men in the relaxation group, but not in the training group—emerged without correction for multiple testing, in small subgroups, and without a clear biological rationale. The researchers are explicit that this should be treated as hypothesis-generating at most and likely reflects chance or non-specific recovery rather than a distinct effect of relaxation therapy.

Why did the exercise fail to move BDNF when the same intervention reliably raises it in healthy people? The authors point to several plausible explanations. Stroke triggers systemic inflammation, autonomic dysregulation, altered platelet dynamics, and often involves medications and comorbidities that all shape circulating BDNF. Reduced exercise capacity in moderately to severely affected patients may also have limited the physiological stimulus. Perhaps most importantly, the intervention was deliberately pragmatic and safety-oriented: moderate-intensity continuous training was the standard, safe approach for early subacute rehabilitation when the trial was designed, not a protocol optimized to maximize neurotrophin release. Recent evidence, including a network meta-analysis in adults, suggests that acute BDNF responses are more consistent after high-intensity interval exercise than after moderate continuous exercise, and the moderate target heart rates used here may simply have been insufficient to trigger meaningful release.

The study also carries an important methodological message for the field. Serum BDNF is an imperfect proxy for brain BDNF signaling. It is heavily influenced by platelet storage and release, by clotting and centrifugation protocols, by circadian rhythms, and by metabolic and pharmacological factors, and serum and plasma measurements are not interchangeable. Recent methodological reviews have highlighted wide variability in how stroke studies handle these pre-analytical variables, complicating comparisons across trials. The trial also did not measure the BDNF Val66Met genotype, which influences activity-dependent BDNF release and could contribute to inter-individual variability. And because blood was not sampled during or immediately after individual exercise sessions, transient within-session BDNF spikes cannot be ruled out.

The authors are careful about what their findings do and do not mean. A null result on peripheral BDNF does not exclude exercise-related effects on central neuroplasticity or on functional recovery through other biological pathways, nor does it rule out acute, higher-intensity, genotype-dependent, or subgroup-specific neurotrophin responses. What it does argue against is the notion that a clinically feasible, four-week, moderate-intensity aerobic programme, delivered in the early subacute window to a heterogeneous and substantially impaired population, produces a robust and sustained elevation of circulating BDNF. For clinicians and researchers, the implication is that future studies of exercise-induced neuroplasticity after stroke should consider denser biomarker sampling around exercise sessions, individualized and potentially higher exercise intensity prescriptions, and biological stratification of patients, rather than assuming that any aerobic programme will uniformly engage the BDNF pathway. The sensitive window after stroke may indeed be real, but the molecular levers required to pull it appear more selective, and more demanding, than the field has often assumed.

Subject of Research: Effects of a four-week moderate-intensity aerobic exercise intervention on serum brain-derived neurotrophic factor in subacute stroke patients

Article Title: Serum BDNF in subacute stroke: effects of a four-week aerobic exercise intervention in the PHYS-STROKE trial

Article References: Rackoll, T., Doppelbauer, L., Neumann, K., & Nave, A. H. (2026). Serum BDNF in subacute stroke: effects of a four-week aerobic exercise intervention in the PHYS-STROKE trial. Frontiers in Stroke, 5, Article 1897859. https://doi.org/10.3389/fstro.2026.1897859

Image Credits: AI Generated

DOI: 10.3389/fstro.2026.1897859

Keywords: stroke, BDNF, aerobic exercise, neuroplasticity, stroke rehabilitation, PHYS-STROKE trial, randomized controlled trial, neurotrophins, Serum, subacute, effects, four-week