Stroke and its consequences: protocol and pilot data of the observational Berlin Long-term Observation of Vascular Events (BeLOVE) stroke stratum

Clinical Question (PICO)

Population: Adults admitted to three Charité, Universitätsmedizin Berlin campuses with acute ischemic stroke, transient ischemic attack (TIA), non-traumatic intracerebral hemorrhage (ICH), retinal stroke, or retinal TIA. Intervention/exposure: There was no assigned intervention; BeLOVE prospectively collects deep clinical, imaging, cardiac, cognitive, patient-reported, biospecimen, and multi-omics phenotypes. Comparator: None prespecified. Outcomes: First major adverse cerebrovascular and cardiovascular event (MACE) over follow-up, plus recurrent stroke, myocardial infarction, heart-failure hospitalization, death, cognition, pain, depression, seizures, disability, and quality of life. Prespecified analyses are descriptive in this report; future analyses will examine exposure–outcome associations and potential risk strata over up to 10 years.

Bottom Line

BeLOVE is a prospective observational platform rather than a treatment trial. In its first study phase (758 participants), the 2-year cumulative incidence of first MACE was 10.7% (95% CI 8.5%–13.2%); first ischemic stroke occurred in 6.6% (95% CI 4.9%–8.6%) and all-cause death in 4.0% (95% CI 2.7%–5.6%). The study establishes a richly phenotyped brain–heart–behavior cohort, but does not show that MRI, multi-omics, or any other component improves outcomes.

Design

Trial type: Prospective, longitudinal, observational, hospital-based cohort; open-label by design, with endpoint assessment supervised by an adjudication committee.
N: 758 in the reported first study phase; subsequent recruitment of the stroke stratum reached 1,584 participants by 31 December 2025.
- Acute ischemic stroke: 574/758 (75.7%) in the study text; the baseline table reports an ischemic-stroke subgroup of 573.
- TIA: 125/758 (16.5%).
- ICH: 30/758 (4.0%).
- Retinal stroke: 20/758 (2.6%); retinal TIA: 9/758 (1.2%).
Randomization: None; no assigned intervention or comparator.
Setting: Three Charité campuses in Berlin, Germany: Campus Mitte, Virchow Klinikum, and Campus Benjamin Franklin.
Enrollment: BeLOVE began 18 July 2017; the reported first study phase ran through 31 December 2020. The stroke subcohort closed recruitment on 31 December 2025.
Mean follow-up: Not reported as a mean; the pilot outcome report uses 2-year follow-up, with planned follow-up to 10 years and structured contacts every 6 months after the initial visits.
Analysis: Descriptive frequencies and medians; Aalen–Johansen cumulative-incidence functions with 95% CIs and competing-risk handling. Planned exposure analyses include time-to-event regression, such as Cox models, adjusted for potential confounders.
Primary outcome: First MACE after study inclusion: ischemic or hemorrhagic stroke, acute myocardial infarction, hospitalization for new or worsening heart failure, or vascular death.

Population

Inclusion Criteria

Adults aged 18 years or older, of any sex.
Acute ischemic stroke, defined by an acute neurological deficit lasting more than 24 hours or a shorter deficit with a corresponding acute ischemic lesion on MRI or CT; selected retinal artery occlusion cases could be included.
Non-traumatic ICH with an acute deficit and fresh hemorrhage on MRI or CT.
TIA with clinical restitution within 24 hours, no acute lesion on imaging, and neurologist verification or an ABCD2 score of at least 3; amaurosis fugax was included.
Written informed consent before participation.

Exclusion Criteria

Pregnancy or breastfeeding.
No health insurance.
Life expectancy under 6 months from a non-cardiovascular cause.
Active cancer or history of organ transplantation.
For ischemic stroke lasting more than 24 hours, no ischemic or hemorrhagic lesion on MRI when the imaging criteria were required.

Baseline Characteristics (overall or representative arm)

Median age 69 years (IQR 59–78); 37% female.
Median admission mRS 1 (IQR 0–3); 69% had mRS 0–2 at admission. The median admission NIHSS was 2 (IQR 1–4) where reported.
Vascular risk factors included hypercholesterolemia in 93%, hypertension in 80%, current or former smoking in 66%, diabetes in 23%, atrial fibrillation in 22%, prior ischemic stroke in 19%, and coronary artery disease in 16%.
Among the ischemic-stroke subgroup, 27% received systemic thrombolysis and 8% underwent mechanical recanalization.

Interventions

BeLOVE phenotyping: Acute assessment during days 0–7, telephone assessments at days 30 and 60, and deep phenotyping at day 90, including clinical variables, mRS, NIHSS, frailty, ECG, echocardiography, cerebral and cardiac MRI, selected EEG, MoCA, CANTAB, PHQ-8, PROMIS-29, EQ-5D-5L, SSQoL, painDETECT, blood and urine sampling, proteomics, metabolomics, genome-wide association studies, immunophenotyping, and clonal hematopoiesis analysis.
Follow-up: Structured telephone visits and/or repeat deep phenotyping every 6 months or annually as scheduled, recording clinical events, disability, cognition, patient-reported outcomes, and quality of life for up to 10 years.
Usual care: Acute stroke and secondary-prevention treatment were not assigned by the study and were delivered in the clinical setting.

Outcomes

Primary Outcome (first MACE, with non-vascular death treated as a competing event):

At 2 years, 73 participants had a MACE; cumulative incidence was 10.7% (95% CI 8.5%–13.2%). No prespecified treatment or exposure subgroups were reported with HRs, ORs, RRs, or P values in this pilot report.
Among the first events that contributed to the composite, strokes accounted for 46 events (60%). The report emphasizes this as a descriptive event composition, not a causal comparison.
At the beginning of observation, 758 participants were at risk; 558 remained at risk for MACE at 2 years. Eighteen non-vascular deaths were competing events, and 108 participants were censored, largely after withdrawal of consent.

Secondary Outcomes:

First ischemic stroke: 45 events; cumulative incidence 6.6% (95% CI 4.9%–8.6%).
First hemorrhagic stroke: 2 events; cumulative incidence 0.3% (95% CI 0.1%–1.0%).
First myocardial infarction: 10 events; cumulative incidence 1.5% (95% CI 0.8%–2.6%).
First hospitalization for heart failure: 12 events; cumulative incidence 1.8% (95% CI 1.0%–3.0%).
All-cause death: 27 events; cumulative incidence 4.0% (95% CI 2.7%–5.6%).
Planned secondary domains include cognition, pain, depression, seizures, disability, and generic and stroke-specific quality of life; this report presents the cohort design rather than their longitudinal results.

Adverse Events / Safety:

There was no randomized intervention and therefore no treatment-safety comparison. During 2-year follow-up, two participants had hemorrhagic stroke and 27 died from any cause; these are outcome events, not adverse-event rates attributable to a study treatment.

Figures

Cumulative incidence plots for MACE and secondary endpoints — *Figure 1.* Cumulative incidence function plots for first MACE with non-vascular death as the competing event and for secondary endpoints including first stroke, first ischemic stroke, all-cause death, myocardial infarction, and first hospitalization for heart failure.

Criticisms

This is a selected, consent-based cohort: aphasic, frail, moribund, and otherwise unable or unwilling to complete extensive visits are under-represented. Generalizing the 10.7% estimate to all stroke survivors will likely miscalibrate risk.
The report is a protocol plus pilot event description, not a validation of a prognostic model. Deep MRI, cardiac testing, and multi-omics are hypotheses and measurement resources, not proven screening tools.
There is no randomized comparator, no causal estimate of a phenotyping strategy, and no intervention pathway tested. A future Cox association cannot by itself prove that acting on a biomarker improves outcomes.
Event denominators are affected by censoring: 108 participants were censored by 2 years, largely after withdrawal of consent. The study reports cumulative incidence with competing risks, but missingness and differential follow-up remain important for later domain-specific outcomes.
The first-phase cohort was predominantly mild to moderate: median NIHSS was 2 and 69% had mRS 0–2 at admission. This improves feasibility but limits external validity to severe stroke, profound aphasia, early death, and patients with active cancer or transplant history.

Funding

The work was substantially supported by the Berlin Institute of Health (BIH) at Charité – Universitätsmedizin Berlin. The BIH BeLOVE Office participated in study design and coordinated recruitment, data collection, and data management. The study was registered in the German Clinical Trials Register as DRKS00023323 on 4 November 2020. Ethics approval was granted by the Charité – Universitätsmedizin Berlin ethics committee (EA1/066/17); participants gave informed consent.

The paper

Authors. Nolte et al.
Title. Stroke and its consequences: protocol and pilot data of the observational Berlin Long-term Observation of Vascular Events (BeLOVE) stroke stratum.
Journal. BMJ Neurology Open.
Year. 2026.
DOI. 10.1136/bmjno-2026-001727
PMCID. PMC13330884

Deep Dive — click to expand

What this is

BeLOVE is an attempt to stop treating stroke as an isolated brain event and instead follow the patient as a connected brain–heart–immune–behavior system. The first 758 participants provide a useful reality check: 2-year cumulative incidence was 10.7% for first MACE, 6.6% for ischemic stroke, and 4.0% for death. Those numbers establish the event burden that a future prediction and prevention programme must handle; they do not yet establish that the proposed MRI, cardiac testing, or multi-omics measurements can change care.

1. Shadow Audit

The paper’s most important result is easy to overread because the cohort is presented as a sophisticated deep-phenotyping platform. The observed 10.7% MACE estimate came from only 73 events, with 558 participants still at risk at 2 years and 108 censored, largely after withdrawal of consent. The composite was also not a generic measure of “everything bad after stroke”: strokes supplied 46 of the MACE events, while clinically covert strokes were excluded from the endpoint. The long-term 10-year promise and the brain–body–brain mechanism are the study’s rationale and infrastructure, not findings demonstrated by this report.

2. Inversion Engine

To invert the paper into a claim that universal deep phenotyping should be adopted now, the measurements would have to produce more than associations: they would need a validated risk model that materially reclassifies patients and a tested management response that improves outcomes. The pilot cannot quantify that treatment effect. A concrete numerical boundary is visible for the event burden itself: any future claim that the 2-year risk is materially below 10% would have to move the point estimate below the current 10.0% threshold, yet the observed estimate is 10.7% with a 95% CI spanning 8.5%–13.2%; the interval therefore does not support a precise threshold claim. The inversion is consequently not “find a different p value,” but demonstrate incremental predictive value and clinical net benefit in an independent cohort, followed by an intervention study.

3. Second-Order Catalyst

The first practice change is not a multi-omics order set. It is a scheduled 90-day post-stroke review that treats recurrent vascular events, cardiac disease, cognition, mood, pain, seizures, disability, and patient-reported quality of life as one follow-up problem. A comprehensive stroke service with existing MRI and research infrastructure—such as the Charité sites—can implement that workflow first, then use six-month contacts to catch events and symptoms that a discharge-only model misses. The speed should be measured in months for the protocol and data capture, not assumed to be an immediate population-wide biomarker rollout.

4. Asymmetric Leverage

The leverage is in the denominator. Seventy-three MACE events among 758 people are too few to validate a large biomarker panel, but 758 deeply characterized participants followed repeatedly can reveal which domains deserve a narrower, cheaper test in the next cohort. The asymmetric payoff is therefore cohort harmonization: a common data-element set plus imaging, PROMs, and endpoint adjudication can make modest predictors useful at scale. Conversely, repeating every expensive test for every survivor would spend a large numerator of staff time to address a relatively small event count without proof that the information changes treatment.

5. Paradigm Destroyer

This study kills the reflex that a technically successful thrombectomy or a good discharge mRS closes the stroke episode. It supports a protocol update for tomorrow morning: at 90 days, review recurrent vascular risk and cardiac symptoms alongside cognition, mood, pain, seizures, disability, and quality of life; repeat structured contact at least every 6 months when feasible. It does not support ordering cardiac MRI, cerebral MRI, or multi-omics indiscriminately, because BeLOVE has not tested that strategy against usual care.

MVP — Minimum Viable Proof

The minimum bedside-relevant statement is: in the first 758 selected BeLOVE participants, first MACE at 2 years occurred with a cumulative incidence of 10.7% (95% CI 8.5%–13.2%), including 6.6% for ischemic stroke and 4.0% for all-cause death. That is enough to justify reliable longitudinal follow-up; it is not enough to justify a new treatment or universal deep-phenotyping panel.

Best Combination

Use BeLOVE as a phenotyping and hypothesis-generation layer alongside established recurrence-risk cohorts, guideline-based secondary prevention, and randomized trials of antithrombotic, lipid, blood-pressure, rhythm-monitoring, and rehabilitation strategies. The randomized literature should determine what clinicians do; BeLOVE can help identify which patients need intensified follow-up and which biological or patient-reported domains explain residual risk. Its best contribution will come when its predictors are externally validated and linked to a prespecified action, rather than treated as a catalogue of interesting measurements.

Overvalue Warning

Do not treat 10.7% as the universal 2-year risk after any stroke: the consent requirement and extensive visit structure select a relatively fit cohort, and active cancer and transplant patients were excluded.
Do not confuse a planned mechanism with a demonstrated mechanism. Brain–heart loops, immune signatures, MRI features, and multi-omics may become useful predictors, but this report does not show incremental accuracy, causal mediation, or improved patient outcomes from any of them.

Educational use notice. The Clotbust Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.