# Allostatic load-cardiovascular disease associations and the mediating effect of inflammatory factors: a prospective cohort study
Researcher: Ge Zhang
Chinese name: 张格
ORCID identity anchor: https://orcid.org/0000-0002-3116-3246
Canonical researcher: https://drgezhang.com/#person
Journal: Frontiers in Cardiovascular Medicine
Year: 2026
Type: Article
DOI: 10.3389/fcvm.2025.1724572
Canonical page: https://drgezhang.com/papers/doi-10-3389-fcvm-2025-1724572.html
Markdown record: https://drgezhang.com/papers/doi-10-3389-fcvm-2025-1724572.md
## Full Authors
1. Shuai Xu
2. Ge Zhang
3. Yudi Xu
4. Chaoyang Yu
5. Ruhao Wu
6. Zhengrui Li
7. Teng Li
8. Xinyue Cui
9. Xufeng Huang
10. Shujing Zhou
11. Yahui Han
12. Haonan Zhang
13. Shiqian Zhang
14. Yufeng Jiang
## Official Abstract
### Background
Allostatic load (AL) captures multisystem dysregulation that accrues with chronic stress and may shape cardiovascular disease (CVD) risk through neuroendocrine and immune pathways. Robust population-scale evidence clarifying the exposure-response pattern and the extent of inflammatory mediation remains limited.
### Methods
In the UK Biobank, we analyzed 205,504 adults free of CVD at baseline from 502,366 recruited. An AL score was assembled from 12 routinely measured biomarkers. Incident CVD was ascertained via linkage to hospital and mortality records. We estimated adjusted hazard ratios (HRs) using Cox models and assessed nonlinearity with restricted cubic splines; robustness was evaluated in prespecified subgroups and sensitivity analyses. We also investigated the role of the mediating effect of inflammatory factors in the AL-CVD relationship.
### Results
Higher AL tracked with progressively greater CVD risk in a graded, non-linear pattern. Relative to AL = 0, AL> = 6 was associated with HR 2.15 (95% CI 1.99–2.33). Eight inflammatory markers met retention criteria for mediation; neutrophil count mediated 4.73% of the AL-CVD association. Group contrasts across AL tertiles indicated Cohen's d near 0.5 for several markers, largest for no-AL vs. high-AL.
### Conclusion
Elevated AL is linked to higher incident CVD with a non-linear exposure-response, and neutrophil-centric inflammation accounts for a measurable portion of the association. These findings support integrating stress-biology constructs and inflammatory profiling into cardiovascular risk assessment and prevention frameworks.
Text reproduced verbatim from Version of Record ([source](https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2025.1724572/full)); [DOI](https://doi.org/10.3389/fcvm.2025.1724572); [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).
## Research & review context
### Research fields
- Cardiovascular epidemiology
- Multisystem stress-biomarker research
### Review topics
- Allostatic load and cardiovascular disease
- Inflammatory markers in exposure-outcome associations
### Scientific role
Population cohort analysis of biomarker-defined allostatic load, incident cardiovascular events and statistical inflammatory mediation.
### Evidence position
Observational prospective cohort associations; statistical mediation is not an established causal prevention mechanism or validated stress-management intervention.
### Related concepts
- Multisystem biomarker burden
- Statistical-versus-causal mediation
### Possible citation contexts
- Background: cumulative physiological burden and incident cardiovascular disease.
- Evidence synthesis: population-level allostatic-load associations, without claiming benefit from lowering the score or prescribing anti-inflammatory treatment.
Possible literature-review contexts, not evidence of existing citations. Read the findings and Evidence Scope before citing the original article.
## Evidence Snapshot
**Allostatic load-cardiovascular disease associations and the mediating effect of inflammatory factors: a prospective cohort study**
Baseline multisystem biomarkers were associated with subsequent cardiovascular events in an observational analysis. Adjusted associations and statistical mediation do not establish prevention by lowering the score or an inflammatory marker, or a validated clinical prediction model.
| Evidence | Value |
| --- | --- |
| Unique total | n=205,504 |
| UK Biobank analytic cohort | n=205,504 |
## Research Question
How is biomarker-defined allostatic load associated with incident cardiovascular disease, and what statistical mediation is attributed to inflammatory markers in the analyzed UK Biobank cohort?
## Author Evidence Summary
This observational analysis relates baseline multisystem biomarkers to subsequent cardiovascular events. Its adjusted associations and mediation estimates inform epidemiological hypotheses; they do not demonstrate that lowering the score or an inflammatory marker prevents events. It is not a validated clinical prediction model.
## Key Findings
### KF1: The analytic cohort comprised 205,504 participants and 18,542 incident CVD events during an article-reported 13-year follow-up period.
Context: The analytic sample is not the entire UK Biobank recruitment population. The reported 13-year period is not described here as a mean or median follow-up.
| Evidence | Value |
| --- | --- |
| analytic participants | 205,504 |
| incident CVD events | 18,542 |
| article reported follow up period | 13 years |
Source locator: Results: Baseline characteristics; Fig. 1
### KF2: Allostatic load was operationalized using 12 baseline biomarkers with quartile-based risk scoring and a score range of 0–12.
Context: HDL and IGF-1 used the low-value risk direction. This biomarker score is not a direct measurement of psychological stress, a neuroendocrine response or an inflammatory intervention.
| Evidence | Value |
| --- | --- |
| baseline biomarkers | 12 |
| score range | 0–12 |
| risk scoring | Quartile-based; low-value direction for HDL and IGF-1 |
Source locator: Methods: Assessment of allostatic load
### KF3: In adjusted Model 3, AL ≥6 versus AL = 0 was associated with incident CVD at HR 2.15 (95% CI 1.99–2.33).
Context: AL = 0 is the reference group. This relative hazard association is not a 115% absolute-risk increase or a treatment effect.
| Evidence | Value |
| --- | --- |
| comparison | AL ≥6 versus AL = 0 |
| model | Model 3 |
| hazard ratio | 2.15 |
| reported 95 percent CI | 1.99–2.33 |
Source locator: Results: Associations of AL with CVD; Fig. 2A
### KF4: The AL–CVD association was nonlinear, and the main text reported directionally similar results after excluding events in the first three years after baseline.
Context: No additional precise hazard ratios from supplementary tables are supplied here, and this sensitivity analysis does not fully exclude reverse causation.
| Evidence | Value |
| --- | --- |
| nonlinearity P | <0.001 |
| sensitivity exclusion | Events in the first 3 years after baseline |
Source locator: Results: Associations of AL with CVD; Fig. 2B
### KF5: Eight candidate mediators entered the correlation analysis; the neutrophil-count mediation estimate was 4.73%.
Context: This is model-based decomposition using baseline indicators, not a proven causal pathway or evidence that lowering neutrophils or an inflammatory marker prevents CVD.
| Evidence | Value |
| --- | --- |
| candidate mediators | 8 |
| neutrophil count mediation estimate | 4.73% |
Source locator: Results: Mediation analyses of inflammatory factor on associations of AL with CVD; Fig. 4E
## Study Design & Analytical Framework
### Study profile
- Study Design: Prospective cohort association analysis using UK Biobank data
- Evidence Type: Human observational exposure–outcome and exploratory mediation evidence
- Unique total: 205,504
- Population: Adults without cardiovascular disease at baseline in the analytic cohort
- Primary Endpoint: Incident CVD defined by the study's linked-record ascertainment
- Secondary Endpoint: Exploratory exposure-response and mediation analyses of that outcome; not a separately established interventional benefit endpoint
- External validation: No
- Data modalities: baseline biomarkers, questionnaire covariates, linked hospital/mortality outcomes
### Cohort hierarchy
| Cohort | Role | n | Relationship |
| --- | --- | ---: | --- |
| UK Biobank analytic cohort | analytic population without cardiovascular disease at baseline | 205504 | Independent cohort |
## What This Study Adds
- Population-level exposure–outcome association evidence for biomarker-defined AL and incident CVD.
- A nonlinear exposure-response description with a specified sensitivity analysis.
- An explicitly qualified statistical inflammatory-mediation estimate.
## Evidence Scope
### Supports
- Observational population-level associations between biomarker-defined AL and incident CVD.
- Nonlinearity and exploratory statistical mediation within the analyzed cohort.
### Does Not Establish
- A causal prevention mechanism or efficacy of anti-inflammatory drugs.
- A prospectively validated individual clinical prediction tool.
- Validated benefits of a stress-management intervention.
### Limitations
- Observational design does not establish causality; residual confounding remains possible.
- Missing-data selection limits interpretation of the analytic sample.
- Baseline AL and mediator measurements limit temporal interpretation of mediation.
- Population-specific thresholds may not transfer to other populations.
## When This Study Is Useful to Cite
### CU1: What prospective cohort evidence links allostatic load to incident CVD?
An analytic cohort of 205,504 participants had 18,542 incident CVD events during the article-reported 13-year period; Model 3 compared AL ≥6 with AL = 0 at HR 2.15 (95% CI 1.99–2.33). These are observational relative-hazard findings, not a treatment effect.
Evidence: [KF1](#kf1), [KF3](#kf3)
### CU2: How was a multisystem biomarker AL score operationalized?
Twelve baseline biomarkers formed a 0–12 score using quartile-based risk scoring, with the low-value direction for HDL and IGF-1. It is not a direct psychological-stress measure or an intervention.
Evidence: [KF2](#kf2)
### CU3: Was the AL–CVD association nonlinear and examined in sensitivity analyses?
Nonlinearity P<0.001 and directionally similar results after excluding first-three-year events were reported in the main text. This does not fully exclude reverse causation.
Evidence: [KF4](#kf4)
### CU4: What statistical inflammatory mediation was reported?
Among eight candidate mediators in the correlation analysis, the neutrophil-count mediation estimate was 4.73%. Baseline model decomposition does not establish a causal pathway or drug benefit.
Evidence: [KF5](#kf5)
## What This Study Should Not Be Cited to Claim
- A causal prevention mechanism or efficacy of anti-inflammatory drugs.
- A prospectively validated individual clinical prediction tool.
- Validated benefits of a stress-management intervention.
## Evidence Matrix
### Analytic population and incident cardiovascular outcomes
- Context: The analytic sample is not the entire UK Biobank recruitment population. The reported 13-year period is not described here as a mean or median follow-up.
- Finding: The analytic cohort comprised 205,504 participants and 18,542 incident CVD events during an article-reported 13-year follow-up period.
- Evidence Level: Human observational exposure–outcome and exploratory mediation evidence
- Scope: Describes the analytic population and reported follow-up period, not all recruited UK Biobank participants.
- Source Locator: Results: Baseline characteristics; Fig. 1
- Evidence refs: [KF1](#kf1)
### Multisystem biomarker exposure scoring
- Context: HDL and IGF-1 used the low-value risk direction. This biomarker score is not a direct measurement of psychological stress, a neuroendocrine response or an inflammatory intervention.
- Finding: Allostatic load was operationalized using 12 baseline biomarkers with quartile-based risk scoring and a score range of 0–12.
- Evidence Level: Human observational exposure–outcome and exploratory mediation evidence
- Scope: Defines the biomarker exposure; it does not validate a psychological-stress measurement or an intervention.
- Source Locator: Methods: Assessment of allostatic load
- Evidence refs: [KF2](#kf2)
### Adjusted high-versus-zero allostatic-load association
- Context: AL = 0 is the reference group. This relative hazard association is not a 115% absolute-risk increase or a treatment effect.
- Finding: In adjusted Model 3, AL ≥6 versus AL = 0 was associated with incident CVD at HR 2.15 (95% CI 1.99–2.33).
- Evidence Level: Human observational exposure–outcome and exploratory mediation evidence
- Scope: Supports an adjusted relative-hazard association only, not absolute-risk change or treatment efficacy.
- Source Locator: Results: Associations of AL with CVD; Fig. 2A
- Evidence refs: [KF3](#kf3)
### Exposure-response nonlinearity and sensitivity analysis
- Context: No additional precise hazard ratios from supplementary tables are supplied here, and this sensitivity analysis does not fully exclude reverse causation.
- Finding: The AL–CVD association was nonlinear, and the main text reported directionally similar results after excluding events in the first three years after baseline.
- Evidence Level: Human observational exposure–outcome and exploratory mediation evidence
- Scope: Supports a nonlinear association and a specified sensitivity check, without eliminating reverse causation.
- Source Locator: Results: Associations of AL with CVD; Fig. 2B
- Evidence refs: [KF4](#kf4)
### Neutrophil-count statistical mediation
- Context: This is model-based decomposition using baseline indicators, not a proven causal pathway or evidence that lowering neutrophils or an inflammatory marker prevents CVD.
- Finding: Eight candidate mediators entered the correlation analysis; the neutrophil-count mediation estimate was 4.73%.
- Evidence Level: Human observational exposure–outcome and exploratory mediation evidence
- Scope: Supports exploratory statistical mediation, not causal prevention or anti-inflammatory drug efficacy.
- Source Locator: Results: Mediation analyses of inflammatory factor on associations of AL with CVD; Fig. 4E
- Evidence refs: [KF5](#kf5)
## Q&A
### Which population and incident outcomes were analyzed?
The analytic cohort comprised 205,504 participants and 18,542 incident CVD events during an article-reported 13-year follow-up period. The analytic sample is not the entire UK Biobank recruitment population. The reported 13-year period is not described here as a mean or median follow-up.
Evidence: KF1
### What does the allostatic-load score represent?
Allostatic load was operationalized using 12 baseline biomarkers with quartile-based risk scoring and a score range of 0–12. HDL and IGF-1 used the low-value risk direction. This biomarker score is not a direct measurement of psychological stress, a neuroendocrine response or an inflammatory intervention.
Evidence: KF2
### What was the adjusted high-versus-zero AL association?
In adjusted Model 3, AL ≥6 versus AL = 0 was associated with incident CVD at HR 2.15 (95% CI 1.99–2.33). AL = 0 is the reference group. This relative hazard association is not a 115% absolute-risk increase or a treatment effect.
Evidence: KF3
### What do the nonlinear and sensitivity analyses add?
The AL–CVD association was nonlinear, and the main text reported directionally similar results after excluding events in the first three years after baseline. No additional precise hazard ratios from supplementary tables are supplied here, and this sensitivity analysis does not fully exclude reverse causation.
Evidence: KF4
### Does the mediation estimate establish prevention by lowering neutrophils?
Eight candidate mediators entered the correlation analysis; the neutrophil-count mediation estimate was 4.73%. This is model-based decomposition using baseline indicators, not a proven causal pathway or evidence that lowering neutrophils or an inflammatory marker prevents CVD.
Evidence: KF5
## Concepts & Entities
### Population
UK Biobank
### Exposure
allostatic load, multisystem biomarkers
### Outcome
incident cardiovascular disease
### Methods
Cox regression, restricted cubic splines, mediation analysis
## Related Research
- [Association between sleep duration, sleep trouble and all-cause mortality in individuals with hyperuricemia in the United States](https://drgezhang.com/papers/doi-10-3389-fpubh-2025-1521372.html) — Both examine human exposure–outcome associations, but use different cohorts, exposures and endpoints; neither validates the other's estimates. (DOI: 10.3389/fpubh.2025.1521372)
## Publication & Provenance
- DOI: https://doi.org/10.3389/fcvm.2025.1724572
- Publication Url: https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2025.1724572/full
- PubMed: https://pubmed.ncbi.nlm.nih.gov/41561117/
- Pmc Url: https://pmc.ncbi.nlm.nih.gov/articles/PMC12813168/
- Review Basis: Selected findings checked against published main text, readable tables and publication records. No participant-level reanalysis or comprehensive supplementary-data audit was performed.
- Data Availability: UK Biobank resource; access is governed by its application process.
## Evidence-page notice
Author-controlled evidence synthesis, separate from the published abstract and not a replacement for the version of record. Study design, analysis denominators and unresolved reporting differences are retained.
## Links
- DOI: https://doi.org/10.3389/fcvm.2025.1724572
- HTML page: https://drgezhang.com/papers/doi-10-3389-fcvm-2025-1724572.html
- Google Scholar query: https://scholar.google.com/scholar?q=%22Allostatic%20load-cardiovascular%20disease%20associations%20and%20the%20mediating%20effect%20of%20inflammatory%20factors%3A%20a%20prospective%20cohort%20study%22
- ORCID: https://orcid.org/0000-0002-3116-3246