The effect of exercise on blood concentrations of angiogenesis markers in older adults: a systematic review and meta-analysis

Aging is associated with impaired angiogenesis and altered circulating angiogenic factors, including declines in VEGF and FGF2 and increases in matrix metalloproteinases, which may relate to higher cardiovascular risk. Exercise improves health, cognition, and cardiovascular fitness in older adults and has been shown to modulate angiogenesis in animal and human studies. However, reported effects of exercise on angiogenesis biomarkers in older adults are inconsistent, and mechanisms remain unclear. This study asked whether exercise alters peripheral concentrations of angiogenesis markers in generally healthy older adults, synthesizing evidence via systematic review and meta-analysis to clarify direction and magnitude of effects and sources of heterogeneity.

Prior reviews suggest exercise can induce angiogenesis, with potentially stronger effects from aerobic versus resistance training. Some biomarkers exhibit transient post-exercise changes, whereas longer-term training may produce sustained alterations. The roles of specific markers are complex: VEGF promotes endothelial proliferation and permeability; endostatin competitively inhibits VEGF signaling and promotes endothelial apoptosis; CD62E (E-selectin), an endothelial adhesion molecule, may modulate anti-angiogenic actions of endostatin and angiostatin; MMP9 participates in extracellular matrix remodeling and can generate anti-angiogenic fragments (e.g., angiostatin). Aging influences these systems, potentially dampening pro-angiogenic signals and elevating proteases. Previous quantitative syntheses were limited, with insufficient trials on VEGF in older adult RCTs, and qualitative reviews have been inconclusive for CD62E and other markers. This review integrates newer studies and applies meta-analytic methods to quantify exercise effects and explore moderators.

Protocol and reporting followed PRISMA guidelines and was registered in PROSPERO (CRD42022334061). Databases searched: MEDLINE, Embase, Cochrane CENTRAL (March 10, 2022), English-language, with exercise and angiogenesis-related MeSH/keywords; reference lists were screened. Inclusion criteria: generally healthy older adults (including overweight/obese), mean age ≥50 years (or implied), pre- and post-exercise measurements of peripheral angiogenesis proteins (serum/plasma/whole blood), and at least moderate-intensity exercise (e.g., ≥40–59% HRR or VO2R, 46–63% VO2max, 64–76% HRmax, or RPE 12–13; or modalities such as running, cycling, resistance training if intensity not explicitly reported). Exclusions: comorbidities/neuropsychiatric conditions affecting angiogenesis markers (e.g., diabetes, schizophrenia), mean age <50 years, significant co-interventions (e.g., high altitude, blood flow restriction, drug trials), or below-moderate intensity (e.g., yoga, easy walking). Where overlapping samples were suspected, the largest cohort was retained and authors were contacted for clarification. Data extraction (by two reviewers) included biomarker means/SDs pre/post (estimating from CI, SE, quartiles, or graphics when necessary), demographics (age, sex, BMI, VO2max), intervention features (type, intensity, duration), assay and blood compartment, compliance, and dropouts. Risk of bias was evaluated using criteria adapted from the Newcastle-Ottawa Scale and Cochrane Risk of Bias Tool; overall potential risk of bias was rated. Statistical analysis: Random-effects meta-analyses computed Hedges’ g SMD with 95% CI; heterogeneity was quantified with I2 and Q-statistics. Prespecified subgroup analyses: exercise type (aerobic, resistance, combined), exercise duration (single session vs ≥4 weeks), population (healthy, overweight, obese), blood compartment (serum vs plasma), and overall potential risk of bias. Meta-regressions (when >10 observations) examined associations with mean age, percent male, exercise dose (session duration × sessions/week × weeks), BMI, and VO2max. Publication bias: Egger’s test, funnel plots, and trim-and-fill; potential outliers explored with leave-one-out analyses (STATA 17). When insufficient for pooling, findings were qualitatively summarized by direction and significance of change.

- Included studies: 44 in the systematic review; 38 contributed to meta-analyses of five proteins (VEGF, CD62E, endostatin, FGF2, MMP9). Most studies rated low potential risk of bias (35 low, 9 unclear). - VEGF: Increased after exercise (SMD [95% CI] = 0.18 [0.03, 0.34], p = 0.02); heterogeneity I2 = 66.8%, Q p < 0.001. Publication bias detected (Egger’s B[SE] = 1.91 [0.62], p = 0.002); trim-and-fill did not adjust effect (SMD unchanged). Leave-one-out analyses around identified outliers yielded estimates within the overall 95% CI. Subgroups: significant increase for aerobic exercise (SMD 0.22 [0.02, 0.42], p = 0.03), but not resistance or combined modalities; trends for ≥4 weeks (SMD 0.13 [−0.003, 0.26], p = 0.06), healthy population (SMD 0.21 [−0.004, 0.42], p = 0.05), and plasma compartment (SMD 0.42 [−0.02, 0.86], p = 0.06). Significant increase observed in low risk-of-bias studies (SMD 0.22 [0.04, 0.40], p = 0.01). Meta-regressions: no significant associations with mean age, percent male, exercise dose, BMI, or VO2max. - CD62E (E-selectin): Decreased after exercise (SMD [95% CI] = −0.72 [−1.42, −0.03], p = 0.04); heterogeneity I2 = 86.8%, Q p < 0.001. Publication bias not detected. Subgroups: significant decreases seen for ≥4-week duration (SMD −0.86 [−1.64, −0.08], p = 0.03) and serum compartment (SMD −1.22 [−2.33, −0.11], p = 0.03). No significant between-subgroup differences detected. - Endostatin: No overall change (SMD [95% CI] = 0.28 [−0.56, 1.11], p = 0.5); heterogeneity I2 = 81.5%, Q p < 0.001. Publication bias detected (Egger’s B[SE] = −7.62 [1.90], p < 0.001); trim-and-fill remained non-significant (estimated SMD 0.54 [−0.33, 1.41]). Subgroup signals suggested higher endostatin after a single session measured in serum and in low risk-of-bias studies (SMD ~0.97 [0.58, 1.36], p < 0.001; I2 = 0%), versus opposite direction in ≥4-week plasma measures (two comparisons; I2 = 0%). - FGF2: No change (SMD [95% CI] = 0.03 [−0.18, 0.23], p = 0.8); heterogeneity I2 = 0%; publication bias not detected. - MMP9: No change (SMD [95% CI] = −0.26 [−0.97, 0.45], p = 0.5); heterogeneity I2 = 89.1%; publication bias not detected. - Qualitative synthesis: One study each reported significant increases in ANGPTL4 and TSP-1 after exercise. ANGPTL2, ANGPTL4, HGF, PDGF-AA, PDGF-BB, VEGF-C, and VEGF-D comparisons generally trended upward post-exercise without consistent statistical significance. MMP2 and PlGF trended downward. Angiogenin showed no significant change. - Overall, findings indicate small-magnitude VEGF increases and moderate-magnitude CD62E decreases post-exercise with substantial heterogeneity across studies.

Findings support exercise-induced modulation of angiogenesis-related biomarkers in older adults, notably increased VEGF and decreased CD62E. VEGF, a central pro-angiogenic mediator acting via VEGFR-2 to enhance endothelial proliferation, migration, and permeability, increased primarily following aerobic exercise, aligning with prior work showing stronger angiogenic and neurotrophic responses to aerobic versus resistance training. Potential mechanistic links include lactate-mediated activation of HCAR1 triggering cerebral VEGF and angiogenesis, suggesting systemic-muscle-to-brain crosstalk. CD62E, an endothelial adhesion molecule implicated in inflammatory signaling and potentially anti-angiogenic processes through interactions with endostatin and angiostatin, decreased after longer-duration exercise, consistent with exercise’s endothelial-protective and anti-inflammatory effects and potential relief of anti-angiogenic tone. The lack of consistent changes in endostatin, FGF2, and MMP9 may reflect limited sample sizes, diverse protocols, and opposing short-term versus training effects. Subgroup and meta-regression analyses suggest that exercise modality (aerobic), measurement compartment (serum vs plasma), population characteristics (healthy vs overweight/obese), and study quality may influence observed effects; however, heterogeneity remained high and demographic moderators were not significant. Publication bias in VEGF and endostatin and inconsistent protocols caution against overgeneralization. Collectively, results indicate that select angiogenic markers respond to exercise in older adults, potentially contributing to vascular and cognitive health benefits, while highlighting the need for standardized protocols and broader biomarker panels.

In generally healthy older adults, exercise is associated with increased circulating VEGF and decreased CD62E, with no overall changes in endostatin, FGF2, or MMP9. Effects appear stronger for aerobic exercise and longer training durations for CD62E, and may depend on blood compartment and study quality. These biomarker shifts may underlie exercise-related angiogenesis and associated health benefits. Future research should include larger, well-controlled studies with standardized exercise prescriptions and sampling protocols, evaluation of additional angiogenic mediators, time-course analyses of acute versus chronic responses, and exploration of mechanistic links to functional outcomes in vascular and brain health.

- Limited number of studies for some biomarkers (endostatin, FGF2, MMP9), reducing power and precision; subgroup analyses often included few comparisons. - High heterogeneity across many analyses (VEGF, CD62E, endostatin, MMP9) likely due to variability in populations, exercise modalities, durations, intensities, and assay methods/blood compartments. - Publication bias detected for VEGF and endostatin; trim-and-fill did not alter VEGF effect, but bias remains a concern. - Nine of forty-four studies had unclear overall risk of bias due to incomplete reporting of interventions, assays/compartments, participant characteristics, or adherence. - Potential confounding by inflammation and other pathways, as some markers (e.g., CD62E) reflect endothelial inflammation in addition to angiogenesis. - Timing of biomarker sampling may miss transient responses; hydration status and exercise-induced hemoconcentration could affect concentrations but were insufficiently reported to analyze. - Generalizability limited to relatively healthy older adults and moderate-or-greater intensity exercise protocols.