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

Background Physical exercise has positive impacts on health and can improve angiogenesis, which is impaired during aging, but the underlying mechanisms of benefit are unclear. This meta-analysis and systematic review investigated the effects of exercise on several peripheral angiogenesis markers in older adults to better understand the relationship between exercise and angiogenesis. Methods MEDLINE, Embase, and Cochrane CENTRAL were searched for original, peer-reviewed reports of peripheral concentrations of angiogenesis markers before and after exercise interventions in older adults (> 50 years). The risk of bias was assessed with standardized criteria. Standardized mean differences (SMD) with 95% confidence intervals (CIs) were calculated from random-effects models. Publication bias was assessed with Egger’s test, funnel plots, and trim-and-fill. A priori subgroup analyses and meta-regressions were performed to investigate heterogeneity where possible. Results Of the 44 articles included in the review, 38 were included in meta-analyses for five proteins. Vascular endothelial growth factor (VEGF) was found to be higher after exercise (SMD[95%CI] = 0.18[0.03, 0.34], p = 0.02), and e-selectin (CD62E) was found to be lower after exercise (SMD[95%CI]= −0.72[−1.42, −0.03], p = 0.04). Endostatin (SMD[95%CI] = 0.28[−0.56, 1.11], p = 0.5), fibroblast growth factor 2 (SMD[95%CI] = 0.03[−0.18, 0.23], p = 0.8), and matrix metallopeptidase-9 (SMD[95%CI] = −0.26[−0.97, 0.45], p = 0.5) levels did not change after exercise. Conclusions Of the five angiogenesis blood markers evaluated in this meta-analysis, only VEGF and CD62E changed with exercise. Although more studies are needed, changes in angiogenesis markers may explain the beneficial effects of exercise on angiogenesis and health in older adults.

competitively inhibits VEGF and increases endothelial cell apoptosis [3,4]. For e-selectin (CD62E), although its role in angiogenesis has not been fully elucidated, it has been suggested to regulate the anti-angiogenic activity of endostatin [5] and angiostatin, another potent angiogenesis inhibitor [6]. Interestingly, metallopeptidase-9 (MMP9) is considered a pro-angiogenic factor that plays a crucial role in the modeling of the extracellular matrix, but it is also involved in the generation of the anti-angiogenic factor, angiostatin [7]. Importantly, several of these angiogenesis markers have been shown to change with age; VEGF and FGF2 can decrease with age, while matrix metalloproteinases can increase [8]. Moreover, the age-related impairment in angiogenesis can be associated with increased risk of cardiovascular diseases [9]. Considering the impairment of angiogenesis during aging [8, 10,11], exploring interventions that can improve angiogenesis can be particularly important in older adults.
In older adults, exercise has a positive impact on overall health and wellness [12], with particular bene ts on mental health and cardiovascular tness [13]. Long-term physical activity of at least moderate intensity has been shown to improve cognitive outcomes in older adults over 50 years old [14]. However, the mechanisms via which exercise confers health bene ts in older adults have not yet been fully elucidated. Physical exercise has been shown to improve angiogenesis in animals and humans [11,[15][16][17][18]. Although angiogenesis marker levels may return to baseline levels within a short interval (about 20 minutes) following acute aerobic and resistance exercise [19], long-term exercise have been associated with sustained alteration in angiogenesis marker levels and thus, may have more pronounced bene ts for older adults [20]. Several reviews have suggested that exercise can change angiogenesis marker levels in healthy adults and even potentially older adults [11,[15][16][17][18], but the ndings from exercise studies with angiogenesis markers are inconsistent and no meta-analysis has yet evaluated the impact of exercise on multiple angiogenesis blood markers in apparently healthy older adults. The aim of this systematic review and meta-analysis was to determine the effect of exercise on various angiogenesis blood markers in generally healthy older adults. This will allow for a better understanding of the angiogenesis processes following physical exercise in older adults.

Data Sources and Search Strategy
This systematic review was conducted according to our pre-de ned protocol registered on the International Prospective Register of Systematic Reviews (CRD42022334061), following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines [21]. English-language literature was searched on March 10, 2022, in Cochrane Central, Embase, and MEDLINE for original peer-reviewed articles. A sample search strategy is included in Additional le 1. Reference lists of included articles were further examined for additional eligible studies. Search strategies for all databases included key and MeSH terms with exercise and key angiogenesis proteins of interest.

Study Selection
Inclusion criteria were as follows: 1) healthy older adults (including overweight and obese older adults); 2) ≥ 50 years of age (mean), or if mean age was not reported, population description implied that the group consisted of participants mostly ≥ 50 years of age; 3) measured serum, plasma or blood concentrations of selected angiogenesis proteins (Additional le 1) before and after an exercise intervention; and 4) the exercise intervention was at least moderate intensity (e.g. ≥40 to 59% heart rate reserve (HRR) or oxygen consumption reserve (VȮ 2 R), 46 to 63% of maximal oxygen consumption (VȮ 2max ), 64 to 76% of maximum heart rate (HR max ) or 12-13 on the rating of perceived exertion (RPE) scale [22]), or if exercise intensity was not reported, exercise described as running, cycling, or resistance training. Exclusion criteria were as follows: 1) studies conducted among populations with medical comorbidities or neuropsychiatric conditions that could impact angiogenesis markers (e.g. diabetes, schizophrenia, etc.) 2) mean age of study participants below 50 years of age; 3) signi cant study co-interventions that may impact the effect of exercise on angiogenesis protein concentrations (e.g. high altitude [23], blood ow restriction [24], or drug trials); or 4) exercise intensity below moderate intensity (e.g. yoga or easy walking).
When studies reported similar overlapping subjects, the larger sample size study was included. If an overlapping sample between studies was suspected, clari cation was sought by emailing the corresponding authors. Studies with different populations or different exercise interventions in the same article were not pooled; a minimum of three studies are needed for a meta-analysis [25]. Studies where data cannot be extracted (e.g., pre-or post-intervention values not provided) were not included in a metaanalysis. If insu cient for a meta-analysis, results from the included studies were qualitatively summarized as numerically increased or decreased and whether the changes were signi cant. Inclusion eligibility was assessed by two independent reviewers, and con icts were resolved by consensus between the reviewers.

Data Extraction and Quality Assessment
Two independent reviewers extracted means and standard deviations (SD) of angiogenesis protein concentrations pre-and post-exercise, population characteristics (e.g. mean age, male proportion, body mass index (BMI), VȮ 2max ), exercise intervention characteristics (e.g. exercise duration, intensity, and type/modality), risk of bias items (e.g. dropout rate, compliance), and other study details (e.g. compartment of blood concentrations, protein quanti cation assay) from included studies into a pre-formatted spreadsheet. When resting concentrations at multiple time points were available, the time-points that were closest to the initiation and completion of the exercise intervention were included for analysis. Mean and SD were estimated when the values were given in con dence intervals (CI), quartiles, standard errors, or graphic gures. Missing data were requested by email from the corresponding authors. Studies with different populations or different exercise interventions in the same article were analyzed separately. The risk of bias was evaluated based on criteria adapted from the Newcastle Ottawa Scale and the Cochrane Collaboration's Risk of Bias Assessment Tool [26,27]. Two independent reviewers assessed the risk of bias, and disagreement was resolved by consensus between the reviewers; the overall risk of bias was assessed based on whether the study methodology may greatly contribute to bias under the present review question.

Statistical Analyses
Standardized mean differences (SMD; hedges' g) with 95% CIs were calculated using the random-effects model as between-study variation was anticipated. Heterogeneity was quanti ed by the I 2 statistic, and the Q statistics were used to test signi cant heterogeneity. If heterogeneity was detected (I 2 > 50%), a prede ned systematic investigation was performed using meta-regressions and sub-group analyses. Subgroup analyses were performed with 1) exercise type, 2) exercise duration, 3) population (healthy, overweight, and obese), 4) blood compartment, and 5) overall potential risk of bias rating. If more than 10 observations were available, meta-regressions with 1) mean age, 2) percent male, 3) exercise dose (calculated by multiplying the exercise duration per session, number of sessions per week, and number of weeks)[28], 4) BMI, and 5) VȮ 2max were also performed. Publication bias was assessed using Egger's test and funnel plots with trim-and-ll to adjust for the bias. Funnel plots were visually inspected for potential outliers which were analyzed using the leave-one-out meta-analysis method in STATA 17.0.

Literature search ndings
A total of 9288 records were found from the literature and two more were identi ed from references of included articles. Forty-four articles met the inclusion criteria with su cient data , and 38 of them were included in the meta-analyses (Additional le 2). Seventeen angiogenesis markers were included in the qualitative synthesis ( Table 1); ve of which were included in meta-analyses: VEGF, endostatin, CD62E, FGF2, and MMP9. Characteristics of the included studies are presented in Additional le 3.

Pre-and Post-intervention Comparisons
In the meta-analyses (Table 2), resting blood concentrations of VEGF were signi cantly higher after exercise

Exploration of heterogeneity
Signi cant heterogeneity was detected in meta-analyses ( VEGF None of the subgroup analyses showed signi cant differences between the subgroups in VEGF (Fig. 3,

Endostatin and MMP9
Endostatin was also analyzed with subgroups of exercise duration, population, measurement compartment, and potential risk of bias, yielding a similar division of the subgroups (Additional les 28-31) with signi cant between-group differences (χ 2 = 18.4, p < 0.001); exercise type was aerobic across all studies, so no subgroup analysis was performed for exercise type. Among the 5 comparisons, endostatin was higher after exercise in the 1 session duration/serum compartment/low potential risk of bias subgroup (SMD [95%CI] = 0.97 [0.58, 1.36], p < 0.001) with reduced heterogeneity (I 2 = 0% p = 0.7), but it seems to go in the opposite direction for the ≥ 4 weeks duration /plasma compartment/unclear potential risk of bias subgroup at a reduced heterogeneity (I 2 = 0% p = 0.9) with the two comparisons included. MMP9 was also analyzed with subgroups of exercise type, duration, population, compartment, and potential risk of bias, showing no signi cant changes in the subgroups with more than one comparison (Additional les 32-36); none of the subgroup analyses showed signi cant differences between the subgroups that had more than one comparison.

Discussion
This systematic review and meta-analysis aimed to examine the changes in angiogenesis markers with exercise in healthy older adults. Among the ve angiogenesis markers that were included in meta-analyses, there were signi cant differences in VEGF and CD62E concentrations after exercise, but no differences in MMP9, FGF2, and endostatin. The increase in VEGF and the decrease in CD62E were heterogeneous with small and medium effect sizes, respectively. The high I 2 values in VEGF and CD62E analyses suggest heterogeneity across studies, and the inconsistency in effect estimates may be contributed by differences between studies, such as demographic variables, exercise parameters, and measurement protocols. Many other proteins could not be included in the meta-analyses but showed a trend to increase with exercise in the qualitative analysis. Our ndings supported the observation that exercise can induce angiogenesis, similar to previous reviews [11,15,16]. VEGF VEGF induces angiogenesis by activating VEGF receptors, particularly VEGF receptor-2, to increase microvascular permeability, endothelial cell proliferation and migration, and the release of matrix metalloproteinases [1]. A recent systematic review and meta-analysis on exercise and in ammatory markers evaluated VEGF, but only found one study from their search of randomized controlled trials, which is not enough for a meta-analysis [73]. Our meta-analysis found an increase in peripheral VEGF after exercise in exercise intervention studies. This result should be interpreted with caution as there might be potential outliers in the analysis, as suggested by the leave-one-out meta-analysis but not the standard trimand-ll; the study characteristics of these comparisons did not reveal any noticeable differences in experimental protocols from other included studies. Although the connection between exercise-induced changes in peripheral VEGF and brain VEGF changes remains to be further investigated, animal studies showed that exercising skeletal muscle may induce VEGF and cerebral angiogenesis through the activation of lactate receptors [74,75].
Our meta-regressions investigated the effects of demographic variables on the VEGF response. VEGF changes after exercise did not show a signi cant association with age or sex, although aging has been associated with decreased angiogenesis and VEGF levels [76], and females were found to have higher VEGF levels than males in adults and the elderly at baseline [77]. Although the meta-regression did not nd BMI to be signi cantly associated with changes in VEGF concentrations after exercise, our subgroup analysis found changes in VEGF to have a trend only in the "healthy" and not overweight or obese population.
Previously, serum VEGF levels were also positively correlated with BMI in non-diseased individuals [78].
Based on the current ndings, age and sex may not affect the VEGF response after exercise in older adults, while it is unclear whether the response may be affected by BMI.
Variations in experimental protocols could also contribute to differences in VEGF response to exercise, as examined in our subgroup analyses. Our subgroup analyses also found that peripheral VEGF increased following aerobic exercise, but not after resistance training, similar to a previous meta-analysis which found increased peripheral BDNF after aerobic exercise but not resistance training [79]. Additionally, a previous review on exercise and angiogenesis suggested that although resistance and aerobic exercise both induce angiogenesis, the effects are stronger in aerobic exercise [11]; this could be attributed to the increased

CD62E
On the other hand, CD62E levels were found to decrease with exercise. CD62E is an adhesion molecule in vascular endothelial cells [84]. It is widely recognized as an endothelial and in ammatory marker regulating leukocyte accumulation, but it has been suggested to also mediate angiogenesis [85]. Although its role in angiogenesis remains to be clari ed, the decreasing CD62E levels found with exercise in this paper may be explained by CD62E's anti-angiogenic actions [5,6]. Our current ndings differ from a recent systematic review that did not nd CD62E levels to change after low-to-moderate-intensity aerobic exercise and resistance exercise, although some other adhesion molecules in that review decreased [86]. This discrepancy may re ect the inclusion of newer studies in the current meta-analysis; also, the previous paper was a qualitative review [86] and not a meta-analysis. Our results are consistent with the notion that CD62E may be more responsive to long-term exercise, as our subgroup analysis showed signi cant changes only in the group that exercised for longer than 4 weeks. However, it should be noted that only one study was identi ed in the other exercise duration subgroup (i.e., 1 session group), thus more studies are needed to examine the effects of a single exercise session on peripheral CD62E levels. The CD62E response was also only signi cant in the serum subgroup and not the plasma subgroup, but this may be due to the similar inclusion of only 4-week exercise duration studies in the serum subgroup.
Endostatin, FGF2, and MMP9 The other three proteins that were included in the meta-analyses, endostatin, FGF2, and MMP9, did not show signi cant changes in association with exercise, but they all had a small sample size. Notably, only three articles were found for endostatin; two of the articles had two different participant groups, and thus, it had ve comparisons in total. Exploring further, the exercise duration, population (obesity), compartment, and quality assessment subgroup analyses all showed that one article was signi cantly different from the other two articles with signi cantly opposite endostatin responses. This nding suggests that exercise and population parameters may in uence endostatin response to exercise, but more research would be needed.
The current ndings also suggest that FGF2, which can upregulate VEGF to induce angiogenesis [87], may not be activated by exercise; alternatively, besides FGF2, VEGF can be activated by other factors and other processes, such as hypoxia inducible factor [88]. Nevertheless, the ndings should be interpreted with caution despite the low heterogeneity in its analysis, since similarly to the articles with endostatin, only three articles were found for FGF2. Moreover, MMP9 only had four comparisons in total from two articles.
Three of the four articles also had an unclear risk of bias. Although the high heterogeneity was lowered in subgroup analyses, the subgroups often included only one or two comparisons. Considering the high heterogeneity and the limited number of articles in the current analyses, more research is needed to examine endostatin, FGF2, and MMP9 changes with exercise.

Limitations and future directions
A few limitations should be considered when interpreting our ndings. Many proteins could not be included in a meta-analysis due to the limited studies available. Amongst the ve proteins that were included in the meta-analyses, only a few studies were identi ed for endostatin, FGF2, and MMP9. Thus, the small sample size may have contributed to the lack of response observed, and the results of endostatin, FGF2, and MMP9 should be interpreted with caution. In addition, despite attempts made to address the differences through subgroup and meta-regression analyses, variations across the studies in study parameters such as exercise interventions and demographics may have contributed to the high heterogeneity from the meta-analyses on VEGF, CD62E, endostatin, and MMP9. Furthermore, nine of forty-four studies included in this paper had an unclear overall risk of bias, and publication bias was a concern speci cally in the VEGF meta-analysis where trim-and-ll was not able to adjust the bias. Also, the protein changes observed may not be speci c to angiogenesis; changes in VEGF and CD62E could re ect changes in other pathways, like in ammation. For example, CD62E has also been shown to improve endothelial in ammation [86] and exercise is known to protect endothelial function by reducing in ammatory factors [89]. Therefore, more studies analyzing the effects of exercise on various angiogenesis markers are needed. Moreover, since some angiogenesis markers have shown transient responses after exercise [19], future research could be done on the duration of exercise's effects on different angiogenesis proteins and the effects of measurement time points. Additionally, dehydration after exercise can also decrease blood volume, affecting blood concentrations [90], but the included studies in this meta-analysis did not provide enough information to examine this effect; future studies could further explore the effects of dehydration on angiogenesis levels.

Conclusions
This systematic review and meta-analysis of angiogenesis blood markers in healthy older adults showed changes after exercise in some angiogenesis blood markers, including VEGF and CD62E. The increase in VEGF after exercise may differ depending on the exercise type, and potentially the blood compartment of measurements and population. The decrease in CD62E may also differ depending on exercise duration and blood compartment of measurements. Our ndings suggest that multiple angiogenesis markers, including VEGF and CD62E, can change with physical exercise in older adults. However, many angiogenesis proteins were only qualitatively reviewed because they could not be included in meta-analyses, suggesting the need for further research. The study of additional angiogenesis blood markers can help elucidate the effects of exercise on angiogenesis.
Abbreviations ≥4 weeks At least 4 weeks of exercise Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests  Meta-analysis of resting VEGF concentrations pre-and-post exercise. Higher VEGF was found after exercise.

Figure 2
Meta-analysis of resting CD62E concentrations pre-and-post exercise. Higher CD62E was found before exercise. Supplementary Files