Effects of Resistance Exercise on Neuroprotective Factors in Middle and Late Life: A Systematic Review and Meta-Analysis

Neuroprotective factors are involved in brain functioning. Although physical exercise has been shown to have a positive influence on these factors, the effect of resistance exercise on them is not well known. This systematic review and meta-analysis aimed to 1) estimate the efficacy of resistance exercise on major neuroprotective factors, such as insulin-like growth factor-1 (IGF-1), brain-derived neurotrophic factor (BDNF), and vascular endothelial growth factor (VEGF), in middle and late life and 2) determine whether the effect is dose dependent. A systematic search was conducted in CINAHL, Cochrane CENTRAL, MEDLINE, Scopus, PEDro, SPORTDiscus, and Web of Science up to November 2022. Random effects models were used to estimate standardized mean differences (SMDs) and their respective 95% confidence intervals (CI) for the effect of resistance exercise on peripheral IGF-1, BDNF or VEGF levels in older adults. Thirty randomized clinical trials with 1247 subjects (53.25% women, 45-92 years) were included in the systematic review, and 27 were selected for the meta-analysis. A significant effect of resistance exercise on IGF-1 levels was observed (SMD: 0.48; 95% CI: 0.27, 0.69), being more effective when performing 3 sessions/week (SMD: 0.55; 95% CI: 0.31, 0.79) but not on BDNF (SMD: 0.33; 95% CI: -0.29, 0.94). The effect on VEGF could not be determined due to the scarcity of studies. Our data support the resistance training recommendation in middle and late life, at a frequency of at least 3 sessions/week, to mitigate the neurological and cognitive consequences associated with aging, mainly through IGF-1.


Introduction
Aging is a natural physiological process associated with cellular and synaptic changes at the brain level related to cognitive processes. Cognitive decline is a slow process that begins in middle life [1]  have an important role in cell proliferation and growth as well as in neuronal development and function [4][5][6]. IGF-1 is a peptide that regulates the effects of growth hormones, and BDNF belongs to the neurotrophin family. Both are essential proteins in brain development and tissue remodelling [7,8]. They provide great benefits to cognition due to their effects on neuroplasticity [5,9]. Moreover, these factors are expressed in some regions of the central nervous system specific to cognition, supporting the idea that their decrease causes cognitive impairment [5,8]. Finally, VEGF, like BDNF, is a neurotrophin with some neuroprotective effect [10]. It is an angiogenic factor that has the capacity to preserve brain cells and slow the deterioration of spatial memory and cognitive impairment [10][11][12]. There is sufficient evidence to support that physical exercise has a positive influence on the release of neuroprotective factors and their cerebral effect by increasing their expression in the central nervous system [13]. Considering that these are peptides that cross the blood-brain barrier, the elevation of their peripheral levels as a consequence of exercise has been reported to favor learning, neurogenesis and angiogenesis [14].
The effect of resistance exercise on these factors has not been sufficiently studied in older adults, and the literature thus far has shown inconclusive results for VEFG [16,17], while they seem to be more consistent for IGF-1 [18] and BDNF [19]. On the other hand, the importance of training parameters such as exercise frequency or intensity in enhancing neuroprotective factors has been described [20], although the magnitude of influence of these training characteristics has not been sufficiently quantified. Therefore, the objectives of this systematic review and meta-analysis were 1) to update and synthesize the available evidence regarding the effect of resistance exercise on key neuroprotective factors at the peripheral level in middle and late life and 2) to determine whether the effect depends on exercise dose.

Methods
The present systematic review and meta-analysis was conducted following the recommendations of the Cochrane Handbook of Systematic Reviews of Interventions [21], and the standards for systematic reviews and meta-analyses of the PRISMA Statement [22] were followed. This review was registered in the PROSPERO database (registration number: CRD420223 02859).

Search strategy
A systematic search was conducted in the following bibliographic databases: CINAHL (via EBSCOhost), Cochrane Central Register of Controlled Trials,
In addition, it was necessary to contact four authors [16,[23][24][25], obtaining a response from only one of them because the data required to carry out the meta-analysis could not be obtained from the articles.
The systematic search was performed independently by two reviewers (E.R.G. and A.T.C.). When there were disagreements, a third researcher made the final decision (V.M.V.).

Eligibility criteria
The inclusion criteria for the systematic review and metaanalysis were as follows: 1) participants: adults with a mean age ≥ 45 years; 2) intervention: resistance exercise (minimum 1 session); 3) comparator: control group; and 4) outcome: concentration of BDNF, IGF-1 or VEGF in serum and/or plasma. Furthermore, the exclusion criteria were as follows: 1) studies other than randomized clinical trials (RCTs); 2) subjects with cognitive impairment.
Two independent reviewers (E.R.G. and A.T.C.) conducted the study selection. When there were disagreements, a third researcher made the final decision (V.M.V.).

Data extraction
The full texts of the included studies were reviewed, and the main data were independently extracted from the included studies by 2 reviewers (E.R.G. and A.T.C.) and synthesized in an ad hoc table including 1) author's name, 2) year of publication, 3) country, 4) population characteristics (final number of participants in each group, proportion of women, age of participants, health status), 5) intervention characteristics (type of intervention in each group, frequency, duration, intensity and volume) and 6) outcome (IGF-1, BDNF or VEGF levels in plasma and/or serum). A third reviewer (V.M.V.) was consulted to resolve disagreements between reviewers.
Continuous data were extracted from the studies (including prepost mean IGF-1, BDNF and VEGF values, Aging and Disease • Volume 14, Number 4, August 2023 1266 standard deviation and sample size of the intervention and control groups). For statistical analysis, all IGF-1, BDNF and VEGF values were transformed to the same unit (ng/mL) (where 1 ng/mL = 1000 pg/mL).

Risk of bias assessment
Two reviewers (E.R.G. and A.T.C.) independently assessed the risk of bias of the included studies using the Cochrane Risk of Bias Tool for Randomized Clinical Trials (RoB 2.0) [21]. Any discrepancies were resolved by a third reviewer (V.M.V.). This tool consists of an assessment based on the following domains: randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of reported outcome. Each of these domains can be assessed as "low risk of bias", "some concerns" and "high risk of bias". Therefore, the overall risk for each of the studies was classified as "low risk of bias" when a low risk of bias was determined for all domains; "unclear risk of bias" when at least one domain had unclear risk but no high risk of bias for any specific domain; and "high risk of bias" when at least one domain was assessed as high risk of bias or as unclear risk of bias in multiple domains [26].

Data analysis
The estimated pooled standardized mean differences (SMDs) of the mean differences for IGF-1, BDNF and VEGF and their 95% confidence intervals (CIs) were calculated using Cohen's d index [21]. When repeated measurements were reported, we considered only the last measurement as the end point. When a study had two intervention groups that performed resistance exercise, they were taken into account as different studies in the analysis of the results. A meta-analysis for each factor was performed using a random-effects model with the DerSimonian and Laird method [27] to determine the effect of resistance exercise on neuroprotective factors compared to a control group. Heterogeneity of results between studies was assessed using the I 2 statistic, which is classified as unimportant (0% to 30%), moderate (30% to 50%), substantial (50% -75%) and high (75% -100%). The corresponding p values were also considered [21]. As recommended by the Cochrane Handbook, when data on standard deviation were not reported, they were estimated using the standard error, the CI or a statistical test (t test, F test or a p value) [21].
A sensitivity analysis was performed to determine the robustness of the estimates by eliminating each study included in the meta-analysis one by one, as well as studies in which the population had any specific health disorder or pathology, to determine whether any represented a large proportion of heterogeneity in the pooled ES. For neuroprotective factors where a significant difference was found after resistance exercise, the doseresponse relationship was estimated by subgroup analysis according to frequency (days/week), sets, exercise intensity, considering high intensity >10 repetitions maximum (RM)(28) and ≥70% RM [29] and lightmoderate intensity ≤10RM and <70% RM and those studies in which indicate that they perform light and/or moderate intensity exercises, the session time (minutes) of the intervention and duration of the exercise program (months). We also conducted a subgroup analysis by sex and a meta-regression model to determine the influence of body mass index and age on this association.
Publication bias was evaluated using Egger's regression asymmetry test, with p values less than 0.10 considered statistically significant. STATA Statistical software, version 16 (StataCorp LLC, College Station, TX, USA) was used to perform the statistical analyses.

Risk of bias assessment
According to the RoB 2.0. tool [26], 11 out of 30 were classified as "high risk of bias", and 19 out of 30 were classified as "unclear risk of bias". The most affected domains were randomization process, deviations from intended interventions and selection of the reported result. The assessment of risk is shown in Supplementary Fig. 2.

Meta-analysis
Aging and Disease • Volume 14, Number 4, August 2023 1268 The pooled SMD of resistance exercise on IGF-1 levels was 0.48 (95% CI: 0.27, 0.69; I 2 = 52.6%, P=0.001) (Fig.  2). In relation to training characteristics, exercise frequency showed significant results with at least 3 days per week (SMD: 0.55; 95% CI: 0.31, 0.79). In relation to the number of sets of the program, both performing 2 or fewer and more than 2 sets showed significant benefits, as well as light-moderate and high-intensity exercises, sessions of more or less than 60 minutes and exercise programs of more or less than 3 months (Table 2). In both men and women, IGF-1 levels increased significantly after resistance exercise (Supplementary Table 2). Meta-regression analysis showed that neither body mass index nor age influenced peripheral IGF-1 levels (P>0.05) in older adults (Supplementary Table 3). BDNF levels were not significantly different when performing resistance exercise versus the control group (SMD: 0.33; 95% CI: -0.29, 0.94; I 2 = 77.8%, P=0.000) (Fig. 2). Subgroup analyses on sex could not be performed due to the limited number of studies.
VEGF was only analysed by one study [17], so it was not included in the meta-analysis. Significant differences were observed between the resistance exercise group and the control group. Abbreviations: BDNF = brain-derived neurotrophic factor, CG = control group, IG = intervention group, IGF-1 = insulin-like growth factor type 1, min = minutes, MMII = lower limbs, MMSS = upper limbs, NR = No reported, OMNI-RES = OMNI-Resistance Exercise Scale, rep = repetition, RM = maximum repetition, VEGF = vascular endothelial growth factor

Sensitivity analysis
The pooled ES estimates for the effect of resistance exercise on IGF-1 and BDNF were not significantly changed in magnitude or direction when removing each study included in the meta-analysis one by one, as well as when eliminating studies in which the population had a specific health disorder or pathology.

Publication bias
There was evidence of publication bias, as seen in the funnel plots and Egger's tests for IGF-1 (P=0.052) and BDNF (P=0.100) (Supplementary Fig. 3).

Discussion
Although several studies have reported the effectiveness of physical exercise on the main neuroprotective factors, there was no updated review comparing the effect of resistance exercise on these factors and determining the dose necessary to achieve this effect in middle and late life. Our data support that resistance exercise increases IGF-1 levels, being more effective when performed at least three sessions per week, independent of the number of sets, the intensity of the exercise, the time of the session and the duration of the exercise program. However, no significant effects were estimated for peripheral BDNF levels, and in the case of VEGF, the scarce number of articles made it impossible to analyse the effect on this factor.
Aging and Disease • Volume 14, Number 4, August 2023 1271 The progressive decline in IGF-1 levels experienced by the population [56,57] is associated with impaired brain function and risk of vascular dementia [56,58]. The available evidence seems to indicate that resistance exercise could mitigate these adverse effects [18,59]. Several factors could justify the effect of resistance exercise on IGF-1 levels, such as the fact that this type of exercise induces anabolic hormonal responses, allowing the direct release of IGF-1 by the liver or indirectly when induced by growth hormone [57]. Furthermore, a positive relationship has been established between IGF-1 levels, strength, and muscle mass because this factor is capable of increasing the proliferation capacity of muscle satellite cells, thereby preventing the loss of muscle mass related to aging [60].
Regarding the dose necessary to achieve this beneficial effect, our data corroborate that three or more sessions per week would be necessary to increase IGF-1 levels, independent of the number of series, the intensity of the exercise, the duration of the sessions or the duration of the exercise program. However, contrary to what has been reported in previous reviews [18,59], which indicated that the increase in IGF-1 levels would only occur in women, our data show that this association is also positive in men, which seems logical considering that as a consequence of exercise, there are increases in growth hormone levels in both sexes, and this is closely related to IGF-1 synthesis [61].
Aging and Disease • Volume 14, Number 4, August 2023 1272 Abbreviations: SMD = standardized mean differences Different meta-analyses have evaluated the effects of exercise interventions on BDNF, reporting significant effects in adolescents and children and in neurodegenerative disorders [62][63][64]. In older adults, Marinus et al. [19] evaluated the impact of resistance exercise on BDNF before and after exercise, observing a significant increase in peripheral levels of this neuroprotective factor. However, our results do not show this effect when compared against a control group, which could be because the BDNF generated in skeletal muscle with contraction is not released into the circulation but would be used to enhance muscle oxidation [65]. This could also be because the release and synthesis of this factor occurs immediately after exercise [66]. However, when exercise ceases, this effect disappears, and the peripheral concentration of BDNF normalizes [65,67], which could indicate that circulating BDNF would be transported to the brain through the blood circulation, where it would cross the blood-brain barrier, achieving greater neuronal survival and synaptogenesis and, therefore, greater brain function and structural changes [66].
Moreover, several studies have shown that aerobic exercise has contradictory effects on VEGF [68,69]. Resistance exercise seems to be effective in the adult population when it is performed with blood flow restriction [70]. However, the lack of studies has not allowed a synthesis of the effect of resistance exercise on this factor. Some limitations of our systematic review and metaanalysis should be acknowledged. First, the lack of studies evaluating the different neuroprotective factors, particularly VEGF, limited the possibility of determining the effect of resistance exercise on this factor. Second, and in relation to the methodological quality of the studies, although in general the quality of the studies was acceptable, a large proportion of them did not provide information about some domains of the RoB 2.0., and the risk of bias was rated as high risk or some concerns. However, to overcome these limitations, sensitivity analyses were performed by eliminating each study included in the meta-analysis one by one. This same process was carried out with those studies in which the population had some pathology or health disorder to provide evidence of the robustness of the results. Third, there was a great heterogeneity of the interventions in terms of type of exercises, volume, frequency, and intensity; however, they have been pooled in a way that can provide conclusive results in relation to the doseresponse.

Conclusion
Our data support a neuroprotective effect of resistance exercise in middle and late life, mainly through its influence on IGF-1. Therefore, physical activity programs targeted to this population should emphasize the promotion of this type of training, with a frequency of at least 3 days/week, to mitigate the neurological and cognitive consequences associated with aging. Because of the scarcity of studies, more clinical trials are needed to consistently establish the neuroprotective effect of resistance exercise, particularly in neuroprotective factors that have not yet been sufficiently studied.

Authors' contributions
ERG participated in the design of the study and contributed to data collection and data reduction/analysis; ATC and CPM participated in the design of the study; DPM participated in the design of the study and