Physical Activity Alters Inflammation in Older Adults by Different Intensity Levels

Abstract

Purpose 

To examine the influence of reallocating time spent at different objectively measured physical activity (PA) behaviors on markers of systemic inflammation in older women with different levels of metabolic risk.

Methods 

Accelerometer-based monitoring of PA was conducted in a population of community-dwelling older women (n = 111; age, 65–70 yr) for determination of daily sedentary time, time in light PA (LPA) and moderate-to-vigorous PA (MVPA). Blood samples were collected for the assessment of the systemic inflammatory markers C-reactive protein (CRP), fibrinogen, and adiponectin. Metabolic risk was assessed by standardized procedures based on definitions for the metabolic syndrome. Data were analyzed by linear regression models based on isotemporal substitution analysis.

Results 

Reallocating 30 min of sedentary time with either time in LPA (β = −0.47; P < 0.05) or MVPA (β = −0.42; P < 0.05) was related to reduced fibrinogen level, whereas no corresponding effect was evident when shifting time in LPA with time in MVPA, while holding sedentary time constant. In contrast, reallocating a 30-min period in sedentary (β = −0.70; P < 0.01) or LPA (β = −0.71; P < 0.01) with MVPA was associated with a significant reduction in CRP level, whereas no impact on CRP was observed when a period of sedentary behavior was replaced with LPA. Importantly, all significant influences on fibrinogen and CRP by displacement of different PA behaviors remained after adjustment for metabolic risk status among participants. No significant associations with adiponectin were observed.

Conclusions 

Altogether, this work supports the existence of different intensity thresholds mediating beneficial effects of PA on important clinical markers of systemic inflammation in older women across different stages of disease prevention.

The occurrence of low-grade chronic systemic inflammation in older adults (inflammaging) is an event contributing to the development of many age-related diseases (¹). Elevated levels of circulating markers of inflammation have been associated with increased risk of diabetes and presence of the METs (²), which represents a cluster of metabolic disorders including abdominal obesity, hypertension, dyslipidemia and hyperglycemia (³). The acute phase proteins C-reactive protein (CRP) and fibrinogen are two laboratory markers commonly used to study the influence of systemic inflammation on age-related diseases (^4,5). Opposed to the action of these two proinflammatory mediators, the adipokine adiponectin is suggested to have anti-inflammatory properties providing protection against metabolic disorders (⁶). Importantly, it has been shown that levels of CRP and fibrinogen are linked to increased CVD risk in both men and women (⁷), an association suggested to be partly mediated by endothelial cell dysfunction (⁸). Notably, the strength of relationships between levels of CRP or fibrinogen on various risk factors of disease and mortality may differ between men and women. For example, less positively graded associations between CRP and mortality have been reported in women compared with men (⁹). In contrast, a twice as strong relationship between fibrinogen level and body mass index has been shown in females compared with males (¹⁰).

Although the influence of physical activity (PA) on metabolic disorders is well established, there is currently a growing attention to its impact on systemic markers of inflammation. Times spent in PA of both light PA (LPA) and moderate-to-vigorous intensity PA (MVPA) as well as in sedentary (^11–15) have previously been associated to systemic inflammation. However, the influence of separate dimensions of PA behaviors (eg. time in MVPA) on inflammation has typically been reported without accounting for the fact that time spent in one activity behavior automatically displaces time spent in another during a day. Moreover, controversies remain regarding dose–response relationships between PA behaviors and different markers of systemic inflammation, where a few (^12,15) but not all (¹⁶) studies suggested that time spent in LPA is associated with reduction in systemic inflammation. Therefore, further investigation of the impact of displacing a given activity behavior with another on systemic inflammation is warranted as such knowledge can infer important public health implications. In this respect, the isotemporal substitution model has recently been employed to investigate health impacts of shifting periods in one PA behavior with another while holding total time constant. Although effects of replacing different PA behaviors on all-cause mortality (¹⁷), cardiovascular risk (¹⁸), and clustered metabolic risk (¹⁹) have been reported in older populations, only one study has recently addressed effects of decreased time in sedentary behavior on inflammation in a middle-age population (²⁰). Nevertheless, knowledge on effects of reallocating time in different PA behaviors on systemic inflammation is lacking in older adults, specifically. Furthermore, PA levels are known to decrease with advancing age, where elderly women spend less daily time in MVPA compared with men (²¹), which makes elderly women in particular a vulnerable group to disease development. Therefore, increased knowledge on how PA behaviors can influence on inflammatory status in elderly women is warranted. Furthermore, varying health status of study participants may moderate potential relationships between PA behavior and level of inflammation. Elderly at risk of cardiovascular and metabolic diseases will likely exhibit an elevated inflammatory status compared to their healthier peers. This in turn may explain contrasting findings whether levels of systemic inflammation are truly age-dependent or rather simple reflections of varying health status (^22,23).

Given paucity of data on relationships between PA behavior and systemic inflammation, encompassing both proinflammatory and anti-inflammatory markers, the aim of this work was to determine the influence of reallocating objectively assessed time in PA behaviors on the acute phase proteins CRP and fibrinogen, and the adipokine adiponectin in older community-dwelling women at different levels of metabolic risk.

METHODS

Participants

A total of 120 community-dwelling older women (65–70 yr) were recruited through advertisement in a local newspaper. Participants were living in an urban area in Sweden and were free of diagnosed coronary heart disease, diabetes mellitus, had no disability with respect to mobility and were nonsmokers. All procedures were conducted according to standards set by the Declaration of Helsinki. Written informed consent was obtained from all participants and the study was approved by ethical committee.

Assessment of physical activity and sedentary behaviors

Accelerometer-based assessment of PA during a week was performed using the Actigraph GT3x activity monitor (Actigraph, Pensacola, Florida) as previously described (¹⁹). In brief, participants wore the monitor during all waking hours (except water activities) and noted nonwear time during awake time. A minimum of 4 d with at least 10 h of wear time per day was required for inclusion. Nonwear time was defined as a minimum of 60 min of continuous zero counts. Counts of >20.000 per minute were discarded as nonhuman movements. Average daily times spent in LPA (100–2019 counts per minute) and MVPA (≥2020 counts per minute) as well as in sedentary (<100 counts per minute) were retrieved (²⁴).

Laboratory measurements and assessment of metabolic risk factors

Blood samples collected by venipuncture were obtained between 7:00 am and 9:00 am after an overnight fast. High-sensitivity CRP (Hs-CRP) was measured using a fully automated immunoturbidimetric assay (Advia 1800; Chemistry System, Siemens, Germany). Fibrinogen was determined using an automated immunoassay method with a polyclonal rabbit antihuman antibody (Dako, Glostrup, Denmark), and adiponectin was assessed using a commercially available ELISA kit (Mercodia, Sweden). Metabolic risk factors included in the IDF definition of the METs (³) were determined. Fasting blood triglycerides and HDL-cholesterol were assessed using chemistry kits from Ortho-Clinical Diagnostics on a Vitros-5.1 analyser platform (Clinical Diagnostics, Raritan, NJ) and blood glucose was assessed using the Reflotron Plus® system (Roche Diagnostics Limited, Rotkreuz, Switzerland). Waist circumference was measured with a steel tape using standardized procedures. Systolic and diastolic blood pressure were measured manually after a 15-min rest in the supine position using a mercury sphygmomanometer. In accordance with the IDF definition of METs, participants were classified as having a high (with METs) or low (without METs) metabolic risk.

Statistical analysis

Data are presented as means ± SD. Differences between high and low metabolic risk groups were determined by one-way ANOVA. The hypothetical change in inflammatory markers when reallocating time spent in PA of a given intensity with a corresponding time frame in PA of another intensity, while holding total time constant, was performed using isotemporal substitution modeling (²⁵). In line with recent work (²⁰), PA and sedentary behaviors were modeled in 30-min periods. To retrieve comparable effect outcomes from variables with different original units of measurement, all variables on inflammatory markers were standardized (z-scores) before regression modeling. Due to skewness, data on CRP were log transformed before standardization. Before selection of inflammatory variables included in isotemporal models, linear associations between each inflammatory marker and times spent in any PA and sedentary behavior were assessed. All models were adjusted by age and accelerometer wear time, whereas isotemporal models were additionally adjusted by differences in metabolic risk (high/low risk) among participants. All models were analyzed using the whole sample as no interaction effects between PA behaviors and metabolic risk groups on inflammatory outcomes were observed. Assumptions for regression models including linearity, homoscedasticity, and multicollinearity between independent variables were checked. Level of significance was set to P < 0.05, which based on our study sample allows detection of smaller effect sizes (≤0.15) with a power of ≥80% when performing all regression models. All statistical analyses were performed using SPSS version 24.

RESULTS

A total of 111 older women with complete data on all variables were included in the final analysis. Nine women had either incomplete accelerometer recordings or missing biological data. Participants wore the accelerometer for 5.8 ± 0.5 d, with a mean daily wear time of 14.2 ± 1.0 h. Based on the IDF definition of METs, 57% and 43% were classified in the low and high metabolic risk groups, respectively. Notably, women with higher metabolic risk had significantly higher levels of CRP and fibrinogen, and lower level of adiponectin, and spent less time in MVPA compared with those with lower metabolic risk (Table 1).

TABLE 1: Subject characteristics (mean ± SD) in all women and stratified by high and low metabolic risks.

Before isotemporal modeling, level of fibrinogen was significantly associated to time spent in all three dimensions of PA and sedentary behaviors, whereas level of CRP was exclusively associated with time in MVPA (Table 2). In contrast, no associations were observed between adiponectin and any of the three dimensions of PA and sedentary behaviors, and therefore, no further analysis on adiponectin was performed (Table 2).

TABLE 2: Associations (β-coefficients, 95% CI) between single dimensions of daily times spent in sedentary, LPA and MVPA, and z scores of CRP, fibrinogen, and adiponectin.

Effects of isotemporal displacement of PA behaviors on CRP and fibrinogen are presented in Table 3. Reallocation of a 30-min period of daily sedentary time with a corresponding period of either LPA (β = −0.47; P < 0.05) or MVPA (β = −0.42; P < 0.01) was related to a significant reduction in fibrinogen level. Interestingly, no significant effect on fibrinogen level was evident when reallocating a 30-min period of LPA with time in MVPA, while holding sedentary time constant. In contrast, reallocating a 30-min period of sedentary (β = −0.70; P < 0.01) or LPA (β = −0.71; P < 0.01) with MVPA was associated with a significant reduction in CRP level, whereas no impact on CRP was observed when reallocating a period of sedentary time with LPA. Importantly, the significant influences on fibrinogen and CRP by displacement of different PA behaviors remained unaltered even when adjusting models with differences in metabolic risk (Table 3). Notably, we also checked the effect of adjusting models with BMI, which did not significantly alter observed outcomes.

TABLE 3: Associated change (β-coefficients, 95% CI) in CRP and fibrinogen when replacing a 30-min time block of sedentary time or time in LPA for higher intensity behaviors, with and without adjustment for metabolic risk status in older women (n = 111).

DISCUSSION

Our study demonstrates that reallocation of daily sedentary time with time spent in PA of different intensities has a significant impact on proinflammatory acute-phase proteins in older women. A key finding is that a hypothetical displacement of time in PA of a given intensity with another impacts differently on levels of fibrinogen and CRP, suggesting different thresholds of PA intensity necessary to infer health-enhancing effects on systemic inflammation. Furthermore, these findings remained evident in older women regardless of metabolic risk, which emphasizes the beneficial health effects of PA in older women across different stages of disease prevention.

To the best of our knowledge, no study has previously explored the influence of reallocating different PA and sedentary behaviors on the acute-phase proteins CRP and fibrinogen in older adults. While influences of PA and sedentary behaviors on levels of CRP and fibrinogen have been previously indicated (^{11–13,15,26–29}), many studies are based on self-reported PA behavior, which hampers ability to determine the influence of different PA intensities on systemic inflammation. Moreover, previous investigations based on objective assessments have typically focused on separate dimensions of PA behaviors, without considering that increased amount of time in one behavior inevitably infers reduced time in another. Notably, favorable effects of hypothetical time reductions in sedentary behavior on C3 complement component, IL-6 and white blood cell count were recently reported in a population of middle-age adults (²⁰). Thus, our findings extend beyond currently available data and reveal that reallocations of time in different PA behaviors impact on systemic inflammation in older women, which is in line with recent studies reporting on effects of reallocating time in PA and sedentary behaviors on clustered metabolic (¹⁹) and cardiovascular risk (¹⁸).

Here we show that hypothetically reducing daily time in sedentary behavior for more active pursuits is related to a reduced fibrinogen level. In contrast, reduced CRP level is exclusively related to increased time in MVPA, with no beneficial effects of reducing sedentary time in favor of LPA time. In fact, when holding sedentary time constant, engaging in less LPA time in favor of MVPA was still related to a reduced CRP with no corresponding effect on fibrinogen. Therefore, it may be hypothesized that liver secretion of different acute phase proteins is PA-intensity sensitive, where spending time in MVPA will infer alterations in liver secretion of CRP through an adipose tissue-dependent modulation of cytokine release. Indeed, circulating CRP levels are strongly linked to adipose tissue through adipocyte expression of various cytokines, including IL-6 and tumor necrosis factor-α (³⁰). Alternatively, PA may modulate hepatic CRP release independently of adiposity. For example, a recent study showed an inverse relationship between time in MVPA and IL-6 levels that remained after controlling for amount of visceral fat (³¹). In our study, the fact that MVPA was related to CRP levels regardless of metabolic risk and thus differences in waist circumference may support an adiposity-independent pathway through which PA is linked to CRP. In contrast to findings on CRP, reduction in fibrinogen level occurred already in the transition from sedentary behaviors to LPA. Fibrinogen plays an important role in the coagulation cascade, including platelet aggregation, which affects blood viscosity and thus influences on cardiovascular function (¹⁰). The transition from sedentariness to PA automatically increases tissue blood flow and is accompanied by cardiovascular protective effects. Thus, reducing time in sedentary behaviors in favor of PA will have beneficial effects on fibrinogen, where even lighter-intensity activities will infer hemodynamic changes likely leading to cardiovascular protection.

As expected, elevated levels of proinflammatory acute-phase proteins were observed in older women at higher metabolic risk compared to those at lower risk, which likely reflects pathophysiological links between inflammatory status and stages of metabolic disease development (³²). Though biological sex may impact on effect size, our observations are in line with previously reported associations between CRP and fibrinogen levels with various risk of disease observed in both men and women (^9,10). Importantly, our data clearly show that the impact of hypothetically shifting times spent in different PA behaviors on acute phase proteins occurs regardless of metabolic risk status, indicating that a health-enhancing effect of PA is not limited to individuals within a specific stage of disease risk. Furthermore, although time spent in lighter PA intensities can alter levels of some acute phase proteins, our findings support current guidelines recommending PA of at least moderate intensity as the threshold required for more substantial health effects in older adults.

In agreement with previous studies indicating the detrimental influence of low adiponectin level on development of metabolic disorders (³³), we show that compared to having low metabolic risk, women with high metabolic risk had lower levels of circulating adiponectin. Importantly, a link between time in different PA behaviors and the anti-inflammatory adipokine adiponectin could not be detected in our study. Interestingly, although an association between adiponectin and time in MVPA has recently been reported, further adjustment by visceral fat mass attenuated this observation (³¹), suggesting a strong influence of adipose tissue on circulating adiponectin. Given the inverse relationship between circulating adiponectin and fat mass, one would expect that a PA-mediated reduction of fat mass would lead to increased adiponectin level. However, a compilation of studies has reported inconsistency in effects of aerobic exercise on circulating adiponectin across samples of various health status (³⁴). Previous studies have shown that the exercise-induced increase in adiponectin can occur both with (³⁵) and without (³⁶) concomitant loss of fat mass, which suggests that circulating levels of adiponectin are not reflected by fat mass alone. Instead, originating from the adipocyte, adiponectin is likely to be more directly influenced by alterations in adipocyte function rather than PA-induced general reductions of total fat mass. Further research is warranted to elucidate mechanisms by which PA may act on circulating adiponectin.

Limitations of this study include the cross-sectional design, which prevents from making inference on causality, and the older women included in the present study may not be representative of a broader population of older adults. Moreover, sleep duration was not included when modeling associations between PA and inflammatory biomarkers. Of note, a previous study based on isotemporal reallocations of time in different PA behaviors showed that differences in sleep durations did not alter observed relationships with cardiovascular disease risk biomarkers (¹⁸). It should also be acknowledged that although markers determined in our study are among the most documented indicators of systemic inflammation, the complexity of inflammatory status is likely explained by additional biomarkers not addressed in the present work. Thus, favorable effects of PA on inflammatory status may be mediated through alterations in biomarkers other than those used in this study.

CONCLUSIONS

In conclusion, this work supports the existence of different intensity thresholds mediating beneficial effects of PA on important clinical markers of systemic inflammation in older women across different stages of disease prevention. Further experimental evidence is warranted to confirm the study findings, which could prove instrumental for development of PA-based health-enhancing strategies in the older population.