Physical activity, chronic kidney disease, and cardiovascular risk: A study in half a million adults

There is a growing prevalence of chronic kidney disease (CKD), a condition associated with a higher cardiovascular disease (CVD) risk. We assessed the association between self‐reported physical activity (PA) and CKD and also studied whether PA attenuates CKD‐associated CVD risk.


| INTRODUCTION
Chronic kidney disease (CKD), conventionally defined by an estimated glomerular filtration rate (eGFR) <60 mL/ min/1.73m 2 and/or present kidney damage markers for ≥3 months regardless of the underlying cause, is estimated to affect 9.1% of the global population, with a mortality toll of ~1.2 million people per year. 1 This condition is associated with a higher prevalence of major cardiovascular disease (CVD) risk factors (hypertension, diabetes, hypercholesterolemia, obesity) 2,3 together with an increased likelihood of CVD events. 1 Physical activity (PA), on the other hand, might play a role in the prevention and treatment of CKD, 4 with recent meta-analytical evidence for a trend toward a doseresponse, inverse association between PA levels and the incidence of CKD. 5 PA might also contribute to reduce the burden imposed by CKD. 4 A systematic review found that greater PA levels were associated with higher survival rates among individuals with CKD. 6 In the same line, a recent prospective study reported that higher PA levels were associated with a lower risk of all-cause mortality, end-stage renal disease, and CVD events among individuals with CKD, with the PA protective effect starting at ≥7.5 metabolic equivalents of task (MET)-h/week (which is equivalent to current international PA recommendations, that is, ≥150 min of moderate-intensity PA per week, ≥75 min of vigorous-intensity PA per week, or a combination thereof). 7Whether PA can mediate the association between CKD, and major CVD risk factors remains, however, unclear.
The aim of the present study was to determine the association between PA levels and CKD.Another purpose was to determine the association between PA levels and CVD risk factors in participants with CKD.

| Experimental design and participants
The study followed both a cross-sectional and a prospective observational design and is reported in accordance with the "Strengthening the Reporting of Observational Studies in Epidemiology" (STROBE) guidelines.A cohort of all Spanish economically active adults (18 to 64 years) insured by a large occupational risk prevention company (Quirónprevención) participated in the study.From 2012 to 2016, participants underwent routine (~1 per year) medical examinations as part of their health insurance coverage.No additional data were available from this time point.Cross-sectional analyses were performed using the first available medical examination ("baseline") for all participants during the study period.Prospective analyses were conducted from baseline to the last available examination in those participants with two or more assessments and meeting one of the following two criteria: (i) having no CKD at baseline with or without CVD risk factors (for the assessment of CKD incidence) or (ii) having CKD at baseline but no CVD risk factors (for the assessment of CVD risk incidence).Participants provided their oral consent, and the local ethics committee approved the protocol (CEI20-004).All procedures were performed in accordance with the Declaration of Helsinki and its amendments.

| Variables
All variables were assessed and registered by a clinician of the above mentioned occupational risk prevention company.Demographic and descriptive variables included date of examination, participant's age, sex, and socioeconomic status.Socioeconomic status-a combination of occupation variables, such as activity and professional status of the household-was obtained from the Spanish census.Thus, a territorial indicator was assigned to each record by merging census street map and territorial indicators information (Spanish census and individuals' code postal) as detailed elsewhere. 80][11] Briefly, participants filled a questionnaire about the frequency, duration and perceived intensity of leisure-time aerobic PA during a typical week, including domestic PA (light and heavy housework, gardening, and "do-it-yourself" tasks), walking (slow, average, brisk, or fast pace), and participation in sports and exercises (e.g., swimming, cycling, tennis).Moderate-intensity PA (i.e., eliciting an average energy expenditure of 3.0 to 5.9 MET) and vigorous-intensity PA (≥6.0 MET) were then identified. 11Accordingly, participants were considered "inactive" (performing neither moderate nor vigorous PA), "insufficiently active" (performing PA but not meeting the guidelines for aerobic PA established by the World Health Organization [WHO], i.e., <150 min/week and <75 min/ week of moderate and vigorous PA, respectively), 12 or "regularly active" (meeting WHO minimum guidelines, i.e., ≥150 min/week of moderate PA or ≥75 min/week of vigorous PA, or a combination thereof).
Participants' estimated glomerular filtration rate (eGFR) was calculated both at baseline and at the end of the follow-up from serum creatinine levels assessed after an overnight fast using the CKD epidemiology (CKD-EPI) collaborative equation (Table S1). 13Thus, CKD occurrence was defined as eGFR <60 mL/min/1.73m 2 . 14The relevant clinician also assessed the presence of major CVD risk factors including diabetes (i.e., medicated against diabetes or fasting glycemia ≥125 mg/dL 15 ); hypercholesterolemia (medicated against hypercholesterolemia or total blood cholesterol concentration ≥240 mg/dL 16 ); hypertension (medicated against hypertension or systolic [SBP]/ diastolic office blood pressure [DBP] ≥140/90 mmHg, with SBP and DBP determined following the guidelines of the European Society of Cardiology/European Society of Hypertension 17 ); and overweight/obesity (body mass index [BMI] ≥25.0 kg/m 2 ).

| Statistical analyses
Data are shown as mean ± standard deviation for continuous data and as percentages for categorical ones.We used logistic regression to determine the association between PA and both the prevalence (cross-sectional analysis) and incidence (prospective analysis) of CKD, with analyses adjusted by sex, age, socioeconomic status, and for the presence of CVD risk factors at baseline (i.e., hypertension, diabetes, hypercholesterolemia, and obesity).Logistic regression was also used to examine the association between CKD and the prevalence (cross-sectional analysis) of CVD risk factors, with analyses for each risk factor also adjusted by sex, age, socioeconomic status, and for the remaining CVD risk factors (e.g., the analysis for hypertension was adjusted by the presence of diabetes, hypercholesterolemia, and obesity).Logistic regression analyses were also performed among participants with CKD and no CVD risk factors at baseline to examine the association between the former and the incidence (prospective analysis) of CVD risk factors, adjusted by sex, age, and socioeconomic status.Given the controversy on the potential beneficial influence of excess BMI on CVD risk among individuals with CKD, 18,19 mediation analyses were performed to elucidate the role of BMI on the association between PA and CVD risk factors.Missing data were not imputed for eventual prospective analyses in those participants with only baseline assessment, in whom the pattern of missing data (i.e., whether missing completely at random, missing at random, or missing not at random) was determined.Statistical analyses were performed with Stata 14.0 (StataCorp, Texas, USA), and the level of significance was set at 0.05.

| RESULTS
Data from 517 917 participants could be included in crosssectional analyses, with 7% (n = 36 108) showing CKD.Participants' descriptive characteristics are displayed in Table 1.A greater proportion of men was observed among participants with CKD, who were older and showed a higher prevalence of hypertension, hypercholesterolemia, and physical inactivity (albeit with a lower mean BMI and a lower prevalence of overweight/obesity in unadjusted analyses) compared to their CKD-free peers.Prospective analyses (median follow-up of 2 years, range 2-5 years) could be performed in a subcohort of 264 581 participants, with missing data following a complete random pattern.During this time, 0.7% of the participants (n = 1734) without CKD developed this condition (characteristics of the participants included in prospective analyses attending to their CKD status shown in Table S2).None of the participants transitioned from having CKD to not having this condition during the follow-up.
In cross-sectional analyses at baseline, being regularly active-but not insufficiently active-was associated with a lower prevalence (−20%) of CKD compared with inactivity (Figure 1A).Yet, prospective analyses in individuals without CKD at baseline (n = 481 809) failed to confirm an inverse association between PA and the odds of developing CKD during the follow-up (Figure 1B).
Compared to participants without CKD, CKD was independently associated with a higher prevalence of diabetes (+5%) and hypertension (+3%) in cross-sectional analyses at baseline (Figure 2A).Prospective analyses among participants with CKD and no CVD risk factors at baseline (n = 264 581) confirmed that CKD was also associated with higher odds (+37%) of developing hypertension during the follow-up compared to participants without CKD, although no association was found for the remainder of CVD risk factors (Figure 2B).
Cross-sectional analyses revealed that, among individuals with CKD, being insufficiently, and especially regularly active was associated with a lower prevalence (ranging from −45% to −7%) of all CVD risk factors compared to inactive participants (Figure 3), albeit a higher prevalence of CVD risk factors was found in physical active people with CKD than in their CKD-free, physically active peers (Figure S1).In the same line, prospective analyses within people with CKD confirmed that being insufficiently, and especially regularly active significantly reduced (by −38% to −4%) the odds of developing all CVD risk factors during the follow-up compared with inactive individuals (Figure 4).On the other hand, the benefits of PA on CVD risk factors among participants with CKD were not significantly mediated by BMI (Figure S2).

| DISCUSSION
The present study indicates that individuals without CKD are more physically active than those with this condition, although no association was found between PA and incident CKD in the middle term (2-year follow-up).It is worth noting, however, that although CKD was overall associated with CVD risk factors (i.e., greater prevalence of hypertension and diabetes at baseline, and higher incidence of hypertension at follow-up), PA attenuated this association, with active individuals with CKD presenting with a considerably lower incidence and prevalence of CVD risk factors than their inactive peers with the same condition.
Previous studies have reported an inverse association between PA and incident CKD, as confirmed by a recent meta-analysis reporting that each 10 MET-h/week of PA were associated with a 2% reduction in CKD risk. 5otably, Parvathaneni et al. 20 recently reported that individuals meeting WHO recommendations of aerobic PA (≥150 min/week) had a 7% lower incidence of CKD during a 24-year follow-up, with a further risk reduction (−11%) in those exceeding these recommendations (>300 min/week).Another study found that higher selfreported PA levels were associated with a lower risk of incident CKD (−6% and −9% in moderately and highly active individuals, respectively) during a ~4year follow-up. 21In turn, although in the present study we found a lower prevalence of CKD among physically active participants, no consistent association was found between PA and incident CKD, which might be due, at least partly, to the short duration of the follow-up (median of 2 years).Longer follow-ups might therefore be needed to confirm the present findings.
Although not confirmed in the present study, several biological mechanisms might support an inverse association between PA and incident CKD.Notably, PA increases nitric oxide availability while attenuating inflammation and oxidative stress, thereby preventing vascular endothelial dysfunction. 22,23Moreover, PA is inversely associated with the risk of several comorbidities associated with the development and progression of CKD, such as obesity, diabetes or hypertension-as supported in fact by the present findings. 9,24In our study, indeed, CKD was associated with CVD risk factors (prevalence of hypertension and diabetes in crosssectional analyses, and incidence of hypertension in prospective analyses), whereas PA levels attenuated this association.Thus, among those participants with CKD, the odds of presenting or developing CVD risk factors was markedly lower in insufficiently active and active individuals compared with their inactive peers.In this Regularly active (%) 20  28   Note: CKD was considered when estimated glomerular filtration rate was <60 mL/min/1.73m 2 . 13etween-group comparisons were performed with the Student's unpaired t-test (for continuous data) or the chi-square test (for categorical data).Regular alcohol drinker: ≥1 drink/day.
T A B L E 1 Descriptive characteristics of study participants at baseline.regard, although a dose-response association was not evident, our prospective analyses suggest a slightly lower incidence of most CVD risk factors among active participants compared with insufficiently active individuals (reduction of 20% vs 14%, 38% vs 36%, and 31% vs 27%, respectively, of the odds of developing hypertension, diabetes, and obesity in active and insufficiently active individuals).Of note, a reduced incidence of major CVD risk factors was observed despite reductions in BMI, even if some reports suggest that a high BMI (i.e., obesity) might provide beneficial effects among individuals with CKD. 18,19Furthermore, the prevalence of CVD risk factors was lower in active individuals with CKD that in their inactive peers without this condition, thereby suggesting that PA levels might be a stronger moderator of CVD risk than the sole presence of CKD.In line with our findings, a recent cohort study showed that individuals meeting PA recommendations had a lower incidence of CVD events and mortality compared with inactive ones during a ~13year follow-up, and even lower values were observed in the most active individuals (e.g., reduction of −43% and −53% in CVD-mortality risk in individuals meeting WHO recommendations and in the most active ones, respectively). 25In the same line, meta-analytical evidence shows that exercise interventions not only improve kidney function among patients with CKD, but also lead to improvements in CVD risk profile (i.e., reductions in blood pressure and body weight). 26ome limitations of the present study should be acknowledged, notably the short follow-up duration (with data only recorded up to year 2016) and the missingness of data at follow-up (largely due to participants moving to a new company, with subsequent changes in risk prevention company).The use of self-reported PA-instead of objective measures such as accelerometers-can also be considered a limitation.Moreover, the lack of information on major potential confounding variables such as diet, medication, as well as the lack of data on major CVD events or mortality can also be considered as study limitations.It is also worth noting that the cross-sectional associations observed between PA levels and incident CKD, [27][28][29] or between CKD and CVD risk factors, 30,31 can be bidirectional and not necessarily reflect a cause-effect association.However, our prospective analyses seem to confirm an association between CKD and the risk of developing hypertension.In turn, the large sample size analyzed-one of the largest studies on this issue thus far-is a major strength.
F I G U R E 3 Cross-sectional association between physical activity levels and the prevalence of major cardiovascular disease risk factors in individuals with chronic kidney disease (CKD).Data are odds ratio and 95% confidence interval, with the reference group being inactive participants with CKD.Analyses were adjusted by age, sex, and socioeconomic status.Analyses for each cardiovascular disease risk factor were additionally adjusted for the remaining risk factors.
The present findings show lower self-reported PA levels among adults with CKD than in their peers without this condition, although being physically active was not associated with a lower risk of developing CKD in the middle term (~2 years).Further research is therefore needed to confirm whether PA can actually protect from CKD in the longer term.On the other hand, although CKD was associated with a higher prevalence (hypertension, diabetes) and incidence (hypertension) of CVD risk factors, being physically active markedly attenuated this association.The latter findings might reinforce the role of PA for the overall management of CKD, or at least for attenuating the potential CVD burden associated with this condition.