Association of kidney function with physical performance: the Maastricht study

This is a cross-sectional analysis of data derived from the Maastricht study population. The Maastricht Study is a prospective, population-based, cohort study performed in the southern part of the Netherlands. The methodology of the study has been previously described [15]. In brief, the Maastricht Study follows an extensive phenotyping approach, aiming to examine the pathophysiology and complications of type 2 diabetes mellitus. Recruitment of individuals aged between 40 and 75 years was performed with stratification based on diabetes mellitus status, with oversampling of those with known type 2 diabetes mellitus. Participants were recruited using mass media campaigns, municipal registries and the regional Diabetes Patient Registry via mailings. This analysis is based on participants who completed the baseline survey (November 2010 to September 2013) providing available data regarding both kidney and physical function. All individuals were fully informed about the study procedures and provided written informed consent. The study procedures were ethically approved by the institutional medical ethical committee (NL31329.068.10) and the Netherlands Health Council (Permit 131,088–105234-PG).

Participants’ physical performance was quantified using the following tests: 6-min walk test, timed chair rise stand test, hand grip strength and isometric strength tests of elbow and knee flexors and extensors.

For the 6-min walk test, participants were instructed to walk at a fast pace without running for 6 min, making turns around two cones, placed 20 m apart in a hallway. Every minute, participants were encouraged to continue walking. After the end of 6 min or when the individual was unable or unwilling to continue, the test was terminated and the covered distance was measured (in meters). For the analyses, the covered distance (m) and the gait speed (s) were captured. For the six-minute walk test the following exclusion criteria applied: history of cardiovascular disease 3 months prior to the test, serious hypertension (systolic blood pressure ≥ 180 mmHg and/or diastolic blood pressure ≥ 110 mmHg), tachycardia (heart rate ≥ 110/min) and mobility limitation, such as the need of a walker or any other condition precluding independent walking.

For the timed chair rise stand test, a 46 cm-high chair with a straight back without armrests was used. Participants were instructed to sit in the middle of the chair, keeping their feet flat on the floor and their arms crossed over the chest. The test required participants to come to a full standing position and sit down again without the support of their arms at the fastest possible pace. For the analysis, the time in seconds needed for the completion of 10 repetitions was measured to one decimal place [16, 17].

Regarding the measurement of handgrip strength, the Jamar handheld dynamometer (SEHAN Corp., Korea-Biometrics Europe BV, Almere) was used. Participants were required to stand straight against a wall, having their upper arm along the trunk with their elbow flexed at 90°. Subsequently, they were instructed to strongly squeeze the dynamometer for 3–5 s, while standardized encouragement was provided. Three measurements were performed for each hand and the maximum strength measurement was recorded in kilograms.

Isometric muscle strength of knee and elbow extensors and flexors was evaluated using a customized set-up with two dynamometers (Futek LSB302, FUTEK Advanced Sensor Technology Inc., Irvine, CA, USA) and was recorded with the M-PAQ (Maastricht Instruments, Maastricht, the Netherlands). Individuals were excluded in case of recent injury or surgery of the right arm or leg. Straps connected to the dynamometer were secured 2 cm above the lateral malleolus and 2 cm proximally from the wrist, with the axis of the dynamometer corresponding to the knee-joint and the elbow-joint axis, respectively. Participants were instructed to flex and extend their right knee and elbow powerfully for 5 s. Three measurements were performed, while the generated force was visualized on a monitor. For the analysis, the joint torques were calculated by multiplying the force measured by the dynamometer by the corresponding moment-arm, defined as the distance between the dynamometer strap and the joint rotation point. The maximal joint torques out of the three trials for elbow and knee flexion and extension were captured in Nm.

The estimated glomerular filtration rate (eGFR) was calculated with the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, which is based on the combination of serum creatinine and serum cystatin-C values (eGFR_Cr-Cys) [18]. In case serum cystatin-C measurements were not available, the creatinine-based CKD-EPI equation was used (eGFR_Cr). For the analyses, eGFR was classified as follows: < 60, 60 to 90 and > 90 ml/min/1.73 m². Renal hyperfiltration was defined as an eGFR exceeding the age-adjusted threshold that was calculated as follows: 130 ml/min/1.73 m² minus 1 ml/min/1.73 m² per year after the age of 40 years [19]. Urinary albumin excretion was measured by two 24-h urine collections. Valid collections were considered those with collection times between 20 and 28 h, with extrapolation to the 24-h excretion performed when the collection time was not exactly 24 h. Urinary albumin excretion was categorized as < 15 mg/24 h, 15–30 mg/24 h and > 30 mg/24 h. According to the KDIGO (Kidney Disease: Improving Global Outcomes) definitions, eGFR values < 60 ml/min/1.73 m² indicated a moderate to severe kidney function decrease (stage ≥ G3a), while urinary albumin excretion > 30 mg/24 h implied moderately to severely increased albuminuria (stage ≥ A2). On the other hand, eGFR > 90 ml/min/1.73 m² indicated normal to high eGFR[20].

Self-reported information was recorded with the use of questionnaires regarding age, sex, ethnicity (Caucasian/other), educational level, smoking habits and total alcohol consumption (in g per day). More specifically, low education referred to primary or lower vocational education. Medium education was defined as intermediate vocational or higher secondary education, while the high level of education corresponded to higher vocational or university education. High alcohol consumption was defined as the intake of more than 7 g and 14 g of alcohol per day for females and males, respectively. The status of diabetes mellitus was assessed by a 7-point 75 gr oral glucose tolerance test. Glucose metabolism was categorized as normal, prediabetes, or type 2 diabetes according to the 2006 World Health Organization criteria [21]. Individuals on anti-diabetic medications without a history of type 1 diabetes were classified as having type 2 diabetes. Daily energy intake (in kcal/day) was estimated by a validated food frequency questionnaire [22], body mass index was estimated by dividing weight (in kilograms) by the square of height (in meters) and waist-to-hip ratio by dividing waist by hip circumference (in cm). Information about mobility status was captured by the 36-Item Short Form Health Survey questionnaire [8], defining mobility limitation as the difficulty in climbing one flight of stairs or walking 500 m. Fasting blood samples were collected to assess the lipid profile of participants. To evaluate office blood pressure, three measurements were performed on the right arm after a 10-min rest and the average systolic and diastolic blood pressure values were recorded. Hypertension was defined by one of the following criteria: average systolic blood pressure ≥ 140 mmHg, average diastolic blood pressure ≥ 90 mmHg [23], or use of anti-hypertensive medications.

Statistical analysis was performed in R-4.0.5. Histograms were visually assessed to examine the distribution of continuous variables. Normally distributed variables were described by their mean and standard deviation; otherwise, the median and interquartile range (IQR) were reported. Comparisons of continuous variables across 3 groups were planned to be conducted using the one-way analysis of variance or the non-parametric Kruskal Wallis test, depending on data normality. Categorical variables were reported as proportions and were statistically compared using the chi-square test or the Fisher’s exact test in case the assumptions of the former were not fulfilled.

Scatterplots were constructed to assess the linearity of association between kidney and physical function markers. To assess non-linear relationships, restricted cubic splines regression analysis was performed with 4 knots placed at the 5th, 35th, 65th and 95th percentiles. In linear regression models, eGFR and urinary albumin excretion were treated as categorical variables, while physical function markers as continuous ones. Participants with eGFR 60–90 ml/min/1.73 m² and urinary albumin excretion < 15 mg/24 h served as the reference groups. Multivariable linear regression models were constructed aiming to take into consideration potentially important covariates, such as basic demographic data, variables that could possibly impact physical function, as well as diabetes mellitus status to account for oversampling. Adjustment for mobility limitation was performed to account for physical disability that would interfere with the proper evaluation of physical performance tests. Specifically, model 1 adjusted sequentially for age, sex, ethnicity (Caucasian or other), educational level (low, medium or high) and diabetes mellitus status (normal glucose metabolism, prediabetes, type 2 diabetes mellitus or other diabetes), model 2 adjusted additionally for body mass index, smoking (never, former or current smoker), mobility limitation and alcohol consumption and model 3 additionally adjusted for systolic blood pressure. Albuminuria models also adjusted for eGFR. Moreover, as an additional analysis, the potential association of physical performance markers with renal hyperfiltration (age-adjusted eGFR > 130 ml/min/1.73 m²) was tested. Participants with renal hyperfiltration were compared to normofiltering ones (age-adjusted eGFR: 60–90 ml/min/1.73 m²). Subgroup analysis was performed by separately analyzing the association of physical function markers with eGFR_Cr-Cys and eGFR_Cr. In addition, stratified analyses were performed based on sex and diabetes mellitus status, testing for the presence of potential interactions. Sensitivity analysis was also conducted by repeating the regression analysis in patients without hypertension and those without dyslipidemia requiring lipid-modifying medications. Statistical significance was defined by a two-sided p-value threshold of 0.05.

The selection of participants is schematically illustrated in Fig. 1. Overall, 296 individuals were completely excluded from the analysis due to missing data about kidney function (n = 91), baseline covariates (n = 182) and physical function tests (n = 20); hence, a total of 7,396 participants were included in the present analysis. The clinical and demographic characteristics of included and excluded individuals are compared in Suppl. Table 1. Kidney function was assessed by the CKD-EPI eGFR_Cr-Cys equation in 3,249 participants with available cystatin-c measurements.

Table 1 presents the main demographic, clinical and physical function characteristics of the included participants, stratified by eGFR categories. The median age of the study population was 61 years, while 50.5% of participants were males. Concerning kidney function, 5.8% of participants had an eGFR < 60 ml/min/1.73 m² and 8.4% of participants had a urinary albumin excretion > 30 mg/24 h. The median eGFR was 84.70 ml/min/1.73 m², while hyperfiltration was observed in 86 participants. More specifically, eGFR 45–60 ml/min/1.73 m² was present in 62 individuals, while 8 participants had an eGFR < 30 ml/min/1.73 m². Individuals with lower eGFR values were older, had a slight male predominance, had a higher waist-to-hip ratio, as well as higher rates of hypertension, diabetes mellitus and dyslipidemia. Moreover, participants with lower eGFR covered significantly shorter 6-min test distance (p-value < 0.001), had worse timed chair rise stand test time (p-value: 0.016) and lower maximal grip strength (p-value < 0.001). Regarding albuminuria, 570 participants presented microalbuminuria, 50 had 24-h albumin excretion over 300 mg/24 h and 2 presented nephrotic-range albuminuria (> 3000 mg/24 h).

The relationship between eGFR and 6-min test distance, gait speed, timed chair rise stand test time and maximal grip strength is depicted in Fig. 2. A non-linear relationship was observed between eGFR, 6-min test distance, gait speed and maximal grip strength, with both low and very high eGFR values being associated with worse physical performance. Multivariable regression analysis (Table 2) indicated that eGFR > 90 ml/min/1.73 m² was associated with significantly shorter 6-min walk distance (β: -6.13 m, 95% confidence intervals-CI: -9.44 to -2.82, model 3) and slower gait speed (β: − 0.02 m/s, 95% CI − 0.03 to − 0.01, model 3), compared to eGFR 60–90 ml/min/1.73 m². In addition, after adjusting for all covariates, eGFR < 60 ml/min/1.73 m² was linked to significantly shorter 6-min walk distance (β: -13.04 m, 95% CI − 19.95 to − 6.13), slower gait speed (β: -0.04 m/s, 95% CI − 0.06 to − 0.02), worse timed chair rise stand test time (β: 0.91 s, 95% CI 0.36 to 1.47) and lower maximal grip strength (β: − 0.83 kg, 95% CI − 1.50 to − 0.15). Figure 3 illustrates the association of eGFR with isometric muscle strength of flexors and extensors of arms and legs, indicating that lower eGFR values correlate to worse isometric muscle strength. Multivariable regression analysis demonstrated that eGFR < 60 ml/min/1.73 m² was linked to significantly lower elbow flexion strength (β: − 3.64 Nm, 95% CI − 7.11 to − 0.16), compared to eGFR 60–90 ml/min/1.73 m².

The association of renal hyperfiltration with physical function markers is exhibited in Suppl. Table 2. The presence of renal hyperfiltration was linked to significantly shorter 6-min walk distance (β: − 18.98 m, 95% CI − 33.09 to − 4.87) and slower gait speed (β: − 0.05 m/s, 95% CI − 0.09 to − 0.01). Subgroup analysis (Suppl. Table 3) indicated that eGFR_Cr > 90 ml/min/1.73 m² was inversely associated with 6-min walk distance, gait speed, maximal grip strength, knee flexion and extension strength, while eGFR_Cr < 60 ml/min/1.73 m² was linked to shorter 6-min walk distance, slower gait speed and worse timed chair rise stand test time. On the other hand, compared to eGFR_Cr-Cys 60–90 ml/1.73 m², eGFR_Cr-Cys > 90 ml/min/1.73 m² was associated with significantly lower maximal grip and knee extension strength, while eGFR_Cr-Cys < 60 ml/min/1.73 m² was associated with significantly shorter 6-min walk distance, lower gait speed, elbow flexion and knee extension strength.

eGFR_Cr-Cys > 90 ml/min/1.73 m² was inversely associated with maximal grip strength and knee extension strength, while eGFR_Cr-Cys < 60 ml/min/1.73 m² was inversely associated with 6-min walk distance, gait speed, elbow flexion and knee extension strength.

Multivariable regression analysis (Table 2) showed that urinary albumin excretion 15–30 mg/24 h was associated with shorter 6-min walk distance (β: − 5.84 m, 95% CI − 11.35 to − 0.33, model 3) and slower gait speed (β: − 0.02 m/s, 95% CI − 0.03 to − 0.00, model 3), compared to albuminuria < 15 mg/24 h. After adjusting for all covariates, urinary albumin excretion > 30 mg/24 h was linked to significantly shorter 6-min walk distance (β: − 12.48 m, 95% CI − 18.28 to − 6.68), slower gait speed (β: − 0.03 m/s, 95% CI − 0.05 to − 0.02), worse timed chair rise stand test time (β: 0.51 s, 95% CI 0.11–1.06), lower maximal grip (β: − 1.34 kg, 95% CI − 1.91 to − 0.76) and elbow flexion strength (β: − 3.31 Nm, 95% CI − 5.80 to − 0.82). No significant association was observed between albuminuria and elbow extension, knee flexion and extension strength.

The outcomes of sensitivity analyses are presented in Table 3. Among participants without hypertension, eGFR > 90 ml/min/1.73 m² was inversely associated with 6-min walk distance, gait speed and knee flexion strength, while urinary albumin excretion > 30 mg/24 h was linked to significantly lower knee flexion strength. Among individuals without dyslipidemia, eGFR > 90 ml/min/1.73 m² was inversely associated with 6-min walk distance, gait speed and knee extension strength, while eGFR < 60 ml/min/1.73 m² was linked to significantly lower knee flexion strength. Additionally, in this subpopulation, albuminuria > 30 mg/24 h was associated with significantly shorter 6-min walk distance, slower gait speed, lower maximal grip and elbow flexion strength.

Stratified analysis based on sex (Suppl. Table 4) demonstrated that eGFR > 90 ml/min/1.73 m² was associated with significantly lower elbow flexion, knee flexion and extension strength only among females (P_interaction < 0.05). Additionally, among males, eGFR > 90 ml/min/1.73 m² was linked to significantly lower maximal grip strength (P_interaction: 0.004), and urinary albumin excretion > 30 mg/24 was associated with significantly lower knee flexion strength (P_interaction: 0.037).

Stratified analysis based on diabetes mellitus status (Suppl. Tables 5–6) indicated that among normoglycemic participants, eGFR > 90 ml/min/1.73 m² was associated with significantly shorter 6-min walk distance and slower gait speed. Interestingly, among individuals with type 2 diabetes mellitus, eGFR > 90 ml/min/1.73 m² was positively associated with elbow flexion and extension strength (P_interaction < 0.05). In diabetic participants, eGFR < 60 ml/min/1.73 m² was linked to shorter 6-min walk distance, slower gait speed and worse timed chair rise stand test time. Albuminuria > 30 mg/24 h was associated with shorter 6-min walk distance and slower gait speed in participants with both prediabetes and type 2 diabetes mellitus. Among those with type 2 diabetes mellitus, urinary albumin excretion > 30 mg/24 h was also linked to worse timed chair rise stand test time and lower maximal grip strength.