Chronic Kidney Disease Defined by Cystatin C Predicts Mobility Disability and Changes in Gait Speed: The Framingham Offspring Study

Abstract

Background.

As creatinine-based estimates of renal function are inaccurate in older adults, an alternative is an estimated glomerular filtration rate (eGFR_cys) based on cystatin C. We examined the prospective association between chronic kidney disease (CKD_cys) as determined by eGFR_cys with the primary outcome of incident mobility disability and the secondary outcome of change in gait speed.

Methods.

Framingham Offspring Study participants older than 60 years and free of mobility disability at baseline (1998–2001) were eligible. Baseline CKD_cys was defined as eGFR_cys less than 60 mL/min/1.73 m². At follow-up (2005–2008), the outcomes of mobility disability, defined as self-reported inability to walk 1/2 mile and/or climb a flight of stairs, and gait speed were measured. Logistic and linear regression models were adjusted for age, sex, body mass index, smoking, diabetes, C reactive protein, and physical activity.

Results.

Of 1,226 participants, 230 (19%) had CKD_cys at baseline. After a mean follow-up of 6.6 years, 185 (15%) developed mobility disability. Of those with CKD_cys, 60 (26%) developed mobility disability. Those with CKD_cys had greater odds of mobility disability in the age- and sex-adjusted (odds ratio [OR] 1.91, 95% CI 1.32, 2.75) and fully adjusted (OR 1.55, 95% CI 1.05, 2.31) models compared with those without CKD_cys. In fully adjusted models, participants with CKD_cys had greater gait speed declines than those without CKD_cys (β = 0.07 [SE 0.02], p = .0022).

Conclusion.

CKD_cys was associated with higher odds of incident mobility disability and greater decline in gait speed, highlighting the loss of physical independence in elders with CKD.

The prevalence in older adults of chronic kidney disease (CKD), or reduced renal function, increased from 27.8% in 1994 to 37.8% in 2004 (1). By 2030, projections predict that there will be more than 70 million older adult Americans (2), resulting in dramatic growth in the number of elders with CKD in the future.

Studies have documented an association between CKD and declining physical function in older adults. The Health, Aging and Body Composition (Health ABC) study found that elders with an estimated glomerular filtration rate (eGFR) less than 60 mL/min/1.73 m² based on serum creatinine were more likely to develop difficulty with walking or climbing stairs (3). The University of Alabama at Birmingham Study of Aging found that elders with an eGFR less than 60 mL/min/1.73 m² based on creatinine had greater odds of difficulties with activities of daily living (4). The Cardiovascular Health Study found that activities of daily living difficulty was associated with elevated levels of creatinine (men: ≥1.5 mg/dL; women: ≥1.3mg/dL) (5).

Creatinine, the current biomarker used to assess kidney function, is derived from muscle, but with age total muscle mass is known to decrease (6). An alternative biomarker is the protein cystatin C (7), produced by nucleated cells at a steady rate and not related to muscle mass. Inker et al (8) demonstrated that eGFR as predicted by cystatin C and creatinine accurately measured renal function at all levels of body mass index (BMI), including those less than twenty.

Mobility-limited elders are less likely to remain in the community (9). Studies have shown that mobility disability is associated with high levels of cystatin C (3). However, guidelines from the Kidney Disease Outcomes Quality Initiative Work Group of the National Kidney Foundation define CKD as eGFR less than 60 mL/min/1.73 m², as eGFR accounts for gender and race (10). Using this definition, studies have shown that CKD (CKD_cr) derived from creatinine-based eGFR (eGFR_cr) is associated with higher rates of kidney disease progression and mortality (11,12). But data using cystatin C-based eGFRs (eGFR_cys) to determine CKD (CKD_cys) are still emerging. CKD_cys is associated with elevated risks of infection-related hospitalizations (13) and cardiovascular events (14), but minimal information exists on functional outcomes. Our main goal was to determine if CKD_cys, as determined by an eGFR_cys less than 60 mL/min/1.73 m², was associated longitudinally with incident mobility disability in community-dwelling older adults. Our secondary goal was to assess if CKD_cys was associated longitudinally with a reduction in gait speed, as gait speed has been associated with mobility disability (15).

Methods

Study Population and Study Design

Established in 1971, the Framingham Offspring Study is an observational cohort study following the children and the children’s spouses of Framingham Heart Study participants (16). At each examination, participants underwent a medical history, physical examination, and laboratory tests. Our primary study was to determine if CKD_cys at examination 7 was associated prospectively with incident mobility disability at examination 8. The secondary study was to determine if CKD_cys at examination 7 was associated with a decrease in gait speed at examination 8. Participants who attended examination 7 (1998–2001) were eligible. Additional inclusion criteria at examination 7 were age more than 60 years, cystatin C data, and free of mobility disability. For the primary analysis, 1,570 participants were eligible. Due to lack of attendance at examination 8 (2005–2008) or missing mobility data at examination 8, 344 were excluded, resulting in a final sample of 1,226 participants for the primary study. For the secondary study, an additional 334 participants were excluded for lack of gait speed data at examinations 7 and 8, giving us a final sample of 892 participants. Mean follow-up was 6.6 years (Figure 1). All participants were white. The institutional review board of the Boston University School of Medicine approved all examinations and procedures, and all participants gave informed consent.

Figure 1.

Study eligibility.

Of the 302 participants excluded from the primary analysis for lack of examination 8 attendance, 94 (31%) had CKD (eGFR < 60 mL/min/1.73 m²) based on cystatin C and 51 (11%) had CKD based on creatinine. Mean age was 70.8 years, and compared with our sample, participants were more likely to be female or smokers, and have diabetes, hypertension, or cardiovascular disease (all p < .05). For the secondary study, comparison of the study sample with those who were ineligible found that the groups did not differ in mean age, proportion of women, mean BMI, or distribution of comorbidities (all p > .05, Supplementary Table 1).

Measurements

Chronic kidney disease.

Frozen serum from fasting blood samples from examination 7 was measured for cystatin C using nephelometry (interassay coefficient of variation [CV]: 3.3%, intraassay CV: 2.4%; Dade Behring Diagnostic, Marburg, Germany) (17). CKD_cys was defined as examination 7 eGFR_cys less than 60 mL/min/1.73 m² by the Chronic Kidney Disease Epidemiology Collaboration equation (eGFR_cys = 127.7 × plasma cystatin C^{− 1.17} × age in years^{− 0.13} × (1.06 if black) × (0.91 if female) (18).

The modified Jaffe method was used to assay creatinine (interassay CV: 2.8%, intraassay CV: 4.0%; Roche Hitachi 911, Roche Diagnostics, Indianapolis, IN) and indirectly calibrated to the Third National Health and Nutrition Examination Survey serum creatinine values (19). CKD_cr was defined as an examination 7 eGFR_cr less than 60 mL/min/1.73 m² by the Chronic Kidney Disease Epidemiology Collaboration equation (males: eGFR = 141 × min[serum creatinine/0.9, 1]^{− 0.411} × max[serum creatinine/0.9, 1]^{− 1.209} × 0.993^{age in years} × 1.159 [if black]; females: eGFR = 141 × min[serum creatinine/0.7, 1]^{− 0.329} × max[serum creatinine/0.7, 1]^{− 1.209} × 0.993^{age in years} × 1.018 × 1.159 [if black]) (20).

Mobility disability and gait speed.

At examinations 7 and 8, technicians using standardized protocols collected self-reported information on ability to walk up stairs to the second floor and walk half a mile (21). Participants had mobility disability if they answered “unable to do” or “does not do” for stairs and/or “need assistance using a device or assistance from another human,” “dependent,” or “does not do during normal day” for walking. For gait speed, the faster of two consecutive trials on a 4-m course walked at usual pace was recorded (22). At examination 7 ancillary visit, time required was recorded to the nearest second, whereas at examination 8, it was recorded to the nearest hundredth of a second.

Covariates

Data were from examination 7. BMI was calculated by dividing weight (in kg) by height (in m²). Cardiovascular disease was defined as Angina, coronary insufficiency (unstable angina), myocardial infarction, stroke/transient ischemic attack, congestive heart failure, and/or intermittent claudication (16). Diabetes was defined as a fasting glucose ≥126 mg/dL or use of insulin or oral hypoglycemic agents. Hypertension was determined by two measurements ≥140/90 mL of mercury or antihypertensive medication use. Smoking was defined as ≥ one cigarette per day in the preceding year. High-density lipoprotein cholesterol was determined from fasting samples. C reactive protein (CRP) was measured by enzyme immunoassay (Hemagen Diagnostics, Inc., Columbia, MD) (23). Proteinuria was from minute dipsticks (Ames Labstix, Elkhardt, IN) for albumin in spot urines, with proteinuria defined as≥ trace protein (24). Self-reports of daily physical activity were assessed using the Framingham Physical Activity Index as previously described (25). Participants reported activities during a routine day. Each activity was assigned a weight based on required oxygen consumption and multiplied by the time spent in 24 hours. The total sum of these values was the final score, with a range of 24–120. A person who slept continuously would have a score of 24, whereas a person with a manual labor occupation would have a score of 42.

Statistical Analysis

The association between CKD_cys at examination 7 and mobility disability at examination 8 was analyzed using logistic regression, adjusting for age and sex. Stepwise logistic regression using the p value of 0.1 to both enter and remain in the model was performed using these covariates: BMI, systolic blood pressure, smoking, proteinuria, high-density lipoprotein cholesterol, diabetes, hypertension, hypertension treatment, and cardiovascular disease, with age and sex forced into the model. Covariates retained were BMI, smoking, and diabetes. As inflammation may have been a mediator, CRP was added to the adjusted models. As it was a potential modifier of the association, physical activity was subsequently input as a covariate. Analyses were repeated using CKD_cr and continuous values of eGFR_cys and eGFR_cr as the exposures.

In the second study, linear regression was used to examine the association of CKD_cys with the change in gait speed between examinations 7 and 8, initially adjusting for age and sex. The covariates selected in the primary analysis were added with examination 7 gait speed. As inflammation and active lifestyle may have mediated the association, CRP and physical activity were added as subsequent covariates. Analyses were repeated with CKD_cr and continuous values of eGFR_cys and eGFR_cr. Adjustments were not made for multiple comparisons as mobility disability was the primary focus. Statistical analysis was performed using SAS version 9.3 (SAS Institute, Inc., Cary NC).

Results

Overall, 1,226 participants with a mean age of 68.0 years and a mean follow-up of 6.6 years were included in the main analysis (Table 1). The majority were female. Mean BMI was 27.9 kg/m². Two hundred thirty (19%) had CKD_cys, 135 (11%) had CKD_cr, and 88 (7%) had both CKD_cys and CKD_cr. Incident mobility disability developed in 185 (15%) participants.

Table 1.

Characteristics of Study Sample (N =1,226)

Characteristic	Mean ± SD or N (%)
Age (y)	68.0±5.4
Females (%)	654 (53%)
Smokers (%)	86 (7%)
Body mass index (kg/m2)	27.9±4.8
Diabetes (%)	147 (12%)
Hypertension (%)	499 (41%)
Cardiovascular disease (%)*	191 (16%)
Proteinuria (%)†	189 (16%)
Estimated glomerular filtration ratecystatin C (eGFRcys), mL/min/1.73 m2‡	72.8±14.4
Estimated glomerular filtrationcreatinine (eGFRcr), mL/min/1.73 m2§	76.4±14.3
Chronic kidney disease by eGFRcys (%)	230 (19%)
Chronic kidney disease by eGFRcr (%)	135 (11%)
Mean follow-up time in years (limits)	6.6 (3.7, 8.7)
Gait speed at examination 8, m/s (4-Meter Course)	1.12±0.26

Notes: *Angina, coronary insufficiency (unstable angina), myocardial infarction, stroke/transient ischemic attack, congestive heart failure, and/or intermittent claudication.

^†Albumin ≥ trace on a dipstick urinalysis.

^‡Chronic Kidney Disease Epidemiology Collaboration equation for cystatin: eGFR_cys = 127.7 × plasma cystatin C^{− 1.17} × age in years^{− 0.13} × (1.06 if black) × (0.91 if female).

^§Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation for creatinine—males: eGFR = 141 × min(serum creatinine/0.9, 1)^{− 0.411} × max(serum creatinine/0.9, 1)^{− 1.209} × 0.993^{age in years} × 1.159 (if black); females: eGFR = 141 × min(serum creatinine/0.7, 1)^{− 0.329} × max(serum creatinine/0.7, 1) ^{− 1.209} × 0.993^{age in years} × 1.018 × 1.159 (if black) (18).

Association of CKD and Incident Mobility Disability

Sixty (26%) participants with CKD_cys developed incident mobility disability compared with 125 (13%) participants without CKD_cys. Twenty-five (18%) of those with CKD_cr developed mobility disability compared with 160 (15%) of those without CKD_cr. After age and sex adjustment, those with CKD_cys had almost twice the odds of incident mobility disability compared with those without CKD_cys (Table 2). The association remained significant with the addition of BMI, diabetes, smoking, CRP, and physical activity. No association between CKD_cr and mobility disability was observed (all odds ratios approximately 1). Analysis using continuous values of eGFR_cys found an association between eGFR_cys and incident mobility disability, with a 1 SD decrease in eGFR_cys resulting in a 21% increase in the odds of the outcome (Supplementary Table 1). Analysis with continuous values of eGFR_cr did not demonstrate an association (p > .05).

Table 2.

Association of Chronic Kidney Disease (CKD) with Incident Mobility Disability* After 6.6 Years of Follow-up (N = 1,226)

CKD Classification	Model	Odds Ratio (95% CI)	p
CKDcystatin C	Age and sex	1.91 (1.32, 2.75)	<.01
	Multivariable†	1.56 (1.06, 2.28)	.02
	Multivariable + CRP + PA	1.55 (1.05, 2.31)	.03
CKDcreatinine	Age and sex	1.11 (0.71, 1.73)	.65
	Multivariable	1.02 (0.60, 1.62)	.92
	Multivariable + CRP + PA	1.03 (0.64, 1.62)	.97

Notes: CRP = C reactive protein; PA = physical activity.

*Self-reported inability to climb one flight of stairs or walk half a mile.

^†Adjusted for age, sex, body mass index, diabetes, and smoking.

Association of CKD With Change in Gait Speed

Among the 892 participants, 152 (17%) had CKD_cys and 107 (12%) had CKD_cr. In age- and sex-adjusted models, participants with CKD_cys had a significantly larger decline in gait speed between examinations 7 and 8 than those without CKD_cys (β = 0.08 [SE 0.02], p = .0004, Table 3). There was minimal attenuation of the association in the fully adjusted model (β = 0.07 [SE 0.02], p = .0022). Similar results were observed for CKD_cr. Gait speed declined in all groups during the 6.6 years of follow-up. After adjustment for age and sex, those with CKD_cys experienced a decline of 0.13 ± 0.01 m/s, and those without CKD_cys had a loss of 0.09±0.02 meters/second (Figure 2). Similarly, CKD_cr participants decreased 0.16±0.01 meters/second, and those without CKD_cr declined 0.08±0.03 meters/second. Gait speed decline was significantly associated with eGFR_cys after adjustment for all covariates (p = .02, Supplementary Table 2), but not with continuous values of eGFR_cr (p = .27).

Table 3.

Association of Chronic Kidney Disease (CKD) with Change in Gait Speed (4-Meter Course) After 6.6 Years of Follow-up (N = 892)

CKD Classification	Model	β (SE)	p
CKDcys	Age and sex	0.08 (0.02)	.0004
	Multivariable†	0.06 (0.02)	.008
	Multivariable + CRP + PA	0.07 (0.02)	.0022
CKDcr	Age and sex	0.07 (0.02)	.0045
	Multivariable	0.07 (0.03)	.008
	Multivariable + CRP + PA	0.07 (0.03)	.0004

Notes: CRP = C reactive protein; PA = physical activity.

^†Adjusted for age, sex, body mass index, diabetes, smoking, and examination 7 gait speed.

Figure 2.

Changes in age- and sex-adjusted mean gait speeds (4-meter course) in those with and without CKD as defined by cystatin C (CKD_cys) or creatinine (CKD_cr). Bars represent standard deviation. CKD_cys was defined as cystatin C-based eGFR less than 60mL/min/1.73 m² by the Chronic Kidney Disease Epidemiology Collaboration equation. CKD_cr was defined as creatinine-based eGFR less than 60mL/min/1.73 m² by the Chronic Kidney Disease Epidemiology Collaboration equation.

Discussion

In community-dwelling older adults, CKD_cys was associated with increased odds of incident mobility disability after adjustment for covariates. We found participants with CKD_cys had slower gait speeds compared with those without CKD_cys, expanding the understanding of the association between cystatin C and mobility disability. The Health ABC study showed that adults older than 70 years in the highest quartile of cystatin C were more likely to have mobility impairment after 4.5 years of follow-up (3). Our study demonstrates that this association exists in a younger population and over a longer period of time.

Evidence has shown that CKD_cys is associated with poor outcomes. In octogenarians, CKD_cys was associated with increased odds of cardiovascular disease (26), and the Cardiovascular Health Study showed that elders with CKD_cys had a higher risk of heart failure and mortality (27). Additionally, they are more likely to be hospitalized for an infection (13). However, such outcomes center on disease and survival, which may not be appropriate for elders focusing on quality of life. We examined the functional outcome of mobility, which is crucial to remaining in the community (9).

Parallel analyses were conducted with CKD_cr, but we did not find an association with mobility disability. The Cardiovascular Health Study demonstrated that risk of activities of daily living disability was not increased with an eGFR_cr less than 66 mL/min/1.73 m²(28). In contrast, Health ABC demonstrated that mobility disability was more likely with an eGFR_cr less than 60 mL/min/1.73 m² (3). However, differences in our results may in part be due to differences in study samples and design. Health ABC was a more ethnically diverse sample and was able to adjust for total body muscle mass, data our study lacked.

Those with reduced muscle mass and CKD can be misclassified with normal renal function when creatinine-based measures of renal function are utilized. Others have shown that an eGFR_cr ≥ 120 mL/min/1.73 m² is associated with increased mortality compared with measurements of 90–119 mL/min/1.73 m² (29). Moreover, elders with low muscle mass are more likely to develop physical function impairments (30). These factors may explain why we did not find an association between CKD_cr and mobility disability.

Our longitudinal analysis of gait speed demonstrated that those with CKD, either based on cystatin or creatinine, had greater losses in gait speed over the follow-up period compared with their non-CKD counterparts. Previously, only cross-sectional data on gait speed in elders were available (31), so our results fulfill a gap in the current understanding of physical performance in this population. Prior work has demonstrated that a gait speed less than 1.0 m/s predicts mobility disability and hospitalizations (32), and speeds less than 0.8 m/s predict greater mortality (33). Together with our evidence, this implies that older adults with CKD are more likely to experience poor outcomes earlier in the life course.

The possible mechanisms underlying the association between CKD and the outcomes of mobility disability and gait speed may be the effects of CKD on musculoskeletal function. Metabolic acidosis is associated with increased muscle catabolism (34), whereas vitamin D deficiency is associated with reduced strength (35). The secondary hyperparathyroidism observed in impaired renal function is associated with decreased bone density (36). Anemia has been linked with exercise intolerance (37).

Recent work has focused on the utilization of cystatin C in combination with other markers, in predicting renal function (8), end-stage renal disease, and mortality (38). Whether this approach would improve the prediction of mobility disability and gait speed in elders with CKD will be an important question to answer. Second, the trajectory of gait speed decline in CKD needs to be further explored. As history of CKD diagnosis was unknown in our study, whether these losses in gait speed occurred simultaneously with CKD onset or some point after cannot be determined. Information regarding continued gait speed decline will be important for design of future interventions.

Strengths of our study include the longitudinal design and the sample size. We suspect that the association between CKD_cys and incident mobility disability is stronger than demonstrated. Of those excluded for missing data, approximately one third had CKD_cys, likely biasing our results toward the null.

Limitations include the lack of ethnic diversity, limiting the applicability of the findings. The National Kidney Foundation defines CKD as decreased kidney function for at least three months, but we based CKD on a single measurement (39). Also, proteinuria was not used to define CKD as we lacked quantitative measurement of urinary protein. Mobility disability was self-reported versus directly assessed, although others have shown self-report of function is comparable (40). Finally, there were slight methodological differences in the assessment of gait speed between examinations 7 and 8.

Conclusion

CKD as defined by cystatin C predicts increased odds of incident mobility disability and greater gait speed decline in community-dwelling older adults. This highlights the vulnerability of elders with CKD to the loss of physical independence.

Supplementary Material

Supplementary material can be found at: http://biomedgerontology.oxfordjournals.org/

Funding

The Framingham Heart Study is supported by the National Heart, Lung, and Blood Institute in collaboration with Boston University (contract number N01-HC-25195). This work was supported by the National Institute on Aging (R01-AG029451) to J.M.Murabito. C.K.L. was supported by the American Federation for Aging Research, the John A. Hartford Foundation (Center of Excellence in Geriatric Medicine), and the U.S. Department of Agriculture, under Agreement no. 1950-51000-068-01S.