ABSTRACT
Background
Previous studies in HIV-infected individuals have demonstrated serum albumin to be strongly associated with kidney function decline, independent of urine albumin and inflammatory markers. Lower serum albumin concentrations may be an under-appreciated risk factor for kidney function decline in elders.
Methods
We performed a cohort analysis in the Health Aging and Body Composition Study, a cohort of well-functioning, bi-racial, community-dwelling elders between the age of 70 and 79 years. We examined the associations of serum albumin concentration with longitudinal kidney function decline by estimated glomerular filtration rate (eGFR). Outcomes included linear eGFR decline, rapid kidney function decline defined as >30% decrease in eGFR, defined as a final eGFR <60 mL/min/1.73 m2 in those with an eGFR >60 mL/min/1.73 m2 at baseline. Cystatin C-based eGFR was calculated at baseline, Year 3 and Year 10.
Results
Mean age was 74 years, and mean eGFR was 73 mL/min/1.73 m2 at baseline. The mean rate of eGFR change was 1.81 mL/min/1.73 m2 per year. After multivariate adjustment, lower serum albumin concentrations were strongly and independently associated with kidney function decline (−0.11 mL/min/1.73 m2 per year for each standard deviation decrease serum albumin; −0.01 to − 0.20) with no attenuation after adjustment for urine albumin and inflammatory markers (−0.12, −0.03 to − 0.22). When divided into quartiles, serum albumin levels ≤3.80 g/dL were associated with increased odds of rapid kidney function decline (odds ratio 1.59; 1.12–2.26) and increased risk of incident chronic kidney disease (incident rate ratio 1.29; 1.03–1.62) relative to levels >4.21g/dL. Urine albumin to creatinine ratio (ACR) was also significantly and independently associated with kidney function decline (−0.08 mL/min/1.73 m2 per year for urine ACR >30 mg/g; −0.82 to − 0.13).
Conclusions
Lower serum albumin levels are strongly and independently associated with kidney function decline in elders, independent of clinical risk factors, urine albumin and measured inflammatory markers.
Keywords: albumin, CKD, ESRD, age, inflammation
INTRODUCTION
Reduced kidney function in the elderly population is strongly and independently associated with cardiovascular disease, heart failure and mortality [1, 2]. The underlying risk factors for the onset and progression of kidney disease in elders are incompletely characterized, but include older age, black race, diabetes, hypertension and cardiovascular disease [3].
One potentially under-appreciated risk factor for kidney function decline is low serum albumin concentration. Previous studies have shown even small decrements in levels of serum albumin concentration to be strongly associated with cardiovascular disease, heart failure and mortality in vulnerable populations, such as elderly and HIV-infected individuals [4–6]. Limited research has linked low albumin levels to kidney function decline. Investigations in the Cardiovascular Health Study found that serum albumin was independently associated with higher risk of kidney function decline, whereas several inflammatory markers had no significant associations [7]. However, the study did not have concurrent measures of urine albumin concentrations, so it could not discern the contribution of urine losses of albumin—an established prognostic marker of kidney disease and potential confounder. Tangri demonstrated that serum albumin concentrations are an important component of a multi-marker predictive model for progression to end-stage renal disease (ESRD) in a cohort of Canadian adults with known chronic kidney disease (CKD) [8]. Recently, we demonstrated that serum albumin levels are an important predictor of kidney function decline in HIV-infected individuals, independent of albuminuria and known liver disease [9].
It remains unclear whether the associations of low serum albumin levels with kidney function decline reflect underlying inflammation, urine albumin, chronic illness or some combination of factors. To our knowledge, no study in a racially diverse cohort of elders has jointly examined the associations of serum and urine albumin concentration with kidney function decline independently of inflammatory markers. In this study, we hypothesized that serum albumin would be an important prognostic biomarker for kidney function decline and CKD risk in an elderly, bi-racial cohort, the Health, and Body Composition Study (Health ABC), and that the associations would be independent and complementary to urine albumin levels.
MATERIALS AND METHODS
Study population
Health ABC enrolled 3075 high-functioning men and women in generally good health between the ages of 70 and 79 years. Recruitment took place in Memphis, TN and Pittsburg, PA between April 1997 and June 1998; inclusion required that the participant planned to remain in the area for at least 3 years. Other inclusion criteria were the ability to perform activities of daily living, walk a quarter of a mile and climb 10 steps, and the absence of life-threatening illness. For this analysis, we included 2598 of the original participants who had serum albumin, urine albumin, baseline cystatin C and at least one follow-up measure of cystatin C. Further details of the Health ABC design and methods have been described previously [10].
Serum albumin
Serum albumin levels were measured at the Health ABC Core Laboratory at the University of Vermont, where the concentration of serum albumin in blood samples was determined using the bromocresol green method (Vitros; Ortho-clinical Diagnostics Inc., Rochester, NY, USA). We analyzed serum albumin concentrations as a continuous variable [per standard deviation (SD)] and a categorical variable divided into quartiles (>4.21, 4.01–4.21, 3.81–4.00, ≤3.80). Since urine albumin values are heavily right skewed, urine albumin was log transformed to the Base 2 and treated as a continuous variable (per doubling); we also dichotomized urine albumin to creatinine ratio (ACR) at 30 mg/g versus lesser values, and categorized urine ACR into quartiles. Secondary predictors included C-reactive protein (CRP), interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) modeled as log transformed to the Base 2 continuous variables.
Outcomes
We determined estimated glomerular filtration (eGFR) rate using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation for cystatin C (eGFRcys). Cystatin C was measured at the baseline visit, Year 3 and Year 10. We chose cystatin C because it is less biased by muscle mass than creatinine, and also because creatinine measurement methods changed during follow-up of Health ABC. We evaluated three potential outcomes: first, we examined eGFRcys as the continuous outcome of annual change expressed as mL/min/1.73 m2 per year; second, we examined rapid kidney function decline defined as a 30% or more decline in eGFR from baseline, as has been recommended [11]; and third, we examined incident reduced eGFR, defined as a final eGFR <60 mL/min/1.73 m2 and a rate of decline of at least 1 mL/min/1.73 m2 per year.
Covariates
For this study, candidate covariates included socio-demographic factors (age, sex, race, clinical site and education level); lifestyle factors (current smoking and alcohol use); and comorbid conditions [body mass index (BMI), diabetes (defined by use of hypoglycemic agents, self-report, fasting plasma glucose >126 mg/dL or an oral glucose tolerance test >200 mg/dL), use of antihypertensive medications, systolic blood pressure, diastolic blood pressure, prevalent heart failure, coronary artery disease, peripheral arterial disease and chronic obstructive pulmonary disease].
Analysis
For continuous outcomes of kidney function decline, we used linear-mixed models with random intercepts and slopes to estimate and compare linear trends in eGFR over time. Rapid kidney function decline was assessed as a dichotomized outcome at >30% decline. For this outcome, we estimated odds ratios using multivariate logistic regression. The change in eGFR was calculated from an ordinary least squares model that takes the observed slope of eGFR over all time points. Incident CKD was also evaluated as a dichotomized outcome for patients reaching the threshold of <60 mL/min/1.73 m2. To prevent this outcome from being reached by small changes in eGFR, we required that participants have at least a 1 mL/min/1.73 m2 annual decline in eGFR, improving the specificity of the outcome. We also excluded individuals with eGFR <60 mL/min/1.73 m2 at baseline. The statistical analysis of this outcome was performed by Poisson regression offset for time, thus allowing us to model the incidence rate ratio. We performed similar analyses for the inflammatory markers CRP, IL-6 and TNF-α, using the same outcomes and statistical methods.
Each outcome was evaluated as a series of models. The first model in each analysis was adjusted for demographic characteristics and clinical risk factors; the second model followed with adjustment for inflammatory markers and self-reported health; and, urine albumin was added to the serum albumin analyses in a third model.
All analyses were performed using IBM SPSS Statistics (Version 23.0, IBM Corp., Armonk, NY, USA) and Stata (StatCorp 2013, Stata Statistical Software: Release 13, StataCorp LP, College Station, TX, USA). A P < 0.05 was considered statistically significant for all analyses including interaction terms.
RESULTS
Among the 2598 participants included in this analysis, the mean eGFR was 73 mL/min/1.73 m2 and the mean ACR was 8.1 mg/g (Table 1). The mean age was 74 years, 39% were African American and 48% were male. Across quartiles of baseline albumin, prevalence of diabetes, hypertension, smoking, alcohol use, coronary artery disease, peripheral arterial disease and CKD appeared similar, as was mean eGFR. Persons with lower serum albumin levels on average had lower urine albumin concentrations.
Table 1.
Characteristics of elders in the health aging and body composition study by serum albumin quartiles
| Parameter | Serum albumin (g/dL) |
||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | All | |
| Range | >4.21 | [AuthorQuery id="AQ7" rid="7"]?>4.01–4.21 | 3.81–4.00 | ≤3.80 | |
| n | 510 | 566 | 646 | 875 | 2598 |
| Age | 73 (3) | 74 (3) | 73 (3) | 74 (3) | 74 (3) |
| Male | 290 (57) | 284 (50) | 299 (46) | 380 (43) | 1253 (48) |
| African American | 185 (36) | 209 (37) | 254 (39) | 374 (43) | 1022 (39) |
| eGFR-cysC | 72 (18) | 73 (18) | 73 (18) | 73 (18) | 73 (18) |
| eGFR<60 | 131 (26) | 133 (24) | 137 (21) | 211 (24) | 612 (24) |
| Urine ACRa | 9.7 [5.2–25.7] | 8.6 [4.9–19.3] | 8.2 [4.6–19.6] | 6.9 [3.9–16.6] | 8.1 [4.4–19.8] |
| Urine albumina | 10.6 [4.9–27.6] | 8.9 [4.6–20.1] | 9.5 [3.8–24.5] | 8.7 [4.1–21.2] | 9.2 [4.2–22.5] |
| Urine ACR >30 | 116 (23) | 88 (16) | 104 (16) | 128 (15) | 436 (17) |
| Smoking | |||||
| Current | 236 (46) | 270 (48) | 299 (46) | 393 (45) | 1198 (46) |
| Alcohol | |||||
| Current | 102 (20) | 126 (22) | 155 (24) | 177 (20) | 560 (22) |
| General health | |||||
| Fair | 61 (12) | 61 (11) | 99 (15) | 136 (16) | 357 (14) |
| Poor | 4 (1) | 8 (2) | 2 (0.3) | 3 (0.3) | 17 (0.8) |
| Prevalent HF | 18 (4) | 10 (2) | 12 (2) | 23 (3) | 64 (3) |
| Prevalent CAD | 113 (23) | 120 (21) | 135 (21) | 167 (20) | 535 (21) |
| BMI | 26.8 (4.3) | 27.2 (4.2) | 27.4 (5.0) | 27.8 (5.1) | 27.4 (4.8) |
| CRPa | 1.48 [0.92–2.65] | 1.56 [0.96–2.91] | 1.74 [0.99–3.18] | 1.88 [1.05–3.47] | 1.66 [0.98–3.06] |
| IL-6a | 1.61 [1.13–2.47] | 1.70 [1.16–2.60] | 1.79 [1.17–2.76] | 1.94 [1.37–2.93] | 1.78 [1.21–2.72] |
| TNF-α | 3.40 (1.62) | 3.43 (1.89) | 3.39 (1.46) | 3.50 (1.71) | 3.44 (1.68) |
| Cholesterol | 212 (41) | 206 (39) | 200 (35) | 197 (37) | 203 (38) |
| LDL | 126 (36) | 124 (35) | 120 (31) | 118 (35) | 122 (35) |
Median [interquartile range].
HF, heart failure; CAD, coronary artery disease; LDL, low-density lipoprotein.
As a linear variable, lower levels of serum albumin concentration were associated with greater decline in eGFR (Table 2). This association remained statistically significant and minimally attenuated when adjusted for demographic risk factors, inflammatory markers, self-reported health status and albuminuria. When divided into quartiles, lower serum albumin levels were strongly associated with eGFR decline in the sequentially adjusted models. Urine albumin did not appear to attenuate the association of serum albumin concentration with eGFR decline. A higher urine ACR as a linear variable, dichotomized and divided into quartiles was also independently associated with kidney function decline after adjustment.
Table 2.
Longitudinal associations of serum and urine albumin with kidney function decline
| Outcome = eGFR decline (mL/min/1.73 m2/year) | ||||
|---|---|---|---|---|
| Unadjusted | Model 1a | Model 2b | Model 3c | |
| β (95% CI) | β (95% CI) | β (95% CI) | β (95% CI) | |
| Serum albumin | ||||
| per SD decrease = 0.31 | −0.07 (0.01, −0.16) | −0.13 (−0.04, −0.22)** | −0.11 (−0.01, −0.20)** | −0.12 (−0.03, −0.22)** |
| Quartiles | ||||
| >4.21 | 0 (ref) | 0 (ref) | 0 (ref) | 0 (ref) |
| 4.01–4.20 | −0.23 (−0.50, 0.04) | −0.31 (−0.58, −0.03)* | −0.26 (−0.55, 0.02) | −0.26 (−0.56, 0.03) |
| 3.81–4.00 | −0.06 (−0.32, 0.20) | −0.17 (−0.45, 0.10) | −0.11 (−0.40, 0.18) | −0.21 (−0.51, 0.08) |
| ≤3.80 | −0.24 (−0.49, 0.02) | −0.38 (−0.65, −0.11)* | −0.32 (−0.61, 0.04)* | −0.35 (−0.64, −0.05)* |
| Urine ACR | ||||
| Urine ACR (per doubling) | −0.09 (−0.13, −0.05)* | −0.07 (−0.12, −0.03)** | −0.09 (−0.14, −0.05)* | −0.08 (−0.17, 0.01) |
| Urine ACR >30 mg/g | −0.60 (−0.86, −0.33)* | −0.53 (−0.80, −0.26)* | −0.60 (−0.88, −0.32)* | −0.47 (−0.82, −0.13)** |
| Quartiles | ||||
| ≤4.00 | 0 (ref) | 0 (ref) | 0 (ref) | 0 (ref) |
| 4.01–7.89 | −0.23 (−0.46, 0.01) | −0.17 (−0.40, 0.06) | −0.16 (−0.39, 0.08) | −0.05 (−0.33, 0.24) |
| 7.90–20.20 | −0.33 (−0.56, −0.10)** | −0.24 (−0.47, −0.01)** | −0.24 (−0.48, −0.0002)** | −0.05 (−0.38, 0.27) |
| >20.20 | −0.56 (−0.83, −0.29)* | −0.40 (−0.67, −0.13)** | −0.47 (−0.76, −0.18)** | −0.16 (−0.62, 0.30) |
Model 1: demographics and risk factors; model adjusted for age, gender, race, site, DM, SBP, HTN meds, smoking, prevalent cardiovascular disease, high-density lipoprotein, low-density lipoprotein.
Model 2: inflammation and health; model further adjusted for CRP, IL-6, TNF-α and self-reported health status.
Model 3: further adjusted for urine albumin (or serum albumin).
P < 0.001.
P < 0.05.
CI, confidence interval.
We assessed the association between serum albumin concentration and rapid kidney function decline (n = 451/2598) (Table 3). As a linear variable, each SD decrease of the serum albumin concentration was associated with a 19% higher odds of rapid decline in an unadjusted model, and there was minimal attenuation in the fully adjusted model. When categorized into quartiles, serum albumin concentrations ≤3.80 g/dL were significantly associated with rapid kidney function decline relative to levels >4.20 g/dL with minimal attenuation after adjustment. The association of urine albumin levels with rapid kidney function decline did not appear to attenuate the association of serum albumin with kidney function decline. Urine ACR was significantly associated with rapid kidney function decline in the fully adjusted model. Levels >30 mg/g were strongly and independently associated with rapid kidney function decline, as well as levels >20.20 mg/g relative to levels ≤4.00 mg/g.
Table 3.
Longitudinal associations of serum and urine albumin concentrations with rapid kidney function decline
| Outcome = rapid eGFR decline >30% (451/2598 = 17%) | ||||||
|---|---|---|---|---|---|---|
| n | No. of patients rapid decline | Unadjusted | Model 1a | Model 2b | Model 3c | |
| OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |||
| Serum albumin | ||||||
| per SD decrease = 0.31 | 2598 | 451 | 1.19 (1.09, 1.27)* | 1.20 (1.09, 1.29)* | 1.18 (1.07, 1.27)** | 1.19 (1.09, 1.28)** |
| Quartiles | ||||||
| >4.21 | 510 | 77 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| 4.01–4.20 | 567 | 98 | 1.21 (0.83, 1.73) | 1.21 (0.84, 1.76) | 1.18 (0.81, 1.71) | 1.23 (0.84, 1.80) |
| 3.81–4.00 | 646 | 91 | 0.99 (0.69, 1.43) | 1.01 (0.69, 1.48) | 0.97 (0.66, 1.42) | 1.00 (0.68, 1.46) |
| ≤3.80 | 875 | 185 | 1.58 (1.14, 2.19)** | 1.62 (1.15, 2.28)** | 1.53 (1.08, 2.16)** | 1.59 (1.12, 2.26)** |
| Urine ACR | ||||||
| Urine ACR (per doubling) | 1.20 (1.14, 1.27)* | 1.17 (1.10, 1.24)* | 1.17 (1.10, 1.24)* | 1.18 (1.11, 1.25)* | ||
| Urine ACR (>30 mg/g) | 436 | 119 | 2.27 (1.75, 2.93)* | 2.00 (1.52, 2.65)* | 1.98 (1.50, 2.62)* | 2.06 (1.55, 2.74)* |
| Quartiles | ||||||
| ≤4.00 | 677 | 90 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| 4.01–7.89 | 662 | 112 | 1.27 (0.92, 1.74) | 1.26 (0.91, 1.74) | 1.25 (0.90, 1.72) | 1.31 (0.94, 1.81) |
| 7.90–20.20 | 631 | 100 | 1.12 (0.80, 1.55) | 1.05 (0.75, 1.46) | 1.03 (0.74, 1.45) | 1.09 (0.78, 1.54) |
| >20.20 | 598 | 146 | 2.14 (1.58, 2.90)* | 1.86 (1.34, 2.57)* | 1.81 (1.30, 2.51)* | 1.93 (1.39, 2.69)* |
Model 1: demographics and risk factors; model adjusted for age, gender, race, site, DM, SBP, HTN meds, smoking, prevalent cardiovascular disease, high-density lipoprotein, low-density lipoprotein and eGFR.
Model 2: inflammation and health; model further adjusted for CRP, IL-6, TNF-α and self-reported health status.
Model 3: further adjusted for urine albumin (or serum albumin).
P < 0.001.
P < 0.05.
CI, confidence interval; OR, odds ratio.
Similar results were found for associations of serum and urine albumin concentrations with incident reduced eGFR (n = 556/1986) (Table 4). The associations of both serum and urine ACR with incident CKD were minimally attenuated by measured risk factors, self-reported health and inflammatory markers. In separate analyses using a binary clinical threshold of serum albumin ≤3.5 g/dL (n = 211), we found an 83% (24–170%) higher risk of rapid kidney function decline and 48% (13–94%) higher risk of incident reduced eGFR relative to levels >4.2 g/dL in the fully adjusted model.
Table 4.
Longitudinal associations of serum and urine albumin concentrations with incident reduced eGFR
| Outcome = incident CKD (556/1986=28%) | ||||||
|---|---|---|---|---|---|---|
| n | No. of patients incident CKD | Unadjusted | Model 1a | Model 2b | Model 3c | |
| IRR (95% CI) | IRR (95% CI) | IRR (95% CI) | IRR (95% CI) | |||
| Serum albumin | IRR (95% CI) | IRR (95% CI) | IRR (95% CI) | IRR (95% CI) | ||
| per SD decrease = 0.31 | 1986 | 556 | 1.09 (0.84–0.98)** | 1.12 (1.06–1.19)* | 1.12 (1.05–1.18)** | 1.13 (1.06–1.19)** |
| Quartiles | ||||||
| >4.21 | 379 | 105 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| 4.0–4.20 | 434 | 123 | 1.04 (0.83–1.30) | 1.13 (0.91–1.40) | 1.04 (0.82–1.31) | 1.03 (0.81–1.31) |
| 3.81–4.00 | 509 | 127 | 0.94 (0.75–1.17) | 0.99 (0.79–1.24) | 0.94 (0.74–1.19) | 0.98 (0.77–1.26) |
| ≤3.80 | 664 | 201 | 1.18 (0.96–1.45) | 1.32 (1.07–1.62)** | 1.26 (1.01– 1.56)** | 1.29 (1.03–1.62)** |
| Urine ACR | ||||||
| Urine ACR (per doubling) | 1.09 (1.05–1.14)* | 1.08 (1.04–1.13)* | 1.07 (1.03–1.12)* | 1.08 (1.04–1.13)* | ||
| Urine ACR (>30 mg/g) | 271 | 108 | 1.79 (1.51– 2.13)* | 1.49 (1.25–1.77)* | 1.51 (1.25–1.81)* | 1.56 (1.30–1 .88)* |
| Quartiles | ||||||
| ≤4.00 | 554 | 135 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| 4.01–7.89 | 533 | 138 | 1.09 (0.89–1.35) | 1.04 (0.85–1.27) | 0.91 (0.73–1.13) | 0.94 (0.76–1.17) |
| 7.90–20.20 | 496 | 136 | 1.18 (0.96–1.46) | 1.04 (0.85–1.28) | 0.95 (0.77–1.17) | 0.99 (0.80–1.22) |
| >20.20 | 381 | 143 | 1.84 (1.50–2.25)* | 1.48 (1.21–1.82)* | 1.38 (1.11–1.71)* | 1.44 (1.16–1.79)* |
Model 1: demographics and risk factors; model adjusted for age, gender, race, site, DM, SBP, HTN meds, smoking, prevalent cardiovascular disease, high-density lipoprotein, low-density lipoprotein and eGFR.
Model 2: inflammation and health; model further adjusted for CRP, IL-6, TNF-α and self-reported health status.
Model 3: further adjusted for urine albumin (or serum albumin).
P < 0.001.
P < 0.05.
CI, confidence interval; IRR, incidence rate ratio.
The inflammatory markers assessed in this analysis did not have consistent associations with kidney function decline (Table 5). None of the inflammatory markers was significantly associated with kidney function decline in the fully adjusted models. Only high CRP levels were significantly associated with rapid decline. All three inflammatory markers—CRP, IL-6 and TNF-α—were initially associated with incident reduced eGFR; however, when adjusted for serum and urine albumin, only TNF-α retained a significant association with the outcome.
Table 5.
Longitudinal associations of inflammatory markers with kidney function decline, rapid kidney function decline and incident CKD
| Outcome = eGFR decline (mL/min/1.73 m2/year) | ||||||
|---|---|---|---|---|---|---|
| Unadjusted | Model 1a | Model 2b | Model 3c | |||
| CRP (per doubling) | — | — | −0.01 (−0.10, 0.07) | −0.02 (−0.09, 0.08) | −0.03 (−0.12, 0.07) | −0.04 (−0.13, 0.05) |
| IL-6 (per doubling) | — | — | −0.02 (−0.12, 0.07) | 0.01 (−0.09, 0.11) | −0.001 (−0.11, 0.11) | 0.03 (−0.08, 0.14) |
| TNF-α (per doubling) | — | — | 0.07 (−0.10, 0.23) | 0.17 (0.001, 0.35)** | 0.18 (0.0004, 0.36)** | 0.18 (−0.003, 0.37) |
| Outcome = rapid eGFR decline >30% | ||||||
| n | No. of patients with rapid decline | |||||
| CRP (per doubling) | 2598 | 451 | 1.19 (1.08, 1.30)* | 1.17 (1.06, 1.30)** | 1.14 (1.01, 1.27)** | 1.14 (1.02, 1.29)** |
| IL-6 (per doubling) | 2598 | 451 | 1.17 (1.03, 1.32)** | 1.15 (1.00, 1.31)** | 1.00 (0.86, 1.17) | 0.96 (0.82, 1.12) |
| TNF-α (per doubling) | 2598 | 451 | 1.18 (0.98, 1.44) | 1.34 (1.06, 1.68)** | 1.30 (1.03, 1.65)** | 1.25 (0.99, 1.59) |
| Outcome = incident CKD | ||||||
| n | No. of patients with incident CKD | |||||
| CRP (per doubling) | 1986 | 556 | 1.15 (1.08, 1.22)* | 1.11 (1.04, 1.18)** | 1.06 (0.98, 1.14) | 1.06 (0.98, 1.15) |
| IL-6 (per doubling) | 1986 | 556 | 1.25 (1.16, 1.35)* | 1.14 (1.05, 1.24)** | 1.05 (0.95, 1.15) | 1.04 (0.94, 1.15) |
| TNF-α (per doubling) | 1986 | 556 | 1.80 (1.54, 2.10)* | 1.40 (1.20, 1.63)* | 1.36 (1.17, 1.59)* | 1.30 (1.11, 1.53)** |
Model 1: demographics and risk factors; model adjusted for age, gender, race, site, DM, SBP, HTN meds, smoking, prevalent cardiovascular disease, high-density lipoprotein, low-density lipoprotein and eGFR.
Model 2: inflammation and health; model further adjusted for CRP, IL-6, TNF-α and self-reported health status.
Model 3: further adjusted for urine albumin.
P < 0.001.
P < 0.05.
DISCUSSION
We found serum albumin concentrations to be strongly associated with eGFR decline, rapid eGFR decline and incident CKD. These findings were independent of albuminuria, self-reported health, biomarkers of inflammation, as well as demographic and clinical risk factors for kidney disease progression. Urine ACR was also significantly associated with eGFR decline, rapid decline and incident CKD in each fully adjusted model. None of the inflammatory markers was consistently associated with kidney function outcomes after adjustment for urine albumin and serum albumin.
Serum albumin concentration is a well-known predictor of mortality in patients with CKD. Only a few studies have examined serum albumin as a risk factor for the development of ESRD [12–14]. In a 2011 retrospective cohort analysis of 1800 men and women from the UK, Wagner et al. found that serum albumin levels were an essential component of a multivariable risk model for mortality in patients with ESRD [12]. Specifically, the authors found that higher serum albumin concentrations, defined as a continuous variable, were associated lower risk of mortality within 3 years of starting dialysis. A second retrospective study of 1000 chronic dialysis patients in Ireland dichotomized serum albumin levels at 3.5 g/dL, and found the lower category to be associated with higher mortality risk in adjusted analyses [13]. One other study of 109 dialysis patients found similar associations of albumin and mortality, which appeared to be mediated by cardiovascular events [15].
Few studies have examined associations of serum albumin with kidney function decline. However, in one study of eight inflammatory markers, including serum albumin, serum albumin was the only laboratory measure associated with kidney function decline [7]. We previously found serum albumin to be a strong and independent risk factor for kidney function decline in HIV-infected women [9], and those associations were only minimally attenuated by albuminuria. No other studies, to our knowledge, have examined serum albumin as an independent risk factor for the development of kidney function decline, rapid decline and incident CKD in a diverse population of elders, accounting for both inflammatory factors and urine albumin. Because serum albumin is a widely available clinical laboratory measurement, it may serve as a useful marker to help identify individuals at higher risk of CKD development in clinical practice.
The underlying mechanisms for these associations are unclear. To our knowledge, there is no direct physiological basis for low serum albumin to cause the development of CKD. Serum albumin levels may be reduced for one of several reasons: liver damage and disease may decrease production; they may decrease as a negative acute phase reactant or in response to inflammation; and substantial and sustained albuminuria can lead to lower serum concentrations. Uncommonly low serum albumin can also reflect malnutrition, but the only malnourished state that consistently produces lower serum albumin is kwashiorkor, a rare disease in the developing world [16]. In this study, albuminuria is clearly not the explanation for these associations since baseline concentrations of urine albumin were low, albuminuria was similar across quartiles of serum albumin at baseline and the associations were not attenuated when adjusted for urine albumin. Inflammation would appear to be the most likely explanation for our findings; however, we evaluated the association of several inflammatory markers—CRP, TNF-α and IL-6—with kidney function decline. IL-6 was not associated with any of the outcomes, while CRP and TNF-α had inconsistent associations. A similar study in Health ABC investigating the associations of these three inflammatory markers with cardiovascular disease also found inconsistent results [17], suggesting that these inflammatory markers do not capture the physiological processes that underlie the associations of serum albumin with adverse outcomes in Health ABC.
We hypothesize that the strong relationship between lower serum albumin and kidney function decline in this study must be multi-factorial, representing general ill health. We could not adjust for prevalence and severity of liver disease in this study, as this may represent one reason why individuals have lower serum albumin and decreased kidney function. Lower albumin concentrations may also reflect much broader abnormalities in processes of inflammation, thrombosis and microvascular function than can be captured by the markers available in this cohort [18–20]. Lower serum albumin and kidney function decline may also be related by multiple factors that are currently poorly understood. These findings are consistent with the results from our previous study of HIV-infected women [9] as well as other studies of serum albumin in ESRD [12, 13].
This study had several strengths. It had a large sample size in a well-characterized cohort with rigorously attained clinical data. The data span 10 years with multiple measurements of cystatin C over that time period. Our study also had several limitations. The cohort was limited to elders and therefore our results may not be generalizable to the younger adult population. Most individuals had preserved kidney function at baseline, and we could not evaluate whether results generalize to populations with more advanced CKD, or with ESRD outcomes. We also cannot fully explain the physiological basis for these findings.
In summary, we found that lower levels of serum albumin are strongly and independently associated with eGFR decline, rapid eGFR decline and incident CKD in a population of well-functioning, community dwelling elders. These findings were independent of albuminuria, three markers of inflammation and other known risk factors for kidney disease progression. Several studies have evaluated utilizing serum albumin as a risk factor for ESRD in both the general and HIV-infected populations [12, 21, 22]. Our findings suggest that serum albumin may further aid clinicians in risk-stratifying patients for the development of kidney disease. Additional studies are required to understand the biological mechanisms linking low serum albumin with kidney function decline.
FUNDING
5R01AG027002- Shlipak (PI).
CONFLICT OF INTEREST STATEMENT
None declared.
REFERENCES
- 1. Fried LF. Kidney function as a predictor of noncardiovascular mortality. J Am Soc Nephrol 2005; 16: 3728–3735 [DOI] [PubMed] [Google Scholar]
- 2. Shlipak MG, Cystatin C.. Mortality risk in the elderly: the health, aging, and body composition study. J Am Soc Nephrol 2005; 17: 254–261 [DOI] [PubMed] [Google Scholar]
- 3. Fox CS, Larson MG, Leip EP. et al. Predictors of new-onset kidney disease in a community-based population. J Am Med Assoc 2004; 291: 844–850 [DOI] [PubMed] [Google Scholar]
- 4. Lang J, Scherzer R, Weekley CC. et al. Serum albumin and short-term risk for mortality and cardiovascular disease among HIV-infected veterans. Aids 2013; 27: 1339–1343 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Goldwasser P, Feldman J.. Association of serum albumin and mortality risk. J Clin Epidemiol 1997 [DOI] [PubMed] [Google Scholar]
- 6. Feldman JG, Gange SJ, Bacchetti P. et al. Serum albumin is a powerful predictor of survival among HIV-1-infected women. J Acquir Immune Defic Syndr 2003; 33: 66–73 [DOI] [PubMed] [Google Scholar]
- 7. Keller C, Katz R, Sarnak MJ. et al. Inflammatory biomarkers and decline in kidney function in the elderly: the Cardiovascular Health Study. Nephrol Dial Transplant 2009; 25: 119–124 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Tangri N. A predictive model for progression of chronic kidney disease to kidney failure. J Am Med Assoc 2011; 305: 1553. [DOI] [PubMed] [Google Scholar]
- 9. Lang J, Scherzer R, Tien PC. et al. Serum albumin and kidney function decline in HIV-infected women. Am J Kidney Dis 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Simonsick EM, Newman AB, Nevitt MC. et al. Measuring higher level physical function in well-functioning older adults: expanding familiar approaches in the Health ABC study. J Gerontol A Biol Sci Med Sci 2001; 56: M644–M649 [DOI] [PubMed] [Google Scholar]
- 11. Levey AS, Inker LA, Matsushita K. et al. GFR decline as an end point for clinical trials in CKD: a scientific workshop sponsored by the National Kidney Foundation and the US Food and Drug Administration. Am J Kidney Dis 2014; 64: 821–835 [DOI] [PubMed] [Google Scholar]
- 12. Wagner M, Ansell D, Kent DM. et al. Predicting mortality in incident dialysis patients: an analysis of the United Kingdom Renal Registry. Am J Kidney Dis 2011; 57: 894–902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Foley RN, Parfrey PS, Harnett JD. et al. Hypoalbuminemia, cardiac morbidity, and mortality in end-stage renal disease. J Am Soc Nephrol 1996; 7: 728–736 [DOI] [PubMed] [Google Scholar]
- 14. Phelan PJ, O'kelly P, Walshe JJ. et al. The importance of serum albumin and phosphorous as predictors of mortality in ESRD patients. Renal Fail 2008; 30: 423–429 [DOI] [PubMed] [Google Scholar]
- 15. Cooper BA, Penne EL, Bartlett LH. et al. Protein malnutrition and hypoalbuminemia as predictors of vascular events and mortality in ESRD. Am J Kidney Dis 2004; 43: 61–66 [DOI] [PubMed] [Google Scholar]
- 16. Friedman AN, Fadem SZ.. Reassessment of albumin as a nutritional marker in kidney disease. J Am Soc Nephrol 2010; 21: 223–230 [DOI] [PubMed] [Google Scholar]
- 17. Cesari M, Penninx BWJH, Newman AB. et al. Inflammatory markers and cardiovascular disease (The Health, Aging and Body Composition [Health ABC] Study). Am J Cardiol 2003; 92: 522–528 [DOI] [PubMed] [Google Scholar]
- 18. Shlipak MG, Katz R, Cushman M. et al. Cystatin-C and inflammatory markers in the ambulatory elderly. Am J Med 2005; 118: 1416.e25–1416.e31 [DOI] [PubMed] [Google Scholar]
- 19. Harris TB, Ferrucci L, Tracy RP. et al. Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med 1999; 106: 506–512 [DOI] [PubMed] [Google Scholar]
- 20. Shlipak MG. Elevations of inflammatory and procoagulant biomarkers in elderly persons with renal insufficiency. Circulation 2002; 107: 87–92 [DOI] [PubMed] [Google Scholar]
- 21. Jotwani V, Li Y, Grunfeld C. et al. Risk factors for ESRD in HIV-infected individuals: traditional and HIV-related factors. Am J Kidney Dis 2012; 59: 628–635 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Sud M, Tangri N, Levin A. et al. CKD stage at nephrology referral and factors influencing the risks of ESRD and death. Am J Kidney Dis 2014; 63: 928–936 [DOI] [PubMed] [Google Scholar]