Abstract
Background:
The estimated glomerular filtration rate (eGFR) from cystatin C (eGFRcys) is often considered a more accurate method to assess GFR compared with an eGFR from creatinine (eGFRcr) in the setting of heart failure (HF) and sarcopenia, because cystatin C is hypothesized to be less affected by muscle mass than creatinine. We evaluated (1) the association of muscle mass with cystatin C, (2) the accuracy of eGFRcys, and (3) the association of eGFRcys with mortality given muscle mass.
Methods and Results:
We included 293 patients admitted with HF. Muscle mass was estimated with a validated creatinine excretion-based equation. Accuracy of eGFRcys and eGFRcr was compared with measured creatinine clearance. Cystatin C and creatinine were 31.7% and 59.9% higher per 14 kg higher muscle mass at multivariable analysis (both P < .001). At lower muscle mass, eGFRcys and eGFRcr overestimated the measured creatinine clearance. At higher muscle mass, eGFRcys underestimated the measured creatinine clearance, but eGFRcr did not. After adjusting for muscle mass, neither eGFRcys nor eGFRcr were associated with mortality (both P > .19).
Conclusions:
Cystatin C levels were associated with muscle mass in patients with HF, which could potentially decrease the accuracy of eGFRcys. In HF where aberrations in body composition are common, eGFRcys, like eGFRcr, may not provide accurate GFR estimations and results should be interpreted cautiously.
Keywords: Heart failure, cystatin C, muscle mass, glomerular filtration rate
Creatinine and cystatin C are the most common endogenous filtration markers used in clinical practice to estimate the glomerular filtration rate (GFR).1,2 The most widely used equations to estimate GFR also include demographic parameters (age, sex, and race) to account for variations in levels of serum creatinine and serum cystatin C that are independent of GFR, such as the level of muscle mass.3
Sarcopenia (muscle loss) and cardiac cachexia (muscle and fat loss) are highly prevalent in patients with severe heart failure (HF),4,5 and may lead to inaccuracies in the estimated GFR (eGFR) when using creatinine (eGFRcr).6 Muscle mass is the most important determinant of creatinine generation7; therefore, patients with severe HF may have low creatinine levels despite low GFR and eGFRcr would overestimate GFR in patients with reduced muscle mass.6 Cystatin C is thought to be affected less by muscle mass than creatinine and is therefore often considered to be a more accurate filtration marker than creatinine in the setting of sarcopenia.8
In severe HF, the eGFR from cystatin C (eGFRcys) has been hypothesized to provide a more accurate estimation of GFR compared with eGFRcr. Prior studies have demonstrated that eGFRcys is more strongly associated with mortality than eGFRcr and hypothesized that the greater accuracy of eGFRcys in the context of sarcopenia observed in HF may account for this observation.9,10 However, whether cystatin C is associated with muscle mass has never been evaluated in HF, and prior reports in chronic kidney disease have challenged the independence of cystatin C from muscle mass.11,12 As such, the goal of the current study was to evaluate (1) the relationship between cystatin C and creatinine with muscle mass estimated from creatinine excretion, (2) the accuracy of eGFRcys and eGFRcr by comparing them with the measured creatinine clearance (mClcr) according to estimated muscle mass, and (3) associations of eGFRcys and eGFRcr with risk of mortality given muscle mass.
Methods
Population
We identified patients admitted for decompensated HF at Yale New Haven Hospital who received intravenous loop diuretics while hospitalized. Inclusion required at least 1 objective sign of volume overload. Patients with significant bladder dysfunction, urinary incontinence, or an inability to comply with urine collection procedures were excluded. This study complied with the Declaration of Helsinki, and all patients provided written informed consent. The study was approved by the Yale Institutional Review Board.
Assessments of GFR
GFR was assessed as mClcr. Before the morning diuretic administration, a blood sample was obtained and patients were asked to empty their bladder. Then, followed by loop diuretic administration, a timed 6-hour urine collection with supervision by study staff was carried out. After 6 hours, patients were asked to empty their bladder to complete the collection and time was recorded. Patients additionally underwent a supervised 18-hour urine collection (to complete 24 hours). The mClcr from the 6-hour timed urine collection was the reference method to determine GFR. The mClcr was calculated as (Urine creatinine× Volume [mL]/Serum creatinine). The result was divided by the recorded time and indexed to 1.73 m2. Patient weight was obtained just before the beginning of the urine collection and was used to calculate body surface area with the Du Bois method. Ideal body weight was used to estimate body surface area if the body mass index (BMI) was greater than 40 kg/m2. The same method was used to measure the clearance of urea nitrogen.
The eGFR was assessed using the Chronic Kidney Disease Epidemiology Collaboration equations for eGFRcr, eGFRcys, and eGFRcr-cys.3
Assessment of Muscle Mass
Muscle mass was estimated from creatinine excretion with a validated equation: Muscle mass (kg) = (18.9× Creatinine excretion [g/d]) +4.1.13 This calculation was done with the 6-hour measured creatinine excretion assuming constant creatinine excretion during 24 hours. Estimated creatinine excretion was ascertained using the Chronic Kidney Disease Epidemiology Collaboration equation. We categorized muscle mass based on the median value of muscle mass (24.3 kg) estimated from Wang et al13 and below the established cutoff point for sarcopenia (<15 kg for female and <20 kg for male).14
Assays
A Randox Imola automated clinical chemistry analyzer was used to measure concentration of urine or serum chemistry parameters. Randox reagents were used in accordance with the manufacturer’s instructions to measure albumin, cystatin C and creatinine (Randox Laboratories, Crumlin, UK). Creatinine measurements are standardized to National Institute of Standards and Technology reference material (SRM 967). Assignment of cystatin C calibrators has been performed at Randox Laboratories by latex enhanced immunoturbidimetry, with reference to material standardized against the International Federation of Clinical Chemistry Reference Standard.
Statistical Analysis
Associations With Muscle mass.
We first examined the unadjusted association of each predictor variable with each log transformed filtration marker using linear regression. We next repeated these analyses adjusting for log transformed mClcr, and finally, adjusting for log transformed mClcr, and for age, sex, race, BMI, and serum albumin. In these models, parameter estimates were transformed as 100 × (ecoeff − 1) such that values can be interpreted as the geometric mean percent difference in the filtration marker for an interquartile range (IQR) difference in continuous predictors (which is the difference between the 25th and 75th percentiles) or for the presence of categorical predictors. For example, the percent difference in the filtration marker per every 14 kg of muscle mass (14 kg was the IQR of muscle mass). Calculated muscle mass was modeled with restricted cubic splines using three knots at recommended percentiles and plotted to show its relationship with cystatin C and creatinine.
Performance of eGFRcys and eGFRcr Compared With mClcr.
Bias was evaluated as the median difference between eGFR and mClcr (eGFR – mClcr; positive values indicate overestimation and negative values indicate underestimation compared with mClcr). Precision was computed as the IQR for the difference. Accuracy was computed as the percentage of patients whose eGFR is outside the 30% of the mClcr (1 – P30); the 30% cutoff value was selected to be consistent with previous publications that have used this value to define accuracy when estimating GFR.3,15 The 95% confidence interval (CI) of these metrics was estimated with 2000 bootstrap replications. Bias in eGFRcr vs eGFRcys was compared with the Wilcoxon matched-pairs signed-rank test. Precision was compared with the bootstrap method and accuracy with the McNemar’s test. In addition, performance was compared in groups of lower and higher muscle mass.
Survival.
We used Cox regression to compare the risk associations between eGFRcr and eGFRcys with death. First, univariate analysis was done with the main predictors: eGFRcr, eGFRcys, mClcr, and muscle mass. Then, each eGFR and mClcr were included in a different model adjusting for muscle mass. Finally, Kaplan-Meier survival curves were plotted to compare groups of high vs low eGFRcr or eGFRcys (based on its median value) stratified by high vs low muscle mass (based on the cutoff point for sarcopenia). P values of less than .05 were considered significant. Stata SE version 14.0 (StataCorp, College Station, TX) was used for statistical analysis.
Sensitivity Analysis.
All of the analyses were repeated adjusting for different estimators of GFR to exclude the possibility that the associations were driven by adjustment for mClcr. Estimators of GFR were the average of mClcr and clearance of urea nitrogen, and the 24-hour mClcr; in 110 patients whose measurements were repeated 1 week later as outpatients, analyses were done adjusting for mClcr derived from a supervised 8-hour urine collection in our clinical research center. In addition, analysis with 6-hour mClcr was repeated including patients with stable values of Cr (change of <0.3 mg/dL).
Results
Population
We included 293 patients who completed the timed-urine collection. Table 1 shows baseline characteristics at admission stratified by the median value for muscle mass (24.3 kg). Patients with lower muscle mass were older, female, non-black, and had a lower BMI (P < .001 for all). Hemoglobin, serum albumin, N-terminal pro b-type natriuretic peptide, and mClcr showed worse values in the group of lower muscle mass (P < .01 for all). At physical examination, the proportion of congestion signs were similar between groups, as were diuretic medications used on the day of the study. Angiotensin-converting enzyme inhibitors or angiotensin receptor blockers were more frequently used in patients with higher muscle mass. Importantly, in patients with a lower muscle mass, the median measured creatinine excretion was only 54% of the estimated creatinine excretion, whereas in patients with a higher muscle mass, the median measured creatinine excretion was 81% of the estimated creatinine excretion.
Table 1.
Baseline Characteristics of Patients
| Variable Demographics |
Overall cohort (N = 293) |
Lower Muscle Mass (Below the Median) (n = 147) |
Higher Muscle Mass (Above the Median) (n = 146) |
P Value |
|---|---|---|---|---|
| Age (years) | 65 ± 13 | 69 ± 12 | 60 ± 13 | <.001 |
| Male (%) | 189 (65) | 74 (50) | 115 (79) | <.001 |
| Black (%) | 86 (29) | 29 (20) | 57 (39) | <.001 |
| Past medical history, ejection fraction and BMI | ||||
| Diabetes (%) | 163 (56) | 78 (53) | 85 (58) | .37 |
| Hypertension (%) | 261 (89) | 132 (90) | 129 (88) | .69 |
| NYHA functional class III–IV (%) | 156 (53) | 87 (59) | 69 (47) | .04 |
| Left ventricular ejection fraction (%) | 34 (23–54) | 34 (23 –54) | 34 (23–55) | .91 |
| Left ventricular ejection fraction <35% (%) | 149 (51) | 74 (50) | 75 (51) | .86 |
| Left ventricular ejection fraction ≥50% (%) | 90 (31) | 46 (31) | 44 (30) | .83 |
| BMI (kg/m2) | 32.5 (27.6–38.8) | 30.0 (26.6–34.8) | 35.9 (29.5–43.0) | <.001 |
| Physical examination | ||||
| Systolic blood pressure (mm Hg) | 114 (103–129) | 113 (102–127) | 118 (106–130) | .10 |
| Jugular venous distention* (%) | 166 (57) | 88 (60) | 78 (53) | .47 |
| Ascites (%) | 17 (6) | 8 (5) | 9 (6) | .68 |
| Peripheral edema (%) | 224 (76) | 114 (78) | 110 (75) | .52 |
| Baseline serum levels | ||||
| Creatinine (mg/dL) | 1.37 (1.05–1.70) | 1.46 (1.11–1.85) | 1.27 (1.03–1.59) | .02 |
| Cystatin C (mg/L) | 1.61 (1.20–2.14) | 1.81 (1.42–2.42) | 1.36 (1.05–1.74) | <.001 |
| Sodium (mmol/L) | 136 (133–139) | 136 (132–139) | 135 (133–138) | .11 |
| Chloride (mmol/L) | 95.7 (93–98) | 96 (93–98) | 96 (94–98) | .9 |
| Hemoglobin (g/dL) | 11.4 ± 2.1 | 10.9 ± 2.1 | 11.9 ± 2.1 | .001 |
| Albumin (g/dL) | 3.60 ± 0.46 | 3.52 ± 0.49 | 3.68 ± 0.40 | .004 |
| NT-proBNP (pg/mL) | 3766 (1888–7966) | 5440 (2911–13226) | 3348 (1220–9175) | <.001 |
| GFR (mL/min/1.73 m2) | ||||
| eGFRcr | 52 (38–69) | 45 (33–59) | 63 (47–78) | <.001 |
| eGFRcys | 40 (27–60) | 33 (22–46) | 51 (36–71) | <.001 |
| mClcr | 40 (26–64) | 27 (19–36) | 62 (47–84) | <.001 |
| Muscle mass and creatinine excretion | ||||
| Estimated muscle mass† (kg) | 24.4 (18.2–32.2) | 18.1 (15.1–21.3) | 32.2 (28.6–38.0) | <.001 |
| Measured creatinine excretion‡ (mg/kg/24 h) | 10.9 (8.0–14.4) | 8.1 (6.5–10.6) | 14.2 (11.6–17.8) | <.001 |
| Measured creatinine excretion‡ (mg/24 h) | 1076 (743–1486) | 743 (584–912) | 1485 (1295–1792) | <.001 |
| Estimated creatinine excretion§ (mg/24 h) | 1594 (1305–1926) | 1382 (1074–1623) | 1831 (1567–2064) | <.001 |
| In-hospital treatment (day of the study) | ||||
| IV Loop diuretic dose (mg/24 hours of IV furosemide equivalents) | 160 (80–320) | 200 (80–360) | 160 (80–320) | .42 |
| Thiazide like diuretic (%) | 14 (5) | 10 (7) | 4 (3) | .17 |
| Angiotensin-converting enzyme inhibitor or angiotensin receptor blocker (%) | 94 (32) | 35 (24) | 59 (40) | .002 |
| Beta-blocker (%) | 181 (62) | 86 (59) | 95 (65) | .25 |
| Aldosterone antagonist (%) | 48 (16) | 22 (15) | 26 (18) | .51 |
Data are mean ± standard deviation, median (quartile 1–quartile 3), or number (%).
Conversion factors for units: creatinine in mg/dL to μmol/L, × 88.4; cystatin C in mg/L to nmol/L, × 74.9; sodium in mmol/L to mEq/L, × 1; chloride in mmol/L to mEq/L, × 1; hemoglobin in g/dL to g/L, × 10; albumin in g/dL to g/L, × 10; NT-proBNP in pg/mL to pmol/L × 0.1182; urine creatinine in mg/24 h to g/24 h, × 0.001.
ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; eUcrV, estimated creatinine excretion; eGFR, estimated glomerular filtration rate; eGFRcr, estimated glomerular filtration rate from creatinine; eGFRcys, estimated glomerular filtration rate from cystatin C; mUcrV, measured creatinine excretion; NT-proBNP, N-terminal pro b-type natriuretic peptide; NYHA, New York Heart Association.
Jugular venous distension was not assessed in 13% of patients.
Calculated from the validated equation of Wang et al13: muscle mass (kg) = 18.9 ×24-hour UcrV (grams) +4.1.
Calculated from the 6-hour timed urine collection.
Estimated from the Chronic Kidney Disease Epidemiology Collaboration equation. P values for the comparison of low vs high muscle mass.
Associations of Cystatin C and Creatinine With Muscle Mass
As shown in Table 2 and Fig. 1, after adjusting for mClcr, both cystatin C and creatinine were associated with muscle mass (percent difference in cystatin C 23.3, 95% CI 15.4–31.9, P < .001; percent difference in creatinine 57.7, 95% CI 51.1–64.5, P < .001). Associations remained significant after adjusting for age, sex, race, BMI, and serum albumin (31.7, 95% CI 22.5–41.6 for cystatin C, and 59.9, 95% CI 52.3–67.7, for creatinine; both P < .001). Cystatin C and creatinine were also associated with BMI (percent difference in cystatin C 13.3, 95% CI 8.9–17.9, P < .001; percent difference in creatinine 9.3, 95% CI 5.4–13.4, P < .001 after adjusting for mClcr, age, sex, race, and serum albumin). Cystatin C showed a trend toward association with serum albumin, whereas creatinine did not (percent difference in cystatin C −4.2, 95% CI −8.5 to 0.4, P = .07; percent difference in creatinine 0.0, 95% CI −4.1 to 4.3, P = .99) after adjusting for mClcr, age, sex, and race. In the sensitivity analysis, results were consistent when different estimators of GFR were used (Supplementary Tables S1.1 to S1.3). In 232 patients, the creatinine was available after 24 hours and 218 had stable values (creatinine change of <0.3 mg/dL); results were also consistent in these patients (Supplementary Table S1.4).
Table 2.
Unadjusted and Adjusted Associations of Variables With Cystatin C and Creatinine
| IQR | Not adjusted Coeff.* (95% CI) |
Adjusted for mClcr Coeff.* (95% CI) |
Adjusted for mClcr, Age, Sex, and Race Coeff.* (95% CI) |
Adjusted for mClcr, Age, Sex, Race, BMI, and Serum Albumin Coeff.* (95% CI) |
|
|---|---|---|---|---|---|
| Cystatin C† (mg/L) | |||||
| Muscle mass† (kg) | 0.567 | −16.1 (−20.8 to −11.2)‡ | 23.3 (15.4 to 31.9)§ | 35.1 (25.2 to 45.7)§ | 31.7 (22.5 to 41.6)§ |
| mClcr† (mL/min/1.73 m2) | 0.894 | −29.7 (−32.9 to −26.3) | — | — | — |
| Age (years) | 17 | 14.8 (8.6 to 21.5) | 1.5 (−3.2 to 6.5) | — | — |
| Female | 1 | 9.7 (−0.3 to 20.7) | 1.4 (−5.9 to 9.1) | — | — |
| Black race | 1 | −10.7 (−19.2 to −1.3) | −1.1 (−8.4 to 7.0) | — | — |
| BMI (kg/m2) | 11 | 0.9 (−4.2 to 6.2) | 11.9 (7.7 to 16.4) | 13.0 (8.5 to 17.6) | — |
| Serum albumin (g/dL) | 0.59 | −11.1 (−16.1 to −5.7) | −4.1 (−8.4 to 0.4) | −4.2 (−8.5 to 0.4) | — |
| Creatinine† (mg/dL) | |||||
| Muscle mass† (kg) | 0.567 | −2.6 (−7.8 to 2.7)‡ | 57.7 (51.1 to 64.5)§ | 62.0 (54.3 to 70.0)§ | 59.9 (52.3 to 67.7)§ |
| mClcr† (mL/min/1.73 m2) | 0.894 | −25.1 (−28.4 to −21.7) | — | — | — |
| Age (years) | 17 | 5.3 (−0.1 to 10.9) | −5.9 (−10.0 to −1.5) | — | — |
| Female | 1 | −6.3 (−13.9 to 2.2) | −12.4 (−18.2 to −6.2) | — | — |
| Black race | 1 | 5.4 (−3.7 to 15.3) | 15.2 (7.2 to 23.8) | — | — |
| BMI (kg/m2) | 11 | −0.1 (−4.6 to 4.7) | 8.8 (4.7 to 13.0) | 9.2 (5.3 to 13.3) | — |
| Serum albumin (g/dL) | 0.59 | −6.2 (−11.1 to −1.1) | −0.1 (−4.3 to 4.5) | 0.0 (−4.1 to 4.3) | — |
Each row shows different models based on the variable in the first column. The second column shows that continuous variables are expressed as interquartile range (which is the difference between the 25th and 75th percentiles) and categorical variables show the presence of the predictor. The third column shows unadjusted associations of variables with cystatin C or creatinine. The fourth column shows associations of variables adjusted for mClcr. The fifth column shows associations adjusting for mClcr, age, sex, and race. The sixth column shows associations adjusting for mClcr, age, sex, race, BMI, and serum albumin. The IQR for nontransformed muscle mass was 14 kg. Conversion factors for units: albumin in g/dL to g/L, × 10.
Parameter estimates were transformed as 100 × (ecoeff – 1) such that values can be interpreted as the geometric mean percent difference in the filtration marker for an interquartile range difference in continuous predictors or for the presence of categorical predictors.
Log-transformed variable.
Association was stronger for cystatin C than for creatinine (P < .001).
Associations were stronger for creatinine than for cystatin C (P < .001). The comparison of associations was done with models that included standardized log-cystatin C or standardized log-creatinine.BMI, body mass index; eGFR, estimated glomerular filtration rate; IQR, interquartile range; mClcr, measured clearance of creatinine.
Fig. 1.
Association of cystatin C and creatinine with muscle mass. After adjusting for mClcr, age, sex and race, muscle mass showed a positive association with both creatinine and cystatin C (both P < .001). The red line and the dashed orange line represent the association of log transformed muscle mass (restricted cubic spline) with log-transformed cystatin C and creatinine, respectively. The blue area represents 95% confidence interval. The x and y axes are on a log-scale; values are shown in natural scale for easier interpretation. CI, confidence interval; Cr, serum creatinine (mg/dL); Cys, serum cystatin C (mg/L); mClcr, measured clearance of creatinine.
Performance of eGFRcys and eGFRcr Compared With mClcr
Supplementary Table S2 shows the performance of eGFRcys and eGFRcr compared with mClcr. In the overall cohort, eGFRcys was unbiased, whereas eGFRcr overestimated mClcr (median difference between eGFRcys and eGFRcr compared with mClcr was −0.9 mL/min/1.73 m2, 95% CI −2.5–0.7 mL/min/1.73 m2vs 6.9 mL/min/1.73 m2, 95% CI 4.7–9.0 mL/min/1.73 m2, respectively, P < .001). However, precision was similar (IQR for the difference eGFRcys 21.4 mL/min/1.73 m2, 95% CI 17.4–25.3 mL/min/1.73 m2vs eGFRcr 20.2 mL/min/1.73 m2, 95% CI 16.9–23.5 mL/min/1.73 m2) as was accuracy (1-P30 of eGFRcys 45.4% vs eGFRcr 48.1%, P = .44).
The performance of both eGFRcys and eGFRcr differed by muscle mass (Fig. 2). In patients with a lower muscle mass, both eGFRcys and eGFRcr overestimated the mClcr (median difference 4.8 mL/min/1.73 m2, 95% CI 3.0–6.6 mL/min/1.73 m2, and 15.8 mL/min/1.73 m2, 95% CI 13.0–18.7 mL/min/1.73 m2, respectively), although the overestimation was greater with eGFRcr (P < .001). In patients with a higher muscle mass, eGFRcys underestimated mClcr but eGFRcr did not (median difference, −9.6 mL/min/1.73 m2, 95% CI −15.0 to −4.3 mL/min/1.73 m2, and −2.3 mL/min/1.73 m2, 95% CI −5.5 to 0.7 mL/min/1.73 m2, respectively), and the underestimation was greater with eGFRcys (P < .001). Accuracy was statistically better with eGFRcys compared with eGFRcr in patients with lower muscle mass, though both performed poorly (1-P30, 48.3%, vs 70.0%, respectively, P < .001); on the contrary, accuracy was worse with eGFRcys compared with eGFRcr in higher muscle mass (1-P30, 42.5%, vs 26.0%, respectively, P < .001).
Fig. 2.
Association of difference in eGFR–mClcr versus muscle mass. As shown, eGFRcys (left) and eGFRcr (right) were affected in the same direction by the amount of muscle mass–overestimation occurred at lower muscle mass and underestimation at higher muscle mass, though the magnitude of the association was greater for eGFRcr. The median difference of eGFRcys and eGFRcr vs mClcr in patients whose muscle mass was under the median (left side of the black line) was 4.8 and 15.8 mL/min/1.73 m2, respectively. In contrast, underestimation occurred in patients whose muscle mass was above the median: the median difference of eGFRcys and eGFRcr vs mClcr was −9.6 and −2.3 mL/min/1.73 m2, respectively. The x axis is on log-scale; values are shown in natural scale for easier interpretation. eGFR, estimated glomerular filtration rate; eGFRcr, estimated glomerular filtration rate from creatinine; eGFRcys, estimated glomerular filtration rate from cystatin C; mClcr, measured clearance of creatinine.
In the sensitivity analysis, results were similar when different estimators of GFR were used (Supplementary Table S2 and Supplementary Figs.).
Survival
At a median follow-up of 394 days, 40 patients had died. In univariate analysis, muscle mass (P = .006), mClcr (P = .044), and eGFRcys (P = .040) were predictors of mortality, whereas eGFRcr was not (P = .41). The magnitude of the association between eGFRcys and mortality was attenuated and lost statistical significance when adjusting for muscle mass (P = .20). Likewise, mClcr did not predict mortality after adjusting for muscle mass (P = .8) (Table 3). Muscle mass, however, remained associated with mortality after adjusting for eGFRcys (hazard ratio 0.42, 95% CI 0.20–0.90, P = .025), and showed a trend after adjusting for mClcr (hazard ratio 0.32, 95% CI 0.10–1.05, P = .06). Fig. 3 shows that, after stratifying by muscle mass below the cutoff value for sarcopenia, a similar survival was observed in high vs low eGFRcys and high vs low eGFRcr.
Table 3.
Univariate and multivariable analysis of survival.
| Variable | Univariate Analysis |
Adjusted for Muscle Mass |
||
|---|---|---|---|---|
| HR (95% CI) | P Value | HR (95% CI) | P Value | |
| Log eGFRcys | 0.53 (0.29–0.97) | .04 | 0.67 (0.36–1.24 | .20 |
| Log eGFRcr | 0.75 (0.38–1.48) | .41 | 0.93 (0.47–1.85) | .83 |
| Log mClcr | 0.63 (0.40–0.99) | .04 | 1.11 (0.53–2.36) | .77 |
| Log muscle mass | 0.37 (0.19–0.75) | .006 | — | — |
Each row represents a different model.
CI, confidence interval; eGFR, estimated glomerular filtration rate; eGFRcr, estimated glomerular filtration rate from creatinine; eGFRcys, estimated glomerular filtration rate from cystatin C; HR, hazard ratio; mClcr, measured clearance of creatinine.
Fig. 3.
Survival according to eGFRcys and eGFRcr stratified by muscle mass. After stratifying by muscle mass below the cutoff value for sarcopenia, similar survival was observed in high vs low eGFRcys (hazard ratio [HR] 0.73, 95% confidence interval [CI] 0.38–1.42, P = .36), but different survival was observed in low vs high muscle mass (HR 2.8, 95% CI 1.44–5.27, P = .002). Likewise, survival was not different in high vs low eGFRcr (HR 0.92, 95% CI 0.49–1.72, P = .78), while it was in low vs high muscle mass (HR 2.96, 95% CI 1.58–5.56, P = .001). Low muscle represents <20 kg of muscle mass for men or <15 kg for women.(17) Low eGFRcys indicates below the median of eGFRcys (40 mL/min/1.73 m2). Low eGFRcr indicates below the median of eGFRcr (52 mL/min/1.73 m2). eGFR, estimated glomerular filtration rate; eGFRcr, estimated glomerular filtration rate from creatinine; eGFRcys, estimated glomerular filtration rate from cystatin C; Musc, muscle mass.
Discussion
The GFR is most commonly estimated using creatinine, but that measure is less accurate in patients with severe sarcopenia owing to the association of creatinine with muscle mass. Cystatin C is hypothesized to be less affected by muscle mass than creatinine, which would provide an advantage in estimating the GFR in patients with sarcopenia and HF. However, in the present study we demonstrated several lines of evidence to suggest that this hypothesis might not be correct, and that the use of eGFRcys to replace eGFRcr is questionable in patients with sarcopenia. First, higher levels of cystatin C were associated with greater muscle mass. Although the magnitude of the association of muscle mass with cystatin C was lower than with creatinine, it remained present and of a reasonable magnitude that might be clinically relevant. Second, although eGFRcys was unbiased compared with mClcr, it led to overestimation and underestimation at lower and higher levels of muscle mass, respectively, which argues against unaffected performance of eGFRcys in this population. Finally, although eGFRcys had a stronger association with mortality than eGFRcr, the association did not persist after adjustment for muscle mass. These findings differ from the commonly held notion that eGFRcys is not affected by sarcopenia in patients with HF.
Like creatinine, cystatin C also has determinants other than GFR. These non-GFR determinants are less well-understood, but are thought to include fat mass and inflammation, both of which might correlate with decreased muscle mass.16 As such, the association of cystatin C with muscle mass might be secondary to these other effects. In this study, patients with lower muscle mass showed a lower BMI and, likely, a lower fat mass. Indeed, it is well-recognized that decreased weight in chronic illness such as cardiac cachexia involves both muscle and fat loss.4 Most studies examining non-GFR determinants have not been able to distinguish muscle mass vs fat mass. For example, in chronic kidney disease, elderly, and the general population, after adjusting for measured GFR, higher BMI, height, and weight were associated with higher cystatin C.12,17-19 However, in 2 studies of otherwise healthy patients with chronic kidney disease, muscle mass, as measured by dualenergy x-ray absorptiometry or creatinine excretion, was positively associated with cystatin C after adjusting for measured GFR.11,12 Notably, in the current cohort of patients with HF, the median BMI was well within the obese range and even the bottom quartile of BMI is considered overweight. Given that it is unlikely that fluid weight solely drove the higher BMI in this cohort, it is possible that muscle mass also directly affects cystatin C. Cystatin C is also thought to be associated with higher levels of inflammation separate from the effect on GFR.12,19 We did observe associations between higher cystatin C, but not creatinine, with lower serum albumin, although the CI crossed 0 after adjustment for GFR. The opposing effect on cystatin C by inflammation and muscle or fat mass might explain the weaker association of cystatin C with muscle mass compared with that of creatinine.
We observed an overestimation with both eGFRcys and eGFRcr at lower muscle, and an underestimation of eGFRcys, but not eGFRcr, at higher levels of muscle mass. The overestimation with both eGFRcys and eGFRcr has been observed in other sarcopenic populations.20 However, the underestimation with eGFRcys at higher levels of muscle mass compared with the unbiased estimation with eGFRcr has not been previously reported. One study did show a correlation between the bias in eGFRcys and BMI (r =−0.50, P = .0001), indicating overestimation at lower BMI and underestimation at higher BMI. Interestingly, the correlation between the bias in eGFRcr and BMI was less strong (r =−0.30, P = .02).16 In the present study, we also found a stronger association between BMI and cystatin C compared with BMI and creatinine. These results would suggest that at higher muscle mass the demographic parameters (age, sex, and race) to account for variations in levels of creatinine and cystatin C help to provide a more accurate estimation of GFR with eGFRcr than with eGFRcys.
Prior reports have demonstrated that the eGFRcys is more strongly associated with mortality than eGFRcr in general population, chronic kidney disease, cirrhosis, and HF.9,10,21,22 It is not clearly understood whether cystatin C predicts prognosis predominantly as a filtration marker or owing to the association of its non-GFR determinants with other risk factors.23 In this study, we found that, although the eGFRcys predicted mortality, associations became nonsignificant when adjusted for muscle mass. This observation suggests that the eGFRcys has greater prediction of events owing its non-GFR associations rather than its more accurate prediction of measured GFR.24
Our results should be interpreted considering several limitations. First, the GFR was not measured with a gold standard technique and the observed error in the performance of eGFR could rather reflect the error in the measured GFR. However, mClcr is a valuable method as long as urine collection and creatinine secretion are taken into account.2 Urine over- or under-collection was mediated using rigorous, timed collection methods, which our team has successfully done in previous studies.25,26 Tubular secretion of creatinine leads to overestimation of measured GFR by mClcr.27 Therefore, if mClcr had introduced significant bias, eGFRcys and eGFRcr would have shown only underestimation compared with mClcr, whereas we showed overestimation. Second, the gold standard technique for body composition (muscle, fat, etc) is cadaver analysis28; thus, no gold standard method for body composition exists. However, muscle mass estimation from creatinine excretion is a well-validated method13,29 and, despite the extrarenal clearance of creatinine at lower GFR, calculation of muscle mass seems to perform similar to energy x-ray absorptiometry and only is affected in predialysis or dialysis patients.30 In addition, muscle mass estimated from creatinine excretion was a strong predictor of mortality. Hence, even though our findings are consistent with alternative gold standard methods in non-HF populations, this remains a weakness because no direct measurements of muscle mass were done. Third, we did not assess fat mass, and hypothesized non-GFR determinant of cystatin C as well as greater loss of fat mass in patients with greater muscle wasting could have contributed to our results. Nevertheless, because patients with HF commonly have both decreased muscle and fat, it does not diminish the importance of our findings that eGFRcys did not perform well in this population. Other limitations of our study are that this was a single-center study and patients with normal BMI were underrepresented. Therefore, we state that our results should be considered hypothesis generating and stimulate further research to determine the role of cystatin C to estimate the GFR in patients with HF and sarcopenia or to explore different filtration markers.
Conclusions
Cystatin C levels are associated with muscle mass in patients with HF. Accordingly, GFR might be overestimated at a lower muscle mass and underestimated at a higher muscle mass. Although this finding might be an indirect association through the impact on the combined loss of muscle and fat mass in chronic illness, the critical message is that, in HF populations where muscle and fat loss are common, eGFRcys, like eGFRcr, should be used with caution. In clinical situations where highly accurate GFR is needed, direct measurements of GFR should be performed.31
Supplementary Material
Acknowledgments
Dr Ivey-Miranda reports funding from the Instituto Mexicano del Seguro Social. Dr Inker reports funding from the National Institutes of Health (NIH), the National Kidney Foundation, and grants from the Paul Teschan Research Fund outside the submitted work. This work was also supported by NIH grants K23HL114868, L30HL115790, R01HL139629, R21HL143092, R01HL128973 (to Dr Testani), K23DK097201 (to Dr Wilson), and 5T32HL007950 (to Dr Griffin). The funding sources had no role in study design, data collection, analysis, or interpretation.
Footnotes
Disclosures
Dr Inker has a patent Precise estimation of GFR from multiple biomarkers (PCT/US2015/044567) issued; funding from Retrophin, Omeros, and Reata Pharmaceuticals for research; contracts with Tufts Medical Center; and consulting agreements with Tricida and Omeros Corp. Dr Inker is on the medical advisory board for the Alport Syndrome Foundation. Dr Rao has a patent Precision treatment of HF and cardiorenal syndrome with royalties paid to Yale University, Dr Rao, and Dr Testani. Dr Tang reports grants and personal fees from Sequana Medical. Dr Testani reports grants and personal fees from Sequana Medical, grants and personal fees from BMS, personal fees from AZ, personal fees from Novartis, grants and personal fees from 3ive labs, personal fees from cardionomic, personal fees from Bayer, grants and personal fees from Boehringer Ingelheim, personal fees from MagentaMed, grants from Otsuka, personal fees from Renalguard, grants and personal fees from Sanofi, grants and personal fees from FIRE1, grants from Abbott, and personal fees from W.L. Gore outside the submitted work. Dr Ivey-Miranda, Dr Griffin, Dr Turner, and Dr Wilson have nothing to disclose.
Supplementary materials
Supplementary material associated with this article can be found in the online version at doi:10.1016/j.cardfail.2020.07.013.
References
- 1.Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney function–measured and estimated glomerular filtration rate. N Engl J Med 2006;354:2473–83. [DOI] [PubMed] [Google Scholar]
- 2.Levey AS, Coresh J, Balk E, Kausz AT, Levin A, Steffes MW, et al. National Kidney Foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Ann Intern Med 2003;139:137–47. [DOI] [PubMed] [Google Scholar]
- 3.Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, et al. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med 2012;367:20–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Rolfe M, Kamel A, Ahmed MM, Kramer J. Pharmacological management of cardiac cachexia: a review of potential therapy options. Heart Fail Rev 2019;24:617–23. [DOI] [PubMed] [Google Scholar]
- 5.Suzuki T, Palus S, Springer J. Skeletal muscle wasting in chronic heart failure. ESC Heart Fail 2018;5:1099–107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Beddhu S, Samore MH, Roberts MS, Stoddard GJ, Pappas LM, Cheung AK. Creatinine production, nutrition, and glomerular filtration rate estimation. J Am Soc Nephrol 2003;14:1000–5. [DOI] [PubMed] [Google Scholar]
- 7.Perrone RD, Madias NE, Levey AS. Serum creatinine as an index of renal function: new insights into old concepts. Clin Chem 1992;38:1933–53. [PubMed] [Google Scholar]
- 8.Vinge E, Lindergard B, Nilsson-Ehle P, Grubb A. Relationships among serum cystatin C, serum creatinine, lean tissue mass and glomerular filtration rate in healthy adults. Scand J Clin Lab Invest 1999;59:587–92. [DOI] [PubMed] [Google Scholar]
- 9.Manzano-Fernandez S, Flores-Blanco PJ, Perez-Calvo JI, Ruiz-Ruiz FJ, Carrasco-Sanchez FJ, Morales-Rull JL, et al. Comparison of risk prediction with the CKD-EPI and MDRD equations in acute decompensated heart failure. J Card Fail 2013;19:583–91. [DOI] [PubMed] [Google Scholar]
- 10.Zamora E, Lupon J, de Antonio M, Vila J, Penafiel J, Galan A, et al. Long-term prognostic value for patients with chronic heart failure of estimated glomerular filtration rate calculated with the new CKD-EPI equations containing cystatin C. Clin Chem 2014;60:481–9. [DOI] [PubMed] [Google Scholar]
- 11.Macdonald J, Marcora S, Jibani M, Roberts G, Kumwenda M, Glover R, et al. GFR estimation using cystatin C is not independent of body composition. Am J Kidney Dis 2006;48:712–9. [DOI] [PubMed] [Google Scholar]
- 12.Stevens LA, Schmid CH, Greene T, Li L, Beck GJ, Joffe MM, et al. Factors other than glomerular filtration rate affect serum cystatin C levels. Kidney Int 2009;75:652–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wang ZM, Gallagher D, Nelson ME, Matthews DE, Heymsfield SB. Total-body skeletal muscle mass: evaluation of 24-h urinary creatinine excretion by computerized axial tomography. Am J Clin Nutr 1996;63:863–9. [DOI] [PubMed] [Google Scholar]
- 14.Cruz-Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyere O, Cederholm T, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing 2019;48:16–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chew-Harris JS, Florkowski CM, George PM, Elmslie JL, Endre ZH. The relative effects of fat versus muscle mass on cystatin C and estimates of renal function in healthy young men. Ann Clin Biochem 2013;50:39–46. [DOI] [PubMed] [Google Scholar]
- 17.Foster MC, Levey AS, Inker LA, Shafi T, Fan L, Gudnason V, et al. Non-GFR determinants of low-molecular-weight serum protein filtration markers in the elderly: AGES-Kidney and MESA-Kidney. Am J Kidney Dis 2017;70:406–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Knight EL, Verhave JC, Spiegelman D, Hillege HL, de Zeeuw D, Curhan GC, et al. Factors influencing serum cystatin C levels other than renal function and the impact on renal function measurement. Kidney Int 2004;65:1416–21. [DOI] [PubMed] [Google Scholar]
- 19.Liu X, Foster MC, Tighiouart H, Anderson AH, Beck GJ, Contreras G, et al. Non-GFR determinants of low-molecular-weight serum protein filtration markers in CKD. Am J Kidney Dis 2016;68:892–900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mindikoglu AL, Dowling TC, Weir MR, Seliger SL, Christenson RH, Magder LS. Performance of chronic kidney disease epidemiology collaboration creatinine-cystatin C equation for estimating kidney function in cirrhosis. Hepatology 2014;59:1532–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Shlipak MG, Matsushita K, Arnlov J, Inker LA, Katz R, Polkinghorne KR, et al. Cystatin C versus creatinine in determining risk based on kidney function. N Engl J Med 2013;369:932–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Yoo JJ, Kim SG, Kim YS, Lee B, Lee MH, Jeong SW, et al. Estimation of renal function in patients with liver cirrhosis: impact of muscle mass and sex. J Hepatol 2019;70:847–54. [DOI] [PubMed] [Google Scholar]
- 23.Tangri N, Inker LA, Tighiouart H, Sorensen E, Menon V, Beck G, et al. Filtration markers may have prognostic value independent of glomerular filtration rate. J Am Soc Nephrol 2012;23:351–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Stevens LA, Levey AS. Chronic kidney disease in the elderly–how to assess risk. N Engl J Med 2005;352:2122–4. [DOI] [PubMed] [Google Scholar]
- 25.Charokopos A, Griffin M, Rao VS, Inker L, Sury K, Asher J, et al. Serum and urine albumin and response to loop diuretics in heart failure. Clin J Am Soc Nephrol 2019;14:712–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Testani JM, Hanberg JS, Cheng S, Rao V, Onyebeke C, Laur O, et al. Rapid and highly accurate prediction of poor loop diuretic natriuretic response in patients with heart failure. Circ Heart Fail 2016;9:e002370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.van Acker BA, Koomen GC, Koopman MG, Krediet RT, Arisz L. Discrepancy between circadian rhythms of inulin and creatinine clearance. J Lab Clin Med 1992;120:400–10. [PubMed] [Google Scholar]
- 28.Buckinx F, Landi F, Cesari M, Fielding RA, Visser M, Engelke K, et al. Pitfalls in the measurement of muscle mass: a need for a reference standard. J Cachexia Sarcopenia Muscle 2018;9:269–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Forbes GB, Bruining GJ. Urinary creatinine excretion and lean body mass. Am J Clin Nutr 1976;29:1359–66. [DOI] [PubMed] [Google Scholar]
- 30.Nielsen PK, Ladefoged J, Olgaard K. Lean body mass by dual energy x-ray Absorptiometry (DEXA) and by urine and dialy-sate creatinine recovery in CAPD and pre-dialysis patients compared to normal subjects. Adv Perit Dial 1994;10:99–103. [PubMed] [Google Scholar]
- 31.Levey AS, Coresh J, Tighiouart H, Greene T, Inker LA. Measured and estimated glomerular filtration rate: current status and future directions. Nat Rev Nephrol 2020;16:51–64. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.