Cystatin C Is More than GFR, and This May Be a Good Thing

For measuring GFR, an exogenous marker is injected into the patient with timed marker levels assayed in the urine and plasma (urinary clearance) or plasma alone (plasma clearance). The essential properties of the exogenous marker are that it is metabolically inert and cleared exclusively by glomerular filtration, but cost, time, and discomfort make measured GFR impractical in most clinical settings. Instead, endogenous markers have been widely used. Unlike exogenous markers, non-GFR determinants of endogenous markers exist but are sometimes difficult to interpret.

Estimated GFR (eGFR) based on serum creatinine (eGFR_Cr) uses age, gender, and race in the estimating equation to model the non-GFR determinants of serum creatinine (largely muscle mass).^1,2 However, these demographics do not fully account for non-GFR determinants of creatinine. In particular, GFR is higher at the same serum creatinine level in healthy individuals with higher muscle mass than those with chronic kidney disease (CKD) and lower muscle mass.^3,4 Cystatin C is another endogenous marker cleared by filtration, and its serum levels are more highly correlated with GFR than serum creatinine.⁵ The non-GFR determinants of cystatin C are less defined and more curious; cystatin C is a 14-kD protease inhibitor with anti-atherosclerotic activity in animal models^6,7 and is a strong predictor of mortality or cardiovascular disease (CVD) among individuals with normal eGFR_Cr^8,9 or CKD.^10–12

In this issue of JASN, Mathisen et al.¹³ provide novel insight into the non-GFR determinants of both serum creatinine and cystatin C. They found that higher serum creatinine levels (or lower eGFR_Cr) had a residual association with higher diastolic BP, not smoking, and increased physical activity. Nonsmokers and physically active individuals may be expected to have higher muscle mass, resulting in higher serum creatinine levels, potentially explaining this association. They also found that higher cystatin C levels (or lower eGFR_cysC) had a residual association with being a smoker, decreased physical activity, higher triglycerides, higher LDL cholesterol, lower HDL cholesterol, and obesity. This residual association with smoking and obesity has been previously reported, but previous studies adjusted for urinary creatinine clearance instead of measured GFR (mGFR).¹⁴ Mathisen et al.¹³ also found increased Framingham risk scores with lower eGFR_cysC but not with lower mGFR or lower eGFR_Cr. These findings argue that eGFR_cysC is a better predictor of CVD than GFR because the non-GFR determinants of cystatin C, possibly its anti-atherosclerotic activity, also reflect cardiovascular risk.

One potential objection to their conclusion is that eGFR may have a residual association with cardiovascular risk factors because mGFR is imprecise. In other words, could eGFR capture a residual association that reflects the true GFR signal missed as a result of error with mGFR? The study by Mathisen et al.¹³ suggests this hypothesis is unlikely. The residual associations with eGFR_Cr were different and sometimes in the opposite direction of residual associations with eGFR_cysC. If residual associations with eGFR could be fully explained by imprecision of mGFR, then residual associations with eGFR_Cr should be similar to residual associations with eGFR_cysC. Furthermore, a sensitivity analysis was performed assuming 30% of the variance in direct GFR measurement was error, but the residual associations with eGFR remained.

How should these findings influence clinical practice? A new cystatin C equation that uses obesity, smoking, and serum lipids to improve the estimation of GFR could be developed. However, such an equation would have substantial drawbacks. Obesity, smoking, and lipids are only correlates of the non-GFR determinants of cystatin C and do not fully capture all of the non-GFR determinants of cystatin C. Incorporating the variables obesity, smoking, and lipids into a new eGFR_cysC may lead to a more accurate estimate of GFR, but, paradoxically, the new eGFR_cysC would be less predictive of CVD than cystatin C alone. In particular, individuals who are obese, smoke, or have dyslipidemia would have their eGFR increased by the new cystatin C equation. In fact, by controlling for these cardiovascular risk factors, the new eGFR_cysC may be less predictive of CVD than GFR! Indeed, the variables used to model non-GFR determinants of a marker fundamentally influences how eGFR predicts outcomes. For example, the use of age to model the non-GFR determinants of serum creatinine inflates mortality risk estimates with eGFR_Cr because age itself is a potent predictor of mortality.¹⁵

The study by Mathisen et al.¹³ has a few limitations worth noting. Patients who reported renal disease, diabetes, or CVD were excluded. Similar studies of less select samples are needed. All analyses were adjusted for age and gender. However, the residual associations between eGFR and cardiovascular risk factors with adjustment only for mGFR should have been provided for three reasons. First, age and gender are variables used to calculate eGFR_Cr. Second, age and gender may correlate with the non-GFR determinants of cystatin C. Third, there has been no adjustment for age and gender with the use of eGFR to define CKD.

Given this residual association of cardiovascular risk factors with cystatin C levels, it might seem that cystatin C should not be used as a kidney function test. Perhaps the focus with cystatin C should be to improve prediction of clinical outcomes instead of optimizing the estimation of GFR. To the extent that cystatin C helps identify patients at higher risk for kidney failure, mortality, and CVD not detected by serum creatinine, it is useful. If the incremental improvement in risk prediction with cystatin C is due in part to its non-GFR determinants, then one might argue that cystatin C is not a pure kidney marker. However, much of the variation in GFR itself is not due to parenchymal injury in the kidney. Indeed, there is no association between mGFR and nephrosclerosis on renal biopsy among normal adults after controlling for age.¹⁶ Variation in GFR can be due to nonrenal factors such as dietary protein intake, volume status, hemodynamics, or even the indexing of GFR to body surface area.¹⁷

Perhaps the most useful application of cystatin C is as a confirmatory test for individuals with an eGFR_Cr <60 ml/min per 1.73 m², where cystatin C identifies the subset with nearly all of the increased risk for kidney failure, cardiovascular events, and mortality.¹⁸ If cystatin C can find high-risk patients for whom targeted management is beneficial, then it is clinically useful. This should be the focus instead of a more exact GFR estimate.

• Copyright © 2011 by the American Society of Nephrology