What Is the Goal With Endogenous Filtration Markers—Estimation of GFR or Prediction of Kidney Outcomes?

Abstract

Related Article, p. 886 Related Article, p. 886 A narrative common to nephrology is the following: “Glomerular filtration rate (GFR) is the best overall marker of kidney function. Therefore, optimizing the estimation of GFR from endogenous filtration markers will improve the diagnosis and management of patients with chronic kidney disease (CKD).” In this issue of the American Journal of Kidney Diseases, Bhavsar and colleagues present data from the African American Study of Kidney Disease and Hypertension (AASK) that challenge this narrative.1Bhavsar N.A. Appel L.J. Kusek J.W. et al.Comparison of measured GFR, serum creatinine, cystatin C, and beta-trace protein to predict ESRD in African Americans with hypertensive CKD.Am J Kidney Dis. 2011; 58: 886-893Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar A cohort of 865 African Americans with hypertensive CKD had a baseline visit with measured GFR and serum levels of endogenous filtration markers. The cohort was followed for the onset of end-stage renal disease (ESRD) or mortality. The investigators found that beta-trace protein (BTP) concentration had stronger discriminatory power than cystatin C concentration or serum creatinine–based estimated GFR (eGFRSCr) for predicting ESRD (highest vs lowest quintile risk ratios of 38, 25, and 14, respectively) and ESRD or mortality (highest vs lowest quintile risk ratios of 11, 10, and 6, respectively). However, BTP was not better correlated with measured GFR than cystatin C level or eGFRSCr (r values of 0.71, 0.72, and 0.78, respectively). In essence, BTP showed the best discriminatory prediction for kidney outcomes and this was not explained by better correlation with GFR. For any analysis comparing different endogenous filtration markers it is important to understand how GFR is measured. GFR is directly determined from the urinary or plasma clearance of an exogenous marker that is biologically inert and exclusively cleared by glomerular filtration. In the study by Bhavsar and colleagues, the exogenous marker iothalamate was used to determine GFR.1Bhavsar N.A. Appel L.J. Kusek J.W. et al.Comparison of measured GFR, serum creatinine, cystatin C, and beta-trace protein to predict ESRD in African Americans with hypertensive CKD.Am J Kidney Dis. 2011; 58: 886-893Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar Large series have found no substantive difference between a simultaneous iothalamate clearance and inulin clearance, the “gold standard” for measured GFR.2Maher F.T. Nolan N.G. Elveback L.R. Comparison of simultaneous clearances of 125-I-labeled sodium iothalamate (Glofil) and of inulin.Mayo Clin Proc. 1971; 46: 690-691PubMed Google Scholar, 3Ott N.T. Wilson D.M. A simple technique for estimating glomerular filtration rate with subcutaneous injection of (125I)Iothalamate.Mayo Clin Proc. 1975; 50: 664-668PubMed Google Scholar Exogenous markers are inherently superior to endogenous markers because they have no physiological role in human metabolism and their infusion or excretion rate can be determined. With endogenous markers, the serum concentration reflects both GFR and non-GFR determinants. The non-GFR determinants include the rates of generation, tubular reabsorption and secretion, and the extrarenal elimination of the marker. Figure 1 provides conceptual scenarios for relating endogenous filtration markers to GFR and outcomes. If the non-GFR determinants of a hypothetical endogenous filtration marker were uniform (constant over time and no variation across the spectrum of health and disease), then this marker could be viewed as representing GFR only (Fig 1, scenario A). Is it plausible that such an endogenous filtration marker even exists? Serum creatinine is the most widely used endogenous filtration marker. The primary non-GFR determinant of creatinine is generation from muscle mass, which shows significant variation in different populations and disease states. eGFRSCr uses age, sex, and race to model muscle mass. This has helped increase the recognition of advanced kidney disease in older white women who generally have less muscle mass than younger black men. However, demographics inadequately capture all variation in muscle mass that affects eGFRSCr.4Rule A.D. Bailey K.R. Schwartz G.L. Khosla S. Lieske J.C. Melton L.J. For estimating creatinine clearance measuring muscle mass gives better results than those based on demographics.Kidney Int. 2009; 75: 1071-1078Crossref PubMed Scopus (86) Google Scholar Importantly, CKD itself is associated with loss of muscle mass and this has made the application of any one eGFRSCr equation to all patient populations problematic.5Murata K. Baumann N.A. Saenger A.K. Larson T.S. Rule A.D. Lieske J.C. Relative performance of the MDRD and CKD-EPI equations for estimating glomerular filtration rate among patients with varied clinical presentations.Clin J Am Soc Nephrol. 2011; 6: 1963-1972Crossref PubMed Scopus (126) Google Scholar Consistent with scenario B (Fig 1), serum creatinine may be a weaker predictor of outcomes than GFR because the rise in serum creatinine with a reduction of GFR is blunted by loss of muscle mass with more severe CKD.6Workeneh B.T. Mitch W.E. Review of muscle wasting associated with chronic kidney disease.Am J Clin Nutr. 2010; 91: 1128S-1132SCrossref PubMed Scopus (198) Google Scholar Bhavsar and colleagues found serum creatinine and eGFRSCr had weaker discriminatory power than measured GFR in predicting ESRD or mortality (highest vs lowest quintile risk ratios of 6, 6, and 10, respectively). A difference in discriminatory power for the endpoint of ESRD alone was not evident, but since eGFRSCr and serum creatinine are often used to determine the onset of ESRD, this finding may have been due to bias. Cystatin C and BTP are low-molecular-weight protein enzymes that have been used as endogenous filtration markers. Enzyme concentrations are controlled by numerous mechanisms and feedback loops, including at the level of transcription of the gene, protein translation, and posttranslational modifications. Hence, there is no clear reason to believe that the generation rate of these enzymes would be uniform across the spectrum of health and disease (Fig 1, scenario A). CKD is associated with a wide array of metabolic abnormalities (eg, acidosis, hyperkalemia, and hyperphosphatemia) and many of these abnormalities are independent predictors of progression to ESRD.7Tangri N. Stevens L.A. Griffith J. et al.A predictive model for progression of chronic kidney disease to kidney failure.JAMA. 2011; 305: 1553-1559Crossref PubMed Scopus (666) Google Scholar The non-GFR determinants of these enzyme levels may also change with CKD severity and inform the risk of progression to ESRD. Receiving an organ transplant, obesity, and inflammation have been associated with higher cystatin C levels independent of GFR.8Rule A.D. Bergstralh E.J. Slezak J.M. Bergert J. Larson T.S. Glomerular filtration rate estimated by cystatin C among different clinical presentations.Kidney Int. 2006; 69: 399-405Crossref PubMed Scopus (323) Google Scholar, 9Stevens L.A. Schmid C.H. Greene T. et al.Factors other than glomerular filtration rate affect serum cystatin C levels.Kidney Int. 2009; 75: 652-660Crossref PubMed Scopus (463) Google Scholar This can be explained by the non-GFR determinants of cystatin C varying with these clinical characteristics. Cystatin C is also the most abundant endogenous inhibitor of cathepsins, a family of proteases involved in atherogenesis.10Lafarge J.C. Naour N. Clement K. Guerre-Millo M. Cathepsins and cystatin C in atherosclerosis and obesity.Biochimie. 2010; 92: 1580-1586Crossref PubMed Scopus (93) Google Scholar If the generation rate of cystatin C increases with atherosclerosis and if atherosclerosis contributes to CKD progression, then cystatin C would be expected to better predict kidney outcomes than GFR (Fig 1, scenario C). Clinical characteristics that associate with the non-GFR determinants of BTP are less clear. BTP is synthesized primarily in the cerebrospinal fluid and can be used to detect cerebrospinal fluid leaks.11Sampaio M.H. de Barros-Mazon S. Sakano E. Chone C.T. Predictability of quantification of beta-trace protein for diagnosis of cerebrospinal fluid leak: cutoff determination in nasal fluids with two control groups.Am J Rhinol. 2009; 23: 585-590Crossref Scopus (21) Google Scholar One could speculate that increased serum levels may occur from cerebrovascular disease, a risk factor for CKD progression.12Kobayashi M. Hirawa N. Morita S. et al.Silent brain infarction and rapid decline of kidney function in patients with CKD: a prospective cohort study.Am J Kidney Dis. 2010; 56: 468-476Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar The proportional contributions of risk factors for decreased kidney function are also remarkably similar to risk factors for cerebrovascular disease.13Turner S.T. Rule A.D. Schwartz G.L. et al.Risk factor profile for chronic kidney disease is similar to risk factor profile for small artery disease.J Hypertens. 2011; 29: 1796-1801Crossref PubMed Scopus (13) Google Scholar Bhavsar and colleagues found BTP and cystatin C levels had stronger discriminatory power than measured GFR in predicting ESRD (highest vs lowest quintile risk ratios of 38, 25, and 13, respectively). An increase in the generation rate of BTP or cystatin C being predictive of ESRD could be consistent with scenario C (Fig 1). Bhavsar and colleagues found that endogenous filtration markers were all associated with kidney outcomes independent of measured GFR.1Bhavsar N.A. Appel L.J. Kusek J.W. et al.Comparison of measured GFR, serum creatinine, cystatin C, and beta-trace protein to predict ESRD in African Americans with hypertensive CKD.Am J Kidney Dis. 2011; 58: 886-893Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar As acknowledged, there are some important limitations to the GFR-adjusted analysis. GFR, like all laboratory tests, has measurement error (within-individual variability and assay error). When GFR is the dependent variable for developing estimating equations (eGFR), this measurement error has a negligible affect; standard regression models minimize the consequences of error in the dependent variable. However, when the dependent variable is a clinical outcome (ESRD) and GFR is a predictor, this measurement error becomes much more relevant. Bhavsar and colleagues used the average of 2 GFR measurements to help decrease this error, but that may not have been sufficient. Furthermore, when measured GFR and endogenous filtration markers are both predictor variables in the same model, there is likely a statistical problem known as multicollinearity. Because of these issues (error in predictor variable, and correlation between predictors), the GFR component of endogenous filtration markers will still contribute to prediction of outcomes. One cannot safely conclude that an association between an endogenous filtration marker and a clinical outcome is not due to GFR, even though nominal adjustment for measured GFR has been done. A more sophisticated statistical analysis is needed to determine the extent the non-GFR versus GFR determinants of the endogenous filtration marker contributes to the prediction of kidney outcomes. Much of the general literature on endogenous filtration markers has focused on how well markers predict clinical outcomes such as ESRD and mortality. But is predictive validity all that matters when evaluating markers? Even though measured GFR appears to be inferior to cystatin C and BTP in predicting kidney outcomes, GFR is an established measure of kidney function. In contrast, it remains unclear how the non-GFR determinants of these markers relate to kidney function and these outcomes. If the non-GFR determinants of cystatin C and BTP reflect systemic conditions such as atherosclerosis or inflammation, then the non-GFR determinants of cystatin C and BTP may be predictive of kidney outcomes but not reflective of underlying functional or structural changes specific to the kidney. To better understand the pathophysiology of CKD it would be helpful to relate markers to underlying kidney pathology in addition to GFR and clinical outcomes. Studies relating GFR and endogenous filtration markers to structural characteristics of CKD by biopsy or imaging (eg, glomerulosclerosis, tubular atrophy, interstitial fibrosis, and cortical thinning) are needed. Interestingly, Bhavsar and colleagues found with multivariable models that included known risk factors for CKD progression (eg, proteinuria), measured GFR had higher discriminatory power for predicting ESRD or mortality than any of the endogenous filtration markers.1Bhavsar N.A. Appel L.J. Kusek J.W. et al.Comparison of measured GFR, serum creatinine, cystatin C, and beta-trace protein to predict ESRD in African Americans with hypertensive CKD.Am J Kidney Dis. 2011; 58: 886-893Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar Perhaps the non-GFR determinants of BTP and cystatin C are correlated with established risk factors for CKD outcomes. Even so, a strong case has been made to incorporate cystatin C and BTP into risk prediction for ESRD. Additional studies are needed, however, to determine how best to incorporate these markers into the management of patients with CKD. A new narrative is needed: “CKD is associated with significant morbidity and mortality. An optimal combination of kidney markers needs to be identified to help determine risk for adverse outcomes, to help characterize the underlying renal pathology, and for targeted interventions to prevent morbidity and mortality.” Financial Disclosure: The authors declare that they have no relevant financial interests. Comparison of Measured GFR, Serum Creatinine, Cystatin C, and Beta-Trace Protein to Predict ESRD in African Americans With Hypertensive CKDAmerican Journal of Kidney DiseasesVol. 58Issue 6PreviewIdentification of persons with chronic kidney disease (CKD) who are at highest risk to progress to end-stage renal disease (ESRD) is necessary to reduce the burden of kidney failure. The relative utility of traditional markers of kidney function, including estimated glomerular filtration rate (eGFR) and serum creatinine level, and emerging markers of kidney function, including cystatin C and beta-trace protein (BTP) levels, to predict ESRD and mortality has yet to be established. Full-Text PDF