- Email: [email protected]
Response to ‘Calculation of glomerular filtration rate using serum cystatin C in kidney transplant recipients’
letter to the editor
Kobayashi and Noiri do not argue against the validity of our reported significant correlations of TT and UFR with mortality risk, but suggest that our report may offer an underestimate of the true magnitude of potential benefits from longer TT and slower UFR. 1. Okamoto K, Kobayashi S, Noiri E. Longer treatment time and slower ultrafiltration in hemodialysis: associations with mortality in the Dialysis Outcomes and Practice Patterns Study. Kidney Int (in press). 2. Saran R, Bragg-Gresham JL, Port FK et al. Longer treatment time and slower ultrafiltration in hemodialysis: associations with reduced mortality in the DOPPS. Kidney Int 2006; 69: 1222–1228. 3. Saran R, Bragg-Gresham JL, Rayner HC et al. Nonadherence in hemodialysis: associations with mortality, hospitalization, and practice patterns in the DOPPS. Kidney Int 2003; 64: 254–262.
differences between the different laboratories may counteract the putative advantages of the new TX formula, (2) this rather disappointing performance of the TX equation may also be due to possible confounders like steroid dosing which may crucially affect Cys C levels.4 To enhance the performance of future Cys C-based glomerular filtration rate equations such cofactors should be taken into account. 1.
2.
R Saran1, JL Bragg-Gresham2, FK Port2 and B Gillespie1
3.
1
Division of Nephrology, Kidney Epidemiology and Cost Center, University of Michigan, Ann Arbor, Michigan, USA and 2Arbor Research Collaborative for Health (Formerly URREA), Ann Arbor, Michigan, USA Correspondence: R Saran, Division of Nephrology, Kidney Epidemiology and Cost Center, University of Michigan, 315 W. Huron, Suite 240, Ann Arbor, Michigan 48103-4262, USA. E-mail: [email protected]
4.
Rule AD, Bergstralh EJ, Slezak JM et al. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: 399–405. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate expressed in ml/min from plasma cystatin C values in mg/l. Scand J Clin Lab Invest 2004; 64: 25–30. Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003; 18: 2024–2031. Poge U, Gerhardt TM, Stoffel-Wagner B et al. {beta} Trace protein is an alternative marker for glomerular filtration rate in renal transplantation patients. Clin Chem 2005; 51: 1531–1533.
U Po¨ge1, T Gerhardt1 and RP Woitas1 1
Department of Internal Medicine I, University of Bonn, Bonn, Germany Correspondence: U Po¨ge, Department of Internal Medicine I, University of Bonn, Bonn D-53179, Germany. E-mail: [email protected]
Calculation of glomerular filtration rate using serum cystatin Response to ‘Calculation of C in kidney transplant recipients glomerular filtration rate using Kidney International (2006) 70, 1878. doi:10.1038/sj.ki.5001843 1 To the Editor: Recently, Rule et al. demonstrated a 19% serum cystatin C in kidney higher glomerular filtration rate at the same cystatin C (Cys transplant recipients’ C) level among patients after renal transplantation in comparison to patients with native kidney disease. Thus, a new Cys C-based formula (glomerular filtration rate76.6 Cys C1.16) was suggested for transplant recipients (TX formula). We analyzed the diagnostic performance of the new TX formula in comparison to two other Cys C formulae (Larsson and Hoek2,3) which are based on the same Cys C assay in a cohort of 108 patients after renal transplantation. Glomerular filtration rate was determined by 99mtechnetiumlabeled diethylenetriamine penta acetate clearance. Results are given in Table 1. Although the Larsson and Hoek formulae were not derived from a transplanted cohort, their diagnostic performances are at least comparable to the TX equation. Thus, two conclusions can be drawn from this analysis: (1) calibration
Kidney International (2006) 70, 1878–1879. doi:10.1038/sj.ki.5001828
We appreciate the work by Po¨ge et al.1 to test the performance of our transplant equation.2 Remarkably, the equation performed well with little bias (1.6 ml/min/ 1.73 m2) in their transplant recipients. There was also little bias with the Larsson3 and Hoek4 equations, which were not specifically developed using transplant recipients. However, we note that our finding of a higher glomerular filtration rate (GFR) in transplant recipients (kidney or other organ) compared to native chronic kidney disease (CKD) patients is consistent with reports by other investigators.5,6 In these centers, one equation cannot accurately estimate GFR in both transplant and native CKD patients unless it includes variables for transplant
Table 1 | Comparison of performance of the different cystatin C based formulae Range Mean estimates (ml/min/1.73 m2) (ml/min/1.73 m2) DTPA Larsson Hoek Rule
39.5 36.3 38.9 37.9
Correlation coefficient
11.8–82.9 7.78–104 8.72–97.4 9.30–101
0.859 0.865 0.862
Median Bias difference (ml/min/1.73 m2) (ml/min/1.73 m2) 3.20 0.58 1.60
4.78 1.50 2.78
Precision (ml/min/ 1.73 m2) 9.59 8.64 9.15
Accuracy within 30% (95% CI) 50% (95% CI) 77.1 77.1 78.0
95.4 97.2 89.0
CI, confidence interval; DTPA, diethylenetriamine penta acetate.
1878
Kidney International (2006) 70, 1877–1883
letter to the editor
Transplant
Native CKD
Hoek
Larsson
GFR (ml/min/1.73 m2)
150 100 60 30 15 10 0.5
0.7
1.0
1.5 2.0 Cystatin C (mg/l)
3.0
4.0 5.0
Figure 1 | Comparison of the transplant (gray solid) and native CKD (black solid) equations with the Hoek (black dashed) and Larsson (gray dashed) equations. Equations are linear or nearly linear on a logarithmic cystatin C (x axis) and logarithmic GFR (y axis) scale.
status or variables that better model the non-GFR factors that affect cystatin C levels. Equations are themselves a statistic of the average GFR at each cystatin C level for a sampled population. As shown in Figure 1, the bias between equations varies depending on the cystatin C level. If one corrected for this bias between equations, there would be no difference in the statistics that reflect model error (R2, root mean square error, precision, or accuracy within 30 or 50%). Additional variables or further sub-population stratification would be needed to decrease model error. It is possible that markers of inflammation7 or glucocorticoid use8–10 would better model the non-GFR variability with cystatin C than transplant status. Given potential calibration differences with cystatin C and ‘gold-standard’ GFR assays, it is a stronger study design to refit equations with new coefficients when comparing different populations, different serum analytes, or different statistical models. In addition, case mix and referral patterns for direct GFR measurement likely differ among centers. To a different extent between centers, patients referred for GFR measurements may be influenced by discordance between the clinical presentation and the serum creatinine. The Larsson equation3 models unstandardized GFR (ml/ min) instead of standardized GFR (ml/min/1.73 m2); however, body surface area is a predictor of unstandardized GFR independent of cystatin C level.11 If we had modeled unstandardized GFR with our data, inclusion of a body surface area term in the model would have increased our R2 from 0.700 to 0.777 in the transplant sample and from 0.807 to 0.864 in the native CKD sample. The extent that the average body surface area in the sample used to derive the Larsson equation differs from 1.73 m2 will contribute to bias between the Larsson equation and equations that predict standardized GFR. It is also worth noting that we used different data transformations for deriving equations compared to the Kidney International (2006) 70, 1877–1883
Hoek equation.4 We regressed ln GFR on ln cystatin C, instead of GFR on 1/cystatin C. If we had regressed GFR on 1/cystatin C with our data, the R2 would have been 0.719 instead of 0.768 in the transplant recipients sample and 0.806 instead of 0.853 in the native CKD sample. In addition, the residual error with a 1/cystatin C model violated the homoscedasticity assumption for linear regression. One might expect a reciprocal relationship between cystatin C and GFR based on clearance physiology: GFR ¼ (cystatin C production rate/serum cystatin C level)non-renal clearance of cystatin C.12 However, our data and those by Larsson et al.3 found the relationship between GFR and cystatin C to be stronger than a reciprocal relationship with an exponential coefficient that was more negative than 1. This suggests that non-GFR factors (production rate or non-renal clearance rate) are not independent of GFR with respect to their effects on serum cystatin C levels, an assumption implicit with 1/cystatin C models.12 This lack of independence between the non-GFR and GFR factors influencing a serum analyte is also a limitation for serum creatinine equations.13,14 1.
2.
3.
4.
5.
6.
7.
8.
9.
10. 11. 12.
13.
14.
Po¨ge U, Gerhardt T, Woitas RP. Calculation of glomerular filtration rate using serum cystatin C in kidney transplant recipients. Kidney Int 2006; 70: 1878. Rule AD, Bergstralh EJ, Slezak JM et al. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: 399–405. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate expressed in ml/min from plasma cystatin C values in mg/l. Scand J Clin Lab Invest 2004; 64: 25–30. Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003; 18: 2024–2031. Hermida J, Romero R, Tutor JC. Relationship between serum cystatin C and creatinine in kidney and liver transplant patients. Clin Chim Acta 2002; 316: 165–170. Bokenkamp A, Domanetzki M, Zinck R et al. Cystatin C serum concentrations underestimate glomerular filtration rate in renal transplant recipients. Clin Chem 1999; 45: 1866–1868. Knight EL, Verhave JC, Spiegelman D et al. Factors influencing serum cystatin C levels other than renal function and the impact on renal function measurement. Kidney Int 2004; 65: 1416–1421. Bjarnadottir M, Grubb A, Olafsson I. Promoter-mediated, dexamethasoneinduced increase in cystatin C production by HeLa cells. Scand J Clin Lab Invest 1995; 55: 617–623. Risch L, Saely C, Reist U et al. Course of glomerular filtration rate markers in patients receiving high-dose glucocorticoids following subarachnoidal hemorrhage. Clin Chimi Acta 2005; 360: 205–207. Berghout A, Wulkan RW, den Hollander JG et al. Cystatin C and the risk of death. N Engl J Med 2005; 353: 842–844. Bokenkamp A. Kidney function itself, and not cystatin C, is correlated with height and weight. Kidney Int 2005; 67: 777–778 (author reply 778–779). Sjostrom P, Tidman M, Jones I. Determination of the production rate and non-renal clearance of cystatin C and estimation of the glomerular filtration rate from the serum concentration of cystatin C in humans [see comment]. Scand J Clin Lab Invest 2005; 65: 111–124. Rule AD, Torres VE, Chapman AB et al. Comparison of methods for determining renal function decline in early autosomal dominant polycystic kidney disease: the consortium of radiologic imaging studies of polycystic kidney disease cohort. J Am Soc Nephrol 2006; 17: 854–862. Rule AD, Larson TS, Bergstralh EJ et al. Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med 2004; 141: 929–937.
AD Rule1 and TS Larson1 1 Divisions of Nephrology, Mayo Clinic, Rochester, Minnesota, USA. E-mail: [email protected]
1879
Kobayashi and Noiri do not argue against the validity of our reported significant correlations of TT and UFR with mortality risk, but suggest that our report may offer an underestimate of the true magnitude of potential benefits from longer TT and slower UFR. 1. Okamoto K, Kobayashi S, Noiri E. Longer treatment time and slower ultrafiltration in hemodialysis: associations with mortality in the Dialysis Outcomes and Practice Patterns Study. Kidney Int (in press). 2. Saran R, Bragg-Gresham JL, Port FK et al. Longer treatment time and slower ultrafiltration in hemodialysis: associations with reduced mortality in the DOPPS. Kidney Int 2006; 69: 1222–1228. 3. Saran R, Bragg-Gresham JL, Rayner HC et al. Nonadherence in hemodialysis: associations with mortality, hospitalization, and practice patterns in the DOPPS. Kidney Int 2003; 64: 254–262.
differences between the different laboratories may counteract the putative advantages of the new TX formula, (2) this rather disappointing performance of the TX equation may also be due to possible confounders like steroid dosing which may crucially affect Cys C levels.4 To enhance the performance of future Cys C-based glomerular filtration rate equations such cofactors should be taken into account. 1.
2.
R Saran1, JL Bragg-Gresham2, FK Port2 and B Gillespie1
3.
1
Division of Nephrology, Kidney Epidemiology and Cost Center, University of Michigan, Ann Arbor, Michigan, USA and 2Arbor Research Collaborative for Health (Formerly URREA), Ann Arbor, Michigan, USA Correspondence: R Saran, Division of Nephrology, Kidney Epidemiology and Cost Center, University of Michigan, 315 W. Huron, Suite 240, Ann Arbor, Michigan 48103-4262, USA. E-mail: [email protected]
4.
Rule AD, Bergstralh EJ, Slezak JM et al. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: 399–405. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate expressed in ml/min from plasma cystatin C values in mg/l. Scand J Clin Lab Invest 2004; 64: 25–30. Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003; 18: 2024–2031. Poge U, Gerhardt TM, Stoffel-Wagner B et al. {beta} Trace protein is an alternative marker for glomerular filtration rate in renal transplantation patients. Clin Chem 2005; 51: 1531–1533.
U Po¨ge1, T Gerhardt1 and RP Woitas1 1
Department of Internal Medicine I, University of Bonn, Bonn, Germany Correspondence: U Po¨ge, Department of Internal Medicine I, University of Bonn, Bonn D-53179, Germany. E-mail: [email protected]
Calculation of glomerular filtration rate using serum cystatin Response to ‘Calculation of C in kidney transplant recipients glomerular filtration rate using Kidney International (2006) 70, 1878. doi:10.1038/sj.ki.5001843 1 To the Editor: Recently, Rule et al. demonstrated a 19% serum cystatin C in kidney higher glomerular filtration rate at the same cystatin C (Cys transplant recipients’ C) level among patients after renal transplantation in comparison to patients with native kidney disease. Thus, a new Cys C-based formula (glomerular filtration rate76.6 Cys C1.16) was suggested for transplant recipients (TX formula). We analyzed the diagnostic performance of the new TX formula in comparison to two other Cys C formulae (Larsson and Hoek2,3) which are based on the same Cys C assay in a cohort of 108 patients after renal transplantation. Glomerular filtration rate was determined by 99mtechnetiumlabeled diethylenetriamine penta acetate clearance. Results are given in Table 1. Although the Larsson and Hoek formulae were not derived from a transplanted cohort, their diagnostic performances are at least comparable to the TX equation. Thus, two conclusions can be drawn from this analysis: (1) calibration
Kidney International (2006) 70, 1878–1879. doi:10.1038/sj.ki.5001828
We appreciate the work by Po¨ge et al.1 to test the performance of our transplant equation.2 Remarkably, the equation performed well with little bias (1.6 ml/min/ 1.73 m2) in their transplant recipients. There was also little bias with the Larsson3 and Hoek4 equations, which were not specifically developed using transplant recipients. However, we note that our finding of a higher glomerular filtration rate (GFR) in transplant recipients (kidney or other organ) compared to native chronic kidney disease (CKD) patients is consistent with reports by other investigators.5,6 In these centers, one equation cannot accurately estimate GFR in both transplant and native CKD patients unless it includes variables for transplant
Table 1 | Comparison of performance of the different cystatin C based formulae Range Mean estimates (ml/min/1.73 m2) (ml/min/1.73 m2) DTPA Larsson Hoek Rule
39.5 36.3 38.9 37.9
Correlation coefficient
11.8–82.9 7.78–104 8.72–97.4 9.30–101
0.859 0.865 0.862
Median Bias difference (ml/min/1.73 m2) (ml/min/1.73 m2) 3.20 0.58 1.60
4.78 1.50 2.78
Precision (ml/min/ 1.73 m2) 9.59 8.64 9.15
Accuracy within 30% (95% CI) 50% (95% CI) 77.1 77.1 78.0
95.4 97.2 89.0
CI, confidence interval; DTPA, diethylenetriamine penta acetate.
1878
Kidney International (2006) 70, 1877–1883
letter to the editor
Transplant
Native CKD
Hoek
Larsson
GFR (ml/min/1.73 m2)
150 100 60 30 15 10 0.5
0.7
1.0
1.5 2.0 Cystatin C (mg/l)
3.0
4.0 5.0
Figure 1 | Comparison of the transplant (gray solid) and native CKD (black solid) equations with the Hoek (black dashed) and Larsson (gray dashed) equations. Equations are linear or nearly linear on a logarithmic cystatin C (x axis) and logarithmic GFR (y axis) scale.
status or variables that better model the non-GFR factors that affect cystatin C levels. Equations are themselves a statistic of the average GFR at each cystatin C level for a sampled population. As shown in Figure 1, the bias between equations varies depending on the cystatin C level. If one corrected for this bias between equations, there would be no difference in the statistics that reflect model error (R2, root mean square error, precision, or accuracy within 30 or 50%). Additional variables or further sub-population stratification would be needed to decrease model error. It is possible that markers of inflammation7 or glucocorticoid use8–10 would better model the non-GFR variability with cystatin C than transplant status. Given potential calibration differences with cystatin C and ‘gold-standard’ GFR assays, it is a stronger study design to refit equations with new coefficients when comparing different populations, different serum analytes, or different statistical models. In addition, case mix and referral patterns for direct GFR measurement likely differ among centers. To a different extent between centers, patients referred for GFR measurements may be influenced by discordance between the clinical presentation and the serum creatinine. The Larsson equation3 models unstandardized GFR (ml/ min) instead of standardized GFR (ml/min/1.73 m2); however, body surface area is a predictor of unstandardized GFR independent of cystatin C level.11 If we had modeled unstandardized GFR with our data, inclusion of a body surface area term in the model would have increased our R2 from 0.700 to 0.777 in the transplant sample and from 0.807 to 0.864 in the native CKD sample. The extent that the average body surface area in the sample used to derive the Larsson equation differs from 1.73 m2 will contribute to bias between the Larsson equation and equations that predict standardized GFR. It is also worth noting that we used different data transformations for deriving equations compared to the Kidney International (2006) 70, 1877–1883
Hoek equation.4 We regressed ln GFR on ln cystatin C, instead of GFR on 1/cystatin C. If we had regressed GFR on 1/cystatin C with our data, the R2 would have been 0.719 instead of 0.768 in the transplant recipients sample and 0.806 instead of 0.853 in the native CKD sample. In addition, the residual error with a 1/cystatin C model violated the homoscedasticity assumption for linear regression. One might expect a reciprocal relationship between cystatin C and GFR based on clearance physiology: GFR ¼ (cystatin C production rate/serum cystatin C level)non-renal clearance of cystatin C.12 However, our data and those by Larsson et al.3 found the relationship between GFR and cystatin C to be stronger than a reciprocal relationship with an exponential coefficient that was more negative than 1. This suggests that non-GFR factors (production rate or non-renal clearance rate) are not independent of GFR with respect to their effects on serum cystatin C levels, an assumption implicit with 1/cystatin C models.12 This lack of independence between the non-GFR and GFR factors influencing a serum analyte is also a limitation for serum creatinine equations.13,14 1.
2.
3.
4.
5.
6.
7.
8.
9.
10. 11. 12.
13.
14.
Po¨ge U, Gerhardt T, Woitas RP. Calculation of glomerular filtration rate using serum cystatin C in kidney transplant recipients. Kidney Int 2006; 70: 1878. Rule AD, Bergstralh EJ, Slezak JM et al. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: 399–405. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate expressed in ml/min from plasma cystatin C values in mg/l. Scand J Clin Lab Invest 2004; 64: 25–30. Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003; 18: 2024–2031. Hermida J, Romero R, Tutor JC. Relationship between serum cystatin C and creatinine in kidney and liver transplant patients. Clin Chim Acta 2002; 316: 165–170. Bokenkamp A, Domanetzki M, Zinck R et al. Cystatin C serum concentrations underestimate glomerular filtration rate in renal transplant recipients. Clin Chem 1999; 45: 1866–1868. Knight EL, Verhave JC, Spiegelman D et al. Factors influencing serum cystatin C levels other than renal function and the impact on renal function measurement. Kidney Int 2004; 65: 1416–1421. Bjarnadottir M, Grubb A, Olafsson I. Promoter-mediated, dexamethasoneinduced increase in cystatin C production by HeLa cells. Scand J Clin Lab Invest 1995; 55: 617–623. Risch L, Saely C, Reist U et al. Course of glomerular filtration rate markers in patients receiving high-dose glucocorticoids following subarachnoidal hemorrhage. Clin Chimi Acta 2005; 360: 205–207. Berghout A, Wulkan RW, den Hollander JG et al. Cystatin C and the risk of death. N Engl J Med 2005; 353: 842–844. Bokenkamp A. Kidney function itself, and not cystatin C, is correlated with height and weight. Kidney Int 2005; 67: 777–778 (author reply 778–779). Sjostrom P, Tidman M, Jones I. Determination of the production rate and non-renal clearance of cystatin C and estimation of the glomerular filtration rate from the serum concentration of cystatin C in humans [see comment]. Scand J Clin Lab Invest 2005; 65: 111–124. Rule AD, Torres VE, Chapman AB et al. Comparison of methods for determining renal function decline in early autosomal dominant polycystic kidney disease: the consortium of radiologic imaging studies of polycystic kidney disease cohort. J Am Soc Nephrol 2006; 17: 854–862. Rule AD, Larson TS, Bergstralh EJ et al. Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med 2004; 141: 929–937.
AD Rule1 and TS Larson1 1 Divisions of Nephrology, Mayo Clinic, Rochester, Minnesota, USA. E-mail: [email protected]
1879