- Email: [email protected]
Comparison of Creatinine Based and Kidney Volume Based Methods of Estimating Glomerular Filtration Rates in Potential Living Kidney Donors
Transplantation/Vascular Surgery
Comparison of Creatinine Based and Kidney Volume Based Methods of Estimating Glomerular Filtration Rates in Potential Living Kidney Donors Yen Seow Benjamin Goh, Mei Wen Fiona Wu, Bee Choo Tai, King Chien Joe Lee, Lata Raman, Boon Wee Teo, Anatharaman Vathsala and Ho Yee Tiong* From the Department of Urology (YSBG, MWFW, KCJL, LR, HYT) and Division of Nephrology, Department of Medicine (BWT, AV), National University Health System, and Saw Swee Hock School of Public Health and Yong Loo Lin School of Medicine, National University of Singapore and National University Health System (BCT), Republic of Singapore
Abbreviations and Acronyms BMI ¼ body mass index CrCl ¼ creatinine clearance CT ¼ computerized tomography dGFR ¼ directly measured GFR eGFR ¼ creatinine based equation estimated GFR GFR ¼ glomerular filtration rate SCr ¼ serum creatinine vGFR ¼ volume based equation estimated GFR Accepted for publication May 28, 2013. Study received institutional review board approval. * Correspondence: Department of Urology, National University Health System, NUHS Tower Block Level 8, 1E Kent Ridge Rd., Republic of Singapore 119228 (telephone: þ6567725642; FAX: þ6567740881; e-mail: [email protected]).
Purpose: Accurate assessment of kidney function is critical to evaluate living kidney donors. Direct glomerular filtration rate measurement using isotopes is currently the gold standard but it is complex and costly. We evaluated the performance of surrogate markers of the glomerular filtration rate in living kidney donors by comparing direct measurement of the rate to the creatinine based equation estimated rate, the kidney volume based estimated rate using a newly developed equation and creatinine clearance. Materials and Methods: We first statistically compared direct glomerular filtration rate measurement to the results of the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) creatinine based equations, and to creatinine clearance in 54 potential renal donors from 2006 to 2010. In 32 donors with cross-sectional computerized tomography available we used measured functional renal volume with age, gender, weight and serum creatinine to estimate the rate based on kidney volume according to a previously reported model. Kidney volume based measurement was compared to direct glomerular filtration rate measurement and assessed against the results of the best performing creatinine based equation. Results: In the first group of 54 donors the correlation index of the estimated glomerular filtration rate according to MDRD and CKD-EPI creatinine based equations, and to creatinine clearance was low compared to direct measurement. In the subset of 32 potential donors the kidney volume based estimated rate correlated better with direct measurement than MDRD equation results with higher accuracy (estimated 87.5% and 75.0% within 30% and 10% of direct rate measurement, respectively). Conclusions: To estimate the glomerular filtration rate in healthy individuals a volume based model correlated better than the MDRD equation, which is the best performing creatinine based equation used to estimate the rate. By providing a more robust estimation of the glomerular filtration rate in healthy potential kidney donors, the volume based model adds value to routine preoperative computerized tomography above that of anatomical evaluation. Key Words: kidney, living donors, kidney function tests, creatinine, organ size
1820
j
www.jurology.com
0022-5347/13/1905-1820/0 THE JOURNAL OF UROLOGY® © 2013 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2013.05.117 Vol. 190, 1820-1826, November 2013 Printed in U.S.A.
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
THE current optimal treatment of end stage renal failure is living kidney transplantation, which is performed as frequently as cadaveric donor kidney transplantation.1 For successful living kidney transplants accurate renal function determination, ideally in a reproducible, cost-effective manner, is critical during potential donor evaluation.2 The accepted reference standard of measuring GFR is direct measurement using radiotracer labeled compounds, which is costly and not readily available.3 Alternatives using creatinine based equations were developed as surrogate determinants of measured GFR. Most of these formulas, including the MDRD and CKD-EPI equations, were derived from patients with impaired renal function and excluded individuals with a GFR of greater than 70 ml/minute/1.73 m2. Although it is currently advocated by the Amsterdam Forum on living donors and widely used to assess donors,2 concern arises regarding the performance of these equations in healthy individuals with limited attempts at validation in kidney donor populations.4 Reports that the volume of a healthy kidney is related to body parameters led Jeon et al to directly correlate the kidney volume of donors measured on contrast enhanced CT with nephrectomized kidney weight and various kidney function measures.5 This correlation of CT measured kidney volume with renal function was incorporated by Herts et al into a novel mathematical model to estimate GFR in potential kidney donors.6 Since cross-sectional CT is the standard practice for evaluating potential kidney donors, this vGFR is potentially convenient and cost-effective. Using renal volume in addition to the standard parameters of age, weight and SCr, the volume based model outperformed the MDRD equation when forward tested in 100 white donors.6 We determined the optimal method of estimating GFR in potential kidney donors at our institution. This was done in 2 parts. 1) We compared the performance of the MDRD and CKD-EPI creatinine based equations, and 24-hour CrCl to dGFR, which is the current reference standard. 2) In a subset of these individuals who donated a kidney and had CT available, we compared the performance of the volume based model and the best performing creatinine based equation.
PATIENTS AND METHODS Study Population This retrospective, institutional review board approved study included 79 consecutive healthy potential renal donors who were reviewed for potential living related kidney donation between June 2006 and September 2010. In 54 potential donors in whom borderline GFR was determined by creatinine based equations we selectively
1821
measured dGFR by 99mTc-diethylenetriaminepentaacetic acid scintigraphy. Baseline demographics and clinical characteristics of the potential donors were retrieved from the transplantation database. All 54 potential kidney donors in whom GFR was measured directly were recruited for study to compare the performance of creatinine based equations. We excluded 22 donors due to dGFR less than 80 ml/minute/1.73 m2 or to voluntary withdrawal after evaluation. In the remaining 32 donors CT was performed and the kidney was donated. These 32 donors were recruited for volume based GFR estimation. All laboratory and anthropometric measurements were made within a month of GFR direct measurement.
Performance Creatinine based equations and CrCl. To study creatinine based equations we determined eGFR in all recruited donors with the MDRD and CKD-EPI equations using the National Kidney Foundation calculator (http://www. kidney.org/professionals/kdoqi/gfr_calculator.cfm). In all donors 1 measurement was made of 24-hour CrCl. Samples were considered adequate if 24-hour urine creatinine was more than 15 and 20 mg/kg per day in females and males, respectively. Our laboratory provides standardized creatinine values traceable to the National Institute of Standards and Technology sample. We used 3 equations. 1) The MDRD equation is GFR in ml/minute/1.73 m2 ¼ 175 (SCr)e1.154 (age)e0.203 0.742 (if female) 1.212 (if black).7 2) The CKD-EPI equation is GFR in ml/minute/1.73 m2 ¼ 141 min(SCr/k,1)a max(SCr/k,1)e1.209 0.993age 1.018 (if female) 1.159 (if black), where SCr is in mg/dl, k is 0.7 for females and 0.9 for males, a is e0.329 for females and e0.411 for males, min represents the minimum of SCr/k or 1 and max indicates the maximum of SCr/k or 1.8 3) For CrCl we used the equation, CrCl ¼ (UCr V)/SCr (adjusted for body surface area using 1.73 m2), where UCr represents 24-hour urine creatinine and V represents 24-hour urine volume. All results were compared against dGFR for precision, accuracy and bias, as described. Kidney volume based equation. In 32 donors with contrast enhanced CT available we first measured kidney volume by a technique adapted from liver volumetry using National Institutes of Health ImageJ tissue segmentation software.9 CT images obtained in the arterial phase were saved in JPEG format. Using ImageJ, the functional nephron mass was manually outlined on the transverse section by a single surgeon, excluding renal sinus fat, cysts, blood vessels and the pelvicalyceal system (fig. 1). The area of outlined functional renal mass was multiplied by slice thickness to obtain renal volume. Total renal volume was calculated by summing all volumes in the measured boundaries of the 2 kidneys after exporting results into WindowsÒ ExcelÒ. Volume measurement and calculation for each kidney required an average of 30 minutes. We calculated vGFR in donors using the regression and forward tested model of Herts et al.6 The equation used was vGFR in ml/minute/1.73 m2 ¼ 70.77 e 0.444(age) þ 0.366(W) þ 0.200(V) e 37.317(SCr), where W represents
1822
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
showed the difference between dGFR and eGFR by the average of 2 measures. These plots represented the trend of errors across ranges of possible GFR values. Statistical analysis was done with STATAÒ, version 11. Statistical evaluations were assessed using the 2-sided test at the conventional 0.05 significance level.
RESULTS Results were analyzed in 2 study populations. The performance of eGFR and CrCl against dGFR was first compared in 54 potential donors. In 32 donors with measured renal volume available on CT and vGFR estimates we compared MDRD eGFR against dGFR. Performance Creatinine based equations and CrCl. Table 1 lists the
Figure 1. Renal volume functional area (yellow outline), as measured by single surgeon using ImageJ. Region of interest (ROI ) manager tabulates area of each measured axial cut.
weight in kg, V represents measured volume in mm3 and SCr is in mg/dl. Results were compared with dGFR for precision, accuracy and bias, and against the best creatinine based eGFR derived from the initial evaluation.
Statistical Analysis Demographic and clinical characteristics of the respective study populations were summarized using the frequency and percent for categorical variables, and the mean SD for continuous variables, which were normally distributed. We assessed the performance of each method in relation to the reference, dGFR, for precision, bias and accuracy. Precision was assessed using the Pearson correlation coefficient and the Lin concordance index.10 The Pearson correlation measures the linear correlation between 2 measures without specifying the degree of correspondence between the value sets. The Lin concordance measures reliability and reproducibility based on the covariation and correspondence of 2 value sets.10 While the Pearson correlation coefficient is immune to biased or unbiased versions for estimating the variance used, the Lin concordance is not. Bias, which is the measure of systematic error, was assessed by the mean difference of (eGFR ½MDRD or CKD-EPI] e dGFR) and (vGFR e dGFR). Accuracy was assessed as the percent of eGFR(MDRD or CKD-EPI) or vGFR values within 10% and 30% of dGFR. Agreement between eGFR and dGFR was evaluated using residual and Bland-Altman plots. Residual plots showed the pattern of error (difference between eGFR and dGFR) over predicted GFR values. Bland-Altman plots
baseline characteristics of 54 donors for the initial comparison of eGFR and CrCl. Mean SD donor SCr was 72.2 19.3 mmol/l and mean dGFR was 91.3 15.8 ml/minute/1.73 m2. Table 2 shows the overall precision, accuracy and bias of eGFR compared to dGFR. The MDRD and CKD-EPI equations demonstrated similar degrees of association with dGFR (Lin concordance 0.280 and 0.277, respectively). Table 2 also shows the index of bias by the mean difference and accuracy by the percent of estimates within 10% and 30% of dGFR for all GFR estimates. The MDRD equation overestimated GFR less than the CKD-EPI equation compared to dGFR (mean difference 0.9 21.7 vs 6.4 18.7 ml/minute/1.73 m2) and CrCl underestimated GFR (mean difference e3.7 35.9 ml/ minute/1.73 m2). MDRD eGFR accuracy was better
Table 1. Characteristics of potential donors overall and subgroup with CT
No. pts Mean SD age at dGFR No. female (%) No. race (%): Chinese Malay Indian Other Mean SD BMI (kg/m2) Mean SD SCr (mmol/l) Mean SD total kidney vol (ml) Mean SD GFR (ml/min/1.73 m2): dGFR MDRD eGFR CKD-EPI eGFR vGFR Mean SD CrCl (ml/min/1.73m2) No. hypertension (%) No. hyperlipidemia (%) No. diabetes mellitus (%) No. smoking (%)
Overall
Subgroup
54 44.5 9.0 31 (57.4)
32 43.1 9.7 18 (56)
32 (59.2) 11 (20.4) 9 (16.7) 2 (3.7) 22.8 3.0 72.2 19.3 e
16 (50.0) 9 (28.1) 5 (15.6) 2 (6.3) 22.3 2.7 72.1 21.4 272.0 78.8
91.3 15.8 92.3 20.1 97.7 15.8 e 87.7 33.1 11 (20.4) 10 (18.5) 2 (3.7) 13 (20.0)
96.2 17.4 94.5 17.4 e 97.5 20.8 e 6 (18.7) 7 (21.8) e 6 (18.7)
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
1823
Table 2. dGFR relation to eGFR by MDRD, CKD-EPI and CrCl in donors overall, and to MDRD eGFR and vGFR in subgroup with CT Overall eGFR MDRD Lin concordance index Pearson correlation coefficient % dGFR: Within 10% Within 30% Mean SD bias (ml/min/1.73 m2)
0.280 0.288 48.1 83.8 0.9 21.7
CKD-EPI 0.277 0.300 31.5 83.3 6.4 18.7
when calculated within 10% of dGFR compared to CKD-EPI estimates (48.1% vs 31.5%). Values were similar when calculated within 30% of dGFR. Figure 2 shows the graphic relationship of eGFR with MDRD, CKD-EPI and dGFR using BlandAltman and residual plots. Residual plots suggested that the MDRD equation tended to underestimate dGFR at lower values and overestimate it at higher values, while the CKD-EPI equation overestimated GFR at all values. Overall the MDRD estimates of GFR had the best precision and accuracy with the least bias in our donor population. Kidney volume based equation. In the subgroup of 32 patients with CT available mean total renal volume was 272.0 78.8 ml. Table 1 list the characteristics
Subgroup CrCl 0.044 0.056 26.4 61.1 e3.7 35.9
vGFR 0.50 0.50 37.5 87.5 1.4 19.1
MDRD eGFR 0.29 0.30 31.3 75.0 e1.6 23.6
of this subgroup of kidney donors, including various calculated GFRs. vGFR performed better than MDRD eGFR and each correlated with dGFR (Lin concordance 0.50 and 0.30, respectively, table 2). The index of accuracy suggested that vGFR was more accurate with 87.5% of estimates within 30% of dGFR compared to 75% for eGFR. Figure 3 shows the graphic relationship of vGFR and eGFR by MDRD with dGFR using Bland-Altman and residual plots. vGFR and eGFR by the MDRD equation tended to underestimate dGFR at lower values and overestimate it at higher values on the residual plots. Overall vGFR estimates outperformed eGFR using MDRD estimates in precision, accuracy and bias.
Figure 2. Distribution of errors when estimating dGFR for observed eGFR using MDRD and CKD-EPI equations. A and B, Bland-Altman plots of nuclear dGFR agreement. Solid line indicates overall measure. Dotted lines indicate 2 SDs. A, MDRD eGFR . B, CKD-EPI eGFR. C and D, residual plots of concordance with nuclear dGFR. C, MDRD eGFR. D, CKD-EPI eGFR.
1824
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
Figure 3. Distribution of errors when estimating dGFR for observed vGFR and eGFR by MDRD equation. A and B, Bland-Altman plots of agreement with nuclear dGFR. Solid line indicates overall measure. Dotted lines indicate 2 SDs. A, vGFR. B, MDRD eGFR. C and D, residual plots of concordance with nuclear dGFR. C, MDRD eGFR. D, CKD-EPI eGFR.
DISCUSSION Living renal transplantation is the definitive therapeutic option for renal failure, offering better outcomes than cadaveric donor kidney transplantation. Comprehensive living donor evaluation is essential to ensure the optimal recipient outcome and donor safety. Accurate assessment of renal function in donors is particularly important since higher measured baseline GFR in donors before transplantation independently predicts improved allograft outcomes in recipients, including better function and a lower risk of rejection.11 In addition, during followup the risk of postoperative CKD is greater in donors who had lower baseline GFR with overestimation of GFR.12,13 This study shows that the MDRD equation had better precision, bias and accuracy among creatinine based estimates. When kidney volume was factored into the model, GFR estimation was enhanced. The gold standard for renal function measurement requires serial blood sampling to calculate the clearance of a substance, the ideal substance being inulin. Current acceptable standards include radiolabeled compounds such as 125I-iothalamate and 99 Tc, which are costly and not readily available.3 At
our institution 99Tc urinary clearance serves as the reference standard, against which we compared eGFR and vGFR. The MDRD and CKD-EPI equations use wellknown variables in individuals, including gender, weight, age and SCr, to estimate renal function in patients with renal impairment.7,8 The MDRD equation was developed in 1,628 patients with renal impairment and SCr 1.2 to 7 mg/dl, while excluding patients with GFR greater than 70 ml/minute/ 1.73 m.2 The CKD-EPI formula was derived from direct nuclear measurements of GFR in each gender of many races in a wide range of clinical scenarios. An understanding of the sample groups of patients used for the MDRD and CKD-EPI equations indicates that correlation results derived from these equations in healthy, potential kidney donors may be suboptimal. The biological relationship between GFR and SCr is more linear in patients with CKD than in healthy individuals. Thus, linear extrapolation of GFR may not apply to healthy donors, in whom renal function exceeds 60 ml/minute/ 1.73 m2.3 While the MDRD and CKD-EPI equations demonstrated a similar concordance coefficient
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
compared to dGFR, the Lin concordance correlation index of 0.3 for each suggested a poor linear association of estimates (precision and reproducibility). The MDRD equation had the most favorable combination of better accuracy, precision and lower bias compared to the CKD-EPI equation and CrCl. The Lin concordance index was between 0.12 and 0.34 for the performance of MDRD, reexpressed MDRD, the Cockcroft-Gault formula and CrCl estimates against 125I-iothalamate GFR in 423 kidney donors.14 Low concordance ratios were also documented for the MDRD, Cockcroft-Gault and Mayo Clinic formulas (0.50, 0.36 and 0.46, respectively) in a randomly selected population of 112 kidney donors.4 In that study the MDRD equation outperformed other creatinine based estimates of GFR, consistent with our findings. While the MDRD equation is suboptimal, current reports support it as the best available equation to estimate renal function in potential donors. CrCl consistently performed most poorly in all studies compared to creatinine based equations.4,14 Because timed urinary collections are cumbersome and prone to error, 24-hour urine collections are not routinely done to estimate renal function.3 Equation estimates of GFR in healthy donors are often lower than directly measured values. The mean difference between GFR estimated from the MDRD equation and direct GFR measurement ranges from e29 to 3.3 ml/minute/1.73 m2.3 For Asian donors at our institution the equations tend to overestimate GFR but with a minimal mean difference using the MDRD formula (e1.6 to 0.9 ml/ minute/1.73 m2). Recent studies identified preoperative kidney volume as an important predictor of graft function in living donor kidney transplantation.15,16 Larger donor renal volume correlated with lower nadir SCr at 1 year in transplant recipients.17 Preoperative kidney volume also independently predicted delayed recovery of kidney function in donors after nephrectomy.5 Beyond transplantation, renal parenchymal volume measurements were used to accurately estimate differential renal function in normal individuals and those with urinary obstruction.18 Kidney volume is inherently related to baseline donor kidney function, as demonstrated in correlation studies using ultrasound measured kidney size.19 Volume based GFR measurements may also serve to predict residual renal function after partial or radical nephrectomy. This is of particular importance in patients with borderline GFR and may aid in counseling on postoperative renal replacement therapy.
1825
We validated that the volume based equation outperformed MDRD eGFR in our Asian population using statistical assessment. Incorporating kidney volume enhanced GFR estimation in healthy donors. In the absence of comorbid disease, including diabetes mellitus and significant hypertension, in young healthy donors the kidneys show less glomerulosclerosis and scarring. Thus, kidney parenchymal volume expectedly correlates with healthy nephron mass and glomerular function correlates with accuracy. The anatomical studies by Nyengaard and Bendtsen revealed a direct linear relationship between kidney weight and total glomerular number using stereological techniques in autopsy specimens from subjects with no kidney disease.20 Contrast enhanced spiral CT in potential donors is the current recommended standard imaging modality before donor nephrectomy, providing the advantage of using vGFR.2 Besides delineating anatomy to aid surgery, GFR can be estimated with an improved correlation at no additional cost or procedure. After validation with larger studies and in different donor populations, vGFR may potentially be the best readily available method of estimating GFR in healthy populations. There were several limitations of our study, including relatively few donors. The patient population may not be readily generalized to other population groups since we included predominantly Asian donors with a normal mean BMI and slight variations (the ideal BMI in Asian population is 18.5 to 23.5 kg/m2). Functional kidney parenchymal volume is relatively lower than that derived from white donors but similar to kidney weight in published Asian series.5 While the volume based equation was derived and forward tested in a white population, the reproducibility of performance in an Asian population may be related to renal function per unit kidney volume, which is independent of race or BMI in a healthy population.
CONCLUSIONS In our healthy Asian population of potential living related renal donors MDRD eGFR had the best concordance, accuracy and bias compared to dGFR. The CT measured, kidney volume based equation of Herts et al6 showed better precision with less bias and improved concordance with dGFR than MDRD eGFR. vGFR could be an initial, complementary screening modality to verify eGFR, especially in healthy populations, which are different than those from whom creatinine based equations of GFR were derived.
1826
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
REFERENCE 1. Port FK, Merion RM, Wolfe RA et al: Trends in organ donation and transplantation in the United States, 1997-2006. Am J Transplant 2008; 8: 911. 2. Delmonico F: A Report of the Amsterdam Forum on the care of the live kidney donor: data and medical guidelines. Transplantation, suppl., 2005; 79: S53. 3. Stevens LA, Coresh J, Greene T et al: Assessing kidney function-measured and estimated glomerular filtration rate. N Engl J Med 2006; 354: 2473. 4. Ibrahim HN, Rogers T, Tello A et al: The performance of three serum creatinine-based formulas in estimating GFR in former kidney donors. Am J Transplant 2006; 6: 1479. 5. Jeon HG, Lee SR, Joo DJ et al: Predictors of kidney volume change and delayed kidney function recovery after donor nephrectomy. J Urol 2010; 184: 1057.
filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: 461. 8. Levey AS, Stevens LA, Schmid CH et al: A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: 604. 9. Dello SAWG, van Dam RM, Slangen JJG et al: Liver volumetry plug and play: do it yourself with ImageJ. World J Surg 2007; 31: 2215. 10. Lin LI: A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989; 45: 255. 11. Issa N, Stephany B, Fatica R et al: Donor factors influencing graft outcomes in live donor kidney transplantation. Transplantation 2007; 83: 593. 12. Tan L, Tai BC, Wu F et al: Impact of kidney disease outcomes quality initiative guidelines on the prevalence of chronic kidney disease after living donor nephrectomy. J Urol 2011; 185: 1820.
6. Herts BR, Sharma N, Lieber M et al: Estimating glomerular filtration rate in kidney donors: a model constructed with renal volume measurements from donor CT scans. Radiology 2009; 252: 109.
13. Ibrahim HN, Foley R, Tan LP et al: Long-term consequences of kidney donation. N Engl J Med 2009; 360: 459.
7. Levey AS, Bosch JP, Lewis JB et al: A more accurate method to estimate glomerular
14. Issa N, Meyer KH, Choure G et al: Evaluation of creatinine-based estimates of glomerular
filtration rate in a large cohort of living kidney donors. Transplantation 2008; 86: 223. 15. Saxena AB, Busque S, Arjane P et al: Preoperative renal volumes as a predictor of graft function in living donor transplantation. Am J Kidney Dis 2004; 44: 877. 16. Poggio ED, Hila S, Stephany B et al: Donor kidney volume and outcomes following live donor kidney transplantation. Am J Transplant 2006; 6: 616. 17. Hugen CM, Polcari AJ, Farooq AV et al: Size does matter: donor renal volume predicts recipient function following live donor renal transplantation. J Urol 2011; 185: 605. 18. Morrisroe SN, Su RR, Bae KT et al: Differential renal function estimation using computerized tomography based renal parenchymal volume measurement. J Urol 2010; 183: 2289. 19. Barai S, Gambhir S, Prasad N et al: Levels of GFR and protein-induced hyperfiltration in kidney donors: a single-center experience in India. Am J Kidney Dis 2008; 51: 407. 20. Nyengaard JR and Bendtsen TF: Glomerular number and size in relation to age, kidney weight, and body surface in normal man. Anat Rec 1992; 232: 194.
Comparison of Creatinine Based and Kidney Volume Based Methods of Estimating Glomerular Filtration Rates in Potential Living Kidney Donors Yen Seow Benjamin Goh, Mei Wen Fiona Wu, Bee Choo Tai, King Chien Joe Lee, Lata Raman, Boon Wee Teo, Anatharaman Vathsala and Ho Yee Tiong* From the Department of Urology (YSBG, MWFW, KCJL, LR, HYT) and Division of Nephrology, Department of Medicine (BWT, AV), National University Health System, and Saw Swee Hock School of Public Health and Yong Loo Lin School of Medicine, National University of Singapore and National University Health System (BCT), Republic of Singapore
Abbreviations and Acronyms BMI ¼ body mass index CrCl ¼ creatinine clearance CT ¼ computerized tomography dGFR ¼ directly measured GFR eGFR ¼ creatinine based equation estimated GFR GFR ¼ glomerular filtration rate SCr ¼ serum creatinine vGFR ¼ volume based equation estimated GFR Accepted for publication May 28, 2013. Study received institutional review board approval. * Correspondence: Department of Urology, National University Health System, NUHS Tower Block Level 8, 1E Kent Ridge Rd., Republic of Singapore 119228 (telephone: þ6567725642; FAX: þ6567740881; e-mail: [email protected]).
Purpose: Accurate assessment of kidney function is critical to evaluate living kidney donors. Direct glomerular filtration rate measurement using isotopes is currently the gold standard but it is complex and costly. We evaluated the performance of surrogate markers of the glomerular filtration rate in living kidney donors by comparing direct measurement of the rate to the creatinine based equation estimated rate, the kidney volume based estimated rate using a newly developed equation and creatinine clearance. Materials and Methods: We first statistically compared direct glomerular filtration rate measurement to the results of the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) creatinine based equations, and to creatinine clearance in 54 potential renal donors from 2006 to 2010. In 32 donors with cross-sectional computerized tomography available we used measured functional renal volume with age, gender, weight and serum creatinine to estimate the rate based on kidney volume according to a previously reported model. Kidney volume based measurement was compared to direct glomerular filtration rate measurement and assessed against the results of the best performing creatinine based equation. Results: In the first group of 54 donors the correlation index of the estimated glomerular filtration rate according to MDRD and CKD-EPI creatinine based equations, and to creatinine clearance was low compared to direct measurement. In the subset of 32 potential donors the kidney volume based estimated rate correlated better with direct measurement than MDRD equation results with higher accuracy (estimated 87.5% and 75.0% within 30% and 10% of direct rate measurement, respectively). Conclusions: To estimate the glomerular filtration rate in healthy individuals a volume based model correlated better than the MDRD equation, which is the best performing creatinine based equation used to estimate the rate. By providing a more robust estimation of the glomerular filtration rate in healthy potential kidney donors, the volume based model adds value to routine preoperative computerized tomography above that of anatomical evaluation. Key Words: kidney, living donors, kidney function tests, creatinine, organ size
1820
j
www.jurology.com
0022-5347/13/1905-1820/0 THE JOURNAL OF UROLOGY® © 2013 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2013.05.117 Vol. 190, 1820-1826, November 2013 Printed in U.S.A.
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
THE current optimal treatment of end stage renal failure is living kidney transplantation, which is performed as frequently as cadaveric donor kidney transplantation.1 For successful living kidney transplants accurate renal function determination, ideally in a reproducible, cost-effective manner, is critical during potential donor evaluation.2 The accepted reference standard of measuring GFR is direct measurement using radiotracer labeled compounds, which is costly and not readily available.3 Alternatives using creatinine based equations were developed as surrogate determinants of measured GFR. Most of these formulas, including the MDRD and CKD-EPI equations, were derived from patients with impaired renal function and excluded individuals with a GFR of greater than 70 ml/minute/1.73 m2. Although it is currently advocated by the Amsterdam Forum on living donors and widely used to assess donors,2 concern arises regarding the performance of these equations in healthy individuals with limited attempts at validation in kidney donor populations.4 Reports that the volume of a healthy kidney is related to body parameters led Jeon et al to directly correlate the kidney volume of donors measured on contrast enhanced CT with nephrectomized kidney weight and various kidney function measures.5 This correlation of CT measured kidney volume with renal function was incorporated by Herts et al into a novel mathematical model to estimate GFR in potential kidney donors.6 Since cross-sectional CT is the standard practice for evaluating potential kidney donors, this vGFR is potentially convenient and cost-effective. Using renal volume in addition to the standard parameters of age, weight and SCr, the volume based model outperformed the MDRD equation when forward tested in 100 white donors.6 We determined the optimal method of estimating GFR in potential kidney donors at our institution. This was done in 2 parts. 1) We compared the performance of the MDRD and CKD-EPI creatinine based equations, and 24-hour CrCl to dGFR, which is the current reference standard. 2) In a subset of these individuals who donated a kidney and had CT available, we compared the performance of the volume based model and the best performing creatinine based equation.
PATIENTS AND METHODS Study Population This retrospective, institutional review board approved study included 79 consecutive healthy potential renal donors who were reviewed for potential living related kidney donation between June 2006 and September 2010. In 54 potential donors in whom borderline GFR was determined by creatinine based equations we selectively
1821
measured dGFR by 99mTc-diethylenetriaminepentaacetic acid scintigraphy. Baseline demographics and clinical characteristics of the potential donors were retrieved from the transplantation database. All 54 potential kidney donors in whom GFR was measured directly were recruited for study to compare the performance of creatinine based equations. We excluded 22 donors due to dGFR less than 80 ml/minute/1.73 m2 or to voluntary withdrawal after evaluation. In the remaining 32 donors CT was performed and the kidney was donated. These 32 donors were recruited for volume based GFR estimation. All laboratory and anthropometric measurements were made within a month of GFR direct measurement.
Performance Creatinine based equations and CrCl. To study creatinine based equations we determined eGFR in all recruited donors with the MDRD and CKD-EPI equations using the National Kidney Foundation calculator (http://www. kidney.org/professionals/kdoqi/gfr_calculator.cfm). In all donors 1 measurement was made of 24-hour CrCl. Samples were considered adequate if 24-hour urine creatinine was more than 15 and 20 mg/kg per day in females and males, respectively. Our laboratory provides standardized creatinine values traceable to the National Institute of Standards and Technology sample. We used 3 equations. 1) The MDRD equation is GFR in ml/minute/1.73 m2 ¼ 175 (SCr)e1.154 (age)e0.203 0.742 (if female) 1.212 (if black).7 2) The CKD-EPI equation is GFR in ml/minute/1.73 m2 ¼ 141 min(SCr/k,1)a max(SCr/k,1)e1.209 0.993age 1.018 (if female) 1.159 (if black), where SCr is in mg/dl, k is 0.7 for females and 0.9 for males, a is e0.329 for females and e0.411 for males, min represents the minimum of SCr/k or 1 and max indicates the maximum of SCr/k or 1.8 3) For CrCl we used the equation, CrCl ¼ (UCr V)/SCr (adjusted for body surface area using 1.73 m2), where UCr represents 24-hour urine creatinine and V represents 24-hour urine volume. All results were compared against dGFR for precision, accuracy and bias, as described. Kidney volume based equation. In 32 donors with contrast enhanced CT available we first measured kidney volume by a technique adapted from liver volumetry using National Institutes of Health ImageJ tissue segmentation software.9 CT images obtained in the arterial phase were saved in JPEG format. Using ImageJ, the functional nephron mass was manually outlined on the transverse section by a single surgeon, excluding renal sinus fat, cysts, blood vessels and the pelvicalyceal system (fig. 1). The area of outlined functional renal mass was multiplied by slice thickness to obtain renal volume. Total renal volume was calculated by summing all volumes in the measured boundaries of the 2 kidneys after exporting results into WindowsÒ ExcelÒ. Volume measurement and calculation for each kidney required an average of 30 minutes. We calculated vGFR in donors using the regression and forward tested model of Herts et al.6 The equation used was vGFR in ml/minute/1.73 m2 ¼ 70.77 e 0.444(age) þ 0.366(W) þ 0.200(V) e 37.317(SCr), where W represents
1822
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
showed the difference between dGFR and eGFR by the average of 2 measures. These plots represented the trend of errors across ranges of possible GFR values. Statistical analysis was done with STATAÒ, version 11. Statistical evaluations were assessed using the 2-sided test at the conventional 0.05 significance level.
RESULTS Results were analyzed in 2 study populations. The performance of eGFR and CrCl against dGFR was first compared in 54 potential donors. In 32 donors with measured renal volume available on CT and vGFR estimates we compared MDRD eGFR against dGFR. Performance Creatinine based equations and CrCl. Table 1 lists the
Figure 1. Renal volume functional area (yellow outline), as measured by single surgeon using ImageJ. Region of interest (ROI ) manager tabulates area of each measured axial cut.
weight in kg, V represents measured volume in mm3 and SCr is in mg/dl. Results were compared with dGFR for precision, accuracy and bias, and against the best creatinine based eGFR derived from the initial evaluation.
Statistical Analysis Demographic and clinical characteristics of the respective study populations were summarized using the frequency and percent for categorical variables, and the mean SD for continuous variables, which were normally distributed. We assessed the performance of each method in relation to the reference, dGFR, for precision, bias and accuracy. Precision was assessed using the Pearson correlation coefficient and the Lin concordance index.10 The Pearson correlation measures the linear correlation between 2 measures without specifying the degree of correspondence between the value sets. The Lin concordance measures reliability and reproducibility based on the covariation and correspondence of 2 value sets.10 While the Pearson correlation coefficient is immune to biased or unbiased versions for estimating the variance used, the Lin concordance is not. Bias, which is the measure of systematic error, was assessed by the mean difference of (eGFR ½MDRD or CKD-EPI] e dGFR) and (vGFR e dGFR). Accuracy was assessed as the percent of eGFR(MDRD or CKD-EPI) or vGFR values within 10% and 30% of dGFR. Agreement between eGFR and dGFR was evaluated using residual and Bland-Altman plots. Residual plots showed the pattern of error (difference between eGFR and dGFR) over predicted GFR values. Bland-Altman plots
baseline characteristics of 54 donors for the initial comparison of eGFR and CrCl. Mean SD donor SCr was 72.2 19.3 mmol/l and mean dGFR was 91.3 15.8 ml/minute/1.73 m2. Table 2 shows the overall precision, accuracy and bias of eGFR compared to dGFR. The MDRD and CKD-EPI equations demonstrated similar degrees of association with dGFR (Lin concordance 0.280 and 0.277, respectively). Table 2 also shows the index of bias by the mean difference and accuracy by the percent of estimates within 10% and 30% of dGFR for all GFR estimates. The MDRD equation overestimated GFR less than the CKD-EPI equation compared to dGFR (mean difference 0.9 21.7 vs 6.4 18.7 ml/minute/1.73 m2) and CrCl underestimated GFR (mean difference e3.7 35.9 ml/ minute/1.73 m2). MDRD eGFR accuracy was better
Table 1. Characteristics of potential donors overall and subgroup with CT
No. pts Mean SD age at dGFR No. female (%) No. race (%): Chinese Malay Indian Other Mean SD BMI (kg/m2) Mean SD SCr (mmol/l) Mean SD total kidney vol (ml) Mean SD GFR (ml/min/1.73 m2): dGFR MDRD eGFR CKD-EPI eGFR vGFR Mean SD CrCl (ml/min/1.73m2) No. hypertension (%) No. hyperlipidemia (%) No. diabetes mellitus (%) No. smoking (%)
Overall
Subgroup
54 44.5 9.0 31 (57.4)
32 43.1 9.7 18 (56)
32 (59.2) 11 (20.4) 9 (16.7) 2 (3.7) 22.8 3.0 72.2 19.3 e
16 (50.0) 9 (28.1) 5 (15.6) 2 (6.3) 22.3 2.7 72.1 21.4 272.0 78.8
91.3 15.8 92.3 20.1 97.7 15.8 e 87.7 33.1 11 (20.4) 10 (18.5) 2 (3.7) 13 (20.0)
96.2 17.4 94.5 17.4 e 97.5 20.8 e 6 (18.7) 7 (21.8) e 6 (18.7)
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
1823
Table 2. dGFR relation to eGFR by MDRD, CKD-EPI and CrCl in donors overall, and to MDRD eGFR and vGFR in subgroup with CT Overall eGFR MDRD Lin concordance index Pearson correlation coefficient % dGFR: Within 10% Within 30% Mean SD bias (ml/min/1.73 m2)
0.280 0.288 48.1 83.8 0.9 21.7
CKD-EPI 0.277 0.300 31.5 83.3 6.4 18.7
when calculated within 10% of dGFR compared to CKD-EPI estimates (48.1% vs 31.5%). Values were similar when calculated within 30% of dGFR. Figure 2 shows the graphic relationship of eGFR with MDRD, CKD-EPI and dGFR using BlandAltman and residual plots. Residual plots suggested that the MDRD equation tended to underestimate dGFR at lower values and overestimate it at higher values, while the CKD-EPI equation overestimated GFR at all values. Overall the MDRD estimates of GFR had the best precision and accuracy with the least bias in our donor population. Kidney volume based equation. In the subgroup of 32 patients with CT available mean total renal volume was 272.0 78.8 ml. Table 1 list the characteristics
Subgroup CrCl 0.044 0.056 26.4 61.1 e3.7 35.9
vGFR 0.50 0.50 37.5 87.5 1.4 19.1
MDRD eGFR 0.29 0.30 31.3 75.0 e1.6 23.6
of this subgroup of kidney donors, including various calculated GFRs. vGFR performed better than MDRD eGFR and each correlated with dGFR (Lin concordance 0.50 and 0.30, respectively, table 2). The index of accuracy suggested that vGFR was more accurate with 87.5% of estimates within 30% of dGFR compared to 75% for eGFR. Figure 3 shows the graphic relationship of vGFR and eGFR by MDRD with dGFR using Bland-Altman and residual plots. vGFR and eGFR by the MDRD equation tended to underestimate dGFR at lower values and overestimate it at higher values on the residual plots. Overall vGFR estimates outperformed eGFR using MDRD estimates in precision, accuracy and bias.
Figure 2. Distribution of errors when estimating dGFR for observed eGFR using MDRD and CKD-EPI equations. A and B, Bland-Altman plots of nuclear dGFR agreement. Solid line indicates overall measure. Dotted lines indicate 2 SDs. A, MDRD eGFR . B, CKD-EPI eGFR. C and D, residual plots of concordance with nuclear dGFR. C, MDRD eGFR. D, CKD-EPI eGFR.
1824
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
Figure 3. Distribution of errors when estimating dGFR for observed vGFR and eGFR by MDRD equation. A and B, Bland-Altman plots of agreement with nuclear dGFR. Solid line indicates overall measure. Dotted lines indicate 2 SDs. A, vGFR. B, MDRD eGFR. C and D, residual plots of concordance with nuclear dGFR. C, MDRD eGFR. D, CKD-EPI eGFR.
DISCUSSION Living renal transplantation is the definitive therapeutic option for renal failure, offering better outcomes than cadaveric donor kidney transplantation. Comprehensive living donor evaluation is essential to ensure the optimal recipient outcome and donor safety. Accurate assessment of renal function in donors is particularly important since higher measured baseline GFR in donors before transplantation independently predicts improved allograft outcomes in recipients, including better function and a lower risk of rejection.11 In addition, during followup the risk of postoperative CKD is greater in donors who had lower baseline GFR with overestimation of GFR.12,13 This study shows that the MDRD equation had better precision, bias and accuracy among creatinine based estimates. When kidney volume was factored into the model, GFR estimation was enhanced. The gold standard for renal function measurement requires serial blood sampling to calculate the clearance of a substance, the ideal substance being inulin. Current acceptable standards include radiolabeled compounds such as 125I-iothalamate and 99 Tc, which are costly and not readily available.3 At
our institution 99Tc urinary clearance serves as the reference standard, against which we compared eGFR and vGFR. The MDRD and CKD-EPI equations use wellknown variables in individuals, including gender, weight, age and SCr, to estimate renal function in patients with renal impairment.7,8 The MDRD equation was developed in 1,628 patients with renal impairment and SCr 1.2 to 7 mg/dl, while excluding patients with GFR greater than 70 ml/minute/ 1.73 m.2 The CKD-EPI formula was derived from direct nuclear measurements of GFR in each gender of many races in a wide range of clinical scenarios. An understanding of the sample groups of patients used for the MDRD and CKD-EPI equations indicates that correlation results derived from these equations in healthy, potential kidney donors may be suboptimal. The biological relationship between GFR and SCr is more linear in patients with CKD than in healthy individuals. Thus, linear extrapolation of GFR may not apply to healthy donors, in whom renal function exceeds 60 ml/minute/ 1.73 m2.3 While the MDRD and CKD-EPI equations demonstrated a similar concordance coefficient
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
compared to dGFR, the Lin concordance correlation index of 0.3 for each suggested a poor linear association of estimates (precision and reproducibility). The MDRD equation had the most favorable combination of better accuracy, precision and lower bias compared to the CKD-EPI equation and CrCl. The Lin concordance index was between 0.12 and 0.34 for the performance of MDRD, reexpressed MDRD, the Cockcroft-Gault formula and CrCl estimates against 125I-iothalamate GFR in 423 kidney donors.14 Low concordance ratios were also documented for the MDRD, Cockcroft-Gault and Mayo Clinic formulas (0.50, 0.36 and 0.46, respectively) in a randomly selected population of 112 kidney donors.4 In that study the MDRD equation outperformed other creatinine based estimates of GFR, consistent with our findings. While the MDRD equation is suboptimal, current reports support it as the best available equation to estimate renal function in potential donors. CrCl consistently performed most poorly in all studies compared to creatinine based equations.4,14 Because timed urinary collections are cumbersome and prone to error, 24-hour urine collections are not routinely done to estimate renal function.3 Equation estimates of GFR in healthy donors are often lower than directly measured values. The mean difference between GFR estimated from the MDRD equation and direct GFR measurement ranges from e29 to 3.3 ml/minute/1.73 m2.3 For Asian donors at our institution the equations tend to overestimate GFR but with a minimal mean difference using the MDRD formula (e1.6 to 0.9 ml/ minute/1.73 m2). Recent studies identified preoperative kidney volume as an important predictor of graft function in living donor kidney transplantation.15,16 Larger donor renal volume correlated with lower nadir SCr at 1 year in transplant recipients.17 Preoperative kidney volume also independently predicted delayed recovery of kidney function in donors after nephrectomy.5 Beyond transplantation, renal parenchymal volume measurements were used to accurately estimate differential renal function in normal individuals and those with urinary obstruction.18 Kidney volume is inherently related to baseline donor kidney function, as demonstrated in correlation studies using ultrasound measured kidney size.19 Volume based GFR measurements may also serve to predict residual renal function after partial or radical nephrectomy. This is of particular importance in patients with borderline GFR and may aid in counseling on postoperative renal replacement therapy.
1825
We validated that the volume based equation outperformed MDRD eGFR in our Asian population using statistical assessment. Incorporating kidney volume enhanced GFR estimation in healthy donors. In the absence of comorbid disease, including diabetes mellitus and significant hypertension, in young healthy donors the kidneys show less glomerulosclerosis and scarring. Thus, kidney parenchymal volume expectedly correlates with healthy nephron mass and glomerular function correlates with accuracy. The anatomical studies by Nyengaard and Bendtsen revealed a direct linear relationship between kidney weight and total glomerular number using stereological techniques in autopsy specimens from subjects with no kidney disease.20 Contrast enhanced spiral CT in potential donors is the current recommended standard imaging modality before donor nephrectomy, providing the advantage of using vGFR.2 Besides delineating anatomy to aid surgery, GFR can be estimated with an improved correlation at no additional cost or procedure. After validation with larger studies and in different donor populations, vGFR may potentially be the best readily available method of estimating GFR in healthy populations. There were several limitations of our study, including relatively few donors. The patient population may not be readily generalized to other population groups since we included predominantly Asian donors with a normal mean BMI and slight variations (the ideal BMI in Asian population is 18.5 to 23.5 kg/m2). Functional kidney parenchymal volume is relatively lower than that derived from white donors but similar to kidney weight in published Asian series.5 While the volume based equation was derived and forward tested in a white population, the reproducibility of performance in an Asian population may be related to renal function per unit kidney volume, which is independent of race or BMI in a healthy population.
CONCLUSIONS In our healthy Asian population of potential living related renal donors MDRD eGFR had the best concordance, accuracy and bias compared to dGFR. The CT measured, kidney volume based equation of Herts et al6 showed better precision with less bias and improved concordance with dGFR than MDRD eGFR. vGFR could be an initial, complementary screening modality to verify eGFR, especially in healthy populations, which are different than those from whom creatinine based equations of GFR were derived.
1826
METHODS OF ESTIMATING GLOMERULAR FILTRATION RATE IN POTENTIAL KIDNEY DONORS
REFERENCE 1. Port FK, Merion RM, Wolfe RA et al: Trends in organ donation and transplantation in the United States, 1997-2006. Am J Transplant 2008; 8: 911. 2. Delmonico F: A Report of the Amsterdam Forum on the care of the live kidney donor: data and medical guidelines. Transplantation, suppl., 2005; 79: S53. 3. Stevens LA, Coresh J, Greene T et al: Assessing kidney function-measured and estimated glomerular filtration rate. N Engl J Med 2006; 354: 2473. 4. Ibrahim HN, Rogers T, Tello A et al: The performance of three serum creatinine-based formulas in estimating GFR in former kidney donors. Am J Transplant 2006; 6: 1479. 5. Jeon HG, Lee SR, Joo DJ et al: Predictors of kidney volume change and delayed kidney function recovery after donor nephrectomy. J Urol 2010; 184: 1057.
filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: 461. 8. Levey AS, Stevens LA, Schmid CH et al: A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: 604. 9. Dello SAWG, van Dam RM, Slangen JJG et al: Liver volumetry plug and play: do it yourself with ImageJ. World J Surg 2007; 31: 2215. 10. Lin LI: A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989; 45: 255. 11. Issa N, Stephany B, Fatica R et al: Donor factors influencing graft outcomes in live donor kidney transplantation. Transplantation 2007; 83: 593. 12. Tan L, Tai BC, Wu F et al: Impact of kidney disease outcomes quality initiative guidelines on the prevalence of chronic kidney disease after living donor nephrectomy. J Urol 2011; 185: 1820.
6. Herts BR, Sharma N, Lieber M et al: Estimating glomerular filtration rate in kidney donors: a model constructed with renal volume measurements from donor CT scans. Radiology 2009; 252: 109.
13. Ibrahim HN, Foley R, Tan LP et al: Long-term consequences of kidney donation. N Engl J Med 2009; 360: 459.
7. Levey AS, Bosch JP, Lewis JB et al: A more accurate method to estimate glomerular
14. Issa N, Meyer KH, Choure G et al: Evaluation of creatinine-based estimates of glomerular
filtration rate in a large cohort of living kidney donors. Transplantation 2008; 86: 223. 15. Saxena AB, Busque S, Arjane P et al: Preoperative renal volumes as a predictor of graft function in living donor transplantation. Am J Kidney Dis 2004; 44: 877. 16. Poggio ED, Hila S, Stephany B et al: Donor kidney volume and outcomes following live donor kidney transplantation. Am J Transplant 2006; 6: 616. 17. Hugen CM, Polcari AJ, Farooq AV et al: Size does matter: donor renal volume predicts recipient function following live donor renal transplantation. J Urol 2011; 185: 605. 18. Morrisroe SN, Su RR, Bae KT et al: Differential renal function estimation using computerized tomography based renal parenchymal volume measurement. J Urol 2010; 183: 2289. 19. Barai S, Gambhir S, Prasad N et al: Levels of GFR and protein-induced hyperfiltration in kidney donors: a single-center experience in India. Am J Kidney Dis 2008; 51: 407. 20. Nyengaard JR and Bendtsen TF: Glomerular number and size in relation to age, kidney weight, and body surface in normal man. Anat Rec 1992; 232: 194.