Living donor kidney transplantation offers the best recipient and allograft outcomes and is the most favorable renal replacement therapy option for patients with end-stage renal disease (ESRD).1 As a result of donating a kidney, living kidney donors eventually incur a loss of approximately one-third of starting glomerular filtration rate (GFR)2 and consequently a higher risk of ESRD.3,4 Furthermore, a lower predonation GFR is a risk factor for progression to ESRD after kidney donation.5,6 As a result, a detailed evaluation of predonation GFR is an important aspect of donor selection.
United Network for Organ Sharing (UNOS) currently requires measurement of GFR using exogenous filtration markers or 24-hour creatinine clearance (CrCl) as part of living kidney donor evaluation.7 The former is the gold standard; however, expense and resource intensiveness limit its widespread use. A survey conducted in 2007 found that 90% of the living donor kidney transplant programs in the US rely on 24-hour CrCl.8 One notable limitation of using CrCl is the susceptibility to technical errors due to inaccurate urine collections. Traditionally, accuracy of 24-hour urine collections is assessed by comparing the measured creatinine excretion rate (CER) to the expected CER of 20–25 mg/kg/day in men and 15–20 mg/kg/day in women.9 These reference values were derived by combining the average creatinine excretion equations from 4 studies with “ideal” weights for height from the 1983 Metropolitan Life Insurance Company tables and have not been externally validated. These do not account for several important determinants of endogenous creatinine generation, including age and race. Ix et al developed and validated 2 equations that provide a more refined assessment of expected CER by incorporating age, race, and serum phosphorous levels (if available) in addition to gender and body weight.10 The usefulness of these equations for assessing completeness of urine collections in the context of CrCl measurement among donor candidates has not been evaluated.
Additionally, it is well established that of the available creatinine-based GFR-estimating equations, the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation provides the least biased, most precise, and most accurate estimate in individuals with normal or only mildly reduced GFR; therefore, it is the preferred equation to calculate estimated GFR (eGFR) in living kidney donor candidates.11 During evaluation of this select population, both CrCl and eGFR are usually obtained if measured GFR is not.8 However, to the best of our knowledge, there are no studies that compare performances of CrCl and eGFR derived using the CKD-EPI creatinine equation against measured GFR for evaluation of renal function in donor candidates.
The goals of this study were 2-fold: (1) to assess how the traditional weight- and gender-based approach of assessing urine collection adequacy compared with the more recently defined and validated approach by Ix et al; and (2) to compare the performance of commonly used metrics of renal function assessment with 125I-iothalamate GFR (iGFR) as the gold standard. Metrics assessed included eGFR by CKD-EPI equation, 24-hour CrCl, and the average of these 2 measures (Avg [CrCl and eGFR]).
MATERIALS AND METHODS
This is a retrospective study of 1504 individuals aged 18 years or older, who were evaluated for kidney donation at the Cleveland Clinic between 1996 and 2013 (Figure 1). Of these, 1412 donor candidates who underwent both a CrCl and iGFR measurement as part of their evaluation were included in the study. The eGFR, CrCl, and iGFR measurements show no significant change over this period of time. This study was approved by the Institutional Review Board at the Cleveland Clinic.
CrCl was calculated from a 24-hour urine sample collected by a donor candidate and then adjusted to body surface area (BSA) of 1.73 m.2 Two separate approaches were used to assess completeness of 24-hour urine collections:9,10
- Approach CER2: 24-hour urine collection was considered accurate by comparing measured CER with CER estimated based on 2 variables—weight and gender, that is, between 20 and 25 mg/kg body weight for men and between 15 and 20 mg/kg body weight for women.
- Approach CER4: 24-hour urine collection was considered accurate if measured CER was within 20% in either direction of CER estimated using Equation D proposed by Ix et al.10 This equation includes 4variables—age, race, weight, and gender:
Estimated CER = 879.89 + 12.51 × weight (kg) − 6.19 × age + (34.51 if black) − (379.42 if female)10
Urinary clearance of 125I-iothalamate was used to measure iGFR.12 A water load was prescribed as follows: 1–2 L the night before the test, 500 mL in the morning before the test, and an additional 10 mL/kg at the beginning of the test followed by continuous administration throughout to ensure a urine output of 200–400 mL/h. Twenty-five mCuries of 125 I-sodium iothalamate was injected subcutaneously. Baseline urine and blood samples were obtained. Two timed urine collections were obtained with blood samples drawn before and after each urine collection. The Packard Minaxi 5000 series counter was used to determine isotope activity. Mean GFR was calculated from the 2 consecutive clearance values and then adjusted for BSA of 1.73 m.2
Baseline characteristics of donor candidates based on the 2 approaches for assessment of accuracy of 24-hour urine collection were analyzed. Additionally, since a substantially higher number of 24-hour urine collections were deemed to be accurate as per CER4, we evaluated differences in characteristics between donor candidates whose urine collections were deemed accurate by CER2 versus those whose urine collections were deemed accurate by CER4 only (ie, accurate as per CER4 but inadequate as per CER2). The Wilcoxon rank-sum test was used to compare continuous variables and χ2 test was used for binary variables.
We evaluated associations between each of the following estimates or measures of renal function against the gold standard iGFR:
- eGFR by CKD-EPI equation11
- CrCl with accuracy of 24-hour urine collection assessed using CER
- Avg (CrCl and eGFR) with accuracy of urine collection assessed using CER4
We selected CER4 for these analyses since the equation used in this approach incorporates age and race in addition to gender and body weight, is derived from a larger cohort, and is validated in an external cohort.10
Associations for different gender, race (black versus non-black), and age were investigated. For each of the above estimates/measures, bias was assessed by median difference, median percentage difference, median absolute difference, and median percentage absolute difference. Accuracy was assessed by percentage of values falling within 10% and 30% of iGFR. For continuous variables, 3-group comparisons were done using the Friedman test and pairwise comparisons using the paired Wilcoxon rank-sum test. For binary variables, 3-group comparisons were done using Cochran’s Q test and pairwise comparisons using McNemar’s test. Multiple pairwise comparisons were corrected using Bonferroni correction. The agreement between each of these measures/estimates of renal function and iGFR was assessed using scatterplots (eg, eGFR versus iGFR) and residual plots (eg, eGFR − iGFR versus iGFR). An internal validation of the models was done using the 0.632+ bootstrap method with 300 repetitions and compared using the mean squared error (MSE).13 All descriptive data were analyzed using the statistical software package R.
There were 1412 individuals who had both iGFR and 24-hour CrCl measured. Seven hundred sixty-nine donor candidates had accurate urine collections as defined by CER4 and 525 donor candidates had accurate urine collection as defined by CER2 (Table 1).
There were 327 donor candidates who had urine collections deemed accurate by CER4 but inaccurate by CER2. Compared with the CER2 group, these donor candidates were older, included a higher proportion of men, and had higher serum creatinine, lower CrCl, and lower iGFR values. Notably, a large majority of these (164/176 or 93.2% among men and 131/151 or 86.8% among women) had CERs below the lower limit of the reference range in CER2 (20 mg/kg/d for men and 15 mg/kg/d for women). Another 83 donor candidates had urine collections accurate by CER2 but inaccurate by CER4. Since CER4 was developed and validated using robust methodology and has been previously shown to provide a less biased estimate of CER than the equations on which CER2 is based, this relatively small group was not analyzed further.9,10
Performance of 24-Hour CrCl, eGFR, and Avg (CrCl and eGFR) Compared With iGFR
Twenty-four-hour CrCl overestimated iGFR with a median bias of 2.2 mL/min/1.73 m2 and median percentage bias of 2.3% overall (Table 2). This observation held true when comparing subpopulations based on race (black and non-black) and age (≤ 42 y and > 42 y; 42 y being the median age for the entire cohort). However, in men, CrCl overestimated iGFR by 6.7 mL/min/1.73 m2 or 6%, and in women, CrCl slightly underestimated iGFR by 1 mL/min/1.73 m2 or 0.8%.
eGFR underestimated iGFR with a median bias of −5.4 mL/min/1.73 m2 and median percentage bias of −5.0%. Comparisons in subpopulations based on gender and age yielded similar numbers; however, in black donor candidates, eGFR overestimated iGFR by 3.2 mL/min/1.73 m2 or 3.5%.
Among the 3 measures, median bias was the lowest using Avg (CrCl and eGFR) at −1.0 mL/min/1.73 m2 or −1.1%. This observation was concordant in male, female, age ≤ 42 years, age > 42 years, and non-black populations (all P < 0.001). However, since both CrCl and eGFR overestimated iGFR in blacks, Avg (CrCl and eGFR) did not provide a better bias statistic in this subpopulation (P was not significant).
Accuracy as measured by the proportion of values falling within 10% of iGFR was moderate for all 3 measures. It was not different between CrCl and eGFR overall or in any of subpopulations. However, accuracy improved using Avg (CrCl and eGFR) when compared with CrCl alone in the entire population, in men and non-black subgroups (P < 0.05). When compared with eGFR alone accuracy improved only in the non-black subgroup.
Accuracy as measured by proportion of values with 30% of iGFR was higher with eGFR compared with CrCl overall and in men, non-black, ≤ 42 years, and >42 years subgroups (P < 0.05) but not in women and blacks. This metric was also improved using Avg (CrCl and eGFR) when compared with CrCl overall, and in men, women, non-black, ≤ 42 years, and >42 years subgroups (P < 0.05) but not in black. When compared with eGFR alone, accuracy improved only in the non-black subgroup.
Figures 2 and 3 show the relationship between CrCl and iGFR, eGFR and iGFR, and Avg (CrCl and eGFR) and iGFR using scatterplots and residual plots, respectively. The scatterplots show better correlation between Avg (CrCl and eGFR) and iGFR (r = 0.50) compared with either measure alone (r = 0.43 with CrCl and r = 0.47 with eGFR). The residual plots show that the mean bias is positive with CrCl, negative with eGFR, and nonsignificant with Avg (CrCl and eGFR). Notably, all 3 estimates/measures overestimate iGFR at lower values (ie, have positive residuals) and underestimate iGFR at higher values (ie, have negative residuals).
Internal Validation of the Models
Internal validation based on the 0.632+ bootstrap method shows that the Avg (CrCl and eGFR) estimate of iGFR (MSE = 209.8, mean bootstrap corrected) outperformed both CrCl (MSE = 223.1) and eGFR (MSE = 225.7) (P < 0.001 for both). The conclusion holds when only patients with iGFR at 90 mL/min/1.73 m2 or less were evaluated (Avg [CrCl and eGFR] versus CrCl, P = 0.008; Avg [CrCl and eGFR] versus eGFR, P < 0.001).
Fifth Percentile Data for iGFR, CrCl, eGFR, and Average of CrCl and eGFR
As noted above and as seen in Figure 3, CrCl, eGFR, and Avg (CrCl and eGFR) underestimated iGFR at higher-normal values. In a donor candidate with an excellent iGFR, for example, 140 mL/min/1.73 m2, a slight underestimation is unlikely to affect his or her candidacy for kidney donation. On the contrary, all 3 metrics overestimated iGFR at lower-normal values where accuracy of these is more likely to be relevant toward decision-making. Table 3 shows the fifth percentile values of iGFR for the entire cohort and for black and non-black subpopulations in various age groups and corresponding values for CrCl, eGFR, and Avg (CrCl and eGFR). The entire range of iGFR, CrCl, eGFR, and Avg (CrCl and eGFR) values for the entire population and various subgroups is provided in Table 4.
A lower GFR in a kidney donor candidate is a risk factor for progression to ESRD.5,6 If 2 otherwise similar kidney donors with predonation GFRs of 120 mL/min/1.73 m2 and 90 mg/min/1.73 m2 and postdonation GFRs of 80 mL/min/1.73 m2 and 60 mg/min/1.73 m2, respectively, were to develop chronic kidney disease that progressed at similar rates, the latter with lower renal function will have a higher risk of reaching ESRD. One study evaluating risk factors for ESRD in kidney donors demonstrated a 36% higher risk of ESRD with 10 mL/min/1.73 m2 decrease in predonation eGFR.6 Consequently, kidney donor evaluation represents one of the few scenarios in nephrology where accurate assessment of GFR is essential. The results from this large single-center study add to our knowledge regarding kidney function assessment in living kidney donor candidates in several ways. First, the age, gender, weight, and race-based equation developed by Ix et al identified a significantly higher proportion of urine collections as accurate; 42.6% of these would have been deemed as inaccurate, mostly under-collections, using the conventional gender- and weight-based methodology. Second and as expected, CrCl overestimated GFR and eGFR underestimated GFR. Third, using Avg (CrCl and eGFR) essentially eliminated the bias in assessment of GFR and led to a modest increase in accuracy. Lastly, our study provides data regarding iGFR and corresponding values of CrCl, eGFR, and Avg (CrCl and eGFR) in living kidney donor candidates, which can serve as a reference in clinical practice, since all transplant centers have access to these 2 methods of GFR evaluation.
Comparing measured CER with estimated CER to confirm accuracy of timed urine collections is the first step in using the data derived from the 24-hour urine collections to assess kidney function. Reference ranges for estimated CERs of 20–25 mg/kg for men and 15–20 mg/kg for women were obtained by combining results from 4 small studies.9 The equation utilized in our study in CER4 provides a more refined assessment of CER.10 In addition to weight and gender, it incorporates 2 important determinants of endogenous creatinine production and excretion—age and race. Furthermore, this equation was developed from a large sample size of 2466 participants from 3 kidney disease trials, and then externally validated in 987 participants from 3 separate trials. Although the CER4 formula is more complex than the CER2, and therefore cannot realistically be remembered, in this era of electronic medical records either formula can be easily programmed into the laboratory system. In addition, using CER4 avoids the cumbersome process of repeating 24-hour urine collections when initial collections are deemed inadequate by CER2. Therefore, we recommend using it as the preferred method for assessment of accuracy of timed urine collections while assessing CrCl in living donor candidates. It is important to highlight that a vast majority (93.2% among men and 86.8% among women) of the collections that were considered accurate by this approach but inaccurate by CER2 had measured CERs below the lower reference limits of 20 mg/kg/day for men and 15 mg/kg/day for women.
Several noteworthy observations were made regarding various surrogates of kidney function assessed in our study. As is already well known, GFR was overestimated by CrCl and underestimated by eGFR.11 Also, the CKD-EPI equation yielded a lower median bias of −5.4 mL/min/1.73 m2 than the values of −11.0 mL/min/1.73 m2 and −16.3 mL/min/1.73 m2 with the modification of diet in renal disease (MDRD) equation and the re-expressed MDRD equation, respectively, as previously published from our cohort of living donors.14 Interestingly, Avg (CrCl and eGFR) provided the least biased estimate of kidney function overall and in all subpopulations except in blacks. A moderate increase in accuracy was also observed. Based on the results of our study, we recommend using Avg (CrCl and eGFR) as the most representative surrogate of renal function. This is especially pertinent as both these measurements are routinely obtained in all living donor candidates who do not have a GFR measured using exogenous makers.
Two recently published studies provide a web-based application that uses eGFR to compute the probability that measured GFR of a donor candidate is above a predefined threshold, most commonly used one being 80 mL/min/1.73 m2.15,16 Based on these, the recently published Kidney Disease: Improving Global Outcomes (KDIGO) Clinical Practice Guideline for living kidney donor evaluation recommends using eGFR as a screening test to identify candidates for whom subsequent testing may not be necessary.11,17-19 However, the data supporting use of 80 mL/min/1.73 m2 as a cut-off for donor candidates is based on recipient outcomes and, in our opinion, should not be used for donor risk assessment.20 In other words, knowledge of whether the donor’s true GFR is above or below 80 mL/min/1.73 m2 does not obviate the need for actual kidney function assessment especially in younger donors.21 For example, from the data provided in Table 4, a predonation iGFR of 90 mL/min/1.73 m2 in a 65-year-old non-black woman is near the median for her expected iGFR and should not preclude donation. However, this same value is below the fifth percentile in the case of a 25-year-old non-black woman and consequently of much greater concern. Tables 3 and 4 provide data on renal function assessed by iGFR, CrCl, eGFR, and Avg (CrCl and eGFR) in otherwise healthy donor candidates and can potentially serve as references for minimum acceptable GFRs during donor evaluations, for each of the methodologies that might be used to assess function.
Our study has the limitations inherent to a single-center design. However, including subjects from only 1 center allowed us to use iGFR measured using uniform and robust methodology similar to that used in the MDRD and the African American Study of Kidney Disease and Hypertension (AASK) trials.22,23 The CKD-EPI equation based on the combined creatinine-cystatin C equation performs better at estimating GFR than the one based on creatinine alone.17 However, we did not have cystatin C values available in our cohort. Lastly, the benefits observed to using the Avg (CrCl and eGFR) equation do not apply to blacks—however, the Avg (CrCl and eGFR) does not appear to have any disadvantages in blacks compared to using just CrCl or just eGFR, so we feel the Avg(CrCl and eGFR) equation could be widely used.
In summary, this study from a large cohort of living kidney donor candidates provides head-to-head comparison of various commonly used methods of kidney function assessment. In the absence of availability of a measured GFR, the eGFR-CrCl average is the best surrogate in all populations except blacks.
1. Saran R, Robinson B, Abbott KC, et al. US renal data system 2018 annual data report: epidemiology of kidney disease in the United States.Am J Kidney Dis2019733S1A7–A8
2. Kasiske BL, Anderson-Haag T, Israni AK, et al. A prospective controlled study of living kidney donors: three-year follow-up.Am J Kidney Dis201566114–124
3. Muzaale AD, Massie AB, Wang MC, et al. Risk of end-stage renal disease following live kidney donation.JAMA2014311579–586
4. Mjøen G, Hallan S, Hartmann A, et al. Long-term risks for kidney donors.Kidney Int201486162–167
5. Grams ME, Sang Y, Levey AS, et al.; Chronic Kidney Disease Prognosis ConsortiumKidney-failure risk projection for the living kidney-donor candidate.N Engl J Med2016374411–421
6. Locke JE, Reed RD, Massie A, et al. Obesity increases the risk of end-stage renal disease among living kidney donors.Kidney Int201791699–703
7. OPTN (Organ Procurement and Transplantation Network)/UNOS (United Network for Organ Sharing)OPTN Policies, Policy 14: Living Donation.Available at http://optn.transplant.hrsa.gov/ContentDocuments/OPTN_Policies.pdf
. Accessed November 3, 2017
8. Mandelbrot DA, Pavlakis M, Danovitch GM, et al. The medical evaluation of living kidney donors: a survey of US transplant centers.Am J Transplant200772333–2343
9. Walser M. Creatinine excretion as a measure of protein nutrition in adults of varying age.JPEN J Parenter Enteral Nutr1987115 Suppl73S–78S
10. Ix JH, Wassel CL, Stevens LA, et al. Equations to estimate creatinine excretion rate: the CKD epidemiology collaboration.Clin J Am Soc Nephrol20116184–191
11. Levey AS, Stevens LA, Schmid CH, et al.; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration)A new equation to estimate glomerular filtration rate.Ann Intern Med2009150604–612
12. Israelit AH, Long DL, White MG, et al. Measurement of glomerular filtration rate utilizing a single subcutaneous injection of 125I-iothalamate.Kidney Int19734346–349
13. Efron B, Tibshirani R. Improvements on cross-validation: the 632+ bootstrap method.J Am Stat Assoc199792548–560
14. Issa N, Meyer KH, Arrigain S, et al. Evaluation of creatinine-based estimates of glomerular filtration rate in a large cohort of living kidney donors.Transplantation200886223–230
15. Huang N, Foster MC, Lentine KL, et al. Estimated GFR for living kidney donor evaluation.Am J Transplant201616171–180
16. Gaillard F, Flamant M, Lemoine S, et al. Estimated or measured GFR in living kidney donors work-up?Am J Transplant2016163024–3032
17. Inker LA, Schmid CH, Tighiouart H, et al.; CKD-EPI InvestigatorsEstimating glomerular filtration rate from serum creatinine and cystatin C.N Engl J Med201236720–29
18. Lentine KL, Kasiske BL, Levey AS, et al. Summary of kidney disease: improving global outcomes (KDIGO) clinical practice guideline on the evaluation and care of living kidney donors.Transplantation20171011783–1792
19. Lentine KL, Kasiske BL, Levey AS, et al. KDIGO clinical practice guideline on the evaluation and care of living kidney donors.Transplantation20171018S Suppl 1S1–S109
20. Nordén G, Lennerling A, Nyberg G. Low absolute glomerular filtration rate in the living kidney donor: a risk factor for graft loss.Transplantation2000701360–1362
21. Kher A, Mandelbrot DA. The living kidney donor evaluation: focus on renal issues.Clin J Am Soc Nephrol20127366–371
22. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of diet in renal disease study group.Ann Intern Med1999130461–470
23. Lewis J, Agodoa L, Cheek D, et al.; African-American Study of Hypertension and Kidney DiseaseComparison of cross-sectional renal function measurements in African Americans with hypertensive nephrosclerosis and of primary formulas to estimate glomerular filtration rate.Am J Kidney Dis200138744–753