ABSTRACT
Background
Creatinine-based equations such as the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) are recommended for estimating glomerular filtration rate (eGFR) in clinical practice, but have reduced performance in advanced stages of chronic kidney disease. However, only rarely studies have evaluated the performance of eGFR by measuring the average of the urinary clearances of creatinine and urea (mClUN-cr) compared with the eGFR equations.
Methods
This cross-sectional study evaluated the usefulness of mClUN-cr in a population of 855 participants who performed a GFR measurement by urinary inulin clearance. The performance of mClUN-cr was compared with those of CKD-EPI 2009 and CKD-EPI 2021, considering three criteria: bias, precision and accuracy.
Results
In the whole sample, the mClUN-cr performed similarly to CKD-EPI equations (2009 and 2021) [precision: 11.5 (95% CI 10.5; 12.5) vs 19.0 (95% CI 17.2; 20.1) and 19.1 (95% CI 17.4; 20.4), and accuracy P30: 97.0 (95% CI 95.8; 98.0) vs 82.0 (95% CI 79.2; 84.4) and 77.2 (95% CI 74.5; 80.0)]. The CKD-EPI equations (2009 and 2021) had the best performance when mGFR was >60 mL/min/1.73 m2. In contrast, the mClUN-cr performed better than others with lowest mGFR values, more noticeable when mGFR was <60 mL/min/1.73 m2.
Conclusions
The study described the best performance of mClUN-cr at GFR levels below 60 mL/min/1.73 m2 and a satisfactory result in the overall cohort. The findings point to a role of this tool, especially for estimating GFR in chronic kidney disease patients in developing countries, when reference measurement of GFR is not available.
Keywords: CKD-EPI equation, creatinine clearance, glomerular filtration rate, inulin, urea clearance
INTRODUCTION
The classification of chronic kidney disease (CKD) is notably based on the glomerular filtration rate (GFR) [1–3]. In everyday practice, measurement of GFR (mGFR) by reference methods (inulin, iohexol, iothalamate, 51Cr-EDTA, etc.) is difficult to perform, and CKD clinical guidelines have recommended equations to estimate GFR (eGFR) as non-invasive alternatives, especially equations based on plasma creatinine (PCr), such as the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [1–5].
The 2009 CKD-EPI equation was developed to provide a more accurate eGFR among individuals with normal or slightly reduced mGFR and is recommended to estimate GFR in adults of any age in North America, Europe and Australia, considering age, sex and ethnicity [1, 2, 6, 7]. This equation was developed using data pooled from 10 studies using iothalamate as a reference method. A new CKD-EPI equation (CKD-EPI 2021) without ethnicity was developed and considered sufficiently accurate for clinical practice, although equations that use two markers (PCr and cystatin C) are more accurate in estimating mGFR than equations with either the PCr or cystatin C level alone [8]. Besides, the CKD-EPI equation cannot be used in all patients in clinical practice, especially patients with severe chronic disease or with low muscle mass or who are undernourished, when PCr does not reflect renal function [1–5, 9].
The determination of urinary creatinine clearance (mClcr) is an alternative method of eGFR evaluation. Nevertheless, mClcr is not ideal for this purpose, partly because PCr depends on non-renal factors, such as age, gender, ethnicity, muscle mass, diet, drugs and proximal tubular secretion (10%–40%) [1, 2, 10–12]. Even with an accurate urine collection, the mClcr overestimates mGFR by 10%–40% [3, 4, 10, 12–18].
Among patients with severe CKD (stages 4–5), the urinary urea clearance (mClUN) significantly underestimated mGFR [13–16, 19]. Since mClcr significantly overestimates mGFR, some authors have proposed to calculate the average between mClcr and mClUN (mClUN-cr) in patients with advanced kidney disease [4, 14, 17, 19, 20]. The 2005 European Best Practices Guidelines recommended this method to estimate GFR in severe CKD stages (stages 4–5) [21]. However, remarkably, there is a scarcity of articles concerning the mClUN-cr [11–14, 16–19]. Therefore, the aim of this study is to evaluate the usefulness of mClUN-cr as an alternative method when reference measurement is not available, and to compare the performance of mClUN-cr with those of CKD-EPI 2009, CKD-EPI 2021 and mClcr, using urinary inulin clearance as a reference method.
MATERIALS AND METHODS
Study population
This retrospective, cross-sectional study considered all 855 consecutive participants who performed a GFR measurement by urinary inulin clearance between July 2003 and July 2013 in a single university hospital (Renal and Metabolic Function Exploration Unit of Edouard Herriot Hospital, Lyon, France). Indications for assessment were: suspected or established renal dysfunction, renal risk or before kidney donation.
Patients were further divided by renal function according to the KDIGO classification as follows: stage 1 (mGFR ≥90 mL/min/1.73 m2), stage 2 (60 ≤ mGFR < 90 mL/min/1.73 m2), stage 3A (45 ≤ mGFR < 60 mL/min/1.73 m2), stage 3B (30 ≤ mGFR < 45 mL/min/1.73 m2) and stage 4–5 (mGFR <30 mL/min/1.73 m2). Stages 4 and 5 CKD were combined due to the low number of patients in each group.
All procedures were conducted in accordance with institutional ethical standards, the 2013 Helsinki Declaration and its later amendments, or with comparable ethical standards. Precisely, an appropriate informed consent was obtained from each participant or his/her legal representatives. The consent form included information on the procedure itself and on the possibility of later use of the data for research purposes.
Laboratory assessments
GFR, mClcr and mClUN measurements
The GFR was measured by renal clearance of inulin (polyfructosan, Inutest; Fresenius Kagi, Graz, Austria). A standard technique was used by a trained staff with a continuous infusion after a priming dose of 30 mg/kg of polyfructosan. Water diuresis was induced by oral administration of 5 mL/kg of water followed by 3 mL/kg every 30 min, combined with an intravenous infusion of 0.9% sodium chloride. This enabled the patients to spontaneously empty their bladder every 30 min. Patients requiring intermittent urethral catheterization were excluded from this study. Three to four urine samples were collected and a blood sample was drawn mid-way through each collection period. Plasma and urine polyfructosan, creatinine and urea measurements were performed to determine mGFR, mClcr and mClUR at the same time using the standard UV/P formula. The clearance values were obtained from the mean values of the three to four clearance periods for each of the parameters. mClUN-cr was calculated as the mean of final result of mClcr and mClUR. All the mGFR results were expressed in 1.73 m2, according to the Dubois formula: body surface area = height0.725 × weight0.425 × 0.007184 [22].
Polyfructosan measurements
Plasma and urine polyfructosan were measured using the same enzymatic method [20] for which we previously checked the imprecision of the assay method (within-run precision values of 0.3% and 0.7%, respectively, and between-run precision values of 3.5%, 1.6% and 2.4% at mean values of polyfructosan 117, 198 and 285 mg/L, respectively) [23].
Creatinine and urea measurements
All creatinine measurements were performed with methods traceable to the National Institute of Standards and Technology [isotope dilution mass spectrometry (IDMS)-calibrated]. From October 2003 to June 2010, plasma and urine creatinine was obtained by a kinetic colorimetric compensated Jaffé technique (Roche Modular, Meylan, France) whose results were standardized by linear regression adjustment versus the concentrations obtained by liquid chromatography mass spectrometry. The calibration equation was as follows: standardized PCr = 0.9395 × Jaffé compensated serum creatinine (in μmol/L) + 4.6964. The coefficient of correlation was 0.97. From June 2010, all plasma and urine creatinine values were obtained by an enzymatic technique (Architect c®, Abbott Diagnostics) traceable to the National Institute of Standards and Technology (NIST SRM 967 and NIST SRM 914). According to KDIGO, the two techniques are relatively similar. Plasma creatinine was expressed in µmol/L. The urea was measured by an enzymatic method and expressed in mmol/L.
GFR estimation
Concomitantly GFR was estimated in each patient by the CKD-EPI equations (2009 and 2021). According to the French recommendations (Haute Autorité de Santé), no correction factor for race and ethnicity in the CKD-EPI 2009 equation should be applied in the European population; therefore, data concerning race and ethnicity were not collected and were not available.
Statistical analysis
The study considered three criteria for performance: bias, precision and accuracy. Bias was defined as the median difference between mGFR and eGFR. Thus, a positive bias indicates an underestimation of mGFR and vice versa. Precision was defined as the interquartile range (IQR) of the differences between mGFR and eGFR. Accuracy was considered under two criteria: (i) the root mean square error (RMSE), calculated as the square root of the difference (log mGFR – log eGFR)2; (ii) the percentage of estimates within ±10% (P10) and ±30% (P30) of the mGFR. A P30 > 90% qualifies an eGFR as satisfactory for clinical interpretation [1]. Agreement was also assessed by the concordance correlation coefficient (CCC) between each eGFR and the mGFR (after logarithmic transformation of their values). The CCC is a measure of agreement that adjusts the Pearson correlation coefficient downward whenever there is a systematic bias between the methods being compared. We compared the correct classification refers to agreement between mGFR and eGFR categories of ≥90, 60–89, 45–59, 30–44 and <30 mL/min/1.73 m2.
The 95% confidence intervals (CIs) were calculated using a bootstrap method BCa (2000 bootstraps). Median biases were compared using Mood's median test. P10 and P30 values were compared using pairwise McNemar, exact, and permutation tests as a post hoc to Cochran Q test. The method of Holm–Bonferroni was used to correct for multiple comparisons.
In the study, the sample size was large and the variability was narrow. Thus, small changes in any variable could lead to small P-values. In order to reduce random bias, the nominal P-value considered for statistical significance was <.005.
The analyses were performed with R for Windows, version 4.1.2 (R-Cran project, http://cran.r-project.org/).
RESULTS
The clinical characteristics of the 855 participants, are shown in Table 1. The participants’ mean age (±SD) was 47.4 ± 14.4 years and 50.4% were women. The mean mGFR (±SD) was 71.3 ± 28.2 mL/min/1.73 m2 (10–168 mL/min/1.73 m2), and 36.5% of measurements were below 60 mL/min/1.73 m2.
Table 1:
The sociodemographic and clinical characteristics of the 855 participants.
| Characteristic | Whole cohort |
|---|---|
| Participants, n (%) | 855 (100.0) |
| Mean age (range), years | 47.4 ± 14.4 (18.0; 83.1) |
| Female sex, n (%) | 431 (50.4) |
| Mean weight, kg | 65.9 ± 15.4 |
| Mean height, m | 1.66 ± 0.09 |
| Mean BSA, m2 | 1.73 ± 0.21 |
| Mean BMI, kg/m2 | 23.8 ± 5.0 |
| BMI ≥30, n (%) | 83 (9.7) |
| Diabetes, n (%) | 128 (15.0) |
| Mean PCr, mg/dL | 1.14 ± 0.50 |
| Mean mGFR in all participants, mL/min/1.73 m2 | 71.3 ± 28.2 |
| mGFR category, n (%) | |
| <30 mL/min/1.73 m2 | 51 (6.0) |
| 30 to <44 mL/min/1.73 m2 | 105 (12.3) |
| 45 to <59 mL/min/1.73 m2 | 156 (18.2) |
| 60 to <90 mL/min/1.73 m2 | 316 (36.9) |
| ≥90 mL/min/1.73 m2 | 227 (26.6) |
| Median albuminuria (IQR), mg/g | 28.0 (8.0; 185.5) |
| Albuminuria, n (%) | |
| UACR <30 mg/g | 667 (75.5) |
| UACR 30–300 mg/g | 164 (18.5) |
| UACR >300 mg/g | 24 (6.0) |
Data are presented as mean ± standard deviation, median (IQR) or n (%).
Comparison of the performance of GFR estimating equations
In the overall population, the CKD-EPI (2009 and 2021) equations and mClcr overestimated the mGFR, whereas the mClUN underestimated it. Although bias was similar for the CKD-EPI equations and mClUN-cr, the precision was superior for mClUN-cr [IQR (95% CI) 11.5 (10.5; 12.5), P < .001], as was accuracy [P10 and P30, 48.0% (44.5; 51.2) and 97.0 (95.8; 98.0), P < .001] and the lower RMSE (95% CI) [0.130 (0.122; 0.146)]. In addition, mClUN-cr had the only CCC greater than 0.9 [0.941 (0.933; 0.948)]. Performance of the mClcr was the worst for all metrics (Table 2). Graphically, the smallest dispersion of the points around the regression line and the closest proximity of the concordance line (45° line) are observed for mClUN-cr (Fig. 1).
Table 2:
Bias, precision and accuracy of the four GFR estimating equations.
| Subjects and criteria | CKD-EPI 2009 equation | CKD-EPI 2021 equation | Creatinine clearance | Urea clearance | Average of creatinine and urea clearances |
|---|---|---|---|---|---|
| Population = 855 | |||||
| Median bias (95% CI) | −3.7 (−4.7; −2.5) | −6.6 (−7.8; −5.5) | −14.0 (−14.0; −13.0)* | 26.0 (25.0; 27.0)* | 5.5 (4.5; 6.0) |
| IQR (95% CI) | 19.0 (17.2; 20.1) * | 19.1 (17.4; 20.4)* | 14.0 (13.0; 15.0)* | 20.0 (18.0; 21.0)* | 11.5 (10.5; 12.5) |
| P10 (95% CI) | 37.1 (33.8; 40.5)* | 36.3 (33.1; 39.4)* | 19.7 (17.2; 22.5)* | 0.8 (0.02; 1.5)* | 48.0 (44.5; 51.2) |
| P30 (95% CI) | 82.0 (79.2; 84.4)* | 77.2 (74.5; 80.0)* | 66.0 (62.8; 69.2)* | 20.1 (17.5; 22.9)* | 97.0 (95.8; 98.0) |
| CCC (95% CI) | 0.844 (0.824; 0.862) | 0.825 (0.803; 0.844)* | 0.825 (0.807; 0.841)* | 0.599 (0.573; 0.624)* | 0.941 (0.933; 0.948) |
| RMSE (95% CI) | 0.234 (0.220; 0.254) | 0.235 (0.221; 0.254) | 0.159 (0.149; 0.171) | 0.149 (0.141; 0.160) | 0.130 (0.122; 0.146) |
| Agreementa (%) | 59.7 (56.6; 63.2)* | 59.3 (56.0; 62.5)* | 54.4 (51.0; 57.8)* | 9.5 (7.7; 11.6)* | 70.3 (67.1; 73.3)* |
Bias is defined as the median difference between mGFR and eGFR. A positive sign indicates underestimation of mGFR, and a negative sign indicates overestimation of mGFR. P10 and P30 are the proportion of eGFR within 10% and 30% of mGFR, respectively.
aAgreement: correct classification refers to agreement between mGFR and eGFR categories of ≥90, 60–89, 45–59, 30–44 and <30 mL/1.73 m2.
*P < .005 between average clearance creatinine–urea and others eGFR.
Figure 1:
Scatterplots showing, for each estimated GFR versus the measured GFR (in mL/min/1.73 m2). The plain line represents the line regression. The dashed lines represent the 95% confidence limits. The dotted lines represent the perfect concordance.
Agreement according to the KDIGO classification
The overall correct classification for mGFR categories of ≥90, 60–89, 45–59, 30–44 and <30 mL/min/1.73 m2 was better for mClUN-cr [70.3% (67.1; 73.3)] than the other eGFR equations (Table 2).
At mGFR ≥90 mL/min/1.73 m2, the CKD-EPI equations (2009 and 2021) performed better than mClcr, mClUN and mClUN-cr, regarding bias [1.0 (−1.5; 3.5) and −2.5 (−5.0; −0.7)] and accuracy P30 [94.3 (90.8; 97.0) and 94.3 (91.2; 97.4)], but had less precision than mClUN-cr [IQR 21.5 (17.0; 25.0) and 20.0 (16.5; 24.0) versus 11.5 (9.5; 13.5)]. The mClUN-cr had the best precision with the smallest value of IQR whatever the mGFR category (Table 3).
Table 3:
Bias, precision and accuracy of the four GFR estimating equations.
| Subjects and criteria | CKD-EPI 2009 equation | CKD-EPI 2021 equation | Creatinine clearance | Urea clearance | Average of creatinine and urea clearances |
|---|---|---|---|---|---|
| mGFR ≥90 mL/min/1.73 m2 (N = 227) | |||||
| Median bias (95% CI) | 1.0 (−1.5; 3.5) | −2.5 (−5.0; −0.7) | −17.0 (−18.0; −14.0) | 43.0 (41.0; 45.0) | 13.5 (12.5; 14.5) |
| IQR (95% CI) | 21.5 (17.0; 25.0) | 20.0 (16.5; 24.0) | 18.0 (15.0; 20.0) | 13.5 (12.0; 15.0) | 11.5 (9.5; 13.5) |
| P10 (95% CI) | 49.7 (43.2; 56.4)* | 52.0 (45.8; 58.2)* | 33.9 (28.2; 40.0)* | 0.5 (0.0; 1.5)* | 35.2 (29.1; 41.4) |
| P30 (95% CI) | 94.3 (90.8; 97.0)* | 94.3 (91.2; 97.4)* | 88.1 (83.7; 92.7) | 6.7 (3.5; 9.7)* | 98.2 (96.5; 99.6) |
| mGFR 60–89 mL/min/1.73 m2 (N = 316) | |||||
| Median bias (95% CI) | −4.0 (−5.8; −2.0) | −8.0 (−9.3; −5.1) | −14.0 (−15.0; −13.0)* | 28.0 (27.0; 29.0)* | 7.0 (6.0; 7.5) |
| IQR (95% CI) | 21.0 (19.0; 23.0) | 21.5 (19.0; 23.0) | 14.0 (12.0; 16.0) | 8.0 (7.0; 10.0) | 9.0 (8.0; 10.0) |
| P10 (95% CI) | 39.6 (34.2; 45.0)* | 36.0 (31.0; 42.0)* | 21.5(17.1; 26.3)* | 0.0 (0.0; 1.0)* | 47.5 (41.8; 52.5) |
| P30 (95% CI) | 86.1 (82.0; 90.0)* | 81.0 (76.6; 85.1)* | 74.4 (69.6; 78.8)* | 14.9 (11.1; 19.0)* | 97.8 (96.0; 99.0) |
| mGFR 45–59 mL/min/1.73 m2 (N = 156) | |||||
| Median bias (95% CI) | −6.0 (−8.5; −3.5) | −8.8 (−11.5; −6.5) | −15.0 (−16.0; −13.0)* | 6.0 (5.0; 8.0) | 2.0 (1.0; 3.0) |
| IQR (95% CI) | 17.0 (14.5; 21.0) | 17.0 (15.0; 21.0) | 9.0 (4.5; 13.5) | 6.0 (5.0; 8.0) | 7.0 (5.5; 9.0) |
| P10 (95% CI) | 27.0 (20.0; 34.0)* | 27.5 (20.5; 34.6)* | 24.5 (21.0; 28.0)* | 0.6 (0.0; 1.9)* | 62.2 (54.5; 70.0) |
| P30 (95% CI) | 74.5 (67.0; 80.8)* | 66.7 (59.0; 74.3)* | 50.6 (43.0; 58.3)* | 26.3 (19.2; 33.3)* | 98.1 (95.5; 100.0) |
| mGFR 30–44 mL/min/1.73 m2 (N = 105) | |||||
| Median bias (95% CI) | −5.0 (−8.5; −2.4) | −7.5 (−10.5; −4.5)* | 14.0 (12.0; 17.0) | −12.0 (−13.0; −12.0)* | 1.0 (−0.5; 2.0) |
| IQR (95% CI) | 13.0 (10.5; 18.0) | 14.0 (11.0; 18.5) | 13.0 (11.0; 14.0) | 6.0 (4.0; 7.0) | 7.5 (5.0; 8.0) |
| P10 (95% CI) | 23.8 (16.2; 32.4)* | 24.8 (17.2; 33.3)* | 6.7 (2.9; 11.4)* | 2.9 (0.0; 6.7)* | 55.2 (45.7; 64.8) |
| P30 (95% CI) | 68.6 (60.0; 77.0)* | 60.0 (50.5; 68.6)* | 36.2 (26.7; 45.7)* | 35.2 (26.7; 44.8)* | 95.3 (90.5; 99.0) |
| mGFR <30 mL/min/1.73 m2 (N = 51) | |||||
| Median bias (95% CI) | −6.0 (−10.0; −3.5) | −7.0 (−12.0; −5.0)* | −9.0 (−12.0; −8.0) | −7.0 (−7.0; −6.0) | −1.5 (−2.5; −0.5) |
| IQR (95% CI) | 14.0 (8.0; 21.0) | 14.0 (8.5; 20.5) | 6.0 (4.0; 10.0) | 3.0 (2.5; 5.5) | 4.0 (2.5; 6.0) |
| P10 (95% CI) | 23.5 (11.8; 35.3) | 17.6 (8.0; 29.4) | 5.9 (0.0; 13.8) | 3.9 (0.0; 9.8) | 51.0 (37.3; 64.7) |
| P30 (95% CI) | 51.0 (37.3; 64.7) | 45.0 (31.3; 58.9) | 23.5 (11.8; 35.3) | 62.7 (49.0; 76.5) | 86.3 (76.7; 94.1) |
P10, percentage of estimated GFRs that lay within range (measured GFR – 10% of measured GFR; measured GFR + 10% of measured GFR); P30, percentage of estimated GFRs that lay within range (measured GFR – 30% of measured GFR; measured GFR + 30% of measured GFR).
*P < .001 between average clearance creatinine–urea and others eGFR.
The CKD-EPI equations (2009 and 2021) had better performance than mClcr, mClUR and mClUN-cr when mGFR was >60 mL/min/1.73 m2. In contrast, the mClUN-cr performed better than the other four with lowest mGFR values, more noticeable when mGFR <60 mL/min/1.73 m2 (Table 3).
The performance of the equations was similar considering the gender of the participants (Table 4).
Table 4:
Bias, precision and accuracy of the four GFR estimating equations.
| Subjects and criteria | CKD-EPI 2009 equation | CKD-EPI 2021 equation | Creatinine clearance | Urea clearance | Average of creatinine and urea clearances |
|---|---|---|---|---|---|
| Males (n = 424) | |||||
| Median bias (95% CI) | −4.6 (−6.4; −3.7) | −8.2 (−9.4; −6.6) | −15.0 (−16.0; −14.0)* | 25.0 (23.0; 26.0)* | 4.0 (3.5; 5.5) |
| IQR (95% CI) | 18.0 (16.5; 21.0)* | 19.0 (16.8; 21.4)* | 14.0 (13.0; 16.0)* | 21.0 (18.0; 23.0)* | 11.5 (10.0; 13.0) |
| P10 (95% CI) | 35.8 (31.6; 40.5)* | 34.9 (30.7; 39.6)* | 17.7 (13.9; 21.5)* | 1.2 (0.2; 2.4)* | 50.7 (46.0; 55.4) |
| P30 (95% CI) | 80.0 (76.2; 83.7)* | 73.8 (69.6; 78.1)* | 62.3 (57.5; 67.0)* | 25.5 (21.5; 29.7)* | 96.4 (94.6; 98.1) |
| CCC (95% CI) | 0.843 (0.814; 0.868) | 0.821 (0.790; 0.849)* | 0.821 (0.794; 0.844)* | 0.624 (0.587; 0.658)* | 0.945 (0.935; 0.954) |
| RMSE (95% CI) | 0.235 (0.215; 0.260) | 0.235 (0.216; 0.260) | 0.160 (0.148; 0.182) | 0.154 (0.142; 0.175) | 0.136 (0.123; 0.163) |
| Agreementa (%) | 58.5 (53.8; 63.2)* | 57.8 (52.8; 62.5)* | 53.8 (48.4; 57.8)* | 11.8 (8.9; 14.8)* | 70.5 (66.3; 75.0)* |
| Females (n = 431) | |||||
| Median bias (95% CI) | −2.0 (−3.7; −0.7) | −4.6 (−6.3; −2.9) | −14.0 (−15.0; −12.0)* | 27.0 (26.0; 28.0)* | 6.0 (5.5; 7.5) |
| IQR (95% CI) | 17.5 (15.8; 20.7)* | 18.4 (15.8; 21.4)* | 13.0 (11.0; 14.5)* | 18.0 (16.0; 20.0)* | 11.5 (10.5; 13.0) |
| P10 (95% CI) | 38.3 (33.8; 42.9)* | 37.6 (33.0; 42.2)* | 14.8 (11.6; 18.1)* | 0.4 (0.0; 1.7)* | 45.2 (40.6; 50.0) |
| P30 (95% CI) | 83.7 (80.3; 87.0)* | 80.5 (76.8; 84.2)* | 69.6 (65.0; 73.8)* | 14.8 (11.8; 18.3)* | 97.5 (95.8; 98.8) |
| CCC (95% CI) | 0.845 (0.816; 0.869)* | 0.829 (0.798; 0.855)* | 0.830 (0.804; 0.852)* | 0.575 (0.538; 0.610)* | 0.936 (0.925; 0.946) |
| RMSE (95% CI) | 0.232 (0.211; 0.261) | 0.232 (0.211; 0.260) | 0.156 (0.144; 0.172) | 0.142 (0.132; 0.154) | 0.122 (0.111; 0.134) |
| Agreementa (%) | 61.0 (56.4; 65.7)* | 60.7 (56.1; 65.4)* | 55.7 (51.0; 60.6)* | 7.5 (5.1; 10.0)* | 70.0 (65.7; 74.3)* |
Bias is defined as the median difference between mGFR and eGFR. A positive sign indicates underestimation of mGFR, and a negative sign indicates overestimation of mGFR. P10 and P30 are the proportion of eGFR within 10 and 30% of mGFR, respectively.
aAgreement: correct classification refers to agreement between mGFR and eGFR categories of ≥90, 60–89, 45–59, 30–44 and <30 mL/1.73 m2.
*P < .005 between average clearance creatinine–urea and others eGFR.
DISCUSSION
The present study compared the performance of mClcr, mClUN-cr and CKD-EPI equations (2009 and 2021) using urinary inulin clearance as the reference method and reported the following main findings: (i) satisfactory performance of mClUN-cr compared with mClcr and CKD-EPI equations in the overall cohort; (ii) best performance of CKD-EPI equations when when mGFR is >60 mL/min/1.73 m2; (iii) best performance of mClUN-cr when mGFR is <60 mL/min/1.73 m2; and (iv) similarity between the CKD-EPI equations 2009 and 2021.
Previous studies have already highlighted the global lack of precision of the eGFR equations [1, 2, 7]. However, only few articles have compared the performance of mClUN-cr and eGFR equations, although urea and PCr measurements are well-established laboratory tests [12, 20]. The mClUN-crhas been used in studies on the contribution of residual GFR to adequacy of dialysis treatment as suggested by the European Best Practices Guidelines [21]. This tool is recommended for patients who are in in the late stages of renal failure. Urea clearance significantly underestimates the GFR. Since the mClcr overestimates this function, one alternative could be to average both the PCr and urea clearances that may give a more accurate eGFR than either clearance by itself [5, 20, 24]. Our study is the first to show a superior performance of mClUN-cr over CKD-EPI equations and mClcr when compared with the clearance of inulin in patient with moderate and advanced CKD (mean GFR 71.3 mL/min/1.73 m2). On the contrary, the utility of mClUN-cr was not demonstrated by Thye et al. in 81 subjects with preserved renal function (mean mGFR 123.66 mL/min/1.73 m2 ) [25].
The 2009 CKD-EPI equation is reasonably accurate for following changes in GFR over time, and is recommended for estimating GFR in adults of any age in North America, Europe and Australia [1, 2, 5, 6]. It was developed in a North American and European population with a wide age range and a mean mGFR of 68.0 mL/min/1.73 m2 [6]. Our analyses showed that the CKD-EPI equation has a good performance (P30 = 82.0%) in our population and the recently published 2021 CKD-EPI equation, without ethnicity, was not statistically or clinically superior to the 2009 CKD-EPI equation (P30 77.2%), although ethnicity was not available in our data.
The CKD-EPI equations (2009 and 2021), as previously shown, are less biased and more accurate than mClcr in the study population. In fact, the mClcr alone is an inaccurate method for the measurement of GFR because it overestimates the GFR as a consequence of the PCr tubular secretion, notably in CKD class 4–5 [1, 3, 11, 13–15, 19]. Also, there are two major errors that can limit the accuracy of the mClcr: an inaccurate urine collection and increasing creatinine secretion. An alternative to increase the reliability of the urine sample would be to estimate the completeness of the collection, monitoring the creatinine excretion by weight. In adults, daily creatinine excretion should be 20–25 mg/kg lean body weight in men and 15–20 mg/kg lean body weight in women [26]. Kim et al. showed that with progressive decline of GFR the mClcr tends to become proportionately higher than the corresponding inulin clearance [27]. Zhang et al. propose that measurement error provides an alternative explanation for the ratio of mClcr/mGFR gets larger at lower mGFR levels [28].
The mCIUN alone is rarely used for kidney function because it underestimates the mGFR [13–16]. A high protein diet, tissue breakdown, major gastrointestinal haemorrhage and corticosteroid therapy can lead to an increase in the urea whereas a low protein diet and liver disease can lead to a reduction. A 40%–50% of filtered urea may be reabsorbed by the tubules, while the proportion is reduced in advanced renal failure [11, 13–15, 17, 18, 24]. When kidney function declines, the degree of this underestimation decreases and the mClUN approximates the mGFR. The explanation for this is that during the course of renal disease, the number of functioning nephrons declines. The capacity to reabsorb water is impaired and the fraction of urea reabsorbed also diminishes [11, 13, 15, 16]. Besides, the technique reliability also depends on proper urine collection.
Although the study showed that the mClUN-cr method could play a role in estimating GFR in late stages of kidney disease, there is still a limitation of urinary collection. Alternatively, some authors have described shorter periods of urinary collection with acceptable performance to estimate GFR. Okuda et al. evaluated the usefulness of the mClUN-cr examined over a 1-h urine collection period in children, compared with urinary inulin clearance [29]. In addition, Uemura et al. compared urinary inulin clearance with 2-h and 24-h urine collection mClcr. They found that 24-h mClcr was approximately 80% of 2-h mClcr, and overestimated GFR approximately 1.3-fold, while the 2-h mClcr overestimated GFR approximately 1.5-fold [30].
The strengths of the study were: (i) the wide ranges in age (18–83 years) and GFR levels (10–168 mL/min/1.73 m2); (ii) the IDMS standardization of PCr measurements according to international recommendations; (iii) the performance of the equations compared with a gold standard (inulin) for GFR measurement; and (iv) the use of rigorous statistical techniques based on a large population. The limitations were: (i) the study populations included few non-Caucasian patients and thus it could not assess the effect of ethnicity for 2009 CKD-EPI equation; and (ii) the data collection was retrospective and the mClUN-cr versus mGFR should be clarified in a prospective study.
This study shows that there is a role for the mClUN-cr in evaluating advanced CKD when the nephrologist wants more precision than the eGFR equations (CKD-EPI) and the reference measurement of GFR is not available. From another perspective, the use of mClUN-cr requires no additional laboratory costs (e.g. exogenous marker) and is easy to use, and so is an interesting option especially in developing countries.
Contributor Information
Luciano Selistre, Hospices Civils de Lyon, Hôpital Edouard Herriot, Service de Néphrologie Dialyse, Hypertension et Exploration Fonctionnelle Rénale, Lyon, France; CAPES, Coordenação de Aperfeiçoamento do Pessoal de Nível Superior, Brasilia, Brazil; Universidade de Caxias do Sul, Programa de Pós-graduação em Ciências da Saúde, Caxias do Sul, Brazil; Hospital Geral de Caxias do Sul, Caxias do Sul, Brazil.
Vandréa de Souza, Universidade de Caxias do Sul, Programa de Pós-graduação em Ciências da Saúde, Caxias do Sul, Brazil; Hospital Geral de Caxias do Sul, Caxias do Sul, Brazil.
Carla Nicola, Universidade de Caxias do Sul, Programa de Pós-graduação em Ciências da Saúde, Caxias do Sul, Brazil.
Laurent Juillard, Hospices Civils de Lyon, Hôpital Edouard Herriot, Service de Néphrologie Dialyse, Hypertension et Exploration Fonctionnelle Rénale, Lyon, France; Inserm UMR-1060, Laboratoire CarMeN, Université Lyon-1, Lyon, France.
Sandrine Lemoine, Hospices Civils de Lyon, Hôpital Edouard Herriot, Service de Néphrologie Dialyse, Hypertension et Exploration Fonctionnelle Rénale, Lyon, France; Inserm UMR-1060, Laboratoire CarMeN, Université Lyon-1, Lyon, France; Laboratoire de Biochimie et Biologie Moléculaire, Groupement Hospitalier Edouard Herriot, Lyon, France.
Laurence Derain-Dubourg, Hospices Civils de Lyon, Hôpital Edouard Herriot, Service de Néphrologie Dialyse, Hypertension et Exploration Fonctionnelle Rénale, Lyon, France; Laboratory of Tissue Biology and Therapeutic Engineering, CNRS UMR 5305, Lyon, France; Centre de Référence des Maladies Rénales Rares, Service de Néphrologie et Rhumatologie Pédiatriques, Hospices Civils de Lyon, Lyon, France.
AUTHORS’ CONTRIBUTIONS
L.S. designed the study, and performed data analysis and statistical considerations. L.D.-D. designed the study, participated in the recruitment of patients and contributed data interpretation. V.S. contributed to conceptualization and design of the study, and wrote the first draft and the final version of the manuscript, which was critically read by L.D.-D., C.N. L.J. and S.L. participated in the recruitment of patients. C.N. evaluated data and critically supervised the final version of the manuscript. All authors revised and approved the final version of the manuscript.
FUNDING
The study had no specific public or private financial support. During the period of analysis, L.S. benefited from a grant from the Brazilian government (CAPES Foundation, Ministry of Education of Brazil, grant number: 88881.156638/2017-01).
DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available on request from the corresponding author.
CONFLICT OF INTEREST STATEMENT
L.J. reports consultancy agreements with AstraZeneca, Baxter, Vifor, Fresenius, Hemotech and Sanofi. All remaining authors have nothing to disclose.
REFERENCES
- 1. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int Suppl 2013;3:1–150. 10.1038/kisup.2012.73. [DOI] [PubMed] [Google Scholar]
- 2. Levey AS, Inker LA, Coresh J.. GFR estimation: from physiology to public health. Am J Kidney Dis 2014;63:820–34. 10.1053/j.ajkd.2013.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Soveri I, Berg UB, Björk Jet al. Measuring GFR: a systematic review. Am J Kidney Dis 2014;64:411–24. 10.1053/j.ajkd.2014.04.010. [DOI] [PubMed] [Google Scholar]
- 4. Boele-Schutte E, Gansevoort RT.. Measured GFR: not a gold, but a gold-plated standard. Nephrol Dial Transplant 2017;32:ii180–4. 10.1093/ndt/gfw441. [DOI] [PubMed] [Google Scholar]
- 5. Evans M, van Stralen KJ, Schön Set al. Glomerular filtration rate-estimating equations for patients with advanced chronic kidney disease. Nephrol Dial Transplant 2013;28:2518–26. 10.1093/ndt/gft226. [DOI] [PubMed] [Google Scholar]
- 6. Levey AS, Stevens LA, Schmid CHet al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604–12. 10.7326/0003-4819-150-9-200905050-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Stevens LA, Zhang YP, Schmid CH.. Evaluating the performance of equations for estimating glomerular filtration rate. J Nephrol 2008;21:797–807. [PMC free article] [PubMed] [Google Scholar]
- 8. Inker LA, Eneanya ND, Coresh Jet al. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med 2021;385:1737–49. 10.1056/NEJMoa2102953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Tynkevich E, Flamant M, Haymann JPet al. Decrease in urinary creatinine excretion in early stage chronic kidney disease. PLoS One 2014;9:e111949. 10.1371/journal.pone.0111949. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Tynkevich E, Flamant M, Haymann JPet al. Urinary creatinine excretion, measured glomerular filtration rate and CKD outcomes. Nephrol Dial Transplant 2015;30:1386–94. 10.1093/ndt/gfv047. [DOI] [PubMed] [Google Scholar]
- 11. Emerson K, Futcher PH, Farr LE. The relation of high and low urea clearances to the inulin and creatinine clearances in children with the nephrotic syndrome. J Clin Invest 1941;20:361–8. 10.1172/JCI101230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Caravaca F, Arrobas M, Luna Eet al. [Differences between the glomerular filtration rate estimated by the MDRD equation and the measurement of creatinine and urea clearance in unselected patients with terminal renal insufficiency]. Nefrologia 2002;22:432–7. [PubMed] [Google Scholar]
- 13. Lubowitz H, Slatopolsky E, Shankel Set al. Glomerular filtration rate. Determination in patients with chronic renal disease. JAMA 1967;199:252–6. 10.1001/jama.199.4.252. [DOI] [PubMed] [Google Scholar]
- 14. Dean RF, McCance RA.. Inulin, diodone, creatinine and urea clearances in newborn infants. J Physiol 1947;106:431–9. 10.1113/jphysiol.1947.sp004224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Lavender S, Hilton PJ, Jones NF.. The measurement of glomerular filtration-rate in renal disease. Lancet 1969;2:1216–8. 10.1016/s0140-6736(69)90752-1. [DOI] [PubMed] [Google Scholar]
- 16. Brouhard BH, Travis LB, Cunningham RJet al. Simultaneous iothalamate, creatinine, and urea clearances in children with renal disease. Pediatrics 1977;59:219–23. 10.1542/peds.59.2.219. [DOI] [PubMed] [Google Scholar]
- 17. Winkler AW, Parra J. The measurement of glomerular filtration. The creatinine, sucrose and urea clearances in subjects with renal disease. J Clin Invest 1937;16:869–77. 10.1172/JCI100913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Coulthard MG, Hey EN, Ruddock V.. Creatinine and urea clearances compared to inulin clearance in preterm and mature babies. Early Hum Dev 1985;11:11–9. 10.1016/0378-3782(85)90114-8. [DOI] [PubMed] [Google Scholar]
- 19. Almond A, Siddiqui S, Robertson Set al. Comparison of combined urea and creatinine clearance and prediction equations as measures of residual renal function when GFR is low. QJM 2008;101:619–24. 10.1093/qjmed/hcn032. [DOI] [PubMed] [Google Scholar]
- 20. Douma CE, Smit W.. When to start dialysis? Nephrol Dial Transplant 2006;21 Suppl 2:ii20–4. 10.1093/ndt/gfl139. [DOI] [PubMed] [Google Scholar]
- 21. Dombros N, Dratwa M, Feriani Met al. European best practice guidelines for peritoneal dialysis. 2 The initiation of dialysis. Nephrol Dial Transplant 2005;20 Suppl 9:ix3–7. 10.1093/ndt/gfi1116. [DOI] [PubMed] [Google Scholar]
- 22. Dubois D, Dubois EF.. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med 1916;17:863–71. 10.1001/archinte.1916.00080130010002. [DOI] [Google Scholar]
- 23. Dubourg L, Hadj-Aissa A, Ferrier B.. Adaptation of an enzymatic polyfructosan assay to clinical practice. Anal Biochem 2010;405:266–8. 10.1016/j.ab.2010.06.016. [DOI] [PubMed] [Google Scholar]
- 24. Traynor J, Mactier R, Geddes CCet al. How to measure renal function in clinical practice. BMJ 2006;333:733–7. 10.1136/bmj.38975.390370.7C. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Thye CK, Lee YW, Jalalonmuhali Met al. The utility of combined urine urea and creatinine clearance for pretransplant renal function assessment. Transplant Proc 2022;54:254–9. 10.1016/j.transproceed.2021.12.023. [DOI] [PubMed] [Google Scholar]
- 26. Walser M. Creatinine excretion as a measure of protein nutrition in adults of varying age. JPEN J Parenter Enteral Nutr 1987;11:73S–8S. 10.1177/014860718701100510. [DOI] [PubMed] [Google Scholar]
- 27. Kim KE, Onesti G, Ramirez Oet al. Creatinine clearance in renal disease. A reappraisal. Br Med J 1969;4:11–4. 10.1136/bmj.4.5674.11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Zhang X, McCulloch CE, Lin Fet al. Measurement error as alternative explanation for the observation that CrCl/GFR ratio is higher at lower GFR. Clin J Am Soc Nephrol 2016;11:1574–81. 10.2215/CJN.12821215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Okuda Y, Hamada R, Uemura Oet al. Mean of creatinine clearance and urea clearance examined over 1 h estimates glomerular filtration rate accurately and precisely in children. Nephrology (Carlton) 2021;26:763–71. 10.1111/nep.13911. [DOI] [PubMed] [Google Scholar]
- 30. Uemura O, Nagai T, Yamakawa Set al. Assessment of kidney function in children by enzymatic determination of 2- or 24-h creatinine clearance: comparison with inulin clearance. Clin Exp Nephrol 2016;20:462–8. 10.1007/s10157-015-1166-3. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that supports the findings of this study are available on request from the corresponding author.