J Nephrol (2015) 28:59–66 DOI 10.1007/s40620-014-0148-y
ORIGINAL ARTICLE
Is there an age cutoff to apply adult formulas for GFR estimation in children? Antonio Azzi • Francois Cachat • Mohamed Faouzi Dolores Mosig • Pascal Ramseyer • Eric Girardin • Hassib Chehade
•
Received: 10 July 2014 / Accepted: 29 September 2014 / Published online: 7 October 2014 Italian Society of Nephrology 2014
Abstract Background Estimation of glomerular filtration rate (eGFR) using a common formula for both adult and pediatric populations is challenging. Using inulin clearances (iGFRs), this study aims to investigate the existence of a precise age cutoff beyond which the Modification of Diet in Renal Disease (MDRD), the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI), or the Cockroft– Gault (CG) formulas, can be applied with acceptable precision. Performance of the new Schwartz formula according to age is also evaluated.
A. Azzi F. Cachat D. Mosig E. Girardin H. Chehade (&) Division of Pediatric Nephrology, Department of Pediatrics, Lausanne University Hospital, Rue Bugnon 46, 1011 Lausanne, Switzerland e-mail: [email protected] A. Azzi e-mail: [email protected] F. Cachat e-mail: [email protected] D. Mosig e-mail: [email protected] E. Girardin e-mail: [email protected] M. Faouzi Institute of Social and Preventive Medicine, Route de la Corniche 10, 1010 Lausanne, Switzerland e-mail: [email protected] P. Ramseyer Division of Pediatric Urology, Department of Pediatrics, Lausanne University Hospital, Rue Bugnon 46, 1011 Lausanne, Switzerland e-mail: [email protected]
Method We compared 503 iGFRs for 503 children aged between 33 months and 18 years to eGFRs. To define the most precise age cutoff value for each formula, a circular binary segmentation method analyzing the formulas’ bias values according to the children’s ages was performed. Bias was defined by the difference between iGFRs and eGFRs. To validate the identified cutoff, 30 % accuracy was calculated. Results For MDRD, CKD–EPI and CG, the best age cutoff was C14.3, C14.2 and B10.8 years, respectively. The lowest mean bias and highest accuracy were -17.11 and 64.7 % for MDRD, 27.4 and 51 % for CKD–EPI, and 8.31 and 77.2 % for CG. The Schwartz formula showed the best performance below the age of 10.9 years. Conclusion For the MDRD and CKD–EPI formulas, the mean bias values decreased with increasing child age and these formulas were more accurate beyond an age cutoff of 14.3 and 14.2 years, respectively. For the CG and Schwartz formulas, the lowest mean bias values and the best accuracies were below an age cutoff of 10.8 and 10.9 years, respectively. Nevertheless, the accuracies of the formulas were still below the National Kidney Foundation Kidney Disease Outcomes Quality Initiative target to be validated in these age groups and, therefore, none of these formulas can be used to estimate GFR in children and adolescent populations. Keywords Child Chronic kidney disease CKD–EPI formula Cockroft–Gault formula Glomerular filtration rate MDRD formula Schwartz formula
Introduction Several formulas have been developed to estimate glomerular filtration rate (eGFR) in both adults and children
123
60
based on serum creatinine (SCreat) or cystatin C (CysC), with varying degrees of precision. The Schwartz formula [1, 2] is the most widely used formula for eGFR in the pediatric population. The Modification of Diet in Renal Disease (MDRD) [3], the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI) [4] and the Cockroft– Gault formulas (CG) [5] are the most popular formulas for eGFR in adults. These adult formulas have proved to be reliable in adult populations, but their application in the pediatric population often provides misleading and conflicting results. Chehade et al. [6] found that the application of these adult formulas is not reliable in children under 18 years of age across all chronic kidney disease (CKD) stages. Pierrat et al. [7] found that the CG formula corrected for body surface area gave an accurate estimation of the GFR in children aged 12 years and above, whereas, the MDRD led to an overestimation of GFR. However, Filler et al. [8] concluded that the CG formula performs poorly in children above 12 years of age. Selistre et al. [9] also cautioned against the use of these formulas in children. All these studies examined the possibility of applying adult formulas in children and adolescents, with different cohorts selected according to the CKD stage [6] or according to a preset arbitrary cutoff age [8, 9]. Instead of choosing an arbitrary age cutoff, which can be misleading, we here used a robust statistical method, the circular binary segmentation (CBS) method, to investigate an age cutoff in children, above which the MDRD, CKD–EPI or CG formulas present good enough accuracy to be usable in a pediatric population. Performance of the new Schwartz formula according to age was also evaluated.
Materials and methods Population After approval by the Lausanne University Hospital ethics board (Local ethical committee approval number 370/12), we compared 503 inulin clearances (iGFRs) for 503 children aged between 33 months and 18 years with their eGFRs using the MDRD, CKD–EPI, CG, and Schwartz formulas. iGFR measurement was performed in our Pediatric Nephrology department between December 2009 and December 2012. The characteristics and classification of patients’ renal disorders are summarized in Table 1. Children with bladder dysfunction, unable to void spontaneously, or in whom bladder catheterization failed, were excluded. During inulin clearances, proper emptying of the bladder was evaluated by comparing the urine output with its osmolality. A decreasing diuresis with a concomitant decreasing urine osmolality was an indication of poor
123
J Nephrol (2015) 28:59–66 Table 1 Characteristics and classification of patients’ renal disorders according to CKD stages Etiologies
CKD stage 1
Obstructive or reflux uropathy
179
Congenital and acquired single kidney
CKD stage 3
CKD stages 4 and 5
82
36
10
25
15
9
1
Polycystic kidney disease
10
8
2
0
Glomerulopathies
5
4
1
0
Other Total = 503
CKD stage 2
64
36
14
2
283
145
62
13
CKD stages 1, 2, 3, 4, and 5 denote inulin glomerular filtration rate [90, 60–89, 30–59, 15–29, and \15 ml/min/1.73 m2, respectively CKD chronic kidney disease
bladder emptying, and the child was then excluded from the study for technical reasons. Analytical methods iGFR measurement All patients fasted overnight before the day of investigation, and drugs interfering with the inulin measurement were omitted before and during the test. iGFR was obtained as follows: two intravenous catheters were inserted, one in each arm and a loading dose of 25 % inulin was administered according to the manufacturer’s protocol (Inutest SPC, Fresenius Kabi Pharma Austria GmbH, Austria). The loading dose was calculated in order to obtain a required plasma concentration of 200–250 mg/l, as follows: loading inulin dose = required plasma concentration 9 estimated inulin distribution volume. The estimated inulin distribution volume corresponds to the extracellular volume and amounts to 25 % of the body weight (BW). Subsequently, inulin was continuously infused over 90 min at a rate given by the required inulin plasma concentration (200–250 mg/l) and the eGFR as follows: infusion rate in mg/min = (required plasma concentration/1,000) 9 eGFR in ml/min. Water diuresis was induced by oral administration of 20 ml/kg of water (maximum 1,200 ml) in the first hour followed by 3 ml/kg/h of water. This was combined with an intravenous infusion of 0.9 % sodium chloride (maximum 300 ml) every 30 min. After a 90-min equilibration period, three timed-urine samples were collected every 30 min, according to the manufacturer’s protocol, with a blood test in the middle of each urine collection. Inulin was measured using the automatic anthrone method of Wright and Gann, with an Autoanalyzer 3 system (high resolution
J Nephrol (2015) 28:59–66
digital colorimeter of SEAL; Bran ? Luebbe, Norderstedt, Germany). The method was calibrated with five point calibrations (target values at 10 mg/100 ml, 20 mg/100 ml, 30 mg/100 ml, 40 mg/100 ml, 50 mg/100 ml) with correlation coefficients of 0.9993 ± 0.0005. The procedure is automatized with software (AACE 6.03, Bran ? Luebbe) which automatically includes corrections for baseline, carryover, sensitivity drift and dilution factor. For the serum, the intra-assay coefficients of variation (CVs) obtained in our laboratory with the internal quality controls were 2.44 % at 10 mg/100 ml, 1.47 % at 30 mg/100 ml and 0.94 % at 40 mg/100 ml, while for urine the intraassay CVs were 1.71 % at 10 mg/100 ml, 1.22 % at 30 mg/100 ml and 1.07 % at 50 mg/100 ml. The interassay CVs obtained in the internal laboratory standards were 2.35 % at 10 mg/100 ml, 2.23 % at 30 mg/100 ml and 0.87 % at 50 mg/100 ml. iGFR was calculated as the mean of the three clearance periods. When the inulin clearance difference between two periods exceeded 20 %, that period was excluded, and iGFR was calculated as the mean of the two valid periods. SCreat measurement Blood sampled on lithium heparinate was processed immediately for SCreat determination. SCreat was measured using the kinetic colorimetric compensated Jaffe´ method which was standardized against the reference isotope dilution mass spectrometry method (IDMS) and was calibrated with the calibrator and procedures as reported by the manufacturer Roche Modular P system (Roche Diagnostics, Mannheim, Germany). The IDMS traceability involved two-point calibration (target values at 0 and 360–390 lmol/l depending on the calibrator lot) and the subtraction of 26 lmol/l from the results, in order to compensate for the non-specific chromogens. The interassay CVs obtained in the laboratory with the internal quality controls are 3.9 % at 45.7 lmol/l and 2.4 % at 108 lmol/l. The intra-assay CVs are 3.3 % at 44.5 lmol/l and 0.7 % at 148 lmol/l. GFR estimation eGFRs were calculated using the MDRD, CKD–EPI, CG, and Schwartz formulas as follows: •
•
For the MDRD formula [3]: eGFR = 175 9 (SCreat)-1.154 9 (age)-0.203 9 [0.742 if female] 9 [1.212 if black], where SCreat is measured by an IDMScalibrated assay and is expressed in mg/dl, age in years. For the CKD–EPI formula [4]: eGFR = a 9 (SCreat/ b)c 9 (0.993)age. The variable (a) depends on race and sex: black women a = 166; black men a = 163; white/
61
other women a = 144; white/other men a = 141. The variable (b) varies on the basis of sex: women b = 0.7; men b = 0.9. The variable (c) takes on the following values on the basis of sex and SCreat measurement: for women, if SCreat B0.7 mg/dl = -0.329; if SCreat [0.7 mg/dl = -1.209; for men, if SCreat B0.9 mg/ dl = -0.411; if SCreat [0.9 mg/dl = -1.209. • For the CG formula [5]: eGFR = [(140 - age) 9 BW/ SCreat] 9 [K = 1.23 if male or 1.04 if female], where age is in years, BW in kg and SCreat in lmol/l. The CG equation estimates absolute GFR (i.e. in ml/min) rather than standardized GFR (i.e. in ml/min/1.73 m2). The latter was hereby used in both the iGFR measurement and the MDRD and CKD–EPI equations. For expressing CG as standardized GFR, CG estimates are multiplied by 1.73 and divided by the body surface area (BSA). In addition, the CG formula was established using the old SCreat measurement method which is the Jaffe´ technique. In this study, we used the compensated Jaffe´ technique, adjusted to the IDMS method. In order to avoid any bias due to using the CG published constants (1.23 for males of any age, and 1.05 for females of any age), which can lead to an inaccurate estimation of GFR, we calculated new constants based on the iGFR for each patient and then calculated the average. The recalculated average constants were 0.95 ± 1.05 and 1.1 ± 0.33 for female and male children, respectively. We then compared the eGFRs from both the CG formula using the published constants and the modified CG formula using the newly established constants, to the iGFRs. • For the Schwartz formula [2]: eGFR = 0.413 9 (height/SCreat), where height is expressed in cm, and SCreat in mg/dl. Statistical analysis Collected data were analyzed using StataCorp. 2011 (Stata Statistical Software: Release 12; StataCorp LP, College Station, TX, USA). Population characteristics (age, weight and height) are summarized as mean and standard deviation (SD), categorical variables are expressed as percentages. To determine the different change points where each adult formula overestimates or underestimates iGFR, we applied the circular binary segmentation (CBS) method [10] for all the bias values of each adult formula according to the child’s age. Bias was defined as the difference between iGFR and eGFR values. Successive change points delimited the age interval (segment) of over/under estimation. Moreover, to define the most accurate age cutoff value of each formula, we calculated for each formula the
123
62
J Nephrol (2015) 28:59–66
accuracy of each segment. Accuracy at 30 % was defined as the percentage of eGFRs values that were within 30 % of the iGFRs. The CBS method allows to segment data through change-point detection using a maximal t test as follows: let R1, R2,… Rn be the sequence of the bias indexed according to the values of the age (age in increasing order) and let Si = R1 ? R2 ?… Ri, 1 B i B n, be the partial sums. When the biases are normally distributed with known variance, the likelihood ratio statistic for testing the null hypothesis that there is no change against the alternative that there is exactly one change at an unknown location I is given by ZB = max1 B i \ n |Zi|, where Zi = {1/i ? 1/(n - i)} - 0.5 {Si/I - (Sn - Si)/(n - i)} [11]. The null hypothesis of no change is rejected if the statistic exceeds the upper ath quantile of the null distribution of ZB and the location of the change-point is estimated to be i such that ZB = |Zi|.
Results Mean ± SD for patients’ age, BW, height and GFRs according to each studied formula are reported in Table 2. Patients’ age ranged between 33 months and 18 years: 42.14 % were girls. Mean ± SD of iGFRs was 91.69 ± 21 ml/min/1.73 m2, ranging from 14 to 191 ml/min/ 1.73 m2. Children with moderate and severe chronic kidney disease represented 44 % of the total study population. When applying the CBS method for all bias values from the MDRD formula, 4 change points were identified at ages Table 2 Patients’ characteristics and mean ± SD values for eGFRs using the MDRD, the CKD–EPI, the Cockroft–Gault, the modified Cockroft–Gault, and the Schwartz formulas Patient characteristics and eGFR values
Mean ± SD
Patients’ age (years)
11.95 ± 4.09
Patients’ body weight (in kg)
43.64 ± 18.69
Patients’ height (cm) eGFR using the MDRD formula (ml/min/1.73 m2)
147.9 ± 22.08 166.04 ± 88.64
eGFR using the CKD–EPI formula (ml/min/ 1.73 m2)
139.60 ± 33.54
eGFR using the Cockroft–Gault formula (ml/min/ 1.73 m2)
103.87 ± 41.56
eGFR using the modified Cockroft–Gault formula (ml/min/1.73 m2)
98.07 ± 38.26
eGFR using the Schwartz formula (ml/min/ 1.73 m2)
95.42 ± 35.70
eGFR estimated glomerular filtration rate, MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration
123
3.3, 6.6, 10.2, and 14.3 years. This result identified 5 age segments: [2.75–3.3[, [3.3–6.6[, [6.6–10.2[, [10.2–14.3[ and [14.3–18] years (Fig. 1a). The mean bias of each segment was, -61.63, -194.35, -113.25, -59.57 and -17.11 ml/min/1.73 m2, respectively. For each segment we calculated the accuracy at 30 %. Only 25, 0, 3.1, 16.7 and 64.7 % of eGFR values were accurate in the five segments, respectively (Table 3). These results show that the MDRD overestimated the GFR in all age groups (Fig. 2) with an accuracy that increased and reached its best value for the age group between 14.3 and 18 years, with a lowest mean difference of 17.11 ml/min/1.73 m2 as compared to iGFR. When using the CKD–EPI formula, the CBS method applied for all bias values detected three change points corresponding to 3.4, 7.5, and 14.2 years. Therefore four segments were identified (Fig. 1b). The first segment, corresponding to children aged between 33 months and 3.3 years, presented a mean bias value of -43.50 ml/min/ 1.73 m2, and for an accuracy of 30 only 25 % of eGFR values were accurate. The second segment, corresponding to children aged between 3.3 and 7.5 years, presented a mean bias value of -77.53 ml/min/1.73 m2 and an accuracy of 1.1 %. The third segment, corresponding to children aged between 7.5 and 14.2 years, presented a mean bias value of -55.14 ml/min/1.73 m2 and an accuracy of 8.49 % (Table 3). The CKD–EPI formula overestimated the GFR at all age groups (Fig. 2); it presented the highest accuracy of 51 % and the lowest mean difference of -27.4 ml/min/1.73 m2 when compared to iGFRs for the age group between 14.2 and 18 years (segment 4). When applying the CBS method for all bias values from the CG formula, a single change point was detected corresponding to age 10.8 years. We therefore identified two segments: 2.75–10.8 years, and 10.8–18 years (Fig. 1c). The mean bias of each segment was 8.31 and -21.33 ml/min/1.73 m2, respectively. For an accuracy of 30, 77.2 and 63.3 % of eGFR values in each segment were accurate respectively (Table 3). These results show that the CG formula underestimated GFR in pediatric patients under 10.8 years of age and overestimated GFR in children above 10.8 years of age (Fig. 2), with a better accuracy and lowest mean difference when compared to iGFR for the first age group. Regarding the modified CG formula, when applying the CBS method for all bias values we found, similarly to the CG formula, one change point corresponding to 10.8 years of age. We also identified two segments, 2.75–10.8 and 10.8–18 years (Fig. 1d). The first segment presented a mean bias value of 13.35 ml/min/1.73 m2 and an accuracy of 74.6 %. The second segment presented a mean bias value of -15.55 ml/min/1.73 m2 and an accuracy of 70.7 %. Similarly to the CG formula, the modified CG
J Nephrol (2015) 28:59–66
63
Fig. 1 Circular binary segmentation (CBS) method for bias values analysis according to age. a Age change points detection using the MDRD formula. b Age change points detection using the CKD–EPI formula. c, d Age change point detection using the CG and the
modified CG formulas, respectively. e Age change point detection using the Schwartz formula. MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration, CG Cockroft–Gault
formula underestimated GFR in pediatric patients below the age of 10.8 years and overestimated GFR for children above this age (Fig. 2), with a better accuracy and a lower mean difference when compared to iGFR for the first age group (Table 3). Regarding the Schwartz formula, when applying the CBS method for all bias values we found, similarly to the CG and the modified CG formulas, one change point corresponding to 10.9 years of age. We also identified 2 segments, 2.75–10.9 and 10.9–18 years (Fig. 1e). The first segment presented a mean bias value of 10.93 ml/min/ 1.73 m2 and an accuracy of 66.2 %. The second segment presented mean bias value of -12.26 ml/min/1.73 m2 and an accuracy of 61.1 % (Table 3).
Discussion The exact determination of GFR is important, especially in patients with renal failure, after transplantation or receiving potential nephrotoxic drugs. The three most commonly used formulas for eGFR in adults are the MDRD, the CKD–EPI, and the CG. The CG formula was developed in 1976 in a white male population with CKD [5]. It was then validated in overweight and obese diabetic populations with preserved renal function [12] and in adult patients under 65 years of age [13], and was approved in 1998 by the US Food and Drug Administration to guide drug dosage for patients with decreased kidney function without being adapted for sex and race. The MDRD formula was
123
64
J Nephrol (2015) 28:59–66
Table 3 Mean bias and accuracy of each age interval which is delimited by the successive change point of over/under estimation of iGFR obtained by applying the CBS method for all bias values from the MDRD, CKD–EPI, Cockroft–Gault, modified Cockroft–Gault, and Schwartz formulas Formulas
Age interval (years) identified by applying the CBS method for formula bias values
MDRD formula
[2.75–3.3[
-61.63
25
[3.3–6.6[
-194.35
0
[6.6–10.2[
-113.25
3.1
[10.2–14.3[
-59.57
16.7
[14.3–18] [2.75–3.3[
-17.11 -43.50
64.7 25
[3.3–7.5[
-77.53
1.1
[7.5–14.2[
-55.14
8.49
[14.2–18]
-27.4
51
CKD–EPI formula
Cockroft– Gault formula
[2.75–10.8]
Modified Cockroft– Gault formula
[2.75–10.8]
Schwartz formula
Mean bias (ml/ min/1.73 m2) of the obtained age interval
Accuracy (%) within 30 % of the obtained age interval
8.31
77.2
-21.33
63.3
13.35
74.6
-15.55
70.7
[2.75–10.9]
10.93
66.2
]10.9–18]
12.26
61.1
]10.8–18]
]10.8–18]
iGFR inulin glomerular filtration rate, eGFR estimated glomerular filtration rate, MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration, CBS circular binary segmentation, Bias difference between iGFR and eGFR values, Accuracy percentage of eGFR values within 30 % of the iGFR values
published in 1999 [3] as a six-variable equation and then subsequently simplified in 2000 to a four-variable equation including SCreat, age, race, and sex [14]. This formula was developed in a group of adult patients with a mean age of 51 years and mean GFR of 40 ml/min/1.73 m2 and was subsequently demonstrated to be more accurate for patients with stage 3 CKD or higher [15, 16]. Of note, the MDRD formula has been validated in the African American adult population [17], in adults with diabetic nephropathy [18] or scleroderma [19], and finally also in healthy adults without renal disease [20, 21]. The National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKFKDOQI) recommends its usage for eGFR in adult patients at risk of kidney diseases. The CKD–EPI formula was published in 2009 [4]. It was developed in a group of adults with a mean age of 47 years and a mean GFR of 67 ml/ min/1.73 m2 in order to overcome the limitations of the MDRD formula. It was also externally validated in various groups such as patients with advanced chronic renal failure
123
Fig. 2 Plot for correlation between bias (difference between iGFR and eGFR) and age, using the MDRD formula (blue curve), the CKD– EPI formula (red curve), the CG formula (green curve), the modified CG formula (yellow curve), and the Schwartz formula (grey curve). iGFR inulin glomerular filtration rate, eGFR estimated glomerular filtration rate. For other abbreviations see Fig. 1
[16] or renal transplant patients [22]. The Schwartz formula was developed in the mid-1970s [1], and revised in 2009 [2] to incorporate the change of serum creatinine measurement method. This formula has been developed in a population with a measured GFR ranging from 15 to 75 ml/min/1.73 m2, and is not adjusted for age or sex. The applicability of the above adult formulas in children is controversial. Several research groups have assessed the accuracy of these adult formulas in different pediatric populations, and found conflicting results. Most of these studies used arbitrary age limits to categorize their population, therefore potentially missing the point beyond which these formulas might be sufficiently accurate. Using a robust statistical approach (CBS method), without choosing a predefined age cutoff, our results showed that for the MDRD and the CKD–EPI formulas the mean bias values decreased with increasing age, which is not really surprising. The lowest mean of bias associated with the highest accuracy was observed in the adolescent population group, beyond an age cutoff of 14.3 and 14.2 years for the MDRD and the CKD–EPI, respectively. However, even above these cutoff values, neither the MDRD nor the CKD–EPI fulfill the NKF-KDOQI target recommendation of C90 % of eGFR estimates to be within 30 % of measured GFR [23]. For the CG and the modified CG formulas, the lower mean bias values and the best accuracies were obtained below an age cutoff of 10.8 years. Although the accuracy obtained with the CG formula was better than the ones with the MDRD or the CKD–EPI formulas, it was still below the required NKF-KDOQI target recommendation for validation. Furthermore, our results showed that
J Nephrol (2015) 28:59–66
the MDRD and the CKD–EPI formulas overestimate the GFR at all age groups. These results are in line with other previous publications [6–8]. In our opinion, the reason why the adult equations perform so poorly in children is due to the fact that there are other parameters not included in the CKD–EPI, MDRD and CG formulas that significantly affect the formulas’ accuracy and results, making them non applicable in pediatric patient groups. In addition, ‘‘age’’ represents an important variable and is a central component of the adult equations. In adults, muscle mass decreases with time and therefore age is entered as: ‘‘(0.993)age’’ and ‘‘(age)-0.203’’ in the MDRD and CKD– EPI formulas and (140-age) in the CG formula, resulting in lower eGFR in older subjects than expected from the SCreat readings. However, the opposite is true for children, where muscle mass increases with age leading to increasing SCreat values, which are more closely related to height than age (which is the concept underlying the pediatric formulas for eGFR). Hence we are not surprised that the adult formulas for eGFR present a large bias in children across all age groups. The Schwartz formula predicts GFR much better than the MDRD or the CKD– EPI formulas. However, we observed a fairly similar prediction of GFR when using the Schwartz formula and the CG formulas. Variation of GFR prediction with the Schwartz formula has already been shown [24–26], especially in the high GFR range. We here demonstrate that the accuracy of the new Schwartz formula is also below the NKF-KDOQI recommendations for it to be usable in our population. Recently, Gao et al. [27] proposed an improvement of the Schwartz linear model in the prediction of GFR using the variable ratio height/creatinine in a quadratic model and published a new quadratic formula adjusted to age and sex. There are a few limitations to our study. First, we did not evaluate any adult formula including CysC as we did not measure the CysC in our patients. Second, we used the compensated Jaffe´ technique for SCreat measurement. The CG formula was developed using the old Jaffe´ technique for SCreat measurement. Nevertheless, we recalculated the formula‘s constants according to our SCreat measurement method and to the patients’ iGFRs, in order to overcome a possible bias related to the difference in the SCreat method of determination. In conclusion, the MDRD and the CKD–EPI formulas present the lowest bias and the best accuracies in adolescents over the age of 14.3 years. However, the accuracies of these formulas are still below the NKF-KDOQI target to be validated in this age group. The CG formula is more accurate in children below 10.8 years but still presents an accuracy below the NKF-KDOQI target guideline. Therefore, none of these formulas can be used to estimate GFR in children or adolescent populations.
65 Conflict of interest
All the authors declare no competing interests.
References 1. Schwartz GJ, Haycock GB, Edelmann CM et al (1976) A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 58:259–263 2. Schwartz GJ, Mun˜oz A, Schneider MF et al (2009) New equations to estimate GFR in children with CKD. J Am Soc Nephrol 20(3):629–637 3. Levey AS, Coresh J, Greene T et al (2006) Chronic Kidney Disease Epidemiology Collaboration: using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 145(4):247–254 4. Levey AS, Stevens LA, Schmid CH et al (2009) CKD–EPI (Chronic Kidney Disease Epidemiology Collaboration) A new equation to estimate glomerular filtration rate. Ann Intern Med 150:604–612 5. Cockroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41 6. Chehade H, Girardin E, Iglesias K et al (2013) Assessment of adult formulas for glomerular filtration rate estimation in children. Pediatr Nephrol 28:105–114 7. Pierrat A, Gravier E, Saunders C et al (2003) Predicting GFR in children and adults: a comparison between Cockroft–Gault, Schwartz, and MDRD formulas. Kidney Int 64:1425–1436 8. Filler G, Foster J, Acker A (2005) The Cockroft–Gault formula should not be used in children. Kidney Int 67:2321–2324 9. Selistre L, De Souza V, Cochat P et al (2012) GFR estimation in adolescents and young adults. J Am Soc Nephrol 23:989–996 10. Venkatraman ES, Olshen AB (2007) A faster circular binary segmentation algorithm for the analysis of array CGH data. Bioinformatics 23:657–663 11. Sen AK, Srivastava MS (1975) On tests for detecting change in mean. Ann Stat 1:98–108 12. Drion I, Joosten H, Santing L et al (2011) The Cockroft–Gault a better predictor of renal function in an overweight and obese diabetic population. Obes Facts 4:393–399 13. Verhave JC, Fesler P, Ribstein J, du Cailar G, Mimran A (2005) Estimation of renal function in subjects with normal serum creatinine levels: influence of age and body mass index. Am J Kidney Dis 46:233–241 14. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D (1999) 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 Med 130:461–470 15. Vervoort G, Willems HL, Wetzels JF (2002) Assessment of glomerular filtration rate in healthy subjects and normoalbuminuric diabetic patients: validity of a new (MDRD) prediction equation. Nephrol Dial Transplant 17:1909–1913 16. Teruel Briones JL, Gomis Couto A, Sabater J et al (2011) Validation of the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI) equation in advanced chronic renal failure. Nefrologia 31:677–682 17. Lewis J, Agodoa L, Cheek D et al (2001) Comparison of crosssectional renal function measurements in African Americans with hypertensive nephrosclerosis and of primary formulas to estimate glomerular filtration rate. Am J Kidney Dis 38:744–753 18. Poggio ED, Wang X, Greene T, Van Lente F, Hall PM (2005) Performance of the modification of diet in renal disease and Cockroft–Gault equations in the estimation of GFR in health and in chronic kidney disease. J Am Soc Nephrol 16:459–466
123
66 19. Kingdon EJ, Knight CJ, Dustan K et al (2003) Calculated glomerular filtration rate is a useful screening tool to identify scleroderma patients with renal impairment. Rheumatology 42:26–33 20. Kang YS, Han KH, Han SY, Kim HK, Cha DR (2005) Characteristics of population with normal serum creatinine impaired renal function and: the validation of a MDRD formula in a healthy general population. Clin Nephrol 63:258–266 21. Lin J, Knight EL, Hogan ML, Singh AK (2003) A comparison of prediction equations for estimating glomerular filtration rate in adults without kidney disease. J Am Soc Nephrol 14:2573–2580 22. Townamchai N, Praditpornsilpa K, Chawatanarat T et al (2013) The validation of estimated glomerular filtration rate equation for renal transplant recipients. Clin Nephrol 79(3):206–213 23. Levey AS, Coresh J et al (2002) National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease
123
J Nephrol (2015) 28:59–66
24.
25.
26.
27.
evaluation, classification, and stratification. Am J Kidney Dis 39(2 Suppl 1):S1–S266 Bacchetta J, Cochat P, Rognant N et al (2011) Which creatinine and cystatin C equations can be reliably used in children? Clin J Am Soc Nephrol 6:552–560 Staples A, LeBlond R, Watkins S et al (2010) Validation of the revised Schwartz estimating equation in a predominantly nonCKD population. Pediatr Nephrol 25:2321–2326 Berg UB, Ba¨ck R, Celsi G et al (2011) Comparison of plasma clearance of iohexol and urinary clearance of inulin for measurement of GFR in children. Am J Kidney Dis 57:55–61 Gao A, Cachat F, Faouzi M et al (2013) Comparison of the glomerular filtration rate in children by the new revised Schwartz formula and a new generalized formula. Kidney Int 83(3): 524–530
ORIGINAL ARTICLE
Is there an age cutoff to apply adult formulas for GFR estimation in children? Antonio Azzi • Francois Cachat • Mohamed Faouzi Dolores Mosig • Pascal Ramseyer • Eric Girardin • Hassib Chehade
•
Received: 10 July 2014 / Accepted: 29 September 2014 / Published online: 7 October 2014 Italian Society of Nephrology 2014
Abstract Background Estimation of glomerular filtration rate (eGFR) using a common formula for both adult and pediatric populations is challenging. Using inulin clearances (iGFRs), this study aims to investigate the existence of a precise age cutoff beyond which the Modification of Diet in Renal Disease (MDRD), the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI), or the Cockroft– Gault (CG) formulas, can be applied with acceptable precision. Performance of the new Schwartz formula according to age is also evaluated.
A. Azzi F. Cachat D. Mosig E. Girardin H. Chehade (&) Division of Pediatric Nephrology, Department of Pediatrics, Lausanne University Hospital, Rue Bugnon 46, 1011 Lausanne, Switzerland e-mail: [email protected] A. Azzi e-mail: [email protected] F. Cachat e-mail: [email protected] D. Mosig e-mail: [email protected] E. Girardin e-mail: [email protected] M. Faouzi Institute of Social and Preventive Medicine, Route de la Corniche 10, 1010 Lausanne, Switzerland e-mail: [email protected] P. Ramseyer Division of Pediatric Urology, Department of Pediatrics, Lausanne University Hospital, Rue Bugnon 46, 1011 Lausanne, Switzerland e-mail: [email protected]
Method We compared 503 iGFRs for 503 children aged between 33 months and 18 years to eGFRs. To define the most precise age cutoff value for each formula, a circular binary segmentation method analyzing the formulas’ bias values according to the children’s ages was performed. Bias was defined by the difference between iGFRs and eGFRs. To validate the identified cutoff, 30 % accuracy was calculated. Results For MDRD, CKD–EPI and CG, the best age cutoff was C14.3, C14.2 and B10.8 years, respectively. The lowest mean bias and highest accuracy were -17.11 and 64.7 % for MDRD, 27.4 and 51 % for CKD–EPI, and 8.31 and 77.2 % for CG. The Schwartz formula showed the best performance below the age of 10.9 years. Conclusion For the MDRD and CKD–EPI formulas, the mean bias values decreased with increasing child age and these formulas were more accurate beyond an age cutoff of 14.3 and 14.2 years, respectively. For the CG and Schwartz formulas, the lowest mean bias values and the best accuracies were below an age cutoff of 10.8 and 10.9 years, respectively. Nevertheless, the accuracies of the formulas were still below the National Kidney Foundation Kidney Disease Outcomes Quality Initiative target to be validated in these age groups and, therefore, none of these formulas can be used to estimate GFR in children and adolescent populations. Keywords Child Chronic kidney disease CKD–EPI formula Cockroft–Gault formula Glomerular filtration rate MDRD formula Schwartz formula
Introduction Several formulas have been developed to estimate glomerular filtration rate (eGFR) in both adults and children
123
60
based on serum creatinine (SCreat) or cystatin C (CysC), with varying degrees of precision. The Schwartz formula [1, 2] is the most widely used formula for eGFR in the pediatric population. The Modification of Diet in Renal Disease (MDRD) [3], the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI) [4] and the Cockroft– Gault formulas (CG) [5] are the most popular formulas for eGFR in adults. These adult formulas have proved to be reliable in adult populations, but their application in the pediatric population often provides misleading and conflicting results. Chehade et al. [6] found that the application of these adult formulas is not reliable in children under 18 years of age across all chronic kidney disease (CKD) stages. Pierrat et al. [7] found that the CG formula corrected for body surface area gave an accurate estimation of the GFR in children aged 12 years and above, whereas, the MDRD led to an overestimation of GFR. However, Filler et al. [8] concluded that the CG formula performs poorly in children above 12 years of age. Selistre et al. [9] also cautioned against the use of these formulas in children. All these studies examined the possibility of applying adult formulas in children and adolescents, with different cohorts selected according to the CKD stage [6] or according to a preset arbitrary cutoff age [8, 9]. Instead of choosing an arbitrary age cutoff, which can be misleading, we here used a robust statistical method, the circular binary segmentation (CBS) method, to investigate an age cutoff in children, above which the MDRD, CKD–EPI or CG formulas present good enough accuracy to be usable in a pediatric population. Performance of the new Schwartz formula according to age was also evaluated.
Materials and methods Population After approval by the Lausanne University Hospital ethics board (Local ethical committee approval number 370/12), we compared 503 inulin clearances (iGFRs) for 503 children aged between 33 months and 18 years with their eGFRs using the MDRD, CKD–EPI, CG, and Schwartz formulas. iGFR measurement was performed in our Pediatric Nephrology department between December 2009 and December 2012. The characteristics and classification of patients’ renal disorders are summarized in Table 1. Children with bladder dysfunction, unable to void spontaneously, or in whom bladder catheterization failed, were excluded. During inulin clearances, proper emptying of the bladder was evaluated by comparing the urine output with its osmolality. A decreasing diuresis with a concomitant decreasing urine osmolality was an indication of poor
123
J Nephrol (2015) 28:59–66 Table 1 Characteristics and classification of patients’ renal disorders according to CKD stages Etiologies
CKD stage 1
Obstructive or reflux uropathy
179
Congenital and acquired single kidney
CKD stage 3
CKD stages 4 and 5
82
36
10
25
15
9
1
Polycystic kidney disease
10
8
2
0
Glomerulopathies
5
4
1
0
Other Total = 503
CKD stage 2
64
36
14
2
283
145
62
13
CKD stages 1, 2, 3, 4, and 5 denote inulin glomerular filtration rate [90, 60–89, 30–59, 15–29, and \15 ml/min/1.73 m2, respectively CKD chronic kidney disease
bladder emptying, and the child was then excluded from the study for technical reasons. Analytical methods iGFR measurement All patients fasted overnight before the day of investigation, and drugs interfering with the inulin measurement were omitted before and during the test. iGFR was obtained as follows: two intravenous catheters were inserted, one in each arm and a loading dose of 25 % inulin was administered according to the manufacturer’s protocol (Inutest SPC, Fresenius Kabi Pharma Austria GmbH, Austria). The loading dose was calculated in order to obtain a required plasma concentration of 200–250 mg/l, as follows: loading inulin dose = required plasma concentration 9 estimated inulin distribution volume. The estimated inulin distribution volume corresponds to the extracellular volume and amounts to 25 % of the body weight (BW). Subsequently, inulin was continuously infused over 90 min at a rate given by the required inulin plasma concentration (200–250 mg/l) and the eGFR as follows: infusion rate in mg/min = (required plasma concentration/1,000) 9 eGFR in ml/min. Water diuresis was induced by oral administration of 20 ml/kg of water (maximum 1,200 ml) in the first hour followed by 3 ml/kg/h of water. This was combined with an intravenous infusion of 0.9 % sodium chloride (maximum 300 ml) every 30 min. After a 90-min equilibration period, three timed-urine samples were collected every 30 min, according to the manufacturer’s protocol, with a blood test in the middle of each urine collection. Inulin was measured using the automatic anthrone method of Wright and Gann, with an Autoanalyzer 3 system (high resolution
J Nephrol (2015) 28:59–66
digital colorimeter of SEAL; Bran ? Luebbe, Norderstedt, Germany). The method was calibrated with five point calibrations (target values at 10 mg/100 ml, 20 mg/100 ml, 30 mg/100 ml, 40 mg/100 ml, 50 mg/100 ml) with correlation coefficients of 0.9993 ± 0.0005. The procedure is automatized with software (AACE 6.03, Bran ? Luebbe) which automatically includes corrections for baseline, carryover, sensitivity drift and dilution factor. For the serum, the intra-assay coefficients of variation (CVs) obtained in our laboratory with the internal quality controls were 2.44 % at 10 mg/100 ml, 1.47 % at 30 mg/100 ml and 0.94 % at 40 mg/100 ml, while for urine the intraassay CVs were 1.71 % at 10 mg/100 ml, 1.22 % at 30 mg/100 ml and 1.07 % at 50 mg/100 ml. The interassay CVs obtained in the internal laboratory standards were 2.35 % at 10 mg/100 ml, 2.23 % at 30 mg/100 ml and 0.87 % at 50 mg/100 ml. iGFR was calculated as the mean of the three clearance periods. When the inulin clearance difference between two periods exceeded 20 %, that period was excluded, and iGFR was calculated as the mean of the two valid periods. SCreat measurement Blood sampled on lithium heparinate was processed immediately for SCreat determination. SCreat was measured using the kinetic colorimetric compensated Jaffe´ method which was standardized against the reference isotope dilution mass spectrometry method (IDMS) and was calibrated with the calibrator and procedures as reported by the manufacturer Roche Modular P system (Roche Diagnostics, Mannheim, Germany). The IDMS traceability involved two-point calibration (target values at 0 and 360–390 lmol/l depending on the calibrator lot) and the subtraction of 26 lmol/l from the results, in order to compensate for the non-specific chromogens. The interassay CVs obtained in the laboratory with the internal quality controls are 3.9 % at 45.7 lmol/l and 2.4 % at 108 lmol/l. The intra-assay CVs are 3.3 % at 44.5 lmol/l and 0.7 % at 148 lmol/l. GFR estimation eGFRs were calculated using the MDRD, CKD–EPI, CG, and Schwartz formulas as follows: •
•
For the MDRD formula [3]: eGFR = 175 9 (SCreat)-1.154 9 (age)-0.203 9 [0.742 if female] 9 [1.212 if black], where SCreat is measured by an IDMScalibrated assay and is expressed in mg/dl, age in years. For the CKD–EPI formula [4]: eGFR = a 9 (SCreat/ b)c 9 (0.993)age. The variable (a) depends on race and sex: black women a = 166; black men a = 163; white/
61
other women a = 144; white/other men a = 141. The variable (b) varies on the basis of sex: women b = 0.7; men b = 0.9. The variable (c) takes on the following values on the basis of sex and SCreat measurement: for women, if SCreat B0.7 mg/dl = -0.329; if SCreat [0.7 mg/dl = -1.209; for men, if SCreat B0.9 mg/ dl = -0.411; if SCreat [0.9 mg/dl = -1.209. • For the CG formula [5]: eGFR = [(140 - age) 9 BW/ SCreat] 9 [K = 1.23 if male or 1.04 if female], where age is in years, BW in kg and SCreat in lmol/l. The CG equation estimates absolute GFR (i.e. in ml/min) rather than standardized GFR (i.e. in ml/min/1.73 m2). The latter was hereby used in both the iGFR measurement and the MDRD and CKD–EPI equations. For expressing CG as standardized GFR, CG estimates are multiplied by 1.73 and divided by the body surface area (BSA). In addition, the CG formula was established using the old SCreat measurement method which is the Jaffe´ technique. In this study, we used the compensated Jaffe´ technique, adjusted to the IDMS method. In order to avoid any bias due to using the CG published constants (1.23 for males of any age, and 1.05 for females of any age), which can lead to an inaccurate estimation of GFR, we calculated new constants based on the iGFR for each patient and then calculated the average. The recalculated average constants were 0.95 ± 1.05 and 1.1 ± 0.33 for female and male children, respectively. We then compared the eGFRs from both the CG formula using the published constants and the modified CG formula using the newly established constants, to the iGFRs. • For the Schwartz formula [2]: eGFR = 0.413 9 (height/SCreat), where height is expressed in cm, and SCreat in mg/dl. Statistical analysis Collected data were analyzed using StataCorp. 2011 (Stata Statistical Software: Release 12; StataCorp LP, College Station, TX, USA). Population characteristics (age, weight and height) are summarized as mean and standard deviation (SD), categorical variables are expressed as percentages. To determine the different change points where each adult formula overestimates or underestimates iGFR, we applied the circular binary segmentation (CBS) method [10] for all the bias values of each adult formula according to the child’s age. Bias was defined as the difference between iGFR and eGFR values. Successive change points delimited the age interval (segment) of over/under estimation. Moreover, to define the most accurate age cutoff value of each formula, we calculated for each formula the
123
62
J Nephrol (2015) 28:59–66
accuracy of each segment. Accuracy at 30 % was defined as the percentage of eGFRs values that were within 30 % of the iGFRs. The CBS method allows to segment data through change-point detection using a maximal t test as follows: let R1, R2,… Rn be the sequence of the bias indexed according to the values of the age (age in increasing order) and let Si = R1 ? R2 ?… Ri, 1 B i B n, be the partial sums. When the biases are normally distributed with known variance, the likelihood ratio statistic for testing the null hypothesis that there is no change against the alternative that there is exactly one change at an unknown location I is given by ZB = max1 B i \ n |Zi|, where Zi = {1/i ? 1/(n - i)} - 0.5 {Si/I - (Sn - Si)/(n - i)} [11]. The null hypothesis of no change is rejected if the statistic exceeds the upper ath quantile of the null distribution of ZB and the location of the change-point is estimated to be i such that ZB = |Zi|.
Results Mean ± SD for patients’ age, BW, height and GFRs according to each studied formula are reported in Table 2. Patients’ age ranged between 33 months and 18 years: 42.14 % were girls. Mean ± SD of iGFRs was 91.69 ± 21 ml/min/1.73 m2, ranging from 14 to 191 ml/min/ 1.73 m2. Children with moderate and severe chronic kidney disease represented 44 % of the total study population. When applying the CBS method for all bias values from the MDRD formula, 4 change points were identified at ages Table 2 Patients’ characteristics and mean ± SD values for eGFRs using the MDRD, the CKD–EPI, the Cockroft–Gault, the modified Cockroft–Gault, and the Schwartz formulas Patient characteristics and eGFR values
Mean ± SD
Patients’ age (years)
11.95 ± 4.09
Patients’ body weight (in kg)
43.64 ± 18.69
Patients’ height (cm) eGFR using the MDRD formula (ml/min/1.73 m2)
147.9 ± 22.08 166.04 ± 88.64
eGFR using the CKD–EPI formula (ml/min/ 1.73 m2)
139.60 ± 33.54
eGFR using the Cockroft–Gault formula (ml/min/ 1.73 m2)
103.87 ± 41.56
eGFR using the modified Cockroft–Gault formula (ml/min/1.73 m2)
98.07 ± 38.26
eGFR using the Schwartz formula (ml/min/ 1.73 m2)
95.42 ± 35.70
eGFR estimated glomerular filtration rate, MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration
123
3.3, 6.6, 10.2, and 14.3 years. This result identified 5 age segments: [2.75–3.3[, [3.3–6.6[, [6.6–10.2[, [10.2–14.3[ and [14.3–18] years (Fig. 1a). The mean bias of each segment was, -61.63, -194.35, -113.25, -59.57 and -17.11 ml/min/1.73 m2, respectively. For each segment we calculated the accuracy at 30 %. Only 25, 0, 3.1, 16.7 and 64.7 % of eGFR values were accurate in the five segments, respectively (Table 3). These results show that the MDRD overestimated the GFR in all age groups (Fig. 2) with an accuracy that increased and reached its best value for the age group between 14.3 and 18 years, with a lowest mean difference of 17.11 ml/min/1.73 m2 as compared to iGFR. When using the CKD–EPI formula, the CBS method applied for all bias values detected three change points corresponding to 3.4, 7.5, and 14.2 years. Therefore four segments were identified (Fig. 1b). The first segment, corresponding to children aged between 33 months and 3.3 years, presented a mean bias value of -43.50 ml/min/ 1.73 m2, and for an accuracy of 30 only 25 % of eGFR values were accurate. The second segment, corresponding to children aged between 3.3 and 7.5 years, presented a mean bias value of -77.53 ml/min/1.73 m2 and an accuracy of 1.1 %. The third segment, corresponding to children aged between 7.5 and 14.2 years, presented a mean bias value of -55.14 ml/min/1.73 m2 and an accuracy of 8.49 % (Table 3). The CKD–EPI formula overestimated the GFR at all age groups (Fig. 2); it presented the highest accuracy of 51 % and the lowest mean difference of -27.4 ml/min/1.73 m2 when compared to iGFRs for the age group between 14.2 and 18 years (segment 4). When applying the CBS method for all bias values from the CG formula, a single change point was detected corresponding to age 10.8 years. We therefore identified two segments: 2.75–10.8 years, and 10.8–18 years (Fig. 1c). The mean bias of each segment was 8.31 and -21.33 ml/min/1.73 m2, respectively. For an accuracy of 30, 77.2 and 63.3 % of eGFR values in each segment were accurate respectively (Table 3). These results show that the CG formula underestimated GFR in pediatric patients under 10.8 years of age and overestimated GFR in children above 10.8 years of age (Fig. 2), with a better accuracy and lowest mean difference when compared to iGFR for the first age group. Regarding the modified CG formula, when applying the CBS method for all bias values we found, similarly to the CG formula, one change point corresponding to 10.8 years of age. We also identified two segments, 2.75–10.8 and 10.8–18 years (Fig. 1d). The first segment presented a mean bias value of 13.35 ml/min/1.73 m2 and an accuracy of 74.6 %. The second segment presented a mean bias value of -15.55 ml/min/1.73 m2 and an accuracy of 70.7 %. Similarly to the CG formula, the modified CG
J Nephrol (2015) 28:59–66
63
Fig. 1 Circular binary segmentation (CBS) method for bias values analysis according to age. a Age change points detection using the MDRD formula. b Age change points detection using the CKD–EPI formula. c, d Age change point detection using the CG and the
modified CG formulas, respectively. e Age change point detection using the Schwartz formula. MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration, CG Cockroft–Gault
formula underestimated GFR in pediatric patients below the age of 10.8 years and overestimated GFR for children above this age (Fig. 2), with a better accuracy and a lower mean difference when compared to iGFR for the first age group (Table 3). Regarding the Schwartz formula, when applying the CBS method for all bias values we found, similarly to the CG and the modified CG formulas, one change point corresponding to 10.9 years of age. We also identified 2 segments, 2.75–10.9 and 10.9–18 years (Fig. 1e). The first segment presented a mean bias value of 10.93 ml/min/ 1.73 m2 and an accuracy of 66.2 %. The second segment presented mean bias value of -12.26 ml/min/1.73 m2 and an accuracy of 61.1 % (Table 3).
Discussion The exact determination of GFR is important, especially in patients with renal failure, after transplantation or receiving potential nephrotoxic drugs. The three most commonly used formulas for eGFR in adults are the MDRD, the CKD–EPI, and the CG. The CG formula was developed in 1976 in a white male population with CKD [5]. It was then validated in overweight and obese diabetic populations with preserved renal function [12] and in adult patients under 65 years of age [13], and was approved in 1998 by the US Food and Drug Administration to guide drug dosage for patients with decreased kidney function without being adapted for sex and race. The MDRD formula was
123
64
J Nephrol (2015) 28:59–66
Table 3 Mean bias and accuracy of each age interval which is delimited by the successive change point of over/under estimation of iGFR obtained by applying the CBS method for all bias values from the MDRD, CKD–EPI, Cockroft–Gault, modified Cockroft–Gault, and Schwartz formulas Formulas
Age interval (years) identified by applying the CBS method for formula bias values
MDRD formula
[2.75–3.3[
-61.63
25
[3.3–6.6[
-194.35
0
[6.6–10.2[
-113.25
3.1
[10.2–14.3[
-59.57
16.7
[14.3–18] [2.75–3.3[
-17.11 -43.50
64.7 25
[3.3–7.5[
-77.53
1.1
[7.5–14.2[
-55.14
8.49
[14.2–18]
-27.4
51
CKD–EPI formula
Cockroft– Gault formula
[2.75–10.8]
Modified Cockroft– Gault formula
[2.75–10.8]
Schwartz formula
Mean bias (ml/ min/1.73 m2) of the obtained age interval
Accuracy (%) within 30 % of the obtained age interval
8.31
77.2
-21.33
63.3
13.35
74.6
-15.55
70.7
[2.75–10.9]
10.93
66.2
]10.9–18]
12.26
61.1
]10.8–18]
]10.8–18]
iGFR inulin glomerular filtration rate, eGFR estimated glomerular filtration rate, MDRD modification of diet in renal disease, CKD–EPI chronic kidney disease epidemiology collaboration, CBS circular binary segmentation, Bias difference between iGFR and eGFR values, Accuracy percentage of eGFR values within 30 % of the iGFR values
published in 1999 [3] as a six-variable equation and then subsequently simplified in 2000 to a four-variable equation including SCreat, age, race, and sex [14]. This formula was developed in a group of adult patients with a mean age of 51 years and mean GFR of 40 ml/min/1.73 m2 and was subsequently demonstrated to be more accurate for patients with stage 3 CKD or higher [15, 16]. Of note, the MDRD formula has been validated in the African American adult population [17], in adults with diabetic nephropathy [18] or scleroderma [19], and finally also in healthy adults without renal disease [20, 21]. The National Kidney Foundation Kidney Disease Outcomes Quality Initiative (NKFKDOQI) recommends its usage for eGFR in adult patients at risk of kidney diseases. The CKD–EPI formula was published in 2009 [4]. It was developed in a group of adults with a mean age of 47 years and a mean GFR of 67 ml/ min/1.73 m2 in order to overcome the limitations of the MDRD formula. It was also externally validated in various groups such as patients with advanced chronic renal failure
123
Fig. 2 Plot for correlation between bias (difference between iGFR and eGFR) and age, using the MDRD formula (blue curve), the CKD– EPI formula (red curve), the CG formula (green curve), the modified CG formula (yellow curve), and the Schwartz formula (grey curve). iGFR inulin glomerular filtration rate, eGFR estimated glomerular filtration rate. For other abbreviations see Fig. 1
[16] or renal transplant patients [22]. The Schwartz formula was developed in the mid-1970s [1], and revised in 2009 [2] to incorporate the change of serum creatinine measurement method. This formula has been developed in a population with a measured GFR ranging from 15 to 75 ml/min/1.73 m2, and is not adjusted for age or sex. The applicability of the above adult formulas in children is controversial. Several research groups have assessed the accuracy of these adult formulas in different pediatric populations, and found conflicting results. Most of these studies used arbitrary age limits to categorize their population, therefore potentially missing the point beyond which these formulas might be sufficiently accurate. Using a robust statistical approach (CBS method), without choosing a predefined age cutoff, our results showed that for the MDRD and the CKD–EPI formulas the mean bias values decreased with increasing age, which is not really surprising. The lowest mean of bias associated with the highest accuracy was observed in the adolescent population group, beyond an age cutoff of 14.3 and 14.2 years for the MDRD and the CKD–EPI, respectively. However, even above these cutoff values, neither the MDRD nor the CKD–EPI fulfill the NKF-KDOQI target recommendation of C90 % of eGFR estimates to be within 30 % of measured GFR [23]. For the CG and the modified CG formulas, the lower mean bias values and the best accuracies were obtained below an age cutoff of 10.8 years. Although the accuracy obtained with the CG formula was better than the ones with the MDRD or the CKD–EPI formulas, it was still below the required NKF-KDOQI target recommendation for validation. Furthermore, our results showed that
J Nephrol (2015) 28:59–66
the MDRD and the CKD–EPI formulas overestimate the GFR at all age groups. These results are in line with other previous publications [6–8]. In our opinion, the reason why the adult equations perform so poorly in children is due to the fact that there are other parameters not included in the CKD–EPI, MDRD and CG formulas that significantly affect the formulas’ accuracy and results, making them non applicable in pediatric patient groups. In addition, ‘‘age’’ represents an important variable and is a central component of the adult equations. In adults, muscle mass decreases with time and therefore age is entered as: ‘‘(0.993)age’’ and ‘‘(age)-0.203’’ in the MDRD and CKD– EPI formulas and (140-age) in the CG formula, resulting in lower eGFR in older subjects than expected from the SCreat readings. However, the opposite is true for children, where muscle mass increases with age leading to increasing SCreat values, which are more closely related to height than age (which is the concept underlying the pediatric formulas for eGFR). Hence we are not surprised that the adult formulas for eGFR present a large bias in children across all age groups. The Schwartz formula predicts GFR much better than the MDRD or the CKD– EPI formulas. However, we observed a fairly similar prediction of GFR when using the Schwartz formula and the CG formulas. Variation of GFR prediction with the Schwartz formula has already been shown [24–26], especially in the high GFR range. We here demonstrate that the accuracy of the new Schwartz formula is also below the NKF-KDOQI recommendations for it to be usable in our population. Recently, Gao et al. [27] proposed an improvement of the Schwartz linear model in the prediction of GFR using the variable ratio height/creatinine in a quadratic model and published a new quadratic formula adjusted to age and sex. There are a few limitations to our study. First, we did not evaluate any adult formula including CysC as we did not measure the CysC in our patients. Second, we used the compensated Jaffe´ technique for SCreat measurement. The CG formula was developed using the old Jaffe´ technique for SCreat measurement. Nevertheless, we recalculated the formula‘s constants according to our SCreat measurement method and to the patients’ iGFRs, in order to overcome a possible bias related to the difference in the SCreat method of determination. In conclusion, the MDRD and the CKD–EPI formulas present the lowest bias and the best accuracies in adolescents over the age of 14.3 years. However, the accuracies of these formulas are still below the NKF-KDOQI target to be validated in this age group. The CG formula is more accurate in children below 10.8 years but still presents an accuracy below the NKF-KDOQI target guideline. Therefore, none of these formulas can be used to estimate GFR in children or adolescent populations.
65 Conflict of interest
All the authors declare no competing interests.
References 1. Schwartz GJ, Haycock GB, Edelmann CM et al (1976) A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 58:259–263 2. Schwartz GJ, Mun˜oz A, Schneider MF et al (2009) New equations to estimate GFR in children with CKD. J Am Soc Nephrol 20(3):629–637 3. Levey AS, Coresh J, Greene T et al (2006) Chronic Kidney Disease Epidemiology Collaboration: using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 145(4):247–254 4. Levey AS, Stevens LA, Schmid CH et al (2009) CKD–EPI (Chronic Kidney Disease Epidemiology Collaboration) A new equation to estimate glomerular filtration rate. Ann Intern Med 150:604–612 5. Cockroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41 6. Chehade H, Girardin E, Iglesias K et al (2013) Assessment of adult formulas for glomerular filtration rate estimation in children. Pediatr Nephrol 28:105–114 7. Pierrat A, Gravier E, Saunders C et al (2003) Predicting GFR in children and adults: a comparison between Cockroft–Gault, Schwartz, and MDRD formulas. Kidney Int 64:1425–1436 8. Filler G, Foster J, Acker A (2005) The Cockroft–Gault formula should not be used in children. Kidney Int 67:2321–2324 9. Selistre L, De Souza V, Cochat P et al (2012) GFR estimation in adolescents and young adults. J Am Soc Nephrol 23:989–996 10. Venkatraman ES, Olshen AB (2007) A faster circular binary segmentation algorithm for the analysis of array CGH data. Bioinformatics 23:657–663 11. Sen AK, Srivastava MS (1975) On tests for detecting change in mean. Ann Stat 1:98–108 12. Drion I, Joosten H, Santing L et al (2011) The Cockroft–Gault a better predictor of renal function in an overweight and obese diabetic population. Obes Facts 4:393–399 13. Verhave JC, Fesler P, Ribstein J, du Cailar G, Mimran A (2005) Estimation of renal function in subjects with normal serum creatinine levels: influence of age and body mass index. Am J Kidney Dis 46:233–241 14. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D (1999) 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 Med 130:461–470 15. Vervoort G, Willems HL, Wetzels JF (2002) Assessment of glomerular filtration rate in healthy subjects and normoalbuminuric diabetic patients: validity of a new (MDRD) prediction equation. Nephrol Dial Transplant 17:1909–1913 16. Teruel Briones JL, Gomis Couto A, Sabater J et al (2011) Validation of the Chronic Kidney Disease Epidemiology Collaboration (CKD–EPI) equation in advanced chronic renal failure. Nefrologia 31:677–682 17. Lewis J, Agodoa L, Cheek D et al (2001) Comparison of crosssectional renal function measurements in African Americans with hypertensive nephrosclerosis and of primary formulas to estimate glomerular filtration rate. Am J Kidney Dis 38:744–753 18. Poggio ED, Wang X, Greene T, Van Lente F, Hall PM (2005) Performance of the modification of diet in renal disease and Cockroft–Gault equations in the estimation of GFR in health and in chronic kidney disease. J Am Soc Nephrol 16:459–466
123
66 19. Kingdon EJ, Knight CJ, Dustan K et al (2003) Calculated glomerular filtration rate is a useful screening tool to identify scleroderma patients with renal impairment. Rheumatology 42:26–33 20. Kang YS, Han KH, Han SY, Kim HK, Cha DR (2005) Characteristics of population with normal serum creatinine impaired renal function and: the validation of a MDRD formula in a healthy general population. Clin Nephrol 63:258–266 21. Lin J, Knight EL, Hogan ML, Singh AK (2003) A comparison of prediction equations for estimating glomerular filtration rate in adults without kidney disease. J Am Soc Nephrol 14:2573–2580 22. Townamchai N, Praditpornsilpa K, Chawatanarat T et al (2013) The validation of estimated glomerular filtration rate equation for renal transplant recipients. Clin Nephrol 79(3):206–213 23. Levey AS, Coresh J et al (2002) National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease
123
J Nephrol (2015) 28:59–66
24.
25.
26.
27.
evaluation, classification, and stratification. Am J Kidney Dis 39(2 Suppl 1):S1–S266 Bacchetta J, Cochat P, Rognant N et al (2011) Which creatinine and cystatin C equations can be reliably used in children? Clin J Am Soc Nephrol 6:552–560 Staples A, LeBlond R, Watkins S et al (2010) Validation of the revised Schwartz estimating equation in a predominantly nonCKD population. Pediatr Nephrol 25:2321–2326 Berg UB, Ba¨ck R, Celsi G et al (2011) Comparison of plasma clearance of iohexol and urinary clearance of inulin for measurement of GFR in children. Am J Kidney Dis 57:55–61 Gao A, Cachat F, Faouzi M et al (2013) Comparison of the glomerular filtration rate in children by the new revised Schwartz formula and a new generalized formula. Kidney Int 83(3): 524–530
