Comparing Estimated and Measured Glomerular Filtration Rate in Children with Posterior Urethral Valve

Abstract

Background:

Posterior urethral valve (PUV) is obstructive uropathy that may lead to chronic kidney disease (CKD) and end-stage renal disease (ESRD) in children. Glomerular filtration rate (GFR) measurement remains the gold standard for renal function measurement. However, due to its less availability and cumbersome, it is not commonly used, and GFR is estimated utilizing various endogenous filtration markers.

Objective:

This study includes pediatric patients with PUV. We aimed to compare the measured GFR (mGFR) with various creatinine-based estimated GFR methods (eGFR).

Materials and Methods:

A single-center retrospective study included 62 treated cases of PUV, postvalve fulguration. The mGFR measured by 99mTc-diethylenetriaminepentaacetate in vitro method and compared with eight eGFR (Schwartz, Cockcroft-Gault [CG], Counahan-Barratt [CB], CKD Epidemiology Collaboration [CKD-EPI], full-age spectrum [FAS] age, FAS height (FAS Ht), Schwartz-Lyon [SL], and Ht independent). Patients were subdivided into different CKD grades and compared with various eGFR.

Discussion:

PUV is a common cause of CKD in children and needs special consideration as there is growth retardation associated with it. It decreases creatinine production and thus fallacies in eGFR measurement. There is a requisite to identify and closely monitor the subset of patients with baseline decreased renal function and therefore at risk of developing ESRD.

Results:

A total of 62 patients were included. Mean age and serum creatinine levels were 8.02 ± 5.53 years and 1.15 ± 0.95 mg/dl (range: 0.4–4.5), respectively. The mean mGFR was 61.6 ± 31.80 mL/min/1.73 m² and a positive variable correlation was 0.46–0.77 between mGFR and eGFR. Based on mGFR, there were 14 (22.6%), 21 (33.8%), 13 (20.9%), 9 (14.5%), and 5 (8.1%) patients in Grades I–V, respectively. The correct classification of the CKD grades was noted in 25 (40.3%), 16 (25.8%), 32 (51.6%), 16 (25.8%), 25 (40.3%), 27 (43.5%), 26 (41.9%), and 28 (45.2%) patients by Schwartz, CG, CB, CKD-EPI, FAS age, FAS Ht, SL, and Ht-independent equation. The eGFR overestimates GFR at the lower level and underestimates at higher levels.

Conclusion:

Our results confirm the considerable limitations of various creatinine-based clearance methods for estimating actual GFR. The creatinine clearance-based eGFR should not replace the measurement of the GFR. An initial measure of the mGFR followed by serial follow-up with the eGFR equation may be done. The most accurate eGFR equations are CB for Grade II, SL or Ht independent for Grade III, FAS age for Grade IV, and SL for Grade V CKD.

INTRODUCTION

The posterior urethral valve (PUV) is the most common cause of congenital lower urinary tract obstruction in boys. The incidence of PUV is estimated at 1:5000–1:8000 in males.[1] It is a common cause of chronic kidney diseases (CKDs) and end-stage renal disease (ESRD).[2,3] Despite early valve fulguration, nearly two-thirds of the children may continue to have bladder dysfunction. They progress to various degrees of bladder decompensation near puberty.[4,5] Significant children (11%–51%) develop progressive renal failure and eventually progress to ESRD.[6,7]

Glomerular filtration rate (GFR) is the best indicator of renal function in children and adolescents. It is paramount to diagnose acute and chronic renal impairment, apply early intervention to avoid ESRD, advise nephrotoxic drugs, and monitor the side effects of medications.[8] The mean baseline GFR in pediatric patients with PUV who ultimately developed renal failure was significantly lower than those without renal failure.[9] Authors have found that mean GFR at presentation significantly correlates with the outcome. The patients with a mean GFR of more than 70.39 ± 24.03 ml/min/1.73 m² did not develop chronic renal failure compared to patients with a GFR of <39.55 ± 19.97 ml. While age at presentation (<1 year vs. >1 year), the presence/absence of vesicoureteral reflux and cortical scars were insignificant with ultimate renal function.[10] Other studies have noted similar findings.[11]

The gold-standard technique to measure GFR (mGFR) is injecting an exogenous glomerular filtration marker.[12] However, it is not widely available, considered invasive, and cumbersome. Due to routine mGFR measurement difficulties, Estimated GFR (eGFR) is measured using clearance of endogenous markers such as creatinine, urea, cystatin C, beta-trace protein, and beta-2 microglobulin.[13] Deng et al. compared eGFR by iohexol clearance with various eGFR equations in pediatric patients. The author found bias ranged from 9.2 to 15.2 ml, with a variable difference within 15% (P₁₅, 34.6%–61.7%) and 30% (P_30,64.2%–82.7%) of mGFR accuracies.[14] Out of endogenous markers, creatinine is the most commonly used. It is secreted by renal tubular cells and filtered by glomerular capillaries. The extrarenal clearance in patients with normal kidney function is small. However, in CKD, as much as two-thirds of excretion can occur by the extrarenal route.[15] A significant number of patients with PUV show growth failure. It results in lower body weight, stunting, and muscle mass.[16] It leads to lower creatinine production; thus, the creatinine-based clearance equation may estimate a spurious high GFR.[17]

There is a lack of data evaluating creatinine-based GFR prediction equations in patients with PUV against a reference standard. The study aimed to compare the gold-standard mGFR with various eGFR. We measured the mGFR using clearance of 99mTc-diethylenetriaminepentaacetate (DTPA). We compared it to the different creatinine clearance equations (Schwartz,[18] Cockcroft-Gault [CG],[19] Counahan-Barratt [CB],[20] CKD Epidemiology Collaboration [CKD-EPI],[21] full-age spectrum [FAS] age,[22] FAS height (FAS Ht),[23] Schwartz-Lyon [SL],[24] and Ht independent.[25]) We intended to know the best eGFR equation for routine clinical practice in PUV patients according to the renal function level.

MATERIALS AND METHODS

Study design and data

A single-center retrospective study was done in a tertiary care hospital between January 2014 and August 2022. They were referred for the GFR measurement. All patients underwent serum creatinine (SCr), blood urea nitrogen (BUN), and serum albumin measurement day as the patients were referred by a nephrologist for the routine assessment of the mGFR. Institutional Ethics Committee approved the retrospective study (2023-66-IM-130).

Inclusion criteria

All patients were cystoscopically proven cases of PUV, postvalve fulguration.

Exclusion criteria

Patients on angiotensin-converting enzyme inhibitors or nonsteroidal anti-inflammatory drugs were excluded from the study.

Measured glomerular filtration rate measurement

GFR was measured at least 6 weeks after valve fulguration. Patients were suggested to have a regular protein vegetarian diet (1 g protein/kg body weight) 10 days before the test. On the study day, all patients reported a fasting state of 6 h and were given oral hydration (5 ml/kg plain water) 30 min before the mGFR. The mGFR was measured using the modified Russell algorithm. One millicurie freshly prepared 99mTc-DTPA was administered intravenously. Venous blood samples (5 ml) were collected from the contralateral arm at 60 and 180 min. The time of injection and blood samples were recorded. After 24 h, 1 ml of plasma from the blood sample and standard was counted in the well counter for 1 min. The plasma clearance of the 99mTc-DTPA was calculated.[26] SCr was measured by the alkaline-picrate method using modified Jaffe kinetics with an auto-analyzer (Randox Imola).

eGFR calculation

We evaluated eight creatinine-based prediction equations for each patient (Schwartz, CG, CB, CKD-EPI, FAS age, FAS Ht, SL, and Ht independent) to calculate several eGFR for each patient. Patients were divided into the various CKD stages based on the mGFR and were compared with this eGFR. CKD stage classification was done as per The Kidney Disease Improving Global Outcomes Guideline.[27]

Statistical analysis

All statistical calculations were done using the computer program SPSS (the Statistical Package for the Social Science; SPSS Inc., Chicago, IL, USA) version 23 for Microsoft Windows. The normality of continuous variables was established, and the variable was normally distributed when the Z value of the skewness was within ± 3.29. The normality of the continuous variable was expressed as means ± standard deviation (SD), while categorical variables were as frequencies (%). Independent samples t-test was used to compare the means between the two groups. At the same time, Chi-square tests (or Fisher's exact test) were performed to test the association between two categorical variables. A linear correlation coefficient was used to detect a correlation between two quantitative variables in one group. The prediction of mGFR was made by multiple linear regression analyses. Bland–Altman plots assessed the bias (mean difference) between eGFR equations. Sample size and power calculation were done retrospectively. Assuming the correlation coefficient between estimated and observed GFR was 0.65. At minimum two-sided 95% confidence interval (CI) and 80% power of the study, the estimated sample size was 53, where the null hypothesis of the correlation was assumed to be 0.4 (0.4–0.59 is considered low-positive correlation). The sample size was estimated using software NCSS statistical Software,Kaysville, Utah, 84037, USA. We achieved power of 85 in the study.

RESULTS

Baseline demographic clinical and glomerular filtration rate measurement

A total of 62 male patients were included in the study. Demographic and clinical variables, including various GFR (mGFR and eGFR), are presented in Table 1. All included patients were normotensive and had different SCr levels. The mean age and the SCr level of the study patients were 8.02 ± 5.53 years (range: 0.4–16 years) and 1.15 ± 0.95 mg/dl (range: 0.4–4.5), respectively. The mean body mass index was 15.9 ± 8.1 (Q1: 13.5 and Q3: 20.3). Similarly, the mean mGFR value was 61.6 ± 31.80 mL/min/1.73 m². The rest of the variables, including various eGFR distribution, is given in Table 1.

Table 1.

Distribution of the demographic, biochemical parameters, measured glomerular filtration rate, and various estimated glomerular filtration rate of study patients (n=62)

Variable	Values (mean±SD)	Median (range)
Age	8.02±5.53	6.00 (0.4–16)
Serum creatinine (mg/dL)	1.15±0.95	0.81 (0.4–4.5)
BUN (mg/dL)	19.33±15.23	15.00 (4.5–79)
Serum albumin	4.40±1.62	4.30
Weight (kg)	28.20±22.32	17.50
Height (cm)	120.12±37.38	110.5
GFR*
mGFR	61.6±31.8	64.5 (10–118)
Various eGFR
Schwartz	56.2±24.4	58.1 (11.7–109.4)
CG	51.9±37.3	42.8 (7.3–161.5)
CB	58.5±25.4	60.5 (11.9–113.9)
CKD-EPI	122.4±46.1	132.2 (14.6–191)
FAS age	56.4±27.7	54.6 (11.5–109.2)
FAS Ht	55.9±25.9	53.6 (10.8–104.7)
SL	52.9±23.6	53.1 (10.3–98.8)
Height independent	53.1±23.4	51.3 (11.2–105.9)

*All GFR value is in mL/min/1.73 m². GFR: Glomerular filtration rate, mGFR: Measured GFR, eGFR: Estimated GFR, CKD: Chronic kidney disease, EPI: Epidemiology, FAS: Full-age spectrum, SD: Standard deviation, BUN: Blood urea nitrogen, CG: Cockcroft-Gault, CB: Counahan-Barratt, SL: Schwartz-Lyon, FAS Ht: FAS height, FAS age: FAS age

Correlation between measured glomerular filtration rate and eGFR measurements

A positive variable correlation (0.46–0.77) existed between mGFR and various eGFR. A strong correlation was found between mGFR and Schwartz (r = 0.77, P < 0.05), CB (r = 0.77, P ≤ 0.05), CKD-EPI (r = 0.69, P > 0.05), FAS age (0.61, P < 0.05), FAS Ht (r = 0.68, P < 0.05), SL (r = 0.73, P < 0.5), and Ht independent (r = 0.69, P < 0.005) equations. However, CG (r = 0.46, P > 0.05) showed only a low-positive correlation with mGFR. In between the eGFR equations, an excellent positive correlation was noted between Schwartz, CB, FAS Ht, SL, and Ht-independent equations.

Correlation between measured glomerular filtration rate and various demographic and clinical measurements

The correlation of mGFR with demographic profile and the clinical variables is given in Table 2. The result showed that there was a moderate but negative correlation between mGFR and SCr (r = −0.604, P < 0.05) and BUN (r = −0.579, P < 0.05). However, no significant correlation was observed with mGFR and other variables, including age, weight, Ht, and serum albumin.

Table 2.

Prediction of measured glomerular filtration rate of the study patients (n=62)

Variable’s	Unstandardized (β)	SE	Standardized (β)	P *
Serum creatinine	−13.41	4.05	−0.40	0.002
BUN	−0.73	0.25	−0.35	0.006
Constant	90.90	5.38		<0.001

*P <0.05 significant. Dependent variable: mGFR, multiple linear regression analysis used. Unadjusted R²=44.1%, adjusted R²=42.2%, one-way ANOVA (P <0.001). mGFR: Measured glomerular filtration rate, BUN: Blood urea nitrogen, SE: Standard error

Prediction of measured glomerular filtration rate using serum creatinine and blood urea nitrogen

Multiple linear regression analyses were used to estimate the regression equation for predicting mGFR. Only SCr and BUN showed significance in univariate and multivariate analysis. Age, Ht, weight, and serum albumin were insignificant. The regression coefficient of SCr (ß = −13.41) and BUN (ß = −0.73) indicated that one unit increase results in −13.41 and −0.73 unit change mGFR, respectively. However, the prediction accuracy of this model was only 44.1% [Table 2]. The respective linear regression equation is given below.

mGFR = (−13.41 × SCr) + (−0.73 × BUN) + 90.90

Relationship between measured glomerular filtration rate chronic kidney disease grades and demographic and clinical values

Patients were divided into five grades of CKD according to the levels of the mGFR. The performance of the various eGFR was compared according to the level of the mGFR. The result showed that the mGFR (taken as the gold-standard method here) identified 14 cases (22.6%) as Grade I, 21 (33.8%) as Grade II, and 13 (20.9%) as Grade III. Grades IV and V were noted in 9 (14.5%) and 5 (8.1%) [Figure 1]. There was no significant difference (P = 0.283) between different CKD stages and age.

Figure 1.

Patients (n = 62) subdivided into the various stages of the CKD by mGFR and different eGFR equations. The number and percentages represent the number of patients in each CKD group by individual equations. CKD EPI: Chronic Kidney Disease-Epidemiology Collaboration, mGFR: Measured glomerular filtration rate, FAS age: Full-age spectrum age, eGFR: estimated glomerular filtration rate, CG: Cockcroft-Gault, Ht independent: Height independent, FAS Ht: FAS height

There was a significant difference between the CKD grade classification by mGFR and various eGFR (P < 0.001). Schwartz correctly classifies only 3/14 (21.4%), 8/21 (38.1%), 8/13 (61.5%), 5/9 (55.6%), and 1/5 (20%) as Grades I–V, respectively. Similar performance of correct and incorrect classification of various grades is shown in Figure 2. The above results found that EPI erroneously overestimates GFR and puts most patients in Grade I. The most sensitive eGFR equations are CB (57.1%) for Grade II, SL and Ht independent (each 61.5%) for Grade III, FAS age (77.1%) for Grade IV, and SL (60%) for Grade V.

Figure 2.

CKD grade classification (correct and incorrect) by different eGFR equations compared to the gold-standard mGFR. The number represents the number of patients in the group by different eGFR equations. (Correct: eGFR and mGFR put individual patients in same CKD grade; incorrect: CKD grades given by mGFR and eGFR do not match). *As mGFR is considered the gold standard, only correct classification is shown. CKD: Chronic kidney disease, eGFR: estimated glomerular filtration rate, mGFR: Measured glomerular filtration rate, CG: Cockcroft-Gault, EPI: Epidemiology, FAS age: Full-age spectrum age, FAS Ht: Full-age spectrum height, HT independent: Height independent

The accurate overall classification of the different CKD grades was noted in only 25 (40.3%), 16 (25.8%), 32 (51.6%), 16 (25.8%), 25 (40.3%), 27 (43.5%), 26 (41.9%), and 28 (45.2%) patients by Schwartz, CG, CB, CKD-EPI, FAS age, FAS (Ht), SL, and Ht-independent equation, respectively [Figure 2].

Bias between each equation

Bland–Altman plots [Figure 3] show the level of consistency in the measurements between mGFR and various individual eGFR equations. Overall, high discordance was observed between various eGFR equations and the mGFR. The result showed that mean bias between mGFR and Schwartz (−5.39, 95% CI = 0.25–10.55, P = 0.040), Ht independent (−8.57, 95% CI = 2.72–14.42, P = 0.005), and SL (−8.73, 95% CI = 3.27–14.19, P = 0.002) were statistically significant indicated consistency in values between two measurements. It was insignificant with CB (−3.08, 95% CI = 02.07–8.23, P = 0.237), FAS age (−5.19, 95% CI=−1.53–11.90, P = 0.128), and FAS Ht (−5.73, 95% CI = −0.22–11.70, P = 0.059), which show inconsistency in mean difference between the measurements.

Figure 3.

The Bland–Altman plot to study the bias between various eGFR equations with mGFR. The X-axis is mGFR, and Y-axis is eGFR substrated by mGFR. Middle error bar showing 95% CI of the mean difference, whereas the upper and lower error bar were showing a 95% CI of the upper and lower limit of the value of mean ± two standard deviations. All eGFR equations show variable mean bias. Various eGFR equations overestimate GFR at a lower level of the mGFR and vice versa. CI: Confidence interval, mGFR: Measured glomerular filtration rate, eGFR: estimated glomerular filtration rate, FAS age: Full-age spectrum age, FAS Ht: Full-age spectrum height

The eGFR is overestimated at the lower mGFR and underestimated at higher levels. The upper 95% limit of agreements was beyond the acceptable limit. Limits of agreement (2 SD by Bland–Altman method) were 35 to 43 (range 78.0) ml/min/1.73 m² for Schwartz, 38 to −42 (range 80.0) ml/min/1.73 m² for CB, 48 to −56 (range for 114) ml/min/1.73 m² for FAS age, 40 to −51 (range 101) ml/min/1.73 m² for FAS Ht, 32 to −51 (range 83) ml/min/1.73 m² for SL, and 37 to −53 (range 90) ml/min/1.73 m² for Ht-independent eGFR equation, respectively.

DISCUSSION

Summary of the result

A total of 62 PUV patients with variable SCr 1.15 ± 0.95 mg/dl (range: 0.4–4.5) were included in the study. A positive correlation (0.46–0.77) between mGFR and various eGFR equations existed. Most of the equations misclassify that the CKD grades are more than half of patients. The most accurate eGFR equations are CB (57.1%) for Grade II, SL and HT independent (each 61.5%) for Grade III, FAS age (77.1%) for Grade IV, and SL (60%) for Grade V.

Posterior urethral valve-chronic kidney disease and estimation or measurement of glomerular filtration rate

PUV is the common cause of CKD and ESRD.[2,3] Despite the valve fulguration, a large proportion of the patients eventually develops ESRD.[6,7] Various coexisting factors, such as renal dysplasia, VUR, late presentation, and urinary tract obstruction, may predispose the failure.[28,29,30,31] We found a high rate of CKD Grades II–V in our study.

The GFR serves as the most reliable marker of kidney function. GFR decreases because of reduced nephrons or filtration rates of the individual single nephron in CKD.[8] GFR cannot be measured directly, so it is measured indirectly by measuring the plasma elimination of the exogenous filtration administered marker. Inulin has considered the ideal substance to estimate GFR. This procedure is cumbersome and challenging.[32] In routine clinical practice, 99mTc-DTPA and 51 Cr-ethylenediaminetetraacetic acid (⁵¹ Cr-EDTA) are utilized, and they correlate strongly with inulin clearance.[33] However, because of the limited availability, expense, and time needed to use exogenous markers to determine an actual GFR, the estimation of GFR from the renal clearance of creatinine is used. In children, GFR estimating equations provide more accurate estimates of kidney function than the single SCr measurement.[34] All the creatinine-based equations do not perform equally well in children. Creatinine production and clearance depend on muscular mass, age, sex, variable absorption, and tubular secretion. Creatinine increases once the GFR has fallen to about 50%–60%. Using SCr alone may delay CKD diagnosis.[35]

We evaluated eight commonly used equations for the comparison with mGFR. The Counahan–Barratt formula has been used in children, primarily in Europe.[20] Studies have shown that mean GFR in patients who ultimately developed renal failure is lower than those without renal failure on an age-by-age basis.[10] Our study also noted a mean mGFR of 61.6 ± 31.8 mL/min/1.73 m², much lower than expected. It is because several children in our study had CKD III–V.

Our study found no significant correlation between age and CKD grades. Bhadoo et al. found a high incidence (42.7%) of CKD in their study, with a mean patient age of 31.3 months. However, the authors found that age at presentation (<1 year vs. >1 year), VUR, and renal cortical scars had no significant correlation with ultimate renal outcome.[10] It highlights that it is difficult to predict future renal impairment in patients with PUV. It stresses the regular follow-up in these patients with serial GFR measurement.

eGFR and their performance in different chronic kidney disease stages

Our study found no single equation suitable that could be used in all grades of CKD. More than half of our patients showed incorrect classification of the CKD grades by the various equations (49.3%–74.2%). There was a significant difference in the CKD stage allotted by the eGFR compared to the mGFR. CKD grades are one of the most important clinical parameters for managing CKD patients.[27] More than 50% of the misclassification would be clinically unacceptable. Schwartz equation was not found to be the most sensitive for any CKD stage.

Clinical implications

As the frequent measurement of the mGFR is cumbersome, we recommend the formulation of a GFR prediction equation for all CKD grades. Baseline measurement of the GFR should be done by plasma filtration, followed by serial follow with any of the above equations. Repeat mGFR should be done in follow-up when there is suspected deterioration in the renal function based on clinical or biochemical parameters. However, the difference between mGFR and eGFR should be kept in mind.

Limitation

There are a few limitations to the study that should be discussed. First, the sample size is limited. The various eGFR equations’ performance may be influenced by multiethnic settings such as muscle mass and nonvegetarian meals. It was a retrospective study. To the best of our knowledge, it is the first study that has compared mGFR with the various SCr-based eGFR equations in PUV patients and included patients with different renal function levels.

CONCLUSION

Our results confirm the considerable limitations of various creatinine-based clearance methods compared to the mGFR. mGFR should not be replaced by the creatinine clearance eGFR equations. None of the equations has a promising performance in Grade I. We recommended the initial measurement of the mGFR by the radionuclide plasma clearance method to assess renal function. However, due to the limited availability of the measurement method, based on the results of CKD grades by mGFR measurement, various eGFR equations may be utilized. The most accurate eGFR equations are CB for Grade II (57.1%), SL or HT independent for Grade III (61.5%), FAS age for Grade IV (71.5%), and SL (60%) for Grade V.

Ethical clearance

Institute Ethics Committee approved the retrospective study (PGI/BE/133/2023).

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

Conference Presentation

A part of the the presented work has been accepted for a poster presentation as SNMI 2021, India.