

ORIGINAL ARTICLE 



Year : 2017  Volume
: 27
 Issue : 2  Page : 124128 

Comparison of serum creatininebased estimating equations with gates protocol for predicting glomerular filtration rate in indian population
AV Mulay^{1}, SM Gokhale^{2}
^{1} Department of Nephrology, Dialysis Unit, Inlaks and Budhrani Hospital, Pune, Maharashtra; Kidney Research and Support Network Trust, Pune, Maharashtra, India ^{2} Department of Nuclear Medicine, Inlaks and Budhrani Hospital, Pune, Maharashtra, India
Date of Web Publication  8Mar2017 
Correspondence Address: A V Mulay Dialysis Unit, Inlaks and Budhrani Hospital, Koregaon Park, Lane No. 1, Pune, Maharashtra India
Source of Support: None, Conflict of Interest: None  Check 
DOI: 10.4103/09714065.200515
In clinical practice, serum creatininebased predicting equations and Gates protocol based on gamma camera imaging of kidneys after injection of Tc99mdiethylenetriaminepentaacetic acid (DTPA) are commonly used to assess glomerular filtration rate (GFR). Comparison of these methods, especially the chronic kidney diseaseepidemiology collaboration (CKDEPI) equation with gold standard method of assessment of GFR by plasma clearance of Tc99mDTPA is not wellstudied in Indian population. We conducted this study to compare GFR estimation by gamma camerabased Gates protocol and serum creatininebased predicting equations with GFR measured by plasma clearance of Tc99mDTPA. One hundred and five patients (65 male and 40 female) underwent Tc99m DTPA renal scan followed by withdrawal of venous blood samples at 2, 3, and 4 h as per predefined protocol. Gates method GFR (GFR_{s}) was assessed using standard protocol. GFR by plasma sampling (GFR_{p}) was calculated by slopeintercept method with provision for corrections. Estimated GFR was calculated by CockroftGault formula, four variable modification of diet in renal disease (MDRD) equation, and CKDEPI equation (GFR_{CG, }GFR_{MDRD}, GFR_{CKDEPI}, respectively). GFR measured by gold standard method (GFR_{p}) was compared with that estimated by other methods by calculating correlation coefficient, bias, precision, and accuracy. GFR estimated by all three estimating equations correlated better than GFR_{s }with GFR_{p. }For estimating GFR_{p}, GFR_{CKDEPI }had highest correlation with GFR_{p }with least bias and highest precision. Gamma camerabased Gates protocol was the least precise and least accurate method for estimating GFR_{p}. To conclude, all three estimating equations based on serum creatinine are superior to Tc99m DTPA scintigraphy for estimating GFR; CKDEPI equation being the most accurate and precise.
Keywords: Chronic kidney diseaseepidemiology collaboration equation, diethylenetriaminepentaacetic acid renogram, estimated glomerular filtration rate, measured glomerular filtration rate
How to cite this article: Mulay A V, Gokhale S M. Comparison of serum creatininebased estimating equations with gates protocol for predicting glomerular filtration rate in indian population. Indian J Nephrol 2017;27:1248 
How to cite this URL: Mulay A V, Gokhale S M. Comparison of serum creatininebased estimating equations with gates protocol for predicting glomerular filtration rate in indian population. Indian J Nephrol [serial online] 2017 [cited 2017 Jul 24];27:1248. Available from: http://www.indianjnephrol.org/text.asp?2017/27/2/124/200515 
Introduction   
Glomerular filtration rate (GFR) is the most important measure of overall kidney function. Accurate estimation of GFR is necessary in a number of clinical situations such as renal donor evaluation, monitoring patients of chronic kidney disease (CKD) for disease progression, monitoring of patients with single kidney or renal transplant recipients, and calculating doses of drugs with narrow therapeutic window such as chemotherapeutic agents.
Inulin clearance is considered the gold standard procedure for the measurement of GFR,^{[1],[2]} but procurement of inulin and the cumbersome procedure pose a challenge for its routine clinical use. GFR can be measured using tracers that are cleared exclusively by glomerular filtration without significant tubular secretion or reabsorption. Ethylenediaminetetraaceticacid (EDTA) and diethylenetriaminepentaaceticacid (DTPA) are such chemicals that are exclusively handled by glomerular filtration. Plasma clearance of radiolabeled EDTA (Cr51EDTA) and DTPA (Tc99m DTPA) emerged in the 1960's and 1970's as a reliable method of estimation of GFR. This technique of GFR estimation involved a single injection of radiolabeled EDTA or DTPA followed by multiple sampling of blood and measuring the clearance by calculating the area under the curve.^{[3],[4]}However, it was found to be labor intensive and was further simplified by restricting the blood sampling to the second of the two exponential components of clearance.^{[5]}This is known as slopeintercept method. This simplification introduced systematic errors in the values of GFR, and various methods of correction have been applied.^{[5],[6],[7]} There could be systematic differences in the values of GFR estimated depending on the substance being used, whether urinary or plasma clearance is used and whether arterial or venous samples are used. It is recommended that clearance of EDTA from venous samples be taken as standard measure of GFR. Small systematic differences have been observed between GFR measurements obtained from EDTA and DTPA clearance.^{[8]} However, they are sufficiently small and plasma clearance of radiolabeled DTPA can be recommended as suitable alternative radiopharmaceutical to EDTA.
There are three methods of estimating the plasma clearance of DTPA  area under curve method, slopeintercept method, and singlesample method. The “area under curve” method is too cumbersome and the “singlesample method” is not precise. The “slopeintercept method” is not only accurate and precise but also provides scope for quality control checks by means of various correlations e.g., Chantler technique.^{[5],[6],[7]}
To avoid repeated blood sampling, kidneys can be imaged sequentially using a gamma camera after injection of Tc99m DTPA and Gates protocol can be used to estimate GFR from the images obtained. Studies have shown that GFR estimated using Gates protocol correlates with GFR measured using plasma sampling method.^{[9]} Gates protocol is commonly used in clinical setting since it is easily available and it does not require multiple blood sampling thus making it more convenient.
In daytoday clinical practice, serum creatinine is used as a marker of kidney function, though this method may be fraught with errors depending on laboratory method of measurement of serum creatinine.^{[10]} A number of serum creatininebased estimating equations have been developed over last four decades to predict GFR. The most commonly used equations are CockroftGault equation normalized for body surface area, fourvariable modification of diet in renal disease (MDRD) equation, and the recently described CKDepidemiology collaboration (CKDEPI) equation. The MDRD study equation and the CKDEPI equation both include variable for age, gender, and race. The CKDEPI equation uses a twoslope “Spline” for the relationship between GFR and age, sex, race, and serum creatinine.^{[11]} When estimated GFR (eGFR) value is <60 ml/min, both the equations were equally accurate. However, at eGFR values of 60–120 ml/min, the CKDEPI equation performed better.^{[11]} However, these equations have been validated in the Caucasian population, and Caucasian patients with CKD. There are very limited data on the performance of the newer CKDEPI equation in Indian population.
We undertook the current study to determine performance of the CKDEPI equation for predicting gold standard GFR as measured by plasma clearance of Tc99m DTPA and comparing it with GFR estimated using CockroftGault equation, 4variable MDRD formula, and Tc99mDTPA imaging using Gates protocol.
Materials and Methods   
Adult persons of either gender (>12 years of age) who had undergone clinical examination and investigations and were advised GFR estimation by plasma clearance of DTPA at the discretion of treating physicians were included in the study. They included healthy subjects who were being evaluated as renal donors as well as patients with stable CKD.
Tc99mdiethylenetriaminepentaacetic acid glomerular filtration rate by plasma sampling and by Gates method
We made sure that the subjects were wellhydrated before the study. They were advised to avoid highprotein diet. The patients were not on any nephrotoxic medications. The height and weight of the patients were noted, for calculation of body surface area. A dose of 10mCi Tc99m DTPA was measured by the “activity method.”^{[12]} A standard solution was prepared as described in the GFR protocol of the nuclear medicine procedure manual of The Ottawa Hospital, Ottawa, Canada.^{[13]} A DTPA renal scan was performed on a dual head gamma camera GEHawkeye4 and the GFRs was calculated by Gates method using vendor provided software. Heparinized blood samples of 7ml each were withdrawn from opposite arm at 2, 3, and 4 h post injection, and the time was recorded. Each of the blood samples was centrifuged for 10 min at 4000 rpm and 100 microliter (µl) of serum was collected in counting tubes with a pipette. An equal amount of standard was pipetted in a counting tube. All the samples plus background were counted for two cycles of 10 min each.^{[12]}
The slopeintercept method with required corrections was integrated in to an Excel program to calculate GFR corrected to body surface area.^{[13]} The GFR thus measured (GFR_{p}) was considered as the gold standard GFR.
Estimated glomerular filtration rate using serum creatinine
Serum creatinine was measured by modified Jaffe's reaction on an EM360 Transasia machine using manufacturer provided reagent.
We calculated GFR by using the following formulae:
 CockroftGault formula corrected for body surface area (GFR_{CG})^{[2]}
 4variable MDRD equation (GFR_{MDRD})^{[2]}
 CKDEPI equation (GFR_{CKDEPI}).^{[11]}
Statistical analysis
We calculated Pearson and Spearman correlation coefficients between GFR_{p } and other methods of estimating GFR, namely CockroftGault formula (GFR_{CG}), fourvariable MDRD equation (GFR_{MDRD}), CKDEPI equation (GFR_{CKDEPI}), and renal scan with Gates method (GFR_{s}). We also calculated the bias as mean difference between GFR_{p } and GFR estimated by each of the other methods. Precision was calculated as standard deviation (SD) of the difference. We calculated the accuracy of each equation as proportion (%) of patients with estimated GFR within 30% of measured gold standard GFR.^{[2]} Bias expresses systematic deviation from the gold standard. Precision expresses the variability (or dispersion) around the bias. Accuracy combines precision and bias. Achieving a high level of accuracy requires low bias and high precision.
We performed BlandAltman analysis and examined whether bias of each method of estimating GFR correlated with the level of true GFR by using Pearson correlation coefficient and linear regression. This provided information whether bias increased significantly at either lower or higher GFR.
In addition, we also examined the subgroup of patients with GFR <60 ml/min to determine the best method of estimating GFR in this subgroup, since this subgroup is more relevant to monitor patients with CKD and for adjusting doses of drugs.
Results   
A total of 105 patients were studied, of which 65 were male and 40 were female. The mean ± SD age was 49.9 ± 16.6 years.
The mean ± SD GFR measured by plasma clearance of Tc99m DTPA was 82.8 ± 29.2 ml/min/1.73 m ^{2}, and it ranged from 12 ml/min/1.73 m ^{2} to 151 ml/min/1.73 m ^{2}. Twenty subjects (19%) had measured GFR ≤60 ml/min/1.73 m ^{2}.
Performance of estimated glomerular filtration rate in the entire group
[Table 1] shows the correlation of GFR_{p } with other methods of predicting GFR. All three serum creatininebased predicting equations correlated better with GFR_{p} than gamma camerabased method. The correlation was greatest with CKDEPI equation (Pearson correlation coefficient 0.7). Bias was least with GFR_{CKDEPI } equation, which also had the best precision among all methods of predicting GFR. MDRD equation had higher bias but better precision as compared to CockroftGault equation.  Table 1: Correlation, bias, and precision of Gates method (GFR_{s}), CockroftGault equation (GFR_{CG}), MDRD equation (GFR_{MDRD}), and CKDEPI equation (GFRCKDEPI) for predicting GFR measured using plasma clearance of Tc99mdiethylenetriaminepentaacetic acid by multiple plasma sampling (GFR_{p})
Click here to view 
[Figure 1] shows scatter plot of GFRs and GFR_{CKDEPI } against GFR_{p}. [Figure 2] shows scatter plot of bias of GFRs and GFR_{CKDEPI } against GFR_{p}.  Figure 1: Scatter plot of estimated glomerular filtration rate by Gates method and by chronic kidney diseaseepidemiology collaboration equation against measured glomerular filtration rate (GFR_{p})
Click here to view 
 Figure 2: Scatter plots of bias for Gates method and chronic kidney diseaseepidemiology collaboration equation
Click here to view 
CKDEPI equation had the highest accuracy (proportion of patients with estimated GFR within 30% of measured gold standard GFR), (73/105, 69.5%) followed closely by that of MDRD equation (65.7%) and CockroftGault equation (64.8%). Gamma camera scanbased Gates method (GFRs) had much lower accuracy (54.9%).
Performance of estimated glomerular filtration rate in subgroup of chronic kidney disease 3–5
We examined subgroup of patients with GFR ≤60 ml/min by plasma sampling, since this subgroup is clinically more relevant for estimating GFR in patients of CKD. In this subgroup also, bias was least with CKDEPI equation, though precision was highest with CockroftGault equation. Gamma camerabased estimated GFR performed worst with highest bias and least precision even in this subgroup. In the subgroup of patients with GFR >60 ml/min, bias and precision was best for CKDEPI equation, whereas MDRD equation had highest bias and gamma camera GFR had the least precision.
Association of bias with measured glomerular filtration rate
Bias with GFR_{CG } had the least slope followed by GFR_{CKDEPI}, both of which were not significant. Thus, bias associated with GFR_{CG } and GFR_{CKDEPI } equations did not depend on true GFR. As against this, bias associated with both GFRs and GFR_{MDRD } had a significant positive correlation with the level of true GFR.
Discussion   
This is the first study to the best of our knowledge to examine the performance of CKDEPI equation in Indian population by comparing it with gold standard measure of GFR, namely plasma clearance of DTPA using multiple plasma sampling. In addition, we also compared CockroftGault equation, 4variable MDRD equation, and Gates protocol with the gold standard to determine the most accurate method of estimating GFR. We found that CKDEPI equation was the most accurate with both least bias and highest precision. Even in patients with Stage 3 to Stage 5 CKD, CKDEPI equation had the least bias. Gates protocol was the least accurate both in the entire group as well as the subgroup of patients with Stage 3 to Stage 5 CKD.
Performance of CKDEPI equation has previously been studied in Asian population. Jeong et al. measured GFR by CrEDTA method in 607 Korean patients and compared the performance of CKDEPI equation with that of MDRD equation.^{[14]} They found that bias was significantly lower with CKDEPI equation, and accuracy of CKDEPI equation was significantly better in patients with GFR >60 ml/min. A study from Pakistan involving 581 participants of age greater than 40 years, also showed that CKDEPI equation had greater accuracy and precision than MDRD equation.^{[15]} Our study confirms better accuracy of CKDEPI equation in Indian population compared to MDRD equation.
Conflicting data exist regarding accuracy of Gates method of estimating GFR. Prasad et al.^{[9]} studied 897 subjects of all levels of GFR and compared MDRD equation and Gates method with twosample plasma clearance of Tc99m DTPA. They found that Gates method had better correlation with measured GFR than MDRD equation at all levels of GFR. Hephzibah et al.^{[16]} found that in voluntary kidney donors, Gates method had poor correlation with GFR measured by twosample plasma clearance method (r = 0.27), whereas CockroftGault formula had somewhat better correlation (r = 0.36). We used threesample plasma clearance method, which may be more accurate than twosample method for measuring GFR by plasma clearance. Our data using threesample method suggests that creatinine based estimating equations, especially the newer CKDEPI equation is superior to Gates method for estimating GFR.
Our study has several limitations. Number of patients with GFR <60 ml/min/1.73 m ^{2} was limited. Thus, caution is warranted while applying these findings in advanced kidney disease. We did not include special populations such as patients with liver disease, advanced heart failure, or advanced malignancy; thus applicability of our finding in these groups will need further study.
Conclusion   
CKDEPI equation is the best predictor of GFR in Indian population. MDRD equation and CockroftGault equation also performed better than Gates method based on renal scintigraphy. Gates method may not be suitable for estimating GFR in potential kidney donors and patients with CKD.
Acknowlegement
We sincerely thank Alan Thibeau, Professional Practice Manager, Medical Radiation Technology, Department of Medical Imaging, The Ottawa Hospital, Ottawa, Canada and Dr. Greg Knoll, Medical Director of Renal Transplantation, Professor of Medicine, Division of Nephrology, University of Ottawa, Canada for their help in establishing the protocol for measurement of GFR by plasma clearance.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References   
1.  Smith H. W. The Kidney Structure and Function in Health and Disease. New York: Oxford University Press; 1951. The reliability of inulin as a measure of glomerular filtration; pp 231238. 
2.  National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am J Kidney Dis 2002;39 2 Suppl 1:S1266. 
3.  Garnett ES, Parsons V, Veall N. Measurement of glomerular filtrationrate in man using a 51Credeticacid complex. Lancet 1967;1:8189. 
4.  BröchnerMortensen J, Giese J, Rossing N. Renal inulin clearance versus total plasma clearance of 51CrEDTA. Scand J Clin Lab Invest 1969;23:3015. 
5.  Chantler C, Garnett ES, Parsons V, Veall N. Glomerular filtration rate measurement in man by the single injection methods using 51CrEDTA. Clin Sci 1969;37:16980. 
6.  Chantler C, Barratt TM. Estimation of glomerular filtration rate from plasma clearance of 51chromium edetic acid. Arch Dis Child 1972;47:6137. 
7.  BröchnerMortensen J. A simple method for the determination of glomerular filtration rate. Scand J Clin Lab Invest 1972;30:2714. 
8.  Biggi A, Viglietti A, Farinelli MC, Bonada C, Camuzzini G. Estimation of glomerular filtration rate using chromium51 ethylene diamine tetraacetic acid and technetium99m diethylene triamine pentaacetic acid. Eur J Nucl Med 1995;22:5326. 
9.  Prasad N, Barai S, Gambhir S, Parasar DS, Ora M, Gupta A, et al. Comparison of glomerular filtration rate estimated by plasma clearance method with modification of diet in renal disease prediction equation and gates method. Indian J Nephrol 2012;22:1037. [Full text] 
10.  Miller WG, Myers GL, Ashwood ER, Killeen AA, Wang E, Thienpont LM, et al. Creatinine measurement: State of the art in accuracy and interlaboratory harmonization. Arch Pathol Lab Med 2005;129:297304. 
11.  Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF rd, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:60412. 
12.  Fleming JS, Zivanovic MA, Blake GM, Burniston M, Cosgriff PS; British Nuclear Medicine Society. Guidelines for the measurement of glomerular filtration rate using plasma sampling. Nucl Med Commun 2004;25:75969. 
13.  White C, Akbari A, Hussain N, Dinh L, Filler G, Lepage N, et al. Estimating glomerular filtration rate in kidney transplantation: A comparison between serum creatinine and cystatin Cbased methods. J Am Soc Nephrol 2005;16:376370. 
14.  Jeong TD, Lee W, Chun S, Lee SK, Ryu JS, Min WK, et al. Comparison of the MDRD study and CKDEPI equations for the estimation of the glomerular filtration rate in the Korean general population: The fifth Korea National Health and Nutrition Examination Survey (KNHANES V1), 2010. Kidney Blood Press Res 2013;37:44350. 
15.  Jessani S, Levey AS, Bux R, Inker LA, Islam M, Chaturvedi N, et al. Estimation of GFR in South Asians: A study from the general population in Pakistan. Am J Kidney Dis 2014;63:4958. 
16.  Hephzibah J, Shanthly N, Oommen R. Comparison of glomerular filtration rate measured by plasma sample technique, Cockroft gault method and gates' method in voluntary kidney donors and renal transplant recipients. Indian J Nucl Med 2013;28:14451. [ PUBMED] [Full text] 
[Figure 1], [Figure 2]
[Table 1]
