Optimal Glomerular Filtration Rate Equations for Various Age Groups, Disease Conditions and Ethnicities in Asia: A Systematic Review

(1) Background: The performance of estimated glomerular filtration rate (eGFR) equations in the Asian population has been widely questioned. The primary objective of this study was to gather evidence regarding optimal GFR equations in Asia for various age groups, disease conditions, and ethnicities. The secondary objective was to see whether the equations based on the combination of creatinine and cystatin C biomarkers if employed are satisfactory across different age groups and disease conditions in various ethnicities in Asia compared to those based on either of the single biomarkers. (2) Methods: Validation studies that had both creatinine and cystatin C-based equations either alone or in combination, validated in specific disease conditions, and those which compared the performance of these equations with exogenous markers were eligible only. The bias, precision, and 30% accuracy (P30) of each equation were recorded accordingly. (3) Results: Twenty-one studies consisting of 11,371 participants were included and 54 equations were extracted. The bias, precision, and P30 accuracies of the equations ranged from −14.54 to 9.96 mL/min/1.73 m2, 1.61 to 59.85 mL/min/1.73 m2, and 4.7% to 96.10%. The highest values of P30 accuracies were found for the JSN-CKDI equation (96.10%) in Chinese adult renal transplant recipients, for the BIS-2 equation (94.5%) in Chinese elderly CKD patients, and Filler equation (93.70%) also in Chinese adult renal transplant recipients. (4) Conclusions: Optimal equations were identified accordingly and it was proven that combination biomarker equations are more precise and accurate in most of the age groups and disease conditions. These can be considered equations of choice for the specific age groups, disease conditions, and ethnicities within Asia.


Introduction
Accurately estimating renal function plays a critical role in diagnosing, staging, and managing chronic kidney disease (CKD) [1]. Glomerular filtration rate (GFR) has gained the status of indicator of choice for assessment of renal function whereas clearance of inulin administered by infusion and its sampling in urine has become a "gold standard" for measuring GFR [2]. Nowadays, the plasma clearance of other exogenous filtration markers e.g., 99mTc-DTPA has been approved to exhibit GFR adequately [3,4] They are easier to perform in comparison to inulin but still, it is laborious work to do. The current estimation of GFR takes into account the individual's race, sex, age, and weight as non-determinants of GFR for serum creatinine (Scr) [5][6][7], as the production of Scr is proportional to these variables as well as to the muscle mass [8]. However, the utilization of Scr for the estimation of GFR is not a method of choice when the level of renal impairment reaches 50% or less [9].

Protocol
The systematic review's protocol has been developed according to Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [26] guidelines and registered in PROSPERO (International Prospective Register of Systematic Reviews; https: //www.crd.york.ac.uk/prospero, registration number: CRD42021276236). However, a meta-analysis was not carried out and 14 elements of the PRISMA checklist are marked as "N/A". Performing a meta-analysis was not feasible as most of the studies included were clinically diverse and had a high risk of bias which will produce misleading results ("See Supplementary Data Table S1").

Eligibility Criteria
We included validation studies that had both creatinine and cystatin C-based equations, either alone or in combination, to estimate GFR in any Asian populations, where validated samples with specific disease conditions, had P30 values of accuracy, and those which compared the accuracy of these equations with measured GFR (mGFR) by exogenous markers (99mTc-DTPA, 51Cr-EDTA, Inulin, and Iohexol) as the gold standard.
The exclusion criteria were the absence of accurate results to evaluate the performance of equations, studies carried out in the general population without specific disease conditions, use of urinary cystatin C or creatinine instead of serum cystatin C or creatinine as GFR marker, review studies, and animal studies. Original validation studies written only in English were considered.

Search Strategy and Study Selection
A systematic search of Pubmed, EMBASE, Cochrane Library, and Web of Science was conducted from 2012 to 2020 by two independent reviewers. The search strategy using a wide range of Medical Subject Headings (MeSH) was as follows: "("Creatinine and Cystatin C" All records were assessed by independent reviewers based on titles and abstracts for inclusion in this review. Abstracts that do not match the inclusion criteria or those that matched the exclusion criteria were discarded. The records that remained and those with abstracts providing adequate information for deciding upon their inclusion were considered for full-text evaluation which was performed independently by the same reviewers. Disagreements were solved by a third reviewer.

Data Collection and Extraction
The selected studies were analyzed by two investigators and the data were extracted by utilizing the standardized system accordingly. The following information was gathered: first author, publication year, sample size, measured GFR (mGFR) value, population, age group, and disease condition in which equations were validated, evaluated creatinine and cystatin C-based equations, reference method for mGFR, creatinine, and cystatin C measurement methods, and whether or not they were traceable to the reference method, bias, precision, and 30% accuracy (P30) of the equation with measured GFR. The populations in the selected studies were categorized according to age groups and specific disease conditions. Age groups were defined on the basis of age ranges; Children + Adolescents = ≤18 years, Adults = 18-60 years, and Elderly = ≥60 years. The optimal equations for each category were then identified on the basis of pre-specified criteria given in KDIGO guidelines (bias, precision, and P30) [27]. Finally, optimal equations were suggested for clinical application. In order to assess the clinical applicability of each suggested equation, disease conditions, age groups, and ethnicities were predicted on the basis of validation studies whereas the clinical settings in which an equation can be applied were approximated on the basis of the study population from which individual equation was derived.

Risk of Bias in Individual Studies and Quality of Systematic Review
The quality and risk of bias of studies were evaluated with the Quality of Diagnostic Accuracy Studies-2 (QUADAS-2) tool [28].

Diagnostic Accuracy Measures and Selection of Optimal Equations
Following the guidelines of KDIGO, accuracy was defined as P30, the percentage of eGFR values within 30% of measured GFR. The bias (mean or median difference) and precision (standard deviation, SD; or interquartile range, IQR of difference) among eGFR and mGFR were recorded accordingly. Optimal equations were selected on the basis of these diagnostic accuracy measures outlined by KDIGO guidelines [27].

Diagnostic Accuracy Measures and Selection of Optimal Equations
Following the guidelines of KDIGO, accuracy was defined as P30, the percentage of eGFR values within 30% of measured GFR. The bias (mean or median difference) and precision (standard deviation, SD; or interquartile range, IQR of difference) among eGFR and mGFR were recorded accordingly. Optimal equations were selected on the basis of these diagnostic accuracy measures outlined by KDIGO guidelines [27].
The populations in the selected studies were categorized according to age groups, specific disease conditions, and ethnicities: Four of them were performed on elderly CKD patients (Four Chinese [48,50,52,54]), eleven in adult and elderly CKD patients (eight Chinese [9,46,55,59,60,[62][63][64], one Japanese [51], one Indian [53], and one multiethnic study [57]), and two in adult renal transplant recipients (one Chinese [56], and one Korean study [61]). Whereas only single studies were found among Chinese children and adolescents with renal injury [65], Chinese adult patients with obstructive nephropathy [49], Japanese adult cirrhotic patients [47], and Chinese adult CKD diabetic and non-diabetic patients [58]. Among the methods used for measurement of serum biomarkers, eleven studies reported the use of immunoturbidimetry, seven studies used immunonephelometry, one used latex enhanced immunoturbidimetric, one used colloidal gold immunoassay and one did not report the method used. The bias (mean or median difference), precision (SD or IQR of difference,) and P30 accuracies of the equations ranged from −14.54 to 9.96, 1.  Table 2 Additionally, the optimal equations have also been suggested along with their clinical applicability in Table 3.

Optimal Equations for Elderly CKD Population
As far as the performance of various equations in Chinese elderly CKD patients is concerned, Guan Changjie et al. demonstrated that a creatinine and cystatin C combination equation, BIS-2 (Berlin Initiative Study) has a better performance compared to MDRD and CKD-EPI equations in estimating glomerular filtration in elderly CKD patients [48]. Similar findings were observed in another validation study performed by them which included 368 elderly CKD patients which also supported the optimal performance of the BIS-2 equation at GFR 30 mL/min/1.73 m 2 or greater in comparison to Feng-Cr-CysC, CKD-EPI-Cr-CysC, and MA-Cr-CysC equations. However, the CKD-EPI-Cr-CysC equation yielded better performance in patients with measured GFR less than 30 mL/min/1.73 m 2 [50].
Similarly, Huang et al. screened different equations which have previously shown high accuracy among the Chinese elderly population. In this study, CG (at mGFR ≤ 60 mL/ min/1.73 m 2 ), Modified CKD-EPI-Cr-CysC for elderly (at mGFR ≥ 60 mL/min/1.73 m 2 ), Standardized SCr and SCysC CKD-EPI (at mGFR ≥ 60 mL/min/1.73 m 2 ) showed optimal performance [52]. On the other hand, modified CKD-EPI-CysC for the elderly and modified CKD-EPI-Cr-CysC for the elderly by Fen li et al., were validated in Chinese which proved the higher accuracy of the modified equations than the original one in the elderly CKD population [54]. So, these identified optimal equations perform best in the Chinese elderly CKD population and should be further validated in other Asian countries and ethnicities in elderly CKD patients.

Optimal Equations for Both Adult and Elderly CKD Population
Min Yang et al. performed a validation study on 632 adult and elderly Chinese CKD patients in which CKD-EPI equations were validated. This study demonstrated the reliability of CKD-EPI-Cr-CysC and CKD-EPI-CysC for assessing CKD stages correctly in these individuals. CKD-EPI-Cr-CysC equation has shown particularly better accuracy and diagnostic value in participants with normal or mildly impaired GFR, while the CKD-EPI-CysC equation performed better in CKD stages 3-4 [59]. These results were consistent with Hua Chi et al. study which showed that the employment of a combination of cystatin C and serum creatinine (CKD-EPI-Cr-CysC) levels show improvement in the bias of equation and achieves greater diagnostic accuracy in patients with renal insufficiency [46]. Another validation study was performed by Kumar et al., where the cystatin-C-based equation showed superior performance in the adult CKD population. Stratification by measured GFR and by gender or age did not change the results. This equation showed P30 values of 81.5% and 69.7% for those having measured GFR ≥60 mL/min/1.73 m 2 and <60 mL/min/1.73 m 2 , respectively [53].
In addition, two prospective cohorts were considered from the Chinese population, and three new equations C-CKD-EPI-Cr, C-CKD-EPI-CysC, and C-CKD-EPI-Cr-CysC were developed in one and validated in the other. The best accuracy with the highest P30 value was depicted by C-CKD-EPI-CysC with relatively lower precision but C-CKD-EPI-Cr-CysC depicted improved accuracy, bias, and precision [60]. C-CKD-EPI-Cr-CysC had also shown the lowest bias in CKD stage three to five among adults and elderly in another validation study by Yue L. et al. [64]. This study also showed optimal performance of two other equations; Xiangya and CKD-EPI-CysC. Feng et al. [9] and Horio et al. [51] also validated eGFR equations in the CKD population and found that Feng-Cr-CysC, Feng-CysC, and CKD-EPI-CysC were ideal equations. In addition, Teo BW et al. [57] performed validation in a multiethnic Asian population with CKD and found that standardized SCr and SCysC CKD-EPI is the best-performed equation for the multiethnic Asian population.
The eGFR equations have also been validated by Xiaoshuang Ye et al. in the Asian population [62]. Among all equations, the Feng-Cr-CysC equation achieved the best performance, although still requires improvement when applying to GFR less than 60 mL/min per 1.73 m 2 in clinical settings. A favorable performance was achieved by the Feng-CysC equation too, which was only less efficient than the Feng-Cr-CysC equation On the other hand, previous studies in Asia have demonstrated the inaccuracy of CKD-EPI-Cr-CysC and CKD-EPI-CysC among aging cohorts with GFR at a moderately severe level of impairment. Thus, the assessment of adaptability and performance of the new FAS equation was assessed in a validation study performed by Yong et al. [63] which proved that the FAS-Cr-CysC equation has superior diagnostic accuracy among whole subjects (in the subgroup rGFR less than 60 mL/min per 1.73 m 2 ), particularly in elderly patients with GFR at a moderately severe level of impairment. Another validation study performed by Pei et al. [55] proved the accuracy of modified MDRD by Pei, modified CKD-EPI by Pei, and Pei (Modified Maclssac) in the adult and elderly Chinese CKD population.
Hence, these identified optimal equations are ideal for implementation in adult and elderly CKD patients in their respective ethnicities in which they have been validated. In addition, these identified equations can be studied in the future and validated in other countries and ethnicities within Asia in CKD patients.

Optimal Equations for Obstructive Nephropathy and Kidney Transplant Recipients
A validation study performed by Chen et al. [49] in Chinese has proved that the CKD-EPI-Cr-CysC equation showed the best performance in obstructive nephropathy patients too. CKD-EPI-CysC equation showed high accuracy and lesser bias in individuals with the lowest GFR. In patients with high GFR, CKD-EPI-Cr-CysC was the best in performance, although both CKD-EPI-CysC and CKD-EPI-Cr-CysC had a huge bias in these patients. In addition, the bias of all the equations was larger in women as compared to men. However, CKD-EPI-Cr-CysC had the highest accuracy for both males and females.
A validation study performed by Yang et al. [61] in the Korean population exhibited better performance of cystatin C equations in kidney transplant recipients with lesser GFR. CKD-EPI-CysC exhibited the least bias and highest precision in patients with measured GFR less than 45 mL/min/1.73 m 2 and all other equations underestimated measured GFR significantly.
In another study performed by Tang et al., [56] in Chinese kidney transplant recipients, JSN-CKDI, Larsson, Rule, 2003 Hoek, Filler, and Grubb equations exhibited the best performances. In this study, it was observed that almost all the equations have elevated net biases when the GFR value was greater than 60 mL/min/1.73 m 2 . It has also been indicated that GFR equations are unsuitable for renal transplant recipients whose GFR values are at a high level.

Optimal Equations for Diabetic CKD Population
It has always been controversial to select an optimal equation for the estimation of GFR in CKD patients with diabetes. Hence, the lower accuracy of GFR estimating equations in diabetic CKD patients led to another Chinese validation study performed by Xie et al. [58] on 215 diabetic CKD patients and 192 non-diabetic CKD patients. Here too, CKD-EPI-Cr-CysC exhibited the best performance among all CKD-EPI equations. However, BMI, CKD status, mGFR, and HbA1c are considered independent factors which are associated with the accuracy of eGFR equations.

Optimal Equations for Renal Injury Children
The glomerular filtration rate is vital for the evaluation of renal function and classification of CKD in children, while the reference technique in children is inconvenient [66]. In Chinese children, the non-availability of data about GFR measurement by renal or plasma clearance of exogenous markers has led to the unavailability of validated estimating tools for GFR in this population. A validation study carried out by Zheng et al. [65] in this age group showed an accurate estimation of GFR by cystatin-C-based CKiD and Filler equations. These equations can be employed as equations of choice in Chinese renal injury children and can further be validated in renal injury children in other populations and ethnicities within Asia.

Optimal Equation for Liver Cirrhosis Population
The assessment of kidney function is of significant importance in managing patients with cirrhosis. Creatinine-based estimating equations do not depict the true renal function because of impairment of liver function and muscle wasting, although serum creatinine is used routinely for this purpose. Cystatin C, by contrast, is not related to liver function and muscle volume. In a Japanese validation study performed by Adachi et al. [47] cystatin C-based equations were examined for assessing the renal function in Japanese cirrhotic patients. The outcome suggested that the eGFR-CysC equation estimated the renal function and predicted the results in a more accurate way as compared to Cr-based estimating equations in patients with cirrhosis. Hence, eGFR-CysC can be considered best while evaluating kidney function in the cirrhotic population.

Discussions
Estimated glomerular filtration rate equations, either alone or with a combination biomarker, in the selected studies, were validated in different ethnicities including Chinese, Japanese, Korean, Malay, and Indian populations. Our review assists clinical researchers, nephrologists, and other clinicians who can apply the identified optimal equations (Table 3) in routine practice to make accurate treatment decisions relying upon the GFR estimation in the Asian population in the respective age groups, ethnicities, and disease conditions. The results of our review emphasize the use of cystatin C and creatinine combination equations for estimating GFR in a multiethnic Asian population in most of the age groups and disease conditions which include chronic kidney disease, diabetes, renal injuries, obstructive nephropathy, renal transplant, and liver cirrhosis ( Table 2). Although equations based only on cystatin C have also shown ideal performances in some of the included studies (Table 2).
Previous studies have also implied the fact that serum creatinine-based equations underestimate and serum cystatin C-based equations overestimate GFR, and the average yield the best estimate [67,68]. Our review has depicted similar findings across different age groups, disease conditions, and ethnicities in Asia. This may be related to factors responsible for the generation of these biomarkers. Serum creatinine generation reflects the amount of lean muscle mass [69]. Body mass index influences serum cystatin C generation [70]. It has been suspected that the degree of ethnic and environmental influences on fat and muscle proportion is rendered less important once the GFR estimates are averaged using a combination of these two markers [57]. Thus, our review also supports the idea that ethnicity adjustment is not required for the combination biomarker equation in multiethnic Asian populations. However, the implementation of cystatin C estimation is difficult in developing countries of Asia because of its associated cost especially for monitoring CKD without any improvement in risk prediction [71] as well as standardization issues for calibration. This has led to the limited use of cystatin C-based equations in the Asian popu-lation and steps needs to be taken to generalize the utilization of such useful biomarker in Asian laboratories.
Some limitations of our review must be taken into account. First, there is a lack of reliable methods for assessment of the risk of publication bias in such systematic reviews as ours which are based on observational studies. Despite the thorough research of the literature, the likelihood of such bias in the present review is not known. Secondly, most of the studies (77%) included in this systematic review came out from the Chinese population. There is a lack of potential validation studies being carried out in other Asian countries for eGFR equations. Furthermore, there is a rigorous need for external validation studies for eGFR equations in Asian countries. Moreover, data is confined to the Asian population only. Thirdly, the results of this review could have high statistical and clinical heterogeneity as the comparisons have limitations in the way in which the validation studies were carried out and the methodology of data reporting in these studies. Moreover, the participants were not recruited from community settings in the majority of included studies. Another limitation of our review includes the use of DTPA scintigraphy (Gates method) in the studies included, although this method is not a reference method as it cannot provide convincing data to support the preferred use of any equation as this method is affected by many factors other than GFR itself and can only be used to determine relative GFR [72]. Additionally, the 99mTc-DTPA was utilized as the reference method for mGFR in almost 80% of studies included in this study. Studies have shown that this method is not suitable for the gold standard for mGFR measurement [73][74][75]. Therefore, errors might exist in the equation evaluations. This systematic review has summarized and suggested the optimal GFR equations for various age groups and disease conditions in Asian populations. This review has illustrated that cystatin C and creatinine combination equations are more precise and accurate in most of the age groups, disease conditions, and ethnicities than equations that are based on either of the biomarkers. Although equations based on cystatin C alone were also ideal in certain age groups and disease conditions. Nephrologists, clinical researchers, and other clinicians can apply suggested equations for accurate decision making and suitable patient outcomes can be achieved in the respective age group, disease condition, and ethnicity in which they have been identified. There is a huge heterogeneity within the continent. Hence, in order to support the findings of this review, further validation studies are required for these optimal equations within other ethnicities and races in Asia in their respective age groups and disease conditions in which they have been identified.
Author Contributions: W.A. conceived, supervised, and conducted the study. A.S. searched, screened, extracted, analyzed the data and wrote the initial draft of the paper. W.A., M.A.K. and S.M. reviewed, edited and approved the paper. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data presented in this study are available in supplementary material.

Conflicts of Interest:
All the authors declared no competing interest. The authors had full responsibility for data collection, data interpretation, and writing of the report.