Comparisons of Five Equations Used for Estimating the Glomerular Filtrations Rate in Chinese Adults

We collected blood biochemical data of 623 chinese adult hospitalised patients within 48 hours before they underwent 99m Tc-DTPA GFR measurement. We computed the bias (mGFR-eGFR), the precision (IQR) ,the accuray (P30and root mean square error (RMSE) relative to mGFR of each equation to evaluate performance. The ROC curves, Kappa value of McNemar test, Bland-Altman plot and the Intraclass correlation coefficient (ICC) were used to evaluate diagnostic accuracy and concordance. severe heart pleural effusion, renal stenosis, urinary obstruction, kidney tumour, severe malnutrition,


Introduction
Chronic kidney disease (CKD) is a global medical problem with a high incidence that affects a large proportion of the world's population [1]. The current prevalence of CKD is approximately 13.4% according to a global systematic review [2]. However, the prevalence of CKD varies widely both within and between countries. On the one hand, there are indeed regional differences in the prevalence of CKD. On the other hand, the diagnosis of CKD using eGFR, the different GFR thresholds for CKD in the elderly, and the use of one-time testing for eGFR or urine protein results to evaluate CKD all lead to different prevalence rates in large-scale epidemiological studies [3]. Population-based studies indicate that CKD epidemiology differs by sex, affecting more women than men, especially with regard to stage G3 CKD. The difference in CKD epidemiology by sex may be due to the natural decline in glomerular filtration rate (GFR) with age, as well as the potential overdiagnosis of CKD through the inappropriate use of GFR equations [4]. Thus, quick and accurate assessment of GFR is important for classifying the stage of chronic kidney disease because it can help assess the patient's current renal function, determine the dose of clinical drugs, and improve judgements regarding the reasonable timing of kidney dialysis or transplantation.
Presently, technetium-99m-diethylenetriaminepentaacetic acid ( 99m Tc-DTPA) renal dynamic imaging is recommended for the measurement of GFR by the Nephrology Committee of the Society of Nuclear Medicine [5] and has been widely accepted as the standard value of GFR in clinical practice [6]. The calculation of the glomerular filtration rate in clinical practice mainly relies on the renal biomarkers serum creatinine (Scr) and serum cystatin C (ScysC). Several studies have shown that ScysC is a better marker than Scr [7][8][9] and is less affected by muscle mass and nutritional status. Current research confirms that within-subject variability in estimated GFR was lower than measured GFR, which indicates that estimates of GFR are at least as reliable as measured GFR for monitoring patients over time [10]. According to the epidemiology of mGFR in China, China has a large population base, with obvious differences in sex, region and race compared to these parameters in white people [11]. Therefore, it is necessary to evaluate the appropriate GFR equation for the Chinese population.
Since the traditional Cockcroft-Gault (CG) equation was widely applied in clinical practice in 1969, the exploration and correction of formulas are progressing consistently. With the standardization of the creatinine and cystatin C determination methods [12,13], the traditional calculation formula is obviously not applicable in the current clinical setting, so the Revised Modification of Diet in Renal Disease (MDRD) equation has emerged, incorporating standard creatinine measured by Roche enzyme assay dating back to isotope dilution mass spectrometry (IDMS) calibration [14]. The 2012 KDIGO guidelines recommend Scr-based and ScysC-based equations and the combined creatininecystatin C Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation used to calculate eGFR for adults [15]. Subsequently, correction can be performed according to Asian race [16].
Since the establishment of the guideline recommendations, there has been progress in the exploration of formulas in recent years, including the Berlin Initiative Study (BIS) equation for people over 70 [17], the full age spectrum (FAS) equation [18,19] (calibrated in Korea [20]), and the latest XiangYa equation based on the Chinese population [21]. At present, the applicability of various equations in Chinese adults still needs to be verified by a large multicentre population study. The purpose of this study is to compare the applicability of current authoritative and popular equations in the population in central China.

Study population
We performed a retrospective study by reviewing the electronic medical records of inpatients from June 2017 to July 2019 at Zhongnan Hospital of Wuhan University (Wuhan, China). The inclusion criteria were as follow: patients who underwent GFR measurement using 99m Tc-DTPA; age ≥ 18 years; and Scr and ScysC concentrations measured within 48 hours before the time measuring 99m Tc-DTPA mGFR. We exclude the participants with acute renal failure, severe heart disease, pleural or abdominal effusion, renal stenosis, urinary obstruction, kidney tumour, severe malnutrition, physical disabilities, during pregnancy or lactation and patients who were taking trimethoprim, cimetidine, cefoxitin or underwent hemodialysis previously. Among the initial 681 selected inpatients, 19 had incomplete data for mGFR(unilateral renal insufficiency or without surface area calibration), 17 lacked ScysC concentration, 14 lacked Scr concentration measured by Jaffe's alkaline picrate method , 8 subjects with Scr concentration too high (Scr levels > 700umol/l). Finally, a total of 623 subjects were included in the study. The detailed characteristics of the study subjects are presented in Table 2.

Laboratory assay
We measured serum creatinine with use of Roche enzymatic method and IDMS calibration ( Beckman AU5800 Coulter chemistry analyzers, Ltd, Tokyo, Japan), traceable to National Institute of Standards and technology(NIST) creatinine standard reference material(SRM 967) [13], with a reported coefficient of variation of 2.3%(reference range:64-104umol/L). we measured serum cystatin C with use of particle-enhanced nephelometric immunoassay(Beckman AU5800 Automatic biochemical analyzers, Ltd, Tokyo, Japan),traceable to international standards [22], with a reported coefficient of variation of 10%(reference range:0-1.2 mg/L). mGFR was measured by using the radioactive isotope 99m Tc-DTPA, Patients were required to have an fasting state on the day of examination, empty the bladder, drink 500 ml water within 30 minutes before examination, get a bolus injection of 5mCi 99m Tc-DTPA via the cubital vein projectile, We perform renal dynamic imaging by a single photon emission computed tomography (CT) scanner, and obtain standardized mGFR by Gates method [23] according to the patient's weight and height.  Table 1.

Statistical analysis
Continuous variables were performed as mean±standard(SD),categorical variables were performed as percentage(%). Normally distributed groups of variables were computed with t-test, Skewed distributed variables were tested using nonparametric tests Wilcoxon and Mann-Whitney U, Categorical variables were computed with chi-square test, P < 0.05 were considered statistically significant. 99m Tc-DTPA mGFR was taken as reference value of actual GFR, Bias was assessed as the median difference between mGFR and eGFR(mGFR-eGFR). Precision was assessed as the interquartile range(IQR) for the differences. Accuracy was assessed as root mean square error (RMSE) relative to mGFR and the percentage of eGFR deviating within 30% of mGFR(P30),which was proposed in K/DOQI clinical practice guideline [24]. Receiver-operating characteristic (ROC) curves were computed for CKD diagnosis according to mGFR and eGFR (CKD are diagnosed based on 2012 clinical practice guideline [15]). Concordance between mGFR and eGFR were compared using the Intraclass correlation coefficient (ICC). The Bland-Altman plot analysis was used to calculate mean difference and precision between mGFR and eGFR. McNemar test and kappa value was used to assess the CKD stages agreement between mGFR and eGFR categories. Subgrouds were based on mGFR<60 or ≥ 60 ml/min/1.73 m 2 and age < 70 or ≥ 70 years old. All statistical analysis were computed using SPSS software(version22.0; SPSS Chicago, IL, USA) and GraphPad Prism for Windows(version7.04; GraphPad software, San Diego, California, IL, USA).

Basic characteristics
Of the 623 patients, 383 were male and 240 were female. The mean age was 56.89 years. There were 480 adults under 70 years old and 143 elderly people over 70 years old. The mean height and weight were 166.13 meters and 66.09 kg. The average values of Scr and cysC were 1.75 mg/dl and 1.74 mg/l, respectively. The primary cause of CKD was chronic glomerulonephritis (33.5%), followed by hypertensive nephropathy (30.2%) and diabetic nephropathy (15.2%), which was consistent with the world epidemiological survey of chronic kidney disease [25]. In the elderly group over 70 years old, the most common primary disease was hypertensive nephropathy (47.6%), followed by chronic nephritis (21.7%) and diabetic nephropathy (16.1%). The mean 99m Tc-DTPA mGFR was 49.31 ml/min/1.73 m 2 , and it was 53.01 ml/min/1.73 m 2 in adults under 70 years old and 36.86 ml/min/1.73 m 2 in the elderly population over 70.

Performance results of different equations in all subjects
The bias, precision and accuracy of each eGFR equation computed for all subjects are shown in Table 3 19.89 Bias was median difference between mGFR and eGFR; IQR was the inter-quartile range of the difference; P30 was the proportion of eGFR within 30% of mGFR, RMSE was root mean square error relative to mGFR; *P < 0.05, **P < 0.01, compared with CKD-EPI Scr

Performance results of different equations in subgroups
The performance of each eGFR equation in the CKD subgroups and age subgroups is shown in Tables 4 and 5. mGFR ≥ 60 ml/min/1.73 m 2 was defined as slightly reduced GFR, and mGFR < 60 ml/min/1.73 m 2 was defined as moderately-to-severely reduced GFR. The results of the subgroup analyses differed from the results of all subjects. In subgroups with mGFR ≥ 60 ml/min/1.  Bias was median difference between mGFR and eGFR; IQR was the inter-quartile range of the difference; P30 was the proportion of eGFR within 30% of mGFR, RMSE was root mean square error relative to mGFR; *P < 0.05, **P < 0.01, compared with CKD-EPI scr Bias was median difference between mGFR and eGFR; IQR was the inter-quartile range of the difference; P30 was the proportion of eGFR within 30% of mGFR, RMSE was root mean square error relative to mGFR; *P < 0.05, **P < 0.01, compared with CKD-EPI scr
The ideal Q value that is applicable should match the different characteristics of the Chinese population [11]. Second, the FAS combined equation assigns Scr/Q Scr and cysC/Q cysC according to weight. The weights of the two factors in this study each account for 50%. In fact, the optimal proportion of creatinine and cystatin C for the calculation results is not yet conclusive. However, regardless of the development of the CKD-EPI equation or FAS equation, the combination of creatinine and cystatin C could reduce the imprecision of either biomarker alone [19,31]. Third, we found that the CKD-EPI cysC equation obviates the race/ethnicity coefficient. However, the calibration coefficients of the various equations in the Chinese race are all greater than 1 [32], and there is a paradox between them. Therefore, it is necessary to carry out multicentre and multiethnic equation verification for Chinese people.
Creatinine is produced by muscle catabolism and secretion by kidney tubular cells. Its concentration is greatly affected by the body's nutritional status and muscle mass. Poor nutritional status, infection along with an inflammatory state, or disability will affect the production of creatinine, leading to an inability to accurately assess renal function, especially between sexes. Cystatin C is less affected by In this study, we also found that the CKD-EPI cysC equation performed well in elderly individuals over 70 years of age. According to the law of GFR decline with age, whether GFR < 60 ml/min/1.73 m 2 is used as the diagnostic criterion for CKD in the elderly population is controversial, which was proposed in the BIS equation study [17]. Epidemiological studies have shown that elderly patients over 65 years old account for 53.07% of the CKD population in China [36].  [21].
In this study, the accuracy P30 of each equation was found to be less than 70%. This is because the plasma biochemical indexes included in this study were from 48 hours before kidney nuclear radiograph, and it was not necessarily the same result as in the morning. This was in line with the current procedures of most hospitals in China. Due to the large population in China, the imbalance between doctors and patients, and limited medical resources, biochemical blood tests and imaging examinations are often unable to be synchronized. Using biochemical test results from within 48 hours to assess renal function is not unacceptable for CKD. In recent reports, the rate of change in biomarkers was shown to be smaller than that in kidney nuclear radiography. The eGFR computed by equations is at least as reliable as the mGFR measured by nuclear medicine for monitoring patients over time [10]. The comparative verification data adopted in this study are more applicable to China's national conditions, and its statistical results have more practical clinical significance.
We verified the effect of the current popular equation on the evaluation of renal function in adult patients in central China. We must take into account the limitations of this study. First, we only enrolled Chinese Han individuals in this single-centre experiment. China is a multi-ethnic and multiregional country. Different living customs lead to diverse human physiques, and different medical conditions contribute to inconsistent diagnoses of CKD in various regions. We will consider involve more ethnic groups in our multi-center verification research. Second, the method of assaying serum creatinine and cystatin C in this study may be different from those used in the derivation studies for the above equations, so was the 99m Tc-DTPA used for mGFR measurement. Although the difference may exist in assaying equipment and materials, the current assaying level can be traced back to international certification standards, as a consequence its results are acceptable. Systematic assessment can be performed to evaluate the influence of different assaying methods on the performance of eGFR equation forward. Third, this was a retrospective validation study, and the concentration of blood biochemical indexes came from one-time testing. We were not able to evaluate the accuracy of eGFR assessed by each equation in terms of a dynamic change in renal function. Considering that the renal function of CKD patients does not change significantly in the short term, it is advisable to evaluate the calculation results of patients in hospitalization and discharging together for a comprehensive analysis in the future study.

Conclusion
The FAS equation based on the combination of Scr and cysC is considered more suitable and simpler for the Chinese population. The CKD-EPI cysC equation is most accurate in individuals with moderately and severely reduced GFR and seniors over 70 years old. The Chinese-developed XiangYa equation is appropriate for normal or mildly injured GFR; however, its prediction accuracy still needs to be verified in individuals with moderately or severely reduced GFR and older adults. Declarations obtained from each subject prior to participation.

Funding
There was no funding obtained for the trial. It was performed in the Zhongnan Hospital of Wuhan University.

Consent for publication
Not applicable.  Figure 1 The ROC curve of equations for diagnose CKD in all subjects The Bland-Altman plot of comparision between eGFR and mGFR