Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Multicenter Study of Creatinine- and/or Cystatin C-Based Equations for Estimation of Glomerular Filtration Rates in Chinese Patients with Chronic Kidney Disease

  • Jia-fu Feng ,

    Contributed equally to this work with: Jia-fu Feng, Ling Qiu

    jiafufeng@yahoo.com.cn

    Affiliation Laboratory Medicine, Mianyang Central Hospital, Mianyang, Sichuan Province, China

  • Ling Qiu ,

    Contributed equally to this work with: Jia-fu Feng, Ling Qiu

    Affiliation Laboratory Medicine, Peking Union Medical College Hospital, Bejing, China

  • Lin Zhang,

    Affiliation Kidney Internal Medical Department, Mianyang Central Hospital, Mianyang, Sichuan Province, China

  • Xue-mei Li,

    Affiliation Kidney Internal Medical Department, Peking Union Medical College Hospital, Bejing, China

  • Yu-wei Yang,

    Affiliation Laboratory Medicine, Mianyang Central Hospital, Mianyang, Sichuan Province, China

  • Ping Zeng,

    Affiliation Laboratory Medicine, Mianyang Central Hospital, Mianyang, Sichuan Province, China

  • Xiu-zhi Guo,

    Affiliation Laboratory Medicine, Peking Union Medical College Hospital, Bejing, China

  • Yan Qin,

    Affiliation Laboratory Medicine, Peking Union Medical College Hospital, Bejing, China

  • Hong-chun Liu,

    Affiliation Laboratory Medicine, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, China

  • Xing-min Han,

    Affiliation Department of Nuclear Medicine, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, China

  • Yan-peng Li,

    Affiliation Department of Nuclear Medicine, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, China

  • Wei Xu,

    Affiliation Laboratory Medicine, The First Bethune Hospital of Jilin University, Jilin, Jilin Province, China

  • Shu-yan Sun,

    Affiliation Laboratory Medicine, The First Bethune Hospital of Jilin University, Jilin, Jilin Province, China

  • Li-qiang Wang,

    Affiliation Laboratory Medicine, The First Bethune Hospital of Jilin University, Jilin, Jilin Province, China

  • Hui Quan,

    Affiliation Laboratory Department, Nuclear Industrial 416 Hospital, Chengdu, Sichuan Province, China

  • Li-jun Xia,

    Affiliation Laboratory Department, Nuclear Industrial 416 Hospital, Chengdu, Sichuan Province, China

  • Hong-zhang Hu,

    Affiliation Laboratory Department, Nuclear Industrial 416 Hospital, Chengdu, Sichuan Province, China

  • Fang-cai Zhong,

    Affiliation Laboratory Department, The First People's Hospital of Neijiang, Neijiang, Sichuan Province, China

  •  [ ... ],
  • Rong Duan

    Affiliation Kidney Internal Medical Department, The First People's Hospital of Neijiang, Neijiang, Sichuan Province, China

  • [ view all ]
  • [ view less ]

Multicenter Study of Creatinine- and/or Cystatin C-Based Equations for Estimation of Glomerular Filtration Rates in Chinese Patients with Chronic Kidney Disease

  • Jia-fu Feng, 
  • Ling Qiu, 
  • Lin Zhang, 
  • Xue-mei Li, 
  • Yu-wei Yang, 
  • Ping Zeng, 
  • Xiu-zhi Guo, 
  • Yan Qin, 
  • Hong-chun Liu, 
  • Xing-min Han
PLOS
x

Abstract

Objective

To establish equations for the estimation of glomerular filtration rates (eGFRs) based on serum creatinine (SCr) and/or serum cystatin C (SCysC) in Chinese patients with chronic kidney disease (CKD), and to compare the new equations with both the reference GFR (rGFR) and the literature equations to evaluate their applicability.

Methods

The 788 Chinese CKD patients were randomly divided into two groups, the training group and the testing group, to establish new eGFR-formulas based on serum CysC and to validate the established formulas, respectively. 99mTc-DTPA clearance (as the rGFR), serum Cr, and serum CysC were determined for all patients, and GFR was calculated using the Cockcroft-Gault equation (eGFR1), the MDRD formula (eGFR2), the CKD-EPI formulas (eGFR3, eGFR4), and the Chinese eGFR Investigation Collaboration formulas (eGFR5, eGFR6). The accuracy of each eGFR was compared with the rGFR.

Results

The training and testing groups' mean GFRs were 50.84±31.36 mL/min/1.73 m2 and 54.16±29.45 mL/min/1.73 m2, respectively. The two newly developed eGFR formulas were fitted using iterative computation: and . Significant correlation was observed between each eGFR and the rGFR. However, proportional errors and constant errors were observed between rGFR and eGFR1, eGFR2, eGFR4, eGFR5 or eGFR6, and constant errors were observed between eGFR3 and rGFR, as revealed by the Passing & Bablok plot analysis. The Bland-Altman analysis illustrated that the 95% limits of agreement of all equations exceeded the previously accepted limits of <60 mL/min •1.73 m2, except the equations of eGFR7 and eGFR8.

Conclusion

The newly developed formulas, eGFR7 and eGFR8, provide precise and accurate GFR estimation using serum CysC detection alone or in combination with serum Cr detection. Differences in detection methods should be carefully considered when choosing literature eGFR equations to avoid misdiagnosis and mistreatment.

Introduction

Chronic kidney disease (CKD) is a serious public health problem worldwide and is usually defined as kidney damage or decreased kidney function with glomerular filtration rates (GFRs) of less than 60 mL/min per 1.73 m2 for 3 months or longer, regardless of cause [1][3]. Based on GFR, CKD is classified into different stages that require stage-specific management. Therefore, accurate measurement of GFR is critical to evaluate the patient's renal function. Currently, the “gold standard” for GFR determination is to measure the clearance of exogenous substances, such as inulin, iohexol, 51Cr-EDTA, 99mTc-DTPA and 125I-iothalamate [4]. However, these measurements are not only time-consuming, labor-intensive and expensive but also require the administration of rare substances; thus, these methods are not routinely used [5]. Therefore, serum or plasma creatinine levels have become the most commonly used markers for GFR determination because of the simplicity and lower costs of this method [6], [7]. GFR can be calculated based on plasma or serum creatinine using the Cockcroft-Gault or the Modification of Diet in Renal Disease (MDRD) study equations [7], [8]. However, using plasma or serum creatinine has significant disadvantages, such as the inability to measure renal function correctly when impairment is 50% or less [7]. Creatinine generation is proportional to muscle mass and related to an individual's age, sex, race and weight [4], [7]. As a result, an increase in serum creatinine may not be observed until a substantial decrease in GFR has occurred.

Cystatin C (CysC) is a cysteine protease inhibitor with a molecular mass of 13 kDa [9]. It has been shown that cystatin C is a more sensitive marker of GFR changes than serum creatinine [10] because its levels are not affected by muscle mass, age, inflammation, fever or exogenous agents [11]. CysC is produced at a constant rate and cleared solely by glomerular filtration [11], and it can be measured easily with particle-enhanced nephelometric immunoassay (PENIA) [12] or particle-enhanced turbidimetric immunoassay (PETIA) [13]. Large amount of studies have shown that CysC is superior to SCr in predicting the function of kidney [14][18]. Therefore, CysC has been used as an alternative endogenous serum marker of GFR [19], and many formulas for GFR estimation have been developed based upon serum CysC determination [20], [21]. However, these formulas were all established based on a small sample and outside of a laboratory, and the “gold standard” measurement of GFR and CysC is also very inconsistent. In addition, these equations have been established in Western populations and they may or may not be suitable for the Chinese population, which requires clinical validation.

The PETIA-CysC method is becoming more common in clinical practice because of its lower cost and more rapid detection than the PENIA-CysC method [22][25]. For creatinine measurement, the enzymatic method (enzymatic-Cr) is widely used in the clinical laboratory because of its low chance of cross-contamination and stable results [26][28]. In the present study, we applied enzymatic-Cr and PETIA-CysC to determine the creatinine levels and CysC levels, respectively. Following the measurements, we established the eGFR equations based on the levels of CysC and/or serum Cr(SCr) in Chinese CKD patients with multi-center cooperation and evaluated the applicability of these GFR estimating equations.

Materials and Methods

Subjects

A total of 788 CKD patients who were referred by nephrologists, diabetologists, cardiologists or general internists were selected from 6 general hospitals between October 2010 and December 2011 in different areas of China, including 421 males (aged 50.4±15.7 years) and 367 females (age,d 51.6±16.6 years). These were 355 patients from North China(Jilin and Beijing, China), 82 patients from Central China(Henan,China), and 351 patients from South China(Mianyang, Chendu and Neijiang, Sichuan, China).All patients met the diagnostic criteria of NKF-KDOQI CKD [1], and the following patients were excluded during the selection: (1) patients with acute kidney disease or acute renal insufficiency; (2) dialysis patients; (3) patients with merger edema, pleural effusion, ascites, thyroid disease, or viral hepatitis (carriers were exceptional); (4) malnourished patients (lower than normal protein, blood urea or urine conductivity); (5) disabled patients; and (6) patients using antibacterial drugs, especially trimethoprim and cimetidine.

A total of 687 cases were chosen randomly for equation development (training group), which included 358 males (aged 50.2±15.6 years) and 329 females (aged 51.5±16.7 years). The remaining 101 cases were assigned to the equation validation (testing group), including 63 males (aged 51.7±16.1 years) and 38 females (aged 52.0±16.1 years).

The study was approved by the Medical Ethics Committee of Mianyang Central Hospital, Peking Union Medical College Hospital, The First Affiliated Hospital of Zhengzhou University, The First Bethune Hospital of Jilin University, Nuclear Industrial 416 Hospital, or The First people's Hospital of Neijiang, and written informed consent was obtained from all subjects.

Sample collection

Blood samples were collected at 8:00 AM. Approximately 5 ml of blood was collected into a BD Vacutainer® SSTTM II ADVANCE tube (Becton Dickinson, USA) for analysis of CysC and SCr concentrations. After one hour, blood samples were centrifuged at 3000 rpm for 15 min, and serum samples were collected and stored at −80°C until analysis within 480 hours.

Detection methods

Because of the distribution of subjects in different regions, reference GFR (rGFR) values can only be determined independently in each study institute. In order to make the inter-institutes variance as small as possible, a identical research program among study institutes was established, which including researcher training, 99mTc-DTPA drug selection (radiochemical purity greater than 95%, percentage of 99mTc-DTPA bound to plasma protein less than 5%), patients' preparation, intravenous injection, blood sampling time point and procedure, regular maintenance of instrument, and radioactivity measurement.

Reference GFR (rGFR) for each subject was measured when blood samples were collected. In all patients included in the six participating study institutes,99mTc-DTPA clearance was measured as a rGFR. rGFR was measured by the dual plasma sampling method [29], standardized by body surface area (BSA), and resulted in the rGFR: rGFR (mL/min per 1.73 m2) = [Dln (P1/P2)/(T2-T1)] exp {[(T1lnP2)−(T2lnP1)]/(T2−T1)}×0.93×1.73/BSA, where D is dosage of drug injected. P1 and P2 is plasma activity at T1(first blood sampling) and T2(second blood sampling,), respectively. Units for D, P1 and P2 were cpm/ml, for T1, T2 was minute.Unlike rGFR, SCr and CysC levels were measured in a single laboratory, using a 7600–020 Automatic Analyzer (Hitachi, Japan), CysC concentrations was measured by PETIA-CysC method and SCr concentrations were measured by an enzymatic method according to the determination of glycine after enzymatic conversion of creatinine to glycine that can be traced back to IDMS, which kits produced by Sichuan Maker Biotechnology Co., Ltd.(Sichuan, China), but CysC reagents are original equipment manufacture(OEM) products that were obtained from Gentian (Moss, Norway), its calibration can be traced back to ERM-DA471/IFCC. Serum CysC measurements were performed using the following instrument settings: Primary wavelength: 546 nm; Secondary wavelength: 700 nm. Temperature: 37°C; Read points: 19–34; sample blank position 16 and spline calibration method. 157 µL assay buffer (reagent 1), and 3.5 µL sample were mixed with 52 µL anti-cystatin C immunoparticles (reagent 2),. The sensitivity of the assay for CysC was 0.03 mg/L. The intra-assay CV was 2.25% (mean, 0.92 mg/L; n = 20), and the day-to-day CV was 3.18% (mean, 0.68 mg/L; n = 30). SCr measurements were performed using the following instrument settings: Primary wavelength: 546 nm; Secondary wavelength: 700 nm. Temperature: 37°C; Read points: 17–34; sample blank position 16 and linear calibration method. 155 µL enzyme working solution (reagent 1), and 2.0 µL sample were mixed with 26 µL chromogen solution (reagent 2),. The sensitivity of the assay for SCr was 2.40 µmol/L. The intra-assay CV was 1.12% (mean, 135.6 µmol/L; n = 20), and the day-to-day CV was 1.33% (mean, 382.5 µmol/L; n = 30).

eGFR calculation formula

Cockcroft-Gault formula [30]:

Simplified MDRD formula [31]:

MDRD/CKD-EPI formula [32]:

eGFR formula of Chinese collaborative group [33]:

Unified measurement units were used in the above 6 formulas and the following eGFR7 and eGFR8 formulas to facilitate comparative analyses: Cr: mg/dl; CysC: mg/L; BSA: m2; age: years; body weight: kg; height: cm.

Statistical analyses

The measurement data are presented as the means±SD, median and range. The differences between genders were estimated by the student t-test for the normal distribution data or the Kolmogorov-Smirnov test for non-normally distribution data.The GFR estimation equations were established based on the Spearman correlation analysis and non-linear regression of parameters including rGFR, CysC, Cr and age. The differences between GFR estimates and rGFR were tested by ANOVA. The consistency was tested by correlation analysis and Bland-Altman analysis, and the previously accepted tolerances was defined as 60 mL/min •1.73 m2 [34]. The differences between rGFR and eGFR were analyzed by Passing-Bablok regression analysis. The difference distributions between eGFR and rGFR were performed by Mountain plot. The deviations are shown as the area between the Bland-Altman regression line of difference and the zero difference line. The precision is represented by the coefficient of repeatability (CR). The accuracies are shown as P15, P30, or P50, which represented the proportion of eGFR within 15%, 30%, and 50% of rGFR (±15%, ±30%, or ±50%). Kappa statistics were used to evaluate the agreement between stages classification from rGFR method and other eGFR methods. The K value can be interpreted as follows: poor agreement (<0.20), fair agreement (0.21–0.40), moderate agreement (0.41–0.60), good agreement (0.61–0.80) and very good agreement (0.81–1.0).The PASW Statistics 18.0 (SPSS Inc., Somers, NY, USA) and MedCalc11.5 (MedCalc Software, Mariakerke, Belguim) software products were used for these statistical analyses. Differences with P<0.05 are considered statistically significant.

Results

eGFR curve fitting based on the concentrations of serum CysC and serum Cr

The basic characteristics of the 788 patients are listed in Table 1. The datasets from the equation development group were tested by Spearman correlation analysis, which revealed negative correlations of rGFR with age, CysC and Cr (r = −0.212, −0.855 and −0.809, all P = 0.000). Further non-linear regression fitting was performed for rGFR with age, CysC and Cr, and iterative calculations were used to establish the equation:

thumbnail
Table 1. Demographic characteristics of the 788 Chinese CKD patients (up line: mean±SD, down line: median, range).

https://doi.org/10.1371/journal.pone.0057240.t001

In addition, the non-linear regression equation of rGFR was established using the CysC single index:

Analysis of the differences between the rGFR and eGFR values estimated by the equations

The aforementioned 8 equations were used to calculate the eGFRs of the 101 cases in the testing group, which were compared to the rGFR using ANOVA (Table 2). The calculated eGFR values were highly correlated (P = 0.000) with, but not significantly different (F = 0.812, P = 0.592) from, the rGFR values. Further Passing & Bablok analysis revealed that each eGFR value had no apparent linear deviation from the rGFR values (all P>0.05). However, among the eGFR values, eGFR1, eGFR2, eGFR4 and eGFR5 all showed significant proportional differences (the 95% CI of slopes did not include B = 1) and significant constant differences (the 95% CI of intercepts did not include A = 0) with eGFR6 (Fig 1-A, B, D-F); eGFR2 showed the most significant errors of both types (a = −21.267, b = 1.487), and eGFR3 also showed a highly significant constant error (Fig 1-C). Only two new equations of the eight did not show significant differences in the proportional errors and the constant errors.

thumbnail
Figure 1. Passing-Bablok plot to analyze and compare eGFR with rGFR.

(a: 95% confidence interval for the intercept; b: 95% confidence interval for the slope; RSD: residual standard deviation; Cusum test, all P>0.05)

https://doi.org/10.1371/journal.pone.0057240.g001

thumbnail
Table 2. Overall limits of agreement between eGFR and rGFR (n = 101).

https://doi.org/10.1371/journal.pone.0057240.t002

Consistency analysis of the estimated eGFR values and the rGFR values

As shown in Table 3, when the eGFR values were compared with the rGFR values, the most significant deviation was observed for eGFR2 (Figure 2-B, 3036 arbitrary units), followed by eGFR1 (Fig. 2-A, 2045 arbitrary units) and eGFR6 (Figure 2-C, 1435 arbitrary units). The deviations of the other 5 eGFR equations were similar to each other, and eGFR7 showed the smallest deviation (Figure 2D, 367 arbitrary units).

thumbnail
Figure 2. Altman-Bland plot: comparison between eGFR and rGFR (n = 101).

https://doi.org/10.1371/journal.pone.0057240.g002

thumbnail
Table 3. Bias, precision and accuracy of eGFR compared with rGFR (n = 101).

https://doi.org/10.1371/journal.pone.0057240.t003

The performed accuracy showed significant differences in P30 and P50 among various eGFR equations (χ2 = 28.341 and 31.399, respectively; P = 0.000 for both). The highest performed accuracies in P30 and P50, 74.26% and 95.05%, respectively, were observed for eGFR7, followed by eGFR8, with 72.28% and 93.07%, respectively. eGFR2 showed the lowest performed accuracies at 43.56% and 73.27%, respectively, followed by eGFR1 at 55.45% and 77.23%, respectively.

Overall, eGFR7 showed the highest performed accuracy for rGFR estimation (Figure 2-B, CR = 28.5 mL/min • 1.73 m2).

As shown in Table 2, the Bland-Altman analysis revealed that the percentage of each eGFR value falling outside the consistency limit was between 3.96 and 6.93%, which were not significantly different (χ2 = 1.483, P = 0.983). However, the eGFR3, eGFR4 and eGFR8 values were slightly lower than the rGFR values (positive deviation), while others were higher. The consistency limits of six equations were all higher than the previously accepted tolerances (<60 mL/min • 1.73 m2), while those of eGFR7 and eGFR8 were within the previously accepted tolerances.

Using the Mountain chart, the newly developed formulas, eGFR7 and eGFR8, and 6 previously reported formulas in the literature were used to estimate the differences in the consistency between the estimated GFR values and the rGFR values (Figure 3). The median deviation of the distribution curve (M, i.e., P50) was used to show their central tendency, with the range of P5P95 (RP5–P95) representing the degree of dispersion (Table 4).

thumbnail
Figure 3. Mountain plot: comparison between newly developed eGFR formulas and various literature eGFR formulas.

https://doi.org/10.1371/journal.pone.0057240.g003

thumbnail
Table 4. Percentiles (P) of difference between eGFR and rGFR (mL/min • 1.73 m2).

https://doi.org/10.1371/journal.pone.0057240.t004

Compared with eGFR7 (M = 0.01, RP5–P95 = 48.94 mL/min • 1.73 m2), eGFR1-6 all showed a larger M of deviation with rGFR (the M were 2.85, 3.77, 6.90, 6.63, 0.88 and 2.51 mL/min • 1.73 m2, respectively) and wider curve distribution (the RP5-P95 were 68.43, 81.01, 52.09, 60.39, 54.04 and 69.09 mL/min • 1.73 m2, respectively), indicating that their GFR estimation was worse than eGFR7 (Figure 3-A, B, and C-F). Among eGFR1-6, eGFR3 and eGFR4 showed an apparent rightward shift in their error distribution curves (P5, M and P95 were all larger than eGFR7) (Figure 3-C and D), indicating overall underestimation of GFR by these two.

Similar results to those reported with eGFR7 were obtained with eGFR8 and all 6 previously reported equations (therefore, not shown, charted repeatedly). As shown in Figure 3-G, the deviation curves of eGFR7 and eGFR8 almost overlapped, indicating basically the same consistency in estimating the GFR.

Table 5 shows how many patients(percentage) were correctly classified for the different stages of CKD, according to GFR estimating equations based upon SCr and/or CysC. Overall a correct classification was achieved in 43.6% to 65.3% of the patients with the traditional eGFR equations and 73.3% to 74.3% with the new eGFR equations. The best percentage (74.3%) was achieved with the eGFR8, however, the eGFR7 gave a very close result 73.3%.

thumbnail
Table 5. Classification of CKD by SCr- and/or CysC-based on eGFR equations(n = 101)*.

https://doi.org/10.1371/journal.pone.0057240.t005

Discussion

GFR is traditionally considered to be the best overall index of the kidney function. Therefore, its accurate detection is critical for early diagnosis, proper staging, effective treatment and monitoring of CKD. The International Association of Nuclear Medicine has recommended the two-serum method to measure the clearance rate of 99mTc-DTPA as the reference method for GFR detection [35]. However, this method is difficult for clinical applications because of its complexity, high costs, high equipment requirement and radioactivity.

The PETIA-CysC is comparable with PENIA-CysC in accuracy, but PETIA-CysC can be detected using various types of automatic biochemical analyzers, so its detection speed is faster than PENIA-CysC and it can be easily and broadly applied in clinical laboratories. Therefore, many clinical laboratories have detected CysC using the PETIA-CysC method [36][40]. In addition, enzymatic detection of serum Cr shows less interference and low cross-contamination between samples [26][28]; therefore, it is also widely used by clinical laboratories.

In this study, we used the two-serum method 99mTc-DTPA clearance rate as the “gold standard” and determined serum Cr with the enzyme-based method. We determined the serial CysC of 687 randomly selected Chinese CKD patients using the PETIA method. Two GFR estimating equations were established based on the top coefficient of the non-linear regression iterative calculation:

The R2 of eGFR8 was higher than that of eGFR7, indicating that using both Cr and age for GFR estimation was not as effective as using CysC alone. Therefore, CysC has the potential to be a good substitute for Cr during the assessment of renal function.

The two equations developed in this study had limits of agreement (57.0 mL/min • 1.73 m2 and 59.5 mL/min • 1.73 m2, respectively) that were within the pre-set values of <60 mL/min • 1.73 m2. They also showed only minor differences in their Mountain deviation distribution curves, with almost the same error (367 vs. 377 arbitrary units), precision (28.5 vs. 29.7 mL/min • 1.73 m2), accuracy (P>0.05), and applicability in GFR estimation. When comparing the newly developed equations and the 6 previously reported ones, only eGFR7 and eGFR8 did not show apparent proportional errors (b's 95% CI contains B = 1) and constant errors (a's 95% CI contains A = 0) from rGFR. All equations showed high linear correlations (r>0.75, P = 0.000) and acceptable Passing & Bablok regression linearity (Cusum test, P>0.05). Among these 6 previous equations, eGFR3 had only constant errors (a's 95% CI = −12.592 and −2.989, not containing 0) from rGFR, but all the other equations had both constant errors and proportional errors. These errors may have been caused by the differences in study subjects, GFR “gold standards”, GFR markers, and GFR determination methods, which may be verified by the following comparisons of the newly developed equations and the literature equations.

In the estimation of GFR consistency, eGFR7 and eGFR8 were both better than eGFR1 and eGFR2 (limits of agreement = 57.0 and 59.5 vs. 98.7 and 111.5 mL/min • 1.73 m2, respectively), and the limits of consistency for the latter two were both significantly higher than the pre-set professional cutoff value (<60 mL/min • 1.73 m2). eGFR7 and eGFR8 also had smaller GFR estimating errors than eGFR1 and eGFR2 (367 and 377 vs. 2045 and 3036 arbitrary units, respectively). eGFR1 and eGFR2 also had worse estimation accuracy (CR = 49.3 and 55.8 vs. 28.5 and 29.7 mL/min • 1.73 m2, respectively) and lower P30 and P50 than eGFR7 and eGFR8 (χ2 test, P<0.05). The serum Cr can be easily affected by many factors, such as gender, age, muscle size, diet, medication and renal secretion and excretion (renal excretion becomes more apparent under pathological conditions), which may lead to the large errors and low accuracy of eGFR1 and eGFR2 in GFR estimation. In addition, the choice of Cr determination may also affect the applicability of eGFR1 and eGFR2. It was reported that eGFR1 was better than eGFR2 when the enzyme-based measurement was selected [41]. A similar conclusion was reached in this study.

The two equations for MDRD/CKD-EPI, eGFR3 and eGFR4, with the limits of agreement being higher than the pre-set values of <60 mL/min • 1.73 m2 at 61.8 and 72.3 mL/min • 1.73 m2, respectively, were also worse than eGFR7 and eGFR8 in the GFR estimation for Chinese CKD patients. The Mountain deviation distribution curves of eGFR3 and eGFR4 both deviated to the right from the “0” point (M = 6.90 and 6.63 mL/min • 1.73 m2, respectively,) with larger P5, M and P95 than those of eGFR7 and eGFR8, indicating significant underestimation of GFR for Chinese CKD patients. Therefore, the MDRD/CKD-EPI equations may only have limited application in Chinese populations.

The two equations from the Chinese cooperative group, eGFR5 and eGFR6, did target the Chinese populations. However, their limits of agreement (65.6 and 80.8 mL/min • 1.73 m2, respectively,)) both exceeded the pre-set values of <60 mL/min • 1.73 m2. Compared with eGFR7 and eGFR8, eGFR5 and eGFR6 showed broadening trends in the Mountain deviation distributions (P5P95 interval distribution width = 54.04 and 69.09 vs. 48.94 and 49.94 mL/min • 1.73 m2, respectively,)), worse precision (CR = 32.8 and 40.4 vs. 28.5 and 29.7 mL/min • 1.73 m2, respectively,)), and worse accuracy (P30: eGFR6 vs. eGFR7, χ2 test, P<0.05; P50: eGFR6 vs. eGFR7 and eGFR8, χ2 test, P<0.05), all of which indicate inferior consistency for rGFR.

In addition to the factor of race, differences in the determination methods for GFR markers, i.e., the different methods of detecting Cr and CysC, may also contribute to the deviation of the results between the newly developed equations and the previously reported ones. In the past, the picric acid assay was used to detect serum Cr (eGFR1-eGFR6 equations). Currently, many laboratories in China prefer the enzyme-based method to detect serum Cr in order to avoid cross contamination from picric acid. Routine CysC tests also include the PETIA and PENIA methods. Because it can be performed in an automatic biochemical analyzer and the detection time can be as short as 5 minutes, PETIA has become the preferred method for CysC routine determinations. The MDRD/CKD-EPI and the Chinese collaboration group, however, used PENIA instead of PETIA to detect serum CysC (eGFR3-eGFR6 equations). In addition, the differences in CysC standards may also cause differences in detection results, which affect the applicability of eGFR equations. The first CysC certified reference material ERM-DA471/IFCC [42] was introduced in 2010, which allows CysC detection to be traced. As a result, studies on CysC-based GFR estimation equations may avoid matrix effects. Once serum CysC detection becomes standardized, its application in GFR evaluation will have greater advantages over Cr detection. In this study, we developed two eGFR equations based on the sole-indicator, CysC, in combination with serum creatinine levels and age, to evaluate renal functions precisely, accurately, simply and quickly with lower costs.

The difference of results between our new equation and traditional eGFR equations, largely may be due to different measured methods for the determination of SCr and CysC. That is, entered enzymatic-Cr results and/or PETIA-CysC results to traditional eGFR equations that were developed by Cr using the Jaffe's kinetic method and/or PENIA-CysC. It is this inadequate use that leads to the eGFR calculation resulting error. Of course, different choices of the "gold standard" in research programs, as well as technical proficiency of the same "gold standard" may also be the reasons of the error. In addition, the population differences in development group, such as the difference of the constituent ratio of the subjects age, sex, CKD stage, and complications, etc., may also be the reason that the results are not consistent between the traditional eGFR equations and this study equation. In this study, there are statistical differences between gender for weight, height, and body surface area. It will worth considering whether they will cause the differences of the traditional eGFR results.

It is worth mentioning that although the meta-analysis demonstrated that CysC is superior to SCr in the determination of the GFR injuries [43], [44]. However, most studies of CysC have focused on fields where the problems of SCr are most apparent, including specific population groups with malnutrition, extensive reduced body surface area, extremely low body mass, or a few comorbidities that have a large influence on the generation of creatinine. Nevertheless, Serum cystatin C may be influenced by factors other than renal function alone, including serum C-reactive protein [45], smoking [46], the subjects with very low GFR[47], thyroid function [48], [49], immunosuppressive therapy [50], and occupational exposure to toxic agents such as lead, cadmium, and arsenic [51], etc. Thus, clinicians must be cautious when interpreting cystatin C levels alone if the subjects encounter these factors.In summary, different fitting eGFR equations with different estimation values can be developed based on different study subjects, different GFR “gold standards”, and different detection methods. The two newly established equations in this study showed effective but significantly different results from previously reported equations, most likely a result of the differences in detection methods and “gold standards”. Currently, enzyme-based Cr detection and CysC PETIA detection are widely used in clinical laboratories. Our study demonstrated that the simple cystatin C formula could achieve a much better diagnostic performance than SCr formula containing more variables.Therefore, careful and complete consideration is required to choose the right eGFR equation for clinical applications to avoid misdiagnosis and errors in treatment.

Supporting Information

Appendix S1.

Representative patient's signed permit for Glomerular Filtration Rate(GFR) Measurement Using Radiochemical 99mTc-DTPA in Chinese Version.

https://doi.org/10.1371/journal.pone.0057240.s001

(PDF)

Appendix S2.

Representative patient's signed permit for publication in PLoS ONE in English Version.

https://doi.org/10.1371/journal.pone.0057240.s002

(PDF)

Author Contributions

Read and approved the final manuscript: All authors. Conceived and designed the experiments: JFF LQ XML. Performed the experiments: XML PZ YQ XMH YPL HCL WX LJX FCZ. Analyzed the data: LQ YWY WX HQ FCZ. Contributed reagents/materials/analysis tools: LZ HCL XZG HZH LQW SYS RD. Wrote the paper: JFF LQ YWY.

References

  1. 1. Levey AS, Atkins R, Coresh J, Cohen EP, Collins AJ, et al. (2007) Chronic kidney disease as a global public health problem: approaches and initiatives- a position statement from Kidney Disease Improving Global Outcomes. Kidney Int 72: 247–259.
  2. 2. Levey AS, Eckardt KU, Tsukamoto Y, Levin A, Coresh J, et al. (2005) Definition and classification of chronic kidney disease: a position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int 67: 2089–2100.
  3. 3. National Kidney Foundation (2002) NFK-K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 39: S1–S266.
  4. 4. Florkowski CM, Chew-Harris JS (2011) Methods of Estimating GFR-Different Equations Including CKD-EPI. Clin Biochem Rev 32: 75–9.
  5. 5. Hojs R, Bevc S, Ekart R, Gorenjak M, Puklavec L (2008) Serum cystatin C as an endogenous marker of renal function in patients with chronic kidney disease. Ren Fail 30: 181–6.
  6. 6. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, et al. (2012) CKD-EPI Investigators. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med 367: 20–9.
  7. 7. Slort PR, Ozden N, Pape L, Offner G, Tromp WF, et al. (2012) Comparing cystatin C and creatinine in the diagnosis of pediatric acute renal allograft dysfunction. Pediatr Nephrol 27: 843–9.
  8. 8. Larsson A, Flodin M, Hansson LO, Carlsson L (2008) Patient selection has a strong impact on cystatin C and Modification of Diet in Renal Disease (MDRD) estimated glomerular filtration rate. Clin Biochem 41: 1355–61.
  9. 9. Salgado JV, Neves FA, Bastos MG, França AK, Brito DJ, et al. (2010) Monitoring renal function: measured and estimated glomerular filtration rates- a review. Braz J Med Biol Res 43: 528–36.
  10. 10. Roos JF, Doust J, Tett SE, Kirkpatrick CM (2007) Diagnostic accuracy of cystatin C compared to serum creatinine for the estimation of renal dysfunction in adults and children- a meta-analysis. Clin Biochem 40: 383–391.
  11. 11. Ferguson MA, Waikar SS (2012) Established and emerging markers of kidney function. Clin Chem 58: 680–9.
  12. 12. Herget-Rosenthal S, Feldkamp T, Volbracht L, Kribben A (2004) Measurement of urinary cystatin C by particle-enhanced nephelometric immunoassay: precision, interferences, stability and reference range. Ann Clin Biochem 41 ((Pt 2)) 111–8.
  13. 13. Voskoboev NV, Larson TS, Rule AD, Lieske JC (2012) Analytic and clinical validation of a standardized cystatin C particle enhanced turbidimetric assay (PETIA) to estimate glomerular filtration rate. Clin Chem Lab Med 50: 1591–6.
  14. 14. Liu J (2012) Evaluation of serum cystatin C for diagnosis of acute rejection after renal transplantation. Transplant Proc 44: 1250–3.
  15. 15. Cai X, Long Z, Lin L, Feng Y, Zhou N, et al. (2012) Serum cystatin C is an early biomarker for assessment of renal function in burn patients. Clin Chem Lab Med 50: 667–71.
  16. 16. Krishnamurthy N, Arumugasamy K, Anand U, Anand CV, Aruna V, et al. (2011) Serum cystatin C levels in renal transplant recipients. Indian J Clin Biochem 26: 120–4.
  17. 17. Kumaresan R, Giri P (2011) A comparison of serum cystatin C and creatinine with glomerular filtration rate in Indian patients with chronic kidney disease. Oman Med J 26: 421–5.
  18. 18. Yap M, Lamarche J, Peguero A, Courville C (2011) Serum cystatin C versus serum creatinine in the estimation of glomerular filtration rate in rhabdomyolysis.J Ren Care. 37: 155–7.19.
  19. 19. Jaisuresh K, Sharma RK, Mehrothra S, Kaul A, Badauria DS, et al. (2012) Cystatin C as a marker of glomerular filtration rate in voluntary kidney donors. Exp Clin Transplant 10: 14–7.
  20. 20. Robles NR, Mena C, Cidoncha J (2012) Estimated Glomerular Filtration Rate from Serum Cystatin C: Significant Differences among Several Equations Results. Ren Fail 34: 871–5.
  21. 21. Hojs R, Bevc S, Ekart R, Gorenjak M, Puklavec L (2010) Serum cystatin C-based formulas for prediction of glomerular filtration rate in patients with chronic kidney disease. Nephron Clin Pract 114: c118–26.
  22. 22. Sohrabian A, Noraddin FH, Flodin M, Fredricsson A, Larsson A (2012) Particle enhanced turbidimetric immunoassay for the determination of urine cystatin C on Cobas c501. Clin Biochem 45: 339–44.
  23. 23. Bargnoux AS, Cavalier E, Cristol JP, Simon N, Dupuy AM, et al. (2011) Cystatin C is a reliable marker for estimation of glomerular filtration rate in renal transplantation: validation of a new turbidimetric assay using monospecific sheep antibodies. Clin Chem Lab Med 49: 265–70.
  24. 24. Bargnoux AS, Perrin M, Garrigue V, Badiou S, Dupuy AM, et al. (2011) Analytical performances of cystatin C turbidimetric assay: which impact on accuracy of glomerular filtration rate estimation in renal transplantation? Clin Chem Lab Med 50: 133–8.
  25. 25. Hansson LO, Grubb A, Lidén A, Flodin M, Berggren A, et al. (2010) Performance evaluation of a turbidimetric cystatin C assay on different high-throughput platforms. Scand J Clin Lab Invest 70: 347–53.
  26. 26. Greenberg N, Roberts WL, Bachmann LM, Wright EC, Dalton RN, et al. (2012) Clin Specificity characteristics of 7 commercial creatinine measurement procedures by enzymatic and Jaffe method principles. Chem 58: 391–401.
  27. 27. Liu WS, Chung YT, Yang CY, Lin CC, Tsai KH, et al. (2012) Serum creatinine determined by Jaffe, enzymatic method, and isotope dilution-liquid chromatography-mass spectrometry in patients under hemodialysis. J Clin Lab Anal 26: 206–14.
  28. 28. Wang X, Xu G, Li H, Liu Y, Wang F (2011) Reference intervals for serum creatinine with enzymatic assay and evaluation of four equations to estimate glomerular filtration rate in a healthy Chinese adult population. Clin Chim Acta 412: 1793–7.
  29. 29. Ma YC, Zuo L, Zhang CL, Wang M, Wang RF, et al. (2007) Comparison of 99mTc-DTPA renal dynamic imaging with modified MDRD equation for glomerular filtration rate estimation in Chinese patients in different stages of chronic kidney disease. Nephrol Dial Transplant 22: 417–23.
  30. 30. Cockcroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16: 31–41.
  31. 31. Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, et al. (2006) Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol 17: 2937–44.
  32. 32. Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, et al. (2008) Estimating GFR using Serum Cystatin C Alone and in Combination with Serum Creatinine: A Pooled Analysis of 3418 Individuals with CKD[J]. Am J Kidney Dis 51: 395–406.
  33. 33. Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, et al. (2007) Improved GFR estimation by combined creatinine and cystatin C measurements. Kidney Int 72: 1535–42.
  34. 34. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307–10.
  35. 35. Blaufox MD, Aurell M, Bubeck B, Fommei E, Piepsz A, et al. (1996) Report of the Radionuclides in Nephrourology Committee on renal clearance. J Nucl Med 37: 1883–90.
  36. 36. Sohrabian A, Noraddin FH, Flodin M, Fredricsson A, Larsson A (2012) Particle enhanced turbidimetric immunoassay for the determination of urine cystatin C on Cobas c501. Clin Biochem 45: 339–44.
  37. 37. Al-Turkmani MR, Law T, Kellogg MD (2008) Performance evaluation of a particle-enhanced turbidimetric cystatin C assay on the Hitachi 917 analyzer. Clin Chim Acta 398: 75–7.
  38. 38. Conde-Sánchez M, Roldán-Fontana E, Chueca-Porcuna N, Pardo S, Porras-Gracia J (2010) Analytical performance evaluation of a particle-enhanced turbidimetric cystatin C assay on the Roche COBAS 6000 analyzer. Clin Biochem 43: 921–5.
  39. 39. Flodin M, Larsson A (2009) Performance evaluation of a particle-enhanced turbidimetric cystatin C assay on the Abbott ci8200 analyzer. Clin Biochem 42: 873–6.
  40. 40. Bargnoux AS, Servel AC, Piéroni L, Dupuy AM, Badiou S, et al. (2012) Accuracy of GFR predictive equations in renal transplantation: validation of a new turbidimetric cystatin C assay on Architect c8000. Clin Biochem 45: 151–3.
  41. 41. Liu X, Lv L, Wang C, Shi C, Cheng C (2012) Comparison of prediction equations to estimate glomerular filtration rate in Chinese patients with chronic kidney disease. Intern Med J 42: e59–67.
  42. 42. Grubb A, Blirup-Jensen S, Lindstrom V, Schmidt C, Althaus H, et al. (2010) First certified reference material for Cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med 48: 1619–21.
  43. 43. Dharnidharka VR, Kwon C, Stevens G (2002) Serum cystatin C is superior to serum creatinine as a marker of kidney function: a meta-analysis. Am J Kidney Dis 40: 221–6.
  44. 44. Roos JF, Doust J, Tett SE, Kirkpatrick CM (2007) Diagnostic accuracy of cystatin C compared to serum creatinine for the estimation of renal dysfunction in adults and children- A meta-analysis. Clin Biochem 40: 383–91.
  45. 45. Stevens LA, Coresh J, Greene T, Levey AS (2006) Assessing kidney function-measured and estimated glomerular filtration rate. N Engl J Med 354: 2473–83.
  46. 46. Sjostrom PA, Jones IL, Tidman MA (2009) Cystatin C as a filtration marker-haemodialysis patients expose its strengths and limitations. Scand J Clin Lab Invest 69: 65–72.
  47. 47. Horio M, Imai E, Yasuda Y, Watanabe T, Matsuo S (2011) Performance of serum cystatin C versus serum creatinine as a marker of glomerular filtration rate as measured by inulin renal clearance. Clin Exp Nephrol 15: 868–76.
  48. 48. Kotajima N, Yanagawa Y, Aoki T, Tsunekawa K, Morimura T (2010) Influence of thyroid hormones and transforming growth factor-b1 on cystatin C concentrations. J Int Med Res 38: 1365–73.
  49. 49. Fricker M, Wiesli P, Brandle M, Schwegler B, Schmid C (2003) Impact of thyroid dysfunction on serum cystatin C. Kidney Int. 63: 1944–7.
  50. 50. Rule AD, Bergstralh EJ, Slezak JM, Bergert J, Larson TS (2006) Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 69: 399–405.
  51. 51. Poręba R, Gać P, Poręba M, Antonowicz-Juchniewicz J, Andrzejak R (2011) Relation between occupational exposure to lead, cadmium, arsenic and concentration of cystatin C. Toxicology 283: 88–95.