Cystatin C (cysC) is a 13-kDa, low-molecular-weight protein with a high isoelectric point [1]. CysC can be freely filtered through the glomerular membrane and synthesized at a relatively constant rate [2]. Owing to its characteristics, cysC is used as an alternative biomarker for predicting the glomerular filtration rate (GFR) [3], a parameter that reflects kidney function. CysC is less affected by age, race, and muscle mass than creatinine [4], which is the most widely applied marker for GFR estimation. Therefore, the 2012 Kidney Disease: Improving Global Outcomes (KDIGO) clinical practice guidelines suggest measuring cysC in patients with chronic kidney disease (CKD) confirmed solely by a creatinine-based estimated GFR (eGFR) of 45–60 mL/min/1.73 m² [5]. Furthermore, cysC can have advantages over creatinine for the rapid detection of an abrupt decrease in eGFR in critically ill adults and children [6, 7].

There are two important issues related to measuring cysC in clinical settings. Systematic measurement errors can result from variability in the measuring systems such as reagents, instruments, calibrators, and control materials [8-12]. More than five manufacturers have produced cysC assay kits using different measurement principles such as immunoassay, nephelometric, turbidimetric, or spectrophotometric methodologies, resulting in various reagents [10, 11]. Uncertainties attributable to imprecision, bias, and drift have been reported [13-16], and the need for cysC measurement standardization has increased. Various equations are also used for eGFR. In 2012, the 2012 cystatin C-based Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI_cys) was developed [8]. Subsequently, the Caucasian, Asian, pediatric, and adult equations (CAPA_eq) [17] and full age spectrum equation (FAS_eq) [18] were developed. Recently, the 2023 cystatin C-based European Kidney Function Consortium equation (EKFC_cys) [19], which had a scaling factor for cysC that did not differ according to race or sex, was reported to have better accuracy than CKD-EPI_cys, CAPA_eq, or FAS_eq.

These two issues for the measurement of cysC have not been resolved, although manufacturers have altered their cysC measurement procedures using the ERM-DA471/IFCC traceable calibrator. Numerous combinations of measuring systems and novel equations have been generated, depending on the laboratory setting and equation selection. Studies on the effect of cysC reagents and equation combinations for eGFR after cysC standardization are limited [12, 15, 20]. We conducted a simulation analysis of cysC measured using two ERM-DA471/IFCC-standardized cysC reagents and analyzed the correlations among eight eGFR combination methods comprising two reagents and four equations as follows: CKD-EPI_cys, CAPA_eq, FAS_eq, and EKFC_cys. We aimed to elucidate the effect of the cysC reagent and eGFR equation combinations for GFR estimation.

The characteristics of the 148 enrolled participants are presented in Table 2. The mean age of the participants was 60.5 years (SD±14.5 years), and 43% were female. The mean cysC was 1.72±1.44 mg/L for Gentian_cys and 1.71±1.35 mg/L for Roche_cys. There was no significant difference in the mean cysC between Gentian_cys and Roche_cys.

The arithmetic mean of the percent difference was −2.74%±6.62%, and the lower and upper limits of the 95% confidence interval (CI) were −15.72% and 10.23%, respectively (Fig. 1A). Using a ±7.61% difference as the limit of concordance, the calculated cysC values exceeded the limits. The percent difference plot showed that the bias might have affected the measurements at low concentrations. The percentage difference converged to zero for higher concentrations, although it again widened for extremely high values >5 mg/L.

The obtained regression equation between Gentian_cys and Roche_cys was Roche_cys=0.13 (95% CI, 0.12–0.15)+0.91 (0.90 to 0.93)×Gentian_cys (Fig. 1B). According to the regression equation, the limit of the maximum acceptable difference, which satisfied a ±7.61% difference, was 0.85–4.40 mg/L based on Gentian_cys. Of the 148 participants, 38 (25.7%) had values <0.85 mg/L and 12 (8.1%) had values ≥4.40 mg/L.

The end point for the linear portion of the regression line was approximately 60 mL/min/1.73 m² based on a visual assessment (Fig. 2A). Comparing the eGFR of Gentian_cys combined with Getian_eq and that of Roche_cys combined with CKD-EPI_cys, the end point for the linear portion of the regression line was 60 mL/min/1.73 m², after which the distance to the ideal line was greater than that of the combination of Gentian_cys and CKD-EPI_cys (Fig. 2B). Bland–Altman analysis results showed that differences between Getian_eq and CKD-EPI_cys increased by approximately 60 mL/min/1.73 m² (Fig. 2C and D).

According to the groups with Gentian_cys values, the eGFR results calculated for each combination are shown in Table 3. The mean eGFR range of all combinations was 63.4±38.0–58.8±28.6 mL/min/1.73 m². Differences in eGFRs between Gentian_cys and Roche_cys were found mainly in the Q4 (6.8–11.7 mL/min/1.73 m²) and Q3 (2.4–3.4 mL/min/1.73 m²) groups rather than in the Q1 (−0.3 to −0.4 mL/min/1.73 m²) or Q2 (−0.8 to −1.0 mL/min/1.73 m²) groups.

According to the groups with Gentian_cys values, comparisons of the ρ_c between the two reagents are described in Supplemental Data Table S2. In the Q1 and Q2 groups, there was a negligible difference between Gentian_cys and Roche_cys, whereas the difference between Gentian_cys and Roche_cys was significant in the Q3 and Q4 groups. Comparisons of the ρ_c among the four equations with the same reagent are described in Supplemental Data Table S3. In the Q1 and Q2 groups, FAS_eq and EKFC_cys resulted in higher eGFR values than CKD-EPI_cys and CAPA_eq. In contrast, FAS_eq, but not EKFC_cys, resulted in lower eGFR values in the Q3 and Q4 groups. The largest difference in eGFR was found between Roche_cys combined with FAS_eq and Gentian_cys combined with CAPA_eq, but this was not the case for Gentian_cys combined with FAS_eq, Roche_cys combined with CAPA_eq, and Roche_cys combined with EKFC_cys (Supplemental Data Table S4).

Initially, simple equations such as “eGFR=A×cysC^–B” were developed for cysC-based GFR calculation [21, 24-27]. Potentially different results obtained with different measurement systems could potentially lead to larger differences in eGFR [8]. Each equation is recommended for use with a specific associated measuring system because of the discrepancy in eGFR among equations despite a fixed cysC value [14, 21, 24, 26, 28]. For the standardization of cysC, a certified reference material, ERM-DA471/IFCC, was produced by the Working Group for the Production of an International Cystatin C Calibrator (WG-SCC) in 2010 [29]. The 2012 cysC-based CKD-EPI_cys was developed to include age and sex factors as well as the standardized cysC value [8], and the KDIGO guidelines recommend use of CKD-EPI_cys for reporting cysC-based eGFR or the alternative equations shown to improve the accuracy of GFR estimation compared with CKD-EPI_cys [5]. Therefore, CKD-EPI_cys appears to be the most broadly applicable equation [30], although CAPA_eq [17], FAS_eq [18], and EKFC_cys [19] were suggested as their GFR estimation showed improved accuracy compared with that of CKD-EPI_cys.

In 2010, ERM-DA471/IFCC was approved for cysC measurements in specific systems with successful calibration and negligible method bias [20, 31, 32]. Using ERM-DA471/IFCC as a calibrator or measurement trueness control material, each manufacturer generated adjusted calibration curves with improved measurement uncertainty [20, 31, 33, 34]. However, the outcome of harmonization and standardization for cysC measurement remains controversial [10, 11], although a few studies have reported improvements in measurement uncertainty for specific measuring systems [20, 31, 34]. Both Gentian_cys and Roche_cys, which use the particle-enhanced immunoturbidimetric (PETIA) method, have low analytical variability and bias [11, 32]. ERM-DA471/IFCC requires dilution because it is a single-level calibrator with a high cysC concentration of 5.48 mg/L. The results of the College of American Pathologists 2014 CYS Survey, international proficiency testing programs [11], and the 2015 survey involving seven clinical laboratories located in France and Belgium [10] showed limitations in the metrological traceability of commercial measuring systems for cysC. We found significant differences between Gentian_cys and Roche_cys at concentrations <0.85 mg/L and cysC concentrations ≥4.40 mg/L calculated using Gentian_cys. Gentian_cys resulted in higher values than Roche_cys at very high cysC concentrations, similar to the results of a previous report [10]. In contrast, the Bland–Altman analysis suggested that Roche_cys shows higher values than Gentian_cys at low concentrations (<0.85 mg/L). These results are in line with a 2015 survey [10], which reported that clinically significant positive bias decreased with higher concentrations of Roche_cys on the Cobas c502 system. Based on our results, the equivalence of cysC values at low concentrations (<0.85 mg/L) between the two PETIA-based measuring systems was unsatisfactory.

A low cysC concentration might be the baseline for evaluating abrupt cysC elevations in critically ill adults and children [35, 36], and the bias at low concentrations might dramatically affect GFR estimations [14, 15]. Recently, a study showed that the predicted concentrations from recalibration against ERM-DA471/IFCC were consistently 17% higher than the previously measured values using calibration with the Siemens nephelometer [31]. These values were 5–6% higher than the previous prediction used for the development of CKD-EPI_cys.

In the current study, CKD-EPI_cys, with an inflection point at 0.8 mg/L of cysC, resulted in the highest ρ_c value between the two reagents in the Q3 and Q4 groups (Supplemental Data Table S2). According to CKD-EPI_cys, the exponent number of cysC is altered from −1.33 to −0.50 at 0.8 mg/L of cysC, which leads to a reduction in effectiveness for bias at low cysC concentrations. Conversely, CAPA_eq and FAS_eq, without a compensation function at low cysC concentrations, resulted in dynamic ρ_c values ranging from 0.73 to 0.93 in the Q3 and Q4 groups, depending on the combination of reagents (Supplemental Data Table S4).

Our study showed that differences in eGFR values between Getian_eq and CKD-EPI_cys within Gentian_cys were increased by approximately 60 mL/min/1.73 m². Although the manufacturer continues to recommend the use of Getian_eq for eGFR using Gentian_cys values after standardization, CKD-EPI_cys and two equations developed after standardization revealed different trends for eGFRs from approximately 60 mL/min/1.73 m² compared with those of Getian_eq. Thus, the initial equation cannot be confidently used in the absence of evidence indicating the superior accuracy of GFR estimations compared with that of CKD-EPI_cys [5].

FAS_eq and EKFC_cys tended to represent higher eGFR values for the Q1 group, which mainly consists of G4 and G5 eGFR categorization, compared with those obtained using CKD-EPI_cys or CAPA_eq. These results might reflect a relatively higher proportion of healthy participants in the FAS_eq development population [18, 37]. In addition, FAS_eq resulted in a less sensitive reflection of the eGFR in the Q3 and Q4 groups with low cysC concentrations compared with that obtained using CKD-EPI_cys, CAPA_eq, or EKFC_cys (Table 3 and Supplemental Data Fig. S1). Because Q_cys rescaling would require only a single rescaling factor, specifically the spectrum of age [19], EKFC_cys would have the advantage of easily reducing bias based on the measured GFR. Compared with FAS_eq, EKFC_cys used a modified Q_cys, and each Q_cys induced a marked difference in eGFR in the Q3 and Q4 groups.

Furthermore, the cumulative frequency distribution of eGFRs from each combination showed the effect of both reagents and equations on the eGFR results (Supplemental Data Fig. S1). The frequency distribution indicated that the equation prominently affects the eGFR at levels less than approximately 60 mL/min/1.73 m², whereas the measuring system and equation have either an additive effect or a decreased effect according to each combination at levels greater than approximately 60 mL/min/1.73 m². In contrast to a previous study [20], our data indicate that the diversity of cysC reagents is a cause of discordance in accuracy in comparison studies of equations.

To overcome the current limitations of the cysC measurement system, methodological improvements are required. First, the most promising candidate, isotope dilution mass spectrometry (IDMS), will be a reference measurement method for cysC [38, 39], and current commercial methods should be calibrated to be traceable with respect to the IDMS method. Second, it is essential to develop novel reference materials for cysC that are assigned clinically useful values and do not require dilution.

Our study has several limitations. First, we could not measure the GFR, which serves as the reference method for the eGFR. Second, we only performed one-point calibration, whereas multi-point calibration based on serial dilutions of a certified reference material or commercial calibrator would help to adjust the calibration factor. Finally, owing to limited resources, we evaluated only a limited number of participants. Due to these limitations, we could not show the accuracy and bias of each cysC value and eGFR equation.

Taken together, our data show that the equivalence of cysC values at low concentrations (<0.85 mg/L) between the two PETIA-based measuring systems is not satisfied after standardization. Results obtained with different measurement systems could lead to larger differences in the eGFR, which vary based on the combination used. Bias and accuracy analyses for cysC-based eGFR equations should be conducted using a standardized measurement system for cysC, as well as measured GFR.