The Precision and Accuracy of Chronic Kidney Disease Epidemiology Collaboration Creatinine, Cystatin C, and Creatinine–Cystatin C Equations in Kidney Transplant Recipients: A Systematic Review and Meta-Analysis Authors: Kanita Mankan, Pakpoom Wongyikul, Nop Khongthon, Adivitch Sripusanapan, Kajohnsak Noppakun, Ekamol Tantisattamo, Pojsakorn Danpanichkul, Phichayut Phinyo, Moragot Chatatikun, Atthaphong Phongphithakchai Published: 2026 ABSTRACT Rationale & ObjectiveCystatin C offers advantages over creatinine for estimating glomerular filtration rate (GFR) because of its reduced dependence on muscle mass. Although the 2024 Kidney Disease: Improving Global Outcomes (KDIGO) guidelines support its use in chronic kidney disease, specific guidance for kidney transplant recipients (KTRs) remains limited. This systematic review and meta-analysis evaluated the precision and accuracy of cystatin C-based, creatinine-based, and combined creatinine- and cystatin C-based equations compared with measured GFR in KTRs.Study DesignSystematic review and meta-analysis.Setting & Study PopulationsPublished studies evaluating the performance of Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations among adult KTRs.Selection Criteria for StudiesThe peer-reviewed studies reporting CKD-EPI–based estimated GFR (eGFR) compared with measured GFR, including data on mean bias and/or accuracy metrics (P30 or P10).Data ExtractionTwo reviewers independently extracted study characteristics, GFR measurement methods, equation type, mean bias, and accuracy outcomes. Discrepancies were resolved through consensus.Analytical ApproachRandom-effects meta-analyses were performed to estimate pooled mean bias and accuracy within 30% (P30) and 10% (P10) of measured GFR, with 95% confidence intervals (CIs). Heterogeneity was quantified using the I2 statistic.ResultsTwenty-six studies met inclusion criteria; 14 provided extractable CKD-EPI bias data (4,041 adult KTRs; 11,997 repeated measures). The combined creatinine–cystatin C equation had the lowest pooled mean bias (0.63 mL/min/1.73 m2; 95% CI, −3.65 to 4.92). Creatinine-based equations underestimated measured GFR (−4.95 mL/min/1.73 m2; 95% CI, −9.10 to −0.81), whereas cystatin C-based equations overestimated it (5.01 mL/min/1.73 m2; 95% CI, 0.54 to 9.48). P30 and P10 accuracies were broadly similar across biomarkers.LimitationsHeterogeneity in GFR measurement methods, assay calibration, and timing post-transplant.ConclusionsAcross heterogeneous KTR studies, the CKD-EPI creatinine–cystatin C equation demonstrates lower mean bias and slightly higher pooled P30 than single-marker equations in population-level assessment. When cystatin C is available, the combined equation is a reasonable choice. Further multicenter validation studies with harmonized biomarker protocols are needed to confirm its clinical utility in KTRs.RegistrationPROSPERO (CRD42024506221). Kidney transplantation is the preferred treatment for patients with kidney failure.1 Accurate estimation of glomerular filtration rate (GFR) is essential for monitoring long-term graft function. Serum creatinine remains the primary biomarker recommended for estimating GFR in kidney transplant recipients (KTRs), but its accuracy may be reduced in patients with low muscle mass and can be influenced by diet, physical activity, and certain medications.2,3 Cystatin C, in contrast, is less affected by muscle mass, and the 2024 Kidney Disease: Improving Global Outcomes (KDIGO) guideline recommends measuring cystatin C to confirm chronic kidney disease or when creatinine-based eGFR may be unreliable.4, 5, 6 However, its role in KTRs remains unclear.2,7,8 Prior evidence also suggests that immunosuppressive therapy may contribute to underestimation of GFR when cystatin C is used in this population.9 The most recent meta-analysis by Pan et al10 compared the diagnostic accuracy of serum creatinine and cystatin C across ten studies conducted before 2012. Although that study highlighted potential advantages of cystatin C, its clinical utility was limited because the analyses relied on raw biomarker concentrations rather than eGFR equations and used heterogeneous cut-off values. As a result, direct evidence examining the performance of cystatin C-based equations for estimating GFR in KTRs remains insufficient. New GFR equations that use both biomarkers, including the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2012 and 2021 equations, have since been developed.8,11 With the rapid growth of research on cystatin C-based eGFR in the past decade, its clinical utility in KTRs may require updated evaluation.12, 13, 14, 15 The objective of this study was to systematically review and meta-analyze the precision of cystatin C- and creatinine-based eGFR equations in KTRs. Materials and Methods This systematic review adhered to the Cochrane Handbook for Systematic Reviews of Interventions and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.16,17 The review protocol was registered with PROSPERO (CRD42024506221). Data sources and search strategy The literature search was conducted using PubMed, Scopus, and EMBASE, covering all records from database inception through February 2024. Specific keywords and controlled vocabulary terms were applied to each database (eg, MeSH terms for PubMed and EMTREE terms for EMBASE), as detailed in Tables S1-S3. The reference lists of previously published systematic reviews and meta-analyses on the same topic were screened manually. Studies identified through these reference lists but not captured in the electronic searches were also included. Study selection and outcomes Cohort or cross-sectional studies were included if they met the following criteria: (1) enrolled adults aged 18 years or older who had undergone kidney transplantation; (2) measured GFR (excluding creatinine clearance) using a recognized reference standard to compare cystatin C-based and creatinine-based eGFR; (3) reported measurement error parameters describing the difference between eGFR and measured GFR (mGFR) for both cystatin C and creatinine, such as mean bias, standard deviation of bias, or accuracy within 30% or 10%, regardless of the equation used; and (4) reported at least 1 diagnostic parameter of interest. Studies were excluded if they were not published in English, lacked accessible full text, were duplicate publications, or did not provide sufficient or relevant diagnostic data. Screening and data extraction After removing duplicates, two investigators (KM and AS) independently screened all the studies and extracted the relevant data. Only studies with accessible full-text manuscripts were included. Extracted variables included study characteristics, design, setting, language, inclusion and exclusion criteria, participant characteristics (age and race), sample size, calibration assay, eGFR equation, measured GFR substance, measured time after transplantation. Any discrepancies were resolved through discussion with the clinical methodologist (PW). Risk of bias assessment Two authors (KM and AS) independently assessed the methodological quality of the included studies using the COSMIN Risk of Bias tool, adapted for research on measurement error. Seven domains were evaluated: (1) stability of the patient during measurement, (2) adequacy of the time interval, (3) likelihood of measurement bias, (4) likelihood of review bias, (5) likelihood of incorporation bias, (6) appropriateness of study conduct, and (7) adequacy of results reporting. Each domain was rated as low risk, some concerns, or high risk of bias.18 Disagreements were resolved through discussion with nephrology experts (KN, ET, and AP) and the clinical methodologist (PW and PP). Data synthesis and analysis All analyses were conducted using R version 4.1.2 (R Foundation for Statistical Computing). A two-sided P < 0.05 was considered statistically significant. Because of substantial heterogeneity among equations within the same biomarker, data pooling was limited to studies that used CKD-EPI equations. Other equations were summarized in the supplementary material. For the primary objective, the pooled mean bias and corresponding 95% confidence intervals (CIs) were estimated using a random-effects meta-analysis. Creatinine-, cystatin C-, and combined creatinine- and cystatin C-based (Crea/Cys) eGFR values derived from different CKD-EPI versions or measurement time points were treated as repeated measures. Mean bias was defined as eGFR minus measured GFR (mGFR). A 95% CI that included zero indicated no significant bias. Narrower CIs were interpreted as indicating greater precision. A clinical threshold of ±5 mL/min/1.73 m2 from zero was applied: CIs entirely within this range were considered clinically indifferent; CIs entirely outside this range were considered clinically different; and CIs that crossed this range were considered uncertain. For the secondary objective, when feasible, pooled P30 and P10 values were estimated using the metaprop command.19 P30 and P10 were defined as the percentages of estimates falling within an error of ±30% and ±10%, respectively. Statistical heterogeneity of pooled effects was assessed using the I2 statistic, with values greater than 75 percent considered to indicate substantial heterogeneity.20 Potential sources of heterogeneity and the robustness of the findings were further examined through subgroup and sensitivity analyses. Funnel plots and Doi plots were used to assess publication bias for continuous and binary outcomes, respectively. A Begg’s test P value <0.05 and an LFK index greater than ±2 were interpreted as evidence of publication bias.21 Publication bias was evaluated when at least 10 studies were available for funnel plot assessment and at least 5 studies for Doi plot assessment. Certainty of evidence Two investigators (KN and PW) independently assessed the quality of evidence for the pooled mean bias of CKD-EPI–based eGFR derived from creatinine-, cystatin C-, and combined Crea/Cys-based equations using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) framework.22 The certainty of the evidence was rated as high, moderate, low, or very low according to study design, risk of bias, consistency, directness, and precision. Results were considered imprecise if the 95% confidence interval of the pooled mean bias exceeded ±5 mL/min/1.73 m2 or if the 95% confidence interval of the pooled P30 or P10 exceeded ±10%.23 Results Search result A total of 10,140 records were identified from the three databases, along with 30 additional records obtained through manual searches. After removing duplicates and screening the records, 26 studies were included in this systematic review, of which 25 were eligible for meta-analysis, totaling 5,189 adult KTRs and 25,687 repeated measures. The PRISMA flow diagram is presented in Figure 1, and a list of excluded studies with the reasons for exclusion is provided in Table S4. Among the included studies, 14 reported measurement error data for CKD-EPI eGFR, including 4,041 adult KTRs and 11,997 repeated measures.Figure 1PRISMA flow diagram. Abbreviation: GFR, glomerular filtration rate. PRISMA flow diagram. Abbreviation: GFR, glomerular filtration rate. Risk of bias assessment of included studies Figure 2 summarizes the risk of bias for the included studies (n = 26) assessing measurement error. All studies were rated as having a low risk of bias with respect to the adequacy of the time interval and the likelihood of measurement, review, and incorporation bias, as well as the appropriateness of study conduct. However, several studies showed some concerns regarding patient stability during measurement and the adequacy of results reporting. Overall, the included studies were judged to have a low risk of bias.Figure 2Risk of bias in included studies. Risk of bias in included studies. Characteristics of included studies assessing measurement error A total of 25 studies reported measurement error for creatinine-, 23 for cystatin C-, and 13 for combined cystatin C- and creatinine-based equations. The mean age of patients ranged from 37.3 to 81.6 years, with measurement periods spanning 11 days to 12 years after transplantation. Most studies conducted measurements more than 6 months after transplantation (Table 1).24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49Table 1Characteristics of Included Studies Assessing Measurement ErrorAuthorsYearSample sizeRacePopulation AgeMeasure GFR SubstanceeGFR substanceEquationeGFRBiasMeasure time (after kidney transplant)Risch et al24200529White51.3 (± 15.8)a125I-Iothalamate clearanceCreatinineMDRD2 (± 11.8)a6 (± 4.5)a yearsCreatinineRecalibrated MDRD1.7 (± 12.6)a6 (± 4.5)a yearsCystatin CLarsson–4.7 (± 10.6)a6 (± 4.5)a yearsWhite et al252005117Mixed52 (± 11)a99mTc-DTPACreatinineMDRD–11.5 (± 12.2)a7.5 (± 10.8)a yearsCystatin CFiller–1.7 (± 11.4)a7.5 (± 10.8)a yearsCystatin CLe Bricon–3.8 (± 11.8)a7.5 (± 10.8)a yearsCystatin CLarsson–16.1 (± 14.2)a7.5 (± 10.8)a yearsCystatin CHoek–10.6 (± 11.7)a7.5 (± 10.8)a yearsWhite et al262007198Mixed53 (± 12)a99mTc-DTPACreatinineOriginal MDRD–10.7 (± 11.6)a7.2 (± 6.9)a yearsCystatin CFiller0.2 (± 11.5)a7.2 (± 6.9)a yearsCystatin CLe Bricon–2.5 (± 11.2)a7.2 (± 6.9)a yearsCystatin CRule–10.1 (± 11.2)a7.2 (± 6.9)a yearsZahran et al272007103Mixed81.62 (± 20.94)aInulin clearance (GFR > 60 mL/min/1.73 m3)CreatinineMDRD–19.86 (± 23.9)a147 monthsCystatin CLe Bricon–16.93 (± 20.28)a147 monthsCystatin CHoek–23.54 (± 20.66)a147 monthsCystatin CFiller–25.5 (± 28.07)a147 monthsCystatin CLarsson–23.87 (± 25.94)a147 monthsCystatin CRule–18.18 (± 29.05)a147 monthsInulin clearance (GFR < 60 mL/min/1.73 m3)CreatinineMDRD6.57 (± 17.13)a147 monthsCystatin CLe Bricon10.26 (± 14.26)a147 monthsCystatin CHoek3.26 (± 14.63)a147 monthsCystatin CFiller6.9 (± 18.62)a147 monthsCystatin CLarsson1.56 (± 23.02)a147 monthsCystatin CRule–0.86 (± 15.62)a147 monthsMaillard et al282008120White53 (NA)Inulin clearanceCreatinineMDRD8.7 (± 18.2)a12 (3-240)b monthsCystatin CLe Bricon2.8 (± 13.1)a12 (3-240)b monthsCystatin CHoek-4 (± 13.1)#12 (3-240)b monthsCystatin CLarsson-5.9 (± 15.4)a12 (3-240)b monthsCystatin CRule-5 (± 13.3)a12 (3-240)b monthsCreatinineMDRD8.7 (± 18.2)a12 (3-240)b monthsCreatinineCockcroft-Gault–2.6 (± 22.5)a6.8 yearsCystatin CStevens–9.5 (± 13.2)a6.8 yearsCreatinineLe Bricon–1.2 (± 13.5)a6.8 yearsPoge et al292008108White48.9 (NA)99mTc-DTPACreatinineMDRDNA75.4 (3-240)b monthsCystatin CHoekNA75.4 (3-240)b monthsCystatin CFillerNA75.4 (3-240)b monthsCystatin CLarssonNA75.4 (3-240)b monthsQutb et al30200997Asian39.8 (± 12.7)a99mTc-DTPACreatinineMDRD7.7 (± 33.5)aNACreatinineNankivell10.3 (± 26.1)aNACystatin CNA0.31 (± 22.4)aNAYeo et al.312009102Asian42 (± 10.47)a51Cr-EDTACreatinineMDRD0.33 (± 12.5)a11.65 (± 2.68)a daysCreatinineMCQ–0.19 (± 11.8)a11.65 (± 2.68)a daysCreatinineChinese MDRD24.27 (± 16.35)a11.65 (± 2.68)a daysCreatinineJapanese MDRD–17.95 (± 11.06)a11.65 (± 2.68)a daysCreatinineJSN-CKDI–20.96 (± 10.89)a11.65 (± 2.68)a daysCystatin CFiller0.03 (± 17.1)a11.65 (± 2.68)a daysCystatin CLe Bricon–6.4 (± 13.98)a11.65 (± 2.68)a daysCystatin CMacisaac–7.98 (± 14.72)a11.65 (± 2.68)a daysCystatin COrebro-cyst–5.67 (± 16.83)a11.65 (± 2.68)a daysCystatin CHoek–12.72 (± 14.23)a11.65 (± 2.68)a daysCystatin CRule–12.84 (± 15.1)a11.65 (± 2.68)a daysCystatin CLarsson–8.17 (± 21.49)a11.65 (± 2.68)a daysBargnoux et al322012157White50.8 (±12.7)a99mTc-DTPACreatinineMDRD3 (± 10.5)a2 yearsCystatin CStevens1.3 (± 11.2)a2 yearsCystatin C and creatinineNR1.7 (± 9.4)a2 yearsTsujita et al33201273Asian49.1 (± 12.8)aInulin clearanceCreatinineJapanese–3.74 (NA)>6 monthsCystatin CJapanese12.64 (NA)>6 monthsCystatin C and creatinineJapanese4.45 (NA)>6 monthsMasson et al342013670White54 (± 14)aInulin clearanceCreatinineCKD-EPI 20124.4 (± 13)a69 (± 77)a monthsCystatin CCKD-EPI 2012–2.82 (± 12)a69 (± 77)a monthsCystatin C and creatinineCKD-EPI 2012-0.54 (± 11)a69 (± 77)a monthsKukla et al352014153Mixed48.9 (±12.4)aIothalamate clearanceCreatinineMDRD6.4 (± 15.8)a>1 yearCreatinineCKD-EPI11.8 (± 16.6)a>1 yearCystatin CCKD-EPI8.6 (± 15)a>1 yearCystatin C and creatinineCKD-EPI9 (± 12.7)a>1 yearCreatinineMDRD10.8 (± 14.4)a>1 yearCreatinineCKD-EPI15.8 (± 15.5)a>1 yearCystatin CCKD-EPI12.3 (± 13.8)a>1 yearCystatin C and creatinineCKD-EPI12.6 (± 11.2)a>1 yearCreatinineMDRD-0.8 (± 15.2)a>1 yearCreatinineCKD-EPI5.4 (± 16.4)a>1 yearCystatin CCKD-EPI2 (± 13.7)a>1 yearCystatin C and creatinineCKD-EPI2.9 (± 12)a>1 yearLuis-Lima et al362015193White56 (± 12.85)aIohexol clearanceCreatinineMDRD7.37 (NA)>6 monthsCreatinineCKD-EPI11.03 (NA)>6 monthsCystatin CCKD-EPI–3 (NA)>6 monthsCystatin CHoek–0.01 (NA)>6 monthsCystatin CLarsson–2.42 (NA)>6 monthsCystatin C and creatinineCKD-EPI–2.26 (NA)>6 monthsCystatin C and creatinineStevens3.36 (NA)>6 monthsMeeusen et al372015568Mixed55 (± 14)aIothalamate clearanceCreatinineCKD-EPI 20090.5 (−2.2 to 3.2)aNRCystatin CCKD-EPI 2012−5.4 (−7.8 to −3.0)aNRCystatin C and creatinineCKD-EPI 2012−4.8 (−7.0 to −2.6)#NRUslu et al38201544Mixed37.3 (± 9.8)a99mTc-DTPACreatinineMDRD24.9 (± 17.5)a3 monthsCystatin CHoek10.2 (± 14)a3 monthsCystatin CLarsson12.9 (± 20.6)a3 monthsCystatin CLe Bricon16.5 (± 13.6)a3 monthsCystatin CFiller23 (± 18)a3 monthsCreatinineMDRD28.4 (± 18.1)a6 monthsCystatin CHoek15.8 (± 13.3)a6 monthsCystatin CLarsson19.6 (± 22.5)a6 monthsCystatin CLe Bricon22.1 (± 12.9)a6 monthsCystatin CFiller29.2 (± 17.6)a6 monthsCreatinineMDRD26.8 (± 14.5)a12 monthsCystatin CHoek14.7 (± 13.8)a12 monthsCystatin CLarsson17.4 (± 23.2)a12 monthsCystatin CLe Bricon21.1 (± 13.4)a12 monthsCystatin CFiller26.9 (± 18)a12 monthsAl-wakeel et al3920168AsianNRInulinCreatinineMDRD–3.2 (± 7.71)a6 monthsCreatinineCKD-EPI–1.1 (± 8.5)a6 monthsCystatin CCKD-EPI–10.2 (± 17.1)a6 monthsCystatin C and creatinineCKD-EPI–6.5 (± 11.6)a6 monthsKeddis et al4020161139White55.9 (±14.3)aIothalamate clearanceCreatinineMDRD–1.9 (± 14.6)a72 monthsCreatinineCKD-EPI 20090.8 (± 13.7)a72 monthsCystatin CCKD-EPI 2012-8.7 (± 11.7)a72 monthsCystatin C and creatinineCKD-EPI 2012–5 (± 10.7)a72 monthsSalvador et al412017594White52 (23 to 75)bIohexol clearanceCreatinineMDRD0.7 (± 12.3)a1 yearCreatinineCKD-EPI15.4 (± 14.6)a1 yearCreatinineNankivell formula B−4.18 (± 13)a1 yearCystatin CCKD-EPI−9.79 (± 11.5)a1 yearCystatin CCAPA−8.58 (± 10.3)a1 yearCystatin C and creatinineCKD-EPI−4.56 (± 10.7)a1 yearCystatin C and creatinineNew one (this study)1.19 (± 7.59)a1 year51Cr-EDTA clearanceCreatinineMDRD1.64 (± 11.1)a10 weeksCreatinineCKD-EPI3.72 (± 13.5)a10 weeksCreatinineNankivell formula B–4.66 (± 11.5)a10 weeksCystatin CCKD-EPI–7.56 (± 9.75)a10 weeksCystatin CCAPA–6.64 (± 10.4)a10 weeksCystatin C and creatinineCKD-EPI–3.26 (± 8.41)a10 weeksCystatin C and creatinineNew one (this study)2.67 (± 7.8)a10 weeksYang et al42201770Asian46 (± 11.5)a99mTc-DTPACreatinineMDRD–8.08 (± 14.9)aNRCreatinineCKD-EPI–14.54 (± 15.1)aNRCystatin CCKD-EPI–8.36 (± 15.8)aNRCystatin C and creatinineCKD-EPI–10.38 (± 14)aNRTang et al432018252Asian38.1 (± 10.1)a99mTc-DTPACreatinineMDRD21.9 (± 13.0)a5.6 (± 7.2)a monthsCreatinineChinese MDRD42.5 (± 18.4)a5.6 (± 7.2)a monthsCreatinineAmerican MDRD22 (± 13.0)a5.6 (± 7.2)a monthsCreatinineIDMS-MDRD16.7 (± 11.7)a5.6 (± 7.2)a monthsCreatinineCKD-EPI27.8 (± 14.9)a5.6 (± 7.2)a monthsCreatinineEPI-Asian32.9 (± 16.2)a5.6 (± 7.2)a monthsCreatinineJSN-CKDI-6.2 (± 8.1)a5.6 (± 7.2)a monthsCystatin CLarsson-15.8 (± 9.9)a5.6 (± 7.2)a monthsCystatin CRule-14.7 (± 10)a5.6 (± 7.2)a monthsCystatin CHoek–13.8 (± 10.1)a5.6 (± 7.2)a monthsCystatin CFiller–3.9 (± 10.2)a5.6 (± 7.2)a monthsCystatin CGrub–9.2 (± 13)a5.6 (± 7.2)a monthsBorjas-Garcia et al44201930NR40.2 (±14.6)aIothalamate clearanceCreatinineCKD-EPI–2.4 (± 9.6)a13 monthsCreatinineMDRD6.8 (± 11.8)a13 monthsCreatinineCockcroft-Gault–24.4 (± 12.8)a13 monthsCreatinineMCQ–5.9 (± 12.1)a13 monthsCreatinineCreatinine clearance10.6 (± 13.7)a13 monthsCystatin CStevens-9.3 (± 31.4)a13 monthsCystatin C and creatinineCKD-EPI-28.4 (± 28.4)a13 monthsUteil et al452020270White47 (± 14)a51Cr-EDTACreatinineMDRD11.1 (± 12.4)a1 yearCreatinineCKD-EPI16.4 (± 14.8)a1 yearCystatin CCKD-EPI15 (± 23.5)a1 yearCystatin C and creatinineCKD-EPI14.1 (± 16.7)a1 yearMeeusen et al462022220Mixed55 (± 14)aIothalamate clearanceCreatinineCKD-EPI 2021−0.05 (−1.67 to 1.36)a2.14 (± 3.2)a yearCystatin C and creatinineCKD-EPI 2021−3.84 (−4.83 to −2.51)aTrans et al47202283White53 (47 to 62)b51Cr-EDTACreatinineMDRD0.2 (± 10)a3 monthsCreatinineCKD-EPI3.4 (± 11)a3 monthsCystatin CCKD-EPI–12.1 (± 9)a3 monthsCystatin C and creatinineCKD-EPI–6.5 (± 8)a3 months6554 (47 to 62)bCreatinineMDRD–1.7 (± 12.7)a12 monthsCreatinineCKD-EPI1.8 (± 13.5)a12 monthsCystatin CCKD-EPI–11.1 (± 7.5)a12 monthsCystatin C and creatinineCKD-EPI–6.4 (± 8.9)a12 monthsKure et al48202382White53 (46 to 62)bIodothalamate clearanceCreatinineCKD-EPI 20216.34 (± 10.85)a3 monthsCystatin C and creatinineCKD-EPI 2021-5.04 (± 7.74)a6454 (47 to 62)bCreatinineCKD-EPI 20214.82 (± 13.63)a12 monthsCystatin C and creatinineCKD-EPI 2021–4.54 (± 8.18)aKrishnamurthy et al49201130Asia43.1 ± 10.099mTc-DTPACreatinineMDRDNR>6 monthsCystatin CLe BriconNR>6 monthsAbbreviations: AKT, After kidney transplant; CAPA, White and Asian pediatric and adult subjects; 51Cr-EDTA, chromium-51 labelled ethylene diamine tetra-acetic acid; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration Equation; eGFR, estimated glomerular filtration rate; 125I-iothalamate, iodine-125 iothalamate; JSN-CKDI, Japanese Society of Nephrology Chronic Kidney Disease Initiative; MCQ, Mayo Clinic Quadratic; MDRD, original Modification of Diet in Renal Disease equation; 99mTc-DTPA, technetium-99m-diethylene triamine penta-acetic acid.aMean (± SD).bmedian (IQR). Table 1Characteristics of Included Studies Assessing Measurement ErrorAuthorsYearSample sizeRacePopulation AgeMeasure GFR SubstanceeGFR substanceEquationeGFRBiasMeasure time (after kidney transplant)Risch et al24200529White51.3 (± 15.8)a125I-Iothalamate clearanceCreatinineMDRD2 (± 11.8)a6 (± 4.5)a yearsCreatinineRecalibrated MDRD1.7 (± 12.6)a6 (± 4.5)a yearsCystatin CLarsson–4.7 (± 10.6)a6 (± 4.5)a yearsWhite et al252005117Mixed52 (± 11)a99mTc-DTPACreatinineMDRD–11.5 (± 12.2)a7.5 (± 10.8)a yearsCystatin CFiller–1.7 (± 11.4)a7.5 (± 10.8)a yearsCystatin CLe Bricon–3.8 (± 11.8)a7.5 (± 10.8)a yearsCystatin CLarsson–16.1 (± 14.2)a7.5 (± 10.8)a yearsCystatin CHoek–10.6 (± 11.7)a7.5 (± 10.8)a yearsWhite et al262007198Mixed53 (± 12)a99mTc-DTPACreatinineOriginal MDRD–10.7 (± 11.6)a7.2 (± 6.9)a yearsCystatin CFiller0.2 (± 11.5)a7.2 (± 6.9)a yearsCystatin CLe Bricon–2.5 (± 11.2)a7.2 (± 6.9)a yearsCystatin CRule–10.1 (± 11.2)a7.2 (± 6.9)a yearsZahran et al272007103Mixed81.62 (± 20.94)aInulin clearance (GFR > 60 mL/min/1.73 m3)CreatinineMDRD–19.86 (± 23.9)a147 monthsCystatin CLe Bricon–16.93 (± 20.28)a147 monthsCystatin CHoek–23.54 (± 20.66)a147 monthsCystatin CFiller–25.5 (± 28.07)a147 monthsCystatin CLarsson–23.87 (± 25.94)a147 monthsCystatin CRule–18.18 (± 29.05)a147 monthsInulin clearance (GFR < 60 mL/min/1.73 m3)CreatinineMDRD6.57 (± 17.13)a147 monthsCystatin CLe Bricon10.26 (± 14.26)a147 monthsCystatin CHoek3.26 (± 14.63)a147 monthsCystatin CFiller6.9 (± 18.62)a147 monthsCystatin CLarsson1.56 (± 23.02)a147 monthsCystatin CRule–0.86 (± 15.62)a147 monthsMaillard et al282008120White53 (NA)Inulin clearanceCreatinineMDRD8.7 (± 18.2)a12 (3-240)b monthsCystatin CLe Bricon2.8 (± 13.1)a12 (3-240)b monthsCystatin CHoek-4 (± 13.1)#12 (3-240)b monthsCystatin CLarsson-5.9 (± 15.4)a12 (3-240)b monthsCystatin CRule-5 (± 13.3)a12 (3-240)b monthsCreatinineMDRD8.7 (± 18.2)a12 (3-240)b monthsCreatinineCockcroft-Gault–2.6 (± 22.5)a6.8 yearsCystatin CStevens–9.5 (± 13.2)a6.8 yearsCreatinineLe Bricon–1.2 (± 13.5)a6.8 yearsPoge et al292008108White48.9 (NA)99mTc-DTPACreatinineMDRDNA75.4 (3-240)b monthsCystatin CHoekNA75.4 (3-240)b monthsCystatin CFillerNA75.4 (3-240)b monthsCystatin CLarssonNA75.4 (3-240)b monthsQutb et al30200997Asian39.8 (± 12.7)a99mTc-DTPACreatinineMDRD7.7 (± 33.5)aNACreatinineNankivell10.3 (± 26.1)aNACystatin CNA0.31 (± 22.4)aNAYeo et al.312009102Asian42 (± 10.47)a51Cr-EDTACreatinineMDRD0.33 (± 12.5)a11.65 (± 2.68)a daysCreatinineMCQ–0.19 (± 11.8)a11.65 (± 2.68)a daysCreatinineChinese MDRD24.27 (± 16.35)a11.65 (± 2.68)a daysCreatinineJapanese MDRD–17.95 (± 11.06)a11.65 (± 2.68)a daysCreatinineJSN-CKDI–20.96 (± 10.89)a11.65 (± 2.68)a daysCystatin CFiller0.03 (± 17.1)a11.65 (± 2.68)a daysCystatin CLe Bricon–6.4 (± 13.98)a11.65 (± 2.68)a daysCystatin CMacisaac–7.98 (± 14.72)a11.65 (± 2.68)a daysCystatin COrebro-cyst–5.67 (± 16.83)a11.65 (± 2.68)a daysCystatin CHoek–12.72 (± 14.23)a11.65 (± 2.68)a daysCystatin CRule–12.84 (± 15.1)a11.65 (± 2.68)a daysCystatin CLarsson–8.17 (± 21.49)a11.65 (± 2.68)a daysBargnoux et al322012157White50.8 (±12.7)a99mTc-DTPACreatinineMDRD3 (± 10.5)a2 yearsCystatin CStevens1.3 (± 11.2)a2 yearsCystatin C and creatinineNR1.7 (± 9.4)a2 yearsTsujita et al33201273Asian49.1 (± 12.8)aInulin clearanceCreatinineJapanese–3.74 (NA)>6 monthsCystatin CJapanese12.64 (NA)>6 monthsCystatin C and creatinineJapanese4.45 (NA)>6 monthsMasson et al342013670White54 (± 14)aInulin clearanceCreatinineCKD-EPI 20124.4 (± 13)a69 (± 77)a monthsCystatin CCKD-EPI 2012–2.82 (± 12)a69 (± 77)a monthsCystatin C and creatinineCKD-EPI 2012-0.54 (± 11)a69 (± 77)a monthsKukla et al352014153Mixed48.9 (±12.4)aIothalamate clearanceCreatinineMDRD6.4 (± 15.8)a>1 yearCreatinineCKD-EPI11.8 (± 16.6)a>1 yearCystatin CCKD-EPI8.6 (± 15)a>1 yearCystatin C and creatinineCKD-EPI9 (± 12.7)a>1 yearCreatinineMDRD10.8 (± 14.4)a>1 yearCreatinineCKD-EPI15.8 (± 15.5)a>1 yearCystatin CCKD-EPI12.3 (± 13.8)a>1 yearCystatin C and creatinineCKD-EPI12.6 (± 11.2)a>1 yearCreatinineMDRD-0.8 (± 15.2)a>1 yearCreatinineCKD-EPI5.4 (± 16.4)a>1 yearCystatin CCKD-EPI2 (± 13.7)a>1 yearCystatin C and creatinineCKD-EPI2.9 (± 12)a>1 yearLuis-Lima et al362015193White56 (± 12.85)aIohexol clearanceCreatinineMDRD7.37 (NA)>6 monthsCreatinineCKD-EPI11.03 (NA)>6 monthsCystatin CCKD-EPI–3 (NA)>6 monthsCystatin CHoek–0.01 (NA)>6 monthsCystatin CLarsson–2.42 (NA)>6 monthsCystatin C and creatinineCKD-EPI–2.26 (NA)>6 monthsCystatin C and creatinineStevens3.36 (NA)>6 monthsMeeusen et al372015568Mixed55 (± 14)aIothalamate clearanceCreatinineCKD-EPI 20090.5 (−2.2 to 3.2)aNRCystatin CCKD-EPI 2012−5.4 (−7.8 to −3.0)aNRCystatin C and creatinineCKD-EPI 2012−4.8 (−7.0 to −2.6)#NRUslu et al38201544Mixed37.3 (± 9.8)a99mTc-DTPACreatinineMDRD24.9 (± 17.5)a3 monthsCystatin CHoek10.2 (± 14)a3 monthsCystatin CLarsson12.9 (± 20.6)a3 monthsCystatin CLe Bricon16.5 (± 13.6)a3 monthsCystatin CFiller23 (± 18)a3 monthsCreatinineMDRD28.4 (± 18.1)a6 monthsCystatin CHoek15.8 (± 13.3)a6 monthsCystatin CLarsson19.6 (± 22.5)a6 monthsCystatin CLe Bricon22.1 (± 12.9)a6 monthsCystatin CFiller29.2 (± 17.6)a6 monthsCreatinineMDRD26.8 (± 14.5)a12 monthsCystatin CHoek14.7 (± 13.8)a12 monthsCystatin CLarsson17.4 (± 23.2)a12 monthsCystatin CLe Bricon21.1 (± 13.4)a12 monthsCystatin CFiller26.9 (± 18)a12 monthsAl-wakeel et al3920168AsianNRInulinCreatinineMDRD–3.2 (± 7.71)a6 monthsCreatinineCKD-EPI–1.1 (± 8.5)a6 monthsCystatin CCKD-EPI–10.2 (± 17.1)a6 monthsCystatin C and creatinineCKD-EPI–6.5 (± 11.6)a6 monthsKeddis et al4020161139White55.9 (±14.3)aIothalamate clearanceCreatinineMDRD–1.9 (± 14.6)a72 monthsCreatinineCKD-EPI 20090.8 (± 13.7)a72 monthsCystatin CCKD-EPI 2012-8.7 (± 11.7)a72 monthsCystatin C and creatinineCKD-EPI 2012–5 (± 10.7)a72 monthsSalvador et al412017594White52 (23 to 75)bIohexol clearanceCreatinineMDRD0.7 (± 12.3)a1 yearCreatinineCKD-EPI15.4 (± 14.6)a1 yearCreatinineNankivell formula B−4.18 (± 13)a1 yearCystatin CCKD-EPI−9.79 (± 11.5)a1 yearCystatin CCAPA−8.58 (± 10.3)a1 yearCystatin C and creatinineCKD-EPI−4.56 (± 10.7)a1 yearCystatin C and creatinineNew one (this study)1.19 (± 7.59)a1 year51Cr-EDTA clearanceCreatinineMDRD1.64 (± 11.1)a10 weeksCreatinineCKD-EPI3.72 (± 13.5)a10 weeksCreatinineNankivell formula B–4.66 (± 11.5)a10 weeksCystatin CCKD-EPI–7.56 (± 9.75)a10 weeksCystatin CCAPA–6.64 (± 10.4)a10 weeksCystatin C and creatinineCKD-EPI–3.26 (± 8.41)a10 weeksCystatin C and creatinineNew one (this study)2.67 (± 7.8)a10 weeksYang et al42201770Asian46 (± 11.5)a99mTc-DTPACreatinineMDRD–8.08 (± 14.9)aNRCreatinineCKD-EPI–14.54 (± 15.1)aNRCystatin CCKD-EPI–8.36 (± 15.8)aNRCystatin C and creatinineCKD-EPI–10.38 (± 14)aNRTang et al432018252Asian38.1 (± 10.1)a99mTc-DTPACreatinineMDRD21.9 (± 13.0)a5.6 (± 7.2)a monthsCreatinineChinese MDRD42.5 (± 18.4)a5.6 (± 7.2)a monthsCreatinineAmerican MDRD22 (± 13.0)a5.6 (± 7.2)a monthsCreatinineIDMS-MDRD16.7 (± 11.7)a5.6 (± 7.2)a monthsCreatinineCKD-EPI27.8 (± 14.9)a5.6 (± 7.2)a monthsCreatinineEPI-Asian32.9 (± 16.2)a5.6 (± 7.2)a monthsCreatinineJSN-CKDI-6.2 (± 8.1)a5.6 (± 7.2)a monthsCystatin CLarsson-15.8 (± 9.9)a5.6 (± 7.2)a monthsCystatin CRule-14.7 (± 10)a5.6 (± 7.2)a monthsCystatin CHoek–13.8 (± 10.1)a5.6 (± 7.2)a monthsCystatin CFiller–3.9 (± 10.2)a5.6 (± 7.2)a monthsCystatin CGrub–9.2 (± 13)a5.6 (± 7.2)a monthsBorjas-Garcia et al44201930NR40.2 (±14.6)aIothalamate clearanceCreatinineCKD-EPI–2.4 (± 9.6)a13 monthsCreatinineMDRD6.8 (± 11.8)a13 monthsCreatinineCockcroft-Gault–24.4 (± 12.8)a13 monthsCreatinineMCQ–5.9 (± 12.1)a13 monthsCreatinineCreatinine clearance10.6 (± 13.7)a13 monthsCystatin CStevens-9.3 (± 31.4)a13 monthsCystatin C and creatinineCKD-EPI-28.4 (± 28.4)a13 monthsUteil et al452020270White47 (± 14)a51Cr-EDTACreatinineMDRD11.1 (± 12.4)a1 yearCreatinineCKD-EPI16.4 (± 14.8)a1 yearCystatin CCKD-EPI15 (± 23.5)a1 yearCystatin C and creatinineCKD-EPI14.1 (± 16.7)a1 yearMeeusen et al462022220Mixed55 (± 14)aIothalamate clearanceCreatinineCKD-EPI 2021−0.05 (−1.67 to 1.36)a2.14 (± 3.2)a yearCystatin C and creatinineCKD-EPI 2021−3.84 (−4.83 to −2.51)aTrans et al47202283White53 (47 to 62)b51Cr-EDTACreatinineMDRD0.2 (± 10)a3 monthsCreatinineCKD-EPI3.4 (± 11)a3 monthsCystatin CCKD-EPI–12.1 (± 9)a3 monthsCystatin C and creatinineCKD-EPI–6.5 (± 8)a3 months6554 (47 to 62)bCreatinineMDRD–1.7 (± 12.7)a12 monthsCreatinineCKD-EPI1.8 (± 13.5)a12 monthsCystatin CCKD-EPI–11.1 (± 7.5)a12 monthsCystatin C and creatinineCKD-EPI–6.4 (± 8.9)a12 monthsKure et al48202382White53 (46 to 62)bIodothalamate clearanceCreatinineCKD-EPI 20216.34 (± 10.85)a3 monthsCystatin C and creatinineCKD-EPI 2021-5.04 (± 7.74)a6454 (47 to 62)bCreatinineCKD-EPI 20214.82 (± 13.63)a12 monthsCystatin C and creatinineCKD-EPI 2021–4.54 (± 8.18)aKrishnamurthy et al49201130Asia43.1 ± 10.099mTc-DTPACreatinineMDRDNR>6 monthsCystatin CLe BriconNR>6 monthsAbbreviations: AKT, After kidney transplant; CAPA, White and Asian pediatric and adult subjects; 51Cr-EDTA, chromium-51 labelled ethylene diamine tetra-acetic acid; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration Equation; eGFR, estimated glomerular filtration rate; 125I-iothalamate, iodine-125 iothalamate; JSN-CKDI, Japanese Society of Nephrology Chronic Kidney Disease Initiative; MCQ, Mayo Clinic Quadratic; MDRD, original Modification of Diet in Renal Disease equation; 99mTc-DTPA, technetium-99m-diethylene triamine penta-acetic acid.aMean (± SD).bmedian (IQR). Characteristics of Included Studies Assessing Measurement Error Abbreviations: AKT, After kidney transplant; CAPA, White and Asian pediatric and adult subjects; 51Cr-EDTA, chromium-51 labelled ethylene diamine tetra-acetic acid; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration Equation; eGFR, estimated glomerular filtration rate; 125I-iothalamate, iodine-125 iothalamate; JSN-CKDI, Japanese Society of Nephrology Chronic Kidney Disease Initiative; MCQ, Mayo Clinic Quadratic; MDRD, original Modification of Diet in Renal Disease equation; 99mTc-DTPA, technetium-99m-diethylene triamine penta-acetic acid. Mean (± SD). median (IQR). There was considerable variability in how stable allograft function was defined. Most studies used a combination of post-transplant duration and laboratory parameters, or considered ongoing kidney disease or medication use. The most common threshold for post-transplant stability was 6 months. Detailed definitions of stable allograft function for each study are summarized in Table S5. The between-study variation also extends to what calibration assay was use in each study (Table S6). Comparison of pooled mean bias between creatinine- and cystatin C-based equations CKD-EPI (14 studies) The formulas for each equation are summarized in Table S7. For the CKD-EPI equation, the subgroup analysis results were primarily derived from CKD-EPI 2012 for cystatin C and Crea/Cys, whereas CKD-EPI 2009 was the main equation used for creatinine. Only three studies evaluated the CKD-EPI 2021 equation. Figure 3 presents the forest plot of the pooled mean bias of CKD-EPI based GFR estimates derived from creatinine, cystatin C, and Crea/Cys. Overall, the pooled mean biases from cystatin C and creatinine were 5.01 mL/min/1.73 m2 (95% CI: 0.54-9.48; I2 = 76%) and −4.95 mL/min/1.73 m2 (95% CI: −9.10 to –0.81; I2 = 81.2%), respectively. A 95% CI for mean bias that does not include zero indicates a statistically significant bias in the GFR estimate. However, the 95% CIs for both biomarkers overlapped the clinical equivalence threshold of ±5 mL/min/1.73 m2, rendering the clinical interpretation inconclusive. Crea/Cys had the smallest pooled mean bias: 0.63 mL/min/1.73 m2 (95% CI −3.66 to 4.91; I2 = 79.5%), indicating clinical equivalence to mGFR.Figure 3Forest plot showing pooled mean bias of CKD-EPI-based eGFR for each biomarker. Abbreviations: CI, confidence Interval; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration formula. Forest plot showing pooled mean bias of CKD-EPI-based eGFR for each biomarker. Abbreviations: CI, confidence Interval; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration formula. Subgroup and sensitivity analyses Subgroup analysis by post-transplantation timing showed that measurements taken more than 6 months after transplantation had pooled mean biases within the clinical threshold, indicating more precise GFR estimates. In contrast, earlier measurements (less than 6months) yielded pooled mean biases outside this range for all biomarkers (Fig 4).Figure 4Sensitivity and subgroup analyses of pooled mean bias of CKD-EPI-based eGFR for each biomarker. Abbreviations: GFR, glomerular filtration rate; CI, confidence interval; 99mTc-DTPA, Technetium-99m-diethylene triamine penta-acetic acid; 51-Cd-EDTA, chromium-51 labelled ethylene diamine tetra-acetic acid. Sensitivity and subgroup analyses of pooled mean bias of CKD-EPI-based eGFR for each biomarker. Abbreviations: GFR, glomerular filtration rate; CI, confidence interval; 99mTc-DTPA, Technetium-99m-diethylene triamine penta-acetic acid; 51-Cd-EDTA, chromium-51 labelled ethylene diamine tetra-acetic acid. Across mGFR subgroups (≥60 vs ≤60 mL/min/1.73 m2; four studies each), all biomarkers demonstrated consistent pooled mean bias, with cystatin C and Crea/Cys showing slightly lower pooled mean bias than creatinine. In Asian populations (2 studies), Crea/Cys showed a higher positive bias, whereas creatinine and cystatin C produced consistent estimates regardless of ethnicity. Sensitivity analyses performed by sequentially removing each reference substance for mGFR showed only minor deviations from the primary results. Sensitivity analysis excluding imputed studies was also consistent with the main findings, although substantial residual heterogeneity persisted across all subgroup and sensitivity analyses (Fig 4). Other formulars Creatinine The MDRD equation was also frequently applied for creatinine-based estimates (20 studies) and showed error comparable to CKD-EPI, with a pooled mean bias of −4.35 mL/min/1.73 m2 (95% CI: −8.86 to 0.16; I2 = 88.9%). In contrast, the Cockcroft–Gault equation showed a clinically significant negative bias (−15.14 mL/min/1.73 m2; 95% CI: −23.47 to −6.80; I2 = 93.0%). Details of other creatinine-based equations are shown in Figure S1. Cystatin C The Le Bricon (8 studies), Hoek (8studies), and Larsson (9 studies) equations were the most frequently used cystatin C-based equations, all developed before CKD-EPI. The Hoek equation showed the lowest pooled mean bias (0.38 mL/min/1.73 m2; 95% CI, −6.33 to 7.10; I2 = 89.1%), followed by the Larsson equation (1.19 mL/min/1.73 m2; 95% CI, −5.97 to 8.33; I2 = 88.0%) and the Le Bricon equation (−6.20 mL/min/1.73 m2; 95% CI, −12.77 to 0.36; I2 = 88.2%). Details of additional cystatin C-based equations are provided in Figure S2. Crea/Cys Four studies proposed their own equations incorporating both creatinine and cystatin C, with mean bias values ranging from −3.36 to 0.98 mL/min/1.73 m2 (Fig S3). Comparison of pooled P10 and P30 between creatinine- and cystatin C-based equations P10 and P30 were reported in seven studies (2,187 patients) and eleven studies (3,383 patients), respectively. As shown in Figure 5A, CKD-EPI–based eGFR calculated using cystatin C, creatinine, or their combination (Crea/Cys) demonstrated comparable P10 accuracy: 23% (95% CI, 15%-31%; I2 = 88.4%), 29% (95% CI, 18%-41%; I2 = 98.1%), and 31% (95% CI, 21%-42%; I2 = 92.8%), respectively. Similarly, P30 accuracy was consistent across biomarkers, with the highest pooled value observed for Crea/Cys at 78% (95% CI, 68%-88%; I2 = 92.9%), followed by cystatin C at 72% (95% CI, 65%-79%; I2 = 92.8%) and creatinine at 71% (95% CI, 60%-82%; I2 = 97.8%; Fig 5B).Figure 5Forest plot showing pooled P10 (A) and P30 (B) of CKD-EPI-based eGFR for each biomarker. Abbreviations: CI, Confidence interval; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration formula; P10, percentage of estimations within an error of ±10%; P30, percentage of estimations within an error of ±30%. Forest plot showing pooled P10 (A) and P30 (B) of CKD-EPI-based eGFR for each biomarker. Abbreviations: CI, Confidence interval; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration formula; P10, percentage of estimations within an error of ±10%; P30, percentage of estimations within an error of ±30%. Across mGFR subgroups (≥60 and <60 mL/min/1.73 m2), all biomarkers showed comparable P10 and P30 accuracy, although CKD-EPI–based eGFR generally showed lower accuracy when GFR was <60 mL/min/1.73 m2 (Figs S4 and S5). Publication Bias Begg’s test showed significant publication bias for the pooled mean bias of CKD-EPI–based eGFR derived from Crea/Cys (P < 0.001) and for the pooled P30 of CKD-EPI–based eGFR derived from Crea/Cys (LFK index: −2.26). However, no significant publication bias was detected for the other outcomes. All funnel plots and Doi plots are presented in Supplementary Figures S6 through S14. GRADE Assessment for Primary and Secondary Outcomes Table 2 summarizes the GRADE quality ratings for the pooled mean bias, P10, and P30 of CKD-EPI–based eGFR derived from creatinine, cystatin C, and combined Crea/Cys-based equations. The certainty of evidence for the pooled mean bias was rated as low across all biomarkers, primarily because of serious inconsistency. The certainty of evidence for pooled P10 and P30 was rated as moderate for all outcomes, except for pooled P30 from Crea/Cys, which was rated as low.Table 2GRADE Quality Rating of the Pooled Primary and Secondary OutcomesCertainty AssessmentNo of PatientsEffectCertaintyImportanceNo of studiesStudy designRisk of biasInconsistencyIndirectnessImprecisionOther ConsiderationsRelative (95% CI)Absolute (95% CI)Pooled mean bias of CKD-EPI-based eGFR from creatinine14MultidesignNot seriousSeriousaNot seriousSeriousbNo publication bias4,297-–4.95 (–9.10 to –0.81)⨁⨁◯◯LowCRITICALPooled mean bias of CKD-EPI-based eGFR from cystatin C10MultidesignNot seriousSeriousaNot seriousSeriousbNo publication bias3,650-5.01 (0.54 to 9.84)⨁⨁◯◯LowCRITICALPooled mean bias of CKD-EPI-based eGFR from cystatin C and creatinine13MultidesignNot seriousSeriousaNot seriousNot seriousPublication bias4,906-0.63 (–3.66 to 4.92)⨁⨁◯◯LowCRITICALPooled P10 of CKD-EPI-based eGFR from creatinine-based eGFR7MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias2,25229 (18 to 41)⨁⨁⨁◯ModerateIMPORTANTPooled P10 of CKD-EPI-based eGFR from cystatin C5MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias1,78023 (15 to 31)⨁⨁⨁◯ModerateIMPORTANTPooled P10 of CKD-EPI-based eGFR from cystatin C and creatinine6MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias2,00031 (21 to 42)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from creatinine11MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias3,87571 (60 to 82)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from cystatin C8MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias4,84572 (65 to 79)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from cystatin C and creatinine10MultidesignNot seriousSeriousaNot seriousNot seriousPublication bias2,00078 (68 to 88)⨁⨁◯◯LowIMPORTANTaHeterogeneity could not be fully explained by the plausible effects identified in the subgroup and sensitivity analyses.bThe width of the confidence interval exceeded the clinical thresholds of ±5 mL/min/1.73 m2 for mean bias or ±10% for P30 and P10. Table 2GRADE Quality Rating of the Pooled Primary and Secondary OutcomesCertainty AssessmentNo of PatientsEffectCertaintyImportanceNo of studiesStudy designRisk of biasInconsistencyIndirectnessImprecisionOther ConsiderationsRelative (95% CI)Absolute (95% CI)Pooled mean bias of CKD-EPI-based eGFR from creatinine14MultidesignNot seriousSeriousaNot seriousSeriousbNo publication bias4,297-–4.95 (–9.10 to –0.81)⨁⨁◯◯LowCRITICALPooled mean bias of CKD-EPI-based eGFR from cystatin C10MultidesignNot seriousSeriousaNot seriousSeriousbNo publication bias3,650-5.01 (0.54 to 9.84)⨁⨁◯◯LowCRITICALPooled mean bias of CKD-EPI-based eGFR from cystatin C and creatinine13MultidesignNot seriousSeriousaNot seriousNot seriousPublication bias4,906-0.63 (–3.66 to 4.92)⨁⨁◯◯LowCRITICALPooled P10 of CKD-EPI-based eGFR from creatinine-based eGFR7MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias2,25229 (18 to 41)⨁⨁⨁◯ModerateIMPORTANTPooled P10 of CKD-EPI-based eGFR from cystatin C5MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias1,78023 (15 to 31)⨁⨁⨁◯ModerateIMPORTANTPooled P10 of CKD-EPI-based eGFR from cystatin C and creatinine6MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias2,00031 (21 to 42)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from creatinine11MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias3,87571 (60 to 82)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from cystatin C8MultidesignNot seriousSeriousaNot seriousNot seriousNo publication bias4,84572 (65 to 79)⨁⨁⨁◯ModerateIMPORTANTPooled P30 of CKD-EPI-based eGFR from cystatin C and creatinine10MultidesignNot seriousSeriousaNot seriousNot seriousPublication bias2,00078 (68 to 88)⨁⨁◯◯LowIMPORTANTaHeterogeneity could not be fully explained by the plausible effects identified in the subgroup and sensitivity analyses.bThe width of the confidence interval exceeded the clinical thresholds of ±5 mL/min/1.73 m2 for mean bias or ±10% for P30 and P10. GRADE Quality Rating of the Pooled Primary and Secondary Outcomes Heterogeneity could not be fully explained by the plausible effects identified in the subgroup and sensitivity analyses. The width of the confidence interval exceeded the clinical thresholds of ±5 mL/min/1.73 m2 for mean bias or ±10% for P30 and P10. Discussion This study is the first systematic review and meta-analysis in a decade to comprehensively evaluate cystatin C-, creatinine-, and combined Crea/Cys-based eGFR equations, particularly CKD-EPI, against mGFR in KTRs.10 Although all biomarkers demonstrated similar accuracy in the pooled P10 and P30 analyses, the pooled mean bias of CKD-EPI–based eGFR derived from Crea/Cys was clinically equivalent to mGFR. In contrast, creatinine and cystatin C showed significant but opposite directional biases (negative and positive, respectively). These findings indicate that cystatin C and Crea/Cys-based CKD-EPI equations may provide clinical value as alternatives to creatinine in KTRs. The apparent superiority of the CKD-EPI Crea/Cys equation over the others may reflect bias cancellation rather than a true improvement in individual-level precision. As observed in the results, creatinine- and cystatin C-based CKD-EPI equations exhibited significant but opposing biases creating conditions for cancellation of mean bias when combined. Bias variance, for example root mean square error, could not be directly compared across studies. However, the similar pooled P10 and P30 values across all three equations are not subject to bias cancellation. This suggests that the observed superiority is likely attributable to a cancellation effect rather than improved precision. Interpretation should further examine the sources of the opposing biases observed between creatinine- and cystatin C-based CKD-EPI equations. In KTRs, physiological factors may alter eGFR estimates. Creatinine-based eGFR is influenced by non-GFR determinants, particularly muscle mass and diet, which may lead to cohort-dependent over- or underestimation.40 Although cystatin C is less affected by muscle mass, it is influenced by non-GFR determinants such as steroid exposure and inflammatory burden. These factors vary over time and across centers in KTRs and may shift cystatin C-based eGFR upward.40 Beyond patient-level factors, differences in mGFR reference methods and assay calibration, particularly the known assay dependence of cystatin C, can further systematically influence bias.11,34,50 Unfortunately, because of reporting limitations, we cannot determine whether the apparent bias cancellation observed in the combined equation reflects a cohort-specific pattern in KTRs or arises from variation in GFR measurement systems across studies. Accordingly, this finding should not be interpreted as a universal improvement in individual-level precision. According to The KDIGO 2024 for CKD guideline, the use of the most accurate validated equations is recommended, preferably those combining creatinine and cystatin C.51 However, most GFR-estimating equations were developed for general or CKD populations; therefore, their applicability to specialized clinical settings may be limited.52 The WhiteWhite subgroup showed a smaller pooled mean bias than the Asian subgroup. This finding is consistent with evidence from non-KTR populations, for which neither equation performs as well in Asian patients, highlighting the need for further validation in this group.53 In 2014, Grubb et al54 proposed a new cystatin C-based equation that included both Asian and White participants, primarily from Japan and Sweden. However, only 1 study has validated this equation in Asian KTRs, and modest accuracy was reported.43 Our study also evaluated the performance of these biomarkers across different levels of kidney function. All biomarkers, when applied to CKD-EPI–based eGFR, showed consistent pooled mean bias among patients with mGFR <60 mL/min/1.73 m2 and those with mGFR ≥60 mL/min/1.73 m2. A previous meta-analysis by Lu-Xi Zou et al55, which compared both biomarkers in 23,667 adults from primary health care settings, reported a similar pattern across both kidney function categories. Refining the GFR assessment is essential for optimizing patient management and reducing misclassification. Although the most recent KDIGO guidelines for kidney transplant recipients2 and the 2017 British Transplantation Society guidelines continue to recommend creatinine rather than cystatin C, our findings provide evidence supporting cystatin C as an alternative biomarker, particularly when combined with creatinine to improve accuracy.56 Further research incorporating long-term clinical outcomes is needed to better define its role in the management of KTRs. Significant heterogeneity was observed. We interpret this heterogeneity as arising from both structural and non-structural sources. Several structural (systematic) heterogeneities may have contributed substantially. First, studies differed in assay calibration and standardization. Estimated GFR equations are designed for standardized serum creatinine (IDMS-traceable) and standardized cystatin C (IFCC/ERM-DA471-traceable) values.57 The differences in calibration protocols can shift biomarker values and thereby the calculated eGFR independently of patient physiology. Second, studies used different mGFR reference methods (and protocols), which are not strictly interchangeable because tracer properties and modeling assumptions can systematically influence measured values.58 In addition, variability in mGFR reproducibility under similar conditions has been reported, ranging from 5.0% to 8.0% for iohexol, approximately 4.0% for chromium-51 labelled ethylene diamine tetra-acetic acid (51Cr-EDTA), and 8.5% for iothalamate.59, 60, 61, 62 Third, body-size normalization and related correction conventions can affect reported GFR; even within a single tracer (eg, 51Cr-EDTA), methodological ordering (eg, body surface area indexing vs Bröchner–Mortensen correction) can change the resulting GFR by approximately 5.5% in a low-body surface area or increasing slow-GFR setting.63 Fourth, timing relative to transplant physiology differed by study. Inclusion of recipients early post-transplant or under high-dose steroids may violate steady-state and change non-GFR determinants, particularly for cystatin C, and therefore shift bias in a way that cannot be treated as purely random.64 Studies that included unstable patients, especially those assessed within six months after transplantation, were rated as having some concerns, and subgroup analyses confirmed greater bias in this subgroup.55,65 Nonstructural heterogeneity likely contributed to the remaining variability, including differences in case-mix (age, comorbid conditions, donor characteristics), underlying kidney disease, and sampling variability, as well as residual within-person biological variation and analytical imprecision. Several limitations should be acknowledged. First, the search was restricted to studies published in English and excluded gray literature, which may have introduced publication bias. However, language bias has reportedly decreased as English-language publications have become more prevalent, and including gray literature may itself introduce bias. Publication bias was assessed, and the certainty of evidence was adjusted accordingly. Second, substantial heterogeneity was observed. Although stringent thresholds (±5 mL/min/1.73 m2 for pooled mean bias and ±10% for P30 and P10) were applied and sensitivity analyses of reported moderators were performed, concerns extend beyond the magnitude of statistical heterogeneity to its nature. The included studies differed in ways that are best understood as structural heterogeneity, including creatinine and cystatin C assay traceability/standardization, mGFR reference methods and protocols, and timing relative to post-transplant physiology. These factors are not random sources of variation that can be fully averaged through pooling; rather, they reflect partially non-interchangeable measurement frameworks and therefore impose a ceiling on the achievable precision, interpretability, and transportability of pooled estimates of bias and accuracy. We encourage future studies to systematically report these moderators to improve interpretability and facilitate meaningful synthesis. Lastly, some outcome data were not directly reported and had to be extracted, imputed, or derived from surrogate measures. Although this may have affected data quality, analyses using standardized reference methods and sensitivity analyses excluding imputed studies yielded results consistent with the primary findings. Conclusion When averaged across heterogeneous KTR studies, the CKD-EPI equation, Crea/Cys demonstrates lower mean bias and slightly higher pooled P30 than single-marker equations. Accordingly, when cystatin C is available, the combined equation is a reasonable choice. However, structural heterogeneity (assay calibration, mGFR methods) and nonsteady state transplant physiology may influence performance estimates. Thus, these pooled estimates should be used to inform equation selection at the population level rather than to infer improved accuracy for individual patients. Further head-to-head multicenter validation studies in KTRs, with harmonized protocols for both biomarkers, are needed to confirm performance consistency in local populations.