Luis-Lima, Sergio; Escamilla-Cabrera, Beatriz; Negrín-Mena, Natalia; Estupiñán, Sara; Delgado-Mallén, Patricia; Marrero-Miranda, Domingo; González-Rinne, Ana; Miquel-Rodríguez, Rosa; Cobo-Caso, María Ángeles; Hernández-Guerra, Manuel; Oramas, Juana; Batista, Norberto; Aldea-Perona, Ana; Jorge-Pérez, Pablo; González-Alayón, Carlos; Moreno-Sanfiel, Miguel; González-Rodríguez, Juan Antonio; Henríquez, Laura; Alonso-Pescoso, Raquel; Díaz-Martín, Laura; González-Rinne, Federico; Lavín-Gómez, Bernardo Alio; Galindo-Hernández, Judith; Sánchez-Gallego, Macarena; González-Delgado, Alejandra; Jiménez-Sosa, Alejandro; Torres, Armando; Porrini, Esteban

Add Journal to My Library
Nephrology Dialysis Transplantation
, Volume Advance Article – May 11, 2018

8 pages

/lp/ou_press/ckd-staging-with-cystatin-c-or-creatinine-based-formulas-flipping-the-O7WlFiSWez

- Publisher
- Oxford University Press
- Copyright
- © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
- ISSN
- 0931-0509
- eISSN
- 1460-2385
- D.O.I.
- 10.1093/ndt/gfy086
- Publisher site
- See Article on Publisher Site

Abstract Background Chronic kidney disease (CKD) affects 10–13% of the population worldwide. CKD classification stratifies patients in five stages of risk for progressive renal disease based on estimated glomerular filtration rate (eGFR) by formulas and albuminuria. However, the reliability of formulas to reflect real renal function is a matter of debate. The effect of the error of formulas in the CKD classification is unclear, particularly for cystatin C–based equations. Methods We evaluated the reliability of a large number of cystatin C and/or creatinine-based formulas in the definition of the stages of CKD in 882 subjects with different clinical situations over a wide range of glomerular filtration rates (GFRs) (4.2–173.7 mL/min). Results Misclassification was a constant for all 61 formulas evaluated and averaged 50% for creatinine-based and 35% for cystatin C–based equations. Most of the cases were misclassified as one stage higher or lower. However, in 10% of the subjects, one stage was skipped and patients were classified two stages above or below their real stage. No clinically relevant improvement was observed with cystatin C–based formulas compared with those based on creatinine. Conclusions The error in the classification of CKD stages by formulas was extremely common. Our study questions the reliability of both cystatin C and creatinine-based formulas to correctly classify CKD stages. Thus the correct classification of CKD stages based on estimated GFR is a matter of chance. This is a strong limitation in evaluating the severity of renal disease, the risk for progression and the evolution of renal dysfunction over time. CKD staging, creatinine, cystatin C, estimated GFR, measured GFR INTRODUCTION The prevalence of chronic kidney disease (CKD) has reached epidemic proportions, affecting ∼10% of the population worldwide [1]. CKD is a risk factor for end-stage renal disease, cardiovascular morbidity, premature death and low quality of life [2]. These adverse outcomes could be prevented by early diagnosis and treatment of renal disease [3]. In 2002, a classification of CKD was proposed by the National Kidney Foundation Kidney Disease Outcomes Quality Initiative [4]. Accordingly, patients with kidney damage are stratified in subgroups of glomerular filtration rate (GFR): CKD-1, >90 mL/min/1.73 m2; CKD-2, 60–89 mL/min/1.73 m2; CKD-3, 30–59 mL/min/1.73 m2; CKD-4, 15–29 mL/min/1.73 m2 and CKD-5, < 15 mL/min/1.73 m2. In 2012, this classification was modified [5], combining albuminuria levels with the stages of CKD to establish a risk score for progressive renal disease. The CKD classification was developed to provide a practical guideline for the diagnosis of renal dysfunction in order to reduce the risks associated with CKD and to prevent disease progression. However, a limitation of the CKD classification is the use of estimated GFR (eGFR) by formulas. Formulas are neither accurate nor precise in reflecting true GFR [6–22], which may lead to over- or underestimation of real GFR and misclassification of patients in CKD stages. Nevertheless, the impact of the error of eGFR in the classification of CKD stages has seldom been evaluated [6–8, 10, 11, 16, 19, 20]. These studies observed that between one-third and one-half of the patients were misclassified in CKD stages. These studies evaluated few formulas, mostly creatinine based, and little evidence is available about cystatin C–based equations in the classification of CKD stages. The present study aimed to evaluate the reliability of a large number of cystatin C and/or creatinine-based formulas in the definition of CKD stages in patients with different clinical situations over a wide range of GFRs. MATERIALS AND METHODS Patients In our centre we perform plasma clearance of iohexol in clinical practice and research. For this study we selected consecutive patients >18 years of age who underwent plasma clearance of iohexol from July 2013 to October 2017 with different clinical conditions: renal transplantation, diabetic nephropathy, pre-dialysis, autosomal dominant polycystic kidney disease (ADPKD), glomerulonephritis, interstitial nephritis, nephroangioesclerosis, cancer patients on chemotherapy, heart failure, cirrhosis, liver transplantation and living kidney donors, among others (Table 1). We created a meta-database of 882 individuals and merged the necessary variables for analysis. Table 1 Clinical characteristics of the patients included in the study N 882 Age (years) 57.4 ± 14.2 Gender (male), n (%) 611 (69.3) No renal disease, n (%) Healthy subjects 50 (5.5) Living kidney donors 89 (9.8) Renal disease, n (%) Kidney transplantation 255 (28.1) Diabetic nephropathy 125 (13.8) Glomerulonephritis 45 (5.0) Interstitial nephritis 45 (5.0) Nephroangioesclerosis 43 (4.7) CKD 38 (4.2) ADPKD 26 (2.9) Primary hyperoxaluria 5 (0.6) Other clinical conditions, n (%) Heart failure 54 (5.9) Cirrhosis 44 (4.8) Liver transplantation 36 (4.0) Oncology 14 (1.5) Other 12 (1.3) Unknown 27 (3.0) Measured GFR (mL/min), mean ± SD 61.0 ± 32.2 Measured GFR range (mL/min) 4.2–173.7 CKD stage (mL/min), n (%) 1 (>90) 186 (20.5) 2 (60–90) 258 (28.4) 3 (30–60) 255 (28.1) 4 (15–30) 181 (19.9) 5 (<15) 28 (3.1) Height (m) 1.68 ± 0.09 Weight (kg) 81.4 ± 17.5 Body mass index (kg/m2), mean ± SD 28. 8 ± 5.5 Body surface area (m2), mean ± SD 1.88 ± 0.31 Serum creatinine (mg/dL), mean ± SD 1.68 ± 1.12 Serum cystatin C (g/dL), mean ± SD 1.75 ± 0.72 24-h creatinine clearance (mL/min), median (IQR) 65.7 (33.2–89.2) 24-h proteinuria (mg/24 h), median (IQR) 765.9 (113.3–650.5) N 882 Age (years) 57.4 ± 14.2 Gender (male), n (%) 611 (69.3) No renal disease, n (%) Healthy subjects 50 (5.5) Living kidney donors 89 (9.8) Renal disease, n (%) Kidney transplantation 255 (28.1) Diabetic nephropathy 125 (13.8) Glomerulonephritis 45 (5.0) Interstitial nephritis 45 (5.0) Nephroangioesclerosis 43 (4.7) CKD 38 (4.2) ADPKD 26 (2.9) Primary hyperoxaluria 5 (0.6) Other clinical conditions, n (%) Heart failure 54 (5.9) Cirrhosis 44 (4.8) Liver transplantation 36 (4.0) Oncology 14 (1.5) Other 12 (1.3) Unknown 27 (3.0) Measured GFR (mL/min), mean ± SD 61.0 ± 32.2 Measured GFR range (mL/min) 4.2–173.7 CKD stage (mL/min), n (%) 1 (>90) 186 (20.5) 2 (60–90) 258 (28.4) 3 (30–60) 255 (28.1) 4 (15–30) 181 (19.9) 5 (<15) 28 (3.1) Height (m) 1.68 ± 0.09 Weight (kg) 81.4 ± 17.5 Body mass index (kg/m2), mean ± SD 28. 8 ± 5.5 Body surface area (m2), mean ± SD 1.88 ± 0.31 Serum creatinine (mg/dL), mean ± SD 1.68 ± 1.12 Serum cystatin C (g/dL), mean ± SD 1.75 ± 0.72 24-h creatinine clearance (mL/min), median (IQR) 65.7 (33.2–89.2) 24-h proteinuria (mg/24 h), median (IQR) 765.9 (113.3–650.5) IQR, interquartile range. Table 1 Clinical characteristics of the patients included in the study N 882 Age (years) 57.4 ± 14.2 Gender (male), n (%) 611 (69.3) No renal disease, n (%) Healthy subjects 50 (5.5) Living kidney donors 89 (9.8) Renal disease, n (%) Kidney transplantation 255 (28.1) Diabetic nephropathy 125 (13.8) Glomerulonephritis 45 (5.0) Interstitial nephritis 45 (5.0) Nephroangioesclerosis 43 (4.7) CKD 38 (4.2) ADPKD 26 (2.9) Primary hyperoxaluria 5 (0.6) Other clinical conditions, n (%) Heart failure 54 (5.9) Cirrhosis 44 (4.8) Liver transplantation 36 (4.0) Oncology 14 (1.5) Other 12 (1.3) Unknown 27 (3.0) Measured GFR (mL/min), mean ± SD 61.0 ± 32.2 Measured GFR range (mL/min) 4.2–173.7 CKD stage (mL/min), n (%) 1 (>90) 186 (20.5) 2 (60–90) 258 (28.4) 3 (30–60) 255 (28.1) 4 (15–30) 181 (19.9) 5 (<15) 28 (3.1) Height (m) 1.68 ± 0.09 Weight (kg) 81.4 ± 17.5 Body mass index (kg/m2), mean ± SD 28. 8 ± 5.5 Body surface area (m2), mean ± SD 1.88 ± 0.31 Serum creatinine (mg/dL), mean ± SD 1.68 ± 1.12 Serum cystatin C (g/dL), mean ± SD 1.75 ± 0.72 24-h creatinine clearance (mL/min), median (IQR) 65.7 (33.2–89.2) 24-h proteinuria (mg/24 h), median (IQR) 765.9 (113.3–650.5) N 882 Age (years) 57.4 ± 14.2 Gender (male), n (%) 611 (69.3) No renal disease, n (%) Healthy subjects 50 (5.5) Living kidney donors 89 (9.8) Renal disease, n (%) Kidney transplantation 255 (28.1) Diabetic nephropathy 125 (13.8) Glomerulonephritis 45 (5.0) Interstitial nephritis 45 (5.0) Nephroangioesclerosis 43 (4.7) CKD 38 (4.2) ADPKD 26 (2.9) Primary hyperoxaluria 5 (0.6) Other clinical conditions, n (%) Heart failure 54 (5.9) Cirrhosis 44 (4.8) Liver transplantation 36 (4.0) Oncology 14 (1.5) Other 12 (1.3) Unknown 27 (3.0) Measured GFR (mL/min), mean ± SD 61.0 ± 32.2 Measured GFR range (mL/min) 4.2–173.7 CKD stage (mL/min), n (%) 1 (>90) 186 (20.5) 2 (60–90) 258 (28.4) 3 (30–60) 255 (28.1) 4 (15–30) 181 (19.9) 5 (<15) 28 (3.1) Height (m) 1.68 ± 0.09 Weight (kg) 81.4 ± 17.5 Body mass index (kg/m2), mean ± SD 28. 8 ± 5.5 Body surface area (m2), mean ± SD 1.88 ± 0.31 Serum creatinine (mg/dL), mean ± SD 1.68 ± 1.12 Serum cystatin C (g/dL), mean ± SD 1.75 ± 0.72 24-h creatinine clearance (mL/min), median (IQR) 65.7 (33.2–89.2) 24-h proteinuria (mg/24 h), median (IQR) 765.9 (113.3–650.5) IQR, interquartile range. Measured GFR Briefly, the morning of the study, 5 mL of iohexol (Omnipaque 300, GE Healthcare, Cardiff, UK) was injected intravenously for 2 min. Afterwards, venous or capillary blood were obtained by finger prick at 120, 180, 240, 300, 360, 420 and 480 min for patients with eGFR ≤40 mL/min/1.73 m2 or at 120, 150, 180, 210 and 240 min for those with eGFR >40 mL/min/1.73 m2. Iohexol was measured in plasma or dried blood spot (DBS) as previously shown [16, 23]. Both methods using plasma or DBS showed excellent agreement and can be considered interchangeable [23]. Iohexol levels were measured by high-performance liquid chromatography (HPLC), as previously decribed by Krutzén et al. [24] and Niculescu-Duvaz et al. [25], with some modifications [23]. For the DBS analysis, a fixed volume of capillary blood (10 µL) was taken by a capillary pipette and deposited on a filter paper. A circle of filter paper containing the whole drop of blood was then punched out for analysis [23]. Plasma clearance of iohexol was calculated according to a one-compartment model and then corrected by the formula proposed by Bröchner-Mortensen [26]. eGFR by formulas Simultaneously with the plasma clearance of iohexol, serum creatinine and cystatin C were determined by calculating 65 equations: 39 creatinine-based, 20 cystatin-C based, and 6 that use both markers (Supplementary data, Table S1). The agreement between formulas and measured GFR (mGFR) was evaluated with the formulas unadjusted for body surface area (BSA). When eGFR was adjusted, we reversed the adjustment of the result by applying the following formula (GFR adjusted = GFR unadjusted/BSA × 1.73). BSA was calculated by the Du-Bois and Du-Bois formula [27]. Biochemistry Creatinine was measured by isotope dilution mass spectrometry–traceable creatinine (cobas c711 module, Roche Diagnostics, Basel, Switzerland) and cystatin C levels by immunonephelometry (BN II System, Siemens Healthcare Diagnostics, Erlangen, Germany), calibrated with ERM-DA471/IFCC. Classification in CKD stages The patients were classified in CKD stages using mGFR and eGFR with 65 formulas (Supplementary data, Table S1). We used the CKD classification of 2002 [4], which does not divide Stage 3 into 3A and 3B, in order to avoid the creation of two additional subgroups with a narrow range of GFRs (15 mL/min). However, we determined the variability of eGFR around the cut-off value of 45 mL/min (Supplementary data, Figure S1A). The agreement between the classification in CKD stages based on eGFR and mGFR was analysed as follows: for each stage we analysed (i) ‘true positive cases’: the subjects correctly classified by a formula in the corresponding stage defined by mGFR; (ii) ‘false positive cases’: subjects who do not belong to the stage of CKD based on mGFR but were incorrectly classified as such by a formula. Both true positive and false positive cases represent 100% of the cases defined by a formula in a given stage of CKD. Finally we analysed the (iii) ‘missing cases’: the subjects that belong to a given stage defined by mGFR but classified in a higher or lower stage by a formula. For example, for CKD-2, mGFR classified 250 patients and 280 by a formula, then 120 were correctly classified by the formula (true positive) and 160 were incorrectly classified in this stage (false positives). Thus true positives + false positives = the number of patients classified by the formula (120 +160 = 280). Finally, the formula did not define in this group 130 subjects that belong to CKD-2 and were classified in different stages, who represent the missing cases. Therefore true positives + missing = the number of patients classified by mGFR (120 +130 = 250). Finally, we evaluated the number of cases in which one stage was skipped and patients were not classified in a consecutive stage but in an even higher or lower subgroup (Supplementary data, Table S3). Sensitivity analyses Since some formulas were developed in specific populations, such as obesity (Salazar-Corcoran), renal transplantation (Nankivell-A and B) or elderly subjects (BIS-1 and 2), we compared the agreement of these equations with mGFR in patients with and without these conditions. Also, to evaluate whether formulas perform better in some specific clinical conditions, such as type 2 diabetes, renal transplantation and chronic nephropaties, we compared the agreement between mGFR and a representative number of formulas in these three subgroups of patients. Statistical analysis Agreement between eGFR and mGFR The performance of eGFR in reflecting mGFR was assessed by statistics of agreement for continuous data, including the concordance correlation coefficient (CCC), total deviation index (TDI) and coverage probability (CP) [28–30]. The CCC varies from 0 to 1 and combines meaningful components of accuracy and precision. A CCC >0.90 reflects optimal concordance between measurements. The TDI captures a large proportion of data within a boundary for allowed differences between two measurements. Empirical TDI was calculated for a theoretical TDI of 10% and a CP of 90%. According to this level of TDI, we defined a priori that the acceptable bias between eGFR and mGFR should be at least 10%. This is based on previous reports and the reproducibility of the method in our laboratory, which is <7%. The CP varies from 0 to 1 and estimates whether a given TDI is less than a prespecified fixed percentage. Interrater agreement between eGFR and mGFR Inter-rater agreement was evaluated using the K-index, which can be interpreted as follows: K < 0.20, poor; K = 0.21–0.40, fair; K = 0.41–0.60, moderate; K = 0.61–0.80, good and K = 0.81–1.00, very good agreement [31]. We used the statistical package AGP (Agreement Program) v.1.0 (IGEKO, SP), available at www.ecihucan.es/lfr/apps/? dir=agreement_installer, which is based on the R code developed by Lin et al. [30]. We also analysed data using SPSS Statistics for Windows, version 17.0 (SPSS, Chicago, IL, USA) and MedCalc Statistical Software version 13.0.2 (MedCalc Software, Ostend, Belgium). RESULTS Patients Of 882 patients, 178 (20.4%) were classified as CKD-1, 252 (28.4%) as CKD-2, 251 (28.3%) as CKD-3, 176 (20.0%) as CKD-4 and 25 (2.7%) as CKD-5 (Table 1). The average age was 57.3 ± 14.2 years, 69% of patients were male and mGFR averaged 60.9 ± 32.2 mL/min (range 4.2–173.7) (Table 1). Classification of CKD stages with cystatin C or creatinine-based formulas CKD-1 (n = 178) The number of patients classified by eGFR ranged from 58 (Lund-Malmö) to 366 (Perkins) (Table 2 and Supplementary data, Table S2). True positive cases averaged 70%, missing cases 30% and false positives 42% for creatinine-based formulas and 88%, 12% and 35%, respectively, for cystatin C–based equations. Most missing cases were classified as CKD-2. However, in 10% of the cases with creatinine-based formulas, one stage was skipped and patients were defined as CKD-3 (Supplementary data, Table S3). For example, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation based on creatinine defined 222 patients as CKD-1; of these, 136 were correctly (true positive) and 86 were incorrectly classified as CKD-1 (false positive) (Table 2). Finally, 42 (24%) of the 178 patients who were classified as CKD-1 by mGFR were not defined as CKD-1 by this formula (missing cases). Table 2 Classification of patients in CKD stages by a representative group of nine creatinine and/or cystatin C–based formulas ‘True positives cases’ represent the subjects that were corretly classified in each CKD stage by eGFR. ‘False positives cases’ represent the patients who were classified in one CKD stage based on eGFR when actually belonging to a different stage. ‘Missing cases’ represent the cases that were not classified in the corresponding CKD stage. a The percentage of false positive cases refers to the number of cases defined in each CKD stage by mGFR (grey column). The percentage of true positive and missing cases refers to the number of cases defined in each CKD stage by eGFR. Table 2 Classification of patients in CKD stages by a representative group of nine creatinine and/or cystatin C–based formulas ‘True positives cases’ represent the subjects that were corretly classified in each CKD stage by eGFR. ‘False positives cases’ represent the patients who were classified in one CKD stage based on eGFR when actually belonging to a different stage. ‘Missing cases’ represent the cases that were not classified in the corresponding CKD stage. a The percentage of false positive cases refers to the number of cases defined in each CKD stage by mGFR (grey column). The percentage of true positive and missing cases refers to the number of cases defined in each CKD stage by eGFR. CKD-2 (n = 252) Patients classified by formulas ranged from 133 (Grubb) to 321 (Nankivell-SPK) (Table 2 and Supplementary data, Table S2). True positives, missing cases and false positives averaged 50% for creatinine-based formulas and 50%, 50% and 36%, respectively, for cystatin C–based equations. Two-thirds of the missing cases were classified as CKD-1 and one-third as CKD-3 for both creatinine and cystatin C–based formulas. For example, the CKD-EPI (cystatin C) equation defined 182 patients as CKD-2; of these, 128 were true positive cases and 54 were false positive cases (Table 2). Also, 124 (49%) of the 252 patients who were classified as CKD-2 by mGFR were not defined as such by this formula (missing cases). CKD-3 (n = 251) Patients classified by formulas in this stage ranged from 80 (Robinson) to 369 (Baracskay) (Table 2 and Supplementary data, Table S2). True positive cases averaged 54%, missing cases 46% and false positive cases 42% for creatinine-based formulas and 69%, 31% and 33%, respectively, for cystatin C–based equations. Missing cases were classified either as CKD-2 (66% for creatinine, 60% for cystatin C), CKD-4 (15%, 32%) or as CKD-1 (18%, 7%). For example, abbreviated modification of diet in renal disease formula (aMDRD) defined 257 patients as CKD-3: 166 of them were true positive and 91 false positive (Table 2). Also, 85 (34%) of the 251 cases classified as CKD-3 according to mGFR were not defined in this stage (missing) based on aMDRD. CKD-4 (n = 176) Patients classified by eGFR ranged from 28 (Perkins) to 213 (Lund-1) (Table 2 and Supplementary data, Table S2). True positive cases averaged 57%, missing cases 43% and false positive cases 25% with creatinine-based equations and 64%, 36%, 26%, respectively, with cystatin C–based formulas. Missing cases were classified as CKD-3 (71%) for both types of formulas and CKD-5 (19% for creatinine, 25% for cystatin C). Also, 10% of the creatinine-based estimations skipped one stage and were defined as CKD-2. For example, the Mayo Clinic Quadratic (MCQ) (cystatin C) formula defined 210 patients; of these, 133 were true positive and 77 were false positive cases (Table 2). Finally, 43 (24%) of the 176 patients who were classified as CKD-4 by mGFR were not defined in the same stage by this formula (missing cases). CKD-5 (n = 25) Patients classified by formulas in this stage ranged from 1 (Le Bricon, Perkins) to 88 (Grubb-1) (Table 2 and Supplementary data, Table S2). True positive cases averaged 30%, missing cases 70% and false positive cases 59% for creatinine-based equations and 44%, 56% and 39% for cystatin C–based formulas. Most missing cases were classified as CKD-4. Again, 10% of the estimations using creatinine skipped one stage and were defined as CKD-3. For example, the Le Bricon equation (cystatin C) defined only 1 patient as CKD-5 (true positive case), instead of 25 that were classified by mGFR. The remaining 24 cases were not defined as CKD-5 by this equation (missing cases). Classification of CKD stages with cystatin C and creatinine-based formulas These formulas showed a number of true positive, false positive and missing cases for each stage of CKD, comparable with the equations that only include cystatin C (Table 2 and Supplementary data, Table S2). For example, the CKD-EPI (creatinine + cystatin C) formula classified 226 patients as CKD-3 instead of 251; of these, 173 were true positive and 53 were false positive cases (Table 2). Finally, 78 (31%) of the 251 patients who were defined as CKD-3 by mGFR were not classified in the same stage by this formula (missing cases). Similar results were observed for more recent equations like the full age spectrum (FAS) combined formula or other approaches based on the mean of two different equations: revised Lund-Malmö (LMR) and Caucasian, Asian, pediatric, and adult cohorts (CAPA) (Table 2 and Supplementary data, Table S2). Interrater agreement between eGFR and mGFR for the classification of CKD stages For creatinine-based algorithms, K-values ranged from 0.530 to 0.204 for the Mayo Clinic Quadratic (MCQ)-CKD and Robinson formulas, respectively (Supplementary data, Table S5). For cystatin C–based algorithms, K-values ranged from 0.632 to 0.245 for the Stevens-2 and Perkins equations, respectively. For the combined creatinine and cystatin C algorithms, K-values ranged from 0.653 to 0.431 for the Stevens and FAS_crcy formulas, respectively. So no formula showed very good strength of agreement (K = 0.81–1.00) in classifying patients in CKD stages. Few formulas, mainly cystatin C based, had good agreement (K = 0.61–0.80) (Supplementary data, Table S5). Agreement between measured GFR and eGFR Creatinine-based formulas The TDI averaged 86%, ranging from 51.1 to 175.4 for the Lund-Malmö (Rv) and Robinson equations, respectively (Supplementary data, Table S4). As an example, modification of diet in renal disease formula (MDRD) formula had a TDI of 54.2, meaning that 90% of the estimations of GFR showed an error ranging from −54 to +54% when compared with mGFR. The CCC averaged 0.74 for all the formulas, reflecting a low level of precision and accuracy, ranging from 0.55 to 0.92 for the Robinson and Mean LMR+CAPA equations, respectively. Finally, the CP averaged 22 for all the formulas, indicating that >78% of the estimations had an error > ±10%. Cystatin C–based formulas The TDI averaged 60%, ranging from 44.2 to 108.7 for the Stevens-2 and Perkins equations, respectively (Supplementary data, Table S4). As an example, the MCQ formula had a TDI of 57.0, meaning that 90% of the estimations of GFR showed an error ranging from −57 to +57% when compared with mGFR. The CCC averaged 0.89 for all the formulas, reflecting a moderate level of precision and accuracy, ranging from 0.74 to 0.94, for the Stevens-2 and Perkins equations, respectively. Finally, the CP averaged 29 for all the formulas, indicating that >71% of the estimations had an error > ± 10%. The CP ranged from 8 to 33 for the Stevens-2 and Perkins equations, respectively. No formula showed 90% of the estimations within bounds of error of ±10% compared with the gold standard. Creatinine and cystatin C–based formulas The TDI averaged 44%, ranging from 38.3 to 50.0 for the Mean LMR+CAPA and Ma equations, respectively (Supplementary data, Table S4). As an example, the CKD-EPI formula had a TDI of 40.6, meaning that 90% of the estimations of GFR showed an error ranging from −41% to +41% when compared with mGFR. The CCC averaged 0.93 for all the formulas, reflecting a moderate level of precision and accuracy, ranging from 0.92 to 0.95 for the BIS-2 and CKD-EPI, and Mean LMR+CAPA equations, respectively. Finally, the CP averaged 32 for all the formulas, indicating that >68% of the estimations had an error > ±10%. Sensitivity analyses Specific formulas did not outperform in the population in which they were designed. The TDI, CCC and CP of the Salazar-Corcoran, Nankivell-A and B or BIS-1 and 2 were comparable in obese and non-obese subjects, renal transplanted patients and non-transplanted patients or young versus elderly subjects, respectively (Supplementary data, Table S6). Also, no formula was particularly reliable in a specific subgroup of patients, such as diabetes, renal transplantation or chronic nephropathies (Supplementary data, Tables S7–S9). DISCUSSION We observed that the error in the classification of CKD stages using eGFR by formulas was very common. The misclassification for all 65 formulas averaged 50% for creatinine-based and 35% for cystatin C–based equations. This error included incorrect classification in a given stage as well as allocation in a consecutive higher or lower stage. Also, in 10% of the cases the error was such that one stage was skipped and patients were classified two stages above or below their current CKD stage. Finally, no clinically relevant improvement in the classification of CKD stages was observed with the use of cystatin C–based formulas compared with creatinine-based algorithms. The error of formulas in the classification of CKD stages is a consequence of the error of eGFR in reflecting mGFR. Previous studies have shown that the error of formulas is frequent and wide, about ±20 to ±30% of mGFR. This has been observed in diabetes [9–15], kidney transplantation [16–20], polycystic kidney disease [21, 22], cancer [32], cirrhosis [33], liver transplantation [34] and patients with human immunodeficiency virus [35]. In the present study, we confirmed the low precision and accuracy of a large group of formulas using specific agreement analysis. TDI values averaged ±86% for creatinine-based formulas, ±60% for cystatin C–based equations and ±44% for those that combine both markers. These TDI values indicate that 90% of the estimations of eGFR are included within a limit of 86%, 60% or 44% of mGFR for each type of equation, respectively. Thus the estimation of real GFR by any formula is extremely variable. In example, in a subject with mGFR of 60 mL/min, eGFR may range from 42 (−30%) to 78 mL/min (+30%) or in another with mGFR of 30 mL/min, eGFR may range from 21 to 39 mL/min. This has important consequences in the allocation of patients in subgroups of GFR. This is illustrated in Figure 1, which shows the eGFR values of the patients of the study had mGFRs of 90, 60 and 30 mL/min. For the cases with an mGFR of 90 mL/min (n = 81), about half of them had eGFR values >90 mL/min and half had <90 mL/min, so ∼50% of them were classified as CKD-1 or CKD-2 and few in CKD-3 (Figure 1B). For those with an mGFR of 30 or 60 mL/min (n = 25 and 58, respectively), about two-thirds had eGFR values higher and one-third had eGFR values lower than the cut-off points. Thus ∼60% of the patients were classified as CKD-2 or 3 and ∼30% as CKD-3 or 4. In a few cases the misclassification was even larger (Figure 1B and C). We defined the CKD stages based on the classification of 2002, which does not divide Stage 3 into two subgroups (3A: 45–59 mL/min and 3B: 30–44 mL/min). However, a wide variability of the estimations was observed around the cut-off value of 45 mL/min, which may indicate that the misclassification within Stage 3 is similar to that observed for the other stages (Supplementary data, Figure S1). Finally, the variability observed around the cut-off points that divide the subgroups explains why 30–50% of patients were incorrectly classified by any formula. FIGURE 1 View largeDownload slide eGFR values of the patients with an mGFR of 90 (pannel A), 60 (B) and 30 mL/min (C) with a representative group of nine creatinine and/or cystatin C–based formulas. Based on repeatability of iohexol plasma clearance in our laboratory, which is about ±5%. We consider all cases included within a ±5% margin of mGFR 90, 60 and 30 mL/min as equal to these values. FIGURE 1 View largeDownload slide eGFR values of the patients with an mGFR of 90 (pannel A), 60 (B) and 30 mL/min (C) with a representative group of nine creatinine and/or cystatin C–based formulas. Based on repeatability of iohexol plasma clearance in our laboratory, which is about ±5%. We consider all cases included within a ±5% margin of mGFR 90, 60 and 30 mL/min as equal to these values. No clinically relevant differences were observed with cystatin C–based equations. In general, true positive cases were higher and false positive and missing cases lower with equations using cystatin C (Table 2 and Supplementary data, Table S2). This led to a lower error in the misclassification of CKD stages, which averaged 35% compared with the 50% error of creatinine-based equations. However, this still represents a wide margin to accept cystatin C as a good marker of GFR. This is in line with previous studies that questioned formulas based on cystatin C [9, 15, 16, 36, 37]. Finally, formulas that combine creatinine and cystatin C do not outperform equations that only use cystatin C. There is no clear explanation for this phenomenon. The rationale of combining both markers has not been clearly explained. From our point of view, creatinine, which is a poor marker of GFR, does not improve either the precision or accuracy of the estimation of renal function in these formulas. So the combined formulas do not represent an alternative for the estimation of renal function in clinical practice. It may be argued that calibrated creatinine and cystatin C may influence the performance of the formulas. However, it is interesting to compare two equations, Effersøe and CKD-EPI_cr, both with a similar TDI (Supplementary data, Table S4). The former was developed in 1957 with uncalibrated creatinine, whereas the latter was designed in 2011 with the calibrated method. Also, more recent formulas have shown a worse TDI than the old equations (MCQ versus Effersøe: Supplementary data, Table S4). In the same line, cystatin C–based formulas like Le Bricon (from 2000) and FAS_cy (from 2017) have a TDI of ∼70%. Although this is indirect evidence, since we do not measure creatinine or cystatin C with the non-calibrated methods, it seems plausible that the calibration does not have a relevant weight in the performance of formulas. The causes of the error of formulas are unclear. Formulas are algorithms mainly based on endogenous biomarkers such as serum creatinine and/or cystatin C. These molecules do not reflect real GFR properly. The synthesis of creatinine is not constant since it is affected by the daily intake of proteins and muscle turnover [38]. Changes in muscle mass also influence creatinine production [39–42]. Secretion and reabsorption by renal tubular cells and extrarenal clearance influence the serum levels of creatinine [43–49]. Serum cystatin C is influenced by obesity, diabetes, hypertension and metabolic syndrome. Thus high levels of this marker may reflect these conditions, and not necessarily renal dysfunction. Our study is in line with previous reports. Froissart et al. [7] evaluated the performance of the Cockcroft–Gault and MDRD formulas in 2095 subjects, showing that ∼30% of the patients were incorrectly classified in CKD stages. Other studies evaluating creatinine-based formulas observed similar results [6, 8, 10, 11, 19, 20, 36]. However, these publications evaluated few formulas and reported the classification in CKD stages only in terms of patients correctly classified or not. On the other hand, the error of cystatin C formulas has been seldom studied [8, 16]. Feng et al. [8] evaluated creatinine and cystatin C–based formulas, showing that ∼30–50% of the cases were incorrectly classified in CKD stages. Our group observed similar results evaluating 51 formulas in 193 renal transplanted patients. Thus, to the best of our knowledge, this is the first study that evaluated the performance of a large number of cystatin C and/or creatinine-based formulas in the classification of the stages of CKD in patients with different clinical situations over a wide range of GFRs. Our study has limitations. First is small number of patients with CKD Stage 5, which may limit the interpretation of our results in this subgroup of patients. However, other studies have shown that the error of eGFR is observed in advanced kidney failure (GFR < 15 mL/min) [19, 50]. Second, we studied mainly a Caucasian cohort and so we can not extrapolate our results to other populations. CONCLUSION We found that one or two patients out of three are incorrectly classified in higher or lower stages of CKD when renal function is estimated by formulas. This was observed for every stage of CKD and for all 65 evaluated formulas. Thus the correct classification of patients in CKD stages based on eGFR is a matter of chance. This is a strong limitation in evaluating the severity of renal disease, the risk for progression and the evolution of renal dysfunction over time. SUPPLEMENTARY DATA Supplementary data are available at ndt online. ACKNOWLEDGEMENTS We thank the Instituto de Tecnologías Biomédicas (ITB), the Instituto de Salud Carlos III (grants PI13/00342 and PI16/01814), REDINREN (RD16/0009/0031), the DISA Foundation, the Spanish Society of Nephrology (SENEFRO) and the IMBRAIN project for support (FP7-RE6-POT-2012-CT2012-31637-IMBRAIN) funded under the 7th Frameworks Programme capacities. FUNDING This study was supported by the Instituto de Salud Carlos III (PI13/00342, PI16/01814), REDINREN (RD16/0009/0031), DISA Foundation, the Spanish Society of Nephrology (SENEFRO) and the IMBRAIN project (FP7-RE6-POT-2012-CT2012-31637-IMBRAIN). S.L.L. is a research fellow supported by the Instituto de Salud Carlos III (grants for Río Hortega specialized health care post-training contracts), ISCIII CM15/00214. E.P. is a researcher in the Program Ramón y Cajal (RYC-2014-16573). CONFLICT OF INTEREST STATEMENT None declared. REFERENCES 1 Trillini M , Perico N , Remuzzi G. Epidemiology of end-stage renal failure: the burden of kidney diseases to global health. In: Orlando G , Remuzzi G , William DF (eds). Kidney Transplantation, Bioengineering, and Regeneration , vol. 1 . Cambridge, MA : Elseiver , 2017 : 5 – 11 Google Scholar CrossRef Search ADS 2 Stenvinkel P. Chronic kidney disease: a public health priority and harbinger of premature cardiovascular disease . J Intern Med 2010 ; 268 : 456 – 467 Google Scholar CrossRef Search ADS PubMed 3 Ketteler M , Block GA , Evenepoel P et al. Executive summary of the 2017 KDIGO chronic kidney disease-mineral and bone disorder (CKD-MBD) guideline update: what's changed and why it matters . Kidney Int 2017 ; 92 : 26 – 36 Google Scholar CrossRef Search ADS PubMed 4 National Kidney Foundation . K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification . Am J Kidney Dis 2002 ; 39(2 Suppl 1) : S1 – 266 5 Chapter 1: Definition and classification of CKD . Kidney Int Suppl (2011) 2013 ; 3 : 19 – 62 CrossRef Search ADS PubMed 6 Murata K , Baumann NA , Saenger AK et al. Relative performance of the MDRD and CKD-EPI equations for estimating glomerular filtration rate among patients with varied clinical presentations . Clin J Am Soc Nephrol 2011 ; 6 : 1963 – 1972 Google Scholar CrossRef Search ADS PubMed 7 Froissart M , Rossert J , Jacquot C et al. Predictive performance of the modification of diet in renal disease and Cockcroft-Gault equations for estimating renal function . J Am Soc Nephrol 2005 ; 16 : 763 – 773 Google Scholar CrossRef Search ADS PubMed 8 Feng JF , Qiu L , Zhang L et al. Multicenter study of creatinine- and/or cystatin C–based equations for estimation of glomerular filtration rates in Chinese patients with chronic kidney disease . PLoS One 2013 ; 8 : e57240 Google Scholar CrossRef Search ADS PubMed 9 Iliadis F , Didangelos T , Ntemka A et al. Glomerular filtration rate estimation in patients with type 2 diabetes: creatinine- or cystatin C-based equations? Diabetologia 2011 ; 54 : 2987 – 2994 Google Scholar CrossRef Search ADS PubMed 10 Rigalleau V , Lasseur C , Perlemoine C et al. A simplified Cockcroft-Gault formula to improve the prediction of the glomerular filtration rate in diabetic patients . Diabetes Metab 2006 ; 32 : 56 – 62 Google Scholar CrossRef Search ADS PubMed 11 MacIsaac RJ , Ekinci EI , Premaratne E et al. The Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equation does not improve the underestimation of glomerular filtration rate (GFR) in people with diabetes and preserved renal function . BMC Nephrol 2015 ; 16 : 198 Google Scholar CrossRef Search ADS PubMed 12 Gaspari F , Ruggenenti P , Porrini E et al. The GFR and GFR decline cannot be accurately estimated in type 2 diabetics . Kidney Int 2013 ; 84 : 164 – 173 Google Scholar CrossRef Search ADS PubMed 13 Rossing P , Rossing K , Gaede P et al. Monitoring kidney function in type 2 diabetic patients with incipient and overt diabetic nephropathy . Diabetes Care 2006 ; 29 : 1024 – 1030 Google Scholar CrossRef Search ADS PubMed 14 Inker LA , Schmid CH , Tighiouart H et al. Estimating glomerular filtration rate from serum creatinine and cystatin C . N Engl J Med 2012 ; 367 : 20 – 29 Google Scholar CrossRef Search ADS PubMed 15 Li HX , Xu GB , Wang XJ et al. Diagnostic accuracy of various glomerular filtration rates estimating equations in patients with chronic kidney disease and diabetes . Chin Med J (Engl) 2010 ; 123 : 745 – 751 Google Scholar PubMed 16 Luis-Lima S , Marrero-Miranda D , González-Rinne A et al. Estimated glomerular filtration rate in renal transplantation: the nephrologist in the mist . Transplantation 2015 ; 99 : 2625 – 2633 Google Scholar CrossRef Search ADS PubMed 17 Gaspari F , Ferrari S , Stucchi N et al. Performance of different prediction equations for estimating renal function in kidney transplantation . Am J Transplant 2004 ; 4 : 1826 – 1835 Google Scholar CrossRef Search ADS PubMed 18 Mariat C , Alamartine E , Barthelemy JC et al. Assessing renal graft function in clinical trials: can tests predicting glomerular filtration rate substitute for a reference method? Kidney Int 2004 ; 65 : 289 – 297 Google Scholar CrossRef Search ADS PubMed 19 Masson I , Flamant M , Maillard N et al. MDRD versus CKD-EPI equation to estimate glomerular filtration rate in kidney transplant recipients . Transplantation 2013 ; 95 : 1211 – 1217 Google Scholar CrossRef Search ADS PubMed 20 Mariat C , Alamartine E , Afiani A et al. Predicting glomerular filtration rate in kidney transplantation: are the K/DOQI guidelines applicable? Am J Transplant 2005 ; 5 : 2698 – 2703 Google Scholar CrossRef Search ADS PubMed 21 Orskov B , Borresen ML , Feldt-Rasmussen B et al. Estimating glomerular filtration rate using the new CKD-EPI equation and other equations in patients with autosomal dominant polycystic kidney disease . Am J Nephrol 2010 ; 31 : 53 – 57 Google Scholar CrossRef Search ADS PubMed 22 Ruggenenti P , Gaspari F , Cannata A et al. Measuring and estimating GFR and treatment effect in ADPKD patients: results and implications of a longitudinal cohort study . PLoS One 2012 ; 7 : e32533 Google Scholar CrossRef Search ADS PubMed 23 Luis-Lima S , Gaspari F , Negrín-Mena N et al. Iohexol plasma clearance simplified by dried blood spot testing . Nephrol Dial Transplant 2017 . doi: 10.1093/ndt/gfx323 24 Krutzén E , Bäck SE , Nilsson-Ehle I et al. Plasma clearance of a new contrast agent, iohexol: a method for the assessment of glomerular filtration rate . J Lab Clin Med 1984 ; 104 : 955 – 961 Google Scholar PubMed 25 Niculescu-Duvaz I , D'Mello L , Maan Z et al. Development of an outpatient finger-prick glomerular filtration rate procedure suitable for epidemiological studies . Kidney Int 2006 ; 69 : 1272 – 1275 Google Scholar CrossRef Search ADS PubMed 26 Bröchner-Mortensen J. A simple method for the determination of glomerular filtration rate . Scand J Clin Lab Invest 1972 ; 30 : 271 – 274 Google Scholar CrossRef Search ADS PubMed 27 Du Bois D , Du Bois EF. A formula to estimate the approximate surface area if height and weight be known. 1916 . Nutrition 1989 ; 5 : 303 – 311 Google Scholar PubMed 28 Lin LI. A concordance correlation coefficient to evaluate reproducibility . Biometrics 1989 ; 45 : 255 – 268 Google Scholar CrossRef Search ADS PubMed 29 Lin L , Hedayat AS , Wu W. A comparative model for continuous and categorical data. In: Lin L , Hedayat AS , Wu W (eds). Statistical Tools for Measuring Agreement . New York : Springer , 2012 : 111 Google Scholar CrossRef Search ADS 30 Lin L , Hedayat AS , Sinha B et al. Statistical methods in assessing agreement . J Am Stat Assoc 2002 ; 97 : 257 Google Scholar CrossRef Search ADS 31 Altman DG. Practical Statistics for Medical Research . London : Chapman and Hall , 1991 32 Craig AJ , Samol J , Heenan SD et al. Overestimation of carboplatin doses is avoided by radionuclide GFR measurement . Br J Cancer 2012 ; 107 : 1310 – 1316 Google Scholar CrossRef Search ADS PubMed 33 Pöge U , Gerhardt T , Stoffel-Wagner B et al. Calculation of glomerular filtration rate based on cystatin C in cirrhotic patients . Nephrol Dial Trasplant 2006 ; 21 : 660 – 664 Google Scholar CrossRef Search ADS 34 Francoz C , Prié D , Abdelrazek W et al. Innaccuracies of creatinine and creatinine-based equations in candidates for liver transplantation with low creatinine: impact on the model for end-stage liver disease score . Liver Transpl 2010 ; 16 : 1169 – 1177 Google Scholar CrossRef Search ADS PubMed 35 Ahlström MG , Kjær A , Gerstoft J et al. Agreement between estimated and measured renal function in an everyday clinical outpatient setting of human immunodeficiency virus-infected individuals . Nephron 2017 ; 136 : 318 – 327 Google Scholar CrossRef Search ADS PubMed 36 White C , Akbari A , Hussain N et al. Chronic kidney disease stage in renal transplantation classification using cystatin C and creatinine-based equations . Nephrol Dial Transplant 2007 ; 22 : 3013 – 3020 Google Scholar CrossRef Search ADS PubMed 37 Zahran A , Qureshi M , Shoker A. Comparison between creatinine and cystatin C-based GFR equations in renal transplantation . Nephrol Dial Transplant 2007 ; 22 : 2659 – 2668 Google Scholar CrossRef Search ADS PubMed 38 Crim MC , Calloway DH , Margen S. Creatine metabolism in men: urinary creatine and creatinine excretions with creatine feeding . J Nutr 1975 ; 105 : 428 – 438 Google Scholar CrossRef Search ADS PubMed 39 Heymsfield SB , Arteaga C , McManus C et al. Measurement of muscle mass in humans: validity of the 24-hour urinary creatinine method . Am J Clin Nutr 1983 ; 37 : 478 – 494 Google Scholar CrossRef Search ADS PubMed 40 Bleiler RE , Schedl HP. Creatinine excretion: variability and relationships to diet and body size . J Lab Clin Med 1962 ; 59 : 945 – 955 Google Scholar PubMed 41 Irving RA , Noakes TD , Irving GA et al. The immediate and delayed effects of marathon running on renal function . J Urol 1986 ; 136 : 1176 – 1180 Google Scholar CrossRef Search ADS PubMed 42 Rennie MJ , Edwards RH , Krywawych S et al. Effect of exercise on protein turnover in man . Clin Sci 1981 ; 61 : 627 – 639 Google Scholar CrossRef Search ADS PubMed 43 Perrone RD , Madias NE , Levey AS. Serum creatinine as an index of renal function: new insights into old concepts . Clin Chem 1992 ; 38 : 1933 – 1953 Google Scholar PubMed 44 Shemesh O , Golbetz H , Kriss JP et al. Limitations of creatinine as a filtration marker in glomerulopathic patients . Kidney Int 1985 ; 28 : 830 – 838 Google Scholar CrossRef Search ADS PubMed 45 Miller BF , Leaf A , Mamby AR et al. Validity of the endogenous creatinine clearance as a measure of glomerular filtration rate in the diseased human kidney . J Clin Invest 1952 ; 31 : 309 – 313 Google Scholar CrossRef Search ADS PubMed 46 Baldwin DS , Sirota JH , Villarreal H. Diurnal variations of renal function in congestive heart failure . Proc Soc Exp Biol Med 1950 ; 74 : 578 – 581 Google Scholar CrossRef Search ADS PubMed 47 Chesley LC. Renal excretion at low urine volumes and the mechanism of oliguria . J Clin Invest 1938 ; 17 : 591 – 597 Google Scholar CrossRef Search ADS PubMed 48 Jones JD , Burnett P. Creatinine metabolism in humans with decreased renal function: creatinine deficit . Clin Chem 1974 ; 20 : 1204 – 1212 Google Scholar PubMed 49 Mitch WE , Collier VU , Walser M. Creatinine metabolism in chronic renal failure . Clin Sci 1980 ; 58 : 327 – 335 Google Scholar CrossRef Search ADS PubMed 50 Tidman M , Sjöström P , Jones I. A comparison of GFR estimating formulae based upon s-cystatin C and s-creatinine and a combination of two . Nephrol Dial Transplant 2007 ; 23 : 154 – 160 Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Nephrology Dialysis Transplantation – Oxford University Press

**Published: ** May 11, 2018

Loading...

personal research library

It’s your single place to instantly

**discover** and **read** the research

that matters to you.

Enjoy **affordable access** to

over 18 million articles from more than

**15,000 peer-reviewed journals**.

All for just $49/month

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Read from thousands of the leading scholarly journals from *SpringerNature*, *Elsevier*, *Wiley-Blackwell*, *Oxford University Press* and more.

All the latest content is available, no embargo periods.

## “Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”

Daniel C.

## “Whoa! It’s like Spotify but for academic articles.”

@Phil_Robichaud

## “I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”

@deepthiw

## “My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”

@JoseServera

DeepDyve ## Freelancer | DeepDyve ## Pro | |
---|---|---|

Price | FREE | $49/month |

Save searches from | ||

Create lists to | ||

Export lists, citations | ||

Read DeepDyve articles | Abstract access only | Unlimited access to over |

20 pages / month | ||

PDF Discount | 20% off | |

Read and print from thousands of top scholarly journals.

System error. Please try again!

or

By signing up, you agree to DeepDyve’s Terms of Service and Privacy Policy.

Already have an account? Log in

Bookmark this article. You can see your Bookmarks on your DeepDyve Library.

To save an article, **log in** first, or **sign up** for a DeepDyve account if you don’t already have one.

All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.

ok to continue