Article
Accuracy of GFR Estimation in Obese Patients Sandrine Lemoine,*†‡§ Fitsum Guebre-Egziabher,†‡§ Florence Sens,†§ Marie-Sophie Nguyen-Tu,| Laurent Juillard,†‡§ Laurence Dubourg,*§| and Aoumeur Hadj-Aissa*§
Abstract Background and objectives Adequate estimation of renal function in obese patients is essential for the classification of patients in CKD category as well as the dose adjustment of drugs. However, the body size descriptor for GFR indexation is still debatable, and formulas are not validated in patients with extreme variations of weight. Design, setting, participants, & measurements This study included 209 stages 1–5 CKD obese patients referred to the Department of Renal Function Study at the University Hospital in Lyon between 2010 and 2013 because of suspected renal dysfunction. GFR was estimated with the Chronic Kidney Disease and Epidemiology equation (CKD-EPI) and measured with a gold standard method (inulin or iohexol) not indexed (mGFR) or indexed to body surface area determined by the Dubois and Dubois formula with either real (mGFRr) or ideal (mGFRi) body weight. Mean bias (eGFR2mGFR), precision, and accuracy of mGFR were compared with the results obtained for nonobese participants (body mass index between 18.5 and 24.9) who had a GFR measurement during the same period of time. Results Mean mGFRr (51.6624.2 ml/min per 1.73 m2) was significantly lower than mGFR, mGFRi, and eGFRCKD-EPI. eGFRCKD-EPI had less bias with mGFR (0.29; 21.7 to 2.3) and mGFRi (21.62; 23.1 to 0.45) compared with mGFRr (8.7; 7 to 10). This result was confirmed with better accuracy for the whole cohort (78% for mGFR, 84% for mGFRi, and 72% for mGFRr) and participants with CKD stages 3–5. Moreover, the Bland Altman plot showed better agreement between mGFR and eGFRCKD-EPI. The bias between eGFRCKD-EPI and mGFRr was greater in obese than nonobese participants (8.7 versus 0.58, P,0.001). Conclusions This study shows that, in obese CKD patients, the performance of eGFRCKD-EPI is good for GFR#60 ml/min per 1.73 m2. Indexation of mGFR with body surface area using ideal body weight gives less bias than mGFR scaled with body surface area using real body weight. Clin J Am Soc Nephrol 9: 720–727, 2014. doi: 10.2215/CJN.03610413
Introduction The increasing prevalence of obesity, especially in the Western world, is associated with the risk to develop kidney disease and the progression of renal disease (1,2). Emerging studies in this last decade indicate that, aside from being a major cause for the development of diabetes and hypertension, obesity may have direct adverse effects on kidney function independent of the other known risk factors (1,3,4). Hormonal factors, oxidative stress, inflammation, and endothelial dysfunction could explain the link between obesity and CKD (5). However, there is still a lack of knowledge about the prevalence and risk factors of obesityrelated kidney disease. Correctly estimating renal function in obese patients is, thus, essential for not only the classification of patients in CKD category but also, better dose adjustment of drugs eliminated by the kidneys in these patients. The most widely used measure of kidney function is eGFR using creatinine-based formulas. There is no formula specially validated for obesity, because eGFR is derived and validated in nonobese 720
Copyright © 2014 by the American Society of Nephrology
individuals with CKD. The Cockroft and Gault formula overestimates GFR in obese patients, because the formula includes body weight. The Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease and Epidemiology (CKD-EPI) equations do not include total weight but have been reported to also overestimate GFR in obese patients (6). Recently, recommendations proposed CKD-EPI as the most reliable method to determine eGFR (7). The accuracy and precision of this equation might still be affected by factors, such as age, muscle mass, diet, and proximal tubule secretion of creatinine. Thus, studies have concluded that no estimation is validated for patients over 75 years, malnourished patients, patients with extreme variations of weight or muscle mass, and patients with a low protein diet (3,4,7). However, MDRD and CKD-EPI equations are more recommended than the Cockroft and Gault formula, because they were developed in North American population, where overweight or obese patients are more prevalent. Inulin clearance is the gold standard measure of GFR. Because GFR varies with weight and height, it is
Departments of *Renal Function Study and †Nephrology, Edouard Herriot Hospital, Hospices Civils de Lyon, Lyon, France; ‡Institut National de la Sante´ et de la Recherche Me´dicale U1060, Laboratoire de Recherche en Cardiovasculaire, Me´tabolisme, Diabe´tologie et Nutrition, Lyon, France; §University of Lyon, Universite´ Lyon 1 Claude Bernard, Lyon, France; and | Unite´ Mixte de Recherche 5305 Centre National de la Recherche Scientifique, Universite´ Claude Bernard, Lyon, France Correspondence: Dr. Sandrine Lemoine, Department of Nephrology, Edouard Herriot Hospital, 5 place d’Arsonval, 69003 Lyon, France. Email: sandrine. [email protected]
www.cjasn.org Vol 9 April, 2014
Clin J Am Soc Nephrol 9: 720–727, April, 2014
evident that GFR comparisons must include some adjustment for body size. However, in obese patients, there is still the problem of the indexation by body surface area (BSA). BSA is disproportionally affected by fat mass in obesity, and the body size descriptor that should be used for this indexation in the setting of obesity is still questionable. To address these fundamental issues, in this study, we compared eGFRCKD-EPI with the gold standard reference method inulin or iohexol clearances nonindexed by BSA or indexed with BSA calculated with either actual body weight or ideal body weight. We evaluated the impact of these alternative body size descriptors on the precision, accuracy, and bias of GFR among obese adults. Because measured GFR (mGFR) is used for drug-dosing purposes and mGFR adjusted to BSA is used for the classification of CKD stages, we compared eGFRCKD-EPI with mGFR, GFR indexed with BSA measured with the Dubois and Dubois formula and ideal weight (mGFRi), and GFR indexed with BSA measured with the Dubois and Dubois formula and real weight (mGFRr).
Materials and Methods Data Collection We retrospectively analyzed the GFR results of 209 obese participants referred in our unit for various nephropathies between February of 2010 and January of 2013 for measurement of inulin clearance because of suspected or established renal dysfunction. We compared these results with the results of a group of nonobese participants who had GFR evaluation during the same period of time. Informed consent was obtained from all patients before measurement of the renal clearance. The consent form contained information on the procedure itself as well as the later use of the information for research purposes. According to the French legislation, concerning the use of a database without direct identification of patients, it was not necessary to obtain an ethical approval (Law 2006–450, April 19, 2004; Commission nationale de l’informatique et des libertés). Creatinine measurement was performed with an enzymatic method with calibration certified by isotope dilution mass spectrometry. Plasma creatinine (PCr) was obtained with the Siemens enzymatic method (on the Dimension Vista System) traceable to National Institute of Standards and Technology creatinine Standard Reference Materials 914 (verified with National Institute of Standards and Technology SRM 967). The GFR measurement was performed with the gold standard method (mGFR): inulin or iohexol clearance. Inulin clearance (INUTEST 25%; Fresenius, Kabi, Austria) was performed with a loading dose of 30 ml/kg that was injected in 10 minutes, with a maintenance dose infusion of a solution of inulin of 40 mg/kg. The urine was collected every 30 minutes, and we performed blood tests in the middle of each period of urine collection (three to four collection periods of 30 minutes). The inulin clearance was calculated in each period to obtain the average. Measurements of plasma and urine polyfructosan concentrations were performed using an enzymatic method (8). For the other remaining 25 patients, we performed iohexol clearance. We injected 8 ml iohexol (300 mg; Omnipaque; GE Healthcare SAS, Vélizy-Villacoublay, France) and weighed
Obesity and GFR, Lemoine et al.
721
the syringe before and after injection. Blood collection was performed at 120, 180, and 240 minutes. The serum iohexol concentration was measured by HPLC (9). The GFR was calculated as GFR=slope3dose/concentration at time 0 corrected with the Bröchner–Mortensen equation (10). GFR was also estimated by CKD-EPI: CKD-EPI ¼ k1 3 ½PCr=k2Þ2k3 3 0:993age ; with k1=141, 143, 163, and 166 for white men and women and black men and women, respectively; k2=0.7 and 0.9 for women and men, respectively; and k3=1.209, 1.209, 0.411, and 0.329 for men with PCr.0.9 mg/dl, women with PCr.0.7 mg/dl, men with PCr#0.9 mg/dl, and women with PCr#0.7 mg/dl, respectively.
Body Size Descriptors The ideal weight was determined by the Lorentz equation (11) ([height (centimeters)2100]2[(height [centimeters] 2150)/4]). The real BSA (BSAr) and ideal BSA were determined by the Dubois and Dubois BSA formula, because it is the most widely and commonly used formula (0.0071843total body weight [kilograms]0.4253height [centimeters]0.725) (12). To assess the BSA that best converts the predicted GFRs, several BSA formulas were used for unadjusted CKD-EPI: formulas by Haycock et al. (13), Gehan and George (14), and Livingston and Lee (15) were used. Obesity was defined using the World Health Organization definition: body mass index (BMI)$30 kg/m2. Reference BMI for nonobese participants was defined by a BMI between 18.5 and 24.9 kg/m2.
Statistical Analyses Statistical analyses were performed with Prism software package (version 6; GraphPad). Mean BSA and mean GFR were compared with a t paired test. The performance of equations was measured by mean bias (eGFRCKD-EPI2mGFR) and mean relative bias ([eGFRCKD-EPI2mGFR]/mGFR) expressed in percentages as well as precision and accuracy. The precision was assessed as an interquartile range for the differences. Accuracy was calculated as the percentage of GFR estimates within 30% deviation of the mGFR. Comparisons of bias and accuracy were performed using t and McNemar’s tests, respectively. The agreement between mGFR and eGFRCKD-EPI was calculated with eGFRCKD-EPI2mGFR on the y axis and eGFRCKD-EPI on the x axis. We analyzed the performance of eGFRCKD-EPI in subgroups of BMI ranges (BMI=30–35, 35–40, and .40 kg/m2). P values,0.05 were considered significant.
Results Characteristics of the Participants In total, 209 obese participants (90 women; 43%) with a mean BMI of 34.864.6 kg/m2 were included (Table 1). In total, 188 nonobese participants with a mean BMI of 22.861.9 kg/m2 were included for comparison. Results of BSA with the Dubois and Dubois formula using real or ideal weight show, as expected, that BSAr is significantly higher than ideal BSA. Results of BSA calculated
722
Clinical Journal of the American Society of Nephrology
Table 1. Patients’ characteristics
Characteristic Age (yr) Weight (kg) Height (cm) Men, n (%) Kidney donor, n (%) Transplant recipient, n (%) Diabetes, n (%) Body mass index (kg/m2), n (%) 30 to ,35 35 to ,40 40 to ,50 .50 Ideal weight (kg) Body surface area (real weight; m2) Body surface area (ideal weight; m2) Creatinine (mg/dl) mGFR (ml/min), n (%) ,15 15–29 30–59 60–89 $90 mGFRr (ml/min per 1.73 m2) mGFRi (ml/min per 1.73 m2) eGFRMDRD (ml/min per 1.73 m2) eGFRCKD-EPI (ml/min per 1.73 m2)
Obese (n=209) 57613.2 (19.4–87) 94.9616 (62.5–195) 16569.8 (163.8–166.4) 119 (57) 3 (1.4) 29 (14) 64 (31) 34.864.6 (30–67.1) 134 (64) 54 (26) 19 (9) 2 (1) 61.367.3 (41–82.6) 260.2 (1.5–2.8) 1.6760.1 (1.2–2.1)b 1.4260.52 (0.4–6.3) 60.2628 (12–146)c 11 (5.2) 20 (9.5) 88 (42.2) 55 (26.3) 35 (16.7) 51.8624 (9–121) 62628 (11–148)c 57.7627 (7–164)c 60.6628 (7–124)c
Nonobese (n=188) 55.5614.5 (19–88) 65.569.5 (47–90)a 16969 (144–193) 110 (59) 4 (3)a 65 (31)a 15 (7)a 22.861.9 (18.5–25)a — — — — — 1.760.17 (1.4–2.2)a — 1.4460.62 (0.7–3.8) 54.8622 (11–105) 5 (2.6) 14 (7.4) 78 (46) 67 (29) 24 (18) 53.8622 (11–114) — 54.6621 (13–120) 53.3622 (18–110)a
Values are expressed as mean6SD (range) or n (%). mGFR, measured GFR nonadjusted with body surface area; mGFRr, GFR indexed with body surface area measured with the Dubois and Dubois equation and real weight; mGFRi, GFR indexed with body surface area measured with the Dubois and Dubois equation and ideal weight; MDRD, Modification of Diet in Renal Disease defined by 1753plasma creatinine (mg/dl)21.1543age (years)20.20330.742 (if woman)31.21 (if black); CKD-EPI, Chronic Kidney Disease and Epidemiology equation. a P,0.05 versus obese population. b P,0.05 real weight body surface area versus ideal weight body surface area. c P,0.05 versus mGFRr.
with other formulas are described in Supplemental Table 1. GFR Measurements Impact of Body Size Descriptors on GFR Measured by Reference Techniques. The mGFRr is significantly lower than mGFR and mGFRi (28.3 and 210.2 ml/min per 1.73 m2, respectively; P,0.001) (Table 1). There is no statistically significant difference between eGFR CKD-EPI (60.6628; 7–124) and mGFR (60.2628; 12–146; P=0.78) or mGFRi (62628; 11–148; P=0.07). The linear regression of eGFR and mGFRi is the closest to the line of identity (Figure 1) compared with mGFRr. eGFRCKD-EPI was well correlated with mGFRr and mGFRi (r=0.89 between eGFRCKD-EPI and mGFRr and r=0.9 between eGFR CKD-EPI and mGFRi; P,0.001). Comparison between eGFRCKD-EPI and mGFR among Obese Participants. The eGFRCKD-EPI (60.6 ml/min per 1.73 m2) has better agreement with mGFR (60.2 ml/min) and mGFRi (61.9 ml/min per 1.73 m2) than mGFRr (51.8 ml/min per 1.73 m2) (Table 1). We, furthermore, analyzed the relative bias, precision, and accuracy of eGFR compared with mGFR, mGFRi, and mGFRr. eGFR CKD-EPI
had significantly more bias with mGFRr than mGFR or mGFRi (P,0.001) (Table 2). The biases between eGFR CKD-EPI and mGFR are not significantly different from the biases between eGFR C K D - E P I and mGFRi (P=0.23). Moreover, accuracy is better between eGFRCKD-EPI and mGFRi than mGFRr (84% versus 72% for CKD-EPI, respectively) and between eGFR CKD-EPI and mGFR than mGFRr (78% versus 72%, respectively) (Table 2). Furthermore, precision between eGFRCKD-EPI and mGFRi is better than mGFRr. Figure 2 confirms the good agreement between eGFR and mGFRi, with the lowest bias between eGFRCKD-EPI and mGFRi. The bias between eGFR and mGFRr is less in nonobese than obese patients, excluding a methodological problem. Compared with the MDRD equation, the CKD-EPI equation showed the lowest bias (mean=21.62 versus 24.67 ml/min per 1.73 m2; P=0.04 for mGFRi; mean=0.29 versus 2.5 ml/min per 1.73 m2 for mGFR; data not shown). Because eGFRCKD-EPI was the least biased equation, additional analyses were performed using this equation. Comparison between CKD-EPI mGFR, mGFRi, and mGFRr and BMI. The Bland Altman plot shows that the biases are important when BMI is up to 40 kg/m2 for
Clin J Am Soc Nephrol 9: 720–727, April, 2014
Obesity and GFR, Lemoine et al.
723
Figure 1. | No statistically significant difference between mGFRi, mGFRr, and eGFR CKD-EPI. Log–log scale regression between mGFRi, mGFRr, and eGFRCKD-EPI. CKD-EPI, Chronic Kidney Disease and Epidemiology equation; mGFRi, GFR indexed with body surface area measured with the Dubois and Dubois equation and ideal weight (milliliters per minute per 1.73 m2); mGFRr, GFR indexed with body surface area measured with the Dubois and Dubois equation and real weight (milliliters per minute per 1.73 m2). The black line is the identity line. The long dashed line is the regression line between eGFR and mGFRr, and the small dashed line is the regression line between eGFR and mGFRi.
Table 2. Bias, precision, and accuracy between eGFRCKD-EPI and mGFR, mGFRr, or mGFRi
CKD-EPI Variables Obese mGFR Bias (ml/min), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min) Accuracy, % (95% confidence interval) mGFRr Bias (ml/min per 1.73 m2), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min per 1.73 m2) Accuracy, % (95% confidence interval) mGFRi Bias (ml/min per 1.73 m2), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min per 1.73 m2) Accuracy, % (95% confidence interval)
0.3a (21.7 to 2.3) 21a (18 to 23) 14 78a (72 to 84) 8.7a,c (7 to 10) 25a,c (21 to 28) 11 72a (66 to 78) 21.6c (23.1 to 0.4) 19c (16 to 21) 11 84c (79 to 89)
Nonobese 1 (20.7 to 2.6) 14 (12 to 15) 8.5 87b (85 to 89) 0.6b (20.6 to 1.8) 16b (15 to 19) 5.7 92b (90 to 93) — — — —
Mean bias is expressed as the mean of eGFR2mGFR. Mean relative bias is expressed as the mean of absolute values of (eGFR2mGFR)/ mGFR. Precision refers to the 25th percentile275th percentile. Accuracy refers to the percentage of GFR estimates that is within 30% of measured mGFR. a P,0.01 versus mGFRi. b P,0.01 versus obese population. c P,0.01 versus mGFR.
724
Clinical Journal of the American Society of Nephrology
Figure 2. | Good agreement and lowest bias between eGFR and mGFRi. The graphical plot of the difference between (eGFRCKD-EPI–mGFR, mGFRr or mGFRi) and eGFR. Mean bias is 0.369.7 ml/min per 1.73 m2 for CKD-EPI versus mGFR, 8.7614.4 ml/min per 1.73 m2 for CKD-EPI versus mGFRr, 21.62613.3 ml/min per 1.73 m2 for CKD-EPI versus mGFRi. The dotted lines represent the limits of agreement (mean difference62 SD) and the dashed line represents the mean difference. mGFRi, GFR indexed with BSA measured with Dubois and Dubois and ideal weight; mGFRr, GFR indexed with BSA measured with Dubois and Dubois and real weight.
Figure 3. | Biases are more important for body mass index (BMI)>40 kg/m2. The graphical plot of the difference between (eGFRCKD-EPI–mGFR, mGFRr or mGFRi) and eGFR. Mean bias is 0.369.7 ml/min per 1.73 m2 for CKD-EPI versus mGFR, 8.7614.4 ml/min per.73 m2 for CKD-EPI versus mGFRr, 21.62613.3 ml/min per 1.73 m2 for CKD-EPI versus mGFRi. The dotted lines represent the limits of agreement (mean difference62 SD) and the dashed line represents the mean difference. mGFRi, GFR indexed with BSA measured with Dubois and Dubois and ideal weight; mGFRr, GFR indexed with BSA measured with Dubois and Dubois and real weight.
mGFR adjusted with BSA (Figure 3). The performances of eGFRCKD-EPI for subgroups of BMI categories are shown in Table 3. For BMI greater than 40 kg/m2, eGFRCKD-EPI shows more bias with mGFRi and mGFR (27.47 and 24.8 ml/min per 1.73 m2, respectively), and accuracies for mGFRr, mGFRi, and mGFR are dramatically reduced (68%, 77%, and 66%, respectively; P,0.001). For BMI range=35–40 kg/m2, precision is better with eGFRCKD-EPI, and accuracies are reduced with mGFRr. Performance of eGFRCKD-EPI According to the mGFR Level. With eGFRCKD-EPI, patients with CKD stages 3 or 4 and 5 are correctly classified if we compare with mGFR or mGFRi (71% or 82%, respectively, are well classified in CKD stage 3; 76% or 81%, respectively, are well classified in CKD stages 4 and 5) (Table 4). Comparison between mGFR and eGFRCKD-EPI Unadjusted to BSA. Several BSA formulas were used to unadjust CKD-EPI to assess the best body size descriptor that converts to the predicted GFR (Table 5). eGFRCKD-EPI adjusted or not to BSA calculated with the Livingston and Lee (15) formula is not substantially different from mGFR (60.6628 and 59.1627.4 ml/min, respectively; P.0.20).
Discussion This study shows that, in obese participants, GFR measured with the gold standard technique (inulin or iohexol clearance) underestimates GFR when adjusted to BSA using the real body weight, whereas there was no
significant difference between mGFR and mGFRr in nonobese participants. eGFR with the CKD-EPI equation is not different from the mGFR or mGFRi, showing that the CKDEPI equation offers a good GFR prediction in obese participants, especially stage 3 CKD patients. There is an emerging recognition of obesity-related kidney disease, and it requires the reassessment of traditional GFR measurements and estimation methods developed in nonobese patients. Most studies relate that formulas are not validated in obese population. Although these equations were established in populations that are not obese, average BMI for the CKD-EPI formula was 28 kg/m2 (16–19). This population was overweight but not obese. Moreover, these formulas have been compared with a reference method adjusted with BSAr. Inulin clearance remains the gold standard, but it is the indexation with the real weight that produces an underestimation of GFR, which is clearly shown in our study. Indeed, if we compare our results with the results found in the population used to develop the CKD-EPI equation, we found the same accuracy for the percentage of GFR estimates that are within 30% of mGFR for the CKD-EPI equation (16). Indeed, muscle mass does not increase proportionally with total body weight but only lean body weight. There is no consensus on the body size descriptor that should be used in the estimation of GFR. Although this indexation has a poor influence in the normal weight population, it can induce important deviations in obese or
Clin J Am Soc Nephrol 9: 720–727, April, 2014
Obesity and GFR, Lemoine et al.
725
Table 3. Performance of CKD-EPI for body mass index category
Bias Mean (95% Confidence Interval)
BMI Categories Body mass index=30–35 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi Body mass index=35–40 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi Body mass index>40 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi
22.3 (0 to 4.8) 10.3a,b (0.2 to 12) 1.26 (20.8 to 3.4)
Accuracy (30%), % (95% Confidence Interval)
Precision
12 15 10
88 (83.6 to 92.4) 82 (76.8 to 87.2) 88 (83.6 to 92.4)
23 (27.1 to 1.2) 5.6b (2.3 to 8.9) 25.5 (29.2 to 21.8)
5.5 11.5 12
72 (67.1 to 76.9) 46 (39.2 to 52.8) 81 (75.7 to 86.3)
24.8 (21.7 to 2) 6.5a,b (0.14 to 12) 27.5b (214 to 20.8)
8.5 14.5 11.5
66 (59.6 to 72.4) 68 (61.7 to 74.3) 77 (71.3 to 82.7)
Relative biases are expressed as the mean of eGFR2mGFR (milliliters per minute per 1.73 m2). Precision refers to the 25th percentile2 75th percentile. Accuracy refers to the percentage of GFR estimates that is within 30% of measured mGFR. BMI, body mass index. a P,0.01 versus mGFRi. b P,0.01 versus mGFR.
Table 4. Number and percentage of subjects correctly classified for CKD stage
Variables
Stage 1 (GFR$90 ml/min per 1.73 m2)
Stage 2 (GFR=60–89 ml/min per 1.73 m2)
Stage 3 (GFR=30–59 ml/min per 1.73 m2)
Stages 4 and 5 (GFR#30 ml/min per 1.73 m2)
mGFR CKD-EPI mGFRr CKD-EPI mGFRi CKD-EPI
35 26 (74%) 21 20 (95%) 38 30 (79%)
54 33 (61%) 49 26 (53%) 65 36 (55%)
90 64 (71%) 98 67 (68%) 79 65 (82%)
30 23 (76%) 41 28 (68%) 27 22 (81%)
Reference method is mGFR, mGFRr, or mGFRi. Values for CKD-EPI are expressed as number of patients (%).
Table 5. Comparison between mGFR and eGFRCKD-EPI or eGFRCKD-EPI unadjusted with body surface area measured using different formulas and ideal or real weight for the obese cohort
With Ideal Weight Variables mGFR (ml/min) CKD-EPI CKD-EPI3DD/1.73 CKD-EPI3Ha/1.73 CKD-EPI3GG/1.73 CKD-EPI3LL/1.73
With Real Weight
Mean6SD
P Value
Mean6SD
P Value
60.2628 60.6628 58.6627 58.6627 58.8628 59.1627
0.12 ,0.001 ,0.001 ,0.001 0.20
60.2628 60.6628 70.2633 73.8634 73.5634 77.3636
0.12 ,0.001 ,0.001 ,0.001 ,0.001
All values are compared with mGFR. P values versus with mGFR. DD, Dubois and Dubois; Ha, Haycock et al. (13); GG, Gehan and George (14); Livingston and Lee (15).
anorectic patients (6,20). BSA was seen as an alternative for reflecting the metabolic rate. Indexing with BSA is essential to make comparisons between subjects, have reference values, and be classified in a status of chronic renal disease (21). Indexation is necessary in patients with normal BMI
with extreme height. However, there is no strong linear relation between GFR and BSA. Scaling GFR to body size may be misleading, and it is controversial in obese subjects. Delanaye et al. (20,22) recommended the use of absolute GFR values instead of the GFRs indexed to BSA,
726
Clinical Journal of the American Society of Nephrology
especially in abnormal body size populations (20,22). However, because indexation is still a matter of debate, we evaluated the accuracy of BSA indexation of measured GFR by using different alternative body size descriptors. Moreover, the indexation with standard BSA (1.73 m2) is also debatable (23,24). BSA of 1.73 m2 is no longer representative of a normal average morphology. This indexing is based on an old principle developed in 1883 by Rubner and colleagues, and it is not totally true (21). Actually, the average estimated BSA is 2.06 m2 for men and 1.83 m2 for women in the United States (25). In clinical practice, if we consider drug dosing, mGFR is used as a reference. Indeed, drug clearance is not proportional to total body weight, and estimation of drug-dosing intervals has been based on the clearance of creatinine. We showed that the CKD-EPI equation has less bias with mGFR, with good accuracy. Although the CKD-EPI equation is an equation adjusted to BSA, we found no difference between eGFRCKD-EPI and mGFR. Because there is no difference between eGFRCKD-EPI and mGFR and given that unadjusted eGFRCKD-EPI results in less precision, it is, therefore, unnecessary to unadjust eGFRCKD-EPI by BSA. If GFR needs to be indexed for BSA, especially for the purpose of CKD stage classification, there is still controversy on how to calculate the BSA in obese subjects. Indeed, equations used to estimate BSA have not been validated in obese individuals and are not stratified by sex (26). Measuring BSA in obese subjects still remains a challenging and unsolved problem, and choosing the best body size descriptor to scale kidney function is difficult (24). Although the Dubois and Dubois formula has already been criticized, we found that BSA and mGFR with another BSA equation (the formula by Livingston and Lee [15]) are not significantly different (Supplemental Table 1) from the results derived from the Dubois and Dubois formula. Our finding is similar to the results already published by Verbraecken et al. (27). Alternative body size descriptors of total body weight could be proposed. Ideal body weight, adjusted body weight, and fat-free weight stratified by sex have been used as estimates of lean body mass for the calculation of eGFR in obese patients (26). In our study, there is good agreement between eGFRCKD-EPI and mGFRi (Figure 1), and mGFRi has less bias than mGFRr, specifically for GFR values above 20 ml/min per 1.73 m2. Lean body weight can be estimated with dual energy x-ray absorptiometry or bioelectric impedance analysis. Janmahasatian et al. (28) have reported that GFR values adjusted with total body weight underestimate GFR in obese patients, whereas GFR scaled by lean body weight was not different between obese and nonobese participants, suggesting that renal function is closely related to lean body mass and not influenced by adipose tissue. Janmahasatian et al. (28) concluded that lean body weight should be used instead of total body weight for estimating creatinine clearance. Another crucial question that needs to be addressed is the threshold of BMI for which eGFRCKD-EPI can be safely used. In our study, eGFRCKD-EPI for obese people with BMI greater than 40 kg/m2 seemed to be less accurate, but because of the limited number of participants in this range of BMI, we cannot make strong conclusions. Hence, over this BMI measurement, we recommend using a gold standard
method adjusted to ideal body weight to classify CKD and mGFR for drug-dosing purposes. Finally, the hallmark of obesity is excess adipose tissue mass or fat mass for a given body weight. The increased body mass in obesity is caused by both increased fat and fat-free mass. The relative contributions of these two components to the excess weight are influenced, in part, by age and an individual’s proportion of fat and fat-free mass before weight gain, assuming adequate protein intake occurs during the weight gain process. However, older individuals and subjects with chronic disease can become obese without a parallel growth of muscle mass. In this case, body weight may not be accurately estimated by the ideal weight. This work shows that the CKD-EPI equation is validated in the obese population up to a BMI range of 40 kg/m2, specifically for GFR levels ,60 ml/min; this level is important, because it is when patients often face CKD-related complications or need dose adjustment for drug. Indexation of mGFR with BSA using ideal body weight gives less bias than mGFR scaled with BSA using real body weight for GFR.20 ml/min per 1.73 m2. Acknowledgments We thank Mélodie Mosca and Pauline Morand for their help in collecting data for this work. Disclosures None.
References 1. Hsu CY, McCulloch CE, Iribarren C, Darbinian J, Go AS: Body mass index and risk for end-stage renal disease. Ann Intern Med 144: 21–28, 2006 2. Iseki K, Ikemiya Y, Kinjo K, Inoue T, Iseki C, Takishita S: Body mass index and the risk of development of end-stage renal disease in a screened cohort. Kidney Int 65: 1870–1876, 2004 3. Fox CS, Larson MG, Leip EP, Culleton B, Wilson PWF, Levy D: Predictors of new-onset kidney disease in a community-based population. JAMA 291: 844–850, 2004 4. Vivante A, Golan E, Tzur D, Leiba A, Tirosh A, Skorecki K, CalderonMargalit R: Body mass index in 1.2 million adolescents and risk for end-stage renal disease. Arch Intern Med 172: 1644–1650, 2012 5. de Jong PE, Verhave JC, Pinto-Sietsma SJ, Hillege HL; PREVEND study group: Obesity and target organ damage: The kidney. Int J Obes Relat Metab Disord 26[Suppl 4]: S21–S24, 2002 6. Michels WM, Grootendorst DC, Verduijn M, Elliott EG, Dekker FW, Krediet RT: Performance of the Cockcroft-Gault, MDRD, and new CKD-EPI formulas in relation to GFR, age, and body size. Clin J Am Soc Nephrol 5: 1003–1009, 2010 7. Earley A, Miskulin D, Lamb EJ, Levey AS, Uhlig K: Estimating equations for glomerular filtration rate in the era of creatinine standardization: A systematic review. Ann Intern Med 156: 785– 795, 2012 8. Dubourg L, Hadj-Aı¨ssa A, Ferrier B: Adaptation of an enzymatic polyfructosan assay to clinical practice. Anal Biochem 405: 266– 268, 2010 9. Cavalier E, Rozet E, Dubois N, Charlier C, Hubert P, Chapelle J-P, Krzesinski JM, Delanaye P: Performance of iohexol determination in serum and urine by HPLC: Validation, risk and uncertainty assessment. Clin Chim Acta 396: 80–85, 2008 10. Bro¨chner-Mortensen J: A simple method for the determination of glomerular filtration rate. Scand J Clin Lab Invest 30: 271–274, 1972 11. Lorentz FH: Der konstitutionsindex der frau. Klin Wochenshr 16: 734–736, 1929 12. Du Bois D, Du Bois EF: A formula to estimate the approximate surface area if height and weight be known. 1916. Nutrition 5: 303–311, 1989
Clin J Am Soc Nephrol 9: 720–727, April, 2014
13. Haycock GB, Schwartz GJ, Wisotsky DH: Geometric method for measuring body surface area: A height-weight formula validated in infants, children, and adults. J Pediatr 93: 62–66, 1978 14. Gehan EA, George SL: Estimation of human body surface area from height and weight. Cancer Chemother Rep 54: 225–235, 1970 15. Livingston EH, Lee S: Body surface area prediction in normalweight and obese patients. Am J Physiol Endocrinol Metab 281: E586–E591, 2001 16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 17. Pai MP: Estimating the glomerular filtration rate in obese adult patients for drug dosing. Adv in Chronic Kidney Dis 17: e53–e62, 2010 18. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D; Modification of Diet in Renal Disease Study Group: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Ann Intern Med 130: 461–470, 1999 19. Friedman AN, Strother M, Quinney SK, Hall S, Perkins SM, Brizendine EJ, Inman M, Gomez G, Shihabi Z, Moe S, Li L: Measuring the glomerular filtration rate in obese individuals without overt kidney disease. Nephron Clin Pract 116: c224–c234, 2010 20. Delanaye P, Radermecker RP, Rorive M, Depas G, Krzesinski JM: Indexing glomerular filtration rate for body surface area in obese patients is misleading: Concept and example. Nephrol Dial Transplant 20: 2024–2028, 2005 21. Pottel LHAH: Is body surface area the appropriate index for glomerular filtration rate? 1–18, 2012 22. Delanaye P, Mariat C, Cavalier E, Krzesinski JM: Errors induced by indexing glomerular filtration rate for body surface area: Reductio ad absurdum. Nephrol Dial Transplant 24: 3593–3596, 2009
Obesity and GFR, Lemoine et al.
727
23. Delanaye P, Krzesinski J-M: Indexing of renal function parameters by body surface area: Intelligence or folly? Nephron Clin Pract 119: c289–c292, 2011 24. Heaf JG: The origin of the 1 x 73-m2 body surface area normalization: Problems and implications. Clin Physiol Funct Imaging 27: 135–137, 2007 25. Ogden CL, Fryar CD, Carroll MD, Flegal KM: Mean body weight, height, and body mass index, United States 1960–2002. Adv Data 347: 1–17, 2004 26. Park EJ, Pai MP, Dong T, Zhang J, Ko CW, Lawrence J, Crentsil V, Zhang L, Xu NN: The influence of body size descriptors on the estimation of kidney function in normal weight, overweight, obese, and morbidly obese adults. Ann Pharmacother 46: 317– 328, 2012 27. Verbraecken J, Van de Heyning P, De Backer W, Van Gaal L: Body surface area in normal-weight, overweight, and obese adults. A comparison study. Metabolism 55: 515–524, 2006 28. Janmahasatian S, Duffull SB, Chagnac A, Kirkpatrick CMJ, Green B: Lean body mass normalizes the effect of obesity on renal function. Br J Clin Pharmacol 65: 964–965, 2008 Received: April 4, 2013 Accepted: December 7, 2013 S.L. and F.G.-E. contributed equally to this work. Published online ahead of print. Publication date available at www. cjasn.org. This article contains supplemental material online at http://cjasn. asnjournals.org/lookup/suppl/doi:10.2215/CJN.03610413/-/ DCSupplemental.
Accuracy of GFR Estimation in Obese Patients Sandrine Lemoine,*†‡§ Fitsum Guebre-Egziabher,†‡§ Florence Sens,†§ Marie-Sophie Nguyen-Tu,| Laurent Juillard,†‡§ Laurence Dubourg,*§| and Aoumeur Hadj-Aissa*§
Abstract Background and objectives Adequate estimation of renal function in obese patients is essential for the classification of patients in CKD category as well as the dose adjustment of drugs. However, the body size descriptor for GFR indexation is still debatable, and formulas are not validated in patients with extreme variations of weight. Design, setting, participants, & measurements This study included 209 stages 1–5 CKD obese patients referred to the Department of Renal Function Study at the University Hospital in Lyon between 2010 and 2013 because of suspected renal dysfunction. GFR was estimated with the Chronic Kidney Disease and Epidemiology equation (CKD-EPI) and measured with a gold standard method (inulin or iohexol) not indexed (mGFR) or indexed to body surface area determined by the Dubois and Dubois formula with either real (mGFRr) or ideal (mGFRi) body weight. Mean bias (eGFR2mGFR), precision, and accuracy of mGFR were compared with the results obtained for nonobese participants (body mass index between 18.5 and 24.9) who had a GFR measurement during the same period of time. Results Mean mGFRr (51.6624.2 ml/min per 1.73 m2) was significantly lower than mGFR, mGFRi, and eGFRCKD-EPI. eGFRCKD-EPI had less bias with mGFR (0.29; 21.7 to 2.3) and mGFRi (21.62; 23.1 to 0.45) compared with mGFRr (8.7; 7 to 10). This result was confirmed with better accuracy for the whole cohort (78% for mGFR, 84% for mGFRi, and 72% for mGFRr) and participants with CKD stages 3–5. Moreover, the Bland Altman plot showed better agreement between mGFR and eGFRCKD-EPI. The bias between eGFRCKD-EPI and mGFRr was greater in obese than nonobese participants (8.7 versus 0.58, P,0.001). Conclusions This study shows that, in obese CKD patients, the performance of eGFRCKD-EPI is good for GFR#60 ml/min per 1.73 m2. Indexation of mGFR with body surface area using ideal body weight gives less bias than mGFR scaled with body surface area using real body weight. Clin J Am Soc Nephrol 9: 720–727, 2014. doi: 10.2215/CJN.03610413
Introduction The increasing prevalence of obesity, especially in the Western world, is associated with the risk to develop kidney disease and the progression of renal disease (1,2). Emerging studies in this last decade indicate that, aside from being a major cause for the development of diabetes and hypertension, obesity may have direct adverse effects on kidney function independent of the other known risk factors (1,3,4). Hormonal factors, oxidative stress, inflammation, and endothelial dysfunction could explain the link between obesity and CKD (5). However, there is still a lack of knowledge about the prevalence and risk factors of obesityrelated kidney disease. Correctly estimating renal function in obese patients is, thus, essential for not only the classification of patients in CKD category but also, better dose adjustment of drugs eliminated by the kidneys in these patients. The most widely used measure of kidney function is eGFR using creatinine-based formulas. There is no formula specially validated for obesity, because eGFR is derived and validated in nonobese 720
Copyright © 2014 by the American Society of Nephrology
individuals with CKD. The Cockroft and Gault formula overestimates GFR in obese patients, because the formula includes body weight. The Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease and Epidemiology (CKD-EPI) equations do not include total weight but have been reported to also overestimate GFR in obese patients (6). Recently, recommendations proposed CKD-EPI as the most reliable method to determine eGFR (7). The accuracy and precision of this equation might still be affected by factors, such as age, muscle mass, diet, and proximal tubule secretion of creatinine. Thus, studies have concluded that no estimation is validated for patients over 75 years, malnourished patients, patients with extreme variations of weight or muscle mass, and patients with a low protein diet (3,4,7). However, MDRD and CKD-EPI equations are more recommended than the Cockroft and Gault formula, because they were developed in North American population, where overweight or obese patients are more prevalent. Inulin clearance is the gold standard measure of GFR. Because GFR varies with weight and height, it is
Departments of *Renal Function Study and †Nephrology, Edouard Herriot Hospital, Hospices Civils de Lyon, Lyon, France; ‡Institut National de la Sante´ et de la Recherche Me´dicale U1060, Laboratoire de Recherche en Cardiovasculaire, Me´tabolisme, Diabe´tologie et Nutrition, Lyon, France; §University of Lyon, Universite´ Lyon 1 Claude Bernard, Lyon, France; and | Unite´ Mixte de Recherche 5305 Centre National de la Recherche Scientifique, Universite´ Claude Bernard, Lyon, France Correspondence: Dr. Sandrine Lemoine, Department of Nephrology, Edouard Herriot Hospital, 5 place d’Arsonval, 69003 Lyon, France. Email: sandrine. [email protected]
www.cjasn.org Vol 9 April, 2014
Clin J Am Soc Nephrol 9: 720–727, April, 2014
evident that GFR comparisons must include some adjustment for body size. However, in obese patients, there is still the problem of the indexation by body surface area (BSA). BSA is disproportionally affected by fat mass in obesity, and the body size descriptor that should be used for this indexation in the setting of obesity is still questionable. To address these fundamental issues, in this study, we compared eGFRCKD-EPI with the gold standard reference method inulin or iohexol clearances nonindexed by BSA or indexed with BSA calculated with either actual body weight or ideal body weight. We evaluated the impact of these alternative body size descriptors on the precision, accuracy, and bias of GFR among obese adults. Because measured GFR (mGFR) is used for drug-dosing purposes and mGFR adjusted to BSA is used for the classification of CKD stages, we compared eGFRCKD-EPI with mGFR, GFR indexed with BSA measured with the Dubois and Dubois formula and ideal weight (mGFRi), and GFR indexed with BSA measured with the Dubois and Dubois formula and real weight (mGFRr).
Materials and Methods Data Collection We retrospectively analyzed the GFR results of 209 obese participants referred in our unit for various nephropathies between February of 2010 and January of 2013 for measurement of inulin clearance because of suspected or established renal dysfunction. We compared these results with the results of a group of nonobese participants who had GFR evaluation during the same period of time. Informed consent was obtained from all patients before measurement of the renal clearance. The consent form contained information on the procedure itself as well as the later use of the information for research purposes. According to the French legislation, concerning the use of a database without direct identification of patients, it was not necessary to obtain an ethical approval (Law 2006–450, April 19, 2004; Commission nationale de l’informatique et des libertés). Creatinine measurement was performed with an enzymatic method with calibration certified by isotope dilution mass spectrometry. Plasma creatinine (PCr) was obtained with the Siemens enzymatic method (on the Dimension Vista System) traceable to National Institute of Standards and Technology creatinine Standard Reference Materials 914 (verified with National Institute of Standards and Technology SRM 967). The GFR measurement was performed with the gold standard method (mGFR): inulin or iohexol clearance. Inulin clearance (INUTEST 25%; Fresenius, Kabi, Austria) was performed with a loading dose of 30 ml/kg that was injected in 10 minutes, with a maintenance dose infusion of a solution of inulin of 40 mg/kg. The urine was collected every 30 minutes, and we performed blood tests in the middle of each period of urine collection (three to four collection periods of 30 minutes). The inulin clearance was calculated in each period to obtain the average. Measurements of plasma and urine polyfructosan concentrations were performed using an enzymatic method (8). For the other remaining 25 patients, we performed iohexol clearance. We injected 8 ml iohexol (300 mg; Omnipaque; GE Healthcare SAS, Vélizy-Villacoublay, France) and weighed
Obesity and GFR, Lemoine et al.
721
the syringe before and after injection. Blood collection was performed at 120, 180, and 240 minutes. The serum iohexol concentration was measured by HPLC (9). The GFR was calculated as GFR=slope3dose/concentration at time 0 corrected with the Bröchner–Mortensen equation (10). GFR was also estimated by CKD-EPI: CKD-EPI ¼ k1 3 ½PCr=k2Þ2k3 3 0:993age ; with k1=141, 143, 163, and 166 for white men and women and black men and women, respectively; k2=0.7 and 0.9 for women and men, respectively; and k3=1.209, 1.209, 0.411, and 0.329 for men with PCr.0.9 mg/dl, women with PCr.0.7 mg/dl, men with PCr#0.9 mg/dl, and women with PCr#0.7 mg/dl, respectively.
Body Size Descriptors The ideal weight was determined by the Lorentz equation (11) ([height (centimeters)2100]2[(height [centimeters] 2150)/4]). The real BSA (BSAr) and ideal BSA were determined by the Dubois and Dubois BSA formula, because it is the most widely and commonly used formula (0.0071843total body weight [kilograms]0.4253height [centimeters]0.725) (12). To assess the BSA that best converts the predicted GFRs, several BSA formulas were used for unadjusted CKD-EPI: formulas by Haycock et al. (13), Gehan and George (14), and Livingston and Lee (15) were used. Obesity was defined using the World Health Organization definition: body mass index (BMI)$30 kg/m2. Reference BMI for nonobese participants was defined by a BMI between 18.5 and 24.9 kg/m2.
Statistical Analyses Statistical analyses were performed with Prism software package (version 6; GraphPad). Mean BSA and mean GFR were compared with a t paired test. The performance of equations was measured by mean bias (eGFRCKD-EPI2mGFR) and mean relative bias ([eGFRCKD-EPI2mGFR]/mGFR) expressed in percentages as well as precision and accuracy. The precision was assessed as an interquartile range for the differences. Accuracy was calculated as the percentage of GFR estimates within 30% deviation of the mGFR. Comparisons of bias and accuracy were performed using t and McNemar’s tests, respectively. The agreement between mGFR and eGFRCKD-EPI was calculated with eGFRCKD-EPI2mGFR on the y axis and eGFRCKD-EPI on the x axis. We analyzed the performance of eGFRCKD-EPI in subgroups of BMI ranges (BMI=30–35, 35–40, and .40 kg/m2). P values,0.05 were considered significant.
Results Characteristics of the Participants In total, 209 obese participants (90 women; 43%) with a mean BMI of 34.864.6 kg/m2 were included (Table 1). In total, 188 nonobese participants with a mean BMI of 22.861.9 kg/m2 were included for comparison. Results of BSA with the Dubois and Dubois formula using real or ideal weight show, as expected, that BSAr is significantly higher than ideal BSA. Results of BSA calculated
722
Clinical Journal of the American Society of Nephrology
Table 1. Patients’ characteristics
Characteristic Age (yr) Weight (kg) Height (cm) Men, n (%) Kidney donor, n (%) Transplant recipient, n (%) Diabetes, n (%) Body mass index (kg/m2), n (%) 30 to ,35 35 to ,40 40 to ,50 .50 Ideal weight (kg) Body surface area (real weight; m2) Body surface area (ideal weight; m2) Creatinine (mg/dl) mGFR (ml/min), n (%) ,15 15–29 30–59 60–89 $90 mGFRr (ml/min per 1.73 m2) mGFRi (ml/min per 1.73 m2) eGFRMDRD (ml/min per 1.73 m2) eGFRCKD-EPI (ml/min per 1.73 m2)
Obese (n=209) 57613.2 (19.4–87) 94.9616 (62.5–195) 16569.8 (163.8–166.4) 119 (57) 3 (1.4) 29 (14) 64 (31) 34.864.6 (30–67.1) 134 (64) 54 (26) 19 (9) 2 (1) 61.367.3 (41–82.6) 260.2 (1.5–2.8) 1.6760.1 (1.2–2.1)b 1.4260.52 (0.4–6.3) 60.2628 (12–146)c 11 (5.2) 20 (9.5) 88 (42.2) 55 (26.3) 35 (16.7) 51.8624 (9–121) 62628 (11–148)c 57.7627 (7–164)c 60.6628 (7–124)c
Nonobese (n=188) 55.5614.5 (19–88) 65.569.5 (47–90)a 16969 (144–193) 110 (59) 4 (3)a 65 (31)a 15 (7)a 22.861.9 (18.5–25)a — — — — — 1.760.17 (1.4–2.2)a — 1.4460.62 (0.7–3.8) 54.8622 (11–105) 5 (2.6) 14 (7.4) 78 (46) 67 (29) 24 (18) 53.8622 (11–114) — 54.6621 (13–120) 53.3622 (18–110)a
Values are expressed as mean6SD (range) or n (%). mGFR, measured GFR nonadjusted with body surface area; mGFRr, GFR indexed with body surface area measured with the Dubois and Dubois equation and real weight; mGFRi, GFR indexed with body surface area measured with the Dubois and Dubois equation and ideal weight; MDRD, Modification of Diet in Renal Disease defined by 1753plasma creatinine (mg/dl)21.1543age (years)20.20330.742 (if woman)31.21 (if black); CKD-EPI, Chronic Kidney Disease and Epidemiology equation. a P,0.05 versus obese population. b P,0.05 real weight body surface area versus ideal weight body surface area. c P,0.05 versus mGFRr.
with other formulas are described in Supplemental Table 1. GFR Measurements Impact of Body Size Descriptors on GFR Measured by Reference Techniques. The mGFRr is significantly lower than mGFR and mGFRi (28.3 and 210.2 ml/min per 1.73 m2, respectively; P,0.001) (Table 1). There is no statistically significant difference between eGFR CKD-EPI (60.6628; 7–124) and mGFR (60.2628; 12–146; P=0.78) or mGFRi (62628; 11–148; P=0.07). The linear regression of eGFR and mGFRi is the closest to the line of identity (Figure 1) compared with mGFRr. eGFRCKD-EPI was well correlated with mGFRr and mGFRi (r=0.89 between eGFRCKD-EPI and mGFRr and r=0.9 between eGFR CKD-EPI and mGFRi; P,0.001). Comparison between eGFRCKD-EPI and mGFR among Obese Participants. The eGFRCKD-EPI (60.6 ml/min per 1.73 m2) has better agreement with mGFR (60.2 ml/min) and mGFRi (61.9 ml/min per 1.73 m2) than mGFRr (51.8 ml/min per 1.73 m2) (Table 1). We, furthermore, analyzed the relative bias, precision, and accuracy of eGFR compared with mGFR, mGFRi, and mGFRr. eGFR CKD-EPI
had significantly more bias with mGFRr than mGFR or mGFRi (P,0.001) (Table 2). The biases between eGFR CKD-EPI and mGFR are not significantly different from the biases between eGFR C K D - E P I and mGFRi (P=0.23). Moreover, accuracy is better between eGFRCKD-EPI and mGFRi than mGFRr (84% versus 72% for CKD-EPI, respectively) and between eGFR CKD-EPI and mGFR than mGFRr (78% versus 72%, respectively) (Table 2). Furthermore, precision between eGFRCKD-EPI and mGFRi is better than mGFRr. Figure 2 confirms the good agreement between eGFR and mGFRi, with the lowest bias between eGFRCKD-EPI and mGFRi. The bias between eGFR and mGFRr is less in nonobese than obese patients, excluding a methodological problem. Compared with the MDRD equation, the CKD-EPI equation showed the lowest bias (mean=21.62 versus 24.67 ml/min per 1.73 m2; P=0.04 for mGFRi; mean=0.29 versus 2.5 ml/min per 1.73 m2 for mGFR; data not shown). Because eGFRCKD-EPI was the least biased equation, additional analyses were performed using this equation. Comparison between CKD-EPI mGFR, mGFRi, and mGFRr and BMI. The Bland Altman plot shows that the biases are important when BMI is up to 40 kg/m2 for
Clin J Am Soc Nephrol 9: 720–727, April, 2014
Obesity and GFR, Lemoine et al.
723
Figure 1. | No statistically significant difference between mGFRi, mGFRr, and eGFR CKD-EPI. Log–log scale regression between mGFRi, mGFRr, and eGFRCKD-EPI. CKD-EPI, Chronic Kidney Disease and Epidemiology equation; mGFRi, GFR indexed with body surface area measured with the Dubois and Dubois equation and ideal weight (milliliters per minute per 1.73 m2); mGFRr, GFR indexed with body surface area measured with the Dubois and Dubois equation and real weight (milliliters per minute per 1.73 m2). The black line is the identity line. The long dashed line is the regression line between eGFR and mGFRr, and the small dashed line is the regression line between eGFR and mGFRi.
Table 2. Bias, precision, and accuracy between eGFRCKD-EPI and mGFR, mGFRr, or mGFRi
CKD-EPI Variables Obese mGFR Bias (ml/min), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min) Accuracy, % (95% confidence interval) mGFRr Bias (ml/min per 1.73 m2), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min per 1.73 m2) Accuracy, % (95% confidence interval) mGFRi Bias (ml/min per 1.73 m2), mean (95% confidence interval) Mean relative bias (%) Precision (ml/min per 1.73 m2) Accuracy, % (95% confidence interval)
0.3a (21.7 to 2.3) 21a (18 to 23) 14 78a (72 to 84) 8.7a,c (7 to 10) 25a,c (21 to 28) 11 72a (66 to 78) 21.6c (23.1 to 0.4) 19c (16 to 21) 11 84c (79 to 89)
Nonobese 1 (20.7 to 2.6) 14 (12 to 15) 8.5 87b (85 to 89) 0.6b (20.6 to 1.8) 16b (15 to 19) 5.7 92b (90 to 93) — — — —
Mean bias is expressed as the mean of eGFR2mGFR. Mean relative bias is expressed as the mean of absolute values of (eGFR2mGFR)/ mGFR. Precision refers to the 25th percentile275th percentile. Accuracy refers to the percentage of GFR estimates that is within 30% of measured mGFR. a P,0.01 versus mGFRi. b P,0.01 versus obese population. c P,0.01 versus mGFR.
724
Clinical Journal of the American Society of Nephrology
Figure 2. | Good agreement and lowest bias between eGFR and mGFRi. The graphical plot of the difference between (eGFRCKD-EPI–mGFR, mGFRr or mGFRi) and eGFR. Mean bias is 0.369.7 ml/min per 1.73 m2 for CKD-EPI versus mGFR, 8.7614.4 ml/min per 1.73 m2 for CKD-EPI versus mGFRr, 21.62613.3 ml/min per 1.73 m2 for CKD-EPI versus mGFRi. The dotted lines represent the limits of agreement (mean difference62 SD) and the dashed line represents the mean difference. mGFRi, GFR indexed with BSA measured with Dubois and Dubois and ideal weight; mGFRr, GFR indexed with BSA measured with Dubois and Dubois and real weight.
Figure 3. | Biases are more important for body mass index (BMI)>40 kg/m2. The graphical plot of the difference between (eGFRCKD-EPI–mGFR, mGFRr or mGFRi) and eGFR. Mean bias is 0.369.7 ml/min per 1.73 m2 for CKD-EPI versus mGFR, 8.7614.4 ml/min per.73 m2 for CKD-EPI versus mGFRr, 21.62613.3 ml/min per 1.73 m2 for CKD-EPI versus mGFRi. The dotted lines represent the limits of agreement (mean difference62 SD) and the dashed line represents the mean difference. mGFRi, GFR indexed with BSA measured with Dubois and Dubois and ideal weight; mGFRr, GFR indexed with BSA measured with Dubois and Dubois and real weight.
mGFR adjusted with BSA (Figure 3). The performances of eGFRCKD-EPI for subgroups of BMI categories are shown in Table 3. For BMI greater than 40 kg/m2, eGFRCKD-EPI shows more bias with mGFRi and mGFR (27.47 and 24.8 ml/min per 1.73 m2, respectively), and accuracies for mGFRr, mGFRi, and mGFR are dramatically reduced (68%, 77%, and 66%, respectively; P,0.001). For BMI range=35–40 kg/m2, precision is better with eGFRCKD-EPI, and accuracies are reduced with mGFRr. Performance of eGFRCKD-EPI According to the mGFR Level. With eGFRCKD-EPI, patients with CKD stages 3 or 4 and 5 are correctly classified if we compare with mGFR or mGFRi (71% or 82%, respectively, are well classified in CKD stage 3; 76% or 81%, respectively, are well classified in CKD stages 4 and 5) (Table 4). Comparison between mGFR and eGFRCKD-EPI Unadjusted to BSA. Several BSA formulas were used to unadjust CKD-EPI to assess the best body size descriptor that converts to the predicted GFR (Table 5). eGFRCKD-EPI adjusted or not to BSA calculated with the Livingston and Lee (15) formula is not substantially different from mGFR (60.6628 and 59.1627.4 ml/min, respectively; P.0.20).
Discussion This study shows that, in obese participants, GFR measured with the gold standard technique (inulin or iohexol clearance) underestimates GFR when adjusted to BSA using the real body weight, whereas there was no
significant difference between mGFR and mGFRr in nonobese participants. eGFR with the CKD-EPI equation is not different from the mGFR or mGFRi, showing that the CKDEPI equation offers a good GFR prediction in obese participants, especially stage 3 CKD patients. There is an emerging recognition of obesity-related kidney disease, and it requires the reassessment of traditional GFR measurements and estimation methods developed in nonobese patients. Most studies relate that formulas are not validated in obese population. Although these equations were established in populations that are not obese, average BMI for the CKD-EPI formula was 28 kg/m2 (16–19). This population was overweight but not obese. Moreover, these formulas have been compared with a reference method adjusted with BSAr. Inulin clearance remains the gold standard, but it is the indexation with the real weight that produces an underestimation of GFR, which is clearly shown in our study. Indeed, if we compare our results with the results found in the population used to develop the CKD-EPI equation, we found the same accuracy for the percentage of GFR estimates that are within 30% of mGFR for the CKD-EPI equation (16). Indeed, muscle mass does not increase proportionally with total body weight but only lean body weight. There is no consensus on the body size descriptor that should be used in the estimation of GFR. Although this indexation has a poor influence in the normal weight population, it can induce important deviations in obese or
Clin J Am Soc Nephrol 9: 720–727, April, 2014
Obesity and GFR, Lemoine et al.
725
Table 3. Performance of CKD-EPI for body mass index category
Bias Mean (95% Confidence Interval)
BMI Categories Body mass index=30–35 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi Body mass index=35–40 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi Body mass index>40 mg/kg2 CKD-EPI mGFR CKD-EPI mGFRr CKD-EPI mGFRi
22.3 (0 to 4.8) 10.3a,b (0.2 to 12) 1.26 (20.8 to 3.4)
Accuracy (30%), % (95% Confidence Interval)
Precision
12 15 10
88 (83.6 to 92.4) 82 (76.8 to 87.2) 88 (83.6 to 92.4)
23 (27.1 to 1.2) 5.6b (2.3 to 8.9) 25.5 (29.2 to 21.8)
5.5 11.5 12
72 (67.1 to 76.9) 46 (39.2 to 52.8) 81 (75.7 to 86.3)
24.8 (21.7 to 2) 6.5a,b (0.14 to 12) 27.5b (214 to 20.8)
8.5 14.5 11.5
66 (59.6 to 72.4) 68 (61.7 to 74.3) 77 (71.3 to 82.7)
Relative biases are expressed as the mean of eGFR2mGFR (milliliters per minute per 1.73 m2). Precision refers to the 25th percentile2 75th percentile. Accuracy refers to the percentage of GFR estimates that is within 30% of measured mGFR. BMI, body mass index. a P,0.01 versus mGFRi. b P,0.01 versus mGFR.
Table 4. Number and percentage of subjects correctly classified for CKD stage
Variables
Stage 1 (GFR$90 ml/min per 1.73 m2)
Stage 2 (GFR=60–89 ml/min per 1.73 m2)
Stage 3 (GFR=30–59 ml/min per 1.73 m2)
Stages 4 and 5 (GFR#30 ml/min per 1.73 m2)
mGFR CKD-EPI mGFRr CKD-EPI mGFRi CKD-EPI
35 26 (74%) 21 20 (95%) 38 30 (79%)
54 33 (61%) 49 26 (53%) 65 36 (55%)
90 64 (71%) 98 67 (68%) 79 65 (82%)
30 23 (76%) 41 28 (68%) 27 22 (81%)
Reference method is mGFR, mGFRr, or mGFRi. Values for CKD-EPI are expressed as number of patients (%).
Table 5. Comparison between mGFR and eGFRCKD-EPI or eGFRCKD-EPI unadjusted with body surface area measured using different formulas and ideal or real weight for the obese cohort
With Ideal Weight Variables mGFR (ml/min) CKD-EPI CKD-EPI3DD/1.73 CKD-EPI3Ha/1.73 CKD-EPI3GG/1.73 CKD-EPI3LL/1.73
With Real Weight
Mean6SD
P Value
Mean6SD
P Value
60.2628 60.6628 58.6627 58.6627 58.8628 59.1627
0.12 ,0.001 ,0.001 ,0.001 0.20
60.2628 60.6628 70.2633 73.8634 73.5634 77.3636
0.12 ,0.001 ,0.001 ,0.001 ,0.001
All values are compared with mGFR. P values versus with mGFR. DD, Dubois and Dubois; Ha, Haycock et al. (13); GG, Gehan and George (14); Livingston and Lee (15).
anorectic patients (6,20). BSA was seen as an alternative for reflecting the metabolic rate. Indexing with BSA is essential to make comparisons between subjects, have reference values, and be classified in a status of chronic renal disease (21). Indexation is necessary in patients with normal BMI
with extreme height. However, there is no strong linear relation between GFR and BSA. Scaling GFR to body size may be misleading, and it is controversial in obese subjects. Delanaye et al. (20,22) recommended the use of absolute GFR values instead of the GFRs indexed to BSA,
726
Clinical Journal of the American Society of Nephrology
especially in abnormal body size populations (20,22). However, because indexation is still a matter of debate, we evaluated the accuracy of BSA indexation of measured GFR by using different alternative body size descriptors. Moreover, the indexation with standard BSA (1.73 m2) is also debatable (23,24). BSA of 1.73 m2 is no longer representative of a normal average morphology. This indexing is based on an old principle developed in 1883 by Rubner and colleagues, and it is not totally true (21). Actually, the average estimated BSA is 2.06 m2 for men and 1.83 m2 for women in the United States (25). In clinical practice, if we consider drug dosing, mGFR is used as a reference. Indeed, drug clearance is not proportional to total body weight, and estimation of drug-dosing intervals has been based on the clearance of creatinine. We showed that the CKD-EPI equation has less bias with mGFR, with good accuracy. Although the CKD-EPI equation is an equation adjusted to BSA, we found no difference between eGFRCKD-EPI and mGFR. Because there is no difference between eGFRCKD-EPI and mGFR and given that unadjusted eGFRCKD-EPI results in less precision, it is, therefore, unnecessary to unadjust eGFRCKD-EPI by BSA. If GFR needs to be indexed for BSA, especially for the purpose of CKD stage classification, there is still controversy on how to calculate the BSA in obese subjects. Indeed, equations used to estimate BSA have not been validated in obese individuals and are not stratified by sex (26). Measuring BSA in obese subjects still remains a challenging and unsolved problem, and choosing the best body size descriptor to scale kidney function is difficult (24). Although the Dubois and Dubois formula has already been criticized, we found that BSA and mGFR with another BSA equation (the formula by Livingston and Lee [15]) are not significantly different (Supplemental Table 1) from the results derived from the Dubois and Dubois formula. Our finding is similar to the results already published by Verbraecken et al. (27). Alternative body size descriptors of total body weight could be proposed. Ideal body weight, adjusted body weight, and fat-free weight stratified by sex have been used as estimates of lean body mass for the calculation of eGFR in obese patients (26). In our study, there is good agreement between eGFRCKD-EPI and mGFRi (Figure 1), and mGFRi has less bias than mGFRr, specifically for GFR values above 20 ml/min per 1.73 m2. Lean body weight can be estimated with dual energy x-ray absorptiometry or bioelectric impedance analysis. Janmahasatian et al. (28) have reported that GFR values adjusted with total body weight underestimate GFR in obese patients, whereas GFR scaled by lean body weight was not different between obese and nonobese participants, suggesting that renal function is closely related to lean body mass and not influenced by adipose tissue. Janmahasatian et al. (28) concluded that lean body weight should be used instead of total body weight for estimating creatinine clearance. Another crucial question that needs to be addressed is the threshold of BMI for which eGFRCKD-EPI can be safely used. In our study, eGFRCKD-EPI for obese people with BMI greater than 40 kg/m2 seemed to be less accurate, but because of the limited number of participants in this range of BMI, we cannot make strong conclusions. Hence, over this BMI measurement, we recommend using a gold standard
method adjusted to ideal body weight to classify CKD and mGFR for drug-dosing purposes. Finally, the hallmark of obesity is excess adipose tissue mass or fat mass for a given body weight. The increased body mass in obesity is caused by both increased fat and fat-free mass. The relative contributions of these two components to the excess weight are influenced, in part, by age and an individual’s proportion of fat and fat-free mass before weight gain, assuming adequate protein intake occurs during the weight gain process. However, older individuals and subjects with chronic disease can become obese without a parallel growth of muscle mass. In this case, body weight may not be accurately estimated by the ideal weight. This work shows that the CKD-EPI equation is validated in the obese population up to a BMI range of 40 kg/m2, specifically for GFR levels ,60 ml/min; this level is important, because it is when patients often face CKD-related complications or need dose adjustment for drug. Indexation of mGFR with BSA using ideal body weight gives less bias than mGFR scaled with BSA using real body weight for GFR.20 ml/min per 1.73 m2. Acknowledgments We thank Mélodie Mosca and Pauline Morand for their help in collecting data for this work. Disclosures None.
References 1. Hsu CY, McCulloch CE, Iribarren C, Darbinian J, Go AS: Body mass index and risk for end-stage renal disease. Ann Intern Med 144: 21–28, 2006 2. Iseki K, Ikemiya Y, Kinjo K, Inoue T, Iseki C, Takishita S: Body mass index and the risk of development of end-stage renal disease in a screened cohort. Kidney Int 65: 1870–1876, 2004 3. Fox CS, Larson MG, Leip EP, Culleton B, Wilson PWF, Levy D: Predictors of new-onset kidney disease in a community-based population. JAMA 291: 844–850, 2004 4. Vivante A, Golan E, Tzur D, Leiba A, Tirosh A, Skorecki K, CalderonMargalit R: Body mass index in 1.2 million adolescents and risk for end-stage renal disease. Arch Intern Med 172: 1644–1650, 2012 5. de Jong PE, Verhave JC, Pinto-Sietsma SJ, Hillege HL; PREVEND study group: Obesity and target organ damage: The kidney. Int J Obes Relat Metab Disord 26[Suppl 4]: S21–S24, 2002 6. Michels WM, Grootendorst DC, Verduijn M, Elliott EG, Dekker FW, Krediet RT: Performance of the Cockcroft-Gault, MDRD, and new CKD-EPI formulas in relation to GFR, age, and body size. Clin J Am Soc Nephrol 5: 1003–1009, 2010 7. Earley A, Miskulin D, Lamb EJ, Levey AS, Uhlig K: Estimating equations for glomerular filtration rate in the era of creatinine standardization: A systematic review. Ann Intern Med 156: 785– 795, 2012 8. Dubourg L, Hadj-Aı¨ssa A, Ferrier B: Adaptation of an enzymatic polyfructosan assay to clinical practice. Anal Biochem 405: 266– 268, 2010 9. Cavalier E, Rozet E, Dubois N, Charlier C, Hubert P, Chapelle J-P, Krzesinski JM, Delanaye P: Performance of iohexol determination in serum and urine by HPLC: Validation, risk and uncertainty assessment. Clin Chim Acta 396: 80–85, 2008 10. Bro¨chner-Mortensen J: A simple method for the determination of glomerular filtration rate. Scand J Clin Lab Invest 30: 271–274, 1972 11. Lorentz FH: Der konstitutionsindex der frau. Klin Wochenshr 16: 734–736, 1929 12. Du Bois D, Du Bois EF: A formula to estimate the approximate surface area if height and weight be known. 1916. Nutrition 5: 303–311, 1989
Clin J Am Soc Nephrol 9: 720–727, April, 2014
13. Haycock GB, Schwartz GJ, Wisotsky DH: Geometric method for measuring body surface area: A height-weight formula validated in infants, children, and adults. J Pediatr 93: 62–66, 1978 14. Gehan EA, George SL: Estimation of human body surface area from height and weight. Cancer Chemother Rep 54: 225–235, 1970 15. Livingston EH, Lee S: Body surface area prediction in normalweight and obese patients. Am J Physiol Endocrinol Metab 281: E586–E591, 2001 16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 17. Pai MP: Estimating the glomerular filtration rate in obese adult patients for drug dosing. Adv in Chronic Kidney Dis 17: e53–e62, 2010 18. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D; Modification of Diet in Renal Disease Study Group: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Ann Intern Med 130: 461–470, 1999 19. Friedman AN, Strother M, Quinney SK, Hall S, Perkins SM, Brizendine EJ, Inman M, Gomez G, Shihabi Z, Moe S, Li L: Measuring the glomerular filtration rate in obese individuals without overt kidney disease. Nephron Clin Pract 116: c224–c234, 2010 20. Delanaye P, Radermecker RP, Rorive M, Depas G, Krzesinski JM: Indexing glomerular filtration rate for body surface area in obese patients is misleading: Concept and example. Nephrol Dial Transplant 20: 2024–2028, 2005 21. Pottel LHAH: Is body surface area the appropriate index for glomerular filtration rate? 1–18, 2012 22. Delanaye P, Mariat C, Cavalier E, Krzesinski JM: Errors induced by indexing glomerular filtration rate for body surface area: Reductio ad absurdum. Nephrol Dial Transplant 24: 3593–3596, 2009
Obesity and GFR, Lemoine et al.
727
23. Delanaye P, Krzesinski J-M: Indexing of renal function parameters by body surface area: Intelligence or folly? Nephron Clin Pract 119: c289–c292, 2011 24. Heaf JG: The origin of the 1 x 73-m2 body surface area normalization: Problems and implications. Clin Physiol Funct Imaging 27: 135–137, 2007 25. Ogden CL, Fryar CD, Carroll MD, Flegal KM: Mean body weight, height, and body mass index, United States 1960–2002. Adv Data 347: 1–17, 2004 26. Park EJ, Pai MP, Dong T, Zhang J, Ko CW, Lawrence J, Crentsil V, Zhang L, Xu NN: The influence of body size descriptors on the estimation of kidney function in normal weight, overweight, obese, and morbidly obese adults. Ann Pharmacother 46: 317– 328, 2012 27. Verbraecken J, Van de Heyning P, De Backer W, Van Gaal L: Body surface area in normal-weight, overweight, and obese adults. A comparison study. Metabolism 55: 515–524, 2006 28. Janmahasatian S, Duffull SB, Chagnac A, Kirkpatrick CMJ, Green B: Lean body mass normalizes the effect of obesity on renal function. Br J Clin Pharmacol 65: 964–965, 2008 Received: April 4, 2013 Accepted: December 7, 2013 S.L. and F.G.-E. contributed equally to this work. Published online ahead of print. Publication date available at www. cjasn.org. This article contains supplemental material online at http://cjasn. asnjournals.org/lookup/suppl/doi:10.2215/CJN.03610413/-/ DCSupplemental.
