Iohexol plasma clearance measurement in older adults with chronic kidney disease sampling time matters

Nephrol Dial Transplant (2015) 30: 1307 1314 doi: 10.1093/ndt/gfv116 Advance Access publication 4 June 2015 Original Articles Iohexol plasma clearance measurement in older adults with chronic kidney disease sampling time matters Natalie Ebert 1, Amina Loesment 1, Peter Martus 2, Olga Jakob 3, Jens Gaedeke 4, Martin Kuhlmann 5, Jan Bartel 6, Mirjam Schuchardt 7, Markus Tölle 7, Tao Huang 7, Markus van der Giet 7 and Elke Schaeffner 1 1 Division of Nephrology, Charité University Medicine, Campus Virchow, Berlin, Germany, 2 Institute of Clinical Epidemiology and Medical Biostatistics, University of Tübingen, Tübingen, Germany, 3 Institute of Biostatistics and Clinical Epidemiology, Charité, Berlin, Germany, 4 Division of Nephrology, Charité University Medicine, Campus Mitte, Berlin, Germany, 5 Department of Nephrology, Vivantes Klinikum im Friedrichshain, Berlin, Germany, 6 Limbach Laboratory, Heidelberg, Germany and 7 Division of Nephrology, Charité University Medicine Campus Benjamin Franklin, Berlin, Germany Correspondence and offprint requests to: Natalie Ebert; E-mail: natalie.ebert@charite.de ABSTRACT Background. Accurate and precise measurement of GFR is important for patients with chronic kidney disease (CKD). Sampling time of exogenous filtration markers may have great impact on measured GFR (mgfr) results, but there is still uncertainty about optimal timing of plasma clearance measurement in patients with advanced CKD, for whom 24-h measurement is recommended. This satellite project of the Berlin Initiative Study evaluates whether 24-h iohexol plasma clearance reveals a clinically relevant difference compared with 5-h measurement in older adults. Methods. In 104 participants with a mean age of 79 years and diagnosed CKD, we performed standard GFR measurement over 5 h (mgfr 300 ) using iohexol plasma concentrations at 120, 180, 240 and 300 min after injection. With an additional sample at 1440 min, we assessed 24-h GFR measurement (mgfr 1440 ). Study design was cross-sectional. Calculation of mgfr was conducted with a one compartment model using the Brochner Mortensen equation to calculate the fast component. mgfr values were compared with estimated GFR values (MDRD, CKD-EPI, BIS1, Revised Lund-Malmö and Cockcroft-Gault). Results. In all 104 subjects, mgfr 1440 was lower than mgfr 300 (23 ± 8 versus 29 ± 9 ml/min/1.73 m 2, mean ± SD; P < 0.001). mgfr 1440 was highly correlated with mgfr 300 (r = 0.9). The mean absolute difference mgfr 300 mgfr 1440 was 5.9 ml/ min/1.73 m 2 corresponding to a mean percentage difference of 29%. In individuals with egfr CKD-EPI 30 ml/min/1.73 m 2, percentage difference of mgfr 300 and mgfr 1440 was even higher (35%). To predict mgfr 1440 from mgfr 300, we developed the correction formula: mgfr 1440 = 2.175 + 0.871 mgfr 300 (1-fold standard error of estimate: ±2.3 ml/min/ 1.73 m 2 ). The GFR estimating equation with the best accuracy and precision compared with mgfr 300 and mgfr 1440 was the Revised Lund Malmö. Conclusions. In elderly CKD patients, measurement of iohexol clearance up to 5 h leads to a clinically relevant overestimation of GFR compared with 24-h measurement. In clinical care, this effect should be bore in mind especially for patients with considerably reduced GFR levels. A new correction formula has been developed to predict mgfr 1440 from mgfr 300. For accurate GFR estimates in elderly CKD patients, we recommend the Revised Lund Malmö equation. Keywords: 24 hours iohexol clearance, elderly, GFR measurement protocol, measured GFR, reduced renal function INTRODUCTION Glomerular filtration rate (GFR) is widely accepted as the best indicator of kidney function and is used for the evaluation, detection and management of chronic kidney disease (CKD). In clinical practice, GFR is most often estimated by the level of serum creatinine and/or cystatin C with the aid of GFR estimating equations, because measuring GFR is time consuming and requires settings where the infrastructure and technical prerequisites are available [1]. In recent years, iohexol, a non-ionic, non-radioactive and low osmolar contrast medium, has gained attractiveness for The Author 2015. Published by Oxford University Press Downloaded from on behalf https://academic.oup.com/ndt/article-abstract/30/8/1307/2324989 of ERA-EDTA. All rights reserved. 1307

ORIGINAL ARTICLE plasma clearance measurement as it is safe; has very little risk of side effects, negligible tubular secretion and protein binding; is not excreted through metabolism; has low extrarenal and biliary clearance and avoids the inconvenience of urine collection [2 6]. It has also been shown to be an unbiased alternative to urinary inulin clearance [7]. The determination of GFR can be obtained by a single bolus injection of iohexol and subsequent blood sampling over a 4 10 h period [8 10]. By calculating the area under the curve (AUC) as a function of time, the blood disappearance curve can be monitored and then mathematically modelled to assess the exact individual GFR [9, 11, 12]. Recommendations for iohexol plasma clearance measurement in individuals with moderately reduced kidney function suggest that sampling time should be postponed to 24 h after injection and in case of very severely reduced kidney function up to 72 h [13, 14]. The exact physiological mechanisms responsible for prolonged plasma sampling are not yet sufficiently known. For the optimal timing to measure GFR using a two-compartment model, the knowledge of the parameters α (fast component distribution constant) and β (slow component distribution constant) should be known. As primarily β is dependent on GFR, consequently the overall sampling time should follow the degree of renal impairment, thus individuals with more severely reduced GFR would require longer sampling time [15]. It has been shown in individuals with reduced kidney function and in transplanted patients that not only did shorter sampling time overestimate true GFR but it also led to a reduced precision of the estimate [15, 16] which has also been demonstrated when renal clearance was calculated with a single plasma sample measurement based on a one-compartment model [17]. In older adults exact assessment of measured GFR (mgfr) is crucial as prevalence of individuals with CKD is higher than in the general population and the risk for harm as a result of inadequate medication dosing and application of contrast media is accelerated [18]. Measuring GFR over the course of 24 h in older adults with considerably reduced kidney function, however, is a logistic challenge. Thus, it is of high clinical relevance to assess whether the additional ( practical and financial) efforts of a 24-h iohexol clearance measurement outweigh the standard measuring of only 5 h. To evaluate whether iohexol plasma clearance measurement in older adults with CKD requires a sampling time of 24 h, we performed iohexol plasma measurements at 120, 180, 240, 300 (mgfr 300 ) and additionally 1440 min (mgfr 1440 ) in 104 study participants aged 70 with CKD. Furthermore, we investigated whether a correction formula could be developed to extrapolate mgfr 1440 values from iohexol clearance measurements over the course of 5 h only. MATERIALS AND METHODS Study design and variables The study sample (n = 104) is part of a satellite project of the Berlin Initiative Study (BIS), a cohort study that investigates the epidemiology of CKD in old age [19]. The primary goal of the present study was to compare iohexol clearance measurement over the course of 5 h with 24-h measurement. In contrast to the original BIS population, the present study participants were under nephrological care and were recruited through cooperating nephrology outpatient clinics at the Charité, the Vivantes Klinik im Friedrichshain, and affiliated private practices (all in Berlin, Germany). Inclusion criteria were age 70 and above, considerably impaired kidney function with externally measured ambulatory serum creatinine level of at least 1.5 mg/dl, a thyroid-stimulating hormone level >0.3 miu/l and no known iodine allergy. All participants gave informed consent for 24 h of iohexol clearance measurement. The local ethics committee approved the study. Variables During the visit, a standardized interview was performed including age, gender, blood pressure, body mass index, history of diabetes mellitus, myocardial infarction, stroke and cancer. Blood samples for serum creatinine and urine samples were obtained before iohexol was applied. Albuminuria was defined as albumin creatinine ratio (ACR) of >30 mg/g. Iohexol measurement All clearance measurements were started between 8:00 and 10:30 a.m., and participants were given specific instructions about food, fluid intake (no caffeine intake), physical activity and avoidance of medication (NSAIDs) that were to be followed prior to and during the examination. Dietary information included material that was sent to the participants instructing them to abstain from high protein foods for 12 h prior to and during the 5 h measurement period; also, the exact composition of meals was explained in personal contact by a specifically trained healthcare worker before the study visit. For the following morning (the 24-h blood sampling), no dietary restrictions were given. Potential candidates for iohexol measurement with oedema, ascites or clinically symptomatic heart failure were excluded from the study. Five millilitres of iohexol solution, containing 3235 mg of iohexol (Accupaque, GE Healthcare Buchler, Braunschweig, Germany) was administered intravenously into an antecubital, forearm or hand vein, and flushed with 10 ml of saline. Five blood samples were contained from the contralateral arm at 120, 180, 240, 300 and additionally at 1440 min after injection. Samples were centrifuged for 10 min at 1500 revolutions per minute within 2 h of collection and transported on dry ice to be stored at 80 C until further analysis at Charité University Hospital. Further details of the iohexol protocol can be found elsewhere [18]. Samples were assayed by high-performance liquid chromatography (HPLC). HPLC analysis of the supernatant was carried out on a Hitachi HPLC system with a Diode array detector (Hitachi) and a Chromolith performance HPLC column (RP-18e, 100 4.6 mm, Merck) and a Chromolith guard column (RP-18e, 5 4.6 mm, Merck) [9, 20]. The coefficient of variation for iohexol was 1.27, 2.36 and 2.84% for index values of 140.86, 60.69 and 24.64 µg/ml, respectively. External quality control was provided by Equalis (Uppsala, Sweden). GFR was calculated using a one-compartment model with the clearance computed from the amount of the marker administered and the AUC of plasma concentration versus time. The mgfr 300 was calculated using blood samples at four time points from 120 until 300 min, and mgfr 1440 was calculated using blood samples at five time points from 120 until 1440 min [9, 13]. In both 1308 N. Ebert et al.

approaches, the fast component was not estimable; thus, the method of Brochner Mortensen was applied [9, 11]. mgfr was corrected for body surface area (BSA) by the factor individual BSA divided by the standard value 1.73 m 2.Weusedthe DuBois DuBois formula for calculation of the BSA [21]. Other blood measures All creatinine samples were analysed at Synlab laboratory Heidelberg using the isotope dilution mass spectrometry (IDMS) traceable enzymatic method from Roche (Crea plus; Roche Diagnostics, Mannheim, Germany) on a Roche Modular-analyzer P-Modul. The inter-assay coefficient of variation for serum creatinine levels was 2.2 and 1.6% at creatinine concentrations of 1.05 and 3.73 mg/dl, respectively. To convert mg/dl to µmol/l, values are to be multiplied by 88.4. Statistical analysis In our study sample, we compared the extended iohexol clearance measurement over 24 h with the standard measurement over 5 h. The analysis was done in three steps: in the first step, mgfr 300 and mgfr 1440 were compared, assuming mgfr 1440 being the more valid measurement ( gold standard ). In the second step, a correction formula for estimating mgfr 1440 from mgfr 300 was constructed. In the third step, the validity of this formula was assessed. In the first step, agreement between uncorrected mgfr 300 and mgfr 1440 was assessed by plotting absolute and percentage differences between the two clearance measurements against the mean (Bland and Altman plot) [22]. Bias, precision and accuracy were assessed by mean bias, standard deviation (SD) of differences, r 2, standard error of estimates and P 30 representing the percentage of mgfr 300 values within 30% of the 24-h measurement value (mgfr 1440 ). Additionally, mean bias, mean percentage bias, r 2 and P 30 were determined for the subgroups with egfr (Chronic Kidney Disease Epidemiology Collaboration equation) [23] below and above the cut-off of 30 ml/min/1.73 m 2. In the second step, a correction formula was constructed using a linear regression model to estimate mgfr 1440 from mgfr 300. The 104 subjects were subdivided randomly into two equal subsamples of 52 subjects each. The first sample was used for development of the formula (learning sample) and the second for internal validation (validation sample). The proposed correction formula, however, was derived from the entire sample of 104 subjects. Additionally, a linear regression model excluding the intercept was calculated to obtain a rough correction factor that corresponds to a percentage reduction of mgfr 300 to estimate mgfr 1440. In the third step, bias, precision and accuracy were determined for the predicted mgfr 1440 in the same way as described for the uncorrected mgfr 300. This was done for the entire sample but not for the subsamples according to egfr CKD-EPI. Finally, we compared egfr and the difference of egfr to mgfr 300 and mgfr 1440, and assessed bias (mean difference of egfr and mgfr), percentage differences [mean difference of (egfr minus mgfr)/mgfr 100)], precision (SD of the difference) and accuracy [ percentage of estimates within 10 and 30% of the mgfr (P 10,P 30 )] for the following five equations: Modification of Diet in Renal Diseases (MDRD) [24], Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) [23], Berlin Initiative Study (BIS1) [18], Revised Lund-Malmö (Rev LM) [25] and Cockcroft-Gault (CG) [26] adjustedto1.73m 2 BSA. Equations are described in detail in the legend of Table 1. RESULTS Study population Mean age of the study population was 79.2 years (range 70 94); 67% of the participants were male, 40% were diabetic, and 94% suffered from arterial hypertension. The mean (±SD) concentration of serum creatinine was 1.91 (±0.6) mg/dl. In all 104 patients, pre-existing externally measured ambulatory creatinine levels were at least 1.50 mg/dl. However, in 16 cases, the Table 1. Main characteristics of study population Characteristics Total sample Number of participants (n) 104 Female, n (%) 34 (33) Age a 79.2 ± 6.1 (70 94) a mgfr 300, ml/min/1.73 m 2 29.3 ± 9.0 (11.0 53.7) a mgfr 1440, ml/min/1.73 m 2 23.4 ± 8.1 (7.2 44.2) Serum creatinine a, mg/dl 1.91 ± 0.60 (1.21 4.77) Serum cystatin C a, mg/l 2.13 ± 0.7 (1.1 5.3) ACR b, mg/g 41.0 (0 6271) BMI a, kg/m 2 28.0 ± 4.5 (19.0 46.1) Body surface area a,m 2 1.9 ± 0.2 (1.5 2.3) Diabetes mellitus, n (%) 42 (40) Hypertension, n (%) 98 (94) Myocardial infarction, n (%) 29 (28) Stroke, n (%) 10 (10) Cancer, n (%) 30 (29) Smoking history, n (%) 90 (87) Systolic blood pressure a, mmhg 132.8 ± 15.1 (87.5 183.5) Diastolic blood pressure a, mmhg 72.8 ± 8.9 (54.0 90.0) Haemoglobin (Hb) a, g/dl 12.6 ± 3.4 (9.1 17.4) Albuminuria c, n (%) 59 (57) Anaemia d, n (%) 51 (49) egfr MDRD, ml/min/1.73 m 2 34.1 ± 10.0 (8.8 58.1) egfr CKD-EPI, ml/min/1.73 m 2 32.8 ± 9.9 (8.1 57.2) egfr BIS1, ml/min/1.73 m 2 33.5 ± 7.8 (12.4 50.8) egfr Revised LM, ml/min/1.73 m 2 29.3 ± 9.2 (9.1 51.9) egfr CG, ml/min/1.73 m 2 32.6 ± 9.3 (10.6 55.0) To convert mg/dl to µmol/l, values are to be multiplied by 88.4. ACR, albumin creatinine ratio; BMI, body mass index; MDRD, modification of diet in renal diseases; CKD-EPI, chronic kidney disease epidemiology collaboration; BIS, Berlin initiative study; LM, Lund-Malmö; CG, Cockcroft-Gault. Equations: MDRD 24 : 175 Creatinine 1.154 age 0.203 0.742 (if female). CKD-EPI 23 : Female 62 µmol/l ( 0.7 mg/dl): 144 (Creatinine/0.7) 0.329 0.993 age. Female >62 µmol/l (>0.7 mg/dl): 144 (Creatinine/0.7) 1.209 0.993 age. Male 80 µmol/l ( 0.9 mg/dl): 141 (Creatinine/0.9) 0.411 0.993 age. Male >80 µmol/l (>0.9 mg/dl): 141 (Creatinine/0.9) 1.209 0.993 age. BIS1 18 3736 Creatinine 0.87 age 0.95 0.82 (if female). Revised Lund-Malmö (LM) 25 X 0.0158 age + 0.438 ln(age) :e Female pcr < 150 µmol/l: X = 2.50 + 0.0121 (150 pcr). Female pcr 150 µmol/l: X = 2.50 0.926 ln(pcr/150). Male pcr < 180 µmol/l: X = 2.56 + 0.00968 (180 pcr). Male pcr 180 µmol/l: X = 2.56 0.926 ln(pcr/180). Cockcroft-Gault (CG) 26 : (140 age) weight/(72 Creatinine) adjusted to 1.73 m 2 body surface area (BSA). a Mean, ± SD, (range). b Median, (range). c Albumin creatinine ratio (ACR) >30 mg/g. d Hb < 12.0 g/dl in women and Hb < 13.0 g/dl in men. ORIGINAL ARTICLE Iohexol sampling time in older adults 1309

ORIGINAL ARTICLE analysis performed in our study laboratory showed values between 1.21 and 1.41 mg/dl. These patients were not excluded from the study. Albuminuria was present in 57% of participants with a median ACR of 41.0 mg/g. With regard to past medical history, 28% had suffered from myocardial infarction, 10% from stroke and 29% from cancer. The mean office systolic and diastolic blood pressure was 133 (±15) and 73 (±9) mmhg, respectively. Mean body mass index (BMI) was 28 (±4.5) kg/m 2,andmeanBSAwas1.9 (±0.2) m 2.Table1 shows a more detailed list of the patients main characteristics. Comparison of 5 versus 24-h iohexol measurement periods Mean iohexol concentration measured at 24 h was 26.5 µg/ml which was considerably higher than 14.17 µg/ml, the concentration predicted by the extrapolation of the slow component from the measurements over the course of 5 h. This result held true for all 104 participants (Figure 1). Consequently, the area under the iohexol elimination curve using an exponential decay model for the slow component was larger when performing the 24-h measurement that resulted in lower mgfr values compared with the 5-h measurement. Both iohexol analyses were highly correlated (r = 0.9). The mean (SD) mgfr of the 5-h analysis (mgfr 300 ) was 29 (±9) ml/ min/1.73 m 2 compared with 23 (±8) ml/min/1.73 m 2 with the 24-h analysis (mgfr 1440 ). Figure 2A shows the Bland Altman plot of both iohexol measurement analyses and displays the raw difference of mgfr 300 minus mgfr 1440 with 5.9 ± 2.6 ml/ min/1.73 m 2 (mean, SD) including 11 ml/min/1.73 m 2 as upper limit and 0.8 ml/min/1.73 m 2 as lower limit of agreement. Accordingly, Figure 2B shows the Bland Altman plot with a mean percentage difference of mgfr 300 minus mgfr 1440 of 29%. Table 2, (1), gives a detailed overview of absolute and percentage bias, precision, r 2 and accuracy of the uncorrected mgfr 1440. The relative frequency of egfr values above/below 30 ml/ min/1.73 m 2 differed by gender: we found egfr > 30 ml/min/ 1.73 m 2 in 71.4% of male and 41.2% of female subjects. This was caused by our inclusion criteria based on serum creatinine level instead of egfr. However, comparing raw and percentage differences between mgfr 300 and mgfr 1440 separately for egfr less than or greater than 30, we did not find large differences between male and female participants: in males, we found a difference of 6.1 ml/min/1.73 m 2 when egfr was 30 and 6.4 ml/min/1.73 m 2 when egfr was >30. In females, we found a difference of 4.5 ml/min/1.73 m 2 when egfr was 30 and 6.2 ml/min/1.73 m 2 when egfr was >30 (data not shown). In individuals with egfr CKD-EPI 30 ml/min/ 1.73 m 2, the absolute difference was smaller than in individuals with egfr CKD-EPI >30mL/min/1.73m 2. However, the percentage difference, which might be of greater relevance, was FIGURE 1: Iohexol plasma concentration over time. Boxes indicate means (line inside box) and quartiles (upper and lower margins of box) of iohexol concentrations at 120, 180, 240, 300 and additionally at 1440 min after injection (antennae are defined by the rule upper-lower box margin ±1.5* inter-quartile range). 1440 predicted from mgfr 300 indicates the mean and quartile of the predicted value for the sample at 1440 min which was extrapolated from the slow component including measurement samples until 300 min. FIGURE 2: (A) Bland Altman plot comparing measured GFR with an overall sampling time of 5 h (mgfr 300 ) and 24 h (mgfr 1440 ). The bias is represented by the solid middle line (5.9 ml/min/1.73 m 2 ). The upper (11 ml/min/1.73 m 2 ) and lower limits (0.8 ml/min/1.73 m 2 ) of the interval of agreement are represented by the dashed lines. Pearson correlation coefficient: r = 0.318, P < 0.001. (B) Bland Altman plot comparing percentage difference of measured GFR with an overall sampling time of 5 h (mgfr 300 ) and 24 h (mgfr 1440 ). The bias is represented by the solid middle line (29%). The upper (65%) and lower limits ( 7%) of the interval of agreement are represented by the dashed lines. Pearson correlation coefficient: r = 0.412, P < 0.001. 1310 N. Ebert et al.

Table 2. Absolute and percentage bias, precision, r 2 and accuracy in the validation sample (n = 52) for mgfr 300 compared with mgfr 1440 and the newly developed mgfr 1440 correction formula Mean absolute bias (ml/min/1.73 m 2 ) SD of differences (ml/min/1.73 m 2 ) Mean percentage a bias ± SD (%) r 2 Standard error of estimate b (ml/min/1.73 m 2 ) (1) uncorrected mgfr 300 5.80 2.63 27.7 ± 14.6 0.93 6.4 63.5 (2) without Intercept (rough estimate) c 0.11 2.31 2.6 ± 11.8 NA 2.3 98.1 (3) with intercept (correction formula) d 0.21 2.27 0.5 ± 11.1 0.93 2.3 98.1 Bias was defined as difference between mgfr 300 and mgfr 1440 as well as the difference of mgfr 1440 and the predicted mgfr 1440 calculated with the rough estimate and the correction formula. Accuracy is expressed as P 30 value, representing the percentage of mgfr 300 or predicted mgfr 1440 values within 30% of the 24 h measurement value (mgfr 1440 ). a Percentage differences (egfr mgfr)/mgfr 100. b Standard error of estimate (1): square root of mean squared differences mgfr 300 mgfr 1440, (2) and (3): residual standard deviation of applying the linear regression equation developed in the learning sample to the data from the validation sample. c Rough estimate: mgfr 1440 = 0.804 mgfr 300. d Correction formula: mgfr 1440 = 2.175 + 0.871 mgfr 300. P 30 (%) Table 3. Absolute difference, percentage difference, correlation and P 30 of mgfr 300 mgfr 1440 (with mgfr 1440 as depending variable) in relation to egfr (CKD-EPI) level ( 30 and > 30 ml/min/1.73 m 2 ) and gender in the total sample (n = 104) Gender GFR level n (%) Absolute difference (ml/min/1.73 m 2 ) a Percentage r 2 P b 30 (%) difference (%) a Male egfr c 30 ml/min/1.73 m 2 20 (28.6) 6.1 ± 3.0 39.1 ± 29.6 0.71 45.0 egfr c >30 ml/min/1.73 m 2 50 (71.4) 6.4 ± 2.4 25.1 ± 11.8 0.90 68.0 Female egfr c 30 ml/min/1.73 m 2 20 (58.8) 4.5 ± 1.8 31.1 ± 15.6 0.90 60.0 egfr c >30 ml/min/1.73 m 2 14 (41.2) 6.2 ± 3.3 22.6 ± 8.2 0.87 92.9 CKD-EPI 23 : Female 62 µmol/l ( 0.7 mg/dl): 144 (Creatinine/0.7) 0.329 0.993 age. Female >62 µmol/l (>0.7 mg/dl): 144 (Creatinine/0.7) 1.209 0.993 age. Male 80 µmol/l ( 0.9 mg/dl): 141 (Creatinine/0.9) 0.411 0.993 age. Male >80 µmol/l (>0.9 mg/dl): 141 (Creatinine/0.9) 1.209 0.993 age. a Mean ± Standard deviation (SD). b Accuracy is expressed as P 30 value, representing the percentage of mgfr 300 within 30% of the 24 h measurement value (mgfr 1440 ). c Calculated with the CKD-EPI equation [23]. as high as 35% in individuals with egfr CKD-EPI 30 ml/min/ 1.73 m 2 compared with 25% in individuals with higher egfr CKD-EPI. Further details can be found in Table 3. Development of the correction formula and the rough estimate The correction formula for predicting mgfr 1440 values from mgfr 300 using linear regression analysis was: mgfr 1440 ¼ 2:175 þ 0:871 mgfr 300 The rough estimate excluding the intercept as described in the Materials and Methods section was mgfr1440 ¼ 0:804 mgfr 300 : This rough estimate (2) provides a simple way to predict mgfr 1440 from mgfr 300 in the clinical setting where it may be sufficient to reduce mgfr 300 values by 20% to obtain an acceptable estimate of mgfr 1440. (11 55 ml/min/1.73 m 2 ). In Appendix Figure A1, we show both prediction of mgfr 1440 from mgfr 300 by linear regression and the rough estimate. Moreover, including a quadratic term for mgfr 300 in the linear regression equation did not change the prediction of mgfr 1440 relevantly. Bias, precision and accuracy of five GFR estimating equations with mgfr 300 and mgfr 1440 in the study population Table 4 shows the metrics of performance of five currently most used egfr equations. With regard to absolute difference of egfr with mgfr 300, the MDRD equation showed the highest bias, whereas the Revised Lund-Malmö (Rev LM) showed the smallest bias, the same held true for the difference between egfr and mgfr 1440. With regard to accuracy, also the Rev LM equation showed the best results in relation to mgfr 300 with 42 and 83% for P 10 and P 30, respectively, and in relation to mgfr 1440 with 26 and 55% for P 10 and P 30, respectively. ORIGINAL ARTICLE Bias, precision and accuracy of the correction formula and the rough estimate For the correction formula, the r 2 was 0.93, the standard error of estimate (residual standard deviation) was 2.26 ml/min/1.73 m 2 and the P 30 value was 98% in the validation sample [Table 2, (3)]. For the rough estimate (reduction of mgfr 300 values by 20%), the standard error of estimate is 2.29 ml/min/1.73 m 2, which is close to the one of the correction formula [Table 2, (2)]. The rough estimate revealed an acceptable prediction of mgfr 1440 in the range of mgfr 300 values of our sample DISCUSSION Measuring instead of estimating GFR in general is time consuming and invasive. Performing mgfr in patients with CKD is an even greater endeavour as it is said to require a 24-h measurement to obtain a more accurate GFR assessment [13, 16]. The practical issues of prolonged measurements again become more challenging in older adults. Several approaches of iohexol measurement have been applied for assessing GFR in elderly populations which often use different overall sampling Iohexol sampling time in older adults 1311

Table 4. Absolute difference, percentage difference, P 10 and P 30 of mgfr 300 and mgfr 1440 in relation to egfr results of five GFR equations in the total sample (n = 104) egfr mgfr Absolute difference (ml/min/1.73 m 2 ) a Percentage P 10 (%) b P 30 (%) b difference (%) a MDRD 300 4.7 ± 6.7 18.9 ± 27.6 30.8 71.2 1440 10.7 ± 6.3 52.0 ± 38.3 5.8 29.8 CKD-EPI 300 3.5 ± 6.7 14.3 ± 27.4 36.5 77.9 1440 9.4 ± 6.2 45.8 ± 36.6 6.7 38.5 BIS1 300 4.1 ± 6.2 19.1 ± 28.1 30.8 76.0 1440 10.1 ± 5.3 52.3 ± 38.2 8.7 30.8 Revised LM 300 0.07 ± 6.4 1.9 ± 24.3 42.3 82.7 1440 5.9 ± 5.7 30.0 ± 32.0 26.0 54.8 CG adjusted for BSA 300 3.3 ± 7.2 15.0 ± 30.2 29.8 70.2 1440 9.2 ± 6.4 46.6 ± 40.3 9.6 38.5 ORIGINAL ARTICLE Bias was defined as difference between estimated GFR (egfr) minus measured GFR (mgfr) for each equation. Percentage differences were defined as (egfr minus mgfr)/mgfr 100. Mean and standard deviation refer to these differences. P 30 and P 10 refer to percentage differences (egfr minus mgfr)/mgfr 100. P 30 is the number of cases with percentage difference at most 30%; P 10 is the number of cases with percentage difference at most 10%. MDRD, modification of diet in renal diseases; CKD-EPI, chronic kidney disease epidemiology collaboration; BIS, Berlin initiative study; LM, Lund-Malmö; CG, Cockcroft-Gault; BSA, body surface area. Equations: MDRD 24 : 175 Creatinine 1.154 age 0.203 0.742 (if female). CKD-EPI 23 : Female 62 µmol/l ( 0.7 mg/dl): 144 (Creatinine/0.7) 0.329 0.993 age. Female >62 µmol/l (>0.7 mg/dl): 144 (Creatinine/0.7) 1.209 0.993 age. Male 80 µmol/l ( 0.9 mg/dl): 141 (Creatinine/0.9) 0.411 0.993 age. Male >80 µmol/l (>0.9 mg/dl): 141 (Creatinine/0.9) 1.209 0.993 age. BIS1 18 3736 Creatinine 0.87 age 0.95 0.82 (if female). Revised Lund-Malmö (LM) 25 X 0.0158 age + 0.438 ln(age) :e Female pcr < 150 µmol/l: X =2.50 + 0.0121 (150 pcr). Female pcr 150 µmol/l: X =2.50 0.926 ln(pcr/150). Male pcr < 180 µmol/l: X =2.56 + 0.00968 (180 pcr). Male pcr 180 µmol/l: X =2.56 0.926 ln(pcr/180). Cockcroft-Gault (CG) 26 : (140 age) weight/(72 Creatinine). a Mean ± Standard deviation (SD). b Accuracy is expressed as P 10 and P 30 values, representing the percentage of egfr values within 10 or 30% of either mgfr 300 or mgfr 1440. times [27 30] leading to diverging methodological approximations of the so called true GFR. In most of these studies, estimates from GFR estimating equations are compared with mgfr results to validate their performance in older adults. However, whether mgfr results of these studies can simply be compared remains questionable. In the present study of 104 elderly CKD patients, we showed that measuring GFR with iohexol plasma clearance leads to a persistent bias depending on the overall sampling time in older adults: in all study participants, GFR values measured over 24 h after injection (mgfr 1440 ) were lower than values measured over only 5 h (mgfr 300 ). This difference was clinically relevant with 6 ml/min/1.73m 2 corresponding to a mean percentage difference of 29%. In those individuals with egfr 30 ml/min/1.73, the mean percentage difference was even higher with 35%, underlining that prolonged sampling time is particularly important in patients with severely decreased kidney function [13]. Our findings from the iohexol clearance measurement in 104 males and females are in agreement with results from Agarwal et al. [15] who measured iothalamate plasma clearance in a small sample of 12 middle-aged men collecting 17 plasma samples over the course of 10 h to evaluate the optimal duration of measurement in CKD patients. They showed that multiple sample measurements over the course of 10 h may improve precision and provide better measures of kidney function compared with shorter sampling time. They also demonstrated that sampling time over 5 h led to an overestimation of measured iothalamate clearance compared with 10 h. This overestimation was also more pronounced in patients with GFR <30 ml/min/1.73 m 2, although this number was limited to six patients. Similar findings have been described in younger kidney transplant recipients where GFR was overestimated when iohexol plasma clearance was measured over thecourseofonly4or6hcomparedwith24h[16]. Since a sample 10 h after injection of the exogenous renal marker is hardly feasible in outpatients, we chose a 24-h sampling period which was well accepted by our study participants. Our results from 24-h iohexol measurement emphasize our assumption that regardless of the exogenous marker, prolonged clearance measurement in CKD patients leads to a more precise approximation of the true GFR. There are clinical scenarios, however, especially in older patients (frailty, long distance to hospital), where measuring iohexol clearance over the course of 24 h may not be feasible. In such cases to predict a more precise mgfr 1440 value from mgfr 300, we developed a new correction formula to be applied in elderly CKD patients only undergoing the short (5-h) iohexol measurement for exact assessment of their kidney function. Additionally, we compared the performance of five GFR estimating equations to mgfr 300 and mgfr 1440 and demonstrated that with regard to bias, precision and accuracy the Revised Lund-Malmö equation showed the best results for both mgfr measurement protocols, leading to the recommendation that this equation gives the best and less biased estimate of GFR in elderly patients with CKD. The remaining four equations showed rather similar bias and commonly overestimate 1312 N. Ebert et al.

mgfr by up to 11 ml/min bearing the risk of drug over-dosing, which is a very critical issue in geriatric patients. Also, accuracy, very important for correctly identifying the degree of kidney failure, was clearly superior for the Revised Lund-Malmö equation, although we believe that these results should be re-evaluated in a larger sample with a wider range of GFR. Although we do not have longitudinal GFR measurements in our cohort and therefore cannot investigate the ability of egfr equations to predict GFR slopes over time, we would like to point out that egfr slope prediction has been shown to be highly inaccurate in diabetic patients [31]. This should be an additional aspect for future research as it is of high clinical relevance in the setting of clinical studies or trials with regard to frail populations. This is the first study comparing plasma clearance sampling times in a reasonably large population of old CKD patients with a mean age of almost 80 years. The main finding of our analysis is that a measurement protocol with a sampling time of 5 h after iohexol injection leads to mgfr results that overestimate GFR values derived from a sampling time of 24 h especially relevant in elderly patients with decreased kidney function. We subsequently showed that up to a precision of about ±2.3 ml/min/ 1.73 m 2 (1-fold standard error of the estimate), the more accurate mgfr 1440 value can be estimated from the shorter more feasible mgfr 300 measurement using the newly developed correction formula. Some limitations of our analysis deserve mention. Firstly, we exclusively measured iohexol plasma clearance in our study population and did not simultaneously assess urinary clearance to validate our results. However, we doubt that urinary clearance measurement would have been the ideal gold standard procedure because of its problematic and error-prone implementation especially relevant in elderly individuals where timed urination is, in the majority of cases, almost impossible and willingness to participate in a clinical study declines dramatically when exact urine collection or even catheterization is involved. Also, in the context of albuminuria assessment, the precision of urine collection has been questioned [32], which is the reason why Kidney Disease: Improving Global Outcomes guidelines recommend albumin creatinine ratio (ACR) out of spot urine instead of 24-h urine collection. At present, it is notable that most other GFR study groups that perform GFR gold standard measurements in older adults also choose the plasma clearance method, most probably for the same reason [31, 33, 34]. Second, it should be noted that part of the negative difference between mgfr 300 and mgfr 1440 may also be due to diurnal variations of GFR [35]. Also Sirota and colleagues [36] observed a slight depression of glomerular activity during deep sleep in the early morning hours. Therefore, nocturnal decrease of glomerular filtration may contribute to a lower overall GFR result when measured over 24 h with the night at the end of this period. In this context, it is important to note that in the present study, all clearance measurements were started between 8:00 and 10:30 a.m. Third, in the course of the study each participant underwent iohexol clearance measurement only once; therefore, we are not able to analyse possible intra-individual variation in iohexol clearance. However, since GFR overestimation was found in literally every patient, we doubt that this would have changed our results considerably. Data by Gaspari et al. [1] showed that the variation of iohexol clearance after repeated measurements was 5.5%. In relation to the mean percentage difference between mgfr 300 and mgfr 1440 of 29% in our population, we still show a difference of at least 18% when taking into account possible intra-individual variation. Still, we believe that this percentage is a rather conservative approach and may be considerably lower with a precise study protocol, where measurements were started at the same time of day and performed under strict dietary and drug (no NSAIDs) control, with precise plasma sample timing and immediate iohexol analysis as well as constant quality control of analysis methods and their external validation, like it was done in our study. Finally, the new correction formula has not been externally validated yet. In conclusion, we demonstrate that in elderly CKD patients, measurement of iohexol clearance up to 5 h leads to a clinically relevant overestimation of GFR compared with 24-h measurement. With deteriorating kidney function within our elderly population, this overestimation becomes even more apparent. The newly developed correction formula can account for measurement variability and increases accuracy (defined as P 10 and P 30 value) of predicting mgfr 1440 from mgfr 300 by 34%. Physicians and researchers should appreciate that there is not only inter gold standard variation due to different exogenous markers [7] but also a considerable intra gold standard variation depending on the period of measurement time. In the era where standardization of the renal biomarkers creatinine and cystatin C is recognized as one of the key elements for highquality GFR estimation, raising awareness of the importance of standardized GFR measurement protocols is past due. ACKNOWLEDGEMENTS The authors thank their colleagues Sima Canaan-Kühl, Julia Lepenies, Edith Fuchs-Lösment, Torsten Mittelstädt, Sylvia Petersen, Wolfram Jabs, Helen Hepburn, Til Leimbach, and Denise Markmann for their support in recruiting elderly CKD patients for the study as well as Katharina Kuschfeldt for her excellent technical assistance. We are indebted to the study participants of the Berlin Initiative Study (BIS). Preliminary data were partly presented at American Society of Nephrology s Kidney Week 2013 (TH-PO269), Atlanta, USA. FUNDING The Berlin Initiative Study (BIS) is funded by the Kuratorium für Dialyse und Nierentransplantation (KfH) Foundation of Preventive Medicine. CONFLICT OF INTEREST STATEMENT None declared. (See related article by Hulter and Krapf. Measured glomerular filtration rate is the goal, but how to measure it? Nephrol Dial Transplant 2015; 30: 1231 1233.) ORIGINAL ARTICLE Iohexol sampling time in older adults 1313

ORIGINAL ARTICLE REFERENCES 1. Gaspari F, Perico N, Remuzzi G. Application of newer clearance techniques for the determination of glomerular filtration rate. Curr Opin Nephrol Hypertens 1998; 7: 675 680 2. van Westen D, Almen T, Chai CM et al. Biliary and total extrarenal clearance of inulin and iohexol in pigs. A source of error when determining gfr as body clearance. Nephron 2002; 91: 300 307 3. Olsson B, Aulie A, Sveen K et al. Human pharmacokinetics of iohexol. A new nonionic contrast medium. Invest Radiol 1983; 18: 177 182 4. Frennby B, Sterner G, Almen T et al. Determination of low glomerular filtration rate using iohexol clearance. Invest Radiol 1994; 29 (Suppl. 2): S234 S235 5. Aakhus T, Sommerfelt SC, Stormorken H et al. Tolerance and excretion of iohexol after intravenous injection in healthy volunteers. Preliminary report. Acta Radiol Suppl 1980; 362: 131 134 6. Back SE, Krutzen E, Nilsson-Ehle P. Contrast media as markers for glomerular filtration: a pharmacokinetic comparison of four agents. Scand J Clin Lab Invest 1988; 48: 247 253 7. Soveri I, Berg UB, Bjork J et al. Measuring GFR: a systematic review. Am J Kidney Dis 2014; 64: 411 424 8. Bird NJ, Peters C, Michell AR et al. Comparison of GFR measurements assessed from single versus multiple samples. Am J Kidney Dis 2009; 54: 278 288 9. Schwartz GJ, Furth S, Cole SR et al. Glomerular filtration rate via plasma iohexol disappearance: pilot study for chronic kidney disease in children. Kidney Int 2006; 69: 2070 2077 10. Gaspari F, Perico N, Ruggenenti P et al. Plasma clearance of nonradioactive iohexol as a measure of glomerular filtration rate. J Am Soc Nephrol 1995; 6: 257 263 11. Brochner-Mortensen J. A simple method for the determination of glomerular filtration rate. Scand J Clin Lab Invest 1972; 30: 271 274 12. Ng DK, Schwartz GJ, Jacobson LP et al. Universal GFR determination based on two time points during plasma iohexol disappearance. Kidney Int 2011; 80: 423 430 13. Nilsson-Ehle P. Iohexol clearance for the determination of glomerular filtration rate: 15 years experience in clinical practice. ejifcc 2002; 13. http// www.ifcc.org/ejifcc/vol13no12/1301200105.htm 14. Sterner G, Frennby B, Hultberg B et al. Iohexol clearance for GFR-determination in renal failure--single or multiple plasma sampling? Nephrol Dial Transplant 1996; 11: 521 525 15. Agarwal R, Bills JE, Yigazu PM et al. Assessment of iothalamate plasma clearance: duration of study affects quality of GFR. Clin J Am Soc Nephrol 2009; 4: 77 85 16. Stolz A, Hoizey G, Toupance O et al. Evaluation of sample bias for measuring plasma iohexol clearance in kidney transplantation. Transplantation 2010; 89: 440 445 17. Jacobsson L. A method for the calculation of renal clearance based on a single plasma sample. Clin Physiol 1983; 3: 297 305 18. Schaeffner ES, Ebert N, Delanaye P et al. Two novel equations to estimate kidney function in persons aged 70 years or older. Ann Intern Med 2012; 157: 471 481 19. Schaeffner ES, van der Giet M, Gaedeke J et al. The Berlin initiative study: the methodology of exploring kidney function in the elderly by combining a longitudinal and cross-sectional approach. Eur J Epidemiol 2010; 25: 203 210 20. Cavalier E, Rozet E, Dubois N et al. Performance of iohexol determination in serum and urine by HPLC: validation, risk and uncertainty assessment. Clin Chim Acta 2008; 396: 80 85 21. Du Bois D, Du Bois E. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med 1916; 17: 863 871 22. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307 310 23. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: 604 612 24. Levey AS, Bosch JP, Lewis JB et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of diet in renal disease study group. Ann Intern Med 1999; 130: 461 470 25. Bjork J, Grubb A, Sterner G et al. Revised equations for estimating glomerular filtration rate based on the Lund-Malmo Study cohort. Scand J Clin Lab Invest 2011; 71: 232 239 26. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976; 16: 31 41 27. Kilbride HS, Stevens PE, Eaglestone G et al. Accuracy of the MDRD (Modification of Diet in Renal Disease) Study and CKD-EPI (CKD Epidemiology Collaboration) equations for estimation of GFR in the elderly. Am J Kidney Dis 2012; 61: 57 66 28. Van Pottelbergh G, Hoste L, Degryse J et al. Measuring the glomerular filtration rate in different age groups using iohexol, the protocol from the Belgian iohexol study. Clin Biochem 2013; 46: 31 36 29. Nyman U, Grubb A, Larsson A et al. The revised Lund-Malmo GFR estimating equation outperforms MDRD and CKD-EPI across GFR, age and BMI intervals in a large Swedish population. Clin Chem Lab Med 2013; 1 10 30. Lopes MB, Araujo LQ, Passos MT et al. Estimation of glomerular filtration rate from serum creatinine and cystatin C in octogenarians and nonagenarians. BMC Nephrol 2013; 14: 265 31. 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 32. Lambers Heerspink HJ, Gansevoort RT, Brenner BM et al. Comparison of different measures of urinary protein excretion for prediction of renal events. J Am Soc Nephrol 2010; 21: 1355 1360 33. Fan L, Levey AS, Gudnason V et al. Comparing GFR Estimating Equations Using Cystatin C and Creatinine in Elderly Individuals. J Am Soc Nephrol 2014; doi: 10.1681/ASN.2014060607 34. Alshaer IM, Kilbride HS, Stevens PE et al. External validation of the Berlin equations for estimation of GFR in the elderly. Am J Kidney Dis 2014; 63: 862 865 35. Van Acker BA, Koomen GC, Arisz L. Drawbacks of the constant-infusion technique for measurement of renal function. Am J Physiol 1995; 268: F543 F552 36. Sirota JH, Baldwin DS, Villarreal H. Diurnal variations of renal function in man. J Clin Invest 1950; 29: 187 192 APPENDIX Figure A1. FIGURE A1: Correction formula (line with crosses) and rough estimate (line with circles) for estimation of mgfr 1440 by mgfr 300. Including a quadratic term for mgfr 300 in the linear regression equation did not change the prediction of mgfr 1440 relevantly. Received for publication: 19.12.2014; Accepted in revised form: 25.3.2015 1314 N. Ebert et al.