Estimating GFR using serum beta trace protein: accuracy and validation in kidney transplant and pediatric populations

original article

http://www.kidney-international.org & 2009 International Society of Nephrology

Estimating GFR using serum beta trace protein: accuracy and validation in kidney transplant and pediatric populations Christine A. White1, Ayub Akbari2,3, Steve Doucette4, Dean Fergusson4, Naser Hussain2, Laurent Dinh5, Guido Filler6, Nathalie Lepage7 and Greg A. Knoll2,3,4 1

Division of Nephrology, Department of Medicine, Queen’s University, Kingston, Ontario, Canada; 2Division of Nephrology, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada; 3Kidney Research Centre, The Ottawa Health Research Institute, Ottawa, Canada; 4Clinical Epidemiology Program, The Ottawa Health Research Institute, Ottawa, Ontario, Canada; 5Division of Nuclear Medicine, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada; 6Division of Nephrology, Department of Pediatrics, University of Western Ontario, London, Ontario, Canada and 7Department of Laboratory Medicine, Children’s Hospital of Eastern Ontario and the University of Ottawa, Ottawa, Ontario, Canada

The limitations of estimates of glomerular filtration rate (GFR) based only on serum creatinine measurements have spurred an interest in more sensitive markers of GFR. Beta-trace protein (BTP), a low-molecular-weight glycoprotein freely filtered through the glomerular basement membrane and with minimal non-renal elimination, may be such a marker. We have recently derived two GFR estimation equations based on BTP. To validate these equations, we measured BTP and the plasma clearance of 99 mTc-DTPA in 92 adult kidney transplant recipients and 54 pediatric patients with impaired kidney function. GFR was estimated using the serum creatinine–based Modification of Diet in Renal Disease (MDRD) Study equation for adults, the Schwartz and updated Schwartz equations in children, and 4 novel BTP-derived equations (our 2 equations and 2 proposed by Poge). In adults, our BTP-based equations had low median bias and high accuracy such that 89–90% of estimates were within 30% of measured GFR. In children, the median bias of our 2 equations was low and accuracy was high such that 78–83% of estimates were within 30% of measured GFR. These results were an improvement compared to the MDRD and Schwartz equations, both of which had high median bias and reduced accuracy. The updated Schwartz equation also performed well. Kidney International (2009) 76, 784–791; doi:10.1038/ki.2009.262; published online 22 July 2009 KEYWORDS: beta trace protein; glomerular filtration rate; kidney transplantation; pediatric transplantation

Correspondence: Greg A. Knoll, Division of Nephrology, The Ottawa Hospital, Riverside Campus, 1967 Riverside Drive, Ottawa, Ontario, Canada K1H 7W9. E-mail: [email protected] Received 28 March 2008; revised 21 April 2009; accepted 2 June 2009; published online 22 July 2009 784

Serum creatinine is a crude marker of glomerular filtration rate (GFR) with several well-described limitations.1,2 It can be noted that its serum concentration is dependent on various non-renal factors, such as muscle mass and turnover, medication use, diet, and non-renal elimination.1 GFR estimation equations derived from serum creatinine have repeatedly been shown to be inaccurate, particularly in patient groups that are distinct from those in whom the equations were derived such as kidney transplant recipients,3–5 patients with abnormal body composition,6 liver dysfunction,7 or relatively well-preserved kidney function.8,9 This is presumably due to the influence of the myriad of factors that affect creatinine production and non-glomerular excretion, which are not adequately ‘corrected for’ in the estimation equations. In the pediatric population, studies have also documented the poor performance of creatininebased GFR estimates.10–12 An ‘updated’ Schwartz equation based on serum creatinine has recently been described, but has not yet been validated in an independent pediatric population.13 Beta-trace protein (BTP) has emerged as an alternate endogenous marker of GFR.14 It is a low-molecular-weight glycoprotein with 168 amino acids that is filtered through the glomerular basement membrane.15 It appears to have minimal non-renal elimination.16 Since its first description as a marker of impaired renal function in 1997,17 BTP has been shown to be a more sensitive marker of GFR than creatinine in patients with chronic kidney disease,18–21 in kidney transplant recipients,22 and in children.19 Until recently, the absence of an equation to convert its serum concentration into an estimate of GFR has limited the clinical utility of BTP. We have recently proposed two GFR estimation equations based on serum BTP (Table 1).23 These were derived from a cohort of 163 kidney transplant recipients (mean age: 53±12 years, 67% men, 90% white), with a mean GFR of 59±23 ml per min per 1.73 m2. The Kidney International (2009) 76, 784–791

original article

CA White et al.: BTP GFR estimation equation validation

Table 1 | Glomerular filtration rate (GFR) estimation equationsa

Table 2 | Patient characteristicsa

Reference

Formula

Characteristic

Whiteb

GFR1=112.1  BTP0.662  urea0.280  (0.880 if female) GFR2=167.8  BTP0.758  creatinine0.204  (0.871 if female)

Male (n (%)) Age (years)

62 (68) 53±12

32 (59) 10±5

Pogeb

GFR1=89.85  BTP0.5541  urea0.3018 GFR2=974.31  BTP0.2594  creatinine0.647

Race (n (%)) White Black Other

87 (95) 2 (2) 3 (3)

NA NA NA

85.9±22.7 169.3±10.2 1.96±0.26

37.9±22 133.1±31 1.16±0.43

Leveyc

GFR=186  creatinine–1.154  age0.203  (0.742 if female)  (1.21 if black)

Schwartzc,d GFR=(k  length)/creatinine, k=0.7 (boys X13), k=0.45 (infantso1), k=0.55 (all others) GFR=(0.413  length)/creatinine Updated Schwartzc,d a

GFR in ml per min per 1.73 m2. b BTP in mg/l; urea in mmol/l; creatinine in mmol/l. c Creatinine in mg per 100 ml. d Length in cm and creatinine in mg per 100 ml.

purpose of this study was to validate these novel BTP-based equations in two independent patients groups consisting of a cohort of adult kidney transplant recipients and a cohort of children with impaired kidney function. The performance of the two BTP-based GFR estimation equations recently proposed by Poge et al.24 and of the updated Schwartz equation13 was also examined. RESULTS Patient characteristics

The clinical characteristics of the study groups are shown in Table 2. The adult cohort had an average age of 53±13 years and a median GFR of 51 (interquartile range (IQR) of 28) ml per min per 1.73 m2 (range: 11–99). All five stages of chronic kidney disease are represented.25 The pediatric cohort had an average age of 10±5 years and a median GFR of 69 (IQR of 38) ml per min per 1.73 m2 (range: 16–97). Equation performance

The concordance correlation coefficients between the measured GFR and the estimation equation GFR are shown in Table 3. In adults, these were the strongest for the White BTP equations. In children, concordance was strongest for the updated Schwartz equation followed by the two White BTP equations. Table 4 shows the performance of the estimation equations. In the adult population, the bias of the White BTP equations was significantly lower than those of the fourvariable Modification of Diet in Renal Disease (MDRD), Poge BTP1, and Poge BTP2 equations (Po0.0001 for all comparisons). They were highly accurate with 89 and 90% of estimates within 30% of the measured GFR for White BTP1 and White BTP2 equations, respectively. In comparison, the four-variable MDRD Study and Poge equations had higher negative bias and had lower accuracy with only 76, 76 and 65% of estimates within 30% of the measured GFR, respectively. The 30% accuracy of the White BTP1 equation was significantly higher than that of the Poge BTP1 Kidney International (2009) 76, 784–791

Adult cohort Pediatric (n=92) cohort (n=54)

Weight (kg) Height (cm) Body surface area (m2) Medication (n (%)) Prednisone Calcineurin inhibitor Sirolimus Mycophenolate mofetil Azathioprine

85 88 4 63 17

Serum creatinine (mmol/l) Serum urea (mmol/l) Serum BTP (mg/l) Median DTPA GFR (ml per min per 1.73 m2) (IQR)

(92) (96) (4) (68) (18)

144±70 10±7 1.29±0.71 51 (28)

CKD stage by measured GFR (n (%)) Stage 1, GFR X90 ml per min per 1.73 m2 Stage 2, GFR 60–89 ml per min per 1.73 m2 Stage 3, GFR 30–59 ml per min per 1.73 m2 Stage 4, GFR 15–29 ml per min per 1.73 m2 Stage 5, GFR o15 ml per min per 1.73 m2

2 26 49 12 3

(2) (28) (53) (13) (3)

NA NA NA NA NA 106±68 9±6 1.67±1.12 69 (38)

8 (15) 25 (46) 12 (22) 9 (17) 0 (0)

BTP, beta-trace protein; CKD, chronic kidney disease; DTPA, diethylenetriaminepentaacetic acid; GFR, glomerular filtration rate; IQR, interquartile range; NA, data not available. a Data expressed as mean±s.d. unless otherwise specified.

Table 3 | Concordance correlation coefficients Equation 4-Variable MDRD study Schwartz Updated Schwartz White BTP1 White BTP2 Poge BTP1 Poge 2 BTP2

Adults (95% CI) 0.676 (0.577, — — 0.829 (0.765, 0.819 (0.751, 0.621 (0.518, 0.511 (0.397,

Pediatrics (95% CI)

0.775)

0.893) 0.886) 0.724) 0.625)

0.681 0.836 0.788 0.779 0.587 0.699

— (0.563, (0.757, (0.676, (0.676, (0.456, (0.571,

0.800) 0.915) 0.881) 0.881) 0.718) 0.827)

BTP, beta-trace protein; CI, confidence interval; MDRD, Modification of Diet in Renal Disease.

(P ¼ 0.012) and Poge BTP2 (Po0.0001) equations. The 30% accuracy of the White BTP2 equation was significantly higher than that of the four-variable MDRD Study equation (P ¼ 0.01), Poge BTP1 (P ¼ 0.007), and Poge BTP2 (Po0.0001) equations. The performance analysis was repeated after subdividing patients by GFR greater (n ¼ 49) or less (n ¼ 43) than the median of 51 ml per min per 1.73 m2 (Table 5). Equation performance varies considerably between the subgroups, particularly for the Poge and MDRD Study equations with substantially increased underestimation (negative bias) of GFR with improved graft function. The White BTP equations 785

original article

CA White et al.: BTP GFR estimation equation validation

Table 4 | Bias, precision, and accuracy of creatinine and BTP estimatesa Accuracy (% within) Median bias

Precision (IQR)

Median % bias

Precision (IQR)

Mean bias

Precision (s.d.)

10%

30%

Adults 4-Variable MDRD White BTP1 White BTP2 Poge BTP1 Poge BTP2

6.0 0.3 0.1 9.3 11.2

14.7 11.9 11.5 12.6 16.8

13.8 0.9 0.4 19.6 21.2

23.7 25.9 23.9 20.5 24.7

9.0 1.5 1.7 11.0 13.3

12.1 10.6 10.5 11.3 12.4

32 36 38 20 18

76 89 90 76 65

Pediatrics Schwartz Updated Schwartz White BTP1 White BTP2 Poge BTP1 Poge BTP2

15.5 6.1 10.3 8.3 19.7 9.0

22.5 14.2 17.5 15.6 22.9 21.1

30.2 9.4 16.6 15.2 28.8 18.6

36.2 26 23.7 23.0 19.9 29.9

18.6 4.9 8.6 7.1 18.4 9.5

17.0 13.1 14.4 16.1 14.2 16.2

17 30 28 24 11 26

49 85 78 83 50 72

BTP, beta-trace protein; IQR, interquartile range; MDRD, Modification of Diet in Renal Disease; 99mTc-DTPA, 99mtechnetium-diethylenetriaminepentaacetic acid. a Bias was defined as the difference between the estimated and the measured (99mTc-DTPA) GFR (estimated GFR – measured GFR); percentage bias was defined as [estimated GFR – measured GFR]/measured GFR  100; precision was defined as the IQR for the median bias and standard deviation of the mean bias; both precision and bias were expressed as ml per min per 1.73 m2; accuracy was defined as the proportion of estimates that were within 10 or 30% of the measured (99mTc-DTPA) GFR.

Table 5 | Bias and 30% accuracy by GFR subgroups (GFR o51 ml per min per 1.73 m2, n=43; GFR 451 ml per min per 1.73 m2, n=49) in the adult kidney transplant cohort Median bias (ml per min per Percentage within 30% 1.73 m2) 4-Variable MDRD study GFRo51 ml per min per 1.73 m2 GFR451 ml per min per 1.73 m2

2.8 24.2

89 65

White BTP1 GFRo51 ml per min per 1.73 m2 GFR451 ml per min per 1.73 m2

2.9 4.9

84 94

White BTP2 GFRo51 ml per min per 1.73 m2 GFR451 ml per min per 1.73 m2

3.6 4.8

84 96

Poge BTP1 GFRo51 ml per min per 1.73 m2 GFR451 ml per min per 1.73 m2

3.6 14.8

81 71

Poge BTP2 GFRo51 ml per min per 1.73 m2 GFR451 ml per min per 1.73 m2

3.4 19.9

84 49

DISCUSSION

BTP, beta-trace protein; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease Study.

show a more consistent performance across the two subgroups. Figure 1 shows the bias of each equation across levels of estimated GFR. In the pediatric group, the White BTP1 equation had significantly lower bias than the Schwartz (Po0.0001) and Poge BTP1 (Po0.0001), but not the Poge BTP2 (P ¼ 0.42), equations (Table 4). Similarly, the White BTP2 equation had significantly lower bias than the Schwartz (Po0.0001) and Poge BTP1 (Po0.0001), but not the Poge BTP2 (P ¼ 0.15), equations. There were no significant differences in bias between the White BTP1 and BTP2 equations and the 786

updated Schwartz equations (P ¼ 0.32, P ¼ 0.79). The accuracy of the White BTP equations was high with 78 and 83% of estimates within 30% of measured GFR for White BTP1 and White BTP2 equations, respectively. In contrast, only 48% of the Schwartz equation estimates and 50 and 72% of the Poge equation estimates were within 30% of the measured GFR. The updated Schwartz equation also showed high accuracy with 85% of estimates within 30% of the measured GFR. There were no significant differences in accuracies between the White BTP, updated Schwartz, and Poge BTP2 equations.

This study reveals that the novel White BTP-based GFR estimation equations are accurate in both an adult and a pediatric population and shows improved estimation performance compared with the standard creatinine-based GFR estimation equations. To date, BTP has been studied to a limited extent in the pediatric population. Filler et al.19 report that the reciprocal of BTP had a significantly higher correlation with GFR than that of serum creatinine, and that the diagnostic accuracy of BTP was improved over creatinine for the detection of a GFR o90 ml per min per 1.73 m2. The poor performance of the Schwartz equation in this study confirms previous reports of reduced accuracy in pediatric patients with impaired renal function.10,11 Seikaly et al.11 showed that the Schwartz equation overestimated GFR by 90% in children with a GFR o50 ml per min per 1.73 m2. Grubb et al.10 report a mean percent bias of 51%, and only 25% of estimates were within 30% of measured GFR using the Schwartz equation in a cohort of 85 children. Similar to our findings, in a recent study Schwartz et al.12 found that the equation overestimated the measured GFR by, on average, 12.2 ml per min per 1.73 m2 and postulated that this can been attributed to Kidney International (2009) 76, 784–791

original article

30

30

20

20

10

Poge BTP1 bias

White BTP1 bias

CA White et al.: BTP GFR estimation equation validation

0 –10 –20 –30

0 –10 –20 –30 –40

–40

–50

–50 0

20

40 60 80 100 White BTP1 estimated GFR

0

120

20

40

60

80

100

120

100

120

Poge BTP1 estimated GFR

30

30

20

20

10

10

Poge BTP2 bias

White BTP2 bias

10

0 –10 –20 –30 –40

0 –10 –20 –30 –40

–50

–50 0

20

40

60

80

100

120

0

20

60

80

4-Variable MDRD study bias

White BTP2 estimated GFR

40 60 80 Poge BTP2 estimated GFR

30 20 10 0 –10 –20 –30 –40 –50 0

20

40

100

120

4-Variable MDRD study estimated GFR

Figure 1 | Estimation equation bias according to level of estimated GFR in ml per min per 1.73 m2. Bias was calculated by subtracting the measured GFR from the estimated GFR and is expressed in ml per min per 1.73 m2. The median bias is represented by the solid line and the first and third quartiles by the dashed lines. BTP, beta-trace protein; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease Study.

differences in creatinine assays. The original Schwartz equation was derived using creatinine measured using a Jaffe reaction, whereas the creatinine measured in the current and recent Schwartz studies were measured using enzymatic methods that yield lower values.26 Recently, a novel creatinine-based estimation equation derived using enzymatic methods was proposed by Schwartz et al.13 This report is the first to validate it in an independent cohort of patients. Creatinine was measured using enzymatic methodologies, and the results show significantly improved performance compared with the traditional Schwartz equation. The performance of the BTP equations in the pediatric population is also encouraging and warrants further examination in other pediatric populations with varying degrees of impaired kidney function. There has been limited data reported on the use of BTP in kidney transplant recipients. Poge et al.22 showed that at a given GFR, BTP had a greater increase above the upper reference value compared with creatinine, although the diagnostic performance was similar by receiver operating characteristic analysis. In a recent publication, Poge et al.24 report on the derivation and validation of BTP-based GFR estimation equations in kidney transplant recipients. They propose two equations, one that included serum urea and the Kidney International (2009) 76, 784–791

other serum creatinine. Similar to our work, the equation containing BTP and urea (Poge BTP1) had a higher R2 and improved performance when validated than that containing BTP and creatinine (Poge BTP2). Furthermore, the Poge BTP1 equation had a bias of only 0.43 ml per min per 1.73 m2 compared with the White BTP1 equation, which overestimated the measured GFR by 9.3 ml per min per 1.73 m2. In this study, the Poge equation underestimated the measured GFR by 11 ml per min per 1.73 m2 whereas the White BTP1 equation only underestimated it by 1.2 ml per min per 1.73 m2. Thus, in both studies, the White BTP1 equation provided estimates that are on average B10 ml per min per 1.73 m2 higher than those calculated using the Poge equation. Both equations were derived using plasma diethylenetriaminepentaacetic acid (DTPA) clearance as the reference standard, and both include BTP and urea. The White equation also contains a gender-correcting factor that did not improve the model fit of the Poge equation.24 It is interesting to note that the model fit using BTP alone was substantially higher in our derivation cohort (R2 ¼ 0.756) than that in the Poge study (R2 ¼ 0.651).23,24 The final model of White equation also had a higher R2 (0.81) compared with the Poge BTP1 equation (0.714).23,24 Even without gender, the White model had an R2 of 0.791, and thus more of the 787

original article

variability in GFR could be explained by BTP and urea in the White model as opposed to the Poge model. The differing results of this current analysis and the Poge study24 can be attributed to several factors. First, differences in populations may also be contributing to between study variance. The adult study populations are similar in gender, race, and age, but the derivation and validation cohorts of Poge et al. have a considerably lower average GFR. The performance of creatinine-based GFR estimation equations is dependent on GFR level in both kidney transplant27,28 and non-transplant populations.8,29 This is presumably due to differences in creatinine production, renal handling of creatinine, and non-renal creatinine elimination at different levels of GFR. Similar to creatinine, the relationship between urea and GFR is confounded by other influences such as volume status,30 cyclosporine use,31 steroid usage,30 liver disease,30 and nutritional state.30 Poge et al.24 examined the difference in performance of the BTP-based estimation equations in subgroups above and below 50 ml per min per 1.73 m2. They found a greater negative bias (5.81 ml per min per 1.73 m2) in the higher GFR subgroup compared with a positive bias of 3.05 ml per min per 1.73 m2 in the lower GFR subgroup for the Poge BTP1 equation. They did not, however, find a significantly different bias for the White BTP1 equation in the two subgroups, although the accuracy did improve in the higher GFR group.24 In our GFR subgroup analysis (4 and o51 ml per min per 1.73 m2) (Table 5), a striking difference in equation performance is evident, particularly for the Poge and MDRD Study equations that is similar to that observed by Poge. Overall, the Poge and MDRD Study equations show improved performance in the lower GFR subgroup, which more closely approximates the GFR of the equations’ derivation and validation cohorts. Second, differences in calibration of the analyte (urea, BTP, and creatinine) assays likely have an important role. The effect of non-standardized creatinine calibration on creatinine-based GFR estimation equations has been well described, and it has led to efforts to internationally harmonize creatinine measurement using assays traceable to the gold standard isotope dilution mass spectrometry (IDMS).32,33 Such efforts have not been attempted for urea, which can be measured using a number of methods based on different analytical principals.34 Details regarding the measurement of urea are not provided by Poge et al.24 Both studies used a Dade Behring immunonephelometric assay to measure BTP, but the platforms on which they were measured were different. To date, differences in BTP assay calibration have not been examined. We modeled the effect of hypothetical systematic changes in urea and BTP assays in our laboratory on the bias of the White BTP1 equation in our adult cohort (results not shown). White BTP1 estimates using 20% higher BTP and urea values are very similar to those obtained using the Poge BTP1 equation and the actual BTP and urea values. Poge et al. did not evaluate the White BTP2 equation. Both studies used assays referable to an IDMS creatinine in the 788

CA White et al.: BTP GFR estimation equation validation

evaluation of the Poge BTP2 equation. Therefore, in theory, the disparate Poge BTP2 estimates cannot be explained by differences in creatinine assay calibration. Third, differences in gold standard GFR technique may also be contributing to the noted study differences. Although both used plasma clearance of DTPA after a single injection, the sampling strategy differed. Poge et al.24 sampled at 1 and 3 h, while we sampled at 2, 3 and 4 h. This may have an impact on GFR measurement, particularly in patients with significantly reduced kidney function where delayed sampling has been shown to improve correlation between tracer plasma and urinary clearance.35,36 The poor performance of the MDRD equation has been well documented elsewhere in renal transplant recipients37 as well as in other patients who differ from the cohort in whom the equation was derived.8,9 It can be noted that the equations significantly underestimate the GFR in patients with well-preserved kidney8 or graft function,37 and modifications to the original equations to account for this are being developed. Despite higher average GFR values, the pediatric cohort had a higher mean BTP concentration compared with the adult kidney transplant patients. The independent effect of age on serum BTP levels has not been fully explored. Filler et al.19 did not find an association between BTP and age in a cohort of pediatric patients with various renal diseases, although there was no correction for GFR and all participants were children. Our data may suggest that age does independently effect on BTP concentrations, but this requires further study. Immunosuppressant medications may also influence BTP concentration, which could also explain the differences observed between the two populations. This needs further exploration in a larger cohort of patients. It is also possible that more significant overestimation of GFR occurred in the pediatric age group by virtue of their increased susceptibility to GFR overestimation, with more impaired GFRs when delayed samples (44 h) are not performed. It is unlikely that other differences in measurement techniques explain this finding given that BTP was measured in the same laboratory using the same assay. In addition, the GFR was measured using identical protocols and tracers, although at different institutions. Strengths of this study include the measurement of GFR and all analytes on the same day, and the utilization of serum creatinine values was calibrated to the Cleveland Clinic for the analyses using the MDRD Study equation and of IDMS traceable creatinine values for the analyses using the Poge BTP2 equation. Limitations should be noted. First, the adult validation population is very similar in terms of demographics and GFR range to the White equations’ derivation cohort. This limits the ability to extrapolate our findings to a less homogeneous population. Second, the population was largely white and the performance of the equations in nonwhite patients cannot be inferred. In addition, we did not have information on race in our pediatric patients. Third, both validation cohorts were relatively small in size, which Kidney International (2009) 76, 784–791

CA White et al.: BTP GFR estimation equation validation

limits the precision of our estimates. Fourth, BTP was measured only once, which may impact our findings, as the intra-patient variability of BTP has not yet been established. Fifth, the reproducibility of BTP remains unknown. Sixth, delayed sampling beyond 4 h was not carried out, which may have led to overestimation of the true GFR in patients with significantly reduced kidney function. Further studies using delayed sampling are required. Finally, BTP is more expensive and is not as commonly available as creatinine, which reduces its appeal as a marker of GFR at this time. In conclusion, this validation study has found that the novel White BTP-based GFR estimation equations are substantially more accurate in adult kidney transplant recipients than accepted creatinine-based GFR estimation equations. The results are also very encouraging in a smaller single-center pediatric population with impaired kidney function Discrepancies in accuracy and precision remain between BTP equations developed at different institutions, which may reflect differences in urea and BTP assay calibration between laboratories, differences in reference standard GFR measurements, and differences in patient populations. More widespread adoption of these equations will require further validation in larger and more diverse patient groups. METHODS AND MATERIALS Study patients Data for the adult group were obtained from patients recruited to two studies. To study novel markers of GFR, 41 stable kidney transplant recipients (o30 mmol/l difference in creatinine between two most recent values) were recruited from the Ottawa Hospital at least 6 months after transplant . Patients were excluded if they were unable to provide informed consent, were pregnant or breastfeeding, had one or more episodes of acute rejection within preceding 3 months, a life expectancy o3 months, or anticipated graft failure within 3 months. These patients were not part of the White BTP equation derivation set, but had their GFR and analytes measured within the same study period (2004–2006), similar to those in the derivation set.23 Data from an additional 51 patients were recruited from a Canadian multicenter trial with GFR and analyte measurement performed between 2006–2008 at study baseline. Inclusion criteria include a four-variable MDRD study estimated GFR 420 ml per min per 1.73 m2, a time since transplant of at least 3 months, and at least 200 mg of proteinuria per day. Exclusion criteria are described in detail elsewhere.38 There were no medication-related exclusion criteria. The studies were approved by each facility’s Research Ethics Board and all patients provided informed consent. The pediatric study group consisted of 54 consecutive children with various kidney diseases and impaired kidney function (GFR o100 ml per min per 1.73 m2) who had been referred to the Children’s Hospital of Eastern Ontario (CHEO) for a DTPA GFR study and who had a BTP measured on the day of their DTPA GFR as part of routine clinical care. Laboratory assessment For both groups, GFR was measured using plasma clearance of radiolabeled 99mtechnetium-DTPA (99mTc-DTPA) through a single injection and plasma sampling at 120, 180, and 240 min after Kidney International (2009) 76, 784–791

original article

injection.3 GFR was then corrected for standard body surface area using the Dubois formula in adults39 and the Mosteller formula in children.40 At the time of the 99mTc-DTPA GFR measurement, blood was sampled for measurement of creatinine, urea, and BTP. BTP was measured for both adult and pediatric cohorts using a nephelometric assay on a BN Dade Behring ProSpec analyzer (Dade Behring, Marburg, Germany) at the same laboratory. The total analytical imprecision (intra-assay plus inter-assay) of the assay calculated from two control materials with concentrations of 1.51 and 7.89 mg/l was 2.33 and 6.5%. For the adult cohort, a modified Jaffe reaction on a Beckman Coulter LX20 Pro Clinical System was used to measure serum creatinine in the same laboratory with the manufacturer’s reagents (Beckman Coulter, Brea, CA, USA). Coefficients of variation (CV) for serum creatinine were 4.9% at 0.6 mg per 100 ml (55 mmol/l), 1.7% at 1.7 mg per 100 ml (150 mmol/ l), and 1.3% at 6.8 mg per 100 ml (600 mmol/l). A Beckman Coulter LX20 Pro Clinical System was also used to measure serum urea. The CV for serum urea was 3.5% at 14.6 mg per 100 ml (5.2 mmol/l), 1.9% at 37.8 mg per 100 ml (13.5 mmol/l) and 1.7% at 66.1 mg per 100 ml (23.6 mmol/l). For the pediatric group, creatinine was measured using an enzymatic assay on the Vitros 250 Chemistry System (Ortho Clinical Diagnostics, Markham, ON, Canada) using manufacturer’s reagents. CVs for serum creatinine were 3% at 0.87 mg per 100 ml (77.6 mmol/l), 1.5% at 2.11 mg per 100 ml (186.8 mmol/l), and 1% at 7.10 mg per 100 ml (628 mmol/l). Urea was measured using an enzymatic assay on the Vitros 250 Chemistry System (Ortho Clinical Diagnostics) using manufacturer’s reagents. The CV for serum urea was 1.3% at 13.7 mg per 100 ml (4.9 mmol/l), 1.6% at 35.5 mg per 100 ml (12.7 mmol/l), and 1.2% at 61.0 mg per 100 ml (21.8 mmol/l). Analysis Estimated GFR was calculated using the two new White BTP equations23 and the Poge BTP equations24 in all participants; the four-variable MDRD study equation41 in adults; and the Schwartz equation;26 and updated Schwartz equation13 in children (Table 1). For the four-variable MDRD study equation, creatinine values were calibrated to the Cleveland Clinic as recommended.32,42 Briefly, 50 Ottawa Hospital samples (range 53–354 mmol/l) were sent to the Cleveland Clinic laboratory. Overall, there was excellent agreement between both laboratories (correlation coefficient of 0.989). The calibrated creatinine was calculated as follows: 1.076(Ottawa Hospital serum creatinine value)7.35 mmol/l. For the Poge BTP2 equation, creatinine values adjusted to the IDMS standard were used for the adult cohort.24,42 Fifty Ottawa Hospital samples (range: 35–500 mmol/l) were analyzed using the ‘old’ calibrators (not referenced to IDMS) and then re-analyzed using set points obtained from the instrument manufacturer’s that would give IDMScompatible results. The following regression equation was obtained: IDMS creatinine ¼ 0.990(Ottawa Hospital serum creatinine value)6.02 mmol/l. The evaluation of the performance of the prediction equations was performed by calculating the bias, precision, and accuracy as recommended in the National Kidney Foundation guidelines on chronic kidney disease.25 Bias was presented as the difference between the estimated GFR and the measured 99mTc-DTPA GFR (estimated GFR–measured GFR) and the percentage difference ([estimated GFRmeasured GFR]/measured GFR  100). A negative bias indicates that the GFR is underestimated by the prediction equation. Precision was defined as the IQR of the median bias and the standard deviation of the mean bias.2 Accuracy was defined as 789

original article

the percentage of GFR estimates lying between 10 and 30% of the measured GFR (using 99Tc-DTPA).2 Differences in equation bias and accuracy were assessed using a Wilcoxon or McNemar’s test as appropriate. To account for multiple comparisons, we used a conservative Bonferroni correction factor.43 As we compared five prediction equations in each cohort, we adjusted our overall a by 5 (overall a ¼ 0.05/5 ¼ 0.01). Therefore, a P-value o0.01 was considered statistically significant. Concordance correlation coefficients between the measured GFR and estimated GFR were also calculated.44

CA White et al.: BTP GFR estimation equation validation

12.

13. 14.

15.

16.

DISCLOSURE

17.

The Physicians Services Incorporated Foundation, Astellas Pharma Canada, and Dade Behring had no role in the study design, data collection, analysis and interpretation, manuscript preparation, or any publication decisions.

18.

ACKNOWLEDGMENTS

This study was funded by the Physicians Services Incorporated Foundation, Astellas Pharma Canada, the Canadian Institute of Health Research, and Dade Behring. We thank the staff and patients from the renal transplant program and from the Children’s Hospital of Eastern Ontario who participated in the study. We acknowledge Alan Thibeau, Nur Jamal, Ian Graham, Lisa Banfield, Sheila Dowell, Philip St Laurent, Julie Noel, Elyse Bienvenue, Martine Blouin, Claudine Messier, and Sunil Thakrar for assistance with the DTPA glomerular filtration rate measurements; Marcella Cheng-Fitzpatrick and Anna Micucci for data management; and, Zeyad Alrayes, Judy Cheesman, Darlene Hackett, Margo McCoshen, Paul McLoughlin, Amy Pocock, Lisa South, and Jennifer Bowes for their invaluable assistance in conducting this study.

19.

20.

21.

22.

23.

24.

REFERENCES 1. Levey AS. Measurement of renal function in chronic renal disease. Kidney Int 1990; 38: 167–184. 2. Levey AS, Coresh J, Balk E et al. National kidney foundation practice guidelines for chronic kidney disease: evaluation, classification and stratification. Ann Intern Med 2003; 139: 137–147. 3. White C, Akbari A, Hussain N et al. Estimating glomerular filtration rate in kidney transplantation: a comparison between serum creatinine and cystatin C-based methods. J Am Soc Nephrol 2005; 16: 3763–3770. 4. Poge U, Gerhardt T, Stoffel-Wagner B et al. Cystatin C-based calculation of glomerular filtration rate in kidney transplant recipients. Kidney Int 2006; 70: 204–210. 5. Mariat C, Alamartine E, Barthelemy JC et al. Assessing renal graft function in clinical trials: can tests predicting glomerular filtration rate substitute for a reference method? Kidney Int 2004; 65: 289–297. 6. Pham-Huy A, Leonard M, Lepage N et al. Measuring glomerular filtration rate with cystatin C and beta-trace protein in children with spina bifida. J Urol 2003; 169: 2312–2315. 7. Poge U, Gerhardt T, Stoffel-Wagner B et al. Calculation of glomerular filtration rate based on cystatin C in cirrhotic patients. Nephrol Dial Transplant 2006; 21: 660–664. 8. Rule AD, Larson TS, Bergstralh EJ et al. Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med 2004; 141: 929–937. 9. Bostom AG, Kronenberg F, Ritz E. Predictive performance of renal function equations for patients with chronic kidney disease and normal serum creatinine levels. J Am Soc Nephrol 2002; 13: 2140–2144. 10. Grubb A, Nyman U, Bjork J et al. Simple cystatin C-based prediction equations for glomerular filtration rate compared with the modification of diet in renal disease prediction equation for adults and the Schwartz and the Counahan–Barratt prediction equations for children. Clin Chem 2005; 51: 1420–1431. 11. Seikaly MG, Browne R, Bajaj G et al. Limitations to body length/serum creatinine ratio as an estimate of glomerular filtration in children. Pediatr Nephrol 1996; 10: 709–711.

790

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

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. Schwartz GJ, Munoz A, Schneider MF et al. New equations to estimate GFR in children with CKD. J Am Soc Nephrol 2009; 20: 629–637. Huber AR, Risch L. Recent developments in the evaluation of glomerular filtration rate: is there a place for beta-trace? Clin Chem 2005; 51: 1329–1330. Urade Y, Hayaishi O. Biochemical, structural, genetic, physiological, and pathophysiological features of lipocalin-type prostaglandin D synthase. Biochim Biophys Acta 2000; 1482: 259–271. Olsson JE, Link H, Nosslin B. Metabolic studies on 125I-labelled beta-trace protein, with special reference to synthesis within the central nervous system. J Neurochem 1973; 21: 1153–1159. Hoffmann A, Nimtz M, Conradt HS. Molecular characterization of betatrace protein in human serum and urine: a potential diagnostic marker for renal diseases. Glycobiology 1997; 7: 499–506. Priem F, Althaus H, Birnbaum M et al. Beta-trace protein in serum: a new marker of glomerular filtration rate in the creatinine-blind range. Clin Chem 1999; 45: 567–568. Filler G, Priem F, Lepage N et al. Beta-trace protein, cystatin C, beta(2)-microglobulin, and creatinine compared for detecting impaired glomerular filtration rates in children. Clin Chem 2002; 48: 729–736. Woitas RP, Stoffel-Wagner B, Poege U et al. Low-molecular weight proteins as markers for glomerular filtration rate. Clin Chem 2001; 47: 2179–2180. Kobata M, Shimizu A, Rinno H et al. Beta-trace protein, a new marker of GFR, may predict the early prognostic stages of patients with type 2 diabetic nephropathy. J Clin Lab Anal 2004; 18: 237–239. Poge U, Gerhardt TM, Stoffel-Wagner B et al. Beta-trace protein is an alternative marker for glomerular filtration rate in renal transplantation patients [see comment]. Clin Chem 2005; 51: 1531–1533. White CA, Akbari A, Doucette S et al. A novel equation to estimate glomerular filtration rate using beta-trace protein. Clin Chem 2007; 53: 1965–1968. Poge U, Gerhardt T, Stoffel-Wagner B et al. Beta-trace protein-based equations for calculation of GFR in renal transplant recipients. Am J Transplant 2008; 8: 608–615. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification and stratification. Am J Kidney Dis 2003; 39: S1–S266. Schwartz GJ, Brion LP, Spitzer A. The use of plasma creatinine concentration for estimating glomerular filtration rate in infants, children, and adolescents. Pediatr Clin North Am 1987; 34: 571–590. White C, Akbari A, Hussain N et al. Chronic kidney disease stage in renal transplantation classification using cystatin C and creatinine-based equations. Nephrol Dial Transplant 2007; 22: 3013–3020. Bosma RJ, Doorenbos CR, Stegeman CA et al. Predictive performance of renal function equations in renal transplant recipients: an analysis of patient factors in bias. Am J Transplant 2005; 5: 2193–2203. Stevens LA, Manzi J, Levey AS et al. Impact of creatinine calibration on performance of GFR estimating equations in a pooled individual patient database. Am J Kidney Dis 2007; 50: 21–35. Dossetor JB. Creatininemia versus uremia. The relative significance of blood urea nitrogen and serum creatinine concentrations in azotemia. Ann Intern Med 1966; 65: 1287–1299. Laskow DA, Curtis JJ, Luke RG et al. Cyclosporine-induced changes in glomerular filtration rate and urea excretion. Am J Med 1990; 88: 497–502. Coresh J, Astor BC, McQuillan G et al. Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis 2002; 39: 920–929. Myers GL, Miller WG, Coresh J et al. Recommendations for improving serum creatinine measurement: a report from the Laboratory Working Group of the National Kidney Disease Education Program [see comment] [Review] [136 refs]. Clin Chem 2006; 52: 5–18. Kessler A, Siekmann L. Measurement of urea in human serum by isotope dilution mass spectrometry: a reference procedure. Clin Chem 1999; 45: 1523–1529. Morton KA, Pisani DE, Whiting Jr JH et al. Determination of glomerular filtration rate using technetium-99m-DTPA with differing degrees of renal function. J Nucl Med Technol 1997; 25: 110–114. Brochner-Mortensen J. Current status on assessment and measurement of glomerular filtration rate. Clin Physiol 1985; 5: 1–17. Kidney International (2009) 76, 784–791

original article

CA White et al.: BTP GFR estimation equation validation

37.

38.

39.

40.

White CA, Huang D, Akbari A et al. Performance of creatinine-based estimates of GFR in kidney transplant recipients: a systematic review. Am J Kidney Dis 2008; 51: 1005–1015. Knoll GA, Cantarovitch M, Cole E et al. The Canadian ACE-inhibitor trial to improve renal outcomes and patient survival in kidney transplantation— study design. Nephrol Dial Transplant 2008; 23: 354–358. DuBois D, DuBois EF. A formula to estimate the approximate surface area if height and weight be known. Arch Inter Med 1916; 17: 863–871. Mosteller RD. Simplified calculation of body-surface area. N Engl J Med 1987; 317: 1098.

Kidney International (2009) 76, 784–791

41.

42.

43.

44.

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 [comment]. Ann Intern Med 1999; 130: 461–470. Levey AS, Coresh J, Greene T et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 2006; 145: 247–254. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B 1995; 57: 289–300. Lin LI. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989; 45: 255–268.

791