Introduction

Adequate measurement of kidney function is important for the management of all children and adolescents. In the neonatal period, there is considerable ontogeny of drug disposition, largely due to developmental changes of nephron formation and recruitment [1]. This was shown elegantly for ceftazidime [2] and for famotidine [3]. Accurate measurement of kidney function is also important for drugs excreted by the kidneys across all other ages. With many kidney diseases, intervention often depends on whether kidney function is normal or abnormal. Further, when there is impaired kidney function, accurate assessment of kidney function is important for initiation of renal replacement therapy [4], listing for renal transplant [5], evaluating interventions, and monitoring changes of function over time [6].

For drug dosing, tubular secretion is perhaps more important than glomerular filtration rate (GFR) [7]. Tubular secretion capability cannot easily be measured and requires assessment of both GFR and renal plasma flow [8]. GFR may be inappropriately high due to glomerular hyperfiltration [9]. Glomerular hyperfiltration is a phenomenon that can occur in various clinical conditions including kidney disease. No single definition of glomerular hyperfiltration has been agreed upon [10]. However, it is thought that glomerular hyperfiltration can be caused by afferent arteriolar vasodilation as seen in patients with diabetes or after a high-protein meal, and/or by efferent arteriolar vasoconstriction owing to activation of the renin-angiotensin-aldosterone system, thus leading to glomerular hypertension [10]. GFR may therefore be inappropriately high for a given nephron endowment. Despite this limitation, it is widely accepted that kidney function is best measured as GFR [11].

GFR cannot be measured directly. The most common method of measuring GFR is based on the concept of clearance. The renal clearance of substance x (Cx) is calculated as:

$$ {{\mathrm{C}}_{\mathrm{x}}}={{{{{\mathrm{U}}_{\mathrm{x}}}\mathrm{V}}} \left/ {{{{\mathrm{P}}_{\mathrm{x}}}}} \right.} $$

where V is the urine flow rate (ml/min), Ux is urine concentration of substance x, and Px is the plasma concentration of substance x. Cx is expressed in milliliters per minute. If the substance is freely permeable across the glomerular capillary and is not synthesized, transported, or metabolized by the kidney, Cx is equal to GFR [12].

Gold-standard measurement of GFR

If a substance in stable concentration in the plasma is physiologically inert, freely filtered at the glomerulus, and is not secreted, reabsorbed, synthesized, or metabolized by the kidney, the amount of that substance filtered at the glomerulus is equal to the amount excreted in the urine [13]. The gold standard for the measurement of GFR is inulin clearance. In 1934, while studying water reabsorption in the renal tubule of amphibians, Richards found that the polysaccharide inulin (a fructose polymer made from the Jerusalem artichoke) is freely filtered through collodion membranes, not absorbed. Inulin is one of the few molecules that satisfies the requirements of an ideal marker of glomerular filtration, namely exclusive elimination by glomerular clearance and no tubular secretion and no non-renal excretion [13].

When considering inulin clearance, a few concepts have to be introduced:

  1. 1.

    Single nephron GFR vs. total GFR: The level of GFR is the product of the single-nephron GFR multiplied by the number of functioning nephrons in both kidneys. In the case of CKD, GFR can be decreased because of a reduction in filtration rate of each nephron and/or a drop in nephron number. Factors leading to decreased renal perfusion may cause a drop in the single nephron GFR. The total GFR serves as the most reliable marker of functioning renal mass [12].

  2. 2.

    Single bolus vs. infusion technique: GFR with an exogenous marker can be measured either by infusion technique [14] or single bolus [15]. The gold standard is an infusion technique that involves a bolus injection, followed by a steady infusion, blood samples at 2, 3, and 4 h and timed urine collection [16]. Small children require catheterization. This is obviously very invasive. The single bolus injection method with its plasma disappearance approach is more convenient and eliminates the need for timed urine collection. However, there are principal disadvantages with an infusion technique. After injections of inulin or other GFR markers, the marker is first distributed in the intravascular space. The volume of distribution is the extracellular space [17]. Time needs to pass to allow for equilibration between both compartments. Only infusion technique with timed urine collection after 3 or 4 h will have saturated both compartments for certain. If samples after single bolus injection of the GFR marker are drawn before equilibration occurs, GFR will be overestimated [18]. The median equilibration time is usually considered to be 20 min in adults, but there is wide inter-individual variability [19]. Children may have longer equilibration times than adults because of the different body composition [20]. The authors are unaware of any studies that have determined equilibration time between the various compartments of inulin in children. Especially in situations such as nephrotic syndrome, edema, and increased extracellular volume, the equilibration time may be much longer. Studies in dogs suggest considerable impact of hydration status and hemodynamic changes [21]. It is not always possible to predict the equilibration time and protocols used for single bolus GFR measurement use fixed sampling times without accommodating to the hydration status of the patient. The difference between infusion versus single bolus injection inulin clearance is on average 9.7 ml/min/1.73 m2 in children, with a tendency to overestimate GFR with the single bolus technique [18]. For a patient with a true GFR of 20 ml/min/1.73 m2, this could mean a delay in transplant listing.

  3. 3.

    The third important concept is the appropriate use of pharmacokinetic modeling for the excretion of inulin. Most centers do not consider this when using either infusion or single bolus injection methods. Agents used to measure glomerular filtration rate (GFR) give a biexponential plasma disappearance curve on multiple peripheral venous sampling between 20 min and 4 h after intravenous injection. These two exponentials are generally regarded to represent equilibration of agent throughout the extracellular fluid (ECF) space and renal clearance, respectively. The arterial plasma clearance curve of GFR agents is triexponential; the first exponential reflects equilibration of agent between plasma and the interstitial space of carcass tissue (mainly muscle and skin). The second exponential is minor compared with the first; it is not clear what it represents. Only the third exponential reflects renal clearance [19]. To adequately address this issue, proper two-compartmental pharmacokinetic modeling needs to be utilized as shown by Van Rossum et al. [22]. In this paper, the need for a late sample at 240 min when using the single bolus injection method is stressed [22]. We recommend at least 3 sampling times and a start no earlier than 90 min. Delayed sampling is important in patients with severe renal function impairment.

If inulin clearance is utilized, we recommend a bolus injection of 5,000 mg of inulin per 1.73 m2, with a maximum dose of 5,000 mg, using Inutest from Fresenius. The inulin has to be infused with a constant rate over 30 s. Extravasation must be excluded as this would result in overestimation of GFR. There have to be 3 or more sampling times, ideally with samples extending to 240 min. For the evaluation, a two-compartmental model using the concentrations at the actual time points must be utilized, ideally with NONMEM or other appropriate pharmacokinetic software programs like WINNONLIN [18]. The limited availability of Inutest and the issues outlined in this chapter may significantly restrict the use of this method. Therefore, alternate methods of measuring GFR have been developed.

GFR methods using radiolabeled isotopes and iohexol

Since the 1970s, radiolabeled isotope techniques for estimation of GFR have replaced inulin clearance [23]. The main advantage of using a radiolabeled compound with characteristics similar to inulin is its immediate determination by counting the radioactivity. Today’s nuclear medicine methods with easily assayed, radiolabeled, and stable compounds that meet the criteria of a marker of GFR, namely clearance only by glomerular filtration without tubular secretion and non-renal elimination, form a new standard. Unfortunately, the widely used iothalamate does undergo tubular secretion [24]. The most widely used single bolus-injection techniques utilize the intravenous injection of a suitable compound at a precisely known amount. Similar to inulin clearance, extravasation will cause significant overestimation of GFR [25]. After the injection, the plasma must be sampled from the opposite arm and concentration is plotted against time.

A number of substances are used for nuclear medicine GFR measurement. They are listed in Table 1. An important concept to introduce is that tracers used for GFR measurement by nuclear medicine show a small amount of plasma protein binding, which may affect their ultrafiltration (see Table 1) [26]. As a consequence, there is slight underestimation of GFR with nuclear medicine techniques by about 10 % compared to inulin [27]. It is also known that the negatively charged glomerular basement membrane forms a restriction to the ultrafiltration of 51Cr-ethylenediamine tetra-acetic acid (EDTA) and 99Tc-diethylenetriamine penta-acetic acid (DTPA) [26]. As a consequence, disorders affecting the serum albumin concentration or the glomerular basement charge, such as nephrotic syndrome, may affect the accuracy of GFR measurement [26].

Table 1 Commonly used nuclear medicine radioisotopes or cold exogenous markers for GFR measurement
Full size table

Most of these nuclear medicine GFR measurements are used as single-bolus injection methods. The same limitations that apply for inulin also apply to these substances. It is generally recognized that at least three sampling points are required. Again, a 240-min sample is necessary, and for lower GFR even further extended sampling times are required. Unfortunately, there is poor standardization among most centers, and only two-point sampling time points are often chosen to reduce the number of venipunctures in children. Similar to the limitations raised with inulin, most centers utilize only one-compartmental models, instead of appropriate pharmacokinetic two-compartmental non-linear models. As with inulin, a single compartmental model results in some overestimation of GFR [28]. Late sampling (4 and/or 5 h concentrations after injection) could overcome some of the problems related to the mathematical models, but this is rarely implemented [29]. For a more accurate assessment of the extracellular water (especially important in children with altered fluid status, for instance if they are on long-term diuretics), an early sampling point is required [30]. Also, if one-compartmental models are employed, appropriate corrections for the overestimation should be employed [28].

Of course, nuclear medicine GFR scans involve radiation exposure. Beta-particles from 51Cr-EDTA are clearly carcinogenic. There are low and high dose methods for 99mTc DTPA GFR scans [31]. While most of the radiation is quickly excreted through the kidneys, the whole long-term body radiation exposure of 99mTc DTPA GFR scans in adults was 2–3 % of the administered radioactivity, probably reflecting the protein-bound fraction. Also, while the radiation exposure is low, 4 % is retained by various tissues at 24 h, 69 % is eliminated by the kidneys with a half time of 1.73 h in healthy adults [26]. The radiation exposure clearly forms a limiting factor towards repeated nuclear medicine GFR scans. The non-ionic low osmolar contrast medium iohexol has become the most commonly used contrast medium for GFR measurements in Europe and may form a good alternative to the radiolabeled GFR markers (Table 1), especially as it has a much lower protein binding.

The lack of standardization of nuclear medicine methods needs to be addressed. Apart from the issues named above, there ought to be a mandatory image of the injection site to assess for extravasation, and a plot of the radioactivity over time should be provided to assess for outliers of the equilibration, especially in hemodynamically unstable patients and patients with significant volume overload. Finally, non-linear modeling using a two-compartmental model should be standard.

In our opinion, if nuclear medicine GFR studies with the plasma disappearance method are employed, the non-compartmental Russell method can be used, but a two-compartmental model is preferable. The Russell method in short: Plasma 99mTc-DTPA is calculated, after correcting for the decay rate from the control sample, based upon a formula by Russell et al. [32]. The formula shown below uses a non-compartmental linear approach after log-transformation of the counts in each of the three samples. The median of the three-GFR values was then standardized by total body surface area. The first measurement should not occur earlier than 90 min.

$$ \begin{array}{*{20}c} {GF{R_{{{i_{Plasma }}}}}={{{\left[ {\frac{{10000\times St}}{{{T_i}-{T_{i-1 }}}}\times \ln \left( {\frac{{{P_{i-1 }}}}{{{P_i}}}} \right)\times \exp \left( {\frac{{{T_{i-1 }}\ln {P_i}-{T_i}\ln {P_{i-1 }}}}{{{T_i}-{T_{i-1 }}}}} \right)} \right]}}^{0.979 }}\quad \mathrm{where}\,i=1,2,3\left( {\mathrm{sample}} \right)} \\ {GF{R_{Plasma }}=median\left( {GF{R_1},GF{R_2},GF{R_3}} \right)\times \left( {\frac{1.73 }{BSA }} \right)\quad \mathrm{where}\,BSA=0.007184\times \left( {heigh{t^{0.725 }}\times weigh{t^{0.425 }}} \right)} \\ \end{array} $$

Nuclear medicine GFR: plasma disappearance versus urinary clearance method

There are two methods typically employed to estimate the GFR: the commonly used plasma disappearance method described above, which examines the rate of decline of a radioisotope in timed plasma samples to determine clearance, and a more labor-intensive urinary method, which combines timed samples of both urine and plasma to determine clearance. Several studies have demonstrated that the accuracy of the plasma disappearance method of GFR estimation is limited when GFR is below 20–30 ml/min/1.73 m2 [33, 34]. Trials assessing GFR estimation equations have used the urinary clearance GFR method to overcome this limitation [35]. However, these studies are limited to advanced CKD in adults. We recently performed a study across all GFR ranges in adults that have not yet been published. We used the following method:

The urinary clearance of 99mTc-DTPA was calculated using urinary clearance equation:

$$ \begin{array}{*{20}c} GF{R_{{{i_{Urine }}}}}=\frac{{{U_i}\times {V_i}}}{{\left( {{T_i}-{T_{i-1 }}} \right)\times \exp \left[ {0.5\times \left( {\ln {P_i}+\ln {P_{i-1 }}} \right)} \right]}}\kern0.75em \mathrm{where}\ i=1,2,3\ (\mathrm{sample}) \hfill \\ \hfill \\ GF{R_{Urine }}=median\left( {GF{R_1},GF{R_2},GF{R_3}} \right)\times \left( {\frac{1.73 }{BSA }} \right)\kern0.75em \mathrm{where}\ BSA=0.007184\times \left( {heigh{t^{0.725 }}\times weigh{t^{0.425 }}} \right) \hfill \\\end{array} $$

where U i is the urinary concentration of 99mTc-DTPA adjusted for isotopic decay rate, V i is the urinary volume and P i is the plasma concentration of 99mTc-DTPA adjusted for the isotopic decay rate. While there was a strong correlation between both the urinary and plasma disappearance method (R2 = 0.86), the plasma disappearance method showed increasing overestimation with lower GFR. Among the 213 samples ranging from chronic kidney disease (CKD) stage 1 through 4, the mean (standard deviation) plasma iGFR of 60.7 (24.9) ml/min/1.73 m2 compared to urinary GFR of 52.0 (28.0) ml/min/1.73 m2, which was statistically significant (p value < 0.001). We concluded that the plasma clearance method of GFR estimation generally yielded higher results than the urinary method, with the bias increasing with advanced CKD stage.

The physician ordering a nuclear medicine GFR in a child needs to be fully aware of the limitations of these methods. Table 2 summarizes the few studies that were performed comparing inulin clearance with the various nuclear medicine methods. Unfortunately, none of them show complete agreement. The small number of patients in these studies points to the need for more studies. With regards to the bias, 51Cr EDTA appears to be the most favorable (Table 2), but its availability is largely restricted to Europe.

Table 2 Comparison of inulin clearance with nuclear medicine methods
Full size table

Magnetic resonance imaging (MRI) for assessment of GFR

Magnetic resonance with dynamic imaging of uptake of contrast agent using a two-compartment model has been proposed for GFR measurement [36], but more clinical studies are needed to establish if this method is sensitive to alterations in disease state, and it requires contrast administration. Functional magnetic resonance imaging enables non-invasive assessment of renal function without contrast media; recently it was reported that the renal blood oxygen-level dependent magnetic technique failed to discriminate between patients with different stages of chronic kidney disease [37]. At this point in time, it is too early to utilize this promising technique, and the use of gadolinium forms a significant limitation. In patients with advanced renal failure gadolinium can cause nephrogenic systemic fibrosis (NSF) characterized by fibrosis of the skin and connective tissue, and occasionally includes internal organs—it develops in days or over several weeks, and can be fatal. Patients with moderate renal disease or with acute kidney injury are at risk of developing NSF, and alternative tests or MRI without gadolinium should be considered.

Children-specific considerations

There are significant developmental changes (ontogeny) during childhood that have a significant effect on the use of any compound for the measurement of GFR. To accommodate for this change, GFR is normalized to body surface area, which is considered to be the best denominator, although the extracellular volume is also considered [28]. However, while all nephrons are terminally differentiated at birth, only the juxtamedullary glomeruli are used at birth. There is continuous recruitment of additional glomeruli until 18–24 months of age [38, 39]. Ideally, GFR should be reported as age-independent z-scores, rather than as absolute GFR or GFR/body surface area [40]. In the first year of life, and during puberty, there are growth spurts and rapid increase of muscle mass, requiring special considerations for the calculation of GFR [41], especially in adolescent males [42]. The most important determinant of GFR is height, which when taken into account led to the Schwartz formula [43], a GFR estimation model based on the height/creatinine ratio. Every nephrologist is familiar with it, and in its most recent iteration, the accuracy is much improved, especially in the range of 25–75 ml/min/1.73 m2 [44]. As outlined above, gold-standard measurements of GFR are cumbersome and invasive and involve the use of an exogenous compound. Those children at risk who require frequent GFR monitoring, such as transplant recipients [45], are already subject to numerous other tests. As such, accurate non-invasive measurements of GFR are essential and endogenous markers are required.

Endogenous surrogate markers of GFR

Because of this, it has become a preferred practice to use an endogenous marker that is produced at a constant rate, shares the features of inulin, and thus eliminates the need for a compound injection. Popper and Mandel proposed the use of serum creatinine in 1937 [46], which remains the most widely used marker for GFR estimation in spite of its shortcomings. Whereas serum endogenous markers are considered as suitable markers of GFR, the National Kidney Foundation guidelines recommend the use of serum marker-based prediction equations for GFR estimation and rules against the use of serum markers alone in the assessment of renal function, as discussed in [47]. However, the limitations of creatinine in children are far greater than that in adults, which has lead to significant research in this field for years.

Limitations of serum creatinine as endogenous marker of GFR

Serum creatinine remains the most widely used endogenous marker to predict GFR. Creatinine is a metabolic product of creatine and phosphocreatine found in muscle, and as such reflects muscle mass and varies little from day to day [48]. Serum creatinine measurement is widely available at low cost. In Canada, the cost of a single serum measurement is 40 cents US, whereas cystatin C is approximately $11.60 for the reagents and standards. On a population basis, creatinine may be a feasible marker of GFR in populations with near normal GFR, but this marker has multiple limitations for individuals. There is substantial inter- and intra-patient variability due to differences in muscle mass [49]. In childhood, there is age and muscle mass dependency of serum creatinine, and accurate assessment of normal GFR even with the use of body length/creatinine ratios remains difficult. In certain pediatric patient clientele, such as patients with spina bifida [50], neuromuscular disease, anorexia nervosa, or liver cirrhosis [51], serum creatinine is completely unusable because of the abnormal muscle mass in these children, who are often wheelchair bound [50]. In addition, the production of creatinine is not constant; the rate of turnover is also variable [52]. It is also known that creatinine undergoes tubular secretion that increases with lower GFR [53]. Creatinine is secreted in the proximal convoluted tubule by active transport similar to that of organic cations. Tubular secretion of creatinine divided by inulin clearance increases progressively from 0.16 in adult patients with normal GFR to 0.92 in patients with GFR less than 40 ml/min [54]. Furthermore, there used to be considerable variability in the reference range for serum creatinine based on the method used for its determination [55]. Proficiency testing surveys, published periodically by the College of American Pathologists, evaluated the variability in creatinine measurement among various methods used by different institutions across the US and demonstrated substantial variability [56]. This will make the validation of older methods to estimate GFR based on serum creatinine, measured by different techniques, subject to debate. Only more recently, the traceability of the creatinine measurements to higher-order reference methods (isotope dilution-mass spectrometry (IDMS) reference method) has improved accuracy of creatinine measurements [57]. This paper established improved pediatric reference intervals that may be adopted by any laboratory serving a similar population (predominantly Caucasian). Clinical implications of creatinine standardization have recently been reviewed [58].

Overall, the method used for serum creatinine is of high importance. Calculation of GFR based on serum creatinine measured by the alkaline picrate method is limited because of method non-specificity, low values in children, particularly in infants, and lack of appropriate GFR formulae. Therefore, enzymatic methods are preferred, as widely suggested in the literature [59]. As outlined above, creatinine measurements have to be IDMS traceable.

Even when correcting for the analytical variability, the problem of tubular secretion of serum creatinine cannot be addressed with the methodology of measurement. Blockade with H2 antagonists revealed promising results that may overcome the significant problem of tubular secretion of serum creatinine [60]. The use of cimetidine protocols in children remains scarce [61]. Creatinine clearance determinations involving timed urine collections may provide greater accuracy, but are difficult for pediatric patients to perform, time-consuming, and impractical for routine use. Small molecular proteins have emerged as superior endogenous markers of GFR [62]. For the new Schwartz formula, cystatin C and urea as well as creatinine measurements are required [44].

Small molecular weight proteins as markers of GFR

Small molecular mass proteins have long been proposed as markers of GFR as they are normally almost freely filtered through the normal glomerular membrane [63]. In a normally functioning kidney these small molecular weight proteins should then be almost completely reabsorbed and degraded by the proximal tubular cells. Several proteins have been tested, such as beta-2 microglobulin and beta-trace protein (analyzed in children in [64]). Beta-2 microglobulin is an acute phase protein, associated with inflammation [65]. However, a recent study in unborn children with congenital obstructive uropathy has demonstrated good results for the measurement of GFR in the fetus [66]. Cystatin C is the best-studied small molecular weight protein as a surrogate GFR marker and its superiority for the detection of mildly impaired GFR was proven in meta-analyses [67].

Cystatin C

The features of cystatin C have been reviewed in detail elsewhere [12]. Earlier studies of the serum level of cystatin C in large patient cohorts have failed to correlate the serum level to any pathophysiological state besides those affecting the glomerular filtration rate, which also is compatible with a stable secretion of cystatin C from most human tissues [68]. However, very large doses of glucocorticoids have recently been described to increase the production of cystatin C [69], whereas low and medium doses of glucocorticoids do not seem to alter its production [70]. Also, thyroid dysfunction can affect cystatin C levels [71], but our group found no association with thyroid function markers in recent studies [72, 73].

The reference values for cystatin C obtained in a carefully selected population are 0.75 ± 0.09 mg/l for children aged 4–19 years, 0.74 ± 0.10 mg/l for males and 0.65 ± 0.09 mg/l for females (aged 20–59 years), and 0.83 ± 0.10 mg/l for older individuals (> or =60 years) [74]. In the first year of life, renal function matures physiologically. Accordingly, much higher cystatin C values up to 2.8 mg/l were found at birth. These are subject to a rapid decline after birth reflecting maturation of kidney function [1]. Age dependency also has to be considered in adults. New reference intervals with more detailed age distribution from central Europe have recently been published on 985 healthy subjects over 25 years old [75].

Studies of the handling of human cystatin C in rats have shown that the plasma renal clearance of cystatin C is 94 % of that of the generally used GFR-marker 51Cr-EDTA and that cystatin C thus is practically freely filtered in the glomeruli [76]. At least 99 % of the filtered cystatin C is degraded in the tubular cells. When the GFR of a set of rats was variably lowered by constricting their aortas above the renal arteries, the renal plasma clearance of cystatin C correlated strongly with that of 51Cr-EDTA, with a linear regression coefficient of 0.99 and with the y-intercept not being statistically different from 0 [77].

After these encouraging studies, additional studies suggested that the reciprocal of cystatin C correlates better with a gold-standard GFR measurement than the reciprocal of serum creatinine [78]. Cystatin C as a marker of GFR was found to be independent of body composition [79]. Development of automated and rapid particle-enhanced immunoturbidimetric and immunonephelometric methods, also more precise than the original radioimmuno- or enzyme-linked immunosorbant assays [80], has allowed large-scale use of serum cystatin C as a clinically useful GFR-marker.

The diagnostic performance of cystatin C in comparison with serum creatinine was first analyzed in 2002 with a meta-analysis of 46 studies, in both adults and children [67]. The pooled data analysis compared correlation coefficients between GFR and the reciprocals of serum creatinine and cystatin C in 3,703 individuals and found significantly better correlations (mean r = 0.816 [95 % confidence interval: 0.804–0.826] versus mean r = 0.742 [95 % confidence interval: 0.726–0.758]). ROC plots were available for a pooled sample size of 997 individuals, again showing a significantly better area under the [ROC] curve (mean = 0.926 [95 % confidence interval: 0.892–0.960] versus mean r = 0.837 [95 % confidence interval: 0.796–0.878]). This meta-analysis suggests that cystatin C is superior to serum creatinine for the detection of impaired GFR in cross-sectional studies. More recently, another meta-analysis of 24 studies (n = 2,007) also confirmed that the diagnostic accuracy favored cystatin C [81], although in this more recent study, the diagnostic odds ratios started to overlap, most probably due to the IDMS testing of serum creatinine. Cystatin C measurement is more expensive than creatinine.

The need for standardization of cystatin C, similar to the IDMS traceability of serum creatinine

It should be highlighted that problems due to the lack of standardization of cystatin C lead to similar problems as highlighted above with regards to serum creatinine and the need for IDMS traceability [57]. Results obtained with the two main assays that are commercially available, namely the DAKO kit (turbidimetric, PETIA) and the Siemens Healthcare assay (nephelometric, PENIA), can be quite different, as recently demonstrated with the publication of reference intervals for healthy term and preterm infants. The results with the PENIA method in our study [82] were significantly lower than those reported by Harmoinen et al. [83]. Similar problems occurred when the results of the CKiD study were re-analyzed using PENIA, which led to substantially better results [44]. The difference in the results between these assays is partially due to different calibrators; DAKO uses recombinant cystatin C and Siemens human cystatin C from patients with tubulopathy. Recently, Anders Grubb reported the first certified standardized reference material for cystatin C [84]. Worldwide standardization, as demonstrated above, is a powerful tool towards improving the diagnostic accuracy of GFR biomarkers, and the authors of this review are convinced that standardization of cystatin C similar to that of creatinine will further enhance the diagnostic performance of cystatin C.

Beta-trace protein

Another potential surrogate marker for GFR measurement is beta-trace protein (BTP) [64], a low molecular weight protein that has been traditionally used as a marker for cerebrospinal fluid leakage. Beta-trace protein shares many of the features of cystatin C as a marker of GFR [64]. Unlike cystatin C, which is strongly positively charged, beta-trace protein has an isoelectric point of 5.5 or lower [85]. Beta-trace protein appears to be a more favorable marker in pregnancy [86] and in newborns [87] and might be better than cystatin C in renal transplant children [88, 89]. Formulae for the estimation of GFR have been developed and validated for adults [90] and for children [91].

Conclusions

In summary, inulin remains the gold standard GFR measurement. However, there is limited availability and its use remains cumbersome and invasive. Nuclear medicine GFR scans have replaced inulin clearance in most centers; however, they are not completely interchangeable. Iothalamate undergoes tubular secretion, which is a significant limitation. 51Cr EDTA and 99Tc DTPA have plasma protein binding that reduces their clearance in comparison to inulin. Iohexol may be the best marker on a theoretical basis because of the low plasma protein binding, and being non radioactive can be re-used in children without radiation burden; however, it demonstrates considerable scatter, and the clinical studies comparing inulin GFR with nuclear medicine scans favor 51Cr EDTA, which has showed the lowest bias and best agreement, both in transplant and non-transplant patients. Significant limitations apply if extravasation occurs, if samples are drawn too early before an equilibration with the entire extracellular volume occurs, and if no appropriate two-compartmental model is used for the analysis. Most centers just use log transformation and proper pharmacokinetic non-linear two-compartmental models with exact time points are rarely employed. The endogenous markers have far greater limitations. Low molecular weight proteins appear to be more favorable than creatinine. Reasonably good agreement between measured and estimated GFR can be achieved with cystatin C, urea and creatinine together. Cystatin C appears to be slightly superior to beta trace protein with the exception of during pregnancy and the neonatal period.