Pharmacokineatic Aspects of Measurement of Glomerular Filtration Rate in the Dog: A Review

Review zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAJ Vet Intern Med zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1998;12:401-414 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAPharmacokinetic Aspects of Measurement of Glomerular Filtration

Rate in the Dog: A Review zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAReidun Heiene and Lars Moe

Glomerular filtration rate (GFR) is estimated by means of clearance, defined as the volume of plasma that has been cleared of a particular substance per unit time. Glomerular filtration rate may be estimated by measuring the renal clearance of a filtration marker using data from both urine and plasma or by plasma clearance using only plasma data. Several alternative pharmacokinetic models are used for the calculation of clearance using various filtration markers with slightly different pharmacokinetic properties. The purpose of this article is to discuss how the choice of marker and pharrnacokinetic model may influence estimated GFR values and to elucidate commonly used methods and reported GFR values in the dog.

Key words: Pharmacokinetics; Renal function.

he basis of urine formation is the ultrafiltration of plas- T ma in the glomerulus. A measure of the magnitude of this filtration therefore is generally accepted to be the best overall index of renal functi0n.l Glomerular filtration rate (GFR) cannot be measured directly but rather is estimated using the clearance of a filtration marker. Cumbersome techniques have been the major obstacle to more extensive use of clearance measurements in a clinical setting. Serum creatinine concentration is a readily available but less sensitive indication of GFR.

Estimation of GFR is widely used in research but can also be very useful in the clinic. In patients with vague clinical symptoms and mildly increased serum creatinine concentration, GFR may indicate whether the kidneys are diseased or not. Estimation of GFR may help to identify intrinsic renal disease from a long list of differential diag- noses in nonazotemic patients with polyuria, proteinuria, or hematuria. Accurate GFR determination may facilitate prognostic decisions in patients with established renal disease. In old and critically ill patients and in therapeutic drug monitoring situations, GFR estimation facilitates correct drug dosage adjustments.

Glomerular filtration rate in the dog may be influenced by several nonrenal factors such as protein intake, hydration status, sodium balance, gender, age, and breed, as well as day-to-day and circadian variations for the individual dog.,-' Thus, the range of reference values in healthy dogs is wide.

Renal clearance of inulin is the accepted reference method for estimation of GFR. The method is labor intensive because it requires continous intravenous infusion, accurate determination of urine production, and a complicated laboratory analysis. The quest for accurate methods that are

From the Center for Companion Animal Health, School c f l Vet- erinary Medicine, University zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAof California, Davis, CA (Heiene); and the Department of Smull Animal Clinical Sciences, Norwegian College of Veterinary Medicine, Oslo. Norway (Moe).

Reprint requests: Reidun Heiene, Department of Small Animal Clin- ical Sciences, Norwegian College of Veterinary Medicine, P.O. Box 8146 Dep., zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA0033 Oslo. Norway; e-mail: [email protected].

Accepted November 5, 1997. Copyright zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA0 1998 by the American College of Veterinary Internal

Medicine 0891 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA-6640/98/1206-0oO1/$3.00/0

simpler to use has resulted in development of new markers of GFR and new kinetic models but also new pitfalls.

Several new methods have been described in recent years. Some of these methods are simple and economical, thus facilitating increased clinical use of GFR estimation. The differences in reported values may be confusing, and possible sources of error should be known when establish- ing a method for estimating GFR. The aims of this review are to explain the basic concepts, underlying assumptions, and methodological problems of the various methods and to discuss some of the results from studies in the dog.

Clearance and Pharmacokinetics Clearance (CL) may be defined zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAas the volume of plasma

cleared of a substance during a given interval of time (mL/ min). Glomerular filtration rate is estimated by clearance measurements under 1 st order conditions (ie, in nonsatur- able systems in which clearance is constant for different plasma concentrations).8 Clearance must be measured during a hernodynamically stable period.

Total body clearance is the sum of all elimination processes in the body, mainly renal clearance and hepatic clearance. Elimination through saliva, sweat, and other routes often is negligible. Plasma clearance reflects total body clearance. Renal clearance equals plasma clearance and total body clearance if the substance is 'excreted only through the kidney.

The term renal or urinary clearance (CL,,,,,) is used to describe clearance calculated using both plasma and urine data, as opposed to plasma clearance (CLplasma), which is calculated using plasma concentration versus time data. Re- nal clearance measures the amount of marker actually excreted in the urine, whereas estimation of renal function by CLplasma depends upon additional assumptions. Renal clearance must be measured after distribution of the substance throughout the body tissues is complete, at which time the tissue : plasma ratio of drug concentrations is constant.8 Re- nal clearance is calculated from the classical formulay

where U = urine flow in d m i n , C, = marker concentration in urine in mg/mL, and C, = marker concentration in plasma in mg/mL. This approach is accurate provided urine collection is complete. In practice, it often is difficult to

402 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAHeiene and Moe

Table zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1. The relationship between the estimated clearance value (CL) and the true glomerular filtration rate (GFR) in various situations of compliance and noncompliance of the filtration marker to the criteria for the ideal filtration marker.

Biological Behavior of the Filtration Marker Renal Clearance Plasma Clearance

Elimination by filtration only CL zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA= GFR CL = GFR Tubular secretion of marker CL > GFR CL > GFR Tubular reabsorption of marker CL < GFR CL < GFR Extrarenal excretion of marker CL = GFR CL > GFR Marker bound to plasma proteins CL < GFR Delayed distribution into tissues CL < GFR CL > GFR Combinations of factors CL < or = or > GFR

CL < GFR

CL < or = or > GFR

make sure that the bladder is empty after collection, and incomplete bladder emptying will lead to underestimation of GFR. Risk of infection due to catheterization also is a complicating factor. The time span of measurement should not exceed 1-2 elimination half-lives of the drug without correction for the logarithmic fall in plasma concentration if mean plasma concentration during the time of measurement is used.

With the traditional inulin method, CL,,,,, is commonly measured at relatively constant blood concentrations during continous intravenous injection of marker, but CL,,,, may also be measured after a single injection, when distribution is complete.

Plasma clearance is calculated by the formula

CLplasma = DIAUC,

where D = dose of marker and AUC = area under the plasma concentration versus time curve. The area under the curve is calculated with specified formulas for different pharmacokinetic models.8 These models represent abstrac- tions from reality, used to facilitate matemathical analysis of the concentration versus time data. The accuracy of the CLplaama estimate depends upon whether the choice of pharmacokinetic model is appropriate.

Other important pharmacokinetic parameters are appar- ent volume of distribution (V,), the elimination rate constant zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA( k ) , and the elimination half life (fl,J. In a l-compartment model, these parameters are defined and related as follows:8

k = CLIV,

C, = AIV, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAtIIz = In 2lk,

where A = amount of drug in the body, C, = plasma concentration at time zero, and CL = total body clearance.

Clearance and V, are called primary pharmacokinetic parameters because they depend only upon physiological vari- ables and not upon each other. The secondary pharmacokinetic parameters, tIIz and k, are sensitive to changes in both CL and V, and therefore are not well suited for evaluation of GFR. Clearance generally is not affected by changes in body fluid volumes such as increases or decreas- es in extracellular fluid volume (ECFV) unless blood flow to the clearing organ is affected.

Commonly used filtration markers have a V, similar to the ECFV because they enter cells only to a limited degree and are not bound to body proteins. One exception is cre-

atinine, which has a V, similar to the volume of total body water.

Although theoretically it is possible to use any pharmacokinetic method for any marker, certain traditions have been established for practical reasons. Estimation of renal clearance by the classical formula from blood and urine data has been the most common approach, but the use of CLplasma methods is becoming more common.

Ideal Marker of Renal Filtration The GFR is equal to the CL,,,, of a filtration marker if

the marker is not protein bound, does not enter red blood cells, and is only excreted through the kidney solely by filtration, with no tubular secretion or reabsorption. In ad- dition, the substance must not be toxic and must not in itself alter GFR. For such a substance, CLplarma equals CL,,,,, provided that the pharmakokinetic method used for calculation of CL,,,,,, is appropriate.

The fructose polymer inulin comes close to meeting these criteria? If a new marker is introduced, it is common to compare the results to CL,,, of inulin for validation of the new method.

Influence of Marker Properties The properties of the filtration marker influence the qual-

ity of the measured clearance value as an estimate of GFR. In the calculations, the filtration marker is assumed to fulfill the criteria specified above. In reality, the marker may ex- hibit some degree of extrarenal clearance, tubular secretion or reabsorption, or protein binding. Table 1 summarizes the consequences if some of the assumptions about the marker are not justified. In Table 1, a hernodynamically stable situation, calculation of CLpIasma according to an appropriate pharmacokinetic method, and complete bladder emptying for the calculation of CL,,,, are assumed.

Most of the protocols used for both renal and plasma clearance measurement are based upon the assumption that distribution of drug in the body is complete within the first 1-2 hours after a single injection of marker or institution of a continuous infusion. If not, errors will be produced because the process of distribution, not only elimination, re- mains an important determinant of the fall in plasma concentration of the marker. These errors are of greater magnitude when working with CLplasma methods because the sampling is usually started immediately. Most CL,,,, sampling is started after a delay in an attempt to sample only when distribution is complete. Distribution into certain tis-

Glomerular Filtration Rate in the Dog 403 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAIV bolus zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAI

el iminat ion zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAFig 1. A 3-compartment model in which a drug is distributed among compartments zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1, 2, and 3 and is eliminated via the central compartment. The movement of drug can be characterized by the transfer rate constants, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAk , ? and k Z , , where k I 2 denotes the rate constant associated with the uptake of drug into compartment 2 from compartment I , and k z , is the rate constant associated with the reverse process. The rate constant k , , , is associated with the loss of drug from compartment 1 by metabolism and excretion."

sues in the body may be very slow for some substances. In these cases, the initial rapid fall in plasma concentration is followed by a second process of slow distribution, after which the plasma concentration primarily is determined by the elimination process. This phenomenon is termed delayed distribution because distribution is delayed relative to the practical procedures commonly used for measuring clearance. If CLplnr;a is greater than CL,,,,,, this may indicate extrarenal excretion, delayed distribution, or a combination of factors (Table 1).

In practice, the exact properties of the marker often are not known. The risk of errors due to violation of the assumptions about the marker is somewhat greater when working with CLplarma because extrarenal clearance will not be detected. If there is extrarenal clearance of the marker, overestimation of GFR is most pronounced at low levels of renal function because extrarenal clearance then constitutes a larger proportion of the total clearance.

Pharmacokinetic Modeling

Common pharmacokinetic models, employed to calculate single dose AUC, can be explained in a simplified way. To facilitate mathematical analysis, the body can be concep- tualized as a series of discrete compartments, as illustrated for a 3-compartment model in Figure 1. After injection, the marker distributes into many physiological compartments, such as liver, adipose tissue, or connective tissue. The combination of compartments produces specific effects upon the log plasma concentration versus time curve. Mathematical- ly, it may appear as though distribution is confined mainly to a few compartments. Elimination primarily take place from the central compartment (compartment 1, Fig 1). The choice of model is based upon a visual or computerized evaluation of the fit of the curve to the data, using the semilogarithmic plot of plasma concentration versus time. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

0 10 20 30 40 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA50 60 70

Minutes after injection

Fig 2. Example of a plasma concentration versus time curve for an assumed I-compartment model after sampling for 1 hour. Mannitol after IV injection in a dog.If1

One-Compartment Model

The body can be viewed as a single compartment if immediate distribution is assumed to occur. This situation produces a monoexponential plasma disappearance curve. A 1-compartment model theoretically corresponds to a straight line (Fig 2) described by the monoexponential func- tion8.I0

C, = C, X e-kl,

in which case

AUC = C,/k,

where C, is defined as the plasma concentration at time t. Immediate distribution does not occur in reality. This error sometimes is considered negligible because the drug is rap- idly distributed, and the 1-compartment model is widely used in pharmacokinetic analyses of drugs. In the context of GFX estimation, however, the error produced will usually be clinically relevant.

Two-Compartment Model

be resolved into 2 straight lines (Fig 3)8J0: A 2-compartment model corresponds to a curve that can

C, = C, X c h l r + Cz X e-hZ1,

in which case

AUC = C,/h, + CZ/Az,

where C, = concentration at the y intercept of the line of slope -A , and C2 = concentration at the y intercept of the line of slope -Az. The plasma concentrations of the initial steep part of the curve are determined by distribution into body tissues, whereas the last part of the curve is primarily influenced by elimination from the body. Hall demonstrated that disturbances of ECFV may prevent resolution of the semilogarithmic curve into 2 straight lines, indicating that the 2-compartment model is inappropriate."

404 Heiene and Moe zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

\ 0.1 I I I I I I I 1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

0 50 100 150 200 250 300 350 400

Minutes after injection zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAFig 3. Plasma concentration versus time for an assumed 2-cornpart- ment model. Iodine after IV injection of iohexol in a dog (Heiene and Moe, in preparation). zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Multicompartmental Models

A 3-compartment model corresponds to a curve that can be resolved into 3 straight lines, and so on. Models involving more than 3 compartments and corresponding polyexponential formulas rarely are used, because the increased accuracy gained by the use of more complicated models is often considered unnecessary. A 5-compartment model of insulin kinetics has been described. Iz

Noncompartmental Models

A noncompartmental approach may sometimes be more accurate. The AUC is calculated using the trapezoidal method, adding the areas of each trapezoid defined by the curve (Fig zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA4) or by numerical integration of the curve.8J1.13 When the decline of drug is exponential, the corresponding log-trapezoidal method is more accurate. In this case, the terminal part of the curve from the last sample to 0 is estimated mathematically, assuming a monoexponential decline after the last sample. When AUC is calculated by the trapezoidal method but the estimated area is large because of early terminaticn of sampling, it has sometimes been called a modified 1-compartment method or area-modified single exponential approach.'J-16 The noncompartmental

l o I

0 50 100 150 200 250 300 350 400


Fig 4. Illustration of the log-trapezoidal method for calculating the area under the plasma concentration versus time curve. Iodine after IV injection of iohexol in a dog (Heiene and Moe, in preparation).

model is attractive because the number of theoretical compartments does not have to be defined. The model requires frequent initial sampling and sampling until most of the marker is eliminated.

Differences zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAin GFR Estimates by Various Models

Figures 2-4 illustrate how AUC is calculated using the different models and how the upper part of the initial AUC is lost when using the 1-compartment model. By starting sampling late, the data may seem to fit the 1-compartment model even if they do not.

Table 2 quantifies the differences produced in GFR estimates for 3 dogs by the use of different pharmacokinetic models when measuring plasma clearance of iohexol (Hei- ene and Moe, in preparation). The notations CLplvsma zo

CL,,,,,,, ,c and CL,,,,,, are used hereinafter to designate total clearance calculated from the plasma disappearance curve in a 2-compartment model, a 1-compartment model, and a noncompartmental AUC method, respectively.

Errors due to inappropriate choice of pharmacokinetic models seldom will be evident from the results, and knowledge about the assumptions behind each model is essential. The influence of the quality of the marker substance, as detailed in Table 1, should also be evaluated.

Table 2. Differences in calculated area under the curve (AUC) and estimated clearance (CL) when employing a 1- compartment model, 2-compartment model, or noncompartmental (trapezoidal) method after injection of iohexol in 3 dogs with similar body size and low, intermediate, or high renal function.

German Golden Alaskan Shepherd

Parameter Retriever Malamute Dog

Body weight (kg) Dose iohexol (mg iodine) AUC (monoexponential, mL.mgh) CL (1 compartment, rnL/min) AUC (biexponential, mLmg/h) CL ( 2 compartment, mL/min) AUC (trapezoidal, mL.mg/h) CL (noncompartmental, mWrnin)

31 I8.900

1101

1028

968

17.2

18.4

19.5

31 19,800

328

362

352

60.3

54.9

56.1

32 19,200

147 129.7 20 1 95.3

188 102.3

Glomerular Filtration Rate in the zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBADog zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA405 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAFurther Application of Noncompartmental Kinetics

The trapezoidal AUC method represents a different branch of pharmacokinetics, called noncompartmental or nonparametric methods, sometimes called model-independent methods because they are independent of a predefined number of compartments. They are not truly model independent because basic assumptions usually are made about constant clearance, linearity and time invariance, and a terminal monoexponential phase.l7-Iy

These basic assumptions also are important if the GFR values are related to ECFV. An estimate of GFRECFV is found without separate calculation of ECFV through the formulas involving the area under the first moment curve (AUMC) and mean residence time (MRT), which are important parameters in noncompartmental pharmacokinetics.20 Important relationships are GFRECFV = l/MRT, where MRT = AUMC/AUC.II Sometimes the mono- or polyexponential formulas are used in calculations involving AUMC and MRT, necessitating evaluation of the underlying assumptions for both the compartmental model and the noncompartmental approach.”

Use of Simplified Methods To avoid urine collection, CLpld,ma often is used instead

of CL,,,, to estimate GFR. The use of a 1-compartment model for the AUC calculation further simplifies the procedure because only 1 or 2 blood samples after the initial sample are needed. The basic assumption of the l-compartment model is immediate mixing into the distribution volume and a monoexponential plasma disappearance curve. This assumption is not really valid, and substantial errors are produced that necessitate correction for the GFR estimates to be valid.

A 1-compartment model will overestimate clearance because of underestimation of AUC because the upper part of the initial AUC is excluded from calculations (Figs 2- 4). The area not used seems small in the figures because of the logarithmic scale, but the magnitude of the error generally is 20-30% when working with filtration markers.2vz5 The size of the excluded area is not proportional to the level of renal function.2J This implies that the excluded area is relatively larger compared to the total AUC when AUC is small, that is, when elimination through the kidney is rapid. The magnitude of the error produced by the 1-compartment model thus tends to be larger when renal function is normal. This fact is illustrated in Table 2, where data from 3 dogs with varying renal function are compared (Heiene and Moe, in preparation).

Formulas that correct for this error are used extensively in human medicine, the most common being that of Br@- ~hner-Mortensen.’~ The reference clearance value predicted by this formula corresponds to the 3- or 4-compartment CLplasma value. This formula has been verified in several subsequent and is similar to corresponding formulas for the dog (Heiene and Moe, in pre~aration).~~]

Another approach commonly employed in human neph- rology calculates CLpla\mr from 1 blood sample only, by means of formulas involving an estimated V, for the marker.31.32 In theory, this 1 sample value could be influenced

by both V, and CL, reducing its usefulness in relation to GFR. Stake, however, demonstrated that a 50% increase or reduction in V, does not produce errors of substantial clinical relevance except in situations with very high or very low renal function.22

A reduction of the number of blood samples also is achieved by calculating the GFRECFV estimate from the terminal elimination rate constant, X2, by means of a correction formula.21 This formula corrects for violation of the assumption of equal concentrations of marker in plasma and interstitial fluid. The GFRECFV estimate has been ex- plored in the dog, and similar correction formulas have been developed.2

The 2-compartment model is commonly considered adequate, although 3-, 4-, and 5-compartment models or noncompartmental models may be more correct. The results from a 2-compartment analysis often are similar to results from a noncompartmental method, as illustrated in Table 2 (Heiene and Moe, in preparation). With current knowledge, it is not evident in which situations the 2-compartment model will be inappropriate, except for some cases of disturbances in ECFV.I1 Future research should clarify this is- sue. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

An alternative CLplaXma method is to use the classical CL,,,, formula but exchange the rate of excretion in urine with the rate of infusion, I X E where I = concentration of marker in the infusate and F = flow rate out of an infusion pump during continuous infusion of marker?

A very different CLplarma method is the noninvasive use of external counting by y-scintigraphy after injection of a radioactive filtration marker. The counts obtained are cali- brated to the results produced by a simultaneous reference method by means of regression analysis.33J4 The underlying assumption of this method is that the relationship between the methods is always constant.

Influence of Sampling Times Upon Plasma Clearance

Figures 5a and 5b are simulated plots of the log plasma concentration versus time curve, exemplifying potential errors due to inappropriate sampling times. A too early start of sampling for a 1-compartment model is illustrated in Figure 5a, and a too early termination of sampling for a 2- compartment model is shown in Figure 5b.

The curve may have a breaking point, when distribution is complete and the terminal monoexponential slope is reached (Fig 5a). When renal function is severely reduced, this point occurs later and at a higher plasma concentration.8 The breaking point is not always evident from a visual in- spection of the plot, but automated procedures in computer programs (eg,WinNonlin 1.0, SCI, Lexington, KY) will calculate the terminal slope from values that satisfy predefined criteria for being close to the monoexponential line. The terminal slope thus starts where the computer program se- lects its first value. Very slow distribution into some parts of body tissues, termed delayed distribution in Table 1, may induce a second, very late breaking point in the curve (Fig 5b). In both cases, the terminal slope is believed to be steeper than it is in reality. Thus, the calculated AUC will be smaller, and GFR will be overestimated. The risk is

406 Heiene and Moe zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1

0.1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA0 100 200 300 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA400 500 600 700


L

al zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAY,

E

g 1

L

0 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAC 0 ._ .- al

C zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA0 m E - a

0.1

0 100 200 300 400 500 600 700

Minutes after injection zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAFig 5. Hypothetical curves and sampling times, illustrating the errors produced in AUC estimates when employing a 1- and a 2-compartment model if distribution is assumed to be complete during sampling when it is not. + , plasma concentration at various time points, 0, the col- lected samples used for the calculation of AUC. -, lines used for the calculations. - - -, lines that should have been used for a more correct AUC estimate. (a) One-compartment model. The slope produced by the 2 samples is steeper than the true monoexponential slope. Sampling should not have been started before approximately 120 minutes. (b) Two-compartment model. The marker distributes slowly into some tissues. making late sampling and the use of a 3-compartment model (illustrated by the dotted lines), or alternatively a noncompartmental method, more correct than the 2-compartment model.

greater when using the 1-compartment model because less of the curve is known, but substantial errors may occur with all models.

The exact time when the terminal monoexponential slope starts is not always clear. In a few studies, curves from individual humans or dogs have been published, and for several markers the curves seem to be monoexponential after 1-2 hour^.^^.^^.^' In a group of 50 dogs with normal or moderately reduced renal function, the monoexponential decline for iohexol started after 0.5-4 hours (Heiene and Moe, in preparation). The time needed to reach the terminal slope was not directly proportional to the level of renal function.

Some studies indicate that inulin may have very slow distribution into parts of the body, reaching equilibration in the extracellular volume 10 hours or more after institution of a constant intravenous i n f ~ s i o n . ~ ~ . ) ~ Similar data are not available for other filtration markers, which have usually

been studied using single injection techniques. If renal function is normal or slightly reduced, most of the marker will be eliminated within 4-6 hours. When renal function is severely reduced, however, most of the marker is still in the body after 6 hours, and it may be necessary to take samples up to 24 or sometimes 48 hour^.^^-'^ zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Using the total curve

When using the 2- or 3-compartment models or the trapezoidal method, sampling ideally should continue until all detectable marker is eliminated. In most studies, however, sampling is terminated after 4-6 hours. The final slope after termination of sampling is estimated, assuming a monoexponential decline. The magnitude of potential error is re- flected by the size of the estimated area relative to the sam- pled part of AUC. The estimated area may be large if renal function is very poor.

Frequent sampling during the first half hour also is necessary. If sampling starts too late, an erronously small AUC will result in overestimation of clearance. There is no exact definition of the necessary number of samplings, although a minimum of 6-8 samples may be used as a rule of thumb. No advantage was found by increasing the numbers of samples from 7 to 10 when using the 2-compartment or trapezoidal method.4n

Using a Reduced Number of Samples

If a simplified model using the corrected 1-compartment estimate is employed, all samples must be taken after the terminal monoexponential phase is reached. Commonly used protocols involve sampling at 3-4 hours after injection of marker, and the situation in Figure 5a represents a typical error that may occur in this situation. Similar errors also may occur if a 1-compartment model is used in the situation with delayed distribution.

One to 4 samples commonly are used for the simplified approaches. One study in humans reported unacceptable errors by reducing the number of samples from 6 to 3 using a 1-compartment model:’ but the errors probably were due to sampling before distribution was complete for all indi- viduals (1, 1.5, and 2 hours). Reduction of the number of samples from 6 to 3 did not weaken the correlation between the total curve clearance estimate and the simplified model in 1 study in dogs3 Similarly, increasing the number of samples used for the 1-compartment model from 2 to 3 did not strengthen the relationship between the methods in another study in dogs (Heiene and Moe, in preparation). Dif- ferent sampling times were examined for a single sample approach when using CLplasma of iohexol in humans. The conclusions were that sampling 3 hours after injection can be recommended for assumed normal renal function, whereas 7 and 24 hours are recommended if renal function is moderately or severely reduced, re~pectively.~~

Standardization of Clearance Values

Clearance values are measured as milliliters per minute per dog, which is clearly difficult to interpret considering the range of body size in different breeds. Some way of relating clearance to body size is necessary to compare the

Glomerular Filtration Rate in zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAthe Dog 407

results to reference values and clearly define the level of renal function, which is the purpose of GFR estimation.

Glomerular filtration rate values most often are reported as milliliters per minute per kilogram in dogs, assuming that CL,,,, is linearly related to body weight. One study indicated that this is not the case for dogs weighing less than 10 kg or more than 50 kg.4L In humans, clearance values usually are related to body surface area (BSA) and reported as mL/min/m2 or zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAmL/min/1.73 m?. This often is considered appropriate because BSA correlates closely with basal met- abolic rate and also kidney size.” There are indications that correct BSA estimation is difficult in the dog,+42, possibly because of the different body shapes in different breeds.

Other authors argue that a more optimal approach is to relate clearance to ECFV, because the regulation of ECFV is closely related to glomerular filtration and GFR varies with ECFV under certain conditions.”.“ These authors have used the biexponential formulas in the noncompartmental model formulas. This approach is not a true standardization procedure because a change in ECFV leads to a change in GFR/ECFV even if GFR is unchanged. This method can be justified if GFR and ECFV usually vary in the same direction in health and disease or if the changes in ECFV are of little importance in clinical situations. These conditions should be studied experimentally before this procedure can gain general acceptance.

An easy solution to the important task of standardization of clearance values is not available because there is no per- fect standard. Nor at this point it is it obvious how much the various standardization methods will deviate from one other in dogs of different weights, with different diseases, with good and poor renal function, or with abnormal ECFV. In 1 study, substantial discrepancies among estimated levels of renal function were found when comparing clearance values related to body weight, BSA, or ECFV. The discrepancies were particularly evident in very large and very small dogs (Heiene and Moe, in preparation).

Method Comparison Studies and Statistical Analysis of the Results

A new diagnostic technique can be considered acceptable only when it is validated by comparison to an accepted reference method. The CL,,,, of inulin is considered the gold standard for measurement of GFR in the dog, although more recent studies may validate CL,,,,, of exogenous creatinine as a reference method in dogs under certain condi- t i o n ~ ? ~ However, these are questions about the qualities of inulin as the ideal marker.Jo Other methods commonly used in human medicine are sometimes used for reference, but these do not adequately validate any new method for use in dogs.

Whichever method is chosen for reference, the statistical evaluation of how the methods compare requires particular consideration. Problems arise when using statistical methods zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAof association for the evaluation of agreement between methods. With very few exceptions, method comparison studies for estimation of GFR in the dog make use of least squares regression analysis or the correlation coefficient. However, this approach is controversial.

Altman and Bland46 argued that both correlation coeffi-

cient and regression analysis are unsuitable for evaluating agreement between methods, for theoretical and mathematical reasons. Correlation is a measure of association, and it is wrong to infer from a high product-moment correlation coefficient, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAr, that the methods can be used interchangeably. Additionally, if the range of measurement for a parameter is large, the corresponding correlation coefficient will be greater than if the range of measurement is low. This phenomenon can be observed by performing the procedures on one half of a data set.

In regression analysis, even if the methods are equally accurate and adequate for the parameter in question, measurement error in the independent variable will lower the expected slope of the curve to less than 1. Although 2 methods may be acceptable, the regression line between the 2 methods may thus be different from the line of identity, depending on the amount of measurement error of the variable chosen as independent. This particular problem can be overcome by errors-in-variable approaches after duplicate sample measurements in the laboratory.47

Alternatively, Altman and Bland proposed a limits of agreement plot where the individual differences between the methods are plotted against their average?h The limits of agreement are the 95% confidence interval of the mean difference. These limits must be evaluated with respect to the clinical significance if a difference of that magnitude is found. This approach is becoming increasingly popular in method comparison studies. The plot simplifies the evaluation of the discrepancy between 2 methods because the differences are directly visualized and evaluated statistical- ly. A confidence interval for the differences is determined. The same difference may be read from a scatter plot of the values where the regression line and the line of identity are shown. However, the magnitude of the difference usually is small compared with the size of the plot, thus making it difficult to judge whether the difference will be of clinical relevance in individual patients.

Regression analysis is familiar to most researchers. A particular method of presentation cannot necessarily be dis- carded as wrong, but the strength and weaknesses of the different methods should be considered when performing and evaluating method comparison studies.

If the old method is known to be better than the new one, as when an approximate or simple method is compared with a very precise one, regression analysis can be used to find a predicted value of the established method from the value of the new method. This calibration approach does not address the question of agreement or comparability and can normally only be justified if it is established that the 2 methods measure the same parameter.

Methodologic Aspects and Results of GFR Studies in Dogs

Because GFR cannot be measured directly, reported values are estimates provided by clearance of the filtration marker. Results from individual studies in dogs are given in Table 3. Reported values often are similar, although substantial discrepancies have occurred in different studies. Several factors contribute to the observed discrepancies. In- dividual nonrenal influences on GFR may partly explain

408 Heiene and Moe zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBATable 3. means, mean zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA5 SEM, or mean (SD).

GFR values reported in healthy dogs and dogs with reduced or varying renal function. Values are reported as

Filtration Marker n Dogs Method Value Units Ref

Healthy Inulin

Creatinine Endogenous

Exogenous

12s'1311-iothalamate

Reduced or variable function Inulin

Creatinine Endogenous

Exogenous

Iohexol

12 12 47 36 5

36 10 12 51 48 36 36 10 30 30 10 25 5

12

6 14 5

42 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA5

12 24 10 78 6

11 14 6

42 8 6

78 6

24 11 14 5

11 12 16 16

I20 24 16 8

16

4.72 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA( 1 32) 3.39 (0.73)

4.60 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA% 0.15 41.8 (13.9) 26 % 3

3.7 (0.77) 2.97 (0.41) 3.49 (0.73)

2.61 4.10 IT 0.14 57.6 (9.3) 42.2 (13.3) 4.09 (0.50) 3.45 (0.7) 7.42 (2.22) 4.10 (0.62) 3.55 ? 0.14

2.8 5.60 (0.77)

17.2-39.2

24.9-70.3 6-5 1

9 c 1

0.1-4.7

40-1 10

2.40 C 0.83 50-191

18.0-40.1 48-130 41-130

1-76

41.02 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAt 0.70

1.48 2 0.34

1-84 40.54 2 0.70 2.34 ? 0.82

35-110

43-1 16

0.24-2.76

0.18-2.45

45-80

25.0-68.2

39-1 16 1.2-3.5 41-106

1-27

1.2-3.0 1.39 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAir 0.37

39-84

10-160

rnL/min/kg mL/min/kg mL/minikg rnL/min/m2 mL/min

mL/min/kg mL/min/kg mL/min/kg mL/min/kg mL/min/kg ,

mL/min/m2 mL/min/m2 mL/min/kg mL/min/kg mL/min/kg mL/min/kg mL/min/kg mL/min/kg mL/min/kg

mL/min mL/min mL/min mL/min mL/min

mL/min/kg mL/min/kg mL/min mL/min mL/min mL/mIn mL/min mL/rnin mL/min/kg mL/min/kg mL/min/kg mL/min mL/min mL/min/kg mL/min mL/min rnL/min mL/min mL/rnin

mL/minJm2 mL/min/LECF mL/min mL/min/kg mL/min/kg

mL/min/kg

15 33 4

48 81

49 2

33 41 4

49 49 2

65 65 2

14 53 15

6 11 68 16 81

34 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA64 10 45 6

25 71 7

16 85 65 45 7 64 69 11 68 25 71 23 23 3

30 23 85

mL/min/m; 23 cL,,a,ma 2c

n, number of dogs in the study; A, anesthetized; C, conscious; E, experimental dogs; H, home-held dogs; Ref, reference number; LECE liter of extracellular fluid.

Glomerular Filtration Rate in the Dog zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA409 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBATable zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA4. Advantages and disadvantages of various filtration markers in relation to practical procedures, laboratory analysis, toxicologic properties, compliance with the criteria for the ideal filtration marker (no extrarenal clearance, no tubular secretion or reabsorption, no binding to plasma proteins or erythrocytes, no effect on GFR, and not toxic).

Marker Advantages Disadvantages

Inulin

Creatinine

Reference substance, values are well

Seems to comply well with most criteria known from research

for the ideal marker

Renal clearance is validated by comparison to renal clearance of inulin

Laboratory analysis (Jaffe) is readily available in most laboratories

EDTA or DTPA Inert substances in the body Plasma clearance methods have been in-

vestigated

Iothalamate

Iohexol

Comparison to inulin has demonstrated

Plasma clearance methods have been in- good agreement

vestigated

Seems to fullfill the criteria for an ideal

Plasma clearance methods can be used Samples can be frozen or sent to labora-

Laboratory analysis is simple Simultaneous urograuhy is possible

marker in humans

tory by mail

High serum levels are usually established by continous infusion

Cumbersome laboratory analysis, proteins and glucose must be removed prior to analysis to avoid interference with inulin analysis

Late distribution may require long-lasting sampling protocols Possible influences upon creatinine levels by endogenous processes in the

Interference by endogenous substances when measuring creatinine by the

More specific laboratory analysis is not readily available, necessitating use

Facilities for preparation and analysis of radioactive drugs

body cannot always be excluded

Jaffe reaction

of exogenous creatinine

must be available

Minor errors in results due to protein binding, chelate instability, or extra-

Inaccurate dose calculation, decay correction, and counting represent po-

Few laboratories perform the analysis

If used in small doses as a radionuclide marker; see EDTA

renal clearance cannot be excluded

tential errors

If used in contrast medium doses: potentially toxic effects in the body Analysis is not readily available in laboratories

May represent toxic strain on renal tissue in rare cases

why the range of GFR values is quite wide in healthy dogs. The filtration marker does not always meet all criteria for the optimal filtration marker. Some of the theoretical and practical advantages or disadvantages of specific markers are stated in Table 4. The choice of pharmacokinetic model influences the results, and in Table 3 the pharmacokinetic model used for the analysis is given.

The groups of dogs in the various studies were different and sometimes poorly defined. Reported values are given in Table 3. Healthy dogs sometimes come from the regular clinical dog population and sometimes from a colony. of young experimental dogs. In dogs with renal disease, the results may be influenced by the type of renal disease and other factors such as abnormal blood pressure or changes in ECFV. Most studies provided values from experimentally induced renal failure only, and these results cannot be directly extrapolated to dogs with clinical diseases. In studies where clinical cases were studied, the type of renal disease was not always reported. Breed, gender, and age variation also may have influenced the results. Old dogs may have chronic lesions in their and medication may change GFR. Because anesthesia may decrease GFR in normal the state of consciousness is given in Table 3. Several studies report values from a small number of dogs only, adding uncertainty to the results.

Inulin zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAMethodologic Aspects

The CL,,,, of inulin has for decades been considered the reference method for estimation of GFR in humans and dog^.^.'^,^ The classical inulin method is complicated and therefore only performed in research laboratories. When used, precautions must be taken against interference from glucose and other endogenous substances.' A high plasma concentration of inulin is established to avoid significant interference from glucose. High plasma concentration is achieved by constant infusion rather than a single injection because of controversy regarding whether inulin would dis- play saturation kinetics at high dosages.4O Sampling is started after distribution is assumed to be complete, usually 1- 2 hours after institution of the inulin injection.

Some researchers have used radiolabeled 14C-inulin, measuring the radioactivity rather than inulin itself. Poten- tial errors primarily related to dissociation of the p-emittor I4C from inulin may occur, but the method has produced results similar to those from the traditional inulin method. Enzymatic methods of analysis have been introduced to avoid interference from endogenous s ~ b s t a n c e s . ~ ~ - ~ ~ The CL,,,,,, methods have not as yet gained acceptance as sub-

410 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAHeiene and Moe zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAstitues for the CL,,,, of inulin because agreement between the methods has not been convincing.16 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Results

In 2 studies involving healthy dogs, CLEnal of 14C-inulin averaged 2.8 mL/minkg in 5 anesthetized dogs and 4.1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAmLI minkg in another group of 10 unanesthetized The mean CLplarma 2c of 14C-inulin in 25 unanesthetized healthy dogs was 3.77 mumidkg in a study where no comparison to CL,,,, was done.14

The effect of gender was studied in 47 healthy dogs where the average CL,,,,, of IT-inulin was 4.60 mL/min/ kg.' No significant difference in clearance values was found between genders, although the repeatability of both methods was higher in female dogs. In the same study, CL,,, of inulin, CL,,, of endogenous creatinine, and CLplprm as measured by the infusion rate related to plasma concentration of inulin were not interchangable numerically. For in- din, CL,,,,,, was higher than CL,,,,,, which may be explained by extrarenal clearance or delayed distribution of inulin.

Extrarenal clearance of inulin can be ignored, but similar findings in humans point to possible distribution of inulin for at least 10-15 hours after injection's.~6.'n.s~. This important question should be adressed in future research.

In 42 dogs with reduced renal mass, the CL,,,,, of inulin was approximately 10% higher than the CL,,,,, of creatinine. I h Late occurrence of complete distribution for inulin could explain some of the differences observed in these studies. The correlation between the methods was lower at clearance values below 1.0 mL/min/kg, which may be explained by the relatively short sampling times used, 75 minutes. This requires large estimated areas in AUC at reduced renal function.

Creatinine

Methodologie Aspects

The body produces creatinine at a constant rate, and creatinine is excreted by glomerular filtration. Serum creatinine concentration therefore is commonly used in clinical work to estimate GFR, but it does not increase above reference values before renal function is severely reduced. Once serum creatinine concentration increases substantially, however, values are inversely correlated with the level of reduction in GFR6.s5.s6 Measurement of creatinine clearance provides a more sensitive estimate of GFR.

The qualities of creatinine as a filtration marker have been questioned for several reasons. Creatinine concentration in plasma is influenced by extrarenal factors such as muscle mass, body weight, food intake, and metabolism by intestinal b a ~ t e r i a ? ~ . ~ ~ Some studies suggest that it also is secreted in the renal Results were available 40 years ago that both favored and disfavored use of creatinine for GFR estimation in the dog.s8 For practical reasons, it became commonly used in canine rned i~ ine ,~~ with GFR measured as the CL,,,, of either endogenous or exogenous ~reatinine.'.'.'~

Most laboratories measure creatinine with a modified Jaf- fe method. The results may be influenced by the serum

content of noncreatinine chromogens, which may be measured as creatinine in the analysis. The measured CLrenul may thus be falsely low because of erroneously high values for plasma creatinine concentration. Additional negative bias also may be introduced in the analysis by various other substances present in serum.60.6' Injection of exogenous creatinine has been widely used for measurement of renal clearance to insure plasma concentrations of a magnitude sufficient to avoid interference from endogenous substances in ~ l a s m a . ' . ~ ~ . ~ ~ For measurement of CLplr,mu by analysis of the plasma disappearence curve, enough creatinine must be injected to avoid influence by endogenous creatinine on the plasma concentration versus time curve.

Results

Exogenous creatinine clearance was more closely correlated to inulin clearance than to endogenous creatinine in some studies,'J whereas other researchers found endogenous creatinine clearance very similar to inulin clearance in anaesthetized dogs.j9 Renal clearance of creatinine was 2.5-4.5 mL/minkg in healthy dogs (Table 3).

One recent study indicated that creatinine was a better filtration marker in the dog than suggested by the early studies, which often based their conclusions on measurements of few dogsa5 The increasing tubular secretion of creatinine with decreasing renal function found in humansb3 apparently does not occur in dogses A close correlation and a ratio close to 1.0 were found between CL,,,, of exogenous creatinine and CL,,,, of IT-inulin when comparing 192 GFR measurements in 78 dogs with a broad range of renal functi0n.4~ In 7 dogs with experimentally induced acute renal failure, the ratio of CL,,,,, of endogenous creatinine to CL,,,,, of inulin was 0.8." A possible explanation is that a modified Jaffe reaction was used for the analysis of creatinine. More specific laboratory analysis of creatinine produced more reliable values for endogenous creatinine clearance, which then was found to be similar to exogenous creatinine clearance through 11 1 measurements of CL,,,, in 24 dogs." CLplarma zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAzc compared with CL,,,, of exogenous creatinine measured in 30 healthy dogs and analyzed by the Jaffe method, with samples taken until 75 minutes after injection, produced mean CLplasma values twice those of CL,,,,,, and poor correlation was found between the meth- o d ~ . ~ ~ zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAThis severe discrepancy may have resulted from early termination of sampling. As expected, the overestimation by CL,,,,, was greater at normal renal function. To con- clude, in spite of considerable controversy regarding creatinine as a filtration marker, recent research indicates that CL,,,,, of both endogenous and exogenous creatinine clearance produces valid GFR estimates in the dog.

Radiolabeled Filtration Markers

Methodologic Aspects

or 1311-iothalamate, s'Cr-ethylenediaminetetraacetic acid (EDTA) and 99qc-diethylenetriarninepentaacetic acid (DTPA) are widely accepted as useful markers of GFR in the dog, based upon research in the dog and upon extrap- olation of findings in humans, where differences between the results from various radionuclides typically have been

Glomerular Filtration Rate in the Dog zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA41 1

less than y9mTc-DTPA has been investigated extensively in the dog and is available at several veterinary institutions today. There are some specific problems related to work with radionuclides, among which the stability of the chelates is of crucial importance. Correct calculation of the injected dose usually is insured by weighing and counting of the syringe before and after injection, and decay correction must be performed. The main limitation of these methods is the need for a laboratory working with nuclear medicine. Although individual measurements are not very expensive, adequate facilities rarely are available outside of academic institutions and research laboratories.

and "lI-iothalamate were among the first nuclides to be investigated for GFR estimation.I5 Iothalamate is an old radiographic contrast agent of higher potential toxicity than newer contrast agents. The iodinated nuclides have long half-lives, making them undesirable for safety reasons. In current research, other radiolabeled filtration markers are used more commonly.

Early studies with y')mT~-DTPA may have been influenced by the fact that some commercial preparations of yymTc- DTPA gave low plasma clearance values, possibly as a result of protein binding and chelate i n ~ t a b i l i t y . ~ ~ . ~ ~ . ~ ~ These problems presently have been largely overcome.J4 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

Results

In 14 dogs, the CLplrrmd of 1251-iothalamate calculated in 3 different ways was compared with the CL,,,,, of inu1in.l' The CLpla,ma Ic overestimated GFR by approximately 30%, whereas CLplrrmlZc was similar to CLplarmdAUC in normal dogs but overestimated GFR in 3 dogs with expanded ECFV. The iothalamate ClplasmaAUc was very close to the CL,,,, of inulin, also in animals with expanded ECFV. These findings indicate that use of the 2-compartment method rather than extrarenal clearance of iothalamate was the explanation for the discrepancy between CL,,,, and Clplasml zc. Mean values up to 4.2 mL/min/kg were found in this study. In another study, mean values of 5.60 mL/min/kg were reported for healthy dogs.ls In zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA5 dogs, CL,,,,, of 12SI-iothalamate was identical to CL,,,, of inulin.hR However, in 2 studies involving 11 and 12 dogs where CL,,,, and CLplaama of radiolabeled iothalamate were measured, CLpla,ms zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAAUC was 13% and 17% higher than the CLrena1.6y.70 This finding may be explained by some extrarenal clearance of iothalamate in the dog, early termination of sampling, or chelate instability.

The CLphSma zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAZr of SICr-EDTA was measured in 34 normal dogs, and in 12 of them in which CL,,,,, of endogenous creatinine was measured, a significant correlation was found?, Glomerular filtration rate estimates were approximately 3 mL/min/kg. The CLplasma I c of 51Cr-EDTA was measured in 48 dogs with renal insufficiency.56 In these 2 studies, normal clearance values were linearly related to body weight except in very large and very small dogs. The overall fractional turnover rate constant of 51Cr-EDTA correlated closely with CLpla,mr Zc. However, the rate constant also is influenced by V,, and it cannot be recommended universally as a replacement for clearance values as an index of renal function in all populations.

In 5 dogs, CL,,,, of y9mTc-DTPA was equal to the CL,,,,,

of In 1 study of 12 dogs, the CLpla\mr zc of yymT~- DTPA was approximately 10% higher than the CL,,,, of y9mT~-DTPA and was similar to the CL,,,, of i~thalamate.~' In a 2nd similarly designed study of 11 dogs, the CL,,,,, zc of 99mT~-DTPA was 12% higher than the CL,,,,, of iothalamate.25 In the same 2 studies, CLplarmn zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAl c values of yymT~- DTPA were approximately 30% higher than CLpls,ma zc Val- ues when sampling was continued until 4 hours after injection. Plasma clearance of yymT~-DTPA was measured by external counting in a study where clearance values were related to ECFV.' The values reported were approximately 5 times the values in milliliters per minute per kilogram, which was expected because ECFV is roughly 20% of body weight.

Iodinated Contrast Media Methodologic Aspects

Nonradioactive iodinated contrast media are easy to use in the clinic and laboratory. They can be used for urography, which sometimes is also indicated in patients with renal disease. Iothalamate, ditrizoate and sodium meglumine have been used as GFR markers in the dog, with laboratory analysis by X-ray fluorescence, 25.7?J3 but toxicity limits the use of these agents. The potential for induction of acute renal failure by use of contrast medium has been a controversial question. The new generation contrast media such as iohexol, iopamidol, and iodixanol are nonionic, low in osmolality, and safer than the older agents.

Iohexol has been evaluated as a filtration marker in the dog. In 3 dogs, 98% of iohexol was excreted unmetabolized in the urine.74 Some species differences may be present with respect to the amount excreted in feces and in urine, but the differences are usually 1-5%.74.75 In human patients with severely impaired renal function, only 6% of the iohexol was excreted in feces after 5 days.38

In several reports in humans, including large prospective studies, iohexol was considered safe even in high-risk pa- t i e n t ~ . ~ ~ . ~ " ~ ~ Its use in patients with poor renal function is well d o c ~ m e n t e d . ~ ~ . ~ ~ In 1 study, the renal effects of iohexol in the dog were mediated by vasodilatation or vasoconstric- tion.R' In dogs with normal renal function, both renal plasma flow and GFR increased after iohexol injection. In dogs with experimentally induced renal insufficiency, low doses of iohexol did not alter these parameters, but higher doses induced transient reductions in renal plasma flow and GFR.

Iodine content can be measured by a simple and accurate X-ray flourescence method (Renalyzer, Provalid AB, Swe- den). Alternative methods of analysis are high pressure liq- uid chromatography, which is labor intensive but allows lower iohexol dosages and smaller sample volumes to be used,*' a ceric arsenite method,R3 or capillary electropho- re~ is . *~ The iodine is stable in the sample, allowing samples to be frozen for several weeks or sent by mail, which simplifies procedures and is especially attractive in small animal practice, because serum samples may be sent to a central laboratory.

Results

On 18 dogs with a wide range of renal function, simultaneous measurements of the plasma disappearance of io-

412 Heiene and Moe

hexol and y9mT~-DTPA were performed.27 The methods followed each other closely, but the zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBACLplusmn zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAzc of 99mT~-DTPA was 15% higher than the CLplaFma zc of iohexol throughout the range of measurement. This discrepancy might be due to extrarenal clearance of Tc-DTPA in the dog or to systematic measurement error, but other explanations are possible. The CLplaama lc was on average 20% higher than CLplasmn zc for both substances when samples were taken until the tracer was below detection level (6-24 hours). Slightly higher estimates of GFR by wmT~-DTPA than by iohexol also were found in 24 dogs with known or suspected renal disease.jo Various simplified pharmacokinetic models were compared with the 2-compartment model, and the results indicate that the correction formulas used in humans also can be used in dogs.

In 7 nephrectomized dogs and 1 normal dog, CLplasmaAUC, CLplasma zc CLplasmalc corrected by the Brgichner-Mortensen formula and uncorrected CLpla5ma , c were compared with CL,,,, of exogenous ~rea t in ine .~~ The clearance values produced by creatinine and iohexol were very similar. The differences between the pharmacokinetic models were very small, probably because most of the dogs had markedly reduced renal function. Several normal cats were used in the same study, where the uncorrected 1-compartment model overestimated zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAGFR as expected. CLplasma Ic corrected by the Brgichner-Mortensen correction formula was measured in 47 healthy dogs using ditrizoate and iopamidol.86 The mean GFR value was 4.24 &midkg (range 3.0-6.5 mL/ minkg), but there was no comparison to reference methods.

Conclusions The CL,,,, of inulid usually is regarded as the reference

method for estimation of GFR, despite some unresolved methodologic questions. The simplicity of various CLplasma methods make them attractive, and some authors have argued that iohexol could replace inulin as a reference substance.''.'' The position of inulin as a gold standard for measurement of GFR thus is challenged, but alternative reference methods need further justification in research. Well- designed method comparison studies and pharmakokinetic studies could bring new insight into the question of an optimal filtration marker. Standardization to body size and optimal sampling times in various models need elucidation. Reference values in healthy dogs should be established, and correction formulas for a reduced number of samples should be calculated in normal dogs and dogs with renal disease.

References I . Levey AS. Use of glomerular filtration rate measurements to as-

sess the progression of renal disease. Semin Nephrol 1989;9:370-379. 2. Finco DR, Coulter DB, Barsanti JA. Simple, accurate method for

clinical estimation of glomerular filtration rate in the dog. Am J Vet Res 198 1 ;42: 1874-1877.

3. Gleadhill A, Peters AM, Mitchell AR. A simple method for measuring glomerular filtration rate in dogs. Res Vet Sci 1995;58:118- 123.

4. Izzat NN, Rosborough JP. Renal function in conscious dogs: Po- tential effect of gender on measurement. Res Exp Med 1989; l89:37 l- 379. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

5. Koopman MG, Koomen GCM, van Acker BAC, et al. Circadian

rhythm in glomerular transport of macromolecules through large port and shunt pathway. Kidney Int 1996;49: 1242-1249. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

6 . Robinson TR, Harbison MH, Bovee KC. Influence of reduce renal mass on tubular secretion of creatinine in the dog. Am J Vet RI 1974;35:487-491.

7. Tabaru H, Finco DR, Brown SA, et al. Influence of hydratic state on renal function of dogs. Am J Vet Res 1993;54:1758-1764.

8. Rowland M, Tozer TN. Clinical Pharmacokinetics: Concepts ar Applications, 3rd ed. Baltimore, MD: Williams &Wilkins; 1995.

9. Smith HW. The Kidney. Structure and Function in Health ar Disease. New York, NY Oxford University Press; 1951.

10. Sapirstein LA, Vidt DG, Mandel MJ, et al. Volumes of distr bution and clearances of intravenously injected creatinine in the do, Am J Physiol 1955; 18 1 :330-336.

11. Hall JE, Guyton AC, Farr BM. A single-injection method fc measuring glomerular filtration rate. Am J Physiol 1977;232:F72-F7

12. Hovorka R, Powrie JK, Smith GD, et al. Five-compartme model of insulin kinetics and its use to investigate action of chlorc quine in NIDDM. Am J Physiol 1993;265:E162-E175,

13. Purves RD. Numerical estimation of the noncompartment pharmacokinetic parameters variance and coefficient of variation I

residence times. J Pharm Sci 1994;83:202-205. 14. Fettman MJ, Allen TA, Wilke WL, et al. Single-injection met1

od for evaluation of renal function with I4C-inulin and 'H-tetraeth lammonium bromide in dogs and cats. Am J Vet Res 1985;46:48; 485.

15. Powers TE. Powers JD, Garg RC. Study of the double isotol single-injection method for estimation of renal function in purebrt Beagle dogs. J Am Vet Med Assoc 1977;38:1933-1936.

16. Brown SA. Evaluation of a single-injection method for esf

mating glomerular filtration rate in dogs with reduced renal functio Am J Vet Res 1994;55:1470-1473.

17. Gillespie WR. Noncompartmental versus compartmental mol elling in clinical pharmacokinetics. Clin Pharmacokinet 199 I ;20:25: 262.

18. Veng-Pedersen I? Mean time parameters in pharmacokinetic Definition, computation and clinical implications (Part 1). Clin Pha macokinet 1989;17:345-366.

19. Veng-Pedersen zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA€? Mean time parameters in pharmacokinetic Definition, computation and clinical implications (Part 2). Clin Phz macokinet 1989;17:424-440.

20. Yamaoka K, Nakagawa T, Uno T. Statistical moments in phz macokinetics. J Pharm Biopharm 1978;6:547-558.

21. Peters AM. Expressing glomerular filtration rate in terms extracellular fluid volume. Nephrol Dial Transplant 1992;7:205-2 IC

22. Riegelman S, Collier P. The application of statistical mome theory to the evaluation of in vivo dissolution time and absorptic time. J Pharm Biopharm 1980;8:509-533.

23. Moe L, Heiene R. Estimation of glomerular filtration rate dogs with wmTc-DTPA and iohexol. Res Vet Sci 1995;58:138-143.

24. Brochner-Mortensen J. A simple method for the determinatic of glomerular filtration rate. Scand J Clin Lab Invest 1972;30:27 274.

25. Powers TA, Stone WA, Grove RB, et al. Radionuclide me surement of differential glomerular filtration rate. Invest Radiol 198 16:59-64.

26. Effersoe H, Rosenkilde P, Groth S, et al. Measurement of ren function with iohexol: A comparison of iohexol, WmTc-DTPA. ai "Cr-EDTA clearance. Invest Radiol 1990;25:778-782.

27. Gaspari E Perico N, Ruggenenti F! et al. Plasma clearance nonradioactive iohexol as a measure of glomerular filtration rate. J A SOC Nehprol 1995;6:257-263.

28. Rehling M, Rabal A. Measurement of glomerular filtration r: in adults: Accuracy of five single-sample plasma clearance methoc Clin Physiol 1989;9:171-182.

29. Sterner G, Frennby B, Hultberg B, et al. Iohexol clearance f

Glomerular Filtration Rate in the Dog 413

GFR-determination in renal failure-Single or multiple plasma sampling? Nephrol Dial Transplant 1996;l I 321-525.

30. Gleadhill A. Evaluation of iohexol zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAas a marker for the clinical measurement of glomerular filtration rate in dogs. Res Vet Sci 1996;

31. Jacobsson L. A method for the calculation of renal clearance based on a single plasma sample. Clin Physiol 1983;3:297-305.

32. Stake G. Monclair T. A single plasma sample method for estimation of the glomerular filtration rate in infants and children using iohexol, I: Establishment of a body weight-related formula for the distribution volume of iohexol. Scand J Clin Lab Invest 1991;51:335- 342.

33. Krawiec DR, Badertscher RR, Twardock AR, et al. Evaluation of zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA"""'Tc-diethylenetriaminepentaacetic acid nuclear imaging for quan- titative determination of the glomerular filtration rate of dogs. Am J Vet Res 1986;47:2175-2179.

34. Krawiec DR. Twardock AR, Badertscher RR, et zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAal. Use for qq"'Tc diethylenetriaminepentaacetic acid for assessment of renal function in dogs with suspected renal disease. J Am Vet Med Assoc 1988;192: 1077-1080.

35. Hellerstein S. Berenbom M. Alon U. et al. The renal clearance and infusion clearance of inulin are similar, but not identical. Kidney Int 1993;44: zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1058-106 I .

36. Estelberger W, Petek W, Zitta S. et al. Determination of glomerular filtration rate by identification of sinistrin kinetics. Eur J Clin Chem Clin Biochem 1995;33:201-209.

37. Stake G, Monn E, Rootwelt K, et al. The clearance of iohexol as a measure of the glomerular filtration rate in children with chronic renal failure. Scand J Clin Lab Invest 1991;51:729-734.

38. Nossen J 0 , Jacobsen JA. Kjiersgaard P. et al. Elimination of the non-ionic X-ray contrast media iodixanol and iohexol in patients with severely impaired renal function. Scand J Clin Lab Invest 1995;55:

39. Frennby B, Sterner G, Almen T, et al. The use of iohexol clearance to determine GFR in patients with severe chronic renal failure- A comparison between different clearance techniques. Clin Nephrol 1995:43:35-46.

40. Florijn KW, Barendregt JN, Lentjes EG, et al. Glomerular filtration rate measurement by "single-shot" injection of inulin. Kidney Int 1994;46:252-259.

41. van den Brom WE, Biewenga WJ. Assessment of glomerular filtration rate in normal dogs: Analysis of the Tr-EDTA clearance and its relation to several endogenous parameters of glomerular filtration. Res Vet Sci 1981:30152-157.

42. Page RL. Macey DW, Thrall DE. Unexpected toxicity associated with body surface area for dosing melphalan in the dog. Cancer Res 1988;48:288-290.

43. Arrington KA, Legendre AM. Tabeling GS. et al. Comparison of body surface area-based and weight-based dosage protocols for doxorubicin administration in dogs. Am J Vet Res 1994;55:1587- 1592.

44. Peters AM. Quantification of renal haemodynamics with radionuclides. Eur J Nucl Med 1991;18:274-286.

45. Finco DR, Brown SA. Crowell WA. et al. Exogenous creatinine clearance as a measure of glomerular filtration rate in dogs with reduced renal mass. Am J Vet Res 1991;52:1029-1032.

46. Altman DG, Bland JM. Measurement in medicine: The analysis of method comparison studies. Statistician 1983;32:307-3 17.

47. Strike PW. Assay method comparison studies. In: Strike PW. ed. Statistical Methods in Laboratory Medicine. Oxford, England: Butter- worth-Heinemann L td 1991:307-330.

48. Cowgill LD. Spangler WL. Renal insufficiency in geriatric dogs. Vet Clin North Am Small Anim Med 1981:11:727-748.

49. Bovee KC. Joyce T. Clinical evaluation of glomerular function: 24-Hour creatinine clearance in dogs. J Am Vet Med Assoc 1979;174: 488-49 I .

60:117-121.

341-350.

50. Degenaar CP, Frenken LAM, van Hooff JF! Enzymatic method for determination of inulin. Clin Chem 1987;33: 1070-1071.

5 1. Soper CPR, Bending MR. Barron JL. An automated enzymatic inulin assay, capable of full sinistrin hydrolysis. Eur J Clin Chem Clin Biochem 1995;34:275.

52. Sugita zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA0. Tomiyama Y, Matsuto T. et al. A new enzymatic method for the determination of inulin. Ann Clin Biochem 1995;32:561- 565.

53. Klopper JE Hauser W, Atkins HL. et al. Evaluation of yvn'Tc- DTPA for the measurement of glomerular filtration rate. J Nucl Med 1972; 13: 107-1 10.

54. Rehling M. M~ l le r ML. Thamdrup B. et al. Simultaneous measurement of renal clearance and plasma clearance of ""'Tc-labelled diethylenetriaminepenta-acetate, 5'Cr-labelled ethylenediaminetetra-acetate and inulin in man. Clin Sci 1984:66:613-619.

55. Finco DR, Brown SA, Vaden SL. et al. Relationship between plasma creatinine concentration and glomerular filtration rate in dogs. J Vet Pharmacol Ther 1995; 18:4 18-42 I . zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA

56. Biewenga WJ, van den Brom WE. Assessment of glomerular filtration rate in dogs with renal insufficiency: Analysis of the "Cr- EDTA clearance and its relation to the plasma concentrations of urea and creatinine. Res Vet Sci 1981:30:158-160.

57. Finco DR. Duncan JR. Blood urea nitrogen and serum creatinine concentrations as indicators of renal dysfunction: A study of I 11 cases and a review of related literature. J Am Vet Med Assoc 1976:168: 593-601.

58. Ladd M, Liddle L, Gagnon JA. Renal excretion of inulin. creatinine and ferrocyanide. at normal and reduced clearance levels in the dog. Am J Physiol 1956;184:505-514,

59. DiBartola SP. Clinical approach and laboratory evaluation of renal disease In: Ettinger SJ, Feldman EC, eds. Textbook of Veterinary Internal Medicine, 4th ed. Philadelphia. PA: WB Saunders; 1995:

60. Jacobs RM, Lumsden JH, Taylor JA, et al. Effects of interferents on the kinetic Jaffe reaction and an enzymatic colorimetric test for serum creatinine concentration determination in cats. cows. dogs and horses. Can J Vet Res 1991;55:150-154.

61. Pemne RD, Madias NE. Levey AS. Serum creatinine as an index of renal function: New insights into old concepts. Clin Chem 1992;38: 1933-1953.

62. Lee K. Behrendt U, Kaczmarczyk G. et al. Estimation of glomerular filtration rate in conscious dogs following a bolus of creatinine. Comparison with simultaneously determined inulin clearance. Pflugers Arch 1983;396:176-178.

63. Shemesh 0, Golbetz H, Kriss JP. et al. Limitations of creatinine as a filtration marker in glomerulopathic patients. Kidney Int 1985:28: 830-838. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA64. Finco DR, Tabaru H, Brown SA. et al. Endogenous creatinine

clearance measurement of glomerular filtration rate in dogs. Am J Vet Res 1993;54: 1575-1578.

65. Labato MA. Ross LA. Plasma disappearance of creatinine as a renal function test in the dog. Res Vet Sci 1991;50253-258.

66. Rootwelt K, Falch D. Sjokvist R. Determination of glomerular filtration rate (GFR) by analysis of capillary blood after single shot injection of ""Tc-DTPA. Eur J Nucl Med 19805:97-102.

67. Carlsen JE, M~ l le r ML, Lund JO, et al. Comparison of four commercial Tc-yyn'(Sn) DTPA preparations used for the measurement of glomerular filtration rate: Concise communication. J Nucl Med 1980:21:126-129.

68. Barbour GL. Crumb CK, Boyd CM. et al. Comparison of inulin. iothalamate, and """Tc-DTPA for measurement of glomerular filtration rate. J Nucl Med 1976;17:317-320.

69. Pihl B. Nosslin B. The single injection technique for determination of renal clearance I. Clearance of iothalamate and iodohippurate in dogs. Scand J Urol Nephrol 1974;8:138-146.

70. Pihl B. The single injection technique for determination of renal

1706-1 7 19.

414 Heiene and Moe

clearance V. A comparison with the continous infusion technique in the dog and in man. Scand J Urol Nephrol 1974;8:147-154.

7 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA1 . Powers TA. Stone WJ. Lowe BA, et al. Radionuclide measurement of differential glomerular filtration rate in urinary tract obstruc- tion. Invest Radiol 1982:17:417-421,

72. Brown SCW, O’Reilly PH. Iohexol clearance for the determination of glomerular filtration rate in clinical practice: Evidence for a new gold standard. J Urol 1991:146:675-679.

73. Westhoff A. Meyer-Lindenberg A, Wohlsein F‘, et al. Determi- nation of the glomerular filtration rate (GFR) in dog by an iodine contrast medium with the Renalyzer-PRX-90. Monatsh Veterinikmed 1993:48:573-582.

74. Miitzel W. Speck zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAU. Pharmacokinetics and biotransformation of iohexol in the rat and the dog. Acta Radiol 1980;335(Suppl):87-92.

75. Ohlsson B, A u k A, Sveen K. et al. Human pharmacokinetics of iohexol. A new nonionic contrast medium. Invest Radiology 1983; 18: 177-182.

76. Katayama H. Yamaguchi K. Takashima T, et al. Adverse reac- tions to ionic and nonionic contrast media: A report from the Japanese committee on the safety of contrast media. Radiolology 1990;175:621- 628.

77. Golman K. Holtz E. AlmCn T. Radiographic contrast media. In: Bach PH, Lock EA. eds. Nephrotoxicity in the Experimental and Clin- ical Situation. Dordrecht: Martinus Nijhoff Publishers; 1987:701-725.

78. O’Reilly PH. Jones DA. Farah NB. Measurement of the plasma

clearance of urographic contrast media for the determination of glomerular filtration rate. J Urol 1988;139:9-11.

79. Rudnick MR, Goldfarb S, Wexler L, et al. Nephrotoxicity of ionic and nonionic contrast media in 1196 patients: A randomized trial. Kidney Int 1995;47:254-261.

80. Stake G. Monn E, Rootwelt K, et al. Influence of urography on renal function in children. A double blind study with metrizoate and iohexol. Acta Radiol 1989;30:643-646.

81. Arakawa K, Suzuki H, Naitoh M, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAet al. Role of adenosine in the renal responses to contrast medium. Kidney Int 1996;49: 1199- 1206.

82. KrutzCn E, Back SE, Nilsson-Ehle zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBAI? Determination of glomerular filtration rate using iohexol clearance and capillary sampling. Scand J Clin Lab Invest 1990:50:279-283.

83. Back SE, Masson F‘, Nilsson-Ehle P. A simple chemical method for the quantification of the contrast agent iohexol. applicable to glomerular filtration rate measurements. Scand J Clin Lab Invest 1988;

84. Rocco MW, Buckalew VM, Moore LC, et al. Capillary electro- phoresis for the determination of glomerular filtration rate using nonradioactive iohexol. Am J Kidney Dis l996;28: 173-177.

85. Brown SA, Finco DR, Boudinot FD, et al. Evaluation of a single injection method, using iohexol, for estimating glomerular filtration rate in cats and dogs. Am J Vet Res 1996:57:105-110.

86. Westhoff A, Meyer-Lindenberg A, Nolte I , et al. Eignung und Aussagekraft der Jodkontrastmittel-Clearance zur Messung der Glom- eruliken Filtrationsrate beim Hund. Kleintierpraxis I994;39:593-606.

48:825-829.

Pharmacokineatic Aspects of Measurement of Glomerular Filtration Rate in the Dog: A Review

Documents