Validated methods of estimating glomerular filtration rate (GFR) in cats requiring only a limited number of samples are desirable.
Validated methods of estimating glomerular filtration rate (GFR) in cats requiring only a limited number of samples are desirable.
To test a single sample method of determining GFR in cats.
The validation population (group 1) consisted of 89 client-owned cats (73 nonazotemic and 16 azotemic). A separate population of 18 healthy nonazotemic cats (group 2) was used to test the methods.
Glomerular filtration rate was determined in group 1 using corrected slope-intercept iohexol clearance. Single sample clearance was determined using the Jacobsson and modified Jacobsson methods and validated against slope-intercept clearance. Extracellular fluid volume (ECFV) was determined from slope-intercept clearance with correction for the 1 compartment assumption and by deriving a prediction formula for ECFV (ECFVPredicted) based on the body weight. The optimal single sample method was tested in group 2.
A blood sample at 180 minutes and ECFVPredicted were optimal for single sample clearance. Mean ± SD GFR in group 1 determined using the Jacobsson and modified Jacobsson formulae was 1.78 ± 0.70 and 1.65 ± 0.60 mL/min/kg, respectively. When tested in group 2, the Jacobsson method overestimated multisample clearance. The modified Jacobsson method (mean ± SD 2.22 ± 0.34 mL/min/kg) was in agreement with multisample clearance (mean ± SD 2.19 ± 0.34 mL/min/kg).
The modified Jacobsson method provides accurate estimation of iohexol clearance in cats, from a single sample collected at 180 minutes postinjection and using a formula based on the body weight to predict ECFV. Further validation of the formula in patients with very high or very low GFR is required.
area under the plasma concentration vs time curve
domestic long hair
domestic short hair
extracellular fluid volume
glomerular filtration rate
high performance liquid chromatography
International Renal Interest Society
urine specific gravity
The overall functional capacity of the kidney relates to the total number of intact functioning nephrons and glomerular filtration rate (GFR) is considered to be the most sensitive method of assessing filtration function. In clinical practice and often in research studies, an estimation of GFR is assumed from the plasma clearance of a filtration marker. Despite the fact that the concentration of most filtration markers is determined in serum, the term plasma clearance used as GFR is dependent on renal plasma flow. One such marker is iohexol, a nonionic radiographic contrast agent, which can be administered exogenously as a single-bolus dose for plasma clearance studies. Iohexol is useful as a clearance marker, as it has been demonstrated in human patients that it has negligible protein binding, is freely filtered, is not secreted into the renal tubule and is not metabolized within the body.[1-3] There are published studies that have used this marker for clearance studies in cats without reporting any adverse effects.[4-15]
Plasma clearance is determined from the concentration of the filtration marker (in the present study iohexol) in multiple blood samples collected over a variable time period (usually 0–360 minutes). The data are then plotted to create a plasma concentration versus time curve. Pharmacokinetic modeling can be applied to the curve to determine the area under the plasma concentration versus time curve (AUC). Clearance is calculated as the dose of filtration marker administered divided by the AUC. The collection of multiple blood samples is not only labor intensive and time consuming, but may also be stressful for the feline patient. In addition, there are increased costs of analysis associated with multiple samples. Techniques that require the collection of only a limited number of samples have become an important goal for clinicians when measuring clearance. A corrected slope-intercept method that requires the collection of only 3 samples has recently been validated for cats.
A further simplification of the limited sampling technique is to determine clearance from a single plasma sample. There are 2 prerequisites for single sample methods. The first is the knowledge of the volume of distribution of the filtration marker (considered, for iohexol, to be approximately the same as extracellular fluid volume [ECFV]). The second, is that the sample is collected at a time postadministration when the distribution phase has reached completion. As renal function declines, this time point is further from the time of administration of the marker.[16, 17] Previous approaches for determining single sample clearance in dogs and cats have applied linear or nonlinear regression analysis to derive equations for single sample methods from multi-sample methods and to select the optimal sampling time.[7, 18-20] Such empirical methods have limitations, as the clearances determined in this manner are not independent measurements. Theoretically, this approach would be valid in a patient with a distribution volume equal to the average of the patients studied and renal function similar to that of the population used to generate the formula. In normal healthy patients, there may be low variation in the distribution volume and the error may be small. In patients with abnormal renal function, however, the range of distribution volumes may be greater, and therefore the error larger.
The objective of the this study was to validate a single sample method for determining iohexol clearance in cats.
Eighty-nine cats (44 female neutered [FN] and 45 male neutered [MN]) with known plasma creatinine concentrations were recruited into the prospective study between January 2009 and January 2010. All cats were reported to be clinically healthy by their owners at the time of examination. Informed consent was obtained from the owners and the study was conducted with approval from the Royal Veterinary College's ethics and welfare committee. Exclusion criteria included any concurrent medical disorder or any evidence of abnormal fluid volume status such as skin tenting on clinical examination.
Food was withheld for 12 hours before clearance measurement, however, free access to water was allowed. An intravenous catheter was placed in the cephalic vein. A bolus dose of iohexol1 (647 mg/kg [300 mg iodine/kg]) was administered intravenously followed by 1 mL of 0.9% saline. The exact dose was determined from the volume administered. The completion of the injection represented time zero. Blood samples of 1 mL each were collected via jugular venipuncture or from a separate intravenous catheter at 120, 180, and 240 minutes and transferred to serum tubes. The exact sampling time was recorded for each cat and used in the clearance calculations. The serum was separated and transferred to storage at −80°C until analysis, which was performed at a commercial reference laboratory2 using an HPLC method previously validated in humans.3
Pharmacokinetic analysis was performed using a commercially available pharmacokinetic computer program.4 Clearance (Cl) was calculated as follows:
where dose is the dose of iohexol administered and AUC is the area under the plasma concentration versus time curve determined using a 1 compartment model. Specifically, the AUC is the ratio of the zero time intercept of the single exponential fitted to the plasma concentrations and the rate constant, β, of the exponential (elimination rate constant). The clearance values were standardized to the body weight (kg). A previously validated cat-specific formula was applied to correct for the 1 compartment assumption.
Single sample clearance (Cl1) was determined from the iohexol concentration in a single sample obtained at each time point (120, 180, and 240 minutes) using the original Jacobsson formula:
where t is time of collection of the sample, ECFV is extracellular fluid volume (determined using methods described below), and Vt is the apparent volume of distribution of iohexol (calculated as the dose administered/plasma concentration at time t). Cl1 was calculated using the equation above and the measurement standardized to the body weight (kg). To correct for the one compartment assumption, a modifying factor (m) was calculated as clearance corrected using the previously validated cat correction formula divided by uncorrected clearance. Cl1 (standardized to body weight) was then recalculated using the Jacobsson equation, replacing ECFV with ECFV/m.
In addition, a modified Jacobsson technique was also validated for determination of Cl1. The principle of this approach is to determine β from a single sample obtained at time t. Following correction for the 1 compartment assumption (see below), β is equal to GFR per unit ECFV (GFR/ECFV). Multiplication by an estimate of ECFV then gives GFR.
Determining β from a single sample requires an estimate of the concentration of the tracer at time zero (C0). Assuming a single compartment the equation is:
where Ct is the concentration at time = t, C0 is the concentration at time = 0, and β is the elimination rate constant. The equation was rearranged to obtain β:
The concentrations, C0 and Ct, were converted to the corresponding distribution volumes so:
As V0 is larger than the actual volume of distribution because of nonimmediate mixing, the equation was rewritten for ECFV:
ECFV is smaller than V0 so β1 > β. In the modified Jacobsson method, ECFV rather than V0 is estimated and therefore it was necessary to determine the factor required to convert β1 to β.
As a result of the 1 compartment assumption, β is close to GFR per unit of ECFV, but not identical to it. By correcting for the 1 compartment assumption, GFR/ECFV can be determined from β using a previously validated cat correction formula5 analogous to that used by Bird et al for human patients:
GFR/ECFV was then multiplied by ECFV (as determined below) to obtain Cl1 (GFR). Cl1 was normalized to body weight (kg).
In addition, a formula to predict ECFV (ECFVPredicted) from body weight was developed by performing linear regression analysis using the data from the cats in group 1.
The data were assessed for normality by visual inspection and by performing the Kolmogorov–Smirnov test. As the data met the assumptions of a Gaussian distribution, parametric testing was used. Correlations between corrected slope-intercept clearance and single sample clearance were explored using Pearson's correlation coefficient. Relationships were evaluated by performing linear regression analysis and determining the coefficient of determination (R2). Agreement was assessed by plotting the difference between the 2 clearances against the average of the 2 values derived for each cat (Bland–Altman plots). Bias was defined as the group mean difference between 2 clearances and the absolute limits of agreement were defined as the group mean difference ± 2SD. Model assumptions and performance of the linear regression analysis were evaluated by examining multicolinearity, standardized residuals, Cook's distances, leverage values, and by performing the Durbin–Watson test. Significance was set at P < .05.
Results are presented as mean ± SD and significance was set at P < .05.
Single sample clearance, determined using either the Jacobsson or modified Jacobsson formulae, found to have the strongest relationship with corrected slope-intercept clearance in group 1, was subsequently tested in a 2nd group of client-owned cats in which multisample iohexol clearance was measured. This was performed to test the validity of the formulae when applied to an independent data set. The multisample data from these cats and the clearance procedure have previously been described. Briefly, the study population consisted of 18 healthy nonazotemic cats. Multisample plasma iohexol clearance was determined from samples collected at 5, 15, 30, 60, 120, 180, 240, and 360 minutes. The AUC was calculated using noncompartmental analysis performed using a commercially available pharmacokinetic computer program.4
The 89 cats included in this study consisted of the following breeds: domestic short hair (DSH [n = 57]), domestic long hair (DLH [n = 7]), Persian (n = 6), Burmese (n = 5), Bengal (n = 4), Russian blue (n = 4), British short hair (n = 3), Ocicat (n = 2), and Balinese (n = 1). The characteristics of the study population are presented in Table 1. Seventy-three cats were considered to be nonazotemic and 16 cats had renal azotemia (plasma creatinine concentration >2.0 mg/dL in association with a reduced urine concentrating ability [USG <1.035]). Eight of the azotemic cats could be categorized as IRIS stage 2 and eight as IRIS stage 3.
|Nonazotemic (n = 73)||Azotemic (n = 16)||All Cats (n = 89)|
|Age (years)||12.39 ± 3.29||13.93 ± 2.67||12.66 ± 3.23|
|Body weight (kg)||4.27 ± 1.27||4.69 ± 1.17||4.34 ± 1.20|
|Plasma creatinine concentration (mg/dL)||1.53 ± 0.30||2.77 ± 0.59||1.75 ± 0.60|
|Plasma urea concentration (mg/dL)||30.97 ± 7.47||49.73 ± 11.79||34.25 ± 11.01|
|USG||1.048 ± 0.021||1.024 ± 0.012||1.043 ± 0.022|
|Slope-intercept clearance (mL/min/kg)||1.80 ± 0.53||0.92 ± 0.36||1.64 ± 0.61|
|ECFVSlope-intercept (mL)||849.28 ± 300.00||823.87 ± 402.93||844.72 ± 318.45|
|ECFVPredicted (mL)||825.22 ± 144.51||875.14 ± 140.00||834.19 ± 144.22|
The modifying factor for use in the Jacobsson formula to correct for the 1 compartment assumption was:
For the modified Jacobsson method, a linear relationship existed between β and β1 at all time points (P < .001). The mean β/β1 at the 180-minute time point was 0.93 (see Fig 1).
Mean ± SD single sample clearance calculated using the Jacobsson and modified Jacobsson formula determined using ECFVSlope-intercept and ECFVPredicted from samples collected at 120, 180, and 240 minutes are presented in Table 2. All single sample methods were significantly correlated with the corrected slope-intercept clearance method (P < .05). Bland–Altman analysis showed excellent agreement between single sample clearance determined using the 180-minute sample and ECFVPredicted and corrected slope-intercept clearance (see Fig 2A and B).
|t||N||Mean ± SD||R 2||P value|
|ECFVSlope-intercept||120||89||1.77 ± 0.76||0.91||<.001|
|180||89||1.80 ± 0.78||0.93||<.001|
|ECFVPredicted||240||89||1.79 ± 0.73||0.92||<.001|
|120||89||1.76 ± 0.79||0.92||<.001|
|180||89||1.78 ± 0.70||0.92||<.001|
|240||89||1.79 ± 0.63||0.85||<.001|
|Modified Jacobsson formula|
|ECFVSlope-intercept||120||89||1.75 ± 0.73||0.90||<.001|
|180||89||1.66 ± 0.67||0.94||<.001|
|ECFVPredicted||240||89||1.66 ± 0.062||0.93||<.001|
|120||89||1.72 ± 0.73||0.90||<.001|
|180||89||1.65 ± 0.60||0.92||<.001|
|240||89||1.65 ± 0.54||0.85||<.001|
The strongest relationship between corrected slope-intercept clearance and single sample clearance using either the Jacobsson formula or the modified Jacobsson formula was based on ECFVSlope-intercept at 180 minutes (R2 = 0.93, P < .001 and R2 = 0.94, P < .001, respectively). The relationship between single sample GFR and slope-intercept GFR was as close when using ECFVPredicted (which requires only a simple calculation from body weight) as when using ECFVSlope-intercept (R2 = 0.92, P < .001 and R2 = 0.93, P < .001, respectively). In addition, there is no alternative way of determining ECFV in clinical practice when using the single sample method. Therefore, when the single sample method was tested in the additional 18 cats against multisample clearance, the sample collected at 180 minutes and ECFVPredicted were used in the formulae.
One cat was excluded from the final model to predict ECFV from body weight as it was not found to meet the model assumptions. Therefore, the prediction formula was derived from data from 88 cats. Mean ± SD ECFVSlope-intercept and ECFVPredicted are presented in Table 1. The derived prediction equation for ECFV from body weight was:
where ECFVPredicted is in mL and BW is body weight in kg.
There was a weak relationship between ECFVSlope-intercept and ECFVPredicted (R2 = 0.234, P < .001). The agreement analysis indicated a negligible bias (0.004 mL), although the limits of agreement were wide (−502.22 to 502.22 mL) and agreement was considered to be poor.
The 18 cats included in this study consisted of the following breeds: domestic short hair (DSH [n = 13]), domestic long hair (DLH [n = 2]), and Bengal (n = 3) with a median (range) age of 4.6 (1.5–18.8) years. Mean ± SD multisample clearance, single sample clearance, ECFVSlope-intercept and ECFVPredicted are presented in Table 3. Single sample clearance calculated using either the Jacobsson or modified Jacobsson formula was significantly correlated with multisample clearance (r = 0.97, P < .001 and r = 0.97, P < .001 respectively). The Bland–Altman plots to assess agreement are presented in Figures 3A and B. Single sample clearance determined using the Jacobsson formula was found to overestimate multisample clearance. In contrast, agreement between single sample clearance determined using the modified Jacobsson formula and multisample clearance was considered to be excellent. The maximum difference in 1 cat was found to be <0.20 mL/min/kg, which is approximately 10% of the mean single sample clearance value.
|Mean ± SD||Range|
|Multisample clearance (mL/min/kg)||2.19 ± 0.34||1.53–2.82|
|Single sample clearance (mL/min/kg)|
|Jacobsson formula||2.42 ± 0.41||1.68–3.23|
|Modified Jacobsson formula||2.22 ± 0.34||1.58–2.88|
|ECFVSlope-intercept (mL)||896.59 ± 126.01||524.37–1240.33|
|ECFVPredicted (mL)||829.86 ± 565.28||565.28–1062.49|
The mean ± SD percent error in cats in group 1 associated with ECFVPredicted and single sample clearance was 20.41 ± 23.97% and 5.95 ± 6.13%, respectively. Azotemic cats had a higher mean ± SD percent error than nonazotemic cats (7.19 ± 4.82% and 5.35 ± 4.75%, respectively). In group 2, the mean ± SD percent error was 12.41 ± 7.72% and 3.91 ± 2.69%, respectively. The percent error associated with single sample clearance compared with multisample clearance or slope-intercept clearances plotted against the percent error associated with ECFVPredicted compared with ECFVSlope-intercept is presented in Figure 4. Generally, only minimal inaccuracies were observed in single sample clearance even if substantial errors in ECFVPredicted arose.
In this study, the original Jacobsson formula for single sample plasma iohexol clearance was considered to show excellent agreement with corrected slope-intercept clearance in cats. However, in a second group of cats it was found to overestimate multisample clearance. The modified Jacobsson single sample formula, in contrast was found to show excellent agreement with both corrected slope-intercept and multisample clearance in this study. It is recommended therefore that the modified Jacobsson method is used for single sample plasma iohexol clearance in cats. The reason for the superior performance of the modified Jacobsson formula over the original Jacobsson formula is unclear. It may be related to the correction factor for the larger volume of V0 compared with ECFV in the original Jacobsson formula (0.0016), which was derived specifically from human rather than feline patients. This correction factor was also based on the empirical formula of Brochner–Mortensen which has previously been shown to have a nonlinear relationship with multisample iohexol clearance in cats and may further explain the inaccuracy of the Jacobsson formula in cats.
In large populations of human patients, single sample iohexol clearance is significantly correlated and in excellent agreement with slope-intercept iohexol clearance corrected using the Brochner–Mortensen formula and with multisample iohexol clearance. Furthermore, single sample iohexol clearance compared with multisample clearance using a different filtration marker also showed good correlation and agreement in human patients. Comparison with a separate filtration marker is recommended when evaluating limited or single sample techniques to provide an independent validation of the measurement. Unfortunately, it was not possible to perform clearance measurements using a separate marker in the current studies as limited sampling methods remain unvalidated in alternative filtration markers in cats. Furthermore, markers for which the distribution volume is similar to iohexol and therefore may potentially be corrected by the previously validated cat correction formula for iohexol, require nuclear medicine facilities not available in this study. However, the single sample formula was tested in a separate group of cats in which multisample clearance measurements had been performed and agreement was considered to be excellent. It would have been optimal to compare the single sample method to the multisample method in all cats. This was not possible becauseof the costs associated with analyzing multiple samples in a large number of cats. In addition, it was not considered ethically justifiable to collect multiple samples from cats, which may be more stressful when a validated limited sampling strategy was available. Considering these factors alongside the good agreement of the limited sampling method with the multisample method, the limited sampling method was deemed the most appropriate reference method for this larger scale study. In human patients, it has been demonstrated that single sample iohexol clearance shows a stronger correlation and better agreement with the gold standard measurement of GFR, urinary clearance of inulin than slope-intercept iohexol clearance. The limits of agreement when comparing single sample to multisample clearance in this study were narrow (−0.12 to 0.19 mL/min/kg) with the maximum difference in any individual cat being approximately 10% of the mean clearance value. However, it is possible with a larger number of cats with larger range of renal function that the limits of agreement may have been wider. The mean percent error associated with single sample clearance compared with corrected slope-intercept clearance was higher in azotemic compared with nonazotemic cats; however, both levels of error may be considered clinically acceptable. Because of the small number of azotemic cats included in this study, it is not possible to compare the performance of single sample clearance with nonazotemic cats. The limits of agreement when comparing slope-intercept clearance to multisample clearance in a previous study were wider (−0.29 to 0.32 mL/min/kg), with the maximum difference being approximately 14% of the mean clearance value. This may suggest that, as in human patients, single sample clearance is more accurate than slope-intercept clearance in cats. The reason for this remains unclear, but may be related to the introduction of more analytical error associated with the analysis of multiple samples. Alternatively, it may be because of the collection of the 1st blood sample when performing slope-intercept clearance that is too early and at a time in some patients when redistribution of iohexol is incomplete.
In this study, the Jacobsson and modified Jacobsson formulae were used to determine single sample clearance. Previous studies in cats have used empirically derived regression formulae to estimate clearance from a single sample.[7, 18-20] A limitation of such regression based correction equations is that they may only be valid in identical populations to those in which the formula was derived. A previous canine study which compared single sample iohexol clearance using the Jacobsson formula with multisample clearance of Tc-DTPA found good correlation between the 2 methods. However, correlation analysis assesses only the strength of a relationship and agreement analysis is required to demonstrate how well 2 methods agree. Agreement analysis was not performed in the canine study.
In this study, out of the 3 sampling time points tested, the optimal sampling time for the collection of the single sample in cats was found to be 180 minutes. In a previous canine study, the same sampling times as used in this study were evaluated for single sample clearance (120, 180, and 240 minutes) and the optimal sampling time was also found to be 180 minutes. The standard sampling time recommended by the British Nuclear Medicine Society when determining single sample clearance using nuclear medicine markers is 180–240 minutes. Other human studies have reported an optimal sampling time of 180 minutes.[25, 31, 32] It is recommended, however, that in patients with a very low GFR, the sampling time is delayed. The original Jacobsson paper reported that sampling time for such patients may be up to 8–9 hours; however, other studies in human patients have suggested a delay of up to 24 hours. A prolonged sampling time is required in patients with very low GFR as there is a delay in the rate of mixing of the marker in the ECFV which may result from an expanded ECFV. If sampling is performed too early in patients with very low GFR, then single sample clearance will overestimate true clearance. A single sample collected at 180 minutes in human patients using the marker iohexol resulted in a small overestimation in clearance at low renal function compared with classical inulin clearance. Inspection of the Bland-Altman plots assessing the agreement between single sample clearance determined using the Jacobsson and modified Jacobsson formulae and corrected slope-intercept clearance did not indicate any consistent over- or underestimation in cats with low clearance measurements. There were, however, a very limited number of cats included in this group with very low GFR. All of the cats in which multisample clearance measurements were performed had normal renal function and therefore it was not possible to assess any errors associated with reduced renal function in this group. Further studies to investigate the optimal sampling time for cats with very low GFR would be required. In clinical practice, it may be difficult to estimate renal function in advance of clearance measurement to predict the most appropriate sampling time. One study involving human patients recommended obtaining the sample at 24 hours in patients with a serum creatinine concentration >200 μmol/L. Another human study selected the most appropriate single sampling time based on an eGFR formula. Such approaches could be explored in cats in further work. However, given that the major clinical indication for GFR determination is in patients with normal or only moderately reduced renal function, the modified Jacobsson technique described, with collection of a sample at 180 minutes, can be considered to be optimal for these cats. Very low values of GFR obtained through measurement of single sample clearance using the method described should be interpreted with caution.
A further consideration when collecting a single sample is that any analytical error will influence the accuracy of the clearance measurement. Additionally, if the entire dose of filtration marker has not been administered intravenously this will also lead to error. The use of multiple samples has the benefit of allowing inspection of the plasma concentration versus time curve which, if abnormal, may suggest intravenous administration did not occur. Studies regarding the stability of iohexol in feline samples remain unpublished, however, data extrapolated from other species and precision studies performed in cats would suggest it is likely to be stable and therefore it is considered by most investigators working with this marker to be one of the advantages of its use. The iohexol assay used in this study has been validated in human but not cat serum, which represents a limitation. In human serum, linearity was established up to 1500 μg/mL. Within-batch precision was 8% at 47 μg/mL and 3% at 384 μg/mL. Absolute recovery for iohexol averaged 90% and the lower limit of detection was 20 μg/mL3. Although this assay has not been validated in cats in this study, the authors found a good agreement with the results obtained with an HPLC method validated in this species.
Extracellular fluid volume determined from bromide space is often considered to be the “gold standard.” However, previous work in the cats has shown that the volume of distribution of bromide and of iohexol determined from slope-intercept clearance, although both considered to approximate to ECFV, are not in agreement.3 The reason for this remains unclear, but may be related to the smaller size of the bromide molecule resulting in different penetration of fluid spaces. In this study, ECFV determined from the volume of distribution of iohexol was considered to be the most appropriate for the single sample clearance method as in the development of the earlier Jacobsson and modified Jacobsson formulae for human patients. A large ECFV, which may be present in edematous patients, would delay mixing of the marker. The sampling time would therefore need to be delayed. A limitation of the single sample method using ECFVPredicted described in this study is the fact that the distribution volume is not measured directly, but predicted using a formula based on the body weight. Use of formulae to predict ECFV from the body weight is the established method in human patients. The prediction formula for ECFV derived in this study was not found to provide a particularly accurate prediction based on the results of the agreement analysis. However, several human studies have shown that an error of up to 50% in the prediction of ECFV will not result in significant error in the clearance estimation if the timing of the sample collection is appropriate.[16, 32] Errors of the magnitude that may be expected to be encountered when calculating single-sample clearance using ECFVPredicted would not be so great as to be considered clinically significant. This is illustrated in Figure 4, which presents the percent error associated with single sample clearance compared with multisample clearance or slope-intercept clearances plotted against the percent error associated with ECFVPredicted compared with ECFV. A very high error in ECFVPredicted would be required to make any impact on clinical management of a patient.
Extracellular fluid volume is reported to be approximately 20% of body weight in cats with a range of between 15 and 30%. In this, the values of ECFVSlope-intercept would also support this. However, it is unclear how ECFV was determined by DiBartola to attain this figure. As discussed earlier, the gold standard method would be considered to be the determination of the dilutional space of bromide. Earlier work in cats has shown that ECFVSlope-intercept is not in agreement with the dilutional space of bromide3 and this is likely to reflect different penetration of fluid spaces by the different markers. Therefore, although the ECFVSlope-intercept supports the figures stated by DiBartola, it is unclear as to whether these figures are in fact accurate if the gold standard method (bromide dilution) is considered. This is an area for future study in cats.
From inspection of the Bland–Altman agreement plots (Fig 2), a small number of outliers can be identified and were studied further. The discrepancies in most of these outliers (4/6) could be explained by the presence of abnormal body size (obesity assessed by body weight and body condition score) that may have resulted in particularly large errors in ECFVPredicted.
The inherent difficulties associated with performing multisample clearance measurement and also calculation of clearance using different pharmacokinetic models may pose conceptual difficulties for veterinary practitioners and therefore only be utilized by clinicians and researchers with specialist knowledge of the area. Therefore, calculation of clearance using the modified Jacobsson formula described in this study, which accurately determines GFR based on administration of a bolus dose of iohexol, collection of a single plasma sample and an estimation of ECFV from BW, would be a simple and attractive method available to all veterinary practitioners.
Conflict of Interest Declaration: Authors disclose no conflict of interest.
Omnipaque300; Amersham Health, Princeton, NJ
Epsom and St Helier University NHS trust, Epsom, UK
Christofides et al. 2006. Ann Clin Biochem 43 Suppl (abstract)
Phoenix WinNonlin 6.0; Pharsight, Sunnyvale, CA
Finch et al. 2010. Nuc Med Comm 31, 451 (abstract)