• View in gallery
    Figure 1—

    Time-attenuation curves depicting mean concentrations of iodinated contrast medium (iodixanol or iohexol) in the abdominal aorta, renal cortex, whole kidney, and renal pelvis in 4 healthy Beagles in a crossover-design study evaluating the effects of changes in analytic variables and contrast medium osmolality on CT-GFRs. Measurements were initiated at the time of contrast medium injection (300 mg I/kg, 3.0 mL/s); there was a 2-week interval between GFR assessments. The CT-GFR values did not differ significantly between contrast medium types, and data from both experiments were combined (8 measurements/time point). The vertical dashed line indicates the second renal cortex peak time. Contrast medium was rapidly excreted to the renal pelvis beginning approximately 60 seconds after the injection started. EU = Enhancement units.

  • 1. Improving Global Outcomes (KDIGO) CKD Work Group. Kidney Disease: KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney 2013; S3:1150.

    • Search Google Scholar
    • Export Citation
  • 2. Svaland MG, Haider T, Langseth-Manrique K, et al. Human pharmacokinetics of iodixanol. Invest Radiol 1992;27:130133.

  • 3. Heglund IF, Michelet ÅA, Blazak WF, et al. Preclinical pharmacokinetics and general toxicology of iodixanol. Acta Radiol Suppl 1995;399:6982.

    • Search Google Scholar
    • Export Citation
  • 4. Jacobsen PB, Blindheim L, Skotland T. Bioanalytical methods for iodixanol and their application to studies on metabolism and protein binding. Acta Radiol Suppl 1995;399:6166.

    • Search Google Scholar
    • Export Citation
  • 5. Tsushima Y, Blomley MJ, Okabe K, et al. Determination of glomerular filtration rate per unit renal volume using computerized tomography: correlation with conventional measures of total and divided renal function. J Urol 2001;165:382385.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6. Hackstein N, Bauer J, Hauck EW, et al. Measuring single-kidney glomerular filtration rate on single-detector helical CT using a two-point Patlak plot technique in patients with increased interstitial space. AJR Am J Roentgenol 2003;181:147156.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Alexander K, Dunn M, Carmel EN, et al. Clinical application of Patlak plot CT-GFR in animals with upper urinary tract disease. Vet Radiol Ultrasound 2010;51:421427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Daghini E, Juillard L, Haas JA, et al. Comparison of mathematic models for assessment of glomerular filtration rate with electron-beam CT in pigs. Radiology 2007;242:417424.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Kishimoto M, Yamada K, Watanabe A, et al. Comparison of excretory urographic contrast effects of dimeric and monomeric non-ionic iodinated contrast media in dogs. J Vet Med Sci 2007;69:713715.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Patlak CS, Blasberg RG. Graphical evaluation of blood-to-brain transfer constants from multiple-time uptake data. J Cereb Blood Flow Metab 1985;5:584590.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Aspelin P, Aubry P, Fransson SG, et al. Nephrotoxic effects in high-risk patients undergoing angiography. N Engl J Med 2003;348:491499.

  • 12. Thomsen HS, Stacul F. CIN: can we forget it? Acta Radiol 2014;55:10271030.

  • 13. Hackstein N, Wiegand C, Rau WS, et al. Glomerular filtration rate measured by using triphasic helical CT with a two-point Patlak plot technique. Radiology 2004;230:221226.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Alexander K, Del Castillo JR, Ybarra N, et al. Single-slice dynamic computed tomographic determination of glomerular filtration rate by use of Patlak plot analysis in anesthetized pigs. Am J Vet Res 2007;68:297304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15. Alexander K, Authier S, del Castillo JR, et al. Patlak plot analysis CT-GFR for the determination of renal function: comparison of normal dogs with autologous kidney transplant dogs. Contrast Media Mol Imaging 2010;5:133139.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. O'Dell-Anderson KJ, Twardock R, Grimm JB, et al. Determination of glomerular filtration rate in dogs using contrast-enhanced computed tomography. Vet Radiol Ultrasound 2006;47:127135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Lakatta EG. Starling's law of the heart is explained by an intimate interaction of muscle length and myofilament calcium activation. J Am Coll Cardiol 1987;10:11571164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 18. Han JK, Kim AY, Lee KY, et al. Factors influencing vascular and hepatic enhancement at CT: experimental study on injection protocol using a canine model. J Comput Assist Tomogr 2000;24:400406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Han JK, Choi BI, Kim AY, et al. Contrast media in abdominal computed tomography: optimization of delivery methods. Korean J Radiol 2001;2:2836.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Fusellier M, Desfontis JC, Madec S, et al. Influence of three anesthetic protocols on glomerular filtration rate in dogs. Am J Vet Res 2007;68:807811.

    • Crossref
    • Search Google Scholar
    • Export Citation

Advertisement

Effects of changes in analytic variables and contrast medium on estimation of glomerular filtration rates by computed tomography in healthy dogs

Yuri MatsudaCooperative Department of Veterinary Medicine, Tokyo University of Agriculture and Technology, Tokyo, Japan 183–8509.

Search for other papers by Yuri Matsuda in
Current site
Google Scholar
PubMed
Close
 DVM
,
Miori KishimotoCooperative Department of Veterinary Medicine, Tokyo University of Agriculture and Technology, Tokyo, Japan 183–8509.

Search for other papers by Miori Kishimoto in
Current site
Google Scholar
PubMed
Close
 PhD
,
Kazuya KushidaDepartment of Veterinary Medicine, Joint Faculty of Veterinary Medicine, Kagoshima University, Kagoshima, Japan 890–0065.

Search for other papers by Kazuya Kushida in
Current site
Google Scholar
PubMed
Close
 DVM
,
Kazutaka YamadaLaboratory of Veterinary Radiology, Azabu University School of Veterinary Medicine, Kanagawa, Japan 252–5201.

Search for other papers by Kazutaka Yamada in
Current site
Google Scholar
PubMed
Close
 PhD
,
Miki ShimizuCooperative Department of Veterinary Medicine, Tokyo University of Agriculture and Technology, Tokyo, Japan 183–8509.

Search for other papers by Miki Shimizu in
Current site
Google Scholar
PubMed
Close
 PhD
, and
Hiroshi ItohCooperative Department of Veterinary Medicine, Tokyo University of Agriculture and Technology, Tokyo, Japan 183–8509.

Search for other papers by Hiroshi Itoh in
Current site
Google Scholar
PubMed
Close
 PhD

Abstract

OBJECTIVE To investigate effects of changes in analytic variables and contrast medium osmolality on glomerular filtration rate estimated by CT (CT-GFR) in dogs.

ANIMALS 4 healthy anesthetized Beagles.

PROCEDURES GFR was estimated by inulin clearance, and dogs underwent CT-GFR with iodinated contrast medium (iohexol or iodixanol) in a crossover-design study. Dynamic renal CT scanning was performed. Patlak plot analysis was used to calculate GFR with the renal cortex or whole kidney selected as the region of interest. The renal cortex was analyzed just prior to time of the second cortical attenuation peak. The whole kidney was analyzed 60, 80, 100, and 120 seconds after the appearance of contrast medium. Automated GFR calculations were performed with preinstalled perfusion software including 2 noise reduction levels (medium and strong). The CT-GFRs were compared with GFR estimated by inulin clearance.

RESULTS There was no significant difference in CT-GFR with iohexol versus iodixanol in any analyses. The CT-GFR at the renal cortex, CT-GFR for the whole kidney 60 seconds after appearance of contrast medium, and CT-GFR calculated by perfusion software with medium noise reduction did not differ significantly from GFR estimated by inulin clearance. The CT-GFR was underestimated at ≥ 80 seconds after contrast medium appearance (whole kidney) and when strong noise reduction was used with perfusion CT software.

CONCLUSIONS AND CLINICAL RELEVANCE Selection of the renal cortex as region of interest or use of the 60-second time point for whole-kidney evaluation yielded the best CT-GFR results. The perfusion software used produced good results with appropriate noise reduction.

IMPACT FOR HUMAN MEDICINE The finding that excessive noise reduction caused underestimation of CT-GFR suggests that this factor should also be considered in CT-GFR examination of human patients.

Abstract

OBJECTIVE To investigate effects of changes in analytic variables and contrast medium osmolality on glomerular filtration rate estimated by CT (CT-GFR) in dogs.

ANIMALS 4 healthy anesthetized Beagles.

PROCEDURES GFR was estimated by inulin clearance, and dogs underwent CT-GFR with iodinated contrast medium (iohexol or iodixanol) in a crossover-design study. Dynamic renal CT scanning was performed. Patlak plot analysis was used to calculate GFR with the renal cortex or whole kidney selected as the region of interest. The renal cortex was analyzed just prior to time of the second cortical attenuation peak. The whole kidney was analyzed 60, 80, 100, and 120 seconds after the appearance of contrast medium. Automated GFR calculations were performed with preinstalled perfusion software including 2 noise reduction levels (medium and strong). The CT-GFRs were compared with GFR estimated by inulin clearance.

RESULTS There was no significant difference in CT-GFR with iohexol versus iodixanol in any analyses. The CT-GFR at the renal cortex, CT-GFR for the whole kidney 60 seconds after appearance of contrast medium, and CT-GFR calculated by perfusion software with medium noise reduction did not differ significantly from GFR estimated by inulin clearance. The CT-GFR was underestimated at ≥ 80 seconds after contrast medium appearance (whole kidney) and when strong noise reduction was used with perfusion CT software.

CONCLUSIONS AND CLINICAL RELEVANCE Selection of the renal cortex as region of interest or use of the 60-second time point for whole-kidney evaluation yielded the best CT-GFR results. The perfusion software used produced good results with appropriate noise reduction.

IMPACT FOR HUMAN MEDICINE The finding that excessive noise reduction caused underestimation of CT-GFR suggests that this factor should also be considered in CT-GFR examination of human patients.

Glomerular filtration rate is an index of renal function that indicates blood plasma volume filtered at the glomeruli expressed per unit of time. Estimated GFR, which is calculated from serum creatinine concentration, is used for the classification of chronic renal disease.1 The CT-GFR method was developed to calculate GFR of each kidney as part of preoperative planning for renal transplantation or nephrectomy. In this method, an iodinated contrast medium is used as a tracer and, similar to inulin, is not metabolized in the body and is not reabsorbed nor excreted at the renal tubules.2–4 Because this method uses CT scan data, renal volume and assessment of vascular anatomy (eg, evaluation for malformation of the renal artery) can be performed for surgical planning in addition to estimation of single-kidney GFR.5 Moreover, the results can be obtained on the same day as the evaluation.

However, because of the lack of a standardized methodology, CT-GFR can overestimate or underestimate the GFR, compared with that assessed by inulin clearance. Some previous studies have included evaluation of the whole kidney (renal cortex and medulla) for CT-GFR estimation,6,7 whereas others have included the renal cortex only.8 However, it has been recommended that CT-GFR analysis should be based on evaluation of the renal cortex alone, because glomerular filtration occurs in the renal cortex.8 Anesthesia is required when performing CT scans of animals, and GFR may also be underestimated as a consequence of reduced blood pressure during anesthesia. To extrapolate CT-GFR data obtained from experimental studies to clinical trials, data on CT-GFR measurements under anesthesia adequate for this purpose are needed. Moreover, iodinated contrast media with different osmolalities may have different contrast effects in kidneys9; however, to the authors’ knowledge, no published studies comparing the effects of such media on estimates of CT-GFR have been published.

The purpose of the study reported here was to investigate the effects of ROI (whole kidney [renal cortex and medulla] vs cortex alone), analysis time, analytic procedures, and iodinated contrast medium type (higher vs lower osmolality) on values of CT-GFR in healthy anesthetized dogs and to compare these values with GFR estimated by inulin clearance as a means to identify potential causes for overestimation or underestimation of GFR by the CT method.

Materials and Methods

Dogs

Four clinically normal female Beagles (15 months old; mean ± SD body weight, 7.2 ± 0.2 kg) were used in the study. Prior to experiments, thoracic and abdominal radiographic examination, hematologic analysis (BUN and serum creatinine concentrations), measurement of serum cystatin C concentrations, urinalysis, and blood pressure measurements were performed to confirm renal and overall health status. Inulin clearance was also measured 2 days before the first experiment involving CT-GFR. Food was withheld for 12 hours prior to each of the GFR evaluations (water was available ad libitum until 2 hours before an experiment to prevent dehydration). There was a 2-week interval between assessments of CT-GFR.

In addition, serum creatinine concentrations were reevaluated and physical examinations were performed for each dog 72 hours and 7 days after administration of contrast medium to assess for the presence of contrast-induced nephropathy or late adverse effects. All experiments were approved by the Animal Experiments Committee at the Tokyo University of Agriculture and Technology.

Study design

Each dog underwent evaluation of GFR by serum inulin clearance, and then underwent CT-GFR with iohexol administration and CT-GFR with iodixanol administration in a crossover-design study. A manual randomization method was used to determine CT-GFR treatment order. The investigation included assessment of potential differences in CT-GFR measurements attributable to the use of contrast media with different osmolalities (iodixanol vs iohexol), compared with that determined by inulin clearance, as well as comparison of GFR determined by inulin clearance with the CT-GFR values when the target renal location for analysis (cortex or whole kidney), the time range for whole kidney analysis (60, 80, 100, or 120 seconds after the appearance of contrast medium [time 0]), analytic methods (manual calculation with a commercially available spreadsheet or automatic calculation by perfusion CT software), or noise reduction levels in the perfusion CT software (strong or medium) were changed.

Serum clearance of inulin

Inulin clearance was considered the reference standard for GFR measurement and was performed at a commercial laboratory.a Briefly, inulinb (100 mg/kg, IV) was administered to dogs, and blood samples (2.0 mL) were collected into tubes containing serum separation gel at 2 and 3 hours after injection. Samples were centrifuged at 1,200 × g, and the serum was collected and shipped with ice packs to the testing laboratory within 2 hours after collection for quantitative analysis.

Anesthesia

All CT procedures were performed in dogs under general anesthesia. Preanesthetic medication included atropine sulfatec (50 μg/kg, SC) and butorphanol tartrated (0.2 mg/kg, IV). Anesthesia was induced by administration of propofole (6.0 mg/kg, IV) through a 20-gauge IV catheter that was aseptically placed in the right cephalic vein, and an endotracheal tube was placed. Respiration was maintained with a mechanical ventilator.f Ventilation settings were as follows: inlet flow, 15 to 20 L/min; peak inspiratory pressure, 10.2 cm H2O; and respiratory rate, 12 breaths/min. Anesthesia was maintained by a continuous IV infusion of propofol (0.2 mg/kg/min). Vital signs (oxygen saturation of hemoglobin as measured by pulse oximetry, end-tidal partial pressure of CO2, body temperature, heart rate, and blood pressure) were monitored during anesthesia. In addition, dogs received infusions of acetated Ringer solution (5.0 mL/kg/h) for 1 hour before and after the CT-GFR procedures to prevent dehydration and renal failure.

CT procedures

Nonionic iodinated contrast medium (iodixanolg [290 mOsm/kg of water] or iohexolh [672 mOsm/kg water]) was administered at a dose of 300 mg of I/kg. Contrast medium was administered by a power injectori through a 20-gauge IV catheter aseptically placed in the left cephalic vein. Contrast medium was stored at 37°C until just prior to use. The CT images were acquired with 64 multidetector-row CT.j

Dogs were placed in ventral recumbency. First, a plain abdominal scan was performed (tube voltage, 100 kV; tube current, 100 mA; slice thickness, 0.5 mm; and rotation speed, 0.5 s/rotation). The angle between the x-axis and a line connecting the left and right renal pelvises was calculated from the images. Next, the dog was rotated by an angle that placed both kidneys in the same transverse section. Contrast medium was administered (3.0 mL/s, IV), and a dynamic scan was initiated at the level of renal pelvis (32 mm scan range; tube voltage, 100 kV; tube current, 80 mA; slice thickness, 0.5 mm; and rotation speed, 0.5 s/rotation). Dynamic scanning was performed at 1.5-second intervals from 0 to 30 seconds after scan initiation, then at 2.0-second intervals from 30 to 60 seconds, then at 20-second intervals from 60 to 120 seconds. Scan-synchronized breath holding was induced to avoid motion artifacts. Obtained images were reconstructed with the following parameters: abdominal reconstruction function, FC03j; iterative reconstruction (dose reduction level strong); and slice thickness, 4.0 mm.

Next, a helical scan was performed to obtain images for renal volume calculation (tube voltage, 100 kV; auto exposure controlled tube current with 300-mA upper limit; SD [a measure of fluctuations in the CT number related to image noise], 8.0; slice thickness, 0.5 mm; and rotation speed, 0.5 s/rotation). Images were reconstructed (reconstruction function, FC03j; slice-thickness, 0.5 mm; and reconstruction interval, 0.25 mm) and were transferred to a workstation.k

Patlak plot analysis

The CT-GFR determinations were made by Patlak plot analysis, which is a 2-compartment model algorithm as follows10:

article image

where indicates the contrast medium concentration in tubules after filtration and indicates the contrast mdium concentration in the vessels. The sum of these is equal to the total contrast medium concentration at time t, designated C(t), in the ROI. The b(t) represents the contrast medium concentration in the aorta at time t; V is the volume of the target renal location (renal cortex or whole kidney in the present study); VB is the blood volume of target renal location; and a is the velocity of blood filtration. Therefore, indicates the GFR per unit volume (mL/cm3/min). The formula can be transformed by dividing both sides by b(t) as follows to calculate GFR from the TAC of artery or tissue:

article image

Concretely on the y-axis is plotted against on the x-axis, C(t) is the CT value at time t for the whole kidney, b(t) is the CT value at time t in the aorta, and ∫0tb(t)dt is the AUC of the aorta during the target time range for analysis (where 0 is the appearance time of contrast medium). After plotting, a linear fitting curve is set up from time 0 to t, to calculate the slope . In addition, is the y-intercept of the fitting curve, which refers to the vascular bed volume (mL/cm3). The GFR value is then corrected for Hct (from a blood sample obtained immediately before induction of anesthesia) because the plasma is the real factor of glomerular filtration:

article image

Manual calculation of CT-GFR

The TACs for contrast effect of the aorta, renal medulla, and renal pelvis at the level of the renal hilum were generated by use of preinstalled software for each dog and for each contrast medium.l These curves were calculated on the basis of CT enhancement units (the net increase of contrast effect [with the precontrast CT value subtracted from the postcontrast CT value]).

The renal cortex ROI was drawn manually by referring to the postcontrast images. The whole kidney ROI was also drawn manually to include the cortex and medulla. Renal volume and renal cortex volume were calculated with the volume rendering function of the workstation, and the AUC was calculated for the aorta TAC with a commercially available spreadsheet.m

The target time range for analysis involving the renal cortex only began at the time of contrast appearance at the renal cortex and ended at the time of contrast appearance at the medulla (the time of second peak attenuation at the cortex). For analysis involving the whole kidney, the target time range began with the time of contrast appearance in the whole kidney ROI and ended at predetermined time points of 60, 80, 100, and 120 seconds after the appearance time.

To calculate GFR of a single kidney from the GFR per unit volume (mL/cm3/min) determined with Patlak plot analysis, the calculated GFR was multiplied by the volume (cm3) of the renal cortex or whole kidney (as applicable for the target renal location). Finally, the GFR per dog (mL/kg/min) was determined by adding the GFRs calculated for the left and right kidneys and dividing the sum by body weight.

CT-GFR analysis with perfusion CT software

Patlak plot analysis was also performed with the preinstalled perfusion CT software.n The matrix size for analysis setting was 256 × 256 pixels. The arterial ROI was placed on the center of the abdominal aorta, and the tissue ROI was placed on renal cortex with the smallest circle size in the software settings. The time range for analysis was identical to that used in the previously described manual analysis. Within the software, noise reduction levels (strong or medium) were selected. Three circular ROIs, the smallest in the software settings, were marked on the renal cortex on the generated flow map to calculate blood flow (mL/100 g/min). Locations with the lowest SD of blood flow values were selected for ROIs. Single-kidney GFR was finally calculated with cortex volume correction and Hct correction performed manually.

Statistical analysis

Data value distributions were assessed by means of the Kolmogorov-Smirnov test or Shapiro-Wilk test. All data were parametric. All statistical analyses were performed with a statistical software package.o Differences between the iodixanol (n = 4) and iohexol (4) treatments were assessed with paired t tests. Differences among GFR values measured by inulin clearance and CT-GFR determined for different target locations (renal cortex or whole kidney), various time ranges for whole kidney analysis, and different noise reduction levels or analytic methods were assessed with post hoc (Bonferroni-Dunn) tests. Values of P < 0.05 were accepted as significant.

Results

No increases in serum creatinine concentration were detected at 72 hours or 7 days after CT-GFR experiments in any dog. There were no other adverse effects observed in any dogs.

Mean ± SD anesthesia time for all CT-GFR experiments was 42.4 ± 11.1 minutes (n = 8). During anesthesia, mean ± SD systolic and mean blood pressures were 147.6 ± 39.3 mm Hg and 107.9 ± 30.1 mm Hg, respectively, and mean ± SD heart rate was 154 ± 30 beats/min (n = 8). The mean renal volume and renal cortex volume were 24.5 ± 3.1 mL and 12.5 ± 1.4 mL (n = 8).

There were no significant differences in CT-GFR at the renal cortex or for the whole kidney at any time point as assessed by use of iohexol and iodixanol (Table 1). Therefore, data for the 2 treatments were combined for subsequent analyses. The TACs depicting mean concentrations of iohexol or iodixanol in the abdominal aorta, renal cortex, whole kidney, and renal pelvis over time are shown (Figure 1). Mean ± SD time of peak attenuation at the aorta was 11.6 ± 1.3 seconds, with a peak CT value of 368.7 ± 45.3 enhancement units. Mean ± SD first and second renal cortex peak attenuation times were 18.0 ± 1.1 seconds and 28.3 ± 1.81 seconds, respectively, with contrast appearing in the renal medulla at 31.4 ± 3.29 seconds and in the renal pelvis at 61.4 ± 5.35 seconds.

Figure 1—
Figure 1—

Time-attenuation curves depicting mean concentrations of iodinated contrast medium (iodixanol or iohexol) in the abdominal aorta, renal cortex, whole kidney, and renal pelvis in 4 healthy Beagles in a crossover-design study evaluating the effects of changes in analytic variables and contrast medium osmolality on CT-GFRs. Measurements were initiated at the time of contrast medium injection (300 mg I/kg, 3.0 mL/s); there was a 2-week interval between GFR assessments. The CT-GFR values did not differ significantly between contrast medium types, and data from both experiments were combined (8 measurements/time point). The vertical dashed line indicates the second renal cortex peak time. Contrast medium was rapidly excreted to the renal pelvis beginning approximately 60 seconds after the injection started. EU = Enhancement units.

Citation: American Journal of Veterinary Research 78, 9; 10.2460/ajvr.78.9.1049

Table 1—

Comparison of mean ± SD CT-GFRs (mL/kg/min) for 4 healthy Beagles that received iodixanol and iohexol in a crossover-design study evaluating the effects of changes in analytic variables and contrast medium osmolality on these values.

VariableIodixanolIohexolP value
Renal cortex3.88 ± 0.104.31 ± 0.560.21
Whole kidney
 60 s3.70 ± 0.633.46 ± 0.410.33
 80 s3.42 ± 0.593.18 ± 0.470.35
 100 s3.21 ± 0.582.98 ± 0.470.34
 120 s2.99 ± 0.472.74 ± 0.420.19
Perfusion CT noise reduction
 Strong3.30 ± 0.233.64 ± 0.460.18
 Medium3.66 ± 0.163.95 ± 0.480.20

Iodixanol (290 mOsm/kg of water) or iohexol (672 mOsm/kg of water) was administered IV at a dose of 300 mg of I/kg (3 mL/s). There was a 2-week interval between the 2 assessments of CT-GFR. Values of P < 0.05 were considered significant.

The mean ± SD GFR estimated by serum inulin clearance in the 4 dogs was 4.23 ± 0.65 mL/kg/min; this value did not differ significantly from that for CT-GFR (with the datasets for iohexol and iodixanol combined [mean of 8 measurements]) from the manual renal cortex analysis (4.09 ± 0.44 mL/kg/min; P = 1.00), from the manual whole kidney analysis at 60 seconds after contrast appearance (3.58 ± 0.51 mL/kg/min; P = 0.26), and from the perfusion software analysis with medium noise reduction (3.80 ± 0.37 mL/kg/min; P = 1.00). However, the CT-GFR measurements determined from the manual whole kidney analysis at 80 (3.30 ± 0.51 mL/kg/min; P = 0.018), 100 (3.09 ± 0.50 mL/kg/min; P < 0.001), and 120 (2.86 ± 0.43 mL/kg/min; P < 0.001) seconds after contrast medium appearance, and from the perfusion software analysis with strong noise reduction (3.26 ± 0.34; P = 0.005) were significantly lower than the GFR estimated by inulin clearance. Further, the mean ± SD CT-GFR estimated by whole kidney analysis at 100 (P = 0.003) and 120 (P < 0.001) seconds after the appearance of contrast medium was lower than that estimated by analysis of the cortex.

Discussion

In the present study, 2 types of iodinated contrast medium with different osmolalities, iohexol and iodixanol, were used in dogs for determination of CT-GFR, with no significant difference in values for the renal cortex or for the whole kidney identified between methods. The apparent contrast effect decreases when a contrast medium with higher osmolality is administered because the contrast medium is diluted by osmotic diuresis in the renal tubules, and results of a previous study9 indicate that the starting time of contrast-derived osmotic diuresis influences the contrast effect at 2 minutes after contrast administration. The lack of a significant osmotic effect on the CT-GFR data in the present study might have been attributable to the time ranges for analysis, which were ≤ 120 seconds. However, these results were obtained from healthy young Beagles, and potential differences related to age or breed should be addressed in future studies.

Furthermore, we found that either of these 2 contrast media can be selected independently according to the status of the patient, because the choice did not significantly affect the CT-GFR. Results of a previous investigation11 suggested that there is a risk of renal failure (contrast-induced nephropathy) when contrast medium with a high osmolality is used. This, together with our study findings, suggests that an isotonic contrast medium such as iodixanol should be used for CT-GFR, which is primarily performed in patients with decreased renal function. However, this theory remains controversial.12

Only CT-GFR assessed at the renal cortex or for the whole kidney within 60 seconds after the appearance of contrast medium did not differ significantly from the GFR estimated by serum inulin clearance. In addition, the CT-GFR estimated by whole kidney analysis at 80, 100, and 120 seconds after the appearance of contrast medium was lower than that estimated by analysis of the cortex alone. This is because the glomerular filtration occurs in the renal cortex. The GFR is not the velocity of urine generation. The renal medulla contains filtrated urine rather than plasma; therefore, this is not the target of Patlak plot analysis. On the other hand, GFR might be considered as a change in the flow rate of contrast medium that is transferred to the medulla; however, it is difficult to adequately evaluate such a change because the contrast medium is concentrated by reabsorption of water at the medulla.

In the analysis involving the whole kidney, a longer time range led to greater underestimation of CT-GFR, compared with GFR determined by inulin clearance. Previous reports5,13 suggest an approximately 100-second time range is suitable for whole kidney analysis in human patients. Results of a study14 in pigs suggested that 120 to 180 seconds should be selected because, in that timeframe, the contrast medium is well mixed with blood and its concentration is constant. Some previous studies15,16 of dogs used a 120-second time range for the Patlak plot analysis. However, canine TACs are different from those generated for people because of differences in heart rate and physical size (which influence contrast medium transfer time and distance to systemic circulation). With the injection protocol performed in the present study, the contrast medium reached renal pelvis in 60 to 80 seconds and reached the ureters in 120 seconds in some dogs (data not shown). In this situation, the prerequisite of Patlak plot analysis, that the tracer should not have disappeared from the ROI during the target time range,10 cannot be satisfied because the renal pelvis and ureter are located outside of the ROI. In fact, the time period in which CT-GFR was not significantly different from the GFR estimated by inulin clearance was only 60 seconds. These results suggested that the time range for analysis should be limited to a maximum of 60 seconds from contrast administration if the whole kidney is selected as a target location in dogs. In addition, the time range should be set after examining the contrast appearance time at the renal pelvis of each dog.

Time range setting is important in analyses limited to the renal cortex. As a prerequisite for Patlak plot analysis, the time range should start when contrast medium appears and end immediately before the contrast medium moves out of the renal cortex. In the present study, the analysis time range for the renal cortex was from appearance of contrast medium at the renal cortex until immediately before its appearance at the renal medulla (approx 30 seconds from contrast administration).

The appearance time at the medulla was almost same as that for the second attenuation peak of the renal cortex (approx 3 seconds afterward). Therefore, the second peak for the renal cortex might be a landmark for the end time of analysis when the TAC of the renal medulla is unknown for analysis involving only the renal cortex (eg, when perfusion CT software is used). However, from this viewpoint, the use of healthy dogs in the present study might have been a limitation. Results of a previous study8 in which the first-moment method was used for GFR analysis suggested that the second cortical peak could be presumed to represent the contrast peak of the renal tubules. However, the TACs from the present study suggest that the second peak might also represent re-circulation of the contrast medium; therefore, it is a possibility that this peak is not relevant to the GFR. Briefly, there is a possibility that the cortical second peak does not always correspond to the timing of contrast transfer to the medulla (the time that should be selected as the analytical criterion) in patients with decreased renal function and an altered circulation status.

The difference in TACs between human subjects and dogs results in underestimation of GFR in dogs because of factors other than the time range for analysis. The protocol for injection of contrast medium in the present study was 300 mg of I/kg delivered at a rate of 3.0 mL/s. According to the Frank-Starling law,17 this injection speed in dogs corresponds to 2 to 3 times that in humans because of the differences in the heart rate. High-speed injection renders the peak attenuation time faster and the peak CT value higher than does a slow injection rate.18,19 A higher peak value tends to increase the AUC for the arterial TAC, and a larger AUC leads to smaller α/V in the formula used for Patlak plot analysis. Eventually, this would lead to underestimation of the GFR, so that, briefly, an injection speed that is too rapid leads to underestimation of GFR. However, our results suggested that the injection speed of 3.0 mL/s is within the allowable limits for dogs because the CT-GFR in the renal cortex analysis was the equivalent of the GFR estimated by inulin clearance in the present study.

Another report16 suggested that anesthesia has a role in the underestimation of CT-GFR in dogs. Anesthetics that provoke hypotension and deep anesthesia should be avoided during CT-GFR examination. Propofol, which was previously shown not to significantly alter GFR, compared with that in awake dogs,20 was used for maintenance of anesthesia in the present study. The effects of anesthesia on GFR are associated with lowered blood pressure. The mean ± SD of mean blood pressure during anesthesia in the present study was 107.9 ± 30.1 mm Hg, and in dogs, renal blood flow and GFR are maintained at a constant level by renal autoregulation when the mean blood pressure is between 65 and 180 mm Hg.16 We considered that the effect of anesthesia on GFR was minimal in the present study because the mean blood pressure did not deviate from this range in any dog.

Our results implied that the perfusion CT softwaren used in the present study can adequately measure CT-GFR in dogs, provided that the noise reduction level is appropriate. In the setting of the present study, application of the medium noise reduction level in the perfusion CT software resulted in a CT-GFR value that did not differ significantly from the GFR estimated by inulin clearance. Iterative reconstruction and noise reduction procedures are routinely performed on perfusion CT analysis because image noise disturbs mathematical procedures. However, excessive noise reduction will make the CT value distribution or border ill-defined in the setting where the SD of CT numbers in a given site is low (eg, when there is less body movement or when a higher radiation dose is used). The present study included a relatively high radiation dose, general anesthesia, and breath-holding techniques; therefore, the CT-GFR value at a medium level of noise reduction had better correspondence with the reference method than did use of a strong noise reduction level. Attention is required when using a higher dose of iodinated contrast medium for CT-GFR analysis, because excess contrast medium provokes streak artifacts. The level of noise reduction in the perfusion CT software should be selected depending on the SD of acquired images. The theory that excessive noise reduction leads to an underestimation of CT-GFR is also useful for human CT-GFR examinations.

Calculation of GFR by use of automated perfusion CT software is of clinical relevance. However, some issues need to be resolved for use of this software in evaluation of dogs. First, the shape, size, and number of ROI require further study. The available ROI shape was circular, the available size was slightly larger than the thickness of renal cortex, and only 1 ROI was available for use. A slight deviation in ROI location leads to a larger partial volume effect in dogs, because canine kidney size is smaller than that of humans (approx 24-mL volume, 25-mm transverse length, and 50-mm sagittal length in the present study). Therefore, a highly skilled operator is essential to ensure repeatability of CT-GFR analysis when perfusion CT software is used.

Important subjects for future investigation are as follows. In the present study, only 4 dogs were involved. However, a larger population size that includes dogs of various breeds, ages, sex and neuter statuses, and renal status should be investigated to better represent actual clinical practice. In addition, results from dogs undergoing other anesthetic regimens for CT-GFR should be compared with those from the present study. Moreover, we believe that highly skilled operators are essential for analysis with perfusion software; therefore, intra- and interobserver differences should be assessed in future investigations.

Acknowledgments

The authors had no external financial support for the study and no conflicts of interest to declare.

ABBREVIATIONS

AUC

Area under the curve

CT-GFR

Glomerular filtration rate estimated by CT

GFR

Glomerular filtration rate

ROI

Region of interest

TAC

Time-attenuation curve

Footnotes

a.

FUJIFILM Monolith Co Ltd, Tokyo, Japan.

b.

Inulead, Fuji Yakuhin Co Ltd, Saitama, Japan

c.

Atropine sulfate injection 0.5 mg, Mitsubishi Tanabe Pharmaceutical Co Ltd, Osaka, Japan.

d.

Vetorphale, Meiji Seika Pharma Co Ltd, Tokyo, Japan.

e.

Propofol intravenous injection 1%, Fresenius Kabi Japan, Tokyo, Japan.

f.

KVS-2100, Kohken Medical Co Ltd, Tokyo, Japan.

g.

Visipaque 320, Daiichi-Sankyo Co Ltd, Tokyo, Japan.

h.

Omnipaque 300, Daiichi-Sankyo Co Ltd, Tokyo, Japan.

i.

Autoenhance A-800, Nemoto Kyorindo, Tokyo, Japan.

j.

Aquilion CXL, Toshiba, Tochigi, Japan.

k.

VirtualPlace Fujin, AZE, Tokyo, Japan.

l.

Dynamic study, Toshiba, Tochigi, Japan.

m.

Microsoft Excel for Mac 2011, version 14.7.1, Microsoft Corp, Redmond, Wash.

n.

Body perfusion, Toshiba, Tochigi, Japan.

o.

SPSS, version 22.0, SPSS Inc, Chicago, Ill.

References

  • 1. Improving Global Outcomes (KDIGO) CKD Work Group. Kidney Disease: KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney 2013; S3:1150.

    • Search Google Scholar
    • Export Citation
  • 2. Svaland MG, Haider T, Langseth-Manrique K, et al. Human pharmacokinetics of iodixanol. Invest Radiol 1992;27:130133.

  • 3. Heglund IF, Michelet ÅA, Blazak WF, et al. Preclinical pharmacokinetics and general toxicology of iodixanol. Acta Radiol Suppl 1995;399:6982.

    • Search Google Scholar
    • Export Citation
  • 4. Jacobsen PB, Blindheim L, Skotland T. Bioanalytical methods for iodixanol and their application to studies on metabolism and protein binding. Acta Radiol Suppl 1995;399:6166.

    • Search Google Scholar
    • Export Citation
  • 5. Tsushima Y, Blomley MJ, Okabe K, et al. Determination of glomerular filtration rate per unit renal volume using computerized tomography: correlation with conventional measures of total and divided renal function. J Urol 2001;165:382385.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6. Hackstein N, Bauer J, Hauck EW, et al. Measuring single-kidney glomerular filtration rate on single-detector helical CT using a two-point Patlak plot technique in patients with increased interstitial space. AJR Am J Roentgenol 2003;181:147156.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Alexander K, Dunn M, Carmel EN, et al. Clinical application of Patlak plot CT-GFR in animals with upper urinary tract disease. Vet Radiol Ultrasound 2010;51:421427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Daghini E, Juillard L, Haas JA, et al. Comparison of mathematic models for assessment of glomerular filtration rate with electron-beam CT in pigs. Radiology 2007;242:417424.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Kishimoto M, Yamada K, Watanabe A, et al. Comparison of excretory urographic contrast effects of dimeric and monomeric non-ionic iodinated contrast media in dogs. J Vet Med Sci 2007;69:713715.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Patlak CS, Blasberg RG. Graphical evaluation of blood-to-brain transfer constants from multiple-time uptake data. J Cereb Blood Flow Metab 1985;5:584590.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Aspelin P, Aubry P, Fransson SG, et al. Nephrotoxic effects in high-risk patients undergoing angiography. N Engl J Med 2003;348:491499.

  • 12. Thomsen HS, Stacul F. CIN: can we forget it? Acta Radiol 2014;55:10271030.

  • 13. Hackstein N, Wiegand C, Rau WS, et al. Glomerular filtration rate measured by using triphasic helical CT with a two-point Patlak plot technique. Radiology 2004;230:221226.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Alexander K, Del Castillo JR, Ybarra N, et al. Single-slice dynamic computed tomographic determination of glomerular filtration rate by use of Patlak plot analysis in anesthetized pigs. Am J Vet Res 2007;68:297304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15. Alexander K, Authier S, del Castillo JR, et al. Patlak plot analysis CT-GFR for the determination of renal function: comparison of normal dogs with autologous kidney transplant dogs. Contrast Media Mol Imaging 2010;5:133139.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. O'Dell-Anderson KJ, Twardock R, Grimm JB, et al. Determination of glomerular filtration rate in dogs using contrast-enhanced computed tomography. Vet Radiol Ultrasound 2006;47:127135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Lakatta EG. Starling's law of the heart is explained by an intimate interaction of muscle length and myofilament calcium activation. J Am Coll Cardiol 1987;10:11571164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 18. Han JK, Kim AY, Lee KY, et al. Factors influencing vascular and hepatic enhancement at CT: experimental study on injection protocol using a canine model. J Comput Assist Tomogr 2000;24:400406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Han JK, Choi BI, Kim AY, et al. Contrast media in abdominal computed tomography: optimization of delivery methods. Korean J Radiol 2001;2:2836.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Fusellier M, Desfontis JC, Madec S, et al. Influence of three anesthetic protocols on glomerular filtration rate in dogs. Am J Vet Res 2007;68:807811.

    • Crossref
    • Search Google Scholar
    • Export Citation

Contributor Notes

Address correspondence to Dr. Kishimoto (285copernicium@gmail.com).