Intraindividual In Vivo Comparison of Gadolinium Contrast Agents for Pharmacokinetic Analysis Using...

ORIGINAL ARTICLE

Intraindividual In Vivo Comparison of Gadolinium ContrastAgents for Pharmacokinetic Analysis Using Dynamic Contrast

Enhanced Magnetic Resonance ImagingJiachao Liang, PhD, Steffen Sammet, MD, PhD, Xiangyu Yang, PhD, Guang Jia, PhD,

Yukihisa Takayama, MD, and Michael V. Knopp, MD, PhD

Purpose: To compare the intraindividual differences of dynamic signalcharacteristics for 3 gadolinium chelates (gadopentetate dimeglumine �Gd-DTPA�, gadodiamide �Gd-DTPA-BMA�, and gadobenate dimeglumine �Gd-BOPTA�) using dynamic contrast enhanced magnetic resonance (MR) im-aging (DCE-MRI) with a preclinical beagle model at 7 Tesla.Method and Materials: Seven beagles were scanned 3 times each with a7-day interval between the scans on a 7T whole body MRI system (Achieva,Philips) using a T/R head coil. Three different Gd contrast agents includingGd-DTPA, Gd-DTPABMA, and Gd-BOPTA were injected in a randomizedorder with a power injector (Spectris, MedRad, Indianola, PA) using dose of0.1 mmol/kg body weight and flow rate of 0.06 mL/s. During imageacquisition and data analysis, the identity of the specific contrast agent usedfor each examination was blinded. A 3D RF-spoiled fast field echo sequencewas used for dynamic scans with in-plane spatial resolution 0.47 � 0.47mm2, temporal resolution of 9.5 seconds, and a total of 60 time points.Regions of interest were drawn within the carotid arteries and muscle tissueto determine semiquantitative parameters including maximum enhancementratio, area under the signal enhancement curve over 90 seconds after contrastinjection (AUC_90), time to maximum signal enhancement (Tmax), andwashout_score. Additionally, quantitative pharmacokinetic parameters weremeasured in muscle tissues by applying 3 separate 2-compartment models;(1) artery input function (AIF) based Tofts model, (2) Brix model withoutAIF, and (3) AIF decomposed refined Brix model.Results: Gd-BOPTA produced higher signal to noise ratio on postcontrastT1- weighted images than the other 2 Gd based contrast agents at 7T.Quantitatively, Gd-BOPTA provided a significantly higher maximum en-hancement ratio (P � 0.01), AUC_90 (P � 0.01) and washout_score (P �0.01) in beagle musculature and cranial vasculature compared with bothGd-DTPA and Gd-DTPA-BMA. Among all the quantitative pharmacoki-netic parameters, only the exchange rate constants (kep) calculated fromthese 3 models did not show a significant difference among the variouscontrast agents.Conclusions: Gd chelate containing MR contrast agents can be used at 7Tfor DCEMRI. Gd-BOPTA demonstrates stronger signal enhancement thanstandard Gd chelates, in concordance with the results of studies at lowerfields. The observed enhancement characteristics for the 3 contrast agentsdemonstrate that the pharmacokinetic parameter kep is more robust invarious models using DCE-MRI than the other pharmacokinetic parameters.This information is important relative to multisite clinical trials and long-term clinical studies that often use several different contrast agents anddifferent models.

Key Words: contrast agent, DCE-MRI, gadobenate dimeglumine, 7T,crossover comparison

(Invest Radiol 2010;45: 233–244)

The number of magnetic resonance (MR) studies which includethe use of MR contrast agents has been steadily increasing.1

Currently, approximately half of all MR studies, world-wide use MRcontrast agents.2 Extracellular gadolinium-based contrast agents arethe major MR contrast media used in clinical applications.3,4 Todate, there are 6 gadolinium(Gd)-based contrast agents approved bythe Food and Drug Administration in the United States: gado-pentetate dimeglumine (Gd-DTPA), gadoteridol (Gd-HP-DO3A),gadodiamide (Gd-DTPA-BMA), gadoversetamide (Gd-DTPA-BMEA), gadobenate dimeglumine (Gd-BOPTA), and gadoxeticacid. As the number of available MR contrast agents grows, knowl-edge of their enhancement characteristics becomes essential. Adetailed understanding of the similarities and differences betweenthe enhancement characteristics of different contrast agents willallow physicians to choose the most appropriate agent for the studybeing ordered. This will contribute to better disease detection andcharacterization, and further improve monitoring and guidance oftreatment planning.5

Functional, dynamic contrast-enhanced MR imaging (DCE-MRI) is currently being investigated for its potential to quantita-tively characterize the microcirculation in tumors and changesduring therapy.6 To accomplish this, T1-weighted acquisitions areused in DCE-MRI to observe the uptake of Gd-based contrastagents, and obtain information related to microvasculature proper-ties. Next, an 8 to 10 minute long dynamic scan is planned inDCE-MRI with a series of repetitions.7 The injection of contrastagents are started after several baseline repetitions during the dy-namic acquisition. The signal intensity (SI) changes during the timecourses of contrast agent uptake are recorded. Using this method-ology, studies have shown DCE-MRI to be a promising method forcomparing the enhancement and pharmacokinetic characteristicsbetween different contrast agents.

Numerous intraindividual comparative studies have been con-ducted evaluating different Gd agents at equivalent dose and equiv-alent magnetic field strengths between 1.5 to 3.0T.8–10 Nonspecificextracellular standard Gd chelates (Gd-DTPA, Gd-DTPA-BMA) areextracted, almost exclusively, by passive glomerular filtrationthrough the kidneys, without secretion or reabsorption. Additionally,these agents have similar pharmacokinetics and T1 relaxivity.3

Furthermore, these standard agents have similar enhancement char-acteristics in DCE-MRI within both the vasculature and in tumors at1.5T.7 Gd-BOPTA is an agent that interacts weakly with protein andhas both renal and hepatobiliary elimination pathways.11,12 Atequivalent doses, Gd-BOPTA has shown higher signal enhancementcharacteristics and has better lesion delineation in both the centralnervous system10,13,14 and noncentral nervous system15,16 imaging

Received October 30, 2008; and accepted for publication (after revision) January16, 2010.

From the Department of Radiology, The Ohio State University, Columbus, OH.Supported by the biomedical structural, functional and molecular imaging enter-

prise, Wright Center of Innovation in Biomedical Imaging at the Ohio StateUniversity, Grant ID: ODOD AGMT TECH 03–051.

Reprints: Michael V. Knopp, MD, PhD, Department of Radiology, The Ohio StateUniversity, 657 Means Hall, 1654 Upham Drive, Columbus, OH 43210.E-mail: [email protected].

Copyright © 2010 by Lippincott Williams & WilkinsISSN: 0020-9996/10/4505-0233

Investigative Radiology • Volume 45, Number 5, May 2010 www.investigativeradiology.com | 233

http://www.investigativeradiology.com

applications. This higher signal enhancement is due to its capacityfor weak and transient interaction with serum albumin.17 Thisinteraction leads to approximately double the in vivo T1 relaxationrate of other standard gadolinium chelates: r1 � 9.7 (mmol/L)�1 � sec�1 as compared with 4.3 and 5.6 (mmol/L) �1 � sec�1 forpurely extracellular agents.17,18

Gd-DTPA was first evaluated for DCE-MRI in 1986 in thebreast.19 The modeling of DCE-MRI was established based onGd-DTPA in the early 1990s.20,21 The majority of reported DCE-MRI studies are based on Gd-DTPA. Because of the developmentand approval of novel contrast agents in the past 2 decades, otherMR contrast agents, particularly agents that interact weakly withprotein, have gained more and more popularity in DCE-MRIstudies.22

Prior studies established the differences in the contrast en-hancement characteristics for vascular and tissue imaging betweenagents with weak protein interactions (Gd-BOPTA) and standardgadolinium chelates (Gd-DTPA, Gd-DTPA-BMA). However, therehas been no prospective study performed in vivo with an intraindi-vidual comparison of the enhancement characteristics between these2 classes of agents. We undertook this prospective study with thehypothesis that there is no difference in the contrast enhancementcharacteristics for DCE-MRI for the quantitative pharmacokineticanalysis of DCE-MRI using Gd-DTPA, Gd-DTPA-BMA, and Gd-BOPTA.

Different enhancement and pharmacokinetics characteristicsof agents with weak protein interactions, will display differentDCE-MRI time-SI curves in DCE-MRI. The arterial input function(AIF) in DCE-MRI will also be different whether there is a differ-ence in the physicochemical properties such as osmolality andviscosity in various Gd agents.2,23 All of these differences willfurther influence the reliability and accuracy of semi-quantitativeand quantitative assessments in DCE-MRI evaluations with orwithout AIF. As DCE-MRI becomes more prevalent in clinical trialsand long-term clinical studies,24,25 establishing reliable and accuratequantitative assessments is becoming an increasing concern forresearchers.26,27 Various selections of contrast agents should be apotential confounding factor that may influence quantitative assess-ments. Recent studies demonstrated the inverse relationship betweenthe molecular weight of contrast agents and quantitative pharmaco-kinetic parameters including transendothelial permeability (KPS) andfractional plasma volume (fpV).28,29 Six Food and Drug Adminis-tration approved Gd-based contrast agents do not vary only in arange from 574 to 1058 Da. It is still important to study the effectsof each Gd-based contrast agent on the quantitative abilities ofDCE-MRI. One of the goals of this investigation was to assess thepotential differences in the quantitative characterization of theDCE enhancement using semi-quantitative approaches, the Toftsmodel,20,30 the Brix model21,31 and the modified Brix model.32

7T MR scanners can provide higher signal to noise ratio(SNR) which may increase temporal or spatial resolution compare

with the possibilities at lower field strength, and thus it is potentiallyable to produce advanced clinical MR applications.33 These advan-tages of 7T may help to differentiate the contrast enhancementcharacteristics between different types of contrast agents.

Because of the high contrast of soft tissues, MR imaging isaccepted as the most sensitive method for brain imaging.34 The useof MR contrast agents for improving visualization of brain imagingwith standard single dose (0.1 mmol/kg body weight) is wellestablished, although numerous studies have shown that the higherdose and dedicated sequences might improve lesion detection anddelineation.35–37 In this study, we imaged the beagle dog brain byadministering single dose contrast agent.

This is an intraindividual crossover comparison study inwhich a preclinical in vivo beagle model for DCE-MRI in ourclinical 7T system is used to compare the signal characteristics oftime-SI curves for 3 different Gadolinium chelates: Gd-DTPA,Gd-DTPA-BMA, and Gd-BOPTA.

METHODS AND MATERIALS

Objects and Gd-Contrast Agents Injection PlanningTo have the highest level of standardization we chose an

established beagle imaging model of 7 healthy beagle subjects thatwill be randomly assigned for the intraindividual comparison cross-over design study. The subjects, which were used under an approvedanimal use protocol, had a mean age of 2 years and a mean weightof 9.6 � 2.5 kg (range, 12.4–7.8 kg). Each subject underwent 3separate MRI examinations within a 7-day interval. Each subjectwas randomly injected with 1 of 3 extracellular Gd-based contrastagents that are available in the global marketplace: Gd-DTPA(Magnevist; Berlex Inc., Montville, NJ), Gd-DTPA-BMA (Omnis-can; GE Healthcare, Princeton, NJ) and Gd-BOPTA (MultiHance;Bracco Diagnostics Inc., Princeton, NJ). A power injector (Spectris,MedRad, Indianola, PA) was used during the contrast agent injec-tion with body weight dosage of 0.1 mmol/kg BW and flow rates of0.06 mL/s. This allows for the injection of contrast agent volumewithin a 30-second window (0.6 mL/s per patient). A total of 20 mLsaline was subsequently injected with 0.3 mL/s flushing rate follow-ing after contrast agent injection. Everyone except the drug dispens-ing person was blinded regarding the use of the contrast agents forthe individual subject. A phantom scan with different concentrationsof contrast agents was performed before the study as well as themethodology for reliable injection validation. The sequence param-eters were optimized to achieve the most linear relationship betweencontrast agent concentration and signal enhancement for the DCE-MRI sequence.

Power Injector SetupTo ensure no safety issue at 7T, we located the power injector

outside of the 7T scanner room and connected the subjects viaextended tubing sets (Fig. 1) through a dedicated waveguide that is

FIGURE 1. Diagram of injection tubing system setup, including one 96� low pressure connector tube from a disposable MRIpower injector kit plus total five 60� low pressure connector tubes. The whole tubing system filled full of Saline, except thefinal ending tube filled full with signal dose contrast agent and saline which were locked by 2 stopcocks before injectionstarted.

Liang et al Investigative Radiology • Volume 45, Number 5, May 2010

234 | www.investigativeradiology.com © 2010 Lippincott Williams & Wilkins234


part of site design for such purposes. For every experiment adisposable MRI kit for spectris solaris (MedRad, Indianola, PA) wasused that contains one 96� low pressure connector tube. The volumeof the tube is 4 mL and we connected a total of 4 60� low pressureconnector tubes (MedRad, Indianola, PA) for the full setup. Thesetubes with a volume of 2.5 mL each connected with a Luer-lock tothe 96� tube. This long tubing was initially filled with saline via apower injector until all air bubbles were removed from the system.Additionally, one final 60� extension tube was prepped by injectingthe amount of contrast agent used in the study followed by thesaline. This contrast agent tube was locked by 2 4-way stopcocksand these 2 stopcocks were connected with prior long tubing and thestudy subject separately. The combined tubing length is 396� longand contains a total volume of 16.5 mL.

With the used flow rate and considering viscosity of thecontrast agents, no elevated inline pressure occurred that would havelet to dilation of the tubing. The tubing was attached to the tableusing adhesive tape. Before connecting the tubing set to the subject,a leakage test was performed. All connecting points between tubeswere visually checked for leakage. The stopcocks were manuallyopened right before the injection started during the dynamic acqui-sitions.

MRI ExaminationsAll dog scans were performed on a 7T whole body MRI

system (Achieva, Philips, Cleveland, OH) using the T/R head coil.High resolution T2-weighted images were acquired using turbo spinecho sequence for the delineation of beagle dog brain and substruc-ture. The parameters for T2-weighted images were: repetition time(TR): 3000 ms, echo time (TE): 70 ms, flip angle: 90°, field of view:150mm; matrix: 512 � 512 with in plane resolution 0.29 � 0.29mm2; slice thickness: 2 mm; gap between slices: 1 mm; 30 slices.

Precontrast and postcontrast axial T1-weighted images wereacquired using 3D-IR-prepared turbo field echo (3D-T1w-TFE)sequence for neurovascular identification and contrast enhancement

comparison (Fig. 2). The parameters for axial T1-weighted imageswere: TR: 11.4 ms, TE: 5.6 ms, inversion short interval time/inversion time: 1200 ms/700 ms, flip angle: 8°, field of view: 150mm; matrix: 512 � 512 with in plane resolution 0.59 � 0.59 mm2;slice thickness: 3 mm; 30 contiguous slices.

DCE-MRI was performed using a 3D T1-weighted RF-spoiled fast field echo (3D-FFE) sequence. The T1 weighted 3D-FFEsequence using the parameters: TR: 7.8 ms; TE: 4 ms; FA: 20°; fieldof view: 120mm; matrix: 256 � 256; in-plane resolution: 0.47 �0.47 mm2; slice thickness: 4 mm; 16 contiguous slices; temporalresolution 9.5 seconds, 60 time points. The acquisition time ofDCE-MRI was about 9 minutes. Contrast agent injection startedafter 5 repetitions.

Data ProcessingSNR assessments were done simultaneous to precontrast

T1-weighted images and postcontrast T1-weighted images usingregions of interest (ROIs) in each examination. ROIs were placed inthe temporalis muscle and the background. Additional sets of ROIswere placed for the assessment of the dynamic acquisitions. ROIswere placed in the temporalis muscle and carotid arteries of bothsides to minimize the influence of B1 inhomogeneity at 7T (Fig. 3).All ROI placements were placed by readers in a blinded fashion andeach session was independent from one another.

SNR was determined by the following formula:

SNR � SI normal muscle tissue/(SI background/1.25) (1)

The noise standard deviation is calculated by (SI back-ground/1.25) in accordance to the assumption of Rician noise inmagnitude images, which is justified by the use of a dual channelT/R coil.38 The SNR assessments were performed using theMedical Image Processing, Analysis, and Visualization softwarepackage.39

FIGURE 2. 7T T1 weighted images before and af-ter Gd contrast agent injection. A, Pre CA injec-tion T1 weighted image. B, Post CA injection T1weighted image after injection of Gd-BOPTA. C,Post CA injection T1 weighted image after injec-tion of Gd-DTPA. D, Post CA injection T1weighted image after injection of Gd-DTPA-BMA.Gd-BOPTA visually observed better signal en-hancement and higher SNR in the postcontrastimages.

Investigative Radiology • Volume 45, Number 5, May 2010 Comparison of Gadolinium Contrast Agents

© 2010 Lippincott Williams & Wilkins www.investigativeradiology.com | 235235


Qualitative assessments from DCE-MRI acquisitions weredone using the DCE-MRI time-SI curves determined from ROIs bysemi-quantitative and quantitative parameters. Time-SI curves, de-rived from carotid artery ROI, were used as AIF. Precontrast, 7minutes postinjection, and total 9 minutes acquisition data were usedin the time-SI curves (Fig. 4).

The semi-quantitative parameters were measured from thetime-SI curves within ROIs in both muscle tissue and carotid artery.This included: maximum enhancement ratio (MER, �%�), area-under-the-curve during 90 seconds since enhancing (AUC_90,�a.u.�), time to maximum signal enhancement (Tmax,�min�), andwashout-score �%�40 as the relative percentage changes between themaximum signal enhancement and the SI at the end of the dynamicscan (Fig. 5).41

Additionally, the SI data was used for pharmacokinetic anal-ysis by fitting the time-SI curves according to the two-compartmentmodels with20,32 and without21 AIF. For comparison, we have used3 different models (Fig. 6): the Tofts model,20 the Brix model,21 andthe AIF decomposed refined Brix model.32

The basic equations of the 2 compartment models are:

Vp

dCp

dt� kepVeCe � kpe � kelVpCp � Kin (2)

Ve

dCe

dt� kpeVpCp � kepVeCe (3)

Where Ce is the CA concentration in the peripheral compart-ment, the extravascular extracellular space (EES). Cp is the CA con-centration in the plasma compartment. Vp and Ve are the correspondingvolumes of these 2 compartments. Kpe and Kep are the one-way contrastagent transfer rate constants between the plasma compartment and theEES compartment. Kel is the contrast agent elimination rate constant.Kin is the contrast agent infusion rate constant. All 3 pharmacokineticmodels are drove from these 2 equations by using different assumptionsabout the plasma compartment (Appendix).

FIGURE 3. Regions of interest (ROIs) were drawn in the dynamic images within the carotid arteries and within left side andright side of the beagle brain muscle tissues. The time-SI curves were generated from these ROIs.




All DCE-MRI data analysis was performed using in-housedeveloped software based on the IDL environment (Interactive DataLanguage, Research Systems Inc., Boulder, CO). PARREC (thePhilips research format) images were used during the analysisinstead of DICOM images to avoid scaling and rescaling factors.

Statistical AnalysisTwo-tailed Student t test was used to compare all semi-

quantitative and quantitative parameters between Gd-BOPTA andother 2 Gd contrast agents (Gd-DTPA and Gd-DTPA-BMA). TheBonferroni method was used for multiple comparisons, with anoverall significance level of 0.1 and 0.05 for semi-quantitative andpharmacokinetic parameters, respectively. A higher significancelevel (P � 0.05) was chosen for the pharmacokinetic parametersbecause they are less prone to noise and temporal resolution, and arethus expected to be more sensitive in detecting subtle differencesbetween the time-SI curves than the empirical semi-quantitativeparameters.

Vertical boxplots for each individual parameter were drawnin the statistic software SPSS (SPSS Inc., Chicago, IL) to comparebetween 3 contrast agents. The boundaries of the box are Tukey’shinges. The inside line identify the median. The length of the box is

the interquartile range (IQR) computed from Tukey’s hinges. As-terisk sign (*) presents the extreme point which has the value morethan 3 IQR’s from the end of a box. Outliers sign (o) represents thevalue more than 1.5 IQR’s but less than 3 IQR’s from the end of thebox.

RESULTS

Qualitative AssessmentsA total of 24 study scans were performed, including one

failed scan (without signalenhancement) because of an incorrectinjection, and 2 failed scans because of motion artifacts from studysubjects. The technical quality of images was deemed sufficient inall other 21 scans. Visual inspection of the time-SI curves demon-strates that Gd-BOPTA has a stronger signal enhancement effectthan Gd-DTPA and Gd-DTPA-BMA at equivalent dosages. Thewash-out rate of Gd-BOPTA is also noticeably faster than the other2 agents (Fig. 3) in both arteries and muscles.

Quantitative AssessmentsThe signal enhancement properties of the 3 contrast agents

(Gd-BOPTA, Gd-DTPA, and Gd-DTPA-BMA) were assessed by

5

6

Comparision of Artery CurvesR

atio

Gd-BOPTAGd-DTPAGd-DTPA-BMA

A

1

2

3

4

Enha

ncem

ent R

0 1 2 3 4 5 6 7 8 90

Time [min]

1.7

1.8

Comparision of Muscle Curves

io

B

1.1

1.2

1.3

1.4

1.5

1.6

Enh

ance

men

t Rat

Gd-BOPTAGd-DTPA

0 1 2 3 4 5 6 7 8 90.9

1

Time [min]

E

Gd DTPAGd-DTPA-BMA

FIGURE 4. The comparisons of time-SI curves between different Gd contrast agent injections (Gd-BOPTA, Gd-DTPA and Gd-DTPA-BMA) for same beagle dog. A, Comparison of artery curves, Gd-BOPTA showed a readily higher signal enhancementthan other 2 Gd contrast agents, and a relative faster wash in and relative faster washout were observed from Gd-BOPTA. B,Comparison of muscle curves, Gd-BOPTA showed a similar higher signal enhancement than other 2 Gd contrast agents innormal muscle curve, and a significant earlier maxima enhancement was observed followed by a faster wash out lope thanother 2 agents.




measuring the signal to noise ratio (SNR) on the axial cross-sectionof temporalis muscle in pre- and postcontrast T1-weighted images(Fig. 2). The SNR increase was 45.3% with Gd-BOPTA. Thiswas significantly lower at 18.8% with Gd-DTPA and 29.8% withGd-DTPA-BMA. The results of the SNR comparison are listed inTable 1.

The pharmacokinetic enhancement characteristics of the 3contrast agents are characterized by the signal enhancement profilesthat were analyzed by the DCE-MRI data sets. The interstitialcontrast agent redistribution process can be quantified with time-SIcurves by using the semi-quantitative analysis of the artery time-SIcurves utilizing the following parameters: the average MER, thewashout_score, AUC_90, and Tmax. The statistical comparison re-sults from cranial vasculature ROIs are listed in Table 2. Amongthese semi-quantitative parameters (Fig. 6), the average MER inGd-BOPTA (4.10 � 0.71 �%�) was significantly higher than inGd-DTPA and Gd-DTPA-BMA. Both the washout_score(61.67% � 4.14%) and AUC_90 (5.97 � 0.52) in Gd-BOPTAwere also significantly higher. An earlier Tmax was also observedfrom the Gd-BOPTA group (0.59 � 0.12 minutes; with Gd-DTPA 0.69 � 0.08 minutes, and Gd-DTPA-BMA 0.68 � 0.10minutes), however, these differences were not determined to bestatistically significant.

Similarly, the musculature uptake properties of the 3 contrastagents were investigated by assessing the muscular time-SI curvesutilizing the same semi-quantitative parameters (Fig. 7). Addition-ally, a more sophisticated, model-based pharmacokinetic analysiswas used (Figs. 8, 9). Statistical comparison results for thosesemi-quantitative parameters are listed in Table 3. MER (0.51 �0.05 �%�), washout_score (20.10% � 10.76%), and AUC_90 (0.42 �0.17) in Gd-BOPTA were significantly higher than in Gd-DTPA andGd-DTPA-BMA. Tmax in Gd-BOPTA (3.76 � 1.33 minutes) wasalso significantly earlier than in Gd-DTPA-BMA, whereas the sig-nificance of the difference between Gd-BOPTA and Gd-DTPA ismarginal (4.60 � 0.89 minutes, P � 0.014).

Pharmacokinetic analysis of the musculature DCE-MRI datawas performed with 3 different models (Fig. 8). Statistical compar-ison results of the pharmacokinetic parameters from those modelsare listed in Table 4.

In Tofts model, Ktrans from the Gd-BOPTA group (0.12 �0.04 minutes�1) was significantly higher than the Gd-DTPA andGd-DTPA-BMA groups. In the Brix model, Amp and kel weresignificantly higher in the Gd-BOPTA group. In the refined Brixmodel, the difference of kep

RB between Gd-BOPTA and the other 2contrast agents were not statistically significant. Further, none of the3 kep

�s kepRB, kep

B , kepRB were determined to be significantly different

between Gd-BOPTA and the other 2 Gd chelates (Fig. 9).

DISCUSSIONThis preclinical in vivo study systematically compared the

signal enhancement behaviors of a weak protein interacting Gd-based contrast agent (Gd-BOPTA) and 2 standard Gd chelates(Gd-DTPA and Gd-DTPA-BMA) at 7T, which shows the feasibilityto provide a better signal to noise ratio. All 3 contrast agents weredemonstrated to have similar enhancement patterns at 7T comparedwith 1.5T and 3T, however, both qualitative visual inspection andquantitative analysis led to rejection of our primary hypothesis thatthere is no difference in the contrast enhancement characteristicsbetween Gd-BOPTA, Gd-DTPA, and Gd-DTPA-BMA. Consistentwith previous findings at lower field strengths,9,10,42 Gd-BOPTAdemonstrated to have a stronger signal enhancement capability thanthe standard Gd chelates at equivalent dosing at 7T. This visualassessment was also verified by the SNR measurement and thesemi-quantitative parameters (MER and AUC_90) from DCE-MRI.Compared with Gd-DTPA and Gd-DTPA-BMA, Gd-BOPTA isknown to have a capacity for weak and transient interactions withserum albumin that enhances proton T1 relaxivity. Whereas thismechanism is supported by solid experimental data at lowerfields,11,12 it has never been explicitly demonstrated at 7T before.

1 5

1.6

Semi-quantitative Parameters

1.4

1.5

Rat

io

MER

WashoutScore

1.2

1.3

nhan

cem

ent

1

1.1E

Tmax

0 1 2 3 4 5 6 7 8 90.9

Time [min]

a

FIGURE 5. Sample time-SI curve ac-quired using Gd-BOPTA of muscle thatwas annotated to identify how the semi-quantitative parameters were deter-mined, including maximum enhance-ment ratio (MER, �%�), time tomaximum signal enhancement (Tmax,�min�) and washout-score �%�.




The stronger signal enhancement behavior of Gd-BOPTA has po-tential clinical significance in that it may improvement of smalllesion visualization and microvasculature quantification at 7T.

Among the semi-quantitative parameters derived from DCE-MRI data, washout_score and AUC_90 were demonstrated to havesignificantly higher values for Gd-BOPTA than Gd-DTPA andGd-DTPA-BMA with equivalent dosage. This contrast-agent-dependent difference in semi-quantitative parameter measurementhas never been reported in literature before, and should be taken intoconsideration in multisite clinical studies in which different contrastagents were involved. In vivo pharmacokinetic characteristics ofGd-based contrast agents are described by a bi-exponential pharma-cokinetic model, and identified by 2 half-life time constants of thedistribution phase (t1/2�) and the elimination phase (t1/2�).Gd-BOPTA has been reported in prior studies to have fasterdistribution and elimination (t1/2� � 0.084 � 0.012 hours, t1/2� � 1.20 � 0.09 hours) than Gd-DTPA (t1/2� � 0.20 � 0.13hours, t1/2� � 1.51 � 0.28 hours).43 In our study, however, theDCE-MRI sequence was executed within 9 minutes, with apostcontrast acquisition of around 7 minutes. Within the 7 min-utes postinjection time frame, the observed signal enhancementbehavior is predominantly because of the distribution phase ofthe contrast agent. As a result, the observed differences inwashout_score and AUC_90 reflect different pharmacokineticproperties of the interstitial distribution of the contrast agentmolecules, rather than their renal/hepatobiliary clearance.

Similar contrast-agent-dependent differences were also ob-served in the pharmacokinetic parameter Ktrans from Tofts model,and Amp and kel

B from Brix model, whereas all the kep values derivedfrom 3 different pharmacokinetic models (Tofts’s kep

T , Brix’s kepB , and

the redefined Brix’s kepRB) are more independent from the choice of

contrast agent. When Tofts and Brix models were first introduced tothe research community in the early 1990s, Gd-DTPA was theprimary and only approved Gd-based MR contrast agent on themarket. Because of the development and approval of novel contrastagents in the past 2 decades, hospitals and research institutes havegained more and more flexibility in the choice of Gd-based contrastagents. Thus, to establish the consistency of semi-quantitative andquantitative DCE-MRI parameters among different contrast agentchoices becomes an issue of crucial importance for multisite andlong-term longitudinal studies because these studies do not tend touse the same contrast agent from center to center. Our data suggestthat the robustness of kep values in contrast agent selection might bean advantage to the other semi-quantitative and pharmacokineticparameters in this scenario. This consideration might gain moreimportance in the future because DCE-MRI pharmacokinetic param-eters are considered to be promising candidates as imaging biomar-kers for tumor diagnosis, treatment monitoring, and prognosis as-sessment.24

Theoretically, kepT and kep

B values should be identical becauseTofts model and Brix model are derived from the same set ofpharmacokinetic equations, only with different a prior assump-tion.20,21 However, the consistency between those 2 parameters wasrarely checked in previous studies, leaving it an open questionwhether different pharmacokinetic models are truly equivalent forreal data. Results from our study address this question directly.Despite the lack of valid statistics because of the small sample size(only 7 beagles), it is clear from Figure 9 that kep

T and kepRB share

reasonably good consistency, whereas kepB is significantly larger than

both of them. The consistency between kepT and kep

RB supports thetheoretically established equivalency between different pharmaco-kinetic models, and thus suggests that pharmacokinetic parametersderived from Tofts model and the refined Brix model are directlycomparable. Given the observation that kpe

RB values are consistently

TABLE 1. Summary of Signal-to-Noise Ratios (SNR) andAverage Percentage Changes Comparing Before and AfterContrast Agent Administration

Gd-BOPTA Gd-DTPA Gd-DTPA-BMA

Precontrast SNR 35.9 � 7.9 36.5 � 10.3 30.1 � 6.4

Postcontrast SNR 52.4 � 14.1 42.1 � 7.7 38.5 � 12.2

Average changes 45.3% 18.8% 29.8%

FIGURE 6. Three 2-compartment pharmacokinetic modelswith or without AIF describe the exchange between theblood plasma compartment and extravascular extracellularspace (EES) compartment. Tofts model (A) uses a known AIF,either simulated from a known blood sample data or mea-sured from an artery ROI in DCE-MRI data sets. Brix model(B) assumes an unknown Cp and derives directly from mus-cle time-SI curve, Ct . Refined Brix model (C) uses AIF to de-compose the exchange rates between the plasma compart-ment and the EES compartment based on original Brixmodel.




comparable to kepRB in magnitude, the basic assumption that kep �� kpe

in Brix model is most likely to be invalid for this data set. Thiswould lead to an overestimation of kep

B as the sum of kep andkpe. A more detailed discussion about this issue was given inreference.25

A higher Ktrans was observed in Gd-BOPTA group. This isassumed to be the result of 2 unique properties of Gd-BOPTAdistinct from other conventional Gd-chelates: high excretion ratios

(32% or 43%) from hepatobiliary system in rats or dogs and weakbinding to proteins.44 The first property caused a significant higherexchange rate k12 for Gd-BOPTA in rat liver.45 This could also bethe reason of a higher Ktrans of Gd-BOPTA in our beagle study usingTofts model. The concept of dynamic relaxivity of protein bindingagents was introduced by considering characteristics of the pharma-cokinetic profile.46 With respect to this concept, it may be necessaryto integrate the impact of dynamic relaxivities to improve pharma-

TABLE 2. Summary of Semi-Quantitative Parameters in Artery (Bonferroni Method, With an Overall Significance Level of 0.1)

Gd-BOPTAMean � SD

Gd-DTPA Gd-DTPA-BMA

Mean � SD P Mean � SD P

MER (%) 4.10 � 0.71 3.45 � 0.57* 0.01 3.48 � 0.64* 0.007

Tmax (min) 0.59 � 0.12 0.69 � 0.08 0.05 0.68 � 0.10 0.06

Washout_score (%) 61.67 � 4.14 53.53 � 6.22* 0.002 55.40 � 5.52* 0.009

AUC_90 (a.u.) 5.97 � 0.52 4.68 � 0.70* 1 � 10�5 4.68 � 0.74* 1 � 10�4

*Observed difference is statistically significant after Bonferroni correction.

FIGURE 7. IA, IIA, IIIA, VIA, are thecomparisons of box plots for semi-quantitative parameters (MER,Tmax, washout-score and AUC_90)from artery time-SI curves betweenGd-BOPTA, Gd-DTPA, and Gd-DTPA-BMA injections. The plotsshowed Gd-BOPTA had higher MER,AUC_90 and washout-score, plus ashorter Tmax. IM, IIM, IIIM, VIM,are the comparisons of box plots forsemi-quantitative parameters frommuscle time-SI curves between Gd-BOPTA, Gd-DTPA and Gd-DTPA-BMA injections. The plots showedGd-BOPTA also had higher MER,AUC_90 and washout-score, plus ashorter Tmax.




cokinetic parameters when Gd-BOPTA or other protein biddingagents are used.

Because of the small body weights of our subjects, themean contrast agent injection volume of all subjects was 2.0 mL(range, 1.5–2.5 mL). We standardized the injection rate to 0.06mL/s followed by saline flashing rate of 0.3 mL/s. This allowedus to have 3 measurement points within a 30-second window ofthe injection duration using temporal resolution of 9.6 seconds inDCE-MRI scan.

In conclusion, we have demonstrated the feasibility ofusing a weak protein interacting Gd-based contrast agent (Gd-BOPTA) and 2 standard Gd chelates (Gd-DTPA and Gd-DTPA-BMA) for semi-quantitative and quantitative pharmacokineticassessments of DCE-MRI data at 7T. As expected, Gd-BOPTAwas demonstrated to have stronger signal enhancement capabilitythan standard Gd chelates, which is reflected in 4 semi-quantita-tive measurements (MER, washout_score, AUC_90, and Tmax)and some pharmacokinetic parameters (Ktrans, ve, Amp, and kel).The pharmacokinetic parameter kep was demonstrated to beconstantly robust against contrast agent selection in variousmodels, which might be an advantage in multisite clinical trialsand long-term clinical studies. Moreover, our data reveals theequivalency between exchange rate constants derived from Toftsmodel and refined Brix model. This finding helps to improve thegenerality of pharmacokinetic modeling as a robust quantitativeassessment method for DCE-MRI data.

FIGURE 8. A, B, are the comparisons of box plots for Tofts model’s pharmacokinetic quantitative parameters (Ktrans, kepT )

from muscle time-SI curves between Gd-BOPTA, Gd-DTPA and Gd-DTPA-BMA. They showed Gd-BOPTA had significanthigher Ktrans values than other 2 Gd contrast agents in muscle curves. kep

T values showed no significant correlation to thedifferent types of contrast agents. C–F, are the comparisons of box plots for Brix and refined Brix model pharmacokineticquantitative parameters (Amp, kep

B , kelB, kep

RB) from muscle time-SI curves between Gd-BOPTA, Gd-DTPA and Gd-DTPA-BMA. The plots showed Gd-BOPTA had significant higher Amp values, significant higher kel

B values than other 2 Gd con-trast agents in muscle curves. kep

B and kepRB values showed no significant correlation to the different types of contrast

agents.

FIGURE 9. The comparisons of clustered box plots for all 3kep�s from muscle time-SI curves between Gd-BOPTA, Gd-DTPA and Gd-DTPA-BMA. The plots showed that AIF de-composed exchange rate constant kep

RB from the refined Brixmodel reveled equivalency results compared with kep

T fromTofts model.




APPENDIX

Derivation of 3 Pharmacokinetic ModelsThis appendix is provided to theoretically review the

development of the 3 different pharmacokinetic models whichwere used in this article, including Tofts model, Brix model andrefined Brix model. The assumption about the linear relationshipbetween the signal enhancement from spoiled fast field echosequence and the contrast agent concentration is described. Thepharmacokinetic parameter estimation based on this assumptionis discussed.

Contrast Concentration from Signal Intensity CurveIn the spoiled gradient echo sequence, the signal intensity is

given by47:

St � PD sin � exp�TE/T2*

1 � exp�TR/T1

1 � cos � exp�TR/T1

(A-1)

Where P is the proton density and D is the scanner gain, TEis the echo time, TR is the repetition time, � is the flip angle, T1

and T2* are the longitudinal and transverse relaxation times

respectively.The effect of Gd-based contrast agent is to change T1 and T2

*.In general, a linear relationship is assumed that:

1

T1

�1

T10

� R1Ct (A-2)

1

T2*

�1

T20*

� R2Ct (A-3)

Where T10 and T20 are the longitudinal and transverse relax-ation times before contrast agent administration, R1 and R2 are thecontrast agent relaxivity for the specific contrast agent and Ct is thecontrast agent concentration at time t. Note: R1 is treated as a knownconstant, and it is different for different contrast agents and differentfield strengths.

When sufficiently short TR and TE are used (in this applica-tion, TR: 7.8 ms, TE: 4 ms), we assume that the conditions includingTR/T10 �� 1, TE � R2 � Ct �� 1 , and TR � R2 � Ct �� 1 are fulfilledfor the sequence parameters taken into the measurement, and a linearapproximation was obtained as follows by a first-order Taylorexpansion. Hence, the reduced signal intensity equation is presentedby equation (A4), where an approximated linear relationship be-tween the contrast agent concentration and the enhanced MR signalis obtained.

Et �St

S0

� 1 � T10 � R1 � Ct (A-4)

Tofts model assumes a known AIF Cp ), either simulated fromknown blood sample data or measured from artery ROI in DCE-MRI data sets. Then we can eliminate Eq. (2) and express Eq. (3) asthe general Eq. (A5) (by Laplace transform)48:

Ce � Ktrans_T 0

t

Cp�exp� � kepT t � ��d� (A-5)

Where KepT is the Kep from Tofts model, and Ktrans_T is the

volume transfer constant. Note: KepT . Ktrans_T/ve � ve is the fractional

volume of EES which is equal to Ve/Vp. In these equations, the intra-

TABLE 3. Summary of All Quantitative Parameters in Muscle (Bonferroni Method, With an Overall Significance Level of 0.1)

Gd-BOPTAMean � SD

Gd-DTPA Gd-DTPA-BMA


MER (%) 0.51 � 0.05 0.38 � 0.06* 5 � 10�6 0.36 � 0.08* 1 � 10�6

Tmax (min) 3.76 � 1.33 4.60 � 0.89 0.014 4.78 � 1.45* 4 � 10�3

Washout_score (%) 20.10 � 10.76 9.13 � 6.06* 0.002 9.14 � 5.78* 0.002

AUC_90 (a.u.) 0.42 � 0.17 0.30 � 0.09* 0.003 0.29 � 0.13* 0.007


TABLE 4. Summary of Pharmacokinetic Parameters in Muscle From 3 Different DCE-MRI Models (Bonferroni Method, Withan Overall Significance Level of 0.05)

Gd-BOPTAMean �

SD

Gd-DTPA Gd-DTPA-BMA


Ktrans (min�1) 0.12 � 0.04 0.08 � 0.02* 0.002 0.07 � 0.03* 6 � 10�4

kepT (min�1) 0.62 � 0.19 0.55 � 0.10 0.21 0.49 � 0.14 0.08

Amp (a.u.) 0.56 � 0.10 0.40 � 0.07* 1 � 10�5 0.37 � 0.09* 1 � 10�7

kepB (min�1) 1.11 � 0.25 1.06 � 0.24 0.37 1.20 � 0.25 0.26

kelB (min�1) 0.06 � 0.02 0.02 � 0.01* 0.001 0.02 � 0.03* 2 � 10�5

kepRB (min1�) 0.58 � 0.17 0.52 � 0.11 0.20 0.46 � 0.15 0.02





vascular contribution to measured tracer concentration isignored.

In this study, AIF was measured directly from an arteryROI in DCE-MRI data and the signal enhancement curves from

both muscle and artery were used to estimate Ktrans_T and KepT ,

via introducing equation (A4) into (A5) as follows:

Ee � Ktrans 0

t

Ep�exp��kepT t � ��d� (A-6)

Where Ktrans � Ktrans_TT10

M R1M

T10P R1

P, T10

M , and T10P are the relaxation

times of muscle and plasma before contrast agent administration. In1.5T, T10

P is higher than T10M by over 30%. In 7T, relaxation time is

longer than at lower field strength, however, T10M and T10

P might be notvary greater than at lower field strength.

R1M and R1

P are the contrast agent relaxivities in muscle andplasma, respectively. Normally, R1 is used from a constant literaturevalue which will vary for different contrast agent types and willslightly decrease at higher field strength, particularly for proteinbinding types. R1 is also slightly different within different tissues,which muscle R1

M is usually higher than blood R1P about 4% in

Gd-DTPA.49

From above equations and discussions, we can see Ktrans

calculated from signal intensity enhancement generates error mainlydependent on relaxation times(T10 ) which have no relationship withcontrast agent type. Kep

T is independent of R1 and T10.Brix model assumes a mono-exponential decay in Cp, which

is seldom supported by observations. Considering equation (2)which replaces the known AIF in Tofts model, the 2 ordinarydifferential equations (ODE) (Eqs. �2� and �3�) can be successfullydecoupled without dependency on Cp after assuming that the ex-change terms are negligible compared with infusion and eliminationterms in the plasma equation.

In this case, the amplitude Amp, the exchange rate constantbetween blood plasma and EES (kep

B ,�min�1�) and the eliminationrate (kel

B,�min�1�) are calculated from equation (A7):

Stt

St0

� 1 �Amp

� {a�exp�kelB t��1�exp��kel

B t��b�exp�kepB t��1�

exp��kepB t�} (A-7)

In which a �kep

B

kelB(kep

B � kelB)

, b �1

kepB � kel

B, t� � t for t � �, and

t� � � for t �.In Brix model, Amp partially indicates R1 value and kel

B

indicates the distribution of the contrast agent. Both are varied bydifferent types of contrast agents. kep

B is independent of the contrastagent, however it does not hold the true equivalency because theinitial assumption excludes Kpe

B term. Contrast agent infusion injec-tion is no longer considering suitable rather considering bolusinjection only.

Refined Brix model uses an AIF decomposed method, whichassumes a bi-exponential decay AIF following bolus injection ofcontrast agent. The solution of this equation is the following:

SAIFt

SAIF0

� 1 �B

1 � 2{1 � kepRB

�1�exp�1t��1�exp1t

�2 � kep

RB

�2�exp�2t��1�exp2t}

Stt

St0

� 1 �A

11 � 2{2�exp�1t��1�exp1t

� 1�exp�2t��1�exp2t} (A-8)

1 � � 2kepRBkel

RB/�kepRB � kpe

RB�2 � � �kep

RB � kpeRB� (A-9)

In which A and B are amplitude parameters for the AIF andtissue ROIs respectively. The refined exchange rate constants (Kep

RB,Kpe

RB) from the above method were also calculated.Because of the assumed bi-exponential AIF and bi-direction

exchange rates in the two-compartment model of the refined Brixmodel, an equivalency might be established between Kep

T from Toftsmodel and from Kep

RB Brix model, which both are independent fromdifferent types of contrast agents.

ACKNOWLEDGMENTSThe authors thank Dr. Petra Schmalbrock, Dr. Seongjin Choi

for helpful discussions and thank Kathleen Lindsey, Jeanne Greenfor the subjects’ preparation.

REFERENCES1. Burns M. The U.S. market for medical imaging contrast media. Las Vegas,

NV: Bio-tech Systems, Inc; 2004.

2. Lin SP, Brown JJ. MR contrast agents: physical and pharmacologic basics. JMagn Reson Imaging. 2007;25:884–899.

3. Bellin MF, van der Molen AJ. Extracellular gadolinium-based contrast media:an overview. Eur J Radiol. 2008;66:160–167.

4. Runge VM. Contrast media research. Invest Radiol. 1999;34:785–790.

5. Bellin MF. MR contrast agents, the old and the new. Eur J Radiol. 2006;60:314–323.

6. Knopp MV, Weiss E, Sinn HP, et al. Pathophysiologic basis of contrastenhancement in breast tumors. J Magn Reson Imaging. 1999;10:260–266.

7. Knopp MV, Tengg-Kobligk H, Choyke PL. Functional magnetic resonanceimaging in oncology for diagnosis and therapy monitoring. Mol Cancer Ther.2003;2:419–426.

8. Essig M. MR imaging of CNS tumors: are all contrast agents created thesame? Neuroradiology. 2006;48(suppl 1):3–8.

9. Runge VM, Biswas J, Wintersperger BJ, et al. The efficacy of gadobenatedimeglumine (Gd-BOPTA) at 3 Tesla in brain magnetic resonance imaging—comparison to 1.5 tesla and a standard gadolinium chelate using a rat braintumor model. Invest Radiol. 2006;41:244–248.

10. Knopp MV, Runge VM, Essig M, et al. Primary and secondary brain tumorsat MR imaging: bicentric intraindividual crossover comparison of gadobenatedimeglumine and gadopentetate dimeglumine. Radiology. 2004;230:55–64.

11. Spinazzi A, Lorusso V, Pirovano G, et al. Safety, tolerance, biodistribution,and MR imaging enhancement of the liver with gadobenate dimeglumine:results of clinical pharmacologic and pilot imaging studies in nonpatient andpatient volunteers. Acad Radiol. 1999;6:282–291.

12. Kirchin MA, Pirovano GP, Spinazzi A. Gadobenate dimeglumine (Gd-BOPTA). An overview. Invest Radiol. 1998;33:798–809.

13. Colosimo C, Demaerel P, Tortori-Donati P, et al. Comparison of gadobenatedimeglumine (Gd-BOPTA) with gadopentetate dimeglumine (Gd-DTPA) forenhanced MR imaging of brain and spine tumours in children. Pediatr Radiol.2005;35:501–510.




14. Runge VM, Parker JR, Donovan M. Double-blind, efficacy evaluation ofgadobenate dimeglumine, a gadolinium chelate with enhanced relaxivity, inmalignant lesions of the brain. Invest Radiol. 2002;37:269–280.

15. Knopp MV, Schoenberg SO, Rehm C, et al. Assessment of gadobenatedimeglumine for magnetic resonance angiography: phase I studies. InvestRadiol. 2002;37:706–715.

16. Schneider G, Maas R, Schultze KL, et al. Low-dose gadobenate dimeglumineversus standard dose gadopentetate dimeglumine for contrast-enhanced mag-netic resonance imaging of the liver: an intra-individual crossover compari-son. Invest Radiol. 2003;38:85–94.

17. Cavagna FM, Maggioni F, Castelli PM, et al. Gadolinium chelates with weakbinding to serum proteins—a new class of high-efficiency, general purposecontrast agents for magnetic resonance imaging. Invest Radiol. 1997;32:780–796.

18. de Haen C, La Ferla R, Maggioni F. Gadobenate dimeglumine 0.5 M solutionfor injection (Multihance) as contrast agent for magnetic resonance imagingof the liver: mechanistic studies in animals. J Comput Assist Tomogr 1999;23(suppl 1):S169–S179.

19. Heywang SH, Hahn D, Schmidt H, et al. MR imaging of the breast usinggadolinium-DTPA. J Comput Assist Tomogr. 1986;10:199–204.

20. Tofts PS, Kermode AG. Measurement of the blood-brain barrier permeabilityand leakage space using dynamic MR imaging. 1. Fundamental concepts.Magn Reson Med. 1991;17:357–367.

21. Brix G, Semmler W, Port R, et al. Pharmacokinetic parameters in CNSGd-DTPA enhanced MR imaging. J Comput Assist Tomogr. 1991;15:621–628.

22. Sardanelli F, Fausto A, Esseridou A, et al. Gadobenate dimeglumine as acontrast agent for dynamic breast magnetic resonance imaging: effect ofhigher initial enhancement thresholds on diagnostic performance. InvestRadiol. 2008;43:236–242.

23. van der Molen AJ, Bellin MF. Extracellular gadolinium-based contrast media:differences in diagnostic efficacy. Eur J Radiol. 2008;66:168–174.

24. Hylton N. Dynamic contrast-enhanced magnetic resonance imaging as animaging biomarker. J Clin Oncol. 2006;24:3293–3298.

25. Yuh WT, Mayr NA, Jarjoura D, et al. Predicting control of primary tumor andsurvival by DCE MRI during early therapy in cervical cancer. Invest Radiol.2009;44:343–350.

26. Yankeelov TE, Luci JJ, Lepage M, et al. Quantitative pharmacokineticanalysis of DCE-MRI data without an arterial input function: a referenceregion model. Magn Reson Imaging. 2005;23:519–529.

27. Lowry M, Zelhof B, Liney GP, et al. Analysis of prostate DCE-MRI:comparison of fast exchange limit and fast exchange regimen pharmacoki-netic models in the discrimination of malignant from normal tissue. InvestRadiol. 2009;44:577–584.

28. Turetschek K, Preda A, Novikov V, et al. Tumor microvascular changes inantiangiogenic treatment: assessment by magnetic resonance contrast mediaof different molecular weights. J Magn Reson Imaging. 2004;20:138–144.

29. de Lussanet QG, Langereis S, Beets-Tan RG, et al. Dynamic contrast-enhanced MR imaging kinetic parameters and molecular weight of dendriticcontrast agents in tumor angiogenesis in mice. Radiology. 2005;235:65–72.

30. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. JMagn Reson Imaging. 1997;7:91–101.

31. Hoffmann U, Brix G, Knopp MV, et al. Pharmacokinetic mapping of thebreast—a new method for dynamic MR mammography. Magn Reson Med.1995;33:506–514.

32. Yang XY, Liang JC, Heverhagen JT, et al. Improving the pharmacokineticparameter measurement in dynamic contrast-enhanced MRI by use of thearterial input function: theory and clinical application. Magn Reson Med.2008;59:1448–1456.

33. Nakada T. Clinical application of high and ultra high-field MRI. Brain Dev.2007;29:325–335.

34. Brantzawadzki M, Norman D, Newton TH, et al. Magnetic-resonance of thebrain—the optimal screening technique. Radiology. 1984;152:71–77.

35. Schneider G, Kirchin MA, Pirovano G, et al. Gadobenate dimeglumine-enhanced magnetic resonance imaging of intracranial metastases: effect ofdose on lesion detection and delineation. J Magn Reson Imaging. 2001;14:525–539.

36. Gasperini C, Paolillo A, Rovaris M, et al. A comparison of the sensitivity ofMRI after double- and triple-dose Gd-DTPA for detecting enhancing lesionsin multiple sclerosis. Magn Reson Imaging. 2000;18:761–763.

37. Erickson BJ, Campeau NG, Schreiner SA, et al. Triple-dose contrast/magne-tization transfer suppressed imaging of “non-enhancing” brain gliomas.J Neurooncol. 2002;60:25–29.

38. Henkelman RM. Measurement of signal intensities in the presence of noise inMR images. Med Phys. 1985;12:232–233.

39. McAuliffe MJ, Lalonde F, McGarry DP, et al. Medical image processing,analysis & visualization in clinical research. In: IEEE Symposium on Com-puter-Based Medical Systems (CBMS 2001). Bethesda, MD: 2008:381–386.

40. Evelhoch JL. Key factors in the acquisition of contrast kinetic data foroncology. J Magn Reson Imaging. 1999;10:254–259.

41. Sherif H, Mahfouz AE, Oellinger H, et al. Peripheral washout sign oncontrast-enhanced MR images of the breast. Radiology. 1997;205:209–213.

42. Pintaske J, Martirosian P, Graf H, et al. Relaxivity of Gadopentetate Dime-glumine (Magnevist), Gadobutrol (Gadovist), and Gadobenate Dimeglumine(Multihance) in human blood plasma at 0.2, 1.5, and 3 Tesla. Invest Radiol.2006;41:213–221.

43. Spinazzi A, Lorusso V, Pirovano G, et al. Multihance clinical pharmacology:biodistribution and MR enhancement the liver. Acad Radiol. 1998;5:S86–S89.

44. Weinmann HJ, Ebert W, Misselwitz B, et al. Tissue-specific MR contrastagents. Eur J Radiol. 2003;46:33–44.

45. Planchamp C, Gex-Fabry M, Dornier C, et al. Gd-BOPTA transport into rathepatocytes: pharmacokinetic analysis of dynamic magnetic resonance im-ages using a hollow-fiber bioreactor. Invest Radiol. 2004;39:506–515.

46. Port M, Corot C, Violas X, et al. How to compare the efficiency ofalbumin-bound and nonalbumin-bound contrast agents in vivo: the concept ofdynamic relaxivity. Invest Radiol. 2005;40:565–573.

47. Haacke EM, Tkach JA. Fast MR-imaging—techniques and clinical-applica-tions. Am J Roentgenol. 1990;155:951–964.

48. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters fromdynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: stan-dardized quantities and symbols. J Magn Reson Imaging. 1999;10:223–232.

49. Donahue KM, Burstein D, Manning WJ, et al. Studies of Gd-Dtpa relaxivityand proton-exchange rates in tissue. Magn Reson Med. 1994;32:66–76.




Intraindividual In Vivo Comparison of Gadolinium Contrast Agents for Pharmacokinetic Analysis Using...

Documents

Transcript of Intraindividual In Vivo Comparison of Gadolinium Contrast Agents for Pharmacokinetic Analysis Using...