three-dimensional high-frequency ultrasound imaging for … · three-dimensional high-frequency...

Three-dimensional High-Frequency Ultrasound Imaging

for Longitudinal Evaluation of Liver Metastases

in Preclinical Models

Kevin C. Graham,1,4Lauren A. Wirtzfeld,

2,5Lisa T. MacKenzie,

1Carl O. Postenka,

4

Alan C. Groom,1Ian C. MacDonald,

1Aaron Fenster,

1,2,5James C. Lacefield,

1,2,3,5

and Ann F. Chambers1,4

Departments of 1Medical Biophysics, 2Biomedical Engineering Graduate Program, and 3Electrical and Computer Engineering,The University of Western Ontario; 4London Regional Cancer Program; and 5Robarts Research Institute, London, Ontario, Canada

Abstract

Liver metastasis is a clinically significant contributor to themortality associated with melanoma, colon, and breastcancer. Preclinical mouse models are essential to the studyof liver metastasis, yet their utility has been limited by theinability to study this dynamic process in a noninvasive andlongitudinal manner. This study shows that three-dimensionalhigh-frequency ultrasound can be used to noninvasively trackthe growth of liver metastases and evaluate potentialchemotherapeutics in experimental liver metastasis models.Liver metastases produced by mesenteric vein injection ofB16F1 (murine melanoma), PAP2 (murine H-ras–transformedfibroblast), HT-29 (human colon carcinoma), and MDA-MB-435/HAL (human breast carcinoma) cells were identified andtracked longitudinally. Tumor size and location were verifiedby histologic evaluation. Tumor volumes were calculatedfrom the three-dimensional volumetric data, with individualliver metastases showing exponential growth. The impor-tance of volumetric imaging to reduce uncertainty in tumorvolume measurement was shown by comparing three-dimensional segmented volumes with volumes estimatedfrom diameter measurements and the assumption of anellipsoid shape. The utility of high-frequency ultrasoundimaging in the evaluation of therapeutic interventions wasestablished with a doxorubicin treatment trial. These resultsshow that three-dimensional high-frequency ultrasoundimaging may be particularly well suited for the quantitativeassessment of metastatic progression and the evaluation ofchemotherapeutics in preclinical liver metastasis models.(Cancer Res 2005; 65(12): 5231-7)

Introduction

Metastasis, the dissemination and growth of cancer cells in asecondary organ, is the leading cause of cancer mortality. Theliver is a frequent metastatic site for melanoma, colon, and breastcancer and therefore an important area of metastasis research.Preclinical animal models, such as the mouse, are essential to thestudy of liver metastasis, yet their utility has been limited bydifficulty in tracking the progression of metastases through time.Noninvasive longitudinal imaging would decrease experimental

variability, provide a more accurate assessment of metastaticprogression and the efficacy of therapeutic interventions, andallow the study of dynamic processes such as tumor vasculari-zation and dormancy.A multitude of preclinical imaging modalities are under

development, including magnetic resonance imaging (MRI), X-raycomputed tomography (CT), positron emission tomography (PET),and fluorescent and bioluminescent imaging, yet no singlemodality should be considered a comprehensive solution forcancer microimaging applications. Each modality possesses aunique combination of strengths and weaknesses that affect theirselection for use in a particular study. In general, desirablecharacteristics in a noninvasive imaging modality would be highresolution to allow detection of minimal disease, cost-effectivenessand rapid image acquisition to facilitate throughput, inherentcontrast between the liver parenchyma and tumor to avoidgenetically encoded or endogenously given contrast agents, andapplicability to a range of liver metastasis models. MRI offers high-resolution imaging yet may be time consuming and relativelyexpensive to purchase and operate (1). X-ray CT also offers highresolution, but poor soft tissue contrast necessitates the use ofradiopaque contrast agents and radiation dosage may limitlongitudinal imaging (1). The resolution of PET does not matchthat of MRI or CT, and the requirement for production andcontainment of radionuclides can make costs prohibitive (1).Fluorescent and bioluminescent imaging offer a relatively cost-effective way to study liver metastases but suffer from poorresolution and the requirement to transfect endogenous reportergenes into the cell line of interest (1, 2). The expression of foreignreporter proteins may lead to increased immunogenicity and thusmust be carefully examined for their effect on the metastatic modelbeing studied (3–5).Ultrasound is an attractive option for preclinical imaging due to

the cost and time efficiencies of this modality. Previous studiesusing high-frequency ultrasound imaging of murine cancer modelsshowed the feasibility of this technique to track s.c. tumorprogression (6). That study concluded that further application ofultrasound imaging would require a fast method for generatingthree-dimensional images. A new high-frequency scanner thatemploys three-dimensional image acquisition methods and recon-struction software developed in our laboratory has addressed thislimitation (7). These developments raised the possibility of usinghigh-frequency ultrasound in the evaluation of clinically relevantmetastatic models, which are difficult to study in a noninvasivefashion.We report here the application of a high-frequency (40 MHz)

ultrasound system with three-dimensional imaging capabilities to

Requests for reprints: Ann F. Chambers, London Regional Cancer Program, 790Commissioners Road East, London, Ontario, Canada, N6A 4L6. Phone: 519-685-8652;Fax: 519-685-8646; E-mail: [email protected].

I2005 American Association for Cancer Research.

www.aacrjournals.org 5231 Cancer Res 2005; 65: (12). June 15, 2005

Research Article

Research. on December 26, 2020. © 2005 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

the study of murine liver metastasis. We showed that the resolutionof high-frequency ultrasound allowed detection of liver metastasesat a minimum size that compared favorably with that of MRI, CT,or optical methods (8–11). The applicability of this technique wasshown by identifying and tracking liver metastases from four tumorcell lines of different tumor origins. The importance of three-dimensional volumetric imaging to reduce uncertainty in volumedetermination was established by comparison of three-dimensionalsegmented volumes with the commonly assumed ellipsoidalvolume calculated from diameter measurements. The utility ofthree-dimensional high-frequency ultrasound in the evaluation ofchemotherapeutics was shown in a preclinical trial with doxoru-bicin. These results show that the cost and time efficiencies oftraditional ultrasound coupled with the three-dimensional capa-bilities and high resolution of this high-frequency ultrasoundsystem make this modality particularly well suited to the study ofliver metastasis in a wide range of preclinical models.

Materials and Methods

Cell culture and experimental metastasis models. B16F1 (12) and

HT-29 (Cat# HTB-38, American Type Culture Collection, Manassas, VA) cells

were maintained in aMEM with L-glutamine, ribonucleosides, and

deoxyribonucleosides (Invitrogen, Carlsbad, CA) supplemented with 10%fetal bovine serum (Sigma, Mississauga, Ontario, Canada). PAP2 cells (13, 14)

were maintained in DMEM (Invitrogen) supplemented with 10% FCS

(Invitrogen). MDA-MB-435/HAL cells were maintained in EMEM (Invitro-

gen) supplemented with 2 mmol/L L-glutamine, 100 Amol/L nonessentialamino acids, 25 mmol/L HEPES buffer, 1 mmol/L sodium pyruvate

(Invitrogen), and 1� MEM vitamin solution (Sigma). The MDA-MB-435/

HAL line was derived from the MDA-MB-435 cell line by an in vivo selectionprocedure for increased metastatic potential (15). All animals were cared for

in accordance to the Canadian Council on Animal Care, under a protocol

approved by the University of Western Ontario Council on Animal Care.

For experimental metastasis assays, mice were anesthetized with an i.p.injection of xylazine/ketamine (2.6 mg ketamine and 0.13 mg xylazine per

20 g body mass). As described previously, anesthetized mice received

mesenteric injections of 0.1 mL of cells suspended in their respective growth

media to target the liver (16). For B16F1 cells, a suspension of 3 � 105 cellswas injected into C57BL/6 mice (Harlan, Indianapolis, IN). For HT-29 cells, a

suspension of 2 � 106 cells was injected into NIH III mice (Charles River,

Wilmington, MA). For PAP2 cells, a suspension of 2 � 105 cells was injectedinto SCID mice (Charles River). For MDA-MB-435/HAL cells, a suspension of

3 � 105 cells was injected into SCID mice (Charles River). All mice were

female and 7 to 11 weeks old at the time of cell injection.

Ultrasound imaging. For ultrasound imaging, the Vevo 660 high-frequency ultrasound system (VisualSonics, Inc., Toronto, Ontario, Canada)

was used. The Vevo 660 is the second generation of a system described

previously (17). The Vevo 660 ultrasound probe has a 40-MHz center

frequency with a 6-mm focal depth. The spatial resolution at the focus is40 � 80 � 80 Am3. Before the first imaging session, the mouse’s abdomen

was depilated with commercial hair removal cream. During imaging, the

mouse was kept under anesthesia with 1.5% isoflurane in oxygen andrestrained on a heated stage. During imaging with the immunodeficient

NIH III and SCID mice, the animals were handled and imaged in a HEPA-

filtered workstation (Microzone Corp., Ottawa, Ontario, Canada). Ultra-

sound is strongly reflected by the ribcage, which hinders imaging of anytissue located beneath the ribs, such as the lungs and a portion of the liver.

Thus, the volume of liver tissue accessible for ultrasound imaging may vary

between animals and between imaging sessions for the same animals. In

general, we found that a significant volume of the left lateral, left medial,and right medial liver lobes were routinely accessible for imaging. During

imaging, two-dimensional images were acquired in the sagittal plane after

ultrasound contact gel was applied to the abdomen. For three-dimensional

imaging, parallel two-dimensional images were acquired by stepping the

transducer in 30-Am intervals in the out-of-plane dimension. Using softwaredeveloped in our laboratory, two-dimensional images were interpolated and

reconstructed online to create a three-dimensional volumetric image (7).

The system can acquire and produce a typical three-dimensional image in

<20 seconds. The three-dimensional reconstruction software is availablethrough VisualSonics, or through the authors for research purposes.

Volume measurements. To determine tumor volume, the boundaries of

a metastasis were outlined within parallel planes separated by 50 Am in the

volumetric image. The total metastasis volume was calculated by summingthe outlined areas and multiplying by the interslice distance (18).

Segmented volumes were compared with ellipsoid volumes estimated using

the formula V = pabc / 6. The measurements for diameters a, b , and c were

obtained from the three-dimensional volumetric images. The sagittal planeshowing the greatest tumor diameter was selected, and the greatest

diameter a measured. The diameter b , perpendicular to a , was then

measured. The volume was then rotated and the transverse plane showingthe largest tumor diameter was selected. The diameter c , perpendicular to

both a and b , was then measured. To determine the % difference between

the ellipsoid and three-dimensional segmented volumes the following

formula was used: 100% � (ellipsoid volume � segmented volume)/{(ellipsoid volume + segmented volume) / 2}.

Longitudinal growth measurements. For longitudinal imaging, the

initial imaging time point was based on previous indications of when

micrometastases could first be detected by ultrasound. Individual livermetastases were identified and a three-dimensional image recorded.

Individual liver metastases were identified on successive imaging dates by

their particular liver lobe location, tumor shape, and proximity to landmarkstructures such as major blood vessels or the liver edges. Landmarks were

all internal to the liver, because the liver lobes move in relation to any

external landmarks such as the ribs. Animals were sacrificed due to

escalating tumor burden, as assessed by ultrasound imaging, or when atleast four imaging time points had been acquired to construct a growth

curve. Approximately 5 minutes was required to locate, identify and image

the liver metastases of each mouse. If the time spent on setup, animal

handling, anesthesia, and recovery is included, the average duration of animaging session was 15 minutes per mouse.

Treatment protocols. The B16F1 liver metastasis model was used to

assess the ability of high-frequency ultrasound to evaluate cytotoxicchemotherapeutic agents. At day 7 after cell injection, the first treatment

with doxorubicin was given. Doxorubicin (Pharmacia, Mississauga, Ontario,

Canada) was given at a previously described treatment schedule (1 mg/kg,

0.1 mL, i.p.) every second day until day 17 after cell injection, for a total ofsix treatments (19). Control animals received saline control injections

(0.1 mL, i.p.). Ultrasound imaging was done from day 8 after cell injection,

the earliest time B16F1 liver metastases could be detected in the images,

until the end of the experiment.Histology. At sacrifice, the mouse liver was excised and fixed in 10%

neutral buffered formalin. Visual inspection validated the tumor size and

location depicted by ultrasound imaging. For histologic confirmation,

formalin-fixed, paraffin-embedded livers were sectioned (4 Am) and stainedwith H&E.

Results

Identification of murine liver metastases using high-frequency ultrasound. To validate ultrasound imaging for thedetection of murine liver metastases, mice were noninvasivelyimaged; once suspected metastases were detected by ultrasound,the animal was sacrificed. Gross pathology and histologic sectionsverified the presence of a tumor, its size, and location. Ultrasoundreliably detected murine liver metastases from the four tumor celllines tested, B16F1, HT-29, MDA-MB-435/HAL, and PAP2, withexcellent agreement among ultrasound images, gross patho-logy, and histologic sections (Fig. 1A-D). Ultrasound imagingproved highly sensitive to small metastases with a minimumdetection size (maximum diameter ! segmented volume) of

Cancer Research

Cancer Res 2005; 65: (12). June 15, 2005 5232 www.aacrjournals.org



0.22 mm ! 0.01mm3, 0.47 mm ! 0.03 mm3, 0.66 mm ! 0.08 mm3,and 0.78 mm ! 0.17 mm3 for B16F1, HT-29, MDA-MB-435/HAL,and PAP2 tumors, respectively. As a point of reference, a volume of0.01 mm3 would be produced by f700 cells, based on theassumption of a spherical cell volume and a cell diameter of 15 Am.High-frequency ultrasound can identify areas of liquefactive

necrosis within metastases. During imaging of B16F1 livermetastases, it was noted that although the metastases were alwaysclearly delineated from the surrounding parenchyma, the metas-tases showed a large amount of heterogeneity in their ultrasoundbackscatter. In a number of metastases, distinct anechoic regions(no texture and dark in appearance) were evident. Histologicexamination of these anechoic regions revealed that they areregions of liquefactive necrosis (Fig. 1E). Liquefactive necrosis isexpected to be anechoic because the breakdown of necrotic cellseliminates the majority of ultrasound scattering structures fromthat region of the tumor.Tracking the growth of individual liver metastases by

noninvasive ultrasound imaging. To show the use of ultrasoundimaging in the longitudinal study of liver metastases, mice werenoninvasively imaged at 2- to 3-day intervals. At sacrifice, grosspathology and histologic sections verified tumor sizes andlocations depicted during ultrasound imaging. The B16F1 metas-tases developed rapidly, forming detectable metastases as early as10 days after cell injection in this experiment (Fig. 2A). The B16F1metastases showed exponential growth with an average volumedoubling time of 1.2F 0.2 (meanF SD) days. The mean correlationcoefficient for fitting an exponential curve was 0.966 F 0.047. TheHT-29 and MDA-MB-435/HAL liver metastases were much slowerto develop, forming detectable metastases at a minimum of 33 daysafter cell injection (Fig. 2B). The HT-29 and MDA-MB-435/HALmetastases also showed exponential growth with doubling timesranging from 3.7 to 4.8 days for the HT-29 metastases and 5.7 to10.4 days for MDA-MB-435/HAL metastases. The correlationcoefficients ranged between 0.972 and 0.993 for HT-29 and between0.635 and 0.889 for MDA-MB-435/HAL. The metastasis HT-29-4 wasnot included in the range of doubling times because its volume didnot increase over the 10-day interval that it was imaged.Representative two-dimensional images from the longitudinalimaging of an individual B16F1 metastasis, B16F1-E, are shown(Fig. 2C). Individual PAP2 liver metastases could not be evaluatedfor longitudinal growth because in this highly aggressive modelnumerous metastases form and quickly fuse. In such cases,ultrasound could be used to monitor increasing tumor burden,instead of the growth of individual metastases, as an indicator oftumor progression.Two-dimensional measurement provides frequent overesti-

mation or underestimation of tumor volume as compared withthree-dimensional measurement. Tumors are often assumed tohave an ellipsoid shape, which allows a volume to be calculatedfrom the maximum widths and length in two-dimensional images.Tumor volumes calculated from this two-dimensional method

Figure 1. Identification of B16F1 (A and E), HT-29 (B ), MDA-MB-435/HAL(C ), and PAP2 (D ) liver metastases by noninvasive ultrasound imaging.Two-dimensional ultrasound images with corresponding histologic sections. Thelocation of each metastasis in the ultrasound image (yellow arrows ) and histologicsection (black arrows ) is indicated. The maximum diameter of the metastasisin the ultrasound image is 1.56 mm (A ), 0.47 mm (B), 0.66 mm (C ), 2.33 mm (D ),and 5.01 mm (E ). Bar on the ultrasound images, 1.00 mm. Bar on the histologicsections, 1.00 mm (A, D , and E) or 0.10 mm (B and C ). E, necrotic areas (N )depicted by ultrasound imaging and confirmed by histology are labeled.

Ultrasound Evaluation of Preclinical Liver Metastases




and from three-dimensional segmentation were compared todetermine if the assumption of an ellipsoid shape was valid forliver metastases. In the liver metastasis models examined here,there were large differences in the measured tumor volumesbetween the three-dimensional and two-dimensional methods. Themean percent difference for B16F1 liver metastases was �8.8 F23.5% (range, �90.1% to 53.2%; Fig. 3A). The negative meanindicates that the ellipsoid volume was on average smaller than thethree-dimensional segmented volume. For the MDA-MB-435/HALmetastases, the mean percent difference was �15.0F 25.3% (range,�45.2% to 23.1%) and for the HT-29 metastases �7.9 F 43.8%(range, �106.5% to 80.5%). Three-dimensional surface rendering

of the liver metastases allowed visualization of the irregular shapesof some tumors (Fig. 3B-C).Three-dimensional ultrasound can be used to monitor

therapeutic response of individual metastases. To assess theability of high-frequency ultrasound to evaluate the efficacy ofcytotoxic chemotherapeutic agents, the B16F1 liver metastasismodel was used. By noninvasively tracking the development ofindividual liver metastases, it was shown that doxorubicinsignificantly increased the doubling time of B16F1 metastasesfrom 1.4 F 0.4 to 1.7 F 0.4 days (t test, P = 0.038; Fig. 4A-B). Theincreased doubling time is apparent in the significant differencebetween the average tumor volumes of doxorubicin-treatedmetastases and the control metastases (Fig. 4C).

Discussion

The importance of preclinical animal models in oncologicalresearch has supported the development of small animal imagingmodalities, including MRI, X-ray CT, PET, and fluorescent and

Figure 2. Tracking the growth of individual liver metastases by noninvasiveultrasound imaging. A, growth curves of B16F1 liver metastases plotted on asemilogarithmic scale. B, growth curves of HT-29 and MDA-MB-435/HALliver metastases plotted on a semilogarithmic scale. C, representativetwo-dimensional ultrasound images of B16F1-E. Sizes of B16F1-E (maximumdiameter ! segmented volume) are 0.50 mm ! 0.06 mm3 (day 10), 1.07 mm! 0.61 mm3 (day 14), and 2.09 mm ! 3.79 mm3 (day 18). Bar on the ultrasoundimages, 1.00 mm.

Figure 3. Discrepancy between tumor volumes obtained from three-dimensional segmentation or diameter measurement with the assumption ofan ellipsoid shape. A, % difference between the ellipsoid volume and thethree-dimensional segmented volume for individual B16F1 metastases is plottedversus the mean volume of those two measurements. Solid bars, mean F 2 SD.B, three-dimensional surface rendering of a B16F1 metastasis in which theellipsoid and three-dimensional segmented volumes are in close agreement(ellipsoid = 5.58 mm3, three-dimensional = 5.79 mm3, % difference = �3.70%).C, three-dimensional surface rendering of a B16F1 metastasis in which theellipsoid and three-dimensional segmented volumes are not in close agreement(ellipsoid = 3.51 mm3, three-dimensional = 5.03 mm3, % difference = �35.60%).

Cancer Research




bioluminescent based systems. Each of these modalities occupiesa niche in noninvasive imaging, based on the unique require-ments and constraints of a particular research study. Importantfactors in choosing the appropriate imaging modality for aparticular study may include the anatomic site being imaged, thedesired resolution and animal throughput, availability of targetedcontrast agents, and cost. In this report, we describe the use ofhigh-frequency ultrasound imaging for the detection andlongitudinal tracking of murine liver metastases. High-frequencyultrasound offers distinct advantages as a cost-effective, rapid,high-resolution, and noninvasive imaging system. The ultrasoundimaging described here was done without exogenous contrastagents or genetic manipulation of the cell lines being studied.This offers significant advantages both in terms of animalthroughput and in the number of animal models able to bestudied.As shown in this report, all four cell lines tested in experimental

liver metastasis models, B16F1, HT-29, MDA-MB-435/HAL, andPAP2, showed inherent ultrasound contrast relative to thesurrounding liver parenchyma. These cell lines are derived fromdifferent primary tumor types, murine melanoma, human coloncarcinoma, human breast carcinoma, and oncogene-transformedmurine fibroblasts, which shows the wide applicability of thistechnique. In all cases, the liver metastases were hypoechoic,appearing darker than the surrounding tissue on ultrasoundimages. Ultrasound imaging was shown highly sensitive withmetastases from all four tumor models detected with maximumdiameters of <0.78 mm. The consistent background image textureof normal liver parenchyma likely contributed to the detection ofsmall tumors; thus, ultrasound seems particularly well suited forimaging liver metastasis models. To determine the absolutedetection limit for any particular cell line, more frequent imagingon a greater number of mice would need to be done.The ability to track the growth of individual liver metastases over

time was shown for B16F1, HT-29, and MDA-MB-435/HAL tumorcell lines. The use of the human cell lines HT-29 and MDA-MB-435/HAL is particularly noteworthy because immunodeficient animals,which must be protected from infection, are used for thesemetastasis models. The ultrasound system was easily adapted tothis requirement by restricting the mice and ultrasound probe to aHEPA-filtered environment. This process would not be possiblewith larger, less portable imaging modalities.The longitudinal imaging trials showed the importance of

noninvasive imaging in allowing analysis on a per metastasisinstead of a per mouse basis. For example, in contrast with theexponential growth seen with other liver metastases in the sameanimal, the metastasis HT-29-4 did not show a significant increasein tumor volume during the 10 days it was imaged. The volume ofthis metastasis was constant at 0.03 mm3. The identification ofmetastases with variable growth patterns would allow furtherinvestigation, such as microdissection and microarray analysis, toelucidate the molecular basis of such variations. The monitoring ofdormant metastases would also permit the study of host-tumorinteractions, such as the role of angiogenesis in the switch from adormant to progressive phenotype (20). The detection sensitivity ofhigh-frequency ultrasound allows the investigation of processes,such as tumor dormancy and angiogenesis, which may occur veryearly in the metastatic process.The longitudinal imaging of B16F1 liver metastases revealed

striking changes in ultrasound image texture during tumordevelopment. The most obvious of these changes was the

Figure 4. A longitudinal ultrasound study shows that doxorubicin (DXR )decreases tumor growth rate and tumor volume in the B16F1 liver metastasismodel. A, the growth of nine B16F1 liver metastases from five mice wastracked longitudinally in the control group. B, growth of 24 B16F1 livermetastases from eight mice was tracked longitudinally in the doxorubicintreatment group. C, mean metastasis volume (points, mean; bars, FSE) foreach imaging time point for the control and doxorubicin treatment groups.*, P < 0.05, significant difference in tumor volume (rank sum test) at theindicated time point.





development of distinct anechoic regions that were shown to beareas of liquefactive necrosis. The ability to detect the formation ofnecrosis may be useful in the assessment of vascular targetingagents and antiangiogenic compounds (21, 22). IncorporatingDoppler blood flow imaging into studies using the high-frequencyultrasound system may further enhance the assessment of tumorvasculature and hemodynamics (23). The cellular or structuralcharacteristics that cause the more subtle changes in ultrasoundbackscatter are currently under investigation in several laboratories(24–26).These studies have focused on imaging the development of

individual metastases because of the unique research opportunitiesthat this approach presents. In time, the models used here willform large coalescing metastases, which can no longer bemonitored for individual growth characteristics. At this stage,ultrasound imaging could be used to measure relative tumorburden between animals. This application will require further studyto determine how tumor burden in acoustically accessible liverareas represent the state of the entire liver.Because s.c. tumor growth is frequently monitored by caliper

measurement and calculation of an ellipsoid volume, we sought todetermine if the approximation of an ellipsoid volume wassufficient for monitoring the growth of liver metastases. It wasshown that tumor volumes calculated from three-dimensional andtwo-dimensional methods yielded vastly different results for manymetastases. For B16F1 metastases, the mean percent volumedifference between the two methods was �8.8 F 23.5%. The largeSD indicates that the two-dimensional method often gives largeoverestimations or underestimations of tumor volume whencompared with the three-dimensional method. Because the truevolume of the metastases could not be determined, it cannot bedefinitely stated that one method is more accurate than the other.However, it is reasonable to suggest that the three-dimensionalmethod is more accurate because there is no assumption of adefined shape, and a three-dimensional image allows the operatorgreater time and control when defining tumor borders. Definitionof tumor borders can be done offline with a three-dimensionalimage, whereas the operator of a two-dimensional system mustidentify maximum tumor diameters during imaging. Furthermore,previous work with a clinical ultrasound system has shown that thethree-dimensional method is more accurate than the two-dimensional method when measuring the volume of regular andirregular shaped phantoms (18, 27). The inaccuracy and variability

brought about by assuming a defined shape could hinder the abilityto track volume changes in slowly growing metastatic models, theability to track subtle responses to therapeutic treatment, and theability to determine if a metastasis is going through a period ofdormancy. The elimination of this uncertainty presents acompelling case for using an imaging modality with three-dimensional imaging capabilities.The use of longitudinal ultrasound imaging in preclinical trials

was shown with the anthracycline chemotherapeutic doxorubicin.Longitudinal assessment of individual liver metastases showed thatdoxorubicin significantly decreased tumor growth rate and tumorvolume in the B16F1 liver metastasis model. Significant differencesin tumor volume were evident at day 12 after cell injection, afteronly three doxorubicin treatments, when the average tumorvolume in the control group was <1.00 mm3 and in the treatedgroup was <0.25 mm3. The ability of high-frequency ultrasound totrack the progression of micrometastases noninvasively allows theevaluation of therapeutic efficacy on sequential stages of tumordevelopment, from early formation to the development of largevascularized metastases, in a single experiment.In summary, this report is the first to describe the use of three-

dimensional high-frequency (40 MHz) ultrasound imaging in thenoninvasive detection and longitudinal evaluation of murine livermetastases. This development is significant in that ultrasoundoffers rapid, cost-effective, high-resolution imaging that can beapplied to a wide range of liver metastasis models without therequirement for contrast agents. Compared with traditional histo-logic methods, ultrasound imaging may provide a more accurateassessment of tumor progression and chemotherapeutic response,which will open new avenues of investigation into dynamicprocesses such as tumor vascularization and tumor dormancy.

Acknowledgments

Received 2/9/2005; revised 3/16/2005; accepted 4/5/2005.Grant support: Canadian Institutes of Health Research, Natural Sciences and

Engineering Research Council of Canada, Canada Foundation for Innovation, OntarioInnovation Trust, Ontario Consortium for Small Animal Imaging, and the University ofWestern Ontario Academic Development Fund.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. David Griggs (Pfizer Global Research and Development,Chesterfield, MO) for providing the MDA-MB-435/HAL cell line, Mike Bygrave fortechnical support, Dr. Alan Tuck for expertise in pathology, and VisualSonics fortechnical support.

References1. Weissleder R. Scaling down imaging: molecularmapping of cancer in mice. Nat Rev Cancer 2002;2:11–8.

2. Choy G, Choyke P, Libutti SK. Current advances inmolecular imaging: noninvasive in vivo bioluminescentand fluorescent optical imaging in cancer research. MolImaging 2003;2:303–12.

3. Stripecke R, Carmen Villacres M, Skelton D, Satake N,Halene S, Kohn D. Immune response to green fluores-cent protein: implications for gene therapy. Gene Ther1999;6:1305–12.

4. Re F, Srinivasan R, Igarashi T, Marincola F, Childs R.Green fluorescent protein expression in dendritic cellsenhances their immunogenicity and elicits specificcytotoxic T-cell responses in humans. Exp Hematol2004;32:210–7.

5. Khanna C, Hunter K. Modeling metastasis in vivo .Carcinogenesis 2005;26:513–23.

6. Turnbull DH, Ramsay JA, Shivji GS, et al. Ultrasound

backscatter microscope analysis of mouse melanomaprogression. Ultrasound Med Biol 1996;22:845–53.

7. Fenster A, Downey DB, Cardinal HN. Three-dimen-sional ultrasound imaging. Phys Med Biol 2001;46:R67–99.

8. Kobayashi H, Saga T, Kawamoto S, et al. Dynamicmicro-magnetic resonance imaging of liver micrometa-stasis in mice with a novel liver macromolecularmagnetic resonance contrast agent DAB-Am64-(1B4M-Gd)(64). Cancer Res 2001;61:4966–70.

9. Yang M, Baranov E, Jiang P, et al. Whole-bodyoptical imaging of green fluorescent protein-express-ing tumors and metastases. Proc Natl Acad Sci U S A2000;97:1206–11.

10. Smakman N, Martens A, Kranenburg O, BorelRinkes IH. Validation of bioluminescence imaging ofcolorectal liver metastases in the mouse. J Surg Res2004;122:225–30.

11. Weber SM, Peterson KA, Durkee B, et al. Imaging ofmurine liver tumor using microCT with a hepatocyte-

selective contrast agent: accuracy is dependent onadequate contrast enhancement. J Surg Res 2004;119:41–5.

12. Fidler IJ. Selection of successive tumour lines formetastasis. Nat New Biol 1973;242:148–9.

13. Hill SA, Wilson S, Chambers AF. Clonal heterogeneity,experimental metastatic ability, and p21 expression inH-ras -transformed NIH 3T3 cells. J Natl Cancer Inst1988;80:484–90.

14. Varghese HJ, Davidson MT, MacDonald IC, et al.Activated ras regulates the proliferation/apoptosisbalance and early survival of developing micrometa-stases. Cancer Res 2002;62:887–91.

15. Schmidt CM, Settle SL, Keene JL, Westlin WF, NickolsGA, Griggs DW. Characterization of spontaneous metas-tasis in an aggressive breast carcinoma model using flowcytometry. Clin Exp Metastasis 1999;17:537–44.

16. Morris VL, MacDonald IC, Koop S, Schmidt EE,Chambers AF, Groom AC. Early interactions of cancercells with the microvasculature in mouse liver and

Cancer Research




muscle during hematogenous metastasis: videomicro-scopic analysis. Clin Exp Metastasis 1993;11:377–90.

17. Foster FS, Zhang MY, Zhou YQ, et al. A newultrasound instrument for in vivo microimaging ofmice. Ultrasound Med Biol 2002;28:1165–72.

18. Tong S, Cardinal HN, McLoughlin RF, Downey DB,Fenster A. Intra- and inter-observer variability andreliability of prostate volume measurement via two-dimensional and three-dimensional ultrasound imaging.Ultrasound Med Biol 1998;24:673–81.

19. Naumov GN, Townson JL, MacDonald IC, et al.Ineffectiveness of doxorubicin treatment on solitarydormant mammary carcinoma cells or late-developingmetastases. Breast Cancer Res Treat 2003;82:199–206.

20. Udagawa T, Fernandez A, Achilles EG, Folkman J,D’Amato RJ. Persistence of microscopic human

cancers in mice: alterations in the angiogenic balanceaccompanies loss of tumor dormancy. FASEB J 2002;16:1361–70.

21. Varghese H, Mackenzie L, Groom A, et al. In vivovideomicroscopy reveals differential effects of thevascular-targeting. Angiogenesis 2004;7:157–64.

22. Thorpe PE. Vascular targeting agents as cancertherapeutics. Clin Cancer Res 2004;10:415–27.

23. Goertz DE, Yu JL, Kerbel RS, Burns PN, Foster FS.High-frequency Doppler ultrasound monitors the effectsof antivascular therapy on tumor blood flow. Cancer Res2002;62:6371–5.

24. Czarnota GJ, Kolios MC, Abraham J, et al. Ultrasoundimaging of apoptosis: high-resolution non-invasivemonitoring of programmed cell death in vitro , in situand in vivo . Br J Cancer 1999;81:520–7.

25. Hunt JW, Worthington AE, Xuan A, Kolios MC,Czarnota GJ, Sherar MD. A model based uponpseudo regular spacing of cells combined with therandomisation of the nuclei can explain the signif-icant changes in high-frequency ultrasound signalsduring apoptosis. Ultrasound Med Biol 2002;28:217–26.

26. Oelze ML, O’Brien WD Jr, Blue JP, Zachary JF.Differentiation and characterization of rat mammaryfibroadenomas and 4T1 mouse carcinomas usingquantitative ultrasound imaging. IEEE Trans MedImaging 2004;23:764–71.

27. Xu HX, Yin XY, Lu MD, Liu GJ, Xu ZF. Estimation ofliver tumor volume using a three-dimensional ultra-sound volumetric system. Ultrasound Med Biol 2003;29:839–46.





2005;65:5231-5237. Cancer Res Kevin C. Graham, Lauren A. Wirtzfeld, Lisa T. MacKenzie, et al. ModelsLongitudinal Evaluation of Liver Metastases in Preclinical Three-dimensional High-Frequency Ultrasound Imaging for

Updated version

http://cancerres.aacrjournals.org/content/65/12/5231

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/65/12/5231.full#ref-list-1

This article cites 27 articles, 5 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/65/12/5231.full#related-urls

This article has been cited by 9 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/65/12/5231To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/content/65/12/5231.full#ref-list-1

http://cancerres.aacrjournals.org/content/65/12/5231.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



three-dimensional high-frequency ultrasound imaging for … · three-dimensional high-frequency...

Documents