Dynamic contrast-enhanced MRI to quantifyVEGF-enhanced tissue-engineered bladder graftneovascularization: Pilot study

Lisa Cartwright,1 Walid A. Farhat,1 Christopher Sherman,2 Jun Chen,3 Paul Babyn,4 Herman Yeger,3

Hai-Ling Margaret Cheng4

1The Hospital for Sick Children, Division of Urology, University of Toronto, Toronto, Ontario, Canada2Department of Pathology, Sunnybrook and Women’s College Hospital, University of Toronto,Toronto, Ontario, Canada3Department of Pediatric Lab Medicine, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada4Department of Diagnostic Imaging, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada

Received 25 July 2005; revised 3 October 2005; accepted 6 October 2005Published online 20 January 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.30648

Abstract: Tissue engineered organs require an immedi-ately perfused vascular tree. Currently, neovascularizationassessment requires animal sacrifice and graft harvest. Inthis pilot study we assess whether neovascularization of anengineered urinary bladder construct is enhanced with vas-cular-derived endothelial growth factor (VEGF) and assess-able non-invasively with dynamic contrast-enhanced MRI(DCE-MRI). Rabbit bladder acellular matrix was hybridizedwith hyaluronic acid (ACM-HA), fortified with one of threeconcentrations (0 ng, 10 ng, or 20 ng per gram of tissue) ofvascular-derived endothelial growth factor (VEGF), andgrafted onto bladders in nine rabbits (3 per VEGF concen-tration). At 1, 2 and 3 weeks, one rabbit from each VEGFgroup underwent DCE-MRI and graft harvest. Microvascu-lature was quantified with scrial optical transverse section-ing of CD31 stained whole mounts using PCl software.Masson trichrome and H&E staining were used to assess

cellular proliferation and fibrosis. There was a significantdifference in mean microvascular area (MVA) between the10 and 20 ng VEGF groups (230187 vs 477894 �m2, P �0.014) but not between the 0 and 10 ng groups (210497 �m2,P � 0.21). Gadolinium uptake increased with MVA andcorrelated with it on linear regression analysis (Pearson’scorrelation coefficient r � 0.71). At three weeks, stromalcellularity was greatest, and fibrosis was least, in the highVEGF group. This preliminary work demonstrates im-proved neovascularization of bladder constructs with VEGFfortification of ACM-HA and the feasibility of MRI as anon-invasive method to assume angiogenesis in tissue engi-neered constructs. © 2006 Wiley Periodicals, Inc. J BiomedMater Res 77A: 390–395, 2006

Key words: tissue engineering; dynamic contrast-enhancedMRI; angiogenesis; acellular matrix; bladder

INTRODUCTION

We utilize bladder acellular matrix (ACM) as a scaf-fold to engineer constructs histologically similar tourinary bladder 1 in which in vivo cellular growthrequires support from early angiogenesis and perfu-sion.2 While graft peripheries may be nourished byhost tissue preceding neovascularization, graft centersdemonstrate fibrosis and contracture without earlyneovascularization.3 We seek to incorporate vascular-derived endothelial growth factor (VEGF) and hyal-uronic acid (HA) into ACM to improve graft function.Theoretically, HA would decrease porosity and in-flammation and act as a carrier for VEGF, which in

turn would aid neovascularization. Additionally, welack methods to non-invasively assess angiogenesis inthese constructs. Magnetic resonance imaging (MRI)has provided measures for tumor angiogenesis in clin-ical and experimental models 4 and was used to assessbenefits of VEGF on cardiac collateral development ina porcine chronic ischemia model.5

In this pilot study, we sought to incorporate VEGFinto a HA–ACM hybrid construct, to determine if itwould improve neovascularization and limit fibrosisin vivo, and to evaluate DCE-MRI as a noninvasivetool to assess neovascularization.

MATERIALS AND METHODS

ACM–HA VEGF constructs

Whole rabbit urinary bladders were halved and acellular-ized using a modified proprietary method. 1 In brief, blad-

Correspondence to: W. A. Farhat, University of Toronto, 555University Ave., Rm 292, Toronto, Ontario, Canada M5G1X8; e-mail: walid.farhat@sickkids.ca

© 2006 Wiley Periodicals, Inc.

ders were washed in sterile phosphate buffer saline (PBS)and then stirred in a hypotonic solution of 10 mM Tris HClpH 8.0, 5 mM EDTA, 1% Triton X-100, Petabloc Plus™ (pro-tease inhibitor) 0.1 mg/mL, and antibiotics/antimycotic at4°C for 24 h to lyse all cellular components. On the secondday, the tissue was placed in a hypertonic solution contain-ing 10 mM Tris HCl pH 8.0, 5 mM EDTA, 1% Triton X-100and 1.5M KCl and stirred for 24 h at 4°C to denature residualproteins. Tissue was then washed in Hanks’ Balanced Saltsolution for 1 h at room temperature twice prior to a 6 henzymatic digestion with DNAse/RNAse solution at 37°Cwith shaking. A final 24 h extraction was performed at 4°Cin 50 mM Tris HCl pH 8.0, 0.25% CHAPS, 1% Triton X-100,and antibiotics/antimycotic with shaking. The resultingACM was finally washed four times in sterile dH2O at 4°Cand then stored in physiological saline. H&E staining andDAPI immunoflourescence confirmed acellularity.

Hyaluronic acid incorporation was done as follows; ACMwas lyophilized (VirTis- Temp: �70°C, Vacuum: 121 mtorr)then rehydrated over 24 h at 37°C in increasing concentra-tions of HA (0.05, 0.1, 0.2, and finally 0.5 mg/100 mL; Sigma,product #: H5388, USA). Alcian blue staining and flouro-phore-assisted carbohydrate electrophoresis glycosamino-glycan assays confirmed HA uptake. The resultant HA–ACM constructs were then dehydrated in alcohol,lyophilized, rehydrated in one of three different VEGF121

(Sigma, #: V3388, USA) concentrations (0 ng, 10 ng, 20 ng/gACM) and maintained in VEGF prior to implantation.

In vitro VEGF uptake and retention

VEGF upake and in vitro retention by porcine ACM wasquantified with ELISA performed in triplicate on proteinextractions from HA–VEGF–ACM samples (0,1.25, 2.5, 5, 10,and 20 ng VEGF/gACM). To determine VEGF release, sam-ples were immersed in MEM for 6, 12, or 24 h. VEGF121 inprotein extraction supernatant was measured using ELISA(QuantiGlo R&D System, Cat No: QVE00B, MN) followingmanufacturer’s protocol, using a Microplate Luminometer(EG&G, Berthold, Germany). To assess the linearity of theassay and create a standard curve, samples containing highconcentrations were diluted to produce samples with valueswithin the dynamic range of the assay {VEGF121 (range: 0,6.4, 32,160, 800, and 4000 pg/mL)}.

Graft implantation

Nine female New Zeland white rabbits (3.0–3.5 kg) werefasted overnight. Anesthesia was induced with Akmezineand maintained with 1.5% Halothane and oxygen on auto-matic ventilatory support with endotracheal intubation. In-travenous Penicillin was administered on induction. AC-M–HA VEGF constructs (n � 3/VEGF concentration, 4 �4–5 � 5 cm2) were implanted extraperitoneally onto theanterior and lateral bladder wall through a lower midlineincision with non-absorbable suture for graft identificationat harvest. The abdomen was closed in two layers. Animalswere recovered, returned to regular animal pens, and ad-

ministered antibiotics (Penicillin and Streptomycin) and an-algesia (Tamgesic i.m.) for one week postoperatively.

At 1, 2, and 3 weeks, one rabbit from each VEGF groupwas anesthetized with intramuscular ketamine, rompun,and atropine (14, 4, and 0.02 mg/kg body weight respec-tively) with pentobarbital maintenance (6–13mg/kg) tocomplete DCE-MRI prior to euthanization with pentobarbi-tal overdose. To decrease the chance of having inflammatorychanges secondary to the sutures used to fix the ACM to thebladder, we ensured that both MRI and histological andimmunohistochemical analysis were made in the centralregion of the graft bladder away from the edges (central 2 �2 cm2 construct grafts and underlying native tissue wereharvested from empty bladders).

Experimental protocols were approved by the AnimalCare Committee of the Hospital for Sick Children and com-plied with the Canadian Council for Animal Care guide-lines.

MRI

MRI was completed on a 1.5-tesla Signa LX whole-bodyMRI scanner (General Electric Systems, Milwaukee, WI) us-ing a cardiac phased array coil with the animal prone. Theradiologist was blinded to VEGF concentrations.

Bladder constructs were identified on T2-weighted multi-slice fast spin-echo (FSE) images with the following param-eters: TR/TE � 4000/105 ms, 8 ETL, 2 NEX, matrix � 256 �192, slice thickness (SL) � 3 mm, field of view (FOV) � 12cm. Dynamic images were then acquired using a 3D T1-weighted fast spoiled gradient recalled echo (SPGR) se-quence prior to, during, and for 8.5 min after rapid intrave-nous injection of 0.1 mmol/kg Gd-DTPA (Magnevist, BerlexCanada, Montreal). The parameters were: TR/TE � 9.3/2.1ms, flip angle (FA) � 15°, bandwidth (BW) � 15.62 kHz,FOV � 12 cm, matrix � 256 � 192 � 12, SL � 3 mm, 1 NEX.Baseline T1-maps were acquired before contrast administra-tion, using the Look–Locker method. High resolution, fat-suppressed, T1-weighted FSE images were also acquiredbefore and following contrast to delineate the enhancingregion (TR/TE � 500/15 ms, 4 ETL, 3 NEX, matrix � 256 �192, SL � 3 mm, FOV � 12 cm).

MRI resolution was 0.47 � 0.63 � 3.0 mm3. ACM waslocalized on sagittal and axial T2-weighted FSE images andconfirmed by comparison to bladder harvest photographs.Four or five contiguous 3-mm image slices through thecentral portion of the ACM construct were identified forsubsequent analysis. On each imaging slice, a region-of-interest (ROI) was outlined to include the most uniformlyenhancing area on T1-weighted images 4 min post-injection.Average ROI signal intensity was used to calculate the post-contrast T1, based on the pre-contrast value and the signalintensity equation for a FSPGR sequence. Tracer concentra-tion was then estimated assuming a linear relationship withthe change in 1/T1. Area under the concentration–timecurve (AUC) was integrated for the ROI over the first 1, 2,and 8 min post-injection. The results were normalized toresting dorsal muscle to reduce variations due to arterialinput. A final AUC value was obtained by averaging over all

QUANTIFICATION OF BLADDER GRAFT NEOVASCULARIZATION BY DCE-MRI 391

slices through the ACM center. Further technique detailspertaining to the MRI have been described by Cheng, et al.6

Immunohistochemistry

Microdissection separated ACM from native bladder andCD31 (vascular endothelial marker) staining was evaluated.Cryosections (140 �m) from the ACM center were blockedfor non-specific binding with 15% normal goat serum inPBS, and incubated overnight at 4° with the anti-humanmonoclonal antibody CD31 (1:30 dilution, clone JC/70,DAKO, product#: M0823, Glostrup, Denmark). Sectionswere incubated with the flourochrome-labeled secondaryantibody, Alexa Fluor 594 chicken anti-mouse IgG (H�L)(1:200 dilution, Molecular probe, #A21201, OR) andmounted in 90% glycerol/PBS containing p-phenylethyl-enediamine (1 mg/mL) to reduce flurochrome bleaching.

Whole mount staining fluorescent images were obtainedwith confocal laser microcsopy in 30 layers (�10 Carl ZeissAxiovert 200) utilizing LSM 510 program software (version3.2), coupled to a color charge-coupled device video camera,digitizing card, and monitor with Simple PCI software toestimate microvascular area (MVA) (�m2).

Histology

Central sections were stained with Masson trichrome andH&E to evaluate cellularity and fibrosis.

Statistical analysis

A two-tailed Student’s t-test was used to determinewhether MRI AUC1 min, 2 min, 8 min and immunohistochem-ical MVAs varied significantly with VEGF, and whetherMRI could predict VEGF dose. MVA and MRI AUC werecompared with Spearman rank correlation and Pearson’slinear regression.

RESULTS

VEGF uptake/retention

VEGF uptake demonstrated by HA–ACM was lin-ear relative to the added VEGF concentrations (Figure1A). In vitro, VEGF was released as early as 6 h withvery little VEGF residual at 24 h regardless of initialVEGF concentration (Figure 1B).

Microvascularity and VEGF

At all time points, higher MVA on CD31 stainingwas seen with increasing levels of VEGF (Figure 2A).

Difference in mean MVA between control and lowVEGF groups did not reach statistical significance(mean � SD: 210,497 � 27,319 �m2 vs. 230,187 �26,460 �m2; p � 0.21). However, the high VEGF meanMVA was significantly higher than the low VEGFmean MVA (477,894 � 123,945 �m2, p � 0.014) (Figure3A).

Histology

Biocompatability was manifested by cellular re-population, little inflammatory response, and similarneovascularization at the implant edge and center.Vascularization was accentuated with higher VEGFlevels. At one week, grafts from all groups demon-strated moderate cellularity and fibrosis, which weremaintained with high VEGF at three weeks, whencontrol and low VEGF grafts demonstrated low cellu-larity and dense fibrosis (Figure 2 B,C).

Figure 1. A: Linear uptake of VEGF by porcine HA–ACMconstruct from 0–20 ng VEGF/g tissue. B: VEGF extractedfrom ACM–HA–VEGF constructs after immersion in MEMdemonstrating rapid release of VEGF in vitro.

392 CARTWRIGHT ET AL.

DCE-MRI

Grafts were identified by thickening and greater en-hancement relative to surrounding bladder. Generally,faster and greater contrast uptake was obtained withmore VEGF. AUC correctly distinguished all VEGFgroups. A significant increase was noted in AUC8 min atall times after implantation between low and high (10and 20 ng) VEGF groups (p � 0.025, figure 3B).

DCE-MRI versus MVA and VEGF

MVA and AUC differences between control andlow VEGF groups were modest (p � 0.21, �0.17). Arelatively small MVA change (�9%) between controland low VEGF groups was accompanied by a muchlarger AUC increase (50% AUC8 min). Comparing lowand high VEGF groups, where MVA difference wasover two-fold (p � 0.01), a statistically significant, butsmaller relative difference was observed in AUC8 min(p � 0.038).

Linear regression analysis showed that all AUCscorrelated with MVA (Pearson’s correlation coefficientr � 0.71, 0.79, 0.82 for AUC8 min, AUC2 min, AUC1 min),and correctly ranked the data, except AUC1 min for one

Figure 3. Comparisons of MVA, VEGF, and DCE-MRI pa-rameter AUC. A: MVA and B: AUC8 min over time by VEGFconcentration. C: AUC8 min correlates with MVA and cor-rectly ranks the data.

Figure 2. Three week samples : A: Composite images ofCD-31 immunostaining demonstrates markedly increasedmicrovascular area with high VEGF concentration comparedwith control and low VEGF concentration. B: H&E stainingwith improved cellularity and decreased fibrosis in the highVEGF group compared with the control and low VEGFgroups. C: Masson trichrome shows improved cellularityand decreased fibrosis in the high VEGF group comparedwith the control and low VEGF groups.

QUANTIFICATION OF BLADDER GRAFT NEOVASCULARIZATION BY DCE-MRI 393

rabbit (Figure 3C). An extensive evaluation of MRIparameters was reported by Cheng et al.6

DISCUSSION

Central fibrosis and contracture in engineered blad-der constructs is theorized to be due to inadequaterapid neovascularization for central graft support,while the graft periphery is rapidly neovascularizedby the surrounding host tissue. Developments wouldbe hastened with a reliable, noninvasive tool to assessangiogenesis. Currently, three-dimensional assess-ment of angiogenesis requires graft harvest. This pilotstudy provides preliminary results supporting im-proved ACM neovascularization after hybridizationwith HA and VEGF, and MRI as a useful noninvasivetool to monitor neovascularization as a starting pointfor further investigation. This approach is promisingbut, given the complexity of the angiogenesis process,it would be rather naı̈ve to think that administration ofa single angiogenic molecule would be sufficient togenerate a well-tempered and durable neovasculariza-tion. Regeneration of functionally normal bladder tis-sue will require advances beyond current tissue engi-neering technology such as biomaterials that interactwith the cells in a controlled and predictable fashion.For instance, the ability to temporally display impor-tant growth factors to the surrounding cells during theregenerative process is highly desirable.

On the basis of the theory that bladder ACM poros-ity 7 has precluded VEGF incorporation, we attemptedto decrease porosity with the macromolecule HA,which might also serve as a carrier for other mole-cules. Functionally, ACM would guide cells, whileincorporated substances would stimulate angiogene-sis, support cellular ingrowth and minimize inflam-mation. We have demonstrated the feasibility of thistechnique with a preliminary dose-response for angio-genesis with VEGF. Differences in cellular prolifera-tion and fibrosis detected over time with varyingVEGF concentrations may be related to angiogenesisor effects of VEGF on other factors such as inflamma-tion. The trend for increased neovascularization, im-proved cellularity, and decreased fibrosis with VEGFwarrants verification with a larger study and othervascular markers.

Using ELISA, VEGF uptake by HA–ACM was linearrelative to the added VEGF concentrations, and intu-itively, difference in vascularization or angiogenesis at3 weeks is secondary and proportional to the amountof incorporated VEGF but may not be fully reflected tothe in vitro release rate, which is depicted in figure 1B.The pharmacokinetics of the incorporated VEGF andin vivo release is what is important and may need to beclosely investigated in future studies. On the other

hand, if these hybrid constructs are to be used forurinary bladder substitution, whereby urine contactwith the construct might result in VEGF or HA loss, asevidenced by rapid in vitro release of VEGF. But evenif VEGF is released quickly in vivo, its release to thesurrounding tissues may stimulate angiogenesis. Fur-thermore, the need for a vascular network for bladdertissue engineering is recognized, the ideal vasculardensity to support appropriate bladder regenerationon an ACM scaffold is unknown.

We chose MRI over ultrasound, CT scan, and PETscan for superior soft tissue contrast and high resolu-tion functional and anatomic imaging. While ultra-sound may identify vascular features in tissues atdifferent levels of resolution,8 anatomic accessibilityand operator dependence may make it less useful inthis application. CT may also be applied functionallyto assess perfusion. However, there had been littlevalidation of ultrasound or CT with accepted angio-genesis surrogates.

In our preliminary evaluation, DCE-MRI showedpromise as useful noninvasive tool for angiogenesisassessment. It is important to note that AUC on MRIcorresponded well with microvascularity, with a moresignificant difference in AUC than in MVA at lowVEGF concentrations. As DCE-MRI provides func-tional and morphological information, this may indi-cate that changes in addition to neovascularizationaffecting the flow regime such as vessel or tissue per-meability occur at lower VEGF concentrations. Addi-tionally, while CD31 stains all vascular tissues regard-less of functional status, DCE-MRI only assesses thosevessels that are perfused.

DCE-MRI evaluation in this study had severallimitations. Gd–DTPA has a low molecular weightand diffuses rapidly through hyperpermeable ves-sels as may be present with neovascularization.While distinction of the graft from the underlyingnormal bladder was difficult, the regions evaluatedenhanced uniformly, and improved delineationwith higher spatial resolution would have been un-likely to improve results significantly. Imaging andanalysis techniques will evolve with our under-standing of vascular properties. Studies are under-way to assess larger contrast agents (less likely toleak through hyperpermeable microvasculature)and finer VEGF concentration increments (widervascular permeability and density distribution) tofurther understand angiogenic modulation in tissueengineering and its imaging correlate.

CONCLUSION

ACM hybridization shows promise to enhance largeorgan tissue engineering. Specifically, initial ACM hy-

394 CARTWRIGHT ET AL.

bridization with HA enabled subsequent incorpora-tion of VEGF, which in turn demonstrated a dose-response with microvascular density in vivocorrelating with DCE-MRI parameters. Over time,higher VEGF concentrations were associated with themaintenance of moderate cellularity and fibrosis,while low concentration and controls demonstrateddecreasing cellularity and increasing fibrosis. Furtherstudies are needed for verification, determination ofoptimal VEGF concentrations and microvascular den-sities, optimization of VEGF incorporation, and searchfor other growth factors that potentiate the regenera-tive capacity of the scaffolds. In addition, correlationof angiogenesis with other MRI parameters, and stan-dardization of the DCE-MRI technique for noninva-sive angiogenesis monitoring in tissue-engineeredconstructs is warranted.

References

1. Reddy PP, Barrieras DJ, Wilson G, Bagli DJ, McLorie GA,Khoury AE Merguerian PA. Regeneration of functional blad-der substitutes using large segment acellular matrix allograftsin a porcine model. J Urol 2000;164:936–941.

2. Nor JE, Peters MC, Christensen JB, Sutorik MM, Linn S, KhanMK, Addison CL, Mooney DJ Polverini PJ. Engineering andcharacterization of functional human microvessels in immuno-deficient mice. Lab Invest 2001;81:453–463.

3. Brown AL, Farhat W, Merguerian PA, Wilson GJ, Khoury AEWoodhouse KA. 22 week assessment of bladder acellular ma-trix as a bladder augmentation material in a porcine model.Biomaterials 2002;23:2179–2190.

4. Taylor JS, Tofts PS, Port R, Evelhoch JL, Knopp M, ReddickWE, Runge VM Mayr N. MR imaging of tumor microcircula-tion: Promise for the new millennium. J Magn Reson Imaging1999;10:903–907.

5. Pearlman JD, Laham RJ Simons M. Coronary angiogenesis:Detection in vivo with MR imaging sensitive to collateral neo-circulation–Preliminary study in pigs. Radiology 2000;214:801–807.

6. Cheng HL, Chen J, Babyn PS Farhat WA. Dynamic Gd-DTPAenhanced MRI as a surrogate marker of angiogenesis in tissue-engineered bladder constructs: A feasibility study in rabbits. JMagn Reson Imaging 2005;21:415–423.

7. Farhat W, Chen J, Erdeljan P, Shemtov O, Courtman D,Khoury A Yeger H. Porosity of porcine bladder acellularmatrix: Impact of ACM thickness. J Biomed Mater Res 2003;67A:970 –974.

8. Ferrara KW, Merritt CR, Burns PN, Foster FS, Mattrey RFWickline SA. Evaluation of tumor angiogenesis with US: Im-aging, Doppler, and contrast agents. Acad Radiol 2000;7:824–839.

QUANTIFICATION OF BLADDER GRAFT NEOVASCULARIZATION BY DCE-MRI 395