imaging and advanced technology - mitweb.mit.edu/hst.035/readingsupplement/04_28_coron... · in the...

IE

MP

*UG

Bc(i(pCi((wisachmtbEfl[hn0Cagcip

Cd

Imaging and Advanced Technology
Michael B. Wallace, Section Editor
n Vivo Molecular Imaging of Colorectal Cancer With Confocalndomicroscopy by Targeting Epidermal Growth Factor Receptor

ARTIN GOETZ,* ALEX ZIEBART,* SEBASTIAN FOERSCH,* MICHAEL VIETH,‡ MAXIMILIAN J. WALDNER,§

ETER DELANEY,� PETER R. GALLE,* MARKUS F. NEURATH,*,¶ and RALF KIESSLICH*

I. Medizinische Klinik und Poliklinik, Universitätsmedizin Mainz, Mainz; ‡Institute of Pathology, Klinikum Bayreuth, Bayreuth; §Institute of Molecular Medicine,niversitätsmedizin Mainz, Mainz, Germany; �Optiscan Pty, Ltd, Notting Hill, Victoria, Australia; and ¶Medizinische Klinik 1, Universitätsklinikum Erlangen, Erlangen,
ermany
scitEra

2pttmolsotrtnbaptvs

conli

cI

See editorial on page 419.

ACKGROUND & AIMS: Epidermal growth factor re-eptor (EGFR) is a therapeutic target for colorectal cancerCRC). However, technical challenges have limited in vivomaging of EGFR in CRCs. Confocal laser endomicroscopyCLE) enables accurate microscopic visualization of CRC inatients during endoscopy. We evaluated the ability to useLE in vivo for instantaneous molecular imaging of EGFR

n CRC models. METHODS: Tumors were grown in micen � 68) from human CRC cell lines that expressed highSW480 cells) or low (SW620 cells) levels of EGFR. Tumorsere visualized in vivo with a handheld CLE probe after

njection of fluorescently labeled EGFR antibodies. EGFR-pecific fluorescence was graded from 0 to 3�. Neoplasticnd non-neoplastic specimens from human colorectal mu-osa were examined. In vivo findings were correlated withistopathology, immunohistochemistry, and fluorescenceicroscopy analyses. RESULTS: CLE analysis of cell cul-

ures confirmed the different expression levels of EGFRetween cell lines. In living animals, CLE differentiatedGFR expression levels between tumor cell limes (meanuorescence, 1.92 � 0.22 [SW480] and 0.59 � 0.21

SW620], P � .0004). CLE analysis of EGFR expression inuman specimens allowed distinction of neoplastic fromon-neoplastic tissues (mean fluorescence, 2.0 � 0.37 vs.25 � 0.16, respectively, P � .0035). CONCLUSIONS:LE can be used for in vivo, molecular analysis of CRCnd to differentiate EGFR expression patterns in xeno-raft tumors and human tissue samples. Because CLEan be performed during endoscopy, in vivo molecularmaging might be used in diagnosis of CRC and toredict response to targeted therapies.

olorectal cancer (CRC) is the second leading cause ofcancer-related deaths in Western countries.1 Early

etection during colonoscopy has been associated with a

ignificantly improved survival.2,3 Autonomous cancerell growth occurs through a multistep progress thatncludes up-regulation of growth factors and their recep-ors such as epidermal growth factor receptor (EGFR).GFR is a transmembrane glycoprotein tyrosine kinaseeceptor that promotes tumor proliferation, invasion,nd metastasis formation and neovascularization.4

In CRC, EGFR overexpression has been reported in5%–94% of cases5–7 and has been associated with a poorrognosis8,9 –11 and early recurrence.12 The central role ofhe EGFR-mediated pathway has made it an attractivearget for anticancer therapy. Cetuximab, a chimeric

onoclonal antibody targeting the extracellular domainf EGFR, competitively prevents binding of the natural

igand and triggers the internalization of the receptor. Ithows clinically beneficial antitumor activity.13,14 EGFRverexpression on the tumor was mandatory for earlyrials evaluating anti-EGFR therapies.15 This view hasecently been challenged by demonstrating kras muta-ions in the downstream signaling cascade as more sig-ificant16,17 and by postulating a lack of relationshipetween EGFR expression and response to targeted ther-py.18,19 Together, all these trials demonstrate efforts torovide an individualized therapy to ensure optimal an-itumor efficacy while at the same time minimizing ad-erse events and the financial burden on the health careystem.20

This requires accurate selection of patients whoould benefit from such a targeted therapy. Numerousptical technologies have been mandated for intravitaloninvasive functional detection of (pre-)neoplastic

esions, such as spectroscopy,21,22 autofluorescence imag-ng,23–25 and induced fluorescence imaging.26,27 Recently,

Abbreviations used in this paper: CLE, confocal laser endomicros-opy; CRC, colorectal cancer; EGFR, epidermal growth factor receptor;HC, immunohistochemistry.

© 2010 by the AGA Institute0016-5085/10/$36.00

doi:10.1053/j.gastro.2009.10.032

GASTROENTEROLOGY 2010;138:435– 446

cdgdtoAiatvsio

(aa(flaGmcpCptad

Ilp(tbwFcg

oLmiinv

pmaxrkps

eeolv�scn(PirFimdbtititctp

aiebl

fiStltrE

Imaging and Advanced Technology continued

4

onfocal laser endomicroscopy (CLE) has been intro-uced, providing immediate microscopy during the on-oing endoscopy. With this novel technique, an intravitalifferentiation among normal, regenerative, and neoplas-ic mucosal changes is possible with high accuracy basedn validated and reproducible morphologic criteria.28 –30

lthough a first study has recently reported molecularmaging with CLE,31 no trial has used labeled antibodiesgainst a well-defined tumor-associated target, and, untiloday, analysis of receptor expression is a domain of exivo immunohistochemistry. The aim of the currenttudy was to evaluate fluorescent targeting of EGFR forn vivo molecular imaging using CLE in a rodent modelf human CRC and in human colon tissue.

Materials and MethodsTumor Cell LinesSW480 and SW620 cells32 were cultured in RPMI

with 10% fetal calf serum, 1% glutamine, 1% streptomycin,nd 1% penicillin) at 37°C (5% CO2). Cells were harvestedfter 24–48 hours. For fluorescence-activated cell sortingFACS), 105 tumor cells were incubated with 2.5 �L ofuorescein isothiocyanate (FITC)-labeled anti-EGFR-ntibody (CBL416F; Chemicon/Millipore, Schwalbach,ermany), isotype control (36875K; Pharmingen, Ger-any), or no antibody (negative control). Immunohisto-

hemistry (IHC) of cytospins of single cell suspensions waserformed with EGFR pharmDx (DakoCytomation,arpinteria, CA). In short, slides were incubated withroteinase K and blocked with peroxidase. After incuba-ion with the primary antibody, labeled polymer wasdded, and specific staining was visualized with 3,3=-iamino-benzidine.Tumor cells cultured in chamber slides (Nalge Nunc

nternational, Naperville, NY) were incubated with FITC-abeled anti-EGFR-antibody without fixation. Cell sus-ensions were first evaluated by confocal microscopy

FIVE1; Optiscan, Notting Hill, Australia), using exactlyhe same settings as for in vivo mouse imaging. Next,ench top fluorescence microscopy of cell suspensionsas performed (IX70; Olympus, Hamburg, Germany).ITC-labeled anti-EGFR was detected at 516 nm, nuclearounterstaining with DAPI (Vector Laboratories, Burlin-ame, CA) at 470 nm.

Animal ModelsTumors were induced by injection of 107 SW480

r SW620 cells in 68 Balb/c nu/nu mice (Charles Riveraboratories, Inc, Sulzfeld, Germany). Subcutaneous tu-ors were induced in the groin of 55 mice, cells were

njected into the spleen after a small left lateral incisionn 6 mice,33 and a 5-mm vital xenograft from a subcuta-eously induced tumor was implanted into a cecal pouch
ia a small abdominal incision in 7 mice.34 Imaging was s
36

erformed after 3– 6 weeks when tumor size approxi-ated 5–10 mm or wasting occurred. For confocal im-

ging, mice were deeply anesthetized using ketamine-ylazine (120 mg/kg and 16 mg/kg intraperitoneally [IP],espectively). After the examination, mice were killed by aetamine-xylazine overdose. Animal procedures were ap-roved by the local review board (Landesuntersuchung-amt Rheinland-Pfalz: 1.5 177-07-04/051-59).

CLE and Staining ProtocolIn the rigid confocal probe (FIVE1; Optiscan), an

xcitation wavelength of 488 nm was delivered, and lightmission was detected at 505–585 nm. Serial en faceptical sections of 475 � 475 �m were obtained with a

ateral resolution of 0.7 �m (1024 � 1024 pixels) atariable imaging depth (adjustable from surface to 250m) but with otherwise predefined and fixed instrument

ettings throughout molecular imaging experiments in-luding laser power of 1000 �W and standardized bright-ess and � control. For tumor morphology, Fluorescein

Alcon Pharma, Freiburg, Germany) and acriflavine (Sigmaharmaceuticals, South Croydon, Victoria, Australia) were

njected intracardially at 100 �g/g and 10 �g/g body weight,espectively, as previously established.35–37 0.1 �g/g OfITC-labeled anti-EGFR antibodies (Chemicon/Millipore)

n 200 �L phosphate-buffered saline were injected 45inutes prior to confocal imaging after a run-in phase

emonstrated that this time frame permits antibodyinding to target sites. For intravital confocal imaging,he tumor was exposed by a small skin or abdominalncision, and the handheld confocal probe was posi-ioned in a way to achieve full contact of the confocalmaging window with the tissue. The complete tumorissue was screened with the confocal probe for fluores-ent signal by carefully moving the probe across theumor surface while at the same time adapting imaginglane depth.Confocal imaging of the tumor before the injection of

nti-EGFR-antibody or after injection of a FITC-labeledsotype control antibody served as negative controls toxclude interference by autofluorescence or unspecificinding. In addition, fluorescence in kidney, spleen, and

iver was evaluated after antibody injection.

Ex Vivo Correlation of EGFR ExpressionFrom the examined sites, tissue specimens were

xed in 4% buffered formalin and embedded in paraffin.erial sections of 4 �m were stained with H&E. Addi-ional specimens from 21 mice were snap frozen iniquid nitrogen. Bench top fluorescence microscopy ofumor cryosections was performed without EGFRestaining (to visualize in vivo bound FITC-labeledGFR antibodies) and with ex vivo nuclear counter-
tain as specified above.

caihcowma

ni6Ffsw(ttpdpm

swqwtrtjcg

sLssbsFcascat

FSc


EGFR immunostaining using the immunohisto-hemical system kit EGFR pharmDx was performedccording to the EGFR pharmDx scoring guidelines. Itncluded the front of invasion from the mean of 3igh-power fields because this region was shown toontain the greatest density of EGFR-positive cells. IHCf tumor cell membranes was scored as follows: 1�,eak; 2�, moderate; and 3�, strong. The absence ofembrane staining or cytoplasmic staining was reported

s negative (0).

Imaging of Human TissueColonoscopic biopsy or surgical specimens of

on-neoplastic (normal, n � 5; unspecific colitis, n � 1;schemia, n � 2) and neoplastic colon tissue (CRC, n �; intraepithelial neoplasia, n � 2) were incubated withITC-labeled anti-EGFR-antibody (Chemicon/Millipore)or 30 minutes at 1:50 dilution at room temperaturehielded from light. Confocal imaging was performedithout further tissue processing using the FIVE1 probe

n � 14) or the confocal endomicroscope (n � 2) that hashe same optical properties (Pentax EC-3870CIFK; Pen-ax, Tokyo, Japan). IHC was performed with EGFRharmDx. Final histopathology served as the gold stan-ard to classify specimens. The study protocol was ap-roved by the Ethics Committee of Rheinland-Pfalz, Ger-any (No. 837.321.03).

igure 1. EGFR staining ex vivo. After FITC-labeling of tumor cells witW480 cells (A) with a high EGFR expression and SW620 cells with a lo
ytospins (C and D).
Statistical AnalysisFrom confocal images captured in vivo, EGFR-

pecific fluorescence intensity was graded as absent (0),eak (1�), moderate (2�), or strong (3�). This semi-uantitative grading was used to match in vivo stainingith scoring in IHC and to provide a scale that is easily

ransferable to clinical practice. Confocal images wereanked independently by 2 investigators. If differences inhe grading occurred, consensus was obtained by againointly evaluating the images (n � 3 tumors). Fluores-ence from bench top microscopy of cryosections wasraded accordingly.

Statistical analysis was performed using the statisticaloftware package GraphPad Prism (v5.00; GraphPad, Inc,a Jolla, CA). Fisher exact test was used to examine thetatistical significance of differences of the fluorescenceignal strength between SW480 and SW620 tumors andetween neoplastic and non-neoplastic human tissueamples. All P values were generated using 2-sided tests.or the primary hypotheses of differences in the fluores-ence signal strength between SW480 and SW620 tumorsnd between neoplastic and non-neoplastic human tissueamples, the global significance level was set at 5%. Be-ause a second analysis investigated the semiquantitativessessment of fluorescence strength, a Bonferroni correc-ion was used to adjust for these 2 tests in both groups;

i-EGFR antibody in vitro, FACS analysis clearly differentiates betweenFR expression (B). These expression patterns are confirmed by IHC of

h antw EG

437

awsvsccc

saaos

emlSsp

flfi

tttwowo

wptmhos(sr

FEipt


4

difference was considered statistically significant if Pas � .025. The comparison of means of the fluorescent

ignal strength was regarded as explorative, and the Palues of the corresponding tests are presented for de-criptive reasons only. Spearman rank correlation coeffi-ient was used to correlate in vivo confocal endomicros-opy and ex vivo bench top fluorescence microscopy onryosections.

ResultsDifferential EGFR Expression by TumorCell LinesFACS analysis confirmed the high EGFR expres-

ion of SW480 cells, whereas SW620 cells only showed anbsent-to-mild EGFR staining. These patterns were par-lleled by a strong specific staining of SW480 cells in IHCf cytospins in contrast to an almost absent specifictaining of SW620 cells (Figure 1).

The ability of handheld CLE to discriminate EGFRxpression was corroborated by visualizing unfixed tu-or cells in cell culture after incubation with FITC-

abeled antibody: A strong cellular signal was seen inW480 cell culture, whereas SW620 cells did not show apecific signal. EGFR-positive tumor cells showed a su-erficial, cytoplasmic, or (peri-)nuclear signal. Bench top

igure 2. Molecular imaging of cell cultures by CLE. Unfixed SW480 ceGFR antibody and examined by handheld confocal microscopy. An in

n EGFR-negative SW620 cells with CLE (A and D). With the same staininatterns, displaying bound antibody in green on SW480 cells (B) but not
o localize EGFR fluorescence to superficial or cytoplasmic cellular compartm
38

uorescence microscopy of SW480 or SW620 cells con-rmed the staining patterns from CLE (Figure 2).

Tumor Morphology in VivoTumor morphology was readily available using

he FIVE1 confocal probe in both SW480 and SW620umors. After acriflavine injection, the densely packedumor cells with an altered nuclear-to-cytoplasmic ratioere visualized. Fluorescein depicted the disorganizedverall tumor tissue structure. In vivo findings correlatedell with ex vivo H&E staining, as demonstrated previ-usly.35

In Vivo Molecular Imaging of Human CRCin Rodent ModelsWith the handheld confocal probe, large areas

ere screened for specific fluorescence by moving therobe gently across the tumor surface. In most SW480umors, CLE found a clear specific signal within the

alignant cells after systemic antibody application. Theigh resolution of the confocal system permitted thebservation of cytoplasmic, cell surface, or perinuclearignal, similar to confocal microscopy of cell culturesFigure 3). In bench top fluorescence microscopy, cryo-ections of SW480 tumors confirmed the intravital fluo-escence patterns after in vivo binding of FITC-labeled

per panels) or SW620 cells (lower panels) were incubated with labeledal signal was observed in EGFR-positive SW480 cells (arrows) but nottocol, bench-top fluorescence microscopy confirmed these expressionW620 cells (E). Overlay with nuclear counterstaining (DAPI blue) helped

lls (uptravitg proon S

ents (C and F).

Fo

FFfl


igure 3. High-resolution molecular imaging in vivo. Within a SW480 tumor, a membranous staining pattern is visualized in vivo (arrows, A). Magnification
f a single cell (boxed area in A) shows accumulation of the fluorescence signal at the level of the cell surface (arrows, B).
igure 4. Molecular imaging of SW480 tumors in live mice. CLE showed a clear specific signal within the tumor tissue after systemic application ofITC-labeled anti-EGFR antibody (arrows, A). Specificity of this signal was confirmed by ex vivo IHC of tumor sections (arrows, B). Ex vivo bench-top
uorescence microscopy confirmed the FITC signal (arrows, green) within the tumor after intravital staining (C; overlay with ex vivo nuclear counterstaining [blue], D).
439

anCIbtfast

iaFtint

ao

ctsaftrt

cwp

FF


4

nti-EGFR-antibodies (Figure 4). In most SW620 tumors,o such specific staining was observed in vivo with theLE or in cryosections after intravital staining (Figure 5).

HC was performed in a subset of tumors that had noteen visualized previously in vivo because intravital an-ibody labeling of the human-mouse xenograft inter-ered with IHC protocols in both paraffin-embeddednd cryopreserved tissue. In SW480 tumors, a specificignal was found in contrast to no such signal in SW620umors.

To corroborate the specificity of in vivo molecularmaging, 3 types of negative controls were performed inddition to imaging EGFR-negative SW620 tumors:irst, no fluorescent signal was seen in tumors beforehe application of FITC-labeled antibody. Second, thenjection of FITC-labeled isotype control antibody didot result in tumor-specific fluorescence. Third, a non-umor-associated specific fluorescence after the EGFR-

igure 5. Molecular imaging of SW620 tumors in live mice. No spe
ITC-labeled anti-EGFR antibody by CLE (A), IHC (B), or bench-top fluoresc
40

ntibody injection in tissue other than tumor was notbserved.However, metastases demonstrated a specific fluores-

ent pattern according to the cell type of the primaryumor. An EGFR-positive liver metastasis (after intra-plenic injection of SW480 cells) showed a strong signalfter in vivo labeling (Figure 6). The metastasis was onlyound by screening the macroscopically normal liver withhe confocal probe in vivo. A targeted biopsy from thisegion correlated this finding to the invasion front of theumor tissue ex vivo.

Quantification of EGFR Expression byFluorescence IntensityThe EGFR-specific signal from confocal images

aptured in vivo was graded from 0 to 3� (Figure 7A)ith a substantial interobserver agreement of 93%. Im-ortantly, the complete tumor tissue was screened for the

signal was observed in a typical SW620 tumor after the injection of
cific ence microscopy (C and D).

siaoF.estttwbSScct

cfs

iotislcfltp

Fibt large


ite showing the strongest fluorescence, which was thenncluded in the evaluation. If tumor tissue was assessedccordingly, a significantly higher EGFR expression wasbserved in SW480 tumors and SW620 tumors (P � .01,igure 7B). This discrimination was also significant (P �

002) if only the presence of any fluorescent signal wasvaluated (0 vs 1� to 3�). In detail, SW480 tumorstained positive in 21 (87.5%) and negative in 3 (12.5%)umors. SW620 showed a specific fluorescence in 7 (41.8%)umors, whereas no such signal was observed in 10 (58.8%)umors. Five of the 7 positive SW620 tumors showed onlyeak fluorescence (1�). Here, unspecific staining could note completely excluded. Mean fluorescence intensity andEM was 1.92 � 0.22 in SW480 tumors and 0.59 � 0.21 inW620 tumors (P � .001, Figure 7C). These results wereonfirmed in bench top fluorescence microscopy (fluores-ence intensity [mean � SEM], 2.60 � 0.22 in SW480

igure 6. Molecular imaging of an EGFR-positive liver metastasis. To minjection of FITC-labeled EGFR antibody, the invasion front of a liver metasiopsy (H&E stain) showed the tumor cells invading the healthy liver tissue (his EGFR-positive metastasis with a central clearance (C), in which the en

umors and 0.36 � 0.24 in SW620 tumors, P � .001). The s

orrelation between bench top fluorescence microscopy onrozen sections and in vivo confocal microscopy was sub-tantial (Spearman, r � 0.81).

EGFR Expression on HumanColorectal TissueTo anticipate a potential use of EGFR targeted

maging in colonoscopy with CLE, the topical applicationf labeled antibody was simulated by incubation of freshissue specimens from healthy colon or neoplastic lesionsn antibody solution. All specimens from neoplastic tis-ue showed a specific signal after EGFR staining (8/8esions), whereas healthy tissue and samples from unspe-ific colitis did not show specific fluorescence. Weakuorescence (1�) was found in only 2 samples of resec-ion specimens after ischemia. Differences between neo-lastic and non-neoplastic specimens were statistically

portal metastasis route, SW480 cells were injected into the spleen. Afterwas delineated against the unstained normal parenchyma (A). A targetedtailed inspection (insets) revealed the cytoplasmic fluorescence pattern of

d nuclei are located, as demonstrated by ex vivo H&E staining (D).

mic atasisB). De

ignificant, if only the absence or presence of any specific

441

flwc0(tw

vi

caovitltt

c

FAfl rs and


4

uorescence was evaluated (P � .003) or if differencesere assessed semiquantitatively (P � .007). Mean spe-

ific fluorescence of human tissue samples was 2.13 �.30 for neoplasia and 0.25 � 0.16 for normal mucosa

P � .002). IHC performed in a subset of surgical resec-ion specimens showed strong (3�) staining correlatingith CLE findings (Figure 8).

DiscussionIn this study, the first evidence is provided that in

ivo endomicroscopic molecular imaging of human CRC

igure 7. Quantification of specific in vivo fluorescence. Signal strength wsignificantly stronger expression was demonstrated in SW480 tumors (lig

uorescence in vivo after EGFR labeling was 0.59 � 0.21 in SW620 tumo

s possible by specifically targeting EGFR with a fluores- w

42

ently labeled antibody. This approach successfully visu-lized and differentiated human CRC tumor types basedn their EGFR expression in a rodent xenograft model inivo. To mimic a potential topical use of such molecularmaging in CLE in humans, labeled antibody was appliedo unfixed human tissue specimens, and successful mo-ecular imaging was achieved. Both intravenous applica-ion in rodents and topical application in fresh humanissue provided adequate contrast for confocal imaging.

Intravenous application of labeled anti-EGFR in theurrent study resulted in a quick delivery to the tumor

aded from absent (0) to weak (�), moderate (��), and strong (��) (A).lumns) than in SW620 tumors (dark columns), P � .01 (B). Mean specific1.92 � 0.22 in SW480 tumors (P � .0004). Bars indicate SEM (C).

as grht co

ithout the necessity to clear the tissue from adherent

tbecpaebtiusc

bflpip

eItsEeacsitsSlm

tom

Faws


issue or mucus. Thus, the complete tumor tissue coulde rapidly screened for the strongest EGFR expressionxploiting the full variable imaging plane depth of theonfocal probe. Topical application may prove moreractical for clinical use during colonoscopy and is prob-bly associated with less immunogenicity and adversevents, although the incubation period may still have toe optimized for clinical application.31 Spraying cathe-ers28 and even enemas26,38 have been advocated for top-cal delivery of targeted dyes. However, in most studiessing CLE so far, systemic contrast has been preferred touperficial staining usually obtained with topical appli-ation.28,30

The percentage of EGRF positivity in human CRC haseen reported with surprisingly high variability rangingrom 25% to 94%.5–7,39 Technical pitfalls might account ateast in part for this observation and might explain re-orts on the inconsistent influence of EGFR expression

n response to targeted therapy.40 – 42 Irregular EGFR ex-

igure 8. Molecular imaging in humans. Tissue samples from a neoplapplication during colonoscopy (A). CLE showed a specific signal withinas 0.25 � 0.16 for normal mucosa vs 2.12 � 0.30 for neoplasia (P � .howed EGFR expression in multiple tumor cells (arrows, D).

ression within a given tumor has been observed,10 and i

ven multiple random thin tissue sections for routineHC may therefore be prone to sampling error. Screeninghe tumor tissue in vivo by taking multiple optical biopsypecimens as performed in our approach may renderGFR patterns more comprehensively, although confocalndomicroscopy is still confined to superficial tumorreas by its limited imaging plane depth. Only 78% con-ordance between EGFR expression in primary CRC andynchronous metastases has been reported.39 This changen receptor expression could explain for the findinghat some SW620 tumors in the current trial demon-trated a specific fluorescence signal, whereas someW480 tumors were EGFR negative. In fact, these 2 cell

ines are derived from the primary tumor (SW480) andetastases (SW620) of the same patient.32

Interestingly, EGFR immunoreactivity has been foundo be inversely correlated to times of fixation and storagef tissue, and the fixative itself has modified EGFR im-unoreactivity.43 These problems might be tackled by an

sion were incubated in labeled EGFR antibody solution to mimic topicallesion (arrows, B). Mean specific fluorescence of human tissue samplesC). Bars indicate SEM. Histopathology confirmed malignancy, and IHC

stic lethe

002,

n vivo approach. IHC is not universally standardized44

443

niapcdTsatbflCb

nrpoesthbllmictammrws

lpiflSscwswnic(ad

bt

atflFeacsoaosi

lxtocwialepte


4

or is the antibody used or its target site,39,45,46 and IHCs therefore challenged by different new modalities suchs in situ hybridization or ligand binding assays.47 Re-roducibility among different trials has been furtheromplicated by the use of different arbitrary cut offs toefine EGFR positivity, ranging from 1% to 10% of cells.herefore, in the current study, images were captured at

tandardized instrument settings, and a semicontinuousssessment scheme was applied. An automated calcula-ion of the specific fluorescence intensity relative to aackground region may help to further objectify suchuorescence measurements at real time in the future.ellular staining patterns such as cytoplasmic or mem-rane activity10 could be reproduced in vivo.

In vivo molecular imaging, ie, the combination ofovel imaging technologies with targeted staining, hasapidly emerged with the perspective to significantly im-act on both basic research and clinical practice. Labelingf EGFR for molecular imaging in whole tissue has beenxamined before in multilayer constructs or slices ofquamous cell dysplasia ex vivo. In this study, a benchop confocal microscope was used, and permeability en-ancing agents were needed prior to application of anti-ody.48 In rodent models, tumors were macroscopically

ocated by luminescence 1–5 days after the injection ofight emitting live bacteria accumulating within the tu-

or tissue.49 Specific targeting of EGFR with a nearnfrared probe permitted intravital fluorescence micros-opy in breast cancer xenografts in rodents.50 However,hese approaches still relied on bulky bench top devicesnd staining protocols not appropriate for use in hu-ans. Our group has recently reported on the use of ainiaturized prototype confocal probe for in vivo fluo-

escent imaging of neuroendocrine tumors in rodentsith carboxyfluorescein-labeled, high-affinity ligands to

omatostatin receptors.35

In humans, both detection and characterization ofesions by molecular endoscopy have been studied. Ailot colonoscopy trial used labeled monoclonal antibod-

es against carcinoembryonic antigen with a prototypeuorescence endoscope for macroscopic tumor detection.pecific fluorescence after topical administration waseen in 19 of 25 carcinomas51 but were, however, notorrelated to in vivo microscopy. A different approachas followed in a recent trial in which a heptapeptide

equence identified from a phage library was conjugatedith fluorescein. The molecular target of this sequence isot yet elucidated, but, after topical administration dur-

ng colonoscopy, dysplastic lesions were visualized byonfocal microscopy with high sensitivity and specificity81% and 82%, respectively).31 Inherently, the problem of

receptor- or antigen-negative tumor has not been ad-
ressed in the trials mentioned above. However, an anti-
44

ody “cocktail” targeting multiple receptors52 or mul-ichannel imaging may prove valuable in the future.

Many studies on molecular imaging used excitationnd emission wavelengths in the near infrared spectrumo achieve deeper tissue imaging. In the current trial,uorescein-based detection was preferred because theIVE1 probe used has identical optical properties as thendomicroscope registered for endoscopy in patients,nd fluorescein has been in routine clinical use for de-ades with a favorable safety profile.53,54 In addition,pecific binding of anti-EGFR-antibodies to tumor tissueccurred rapidly within a time frame still acceptable forcolonoscopy setting. Although meticulous evaluation

f potential clinical applications is mandated, this en-ures that an immediate transfer from bench to bedsides feasible from a technical point of view even today.

In summary, the current trial demonstrates that mo-ecular imaging is feasible in vivo by targeting EGFR in aenograft model of human CRC and ex vivo on humanissue. Discrimination of tumor cells was possible basedn their molecular signature. As a perspective for furtherlinical use, this approach could ideally be combinedith a high-resolution macroscopic fluorescent imaging

nstrument. Such a combined approach would then offerred-flag technique for macroscopic detection (a “mo-

ecular chromoendoscopy”) and simultaneous confocalndomicroscopy to detect immediately the expressionatterns of a lesion during ongoing endoscopy. With this,argeted individualized therapies could be based on annhanced diagnosis and accurate selection of patients.

References

1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CACancer J Clin 2008;58:71–96.

2. Selby JV, Friedman GD, Quesenberry CP Jr, et al. A case-controlstudy of screening sigmoidoscopy and mortality from colorectalcancer. N Engl J Med 1992;326:653–657.

3. Levin B, Lieberman DA, McFarland B, et al. Screening and sur-veillance for the early detection of colorectal cancer and adeno-matous polyps, 2008: a joint guideline from the American CancerSociety, the US Multi-Society Task Force on Colorectal Cancer,and the American College of Radiology. Gastroenterology 2008;134:1570–1595.

4. Herbst RS, Shin DM. Monoclonal antibodies to target epidermalgrowth factor receptor-positive tumors: a new paradigm for cancertherapy. Cancer 2002;94:1593–1611.

5. Ciardiello F, Tortora G. EGFR antagonists in cancer treatment.N Engl J Med 2008;358:1160–1174.

6. Salomon DS, Brandt R, Ciardiello F, et al. Epidermal growthfactor-related peptides and their receptors in human malignan-cies. Crit Rev Oncol Hematol 1995;19:183–232.

7. Normanno N, De Luca A, Bianco C, et al. Epidermal growth factorreceptor (EGFR) signaling in cancer. Gene 2006;366:2–16.

8. Nicholson RI, Gee JM, Harper ME. EGFR and cancer prognosis.Eur J Cancer 2001;37(Suppl 4):S9–S15.

9. Abd El All HS, Mishriky AM, Mohamed FA. Epidermal growth factorreceptor in colorectal carcinoma: correlation with clinico-patho-
logical prognostic factors. Colorectal Dis 2008;10:170–178.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4


0. Goldstein NS, Armin M. Epidermal growth factor receptor immuno-histochemical reactivity in patients with American Joint Committeeon Cancer Stage IV colon adenocarcinoma: implications for a stan-dardized scoring system. Cancer 2001;92:1331–1346.

1. Grunwald V, Hidalgo M. Developing inhibitors of the epidermalgrowth factor receptor for cancer treatment. J Natl Cancer Inst2003;95:851–867.

2. Galizia G, Lieto E, Orditura M, et al. Epidermal growth factorreceptor (EGFR) expression is associated with a worse prognosisin gastric cancer patients undergoing curative surgery. WorldJ Surg 2007;31:1458–1468.

3. Saltz LB, Meropol NJ, Loehrer PJ Sr, et al. Phase II trial ofcetuximab in patients with refractory colorectal cancer that ex-presses the epidermal growth factor receptor. J Clin Oncol 2004;22:1201–1208.

4. Cunningham D, Humblet Y, Siena S, et al. Cetuximab mono-therapy and cetuximab plus irinotecan in irinotecan-refractorymetastatic colorectal cancer. N Engl J Med 2004;351:337–345.

5. Jonker DJ, O’Callaghan CJ, Karapetis CS, et al. Cetuximab for thetreatment of colorectal cancer. N Engl J Med 2007;357:2040–2048.

6. Lievre A, Bachet JB, Boige V, et al. KRAS mutations as an indepen-dent prognostic factor in patients with advanced colorectal cancertreated with cetuximab. J Clin Oncol 2008;26:374–379.

7. De Roock W, Piessevaux H, De Schutter J, et al. KRAS wild-typestate predicts survival and is associated to early radiologicalresponse in metastatic colorectal cancer treated with cetuximab.Ann Oncol 2008;19:508–515.

8. Italiano A, Follana P, Caroli FX, et al. Cetuximab shows activity incolorectal cancer patients with tumors for which FISH analysisdoes not detect an increase in EGFR gene copy number. Ann SurgOncol 2008;15:649–654.

9. Chung KY, Shia J, Kemeny NE, et al. Cetuximab shows activity incolorectal cancer patients with tumors that do not express theepidermal growth factor receptor by immunohistochemistry. J ClinOncol 2005;23:1803–1810.

0. Schrag D. The price tag on progress—chemotherapy for colorec-tal cancer. N Engl J Med 2004;351:317–319.

1. Yoshida S, Tanaka S, Hirata M, et al. Optical biopsy of GI lesionsby reflectance-type laser-scanning confocal microscopy. Gastroi-ntest Endosc 2007;66:144–149.

2. Nakao M, Yoshida S, Tanaka S, et al. Optical biopsy of earlygastroesophageal cancer by catheter-based reflectance-typelaser-scanning confocal microscopy. J Biomed Opt 2008;13:054043.

3. Curvers WL, Singh R, Song LM, et al. Endoscopic tri-modal imag-ing for detection of early neoplasia in Barrett’s oesophagus: amulti-centre feasibility study using high-resolution endoscopy,autofluorescence imaging and narrow band imaging incorporatedin one endoscopy system. Gut 2008;57:167–172.

4. van den Broek FJ, Fockens P, van Eeden S, et al. Endoscopictri-modal imaging for surveillance in ulcerative colitis: random-ised comparison of high-resolution endoscopy and autofluo-rescence imaging for neoplasia detection; and evaluation ofnarrow-band imaging for classification of lesions. Gut 2008;57:1083–1089.

5. Matsuda T, Saito Y, Fu KI, et al. Does autofluorescence imagingvideoendoscopy system improve the colonoscopic polyp detec-tion rate? A pilot study. Am J Gastroenterol 2008;103:1926–1932.

6. Mayinger B, Neumann F, Kastner C, et al. Early detection ofpremalignant conditions in the colon by fluorescence endoscopyusing local sensitization with hexaminolevulinate. Endoscopy
2008;40:106–109.
7. Mayinger B, Neidhardt S, Reh H, et al. Fluorescence induced with5-aminolevulinic acid for the endoscopic detection and follow-upof esophageal lesions. Gastrointest Endosc 2001;54:572–578.

8. Kiesslich R, Burg J, Vieth M, et al. Confocal laser endoscopy fordiagnosing intraepithelial neoplasias and colorectal cancer invivo. Gastroenterology 2004;127:706–713.

9. Kiesslich R, Gossner L, Goetz M, et al. In vivo histology ofBarrett’s esophagus and associated neoplasia by confocallaser endomicroscopy. Clin Gastroenterol Hepatol 2006;4:979–987.

0. Polglase AL, McLaren WJ, Skinner SA, et al. A fluorescenceconfocal endomicroscope for in vivo microscopy of the upper- andthe lower-GI tract. Gastrointest Endosc 2005;62:686–695.

1. Hsiung PL, Hardy J, Friedland S, et al. Detection of colonicdysplasia in vivo using a targeted heptapeptide and confocalmicroendoscopy. Nat Med 2008;14:454–458.

2. Leibovitz A, Stinson JC, McCombs WB III, et al. Classification ofhuman colorectal adenocarcinoma cell lines. Cancer Res 1976;36:4562–4569.

3. Price JE, Daniels LM, Campbell DE, et al. Organ distribution ofexperimental metastases of a human colorectal carcinoma in-jected in nude mice. Clin Exp Metastasis 1989;7:55–68.

4. Flatmark K, Maelandsmo GM, Martinsen M, et al. Twelve colo-rectal cancer cell lines exhibit highly variable growth and meta-static capacities in an orthotopic model in nude mice. Eur JCancer 2004;40:1593–1598.

5. Goetz M, Fottner C, Schirrmacher E, et al. In vivo confocal real-time mini-microscopy in animal models of human inflammatoryand neoplastic diseases. Endoscopy 2007;39:350–356.

6. Goetz M, Memadathil B, Biesterfeld S, et al. In vivo subsurfacemorphological and functional cellular and subcellular imaging ofthe gastrointestinal tract with confocal mini-microscopy. World JGastroenterol 2007;13:2160–2165.

7. Kiesslich R, Goetz M, Angus EM, et al. Identification of epithelialgaps in human small and large intestine by confocal endomicros-copy. Gastroenterology 2007;133:1769–1778.

8. Endlicher E, Gelbmann CM, Knuchel R, et al. Hexaminolevulinate-induced fluorescence endoscopy in patients with rectal adenomaand cancer: a pilot study. Gastrointest Endosc 2004;60:449–454.

9. Bibeau F, Boissiere-Michot F, Sabourin JC, et al. Assessment ofepidermal growth factor receptor (EGFR) expression in primarycolorectal carcinomas and their related metastases on tissuesections and tissue microarray. Virchows Arch 2006;449:281–287.

0. Erlichman C, Sargent DJ. New treatment options for colorectalcancer. N Engl J Med 2004;351:391–392.

1. Valentini AM, Pirrelli M, Caruso ML. EGFR-targeted therapy incolorectal cancer: does immunohistochemistry deserve a role inpredicting the response to cetuximab? Curr Opin Mol Ther 2008;10:124–131.

2. Saltz L. Epidermal growth factor receptor-negative colorectal can-cer: is there truly such an entity? Clin Colorectal Cancer 2005;5(Suppl 2):S98–S100.

3. Atkins D, Reiffen KA, Tegtmeier CL, et al. Immunohistochemicaldetection of EGFR in paraffin-embedded tumor tissues: variationin staining intensity due to choice of fixative and storage time oftissue sections. J Histochem Cytochem 2004;52:893–901.

4. Derecskei K, Moldvay J, Bogos K, et al. Protocol modificationsinfluence the result of EGF receptor immunodetection by EGFRpharmDx in paraffin-embedded cancer tissues. Pathol Oncol Res2006;12:243–246.

5. Bhargava R, Chen B, Klimstra DS, et al. Comparison of two
antibodies for immunohistochemical evaluation of epidermal
445

4

4

4

4

5

5

5

5

5

R

MLm

A

BG

A

C

tKgc

F


4

growth factor receptor expression in colorectal carcinomas,adenomas, and normal mucosa. Cancer 2006;106:1857–1862.

6. Penault-Llorca F, Cayre A, Arnould L, et al. Is there an immu-nohistochemical technique definitively valid in epidermalgrowth factor receptor assessment? Oncol Rep 2006;16:1173–1179.

7. Moroni M, Sartore-Bianchi A, Veronese S, et al. EGFR FISH incolorectal cancer: what is the current reality? Lancet Oncol 2008;9:402–403.

8. Hsu ER, Anslyn EV, Dharmawardhane S, et al. A far-red fluores-cent contrast agent to image epidermal growth factor receptorexpression. Photochem Photobiol 2004;79:272–279.

9. Yu YA, Shabahang S, Timiryasova TM, et al. Visualization oftumors and metastases in live animals with bacteria and vacciniavirus encoding light-emitting proteins. Nat Biotechnol 2004;22:313–320.

0. Ke S, Wen X, Gurfinkel M, et al. Near-infrared optical imaging ofepidermal growth factor receptor in breast cancer xenografts.Cancer Res 2003;63:7870–7875.

1. Keller R, Winde G, Terpe HJ, et al. Fluorescence endoscopy usinga fluorescein-labeled monoclonal antibody against carcinoembry-onic antigen in patients with colorectal carcinoma and adenoma.Endoscopy 2002;34:801–807.

2. Barrett T, Koyama Y, Hama Y, et al. In vivo diagnosis of epidermalgrowth factor receptor expression using molecular imaging with acocktail of optically labeled monoclonal antibodies. Clin CancerRes 2007;13:6639–6648.

3. Kiesslich R, Hurlstone DP, Dunbar K, et al. Confocal Laser En-domicroscopy is an effective and safe tool in GI-Endoscopy.
Gastroenterology 2008;134:A191. f
46

4. Pacurariu RI. Low incidence of side effects following intravenousfluorescein angiography. Ann Ophthalmol 1982;14:32–36.

Received June 6, 2009. Accepted October 14, 2009.

eprint requestsAddress requests for reprints to: Martin Goetz, MD, PhD, I.edizinische Klinik und Poliklinik, Universitätsmedizin Mainz,

angenbeckstr. 1, 55131 Mainz, Germany. e-mail: [email protected]; fax: (49) 6131-17-5552.

cknowledgmentsThe authors thank Kathrin Kuhr, Institut für Medizinische

iometrie, Epidemiologie und Informatik, Universitätsmedizin Mainz,ermany, for help with the statistical analysis.Aspects of this trial are part of the doctor of medicine theses of

.Z. and S.F.

onflicts of interestThe authors disclose the following: Peter Delaney is director of

echnology at Optiscan Pty, Ltd, Notting Hill, Victoria, Australia. Ralfiesslich and Markus F. Neurath report having received researchrants from Pentax, Tokio, Japan. The remaining authors disclose noonflicts.

undingSupported by a government grant of the “Stiftung Rheinland-Pfalz

ür Innovation” (961-38 62 61/821-Projekt 821) (to M.G.).

mailto:[email protected]

mailto:[email protected]

imaging and advanced technology - mitweb.mit.edu/hst.035/readingsupplement/04_28_coron... · in the...

Documents