Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/5877907

Assessmentofindocyaninegreen-labeledcetuximabtodetectxenograftedheadandneckcancercelllines

ARTICLEinOTOLARYNGOLOGYHEADANDNECKSURGERY·DECEMBER2007

ImpactFactor:2.02·DOI:10.1016/j.otohns.2007.06.736·Source:PubMed

CITATIONS

19

READS

39

7AUTHORS,INCLUDING:

JohnPGleysteen

OregonHealthandScienceUniversity

9PUBLICATIONS178CITATIONS

SEEPROFILE

EbenL.Rosenthal

StanfordUniversity

239PUBLICATIONS3,253CITATIONS

SEEPROFILE

Availablefrom:EbenL.Rosenthal

Retrievedon:05February2016

Otolaryngology–Head and Neck Surgery (2007) 137, 729-734

ORIGINAL RESEARCH—HEAD AND NECK CANCER

Assessment of indocyanine green–labeled

cetuximab to detect xenografted head

and neck cancer cell lines

Kirk P. Withrow, MD, John P. Gleysteen, BS, Ahmad Safavy, PhD,Joni Skipper, MD, Renee A. Desmond, DVM, PhD, Kurt Zinn, DVM, PhD,

and Eben L. Rosenthal, MD, Birmingham, AL

OBJECTIVE: The aim of this study is to determine the efficacyof indocyanine green (ICG) conjugated to antiepidermal growthfactor receptor antibody (cetuximab) to image head and neckcancer.STUDY DESIGN: Mice (n � 3) were injected with unconju-gated ICG and imaged at 100-second intervals for a total of 1000seconds to assess imaging characteristics. Mice (n � 10) xe-nografted with SCC-1 cells were then systemically injected withcetuximab conjugated to indocyanine green and imaged overa 72-hour period. To assess the sensitivity and specificity, xe-nografted tumors underwent subtotal resections and then wereassessed for residual disease by fluorescence stereomicroscopy andconfirmed by histology.RESULTS: Tumors demonstrated excellent fluorescence 24hours after injection of cetuximab-ICG. There was a direct rela-tionship between fluorescence and the given dose of cetuximab-ICG. Following subtotal resection, we found fluorescence corre-lated with a sensitivity of 78.4% and specificity of 96%.CONCLUSIONS: This study provides evidence that supportsfurther preclinical investigation of cetuximab in the evaluation ofsurgical margins, but linkage to ICG lacks the sensitivity for use ina clinical setting.© 2007 American Academy of Otolaryngology–Head and NeckSurgery Foundation. All rights reserved.

Determination of adequate surgical margins in head andneck cancer oncologic resections relies on gross inspec-

tion and palpation by the surgeon. Unfortunately, these sub-jective methods result in positive or close surgical margins inalmost 40% of cases.1,2 Because positive margins predict poorsurvival,3 more accurate identification of intraoperative mar-gins is likely to improve patient outcomes. The identification ofcancer-specific optical contrast agents could provide surgeonswith real-time, objective information about the presence ofresidual disease, thereby reducing the number of patients withhistologically positive surgical margins. Furthermore, accurateidentification of tumor margins may decrease postoperative

Received October 30, 2006; revised April 17, 2007; accepted June 26,

2007.

0194-5998/$32.00 © 2007 American Academy of Otolaryngology–Head and Necdoi:10.1016/j.otohns.2007.06.736

morbidity by minimizing the resection of uninvolved tissues.Optical imaging has revolutionized molecular biology in thepast 5 to 10 years and is showing considerable promise for usein cancer detection.4-6

Fluorescent compounds (fluorophores) absorb light at onewavelength and emit at another wavelength. Fluorescentagents have poor tissue penetration for whole-body imaging,but this may not be a significant problem in the surgical settingwhere the tumor is exposed. The potential clinical utility of afluorescent agent will depend on the specificity of the agentand toxicity of the cancer-specific probe. Indocyanine green(ICG) is a tricarbocyanine dye and is currently the only near-infrared (NIR) fluorophore available for human use. It hasproven to be medically useful because its peak fluorescence(800-810 nm in circulation) is higher than the auto-fluores-cence exhibited by normal tissue. ICG has been successfullyused to measure hepatic function and cardiac output, as well asto perform ophthalmic angiography. Recently, ICG has beenunder investigation as a potential optical contrast agent forsentinel lymph node mapping.7,8 It has also been conjugated toantibodies for in vitro detection of tumor markers in gastroin-testinal cancers.9,10

Although fluorescent stereomicroscopy has been appliedto image fluorescent agents in the operating room, uncon-jugated ICG is not targeted to the tumor and remains non-specific.6 Unfortunately, the use of tumor targeting agentslinked to fluorescent probes has not yet been performed inhumans. We sought to determine if ICG conjugated to anFDA-approved anti-EGFR antibody (cetuximab) could beused to detect head and neck cancer in a preclinical model.The use of an anti-EGFR antibody was deemed appropriatebecause 80% to 90% of head and neck cancers overexpressEGFR.11 We have previously shown that cetuximab conju-gated to a non–FDA-approved fluorescent dye could beused to image xenografted head and neck tumors.4 Clinicaltranslation of this fluorescent imaging would be greatly

k Surgery Foundation. All rights reserved.

730 Otolaryngology–Head and Neck Surgery, Vol 137, No 5, November 2007

facilitated if both the targeting agent (cetuximab) and thefluorophore were approved for human use. There are nocurrent studies assessing the utility of systemically injectedICG conjugated to an antibody.

MATERIALS AND METHODS

ReagentsTo determine baseline fluorescent and pharmacologicalproperties of ICG in our murine model, unlabeled ICG(IC-Green, Akorn, Inc., Somerset, MD) was utilized. ICG isa sterile, water-soluble, tricarbocyanine dye with a peakspectral absorption at 800 to 810 nm in plasma or blood; itsapproximate molecular weight is 774.96. We used cetux-imab (Erbitux, ImClone Systems Incorporated, New York,NY), a recombinant, human/mouse chimeric monoclonalantibody that binds specifically to the extracellular domainof the human EGFR. Cetuximab is composed of the Fvregions of a murine anti-EGFR antibody with human IgG1heavy and kappa light chain constant regions and has anapproximate molecular weight of 152 kDa. ICG (ICG,Dojindo Molecular Technologies, Gaithersburg, MD) wasconjugated to cetuximab for use as the NIR fluorescentmarker.

Synthesis of ICG-Cetuximab ConjugateThe antibody solution was buffer-exchanged with a 100mM, pH 9.5 phosphate-buffered saline (PBS) containing5 mM ethylenediamine tetraacetic acid in a Microcon 50 kmolecular weight cut-off concentrator (Amicon, Bedford,MA). The cetuximab concentration was set to about 5mg/mL by volume adjustment. The solution was cooled inan ice bath and a 2-mg/mL solution of the ICG-NHS indimethylsulfoxide (DMSO) was added to a final ICG:cetux-imab molar ratio of 10. The reaction mixture was incubatedat this temperature for 1 hour, after which time the greenhomogeneous solution was loaded into a Sephadex G-25PD-10 column (Amersham Pharmacia AB, Uppsala, Swe-den) pre-equilibrated with 1� Dulbecco’s PBS (DPBS,Mediatech, Herndon, VA). The column was eluted withDPBS and 1-mL fractions were collected. Fractions 3 and 4containing the green-colored conjugate were pooled. Theresulting conjugation number (ICG:cetuximab molar ratio)was determined by MALDI/TOF mass spectrometry andfound to be approximately 1:1.

Murine Human Tumor Xenograft Flank

ModelSCID male mice, ages 4 to 6 weeks (Charles River Laborato-ries, Wilmington, MA), were obtained and housed in accor-dance with the institution’s IACUC (Institutional Animal Careand Use Committee) guidelines. All experiments were con-ducted and the animals euthanized according to our institu-tion’s IACUC guidelines. SCC-1 cells were obtained (Thomas

Carey, University of Michigan, Ann Arbor, MI) and main-

tained in DMEM containing 10% fetal bovine serum andsupplemented with L-glutamine, penicillin, and streptomycinand incubated at 37°C in 5% CO2. SCC-1 cells (2 � 106) wereinjected into the flank of SCID mice. The tumors were imagedwhen they reached 5 mm to 10 mm in size.

ImagingHuman tumor xenografts were imaged using a custom-builtLeica fluorescent stereomicroscope (Leica MZFL3 Stereoresearch microscope; Leica Microsystems, Bannockburn,IL) fitted with an ICG filter and an ORCA ER charge-coupled device (CCD) camera (Hamamatsu, Bridgewater,NJ) to allow for real-time imaging of fluorescence. ICGfilter (filter set 41030, Chroma Technology Corp., Rocking-ham, VT) has an excitation spectrum between 750 and 800nm and an emission spectrum of 820 to 875 nm. Gross,brightfield, and fluorescent images were obtained for eachresection.

Calculating ICG Dose/Time CurvesUnconjugated ICG was reconstituted in accordance with themanufacturer’s protocol immediately prior to injection. Ini-tial images were obtained to ensure that no autofluorescencewas present. Three doses of plain ICG (50 ug, 100 ug, and150 ug) were injected systemically via tail vein. The mice(n � 3) were then imaged initially and at 100-second inter-vals. The last time point was 1000 seconds postinjection.

The cetuximab-ICG conjugate was delivered to micebearing xenografts (n � 2 per group) at different doses: 50,70, 90, 140, 350 ug prior to imaging. Mice were imaged at0, 2, 4, 24, and 72 hours.

Sensitivity and Specificity ModelSubtotal surgical resections were performed on euthanizedmice (n � 8) approximately 24 hours after systemic injectionof cetuximab-ICG. Fluorescent, stereomicroscopic imagingwas performed before and after each subtotal resection andbiopsy to determine if any residual fluorescence was present.Using a microcup biopsy forceps (Hartmann Herzfeld cup-shaped alligator forceps, 2 mm cup size, Medtronic, Jackson-

Figure 1 Unconjugated ICG has short in vivo time course.Fluorescent stereomicroscopy detected unconjugated ICG (50,100, and 150 ug) in xenografted tumors. Luminosity peaks at 100

seconds after systemic injection.

731Withrow et al Assessment of indocyanine green–labeled . . .

ville, FL), areas of persistent fluorescence were sampled underfluorescent guidance using the ICG filter. This process wasrepeated until a nonfluorescent background was obtained, atwhich point an equal number of biopsies were taken from thepreviously fluorescent areas within the tumor bed. Serial his-tological sections and routine H&E staining were performedon each biopsy specimen to assess the presence of any residualtumor. A total of 60 biopsies (30 fluorescent and 30 nonfluo-rescent) were available for evaluation.

Figure 2 Cetuximab-ICG can be detected in vivo at 24 hours.Luminosity was detected in xenografted head and neck tumors(5-10 mm in size) by fluorescent stereomicroscopy after systemicinjection of cetuximab-ICG.

Figure 3 Mock tumor resections were performed to determine(A,D,G), with stereomicroscopy (B,E,H), and with near-infrared fl

underwent subtotal resections (G,H,I).

Statistical AnalysisSensitivity, specificity, positive predictive value, negativepredictive value, and confidence limits were calculated foreach comparison using the normal approximation of thestandard error for proportions. A 95% confidence interval(CI) for each proportion was calculated according to theefficient-score method (corrected for continuity). A poweranalysis was based on testing the hypothesis that the sensi-tivity of the fluorescence procedure exceeds a minimallyacceptable level of 0.75 (null hypothesis). A sample size of30 samples would provide over 0.90 power to detect a truesensitivity of 0.95 over the null hypothesis sensitivity of0.75, based on a level of significance of � � 0.05.

RESULTS

To determine the imaging properties of plain ICG in ourmodel system as compared to the cetuximab-conjugatedICG, unconjugated ICG was administered into tumor-bear-ing mice and imaged over time. The luminosity increasedwhen the dose was increased from 50 ug to 100 ug, but asignificant change in luminosity was not observed when thedose was increased from 100 ug to 150 ug (Fig 1). Peakluminosity occurred at approximately 100 seconds aftersystemic injection. At 1 hour after injection, luminosity inthe tumors returned to background levels.

ivity and specificity. Tumors were imaged using a digital camerance (C,F,I). After removal of the murine skin (D,E,F), the tumors

sensituoresce

732 Otolaryngology–Head and Neck Surgery, Vol 137, No 5, November 2007

Cetuximab-ICG was systemically administered to SCIDmice (n � 2 per group) bearing SCC-1 subcutaneous xeno-grafts at multiple doses (50, 70, 90, 140, 350 ug) andimaged over time using a fluorescent filter with an emissionrange of 820 to 875 nm. A significant improvement inluminosity was noted when the dose of cetuximab-ICGadministered was escalated above 90 ug (Fig 2). Animalswere imaged out to 72 hours with no significant degenera-tion of luminosity.

To assess the sensitivity and specificity of cetuximab-ICGin detecting histologically malignant surgical margins in apreclinical model, subtotal resection of flank tumors (n � 8)was performed 24 hours after systemic injection of 350 ug ofthe cetuximab-ICG conjugate. Prior to resection, fluorescent

Figure 4 Biopsies were obtained from the tumor bed after near-fluorescent (B,D,F,H) images were obtained before and after each w

2-mm cupped forceps until negative background achieved. Additional bi

images were obtained (Fig 3). Real-time fluorescent imagingwas used to guide multiple biopsies using 2-mm cupped for-ceps from the wound bed after near-total resection (Fig 4).Serial sectioning and H&E staining of biopsy specimens wereperformed to confirm that the presence or absence of fluores-cence correlated with histologically malignant tissue. Areas ofpositive fluorescence within the wound could predict the pres-ence of tumor in 29 of 30 biopsies, a specificity of 96.0%(Table 1). When all fluorescent tissue had been removed,biopsies were obtained from areas of the wound bed thatdemonstrated nonfluorescence. Disease was found in 8 of 30biopsies taken from nonfluorescent areas, a sensitivity of78.4% (Table 1). Representative images of true- and false-positive biopsies as well as true- and false-negative biopsies

section. Brightfield stereomicroscopy (A,C,E,G) and near-infraredbed biopsy. Areas of fluorescence (arrows) underwent biopsy with

total reound

opsies of the fluorescent negative wound bed were then obtained.

733Withrow et al Assessment of indocyanine green–labeled . . .

are shown (Fig 5). Examination of serial histological section-ing of false-negative biopsy specimens revealed an averagesize of 3.4 mm2 (0.6-9.0 mm2).

DISCUSSION

Intraoperative assessment of tumor margins is currentlydone by gross examination of the tumor bed and frozen-section histology. Although frozen sections can accuratelyidentify residual disease, only a fraction of the wound bedcan be assessed by this technique. Furthermore, frozen-section processing takes over 30 to 40 minutes to process. Itis possible that real-time intraoperative detection of tumors

Figure 5 Representative biopsies. Biopsies obtained from arenegative (A), while those with tumor present were false negatives (

Table 1

Specificity and sensitivity of ICG labeled cetuximab

for xenografted tumors

Lower95% CI

Upper95% CI

Specificity96.0% 76.0% 100.0%

Sensitivity78.4% 61.3% 89.6%

Positive predictive value97.0%

Negative predictive value73.3%

CI, confidence interval.

tumor were considered true positive (C), while those that did not were

would improve the ability of surgeons to accurately identifymicroscopic residual disease. Our results suggest that atdoses approaching therapeutic levels for animal studies(�250 ug) tumors were well visualized by stereomicros-copy. In mock surgical resections, a sensitivity of 78% andspecificity of 96% could be achieved. The detection ofresidual disease after an intentional subtotal resection of axenografted flank tumor was used to approximate the de-tection of microscopic disease at the margins after a headand neck cancer ablative procedure in the clinical setting.Unfortunately, modeling oncologic resections in a murinemodel has inherent limitations. Although the tumors in miceare significantly smaller, less invasive, and more clearlyvisible than in the clinical setting, the model helps to quan-tify this technique and allow improvements in the design ofthe fluorescent probe and targeting molecule in preclinicalmodels. It is very unlikely that the results will translatedirectly to the clinical environment.

Because of poor tissue penetration of fluorescent agents,the potential of optical imaging has been largely neglectedby the imaging community. Fluorescent optical imaging iscurrently being investigated by surgical oncologists for usein primary resections in neurosurgery5 and mapping ofregional metastasis in other disease states.6 However, theseagents remain nonspecific for cancer. The development ofantibodies designed to selectively bind to tumor markers hasthe potential to deliver and concentrate near-infrared flu-orophores in tumors and enhance visualization in the oper-ating room or clinic. Successful application of ICG as anoptical agent for detection of cancer in patients has been metwith multiple obstacles. Most significantly, systemic injec-

onfluorescence with no evidence of tumor were considered trueow marks tumor). Specimens from fluorescent areas that contained

as of nB; arr

considered false positive (D). Bar � 0.5 mm.

734 Otolaryngology–Head and Neck Surgery, Vol 137, No 5, November 2007

tion of unconjugated ICG is cleared within minutes from thecirculatory system and requires repeated dosing for clinicaluse. It has been most successfully used to map the circula-tory, lymphatic, or biliary system. ICG has been success-fully conjugated to anticarcinoembryonic antigen to detectgastric cancer in biopsy specimens ex vivo.12 Unlike thisstudy and others like it,13 we systemically administered theICG-antibody conjugate to detect tumor xenografts in vivo.Because tumor luminosity persists beyond the initial imag-ing period of 1 hour when unconjugated ICG is no longervisible, it is unlikely that the ICG dissociates from cetux-imab in circulation. Furthermore, imaging of this conjugateover the course of 72 hours failed to demonstrate a signif-icant degradation of fluorescence.4

We have previously demonstrated that cetuximab (whichis FDA approved in head and neck cancer) conjugated toCy5.5 (GE Healthcare, Piscataway, NJ) can accurately im-age head and neck cancer xenografts.4 In these studies wedemonstrate that a nonspecific antibody conjugated to thefluorescent dye is not concentrated within the tumor. Un-fortunately, Cy5.5 is not currently FDA approved for humanuse. To facilitate the clinical translation of this technique,we sought to determine if an FDA-approved fluorophorecould be used with similar results. Unlike the cetuximab-Cy5.5 conjugate, which could be well visualized at a dose of50 ug, the cetuximab-ICG agent required significantlyhigher doses (350 ug) to achieve similar luminosity inten-sities. This is consistent with mass spectroscopy analysis ofthe conjugated antibody, which demonstrated only one mol-ecule of bound ICG per molecule of cetuximab as opposedto five molecules of the Cy5.5 fluorophore bound to eachmolecule of antibody (data not shown).

The clinical utility of this model will depend upon improv-ing the sensitivity and specificity. Furthermore, animal modelswill not be able to predict background expression on mucosalsurfaces. Although we have previously demonstrated that en-grafted human skin has minimal fluorescence after injection offluorescently labeled cetuximab, the background fluorescenceof mucosal surfaces remains unknown. Ultimately, clinicaltrials will be required to determine if high EGFR expression indysplastic mucosa that commonly surrounds invasive head andneck cancers translates into higher background fluorescence.

This data supports the need for further investigation of cancer-specific contrast agent to guide surgical therapy or detect diseasein the clinical setting. Although clinical translation will be facili-tated by the use of fluorescence agents currently approved for usein humans, other fluorophores with improved optical characteris-tics may result in better sensitivity and specificity.

AUTHOR INFORMATION

From the Department of Surgery, Division of Otolaryngology–Head andNeck Surgery (Drs Withrow, Gleysteen, Skipper, and Rosenthal), Divisionof Radiation Biology, Department of Radiation Oncology (Dr Safavy), andDepartment of Medicine (Drs Desmond and Zinn), University of Alabama

at Birmingham.

Corresponding author: Eben L. Rosenthal, MD, Division of Otolaryngol-ogy, University of Alabama at Birmingham, BDB Suite 563, 1808 7thAvenue South, Birmingham, AL 35294-0012.

E-mail address: oto@uab.edu.

AUTHOR CONTRIBUTIONS

Eben Rosenthal, study design, writer; Kirk Withrow, data collection,writer; Ahmad Safavy, experimental studies; John Gleysteen, data anal-ysis, writer; Joni Skipper, data collection; Renee Desmond, statisticalanalysis; Kurt Zinn, study design.

FINANCIAL DISCLOSURE

This work was supported by a grant from the American Cancer Society(RSG-06-1006-01-CCE).

REFERENCES

1. Ravasz LA, Slootweg PJ, Hordijk GJ, et al. The status of the resectionmargin as a prognostic factor in the treatment of head and neckcarcinoma. J Craniomaxillofac Surg 1991;19:314–8.

2. O’Brien CJ, McNeil EB, McMahon JD, et al. Significance of clinicalstage, extent of surgery, and pathologic findings in metastatic cutaneoussquamous carcinoma of the parotid gland. Head Neck 2002;24:417–22.

3. Bernier J, Vermorken JB, Koch WM. Adjuvant therapy in patientswith resected poor-risk head and neck cancer. J Clin Oncol 2006;24:2629–35.

4. Rosenthal EL, Kulbersh BD, Duncan RD, et al. In vivo detection ofhead and neck cancer orthotopic xenografts by immunofluorescence.Laryngoscope 2006;116:1636–41.

5. Stummer W, Pichlmeier U, Meinel T, et al. Fluorescence-guidedsurgery with 5-aminolevulinic acid for resection of malignant glioma:a randomised controlled multicentre phase III trial. Lancet Oncol2006;7:392–401.

6. Tanaka E, Choi HS, Fujii H, et al. Image-guided oncologic surgeryusing invisible light: completed pre-clinical development for sentinellymph node mappingl. Ann Surg Oncol 2006;13:1671–81.

7. Parungo CP, Ohnishi S, De Grand AM, et al. In vivo optical imagingof pleural space drainage to lymph nodes of prognostic significance.Ann Surg Oncol 2004;11:1085–92.

8. Parungo CP, Ohnishi S, Kim SW, et al. Intraoperative identification ofesophageal sentinel lymph nodes with near-infrared fluorescence im-aging. J Thorac Cardiovasc Surg 2005;129:844–50.

9. Muguruma N, Ito S, Hayashi S, et al. Antibodies labeled with fluores-cence-agent excitable by infrared rays. J Gastroenterol 1998;33:467–71.

10. Katayama A, Bandoh N, Kishibe K, et al. Expressions of matrixmetalloproteinases in early-stage oral squamous cell carcinoma aspredictive indicators for tumor metastases and prognosis. Clin CancerRes 2004;10:634–40.

11. Ang KK, Andratschke NH, Milas L. Epidermal growth factor receptorand response of head-and-neck carcinoma to therapy. Int J RadiatOncol Biol Phys 2004;58:959–65.

12. Ito S, Muguruma N, Kusaka Y, et al. Detection of human gastriccancer in resected specimens using a novel infrared fluorescent anti-human carcinoembryonic antigen antibody with an infrared fluores-cence endoscope in vitro. Endoscopy 2001;33:849–53.

13. Inayama K, Ito S, Muguruma N, et al. Basic study of an agent forreinforcement of near-infrared fluorescence on tumor tissue. Dig Liver

Dis 2003;35:88–93.