Photodiagnosis and Photodynamic Therapy (2006) 3, 259—265

DAFE autofluorescence assessment of oral cavity,larynx and bronchus in head and neckcancer patients

David Fielding FRACP, MDa,∗, Julie Agnewb,David Wrightb, Robert Hodgeb

a Department of Thoracic Medicine, Royal Brisbane and Women’s Hospital,Herston Rd., Herston 4029, Australiab Department of ENT Surgery, Royal Brisbane and Women’s Hospital,

Herston Rd., Herston 4029, Australia

Available online 28 August 2006

KEYWORDSAutofluorescencebronchoscopy;Lung cancer;Panendoscopy;Larynx cancer

Summary Head and neck (H&N) cancer patients have a high incidence of localfield change as well as second primary lung tumours. We have applied the WolfDiagnostic Autofluorescence Endoscopy (DAFE) system in a novel way, combining aut-ofluorescence evaluation of both H&N region and bronchial. New H&N cases as wellas ‘‘old’’ cases with symptoms were included. The DAFE procedure was done sep-arate to panendoscopy. The H&N region was examined first; images were recordedof the known primary with reference to subsequent resection margins, as well asof adjacent mucosa inspecting for additional abnormal sites. Then autofluorescencebronchoscopy was performed. Changes in management were only on the basis of his-tology taken because of abnormal autofluorescence. One hundred and seven caseswere referred, including 96 new cases and 11 old cases. Autofluorescence exami-nation of H&N detected sites which led to change of management in 11 patients.This included additional sites in nine patients (which then either had extra surgeryor radiotherapy as a result) and wider resection margins were made possible in twopatients. In the bronchus there were 21 significant lesions in 16 patients. Immediatemanagement change occurred in one invasive cancer, two microinvasive carcino-mas and three carcinomas in situ (in four patients). There were 15 sites of severeand moderate dysplasia in 12 patients. None of these bronchoscopic lesions weredetectable on CT chest. Therefore, overall an immediate change in management

occurred in 15 of 107 patients (14% of patients). This combined procedure yielded asignificant number of lesions particularly as a simple addition to preoperative work-up in new H&N cancer cases.© 2006 Elsevier B.V. All rights reserved.

∗ Corresponding author. Tel.: +61 7 36367631; fax: +61 7 36365651.E-mail address: david fielding@health.qld.gov.au (D. Fielding).

1572-1000/$ — see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.pdpdt.2006.07.004

ohasrFcaWflp

M

PlidilocPtwr

wscaiurwsa2u(abtrpifiotA

260

Introduction

Tissue autofluorescence is an established mode ofdetecting preneoplasia and in situ carcinoma inmucosal surfaces [1,2]. The most common applica-tion of fluorescence detection is in the bronchialtree [3—6]. Patients with squamous carcinoma ofthe head and neck are well known to have highrisk of synchronous and metachronous carcinomasin the lung and other areas of the upper aerodiges-tive tract [7—9]. For this reason patients are usuallyreferred for pan-endoscopy prior to definitive treat-ment [10]. The upper aerodigestive tract is easilyaccessible to flexible and rigid optics used in aut-ofluorescence diagnosis. Because of the commonoccurrence of field change disease in these regions,there is potential for fluorescence detection ofadditional significant lesions within the oral cavity,oropharynx and glottic regions. Fluorescence bron-choscopy has been used in head and neck patientsbut usually as surveillance rather than to detectsynchronous primaries at the time of initial diagno-sis [11]. The potential exists for the clinical applica-tion of using autofluorence to inspect the upper air-way as well as the bronchial tree at panendoscopy.

Tissue auto fluorescence has been investigatedin the oral cavity as reported by Arens et al. [12],Svistun et al. [13], and Onizawa et al. [14]. Inthese studies known tumours were assessed for afluorescence difference from the adjacent normalmucosa; there was correlation between increasingdegrees of fluorescence abnormality and increas-ing degrees of epithelial malignancy. Arens showedlesions of the larynx to have diminished fluores-cence using the filtered blue light (375—440 nm)of the Storz D Light system [12]. Generally speci-ficity was high at 84% for high grade dysplasia andmalignancy. Causes for false positive results weremarked hyperkeratosis, scarring and inflammation,although generally these were an uncommon findingand did not limit the overall utility of the technique.Autofluorescence of freshly resected oral tissue hasshown best specificity for discriminating malignantfrom benign tissue when incident light is 400 nm andobservation is at 530 nm [14]. Despite this potentialapplication, autofluorescence is not widely used todetect synchronous disease at panendososcopy inthese patients, and has not been reported in termsof its ability to change management.

Our study looked at clinical aspects of thepotential application of autofluorescence at panen-doscopy. Some data exists for the application of

tissue fluorescence using the topical photosensi-tizer aminolaevulinic acid (ALA) in the oral cavity[15] however in this study we only used auto fluo-rescence. The aims were firstly to define margins

mppc

D. Fielding et al.

f known or small localized primary tumours in theead and neck to assist with local excisions suchs laser excision. Secondly we looked for additionalites of in situ carcinoma and dysplasia within theseegions which might alter surgical management.inally, as part of the same procedure autofluores-ence bronchoscopy was performed with the usualim of detecting preneoplastic and in situ lesions.e sought to determine whether the addition ofuorescence to standard work-up would have theotential to change management.

ethods

atients were selected for this procedure from aarge multi-disciplinary Head and neck cancer clinicn a prospective fashion. Patients were either newlyiagnosed with a head and neck cancer which wasn the process of being worked up for a definitiveocal treatment including surgery or radiotherapy,r were in a follow-up phase having received priorurative treatment for a head and neck cancer.atients had usually undergone biopsy to confirmhe primary head and neck cancer under standardhite light conditions. They were at the point of

eferral for pan endoscopy.Patients who had bulky tumours (T3 or T 4)

ere not included as the finding of additional smallites of tissue abnormality was thought unlikely tohange management. All patients had chest X-raynd chest CT scan in addition to other CT scansn workup of the head and neck primary. Patientsndergoing fluorescence bronchoscopy did not haveadiological evidence of lung cancer. Proceduresere performed in a bronchoscopy suite under

tandard conscious sedation. Local anaesthetic waspplied to the oral cavity with mist spray of 2 ml% lignocaine. Auto fluorescence was performedsing the Diagnostic Autofluorescence EndoscopyDAFE) system (Wolf) [16]. This utilises a detach-ble camera for attachment to either a fiberopticronchoscope or a rigid optic. The camera receiveshe autofluorescence signal which is displayed ineal time on the monitor, with red indicating neo-lasia and green or light pink or white represent-ng non malignant tissue. The camera also provideseedback to the light source such that overall lightntensity is maintained; in more voluminous partsf the airway therefore greater light output fromhe light source occurs. Only additional sites ofF abnormality were acted upon, not the abnor-

al fluorescence of the known primary tumour. Therocedures were performed in two parts; the firstart was assessment of the floor of mouth, buc-al mucosa, and tonsil regions in patients known to

DAFE autofluorescence assessment of oral cavity, larynx and bronchus in head and neck cancer 261

Table 1 Site of new head and neck tumours

Site of head and neck primary Number of cases

Larynx 36Supraglottis 12Floor of mouth, alveolus 22Oropharynx 17

hsAtdbspidtoaCtai

R

TnphmynpbT

otpm

wAwtsdwd

Figure 1 White light (a) and fluorescence (b) image ofa known squamous cell carcinoma of floor of mouth. Thefluorescence image is bright and clear demarcation ofnormal and abnormal mucosa is possible. Note in the leftot

clftcTat

Unknown primary 9

Total 96

ave tumours in these regions. This was done in aitting position without administration of sedation.rigid optic specific to the Wolf system was used in

his location. Secondly with the patient supine theedicated 4.5 mm diameter flexible fluorescenceronchoscope was used to inspect the oropharynx,upraglottis and glottis, then the bronchoscopy waserformed in standard fashion. Where abnormal-ties were detected in the head and neck regionigital video images were recorded so that at theime biopsy could be performed subsequently. Somef the oral cavity assessments were performed usingdedicated rigid 4 mm fluorescence optic (Wolf).

hanges in management were considered only onhe basis of biopsy histology taken as a result ofbnormal tissue autofluorescence rather than themages themselves.

esults

here were 96 new cases who had both head andeck fluorescence inspection, and 11 cases withrior definitive treatment of H&N carcinoma whoad only fluorescence bronchoscopy. There were 90ales and 17 females with a mean age of 62 ± 9.1

ears. Table 1 lists the site of the primary carci-oma in the newly diagnosed head and neck canceratients. The mean time for a procedures includingoth head and neck and bronchus was 16 ± 6 min.here were no side-effects.

Known primary sites clearly showed marked aut-fluorescence abnormalities (Fig. 1). The DAFE sys-em provided excellent images even in more openarts of the oral cavity because of the camera auto-atically adjusting incident light intensity.DAFE detected significant sites in the H&N region

hich led to change of management in 11 patients.dditional sites were found in nine patients andider resection margins were made possible in

wo patients. Table 2 lists the location of these

ites. There were five cases of additional sitesetected around the larynx. One of these casesas a 65-year-old man who had previously beeniagnosed with a Right vocal cord squamous cell

nwow

f the fluorescence image a red fluorescence point dueo salivary gland opening.

arcinoma. This patient had undergone two whiteight assessments by ENT surgeons prior to referralor the fluorescence procedure. The DAFE examina-ion revealed an additional lesion on the Left falseord, approximately 3 mm in diameter (Fig. 2).herefore, in the subsequent ENT procedure, thisdditional tumour was biopsied and frozen sec-ion revealed microinvasive squamous cell carci-

oma. This was treated by laser excision alongith the planned laser of the known of tumourn the right vocal cord. Another of these casesas a 70-year-old lady who had contralateral vocal

262 D. Fielding et al.

Table 2 Sites of additional H&N tumour found with DAFE inspection of H&N region

Primary site Additional disease Number of patients Histology of additional site

Larynx Contralateral cord or supraglottis 5 Carcinoma (3)CIS (1)Severe dysplasia (1)

Pyriform fossa Local field change 1 CarcinomaFloor of mouth Tongue base, alveolus 2 Severe dysplasia (2)Soft palate Local field change 1 CarcinomaUnknown primary Tongue base, tonsil 2 Carcinoma (2)

cord carcinoma found by fluorescence which meantthat radiotherapy was given rather than the previ-ously planned surgery. Another case was a patientwho had previously had laser excision on Left vocalcord who had a Right cord carcinoma identified byfluorescence amongst non specific post radiother-apy inflammation; this patient had laryngectomy.A fourth case had a biopsy proven carcinoma of

Figure 2 (a) White light appearance of larynx showingR cord tumour. (b) DAFE appearance showing the abnor-mal R cord as well as an additional lesion anteriorly onL false cord. Frozen section biopsy of this additional siteshowed in situ carcinoma; both sites were treated withlaser excision.

right arytenoid after a standard white light H&Nassessment. Subsequent fluorescence assessmentshowed widespread disease of contralateral supra-glottic region; biopsies showed invasive carcinomain these regions. Radiotherapy was therefore giveninstead of surgery which had been planned. In the

Figure 3 Case of unknown primary of head and neck.Subtle changes in R tonsil on white light (a) are clearlyshown as abnormal on fluorescence image (b).

DAFE autofluorescence assessment of oral cavity, larynx and bronchus in head and neck cancer 263

tongue base and alveolus there were two caseswhere biopsies showed severe dysplasia. One casewas therefore treated with radiotherapy ratherthan surgery due to contralateral disease. The sec-ond had an abnormal PET scan showing diffuseright alveolar disease, and fluorescence was ableto localise this abnormality, clarifying the reasonsfor the PET change, and allowing an observationapproach rather than repeat treatment to an areaof local recurrence. Two cases of unknown primarydisease had prior standard white light H&N exami-nation (Fig. 3). In both cases fluorescence was read-ily able to identify the subsequent histological site

Figure 4 Squamous cell carcinoma in the right Pyriformfossa which was assessed prior to laser excision. The auto-fluorescence pictures demonstrate the margins to theapproximately 2—3 mm greater than can be seen understandard white light conditions. This directed the laserexcision to take a wider margin.

Figure 5 Example of moderate-severe dysplasia detec-ted by fluorescence examination (b) after prior standardwv

oieoaagc

mistapw

hite light examination only showed the macroscopic Rocal cord tumour.

f tumour. Biopsies could be directed to these sitesmmediately. There were two cases where laserxcision was to be performed (Fig. 4). The marginsf the tumour were clearly seen to extend furtherway from the subtle white light abnormalities,llowing the surgeon to extend the resection mar-ins further. In both cases resection margins werelear on final histology.

Not included in this list is an example ofoderate-severe dysplasia seen on the L vocal cord

n a patient with histologically proven R vocal cordquamous cell carcinoma (Fig. 5). This patient waso receive radiotherapy as part of a clinical study

nd as such management was not altered. Had thisatient been sent for laser resection this lesionould have required careful surveillance.

264

Table 3 Significant histology results of DAFE fluores-cence bronchoscopic biopsies

Histology Number of lesions

Moderate—severe dysplasia 15Carcinoma in situ 3Micronvasive carcinoma 2

im

D

WbcpotmpttctOtidbd7pcboaehSb

gTnrsrsbtsa

amsgd

Carcinoma 1

With respect to fluorescence bronchoscopy, intotal from all patients there were 124 biopsiesat an average of 1.2 ± 1.2 biopsies per patient.Table 3 lists the biopsies from fluorescence bron-choscopy which had moderate dysplasia or worseon histology; (the remaining histologies were milddysplasia, metaplasia, inflammation or normal). Inthe bronchus there were 21 significant lesions in16 patients. All were not detectable on CT chest.Four patients had six lesions showing carcinomaor carcinoma in situ. Of these four patients, allreceived radiotherapy either in combination withtreatment of the new head and neck primary, oras sole treatment. In addition one of these fourhad endobronchial electrocautery to a contrateralcarcinoma-in-situ lesion. The other 12 patients arehaving repeat DAFE biopsies of the known moderateto severe dysplasias, as are those with mild dyspla-sia at a longer interval.

Therefore, overall an immediate change in man-agement occurred in 15 patients (14% of cases) witha further 12 cases still under review.

The best sites for H&N fluorescence assessmentwere the floor of mouth, epiglottis, larynx and falsecord regions, and pyriform fossae. False positivebiopsies were obtained in five cases where fluores-cence was positive but biopsies showed no malig-nancy. These were in the base of tongue region,and the tonsils, presumably due to strong fluores-cence changes in lymphoid tissue in these regions.As listed in Table 2 fluorescence could provide use-ful information from these sites where there was aclear difference in fluorescence signal when com-paring the two sides; that is there was a clearchange compared to the background signal. Oneother artefact of note was the salivary gland open-ings in the floor of mouth. Fig. 1 shows an exampleof the bright fluorescence as obtained from theseducts. On occasion only the opening of the ductwas seen on tissue auto fluorescence whereas inother cases the length of the duct traversing the

flora of mouth could be seen. The surface of thetongue generally fluoresced very strongly and eval-uation of tongue tumour margins was difficult. Thiswas presumably due to an abundance of fluophores

omfli

D. Fielding et al.

n the actively dividing cells on the tongue surfaceucosa.

iscussion

e have demonstrated that autofluorescence cane simply used in the oral cavity and larynx tohange management in head and neck canceratients. We have also shown that the additionf autofluorescence bronchoscopy allows simul-aneous inspection of the other region com-only involved by second primary disease in theseatients, the bronchus. Panendoscopy with conven-ional white light has been reported to show asymp-omatic synchronous primary disease in up to 3% ofases of new head and neck cancer, including bothhe head and neck region and the bronchus [10].ur results are a proof of principle which suggesthe possible benefits of adding fluorescence exam-nation to a panendoscopy. They allow us to plan aetailed randomised study to determine the overallenefits of adding autofluorescence to a panen-oscopy. This will include the use of a dedicated0◦ fluorescence rigid pharyngoscope for the initialart of the assessment. As expected it a signifi-ant number of patients had radiologically occultut significant bronchial mucosal lesions. Detectionf these lesions is clearly important in these headnd neck cancer patients, as up to 30% higher canventually succumb to second primary lung canceraving being cured of the original primary tumour.tudies have already shown benefits of fluorescenceronchoscopy in these patients.

We found it easiest to image the larynx, supra-lottic region, pyriform fossae, and floor of mouth.he larynx is particularly suited to this technique,ot only because of the ease of inspecting thisegion given the strong normal autofluorescenceignals from the laryngeal tissue, but also becauseegional sites of dysplasia are common. The ton-il and tongue base were more difficult to inspectecause of background strong autofluorescence dueo lymphoid tissue. None the less it was still be pos-ible to detect relative differences in local tissueuto fluorescence.

We consider this technique to have potential inssisting ENT surgeons in detection of unknown pri-aries. In the two cases with this was useful in our

eries one case had been identified by the ENT sur-eon as a possible site prior to referral. Clearly aetailed blinded randomised study is in these types

f patients is required before it can be recom-ended as a tool in this setting. It may be thatuorescence could limit the number of sites requir-

ng biopsy around the upper aerodigestive tract.

D rynx 5

om

R

i-s-ofrg

s-erer

[ n-o-

[ E,inol

[ ct2:

[ n-ms-

[ c-i-

[ yd

er

AFE autofluorescence assessment of oral cavity, la

In conclusion this simple addition to the work upf cases of head and neck carcinoma can changeanagement.

eferences

[1] Hung J, Lam S, LeRiche JC, et al. Autofluorescence ofnormal and malignant bronchial tissue. Laser Surg Med1991;11:99—105.

[2] Ikeda N, Hiyoshi T, Kakihana M, Honda H, Kato Y. Histopatho-logical evaluation of fluorescence bronchoscopy usingresected lungs in cases of lung cancer. Lung Cancer2003;41:303—9.

[3] Lam S, Kennedy T, Unger M, et al. Localization of bronchialintraepithelial neoplastic lesions by fluorescence bron-choscopy. Chest 1998;113:696—702.

[4] Kennedy T-C, Lam S, Hirsch F-R. Review of recent advancesin fluorescence bronchoscopy in early localization of centralairway lung cancer. Oncologist 2001;6:257—62.

[5] Weigel TL, Kosco PJ, Dacic S, Rusch VW, Ginsberg RJ,Luketich JD. Postoperative fluorescence bronchoscopicsurveillance in non-small cell lung cancer patients. Ann Tho-rac Surg 2001;71:967—70.

[6] Weigel TL, Yousem S, Dacic S, Kosco PJ, Siegfried J, LuketichJD. Fluorescence bronchoscopic surveillance after curativesurgical resection for non small cell lung cancer. Ann SurgOncol 2000;7:176—80.

[7] Laccorreye O, Vievers FD, Hans S, Brasnu FD, Garcia D, Lac-courreye FL. Metachronous second primary cancers afterpartial laryngectomy for invasive sqaumous cell carcinomaof the true vocal cord. Ann Otol Rhinol Laryngol 2002;111(3Pt 1):204—9.

[

and bronchus in head and neck cancer 26

[8] Morita M, Araki K, Saeki H, et al. Risk factors for multcentric occurrence of carcinoma in the upper aerodigetive tract-analysis with a serial histologic evaluationthe whole resected-esophagus including carcinoma. J SuOncol 2003;83:216—21.

[9] Tabor MP, Brakenhoff RH, van-Houten VM, et al. Persitence of genetically altered fields in head and neck cancpatients: biological and clinical implications. Clin CancRes 2001;7(6):1523—32.

10] Stoeckli SJ, Zimmerman R, Schmid S. Role of routine panedoscopy in cancer of the upper aerodigestive tract. Otlaryngol Head Neck Surg 2001;124:208—12.

11] Venmans B, Ton JM, van Boxem Smit EF, Postmus PSutedja T. Bronchial intraepithelial neoplastic lesionshead and neck cancer patients. Ann Otol Rhinol Laryng2001;110:635—8.

12] Arens C, Glanz H, Dreyer T, Malzahn K. Compaendoscopy of the larynx. Ann Otol Rhinol Laryngol 2003;11113—9.

13] Svistun E, Alizadeh Naderi R, El Naggar A, Jacob R, Gillewater A, Richards Kortum R. Vision enhancement systefor detection of oral cavity neoplasia based on autofluorecence. Head Neck 2004;26:205—15.

14] Onizawa K, Okamurab N, Saginoyac H, Yoshidaa H. Charaterization of autofluorescence in oral squamous cell carcnoma. Oral Oncol 2003;69:150—6.

15] Betz CS, Stepp H, Janda P, et al. A comparative studof normal inspection, autofluorescence and 5ALA inducePPIX fluorescence for oral cancer diagnosis. Int J Canc

2002;97:245—52.

16] Goujon D, Glanzmann T, Gabrecht T, et al. Detection ofearly bronchial carcinoma by imaging of the tissue autoflu-orescence. Proc SPIE 2001;4432:131—8.