ARTICLE IN PRESS

Optical Biopsy of Human Bladder Neoplasia With In Vivo

Confocal Laser Endomicroscopy

Geoffrey A. Sonn, Sha-Nita E. Jones, Tatum V. Tarin, Christine B. Du,Kathleen E. Mach, Kristin C. Jensen and Joseph C. Liao*,†

From the Departments of Urology (GAS, SEJ, TVT, CBD, KEM, JCL) and Pathology (KCJ), and Stanford Cancer Center (JCL), StanfordUniversity School of Medicine, Stanford, and Veterans Affairs Palo Alto Health Care System, Palo Alto (KCJ, JCL), California

Purpose: Confocal laser endomicroscopy is a new endoscopic imaging technologythat could complement white light cystoscopy by providing in vivo bladder his-topathology. We evaluated confocal laser endomicroscopy by imaging normal,malignant appearing and indeterminate bladder mucosa in a pilot study.Materials and Methods: Patients scheduled to undergo transurethral resectionof bladder tumors were recruited during a 3-month period. After standard cys-toscopy fluorescein was administered intravesically and/or intravenously as acontrast dye. A 2.6 mm probe based confocal laser endomicroscope was passedthrough a 26Fr resectoscope to image normal and abnormal appearing areas. Theimages were collected with 488 nm excitation at 8 to 12 frames per second. Theendomicroscopic images were compared with standard hematoxylin and eosinanalysis of transurethral resection of bladder tumor specimens.Results: Of the 27 recruited patients 8 had no cancer, 9 had low grade tumors, 9had high grade tumors and 1 had a low grade tumor with a high grade focus.Endomicroscopic images demonstrated clear differences between normal mucosa,and low and high grade tumors. In normal urothelium larger umbrella cells areseen most superficially followed by smaller intermediate cells and the less cellu-lar lamina propria. In contrast, low grade papillary tumors demonstrate denselyarranged but normal-shaped small cells extending outward from fibrovascularcores. High grade tumors show markedly irregular architecture and cellularpleomorphism.Conclusions: We report the first study to our knowledge of in vivo confocal laserendomicroscopy in the urinary tract. Marked differences among normal urothe-lium, low grade tumors and high grade tumors were visualized. Pending furtherclinical investigation and technological improvement, confocal laser endomicros-copy may become a useful adjunct to conventional cystoscopy.

Key Words: microscopy, confocal; urinary bladder neoplasms; anatomy and

Abbreviations

and Acronyms

TURBT � transurethral resectionof bladder tumor

Submitted for publication January 29, 2009.Study received institutional review board ap-

proval.Clinical Trial Registration NCT00801762 (www.

clinicaltrials.gov).* Correspondence and requests for reprints: De-

partment of Urology, Stanford University School ofMedicine, 300 Pasteur Drive, S-287, Stanford, Cal-ifornia 94305-5118 (telephone: 650-852-3284; FAX:650-849-0319; e-mail: jliao@stanford.edu).

† Recipient of Stanford University CancerCenter Developmental Cancer Research Award.

To view the accompanying video, pleasesee the online version of this article (Vol-ume 182, Number 4) at www.jurology.com.

histology; biopsy

DIAGNOSTIC cystoscopy is an integralpart of urinary tract evaluation, partic-ularly for bladder cancer. Nevertheless,numerous well recognized limitationsinclude operator dependency, differen-tiation of inflamed mucosa from flat

neoplastic lesions and lack of real-time

0022-5347/09/1824-1299/0THE JOURNAL OF UROLOGY®

Copyright © 2009 by AMERICAN UROLOGICAL ASSOCIATION

tumor grade or stage information. Pro-cedures to obtain tissue diagnosis in-cluding TURBT and random bladderbiopsy carry risks of anesthesia, bleed-ing and bladder perforation in additionto the several day delay to obtain

pathological results. Due in large part

Vol. 182, 1299-1305, October 2009Printed in U.S.A.

DOI:10.1016/j.juro.2009.06.039www.jurology.com 1299

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA1300ARTICLE IN PRESS

to its recurrent nature and the shortcomings of cystos-copy for diagnosis and surveillance, bladder cancer isthe most expensive cancer from diagnosis to death inthe United States.1

Widespread interest exists in applying new technol-ogies to improve the diagnostic accuracy of standardwhite light cystoscopy. The bladder is particularly wellsuited to evaluate novel endoscopic imaging modalitiesgiven the easy accessibility and the option of intraves-ical instillation of pharmacological agents to providetissue contrast. Two promising technologies includehexylaminolevulinate fluorescence cystoscopy2 and op-tical coherence tomography.3 However, neither methodprovides the cellular resolution that could possiblyreduce or eliminate the need for TURBT.

Confocal microscopy is a powerful imaging toolwidely used in basic research.4 Using a laser lightsource and fluorescent dyes as contrast agents, confo-cal microscopy offers high resolution, dynamic, subsur-face optical sectioning of biological systems.4–6 Whileconventional confocal microscopes are too large to usein vivo, recent advances in instrument miniaturizationhave led to the development of flexible, fiberoptic con-focal microscopes that can be passed through theworking channel of standard endoscopes.7–10 This en-ables in vivo microscopy, called confocal laser endomi-croscopy or fibered confocal microscopy. Applicationof confocal laser endomicroscopy in the gastroin-testinal tract for esophageal, stomach and coloncancer detection has moved beyond early proof ofconcept evaluation to large clinical trials.11–17 Todate, in vivo use in the urinary tract has beenlimited to 2 animal studies.18,19

Recently we reported on confocal laser endomi-croscopy of fresh radical cystectomy specimens usingintravesical fluorescein, a Food and Drug Adminis-tration approved drug, as the fluorescent contrastagent.20 We noted cellular resolution and microar-chitectural features of normal appearing and can-cerous regions that correlated with standard hema-toxylin and eosin staining.

In this study we hypothesized that confocal laserendomicroscopy could be used with conventionalcystoscopy to provide real-time histological informa-tion that may help distinguish normal from neoplas-tic urothelium in vivo. We documented the confocalmicroscopic appearance of normal urothelium, andcompared it to low and high grade tumors in pa-tients undergoing TURBT.

MATERIALS AND METHODS

Confocal Microscopy Imaging SystemThe Cellvizio® fibered confocal microscope used in thisstudy consisted of a 2.6 mm diameter flexible ConfocalMiniprobe™ attached to a laser scanning unit. A 488 nm

laser delivered the excitation beam for scanning that was

transmitted through an imaging bundle containing morethan 30,000 optical fibers. Several optical lenses at thedistal end of the probe focused the beam into the tissue,producing images with a depth of penetration of 60 �m, alateral resolution of 1 �m and a field of view 240 �m indiameter. Fluorescence was collected by the same lens andrefocused back into the illumination fiber such that asingle optical fiber acted as the illumination point sourceand the detection pinhole. The images were collected at arate of 8 to 12 frames per second through direct contact ofthe urothelium with the probe tip. The prototype probesused for this study were sterilized with the STERIS®system before each use.

Imaging ProtocolThe Stanford University and Veterans Affairs Palo AltoHealth Care System institutional review board approvedthe research protocol (ClinicalTrials.gov IdentifierNCT00801762). All patients scheduled to undergo TURBTbased on abnormal office cystoscopy from September toDecember 2008 were recruited for the study. A singlesurgeon (JCL) performed each case including the confocalimage acquisition. With the patient under spinal or gen-eral anesthesia standard white light cystoscopy was per-formed with a 21Fr rigid cystoscope or a 26Fr resectoscopewith a visualizing obturator (Karl Storz Endoscopy, Cul-ver City, California). Fluorescein sodium (InternationalMedications Systems, Ltd., South El Monte, California)then served as a contrast agent through intravesical in-stillation and/or intravenous administration. For intra-vesical instillation 300 to 500 cc 0.1% fluorescein dilutedin normal saline was instilled into the bladder via a Foleycatheter and left indwelling for 5 minutes. For intrave-nous administration 1 ml 10% fluorescein was used. Insubjects given fluorescein via both routes the intravesicalfluorescein was always administered first. Endomicros-copy video sequences were obtained by passing the probethrough the working channel of a standard Storz 26Frresectoscope (fig. 1). This allowed for simultaneous confo-cal microscopy and white light cystoscopy using a 0-degreeor 30-degree lens. Normal and abnormal appearing areas ofurothelium were imaged. White light cystoscopy images andvideos were recorded and stored on compact discs for eachparticipant. Endomicroscopy video collection was controlledwith a foot pedal, and the images were viewed in real timeand stored as digital video files. In 9 cases biopsies weretaken of the normal appearing area that had been imaged

Figure 1. A, 2.6 mm probe passed through working channel of26Fr Storz resectoscope. Blue laser transmitted through fiber-optic probe from Laser Scanning Unit. B, probe in direct contact

with bladder urothelium during confocal imaging.

hrocyt

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA 1301ARTICLE IN PRESS

with the confocal microscope to confirm the lack of malig-nancy. In another 11 cases in which intravenous fluoresceinwas used the prostatic urethra was also imaged.

Figure 2. Comparison of H & E and confocal endomicroscopic ipolygonal superficial cells consistent with umbrella cells. B, smcellular lamina propria containing blood vessels filled with eryt

Figure 3. A, endomicroscopic image of tumor resection bed aftefrom this site demonstrating fibers of muscularis propria. C, end

of fat lobules. D, H & E stain of biopsy from this site confirming prese

After acquiring endomicroscopic images standard TURBTwas performed using cold cup biopsy forceps followed bymonopolar electrocauterization or the bipolar loop (Gyrus

of normal bladder mucosa with fluorescein staining. A, large,eeper urothelial cells consistent with intermediate cells. C, lesses.

T showing fibers of muscularis propria. B, H & E stain of biopsyscopic image of resection bed revealing distinctive appearance

magesaller, d

r TURBomicro

nce of fat in specimen.

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA1302ARTICLE IN PRESS

ACMI, Southborough, Massachusetts). In 9 cases furtherendomicroscopic images were obtained after completingTURBT to evaluate the tumor bed. The biopsy specimenswere then sent for standard pathological assessment withhematoxylin and eosin staining. The endomicroscopic im-ages were further analyzed and compared with the hema-toxylin and eosin sections of the corresponding tissues.

RESULTS

Between September and December 2008, 28 pa-tients (mean age 73 years, range 47 to 90) werescheduled to undergo TURBT at our institution. Allpatients were recruited for the study and 1 declined,leaving a study population of 27 patients. Real-timeendomicroscopy video sequences of the bladder mu-cosa were collected in vivo after intravesical and/orintravenous administration of fluorescein. Mean du-ration of the endomicroscopy portion was 18 minutesoverall (range 3 to 35), 21 minutes for the first 13cases and 16 minutes for the remaining 14 cases. Nosystemic toxicity or hypersensitivity reactions fromintravesical or intravenous fluorescein were ob-served. Patients who received intravenous fluores-cein had green tinged urine for approximately 24hours.

Intravenous vs Intravesical Fluorescein

All patients received fluorescein intravenously (10),intravesically (5) or via both routes (12). Those givenfluorescein via both routes underwent imaging afterintravesical and again after intravenous fluoresceinadministration. The first subject received intrave-nous fluorescein before cystoscopy. However, due tothe difficulty performing cystoscopy with the greenfluorescein dye in the bladder, subsequent patientswere given intravenous fluorescein only after com-pletion of white light cystoscopy. We found subtledifferences in image quality and contrast patternbetween the 2 administration routes. Within 1minute of injection of intravenous fluorescein, con-trast material was visible in the blood vessels of thelamina propria, providing striking images of eryth-rocytes moving within the vessels. For several min-utes the fluorescein diffused into the surroundingtissue providing contrast for visualization of bladdercellular architecture. After approximately 10 min-utes the fluorescein began to be excreted into thebladder. With intravesical fluorescein image qualitybegan to deteriorate after approximately 15 min-utes, likely due to fluorescein washout. Interestinglyin several patients given only intravesical fluores-cein the vasculature in the lamina propria was vi-sualized, suggesting that the fluorescein is able todiffuse through the transitional epithelium into thelamina propria. We were unable to image the blad-

der using the confocal probe without fluorescein due

to the lack of significant autofluorescence of urothe-lial cells.

Normal Histology

Confocal laser endomicroscopy was performed onnormal appearing areas of the bladder mucosa in allpatients. We successfully visualized the normalbladder urothelium and lamina propria (fig. 2, videosegment 1). Within the urothelial layer the moresuperficial cells were consistently larger than thedeeper layer, corresponding to the known size dif-ference between the larger umbrella cells and thesmaller intermediate cells seen on hematoxylin andeosin microscopy. The urothelial cells existed in anorganized network highlighted by the fluoresceinstaining of the extracellular matrix of the cells. Thisvisualized pattern is unique to confocal microscopybecause standard histopathological processing doesnot accentuate cellular borders and the tissue is cutin a plane transverse to the mucosal surface. Cellu-lar nuclei were not visualized because fluoresceindoes not cross the cell membrane. However, cellularoutlines were obvious due to the extracellular dis-persion of fluorescein. By applying pressure with theprobe the underlying lamina propria, with its vas-cular network and less cellular connective tissue,was visualized. Individual erythrocytes were consis-tently seen flowing within vessels of the lamina pro-pria after giving intravenous fluorescein, and lesscommonly after intravesical administration. In gen-eral we noted decreased image quality with bladderoverdistention. All 9 cases in which random bladder

Figure 4. Endomicroscopic image of prostatic urethra with ap-

pearance suggestive of prostatic glands and intervening stroma.

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA 1303ARTICLE IN PRESS

biopsies were obtained from a normal appearingarea that had been imaged with the confocal micro-scope showed no evidence of malignancy. Endomi-croscopic imaging of the resection bed after biopsyyielded images of muscularis propria fibers (fig. 3, Aand B) and fat (fig. 3, C and D). In 2 of the 11patients in whom the prostatic urethra was scannedimages suggestive of prostatic glands and stromawere identified (fig. 4). In the other 9 patients bloodvessels were visible but no glands or stroma wereseen.

Bladder Tumors

Of the 27 patients 8 had no cancer (3 with normalurothelium only, 1 with cystitis cystica, 2 with in-flammation only, 1 with a dense lymphocytic infil-trate, and 1 with mild to moderate dysplasia andinflammation), 9 had low grade tumors, 9 had highgrade tumors including 1 with carcinoma in situ,and 1 had a predominately low grade tumor with ahigh grade focus. Of the patients with high gradetumors 4 underwent subsequent radical cystopros-tatectomy. In 2 patients endomicroscopic imageswere not obtained due to an anterior tumor locationthat made contacting the tumor with the probe tipimpossible. Apart from these patients confocal laserendomicroscopy documented clear differences be-

Figure 5. A, endomicroscopic image of low grade, papillary tneoplastic urothelial cells. B, corresponding H & E slide. C, end

D, corresponding H & E slide.

tween normal mucosa, and low and high grade tu-mors in most cases. In contrast to the normal cellulararchitecture, low grade papillary tumors demon-strated densely arranged but uniformly shaped smallcells extending outward from fibrovascular cores (fig. 5,video segment 2). High grade tumors showed mark-edly disorganized architecture and variably sizedcells that were much different from normal urothe-lium or low grade tumors (fig. 6). Characterization ofthe nuclear appearance of the cells in these tumorswas not possible because fluorescein does not enterthe cells. In addition, with a 60 �m penetrationdepth the confocal probe used in this study cannotvisualize the muscularis propria through the intacturothelium and, therefore, does not offer informa-tion about muscle invasion for staging.

DISCUSSION

We report the first in vivo microscopic evaluation ofhuman bladder urothelium using a confocal laserendomicroscope. Confocal laser endomicroscopy en-ables subsurface imaging of living cells in the blad-der mucosa during cystoscopy. Fluorescein, admin-istered intravesically or intravenously, is a safe andeffective contrast agent that permits visualization ofthe urinary tract with the confocal endomicroscope.

with central fibrovascular core surrounded by well organizedscopic image of low grade papillary tumor in another patient.

umoromicro

mage

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA1304ARTICLE IN PRESS

The confocal probe is passed through the workingchannel of a standard resectoscope, offering simul-taneous confocal microscopy and cystoscopy. The ac-quired endomicroscopic images display sufficientresolution to distinguish cell size and architecture,allowing the differentiation of different depths inthe urothelium and the lamina propria. Most impor-tantly we noted marked differences between normalmucosa, and low and high grade tumors in mostcases.

Our feasibility study has several limitations, par-ticularly the relatively small sample size and thenonquantitative nature of the image analysis. Inaddition, due to the small size of the probe and thefield of view, image acquisition is exquisitely sensi-tive to motion introduced by the subject (ie dia-phragmatic motion or arterial pulsation) or the op-erator. Also, a probe developed for gastrointestinalendoscopy was used whose 2.6 mm size necessitatesthe use of a 26Fr rigid resectoscope instead of aflexible cystoscope. Lacking the ability to manipu-late the probe tip with a rigid cystoscope makesestablishing the required perpendicular contact withthe tissue impossible in cases where the area ofinterest resides on the anterior bladder wall. Accord-ingly due to their location in the bladder some tu-mors in this study could not be imaged. In addition,

Figure 6. A, endomicroscopic image of high grade, noninvasivirregularity. B, corresponding H & E slide. C, endomicroscopic i

given the 2.6 mm probe tip size with a 240 �m field

of view and the requirement for direct contact withthe urothelium, imaging the entire bladder is notpractical. Rather the proposed role of confocal laserendomicroscopy is to provide real-time histology ofareas highlighted by white light cystoscopy or otherimaging modalities such as fluorescence cystoscopy.

Approximately 10 minutes after intravenous ad-ministration fluorescein is excreted into the bladder,potentially interfering with cystoscopic visualiza-tion. In contrast to most other hollow organs thebladder offers a direct route for topical administra-tion of medications. Our study demonstrates thatintravesical fluorescein instillation adequately pro-vides cellular resolution and microarchitecture com-parable to intravenous administration. In addition,intravesical fluorescein can be rinsed out of the blad-der, thereby regaining excellent clarity while main-taining sufficient cellular contrast to permit micros-copy. Nevertheless, image quality does deteriorateafter approximately 15 minutes likely due to re-peated bladder irrigation. However, intravesical flu-orescein instillation may be repeated if necessary.Finally fluorescein does not permit visualization ofcell nuclei as noted in similar studies in the gastro-intestinal tract.15 Rather it nonspecifically stainsthe extracellular matrix of normal and abnormal

helial tumor with extensive architectural disarray and cellularof another high grade tumor. D, corresponding H & E slide.

e urot

bladder mucosa. Due to a concern of potential carci-

CONFOCAL LASER ENDOMICROSCOPY OF BLADDER NEOPLASIA 1305ARTICLE IN PRESS

nogenicity we did not use acriflavine hydrochloride,a contrast agent capable of nuclear staining.

Despite these limitations confocal laser endomi-croscopy may prove to be a useful adjunct to conven-tional cystoscopy. The ability to differentiate nor-mal, benign and neoplastic tissues in real time withmicroscopic resolution during cystoscopy would pro-vide valuable diagnostic information that could im-pact clinical decision making such as avoiding un-necessary biopsies, allowing more targeted biopsiesand confirming the low grade nature of tumors beingmanaged expectantly. The possibility of using intra-vesical fluorescein to generate interpretable imagesmakes the use of this technology in the clinic settingmore feasible. To achieve that goal evaluation withsmaller probes that can pass through a flexible cys-toscope will be necessary. We are developing objec-tive criteria for identifying and grading bladder tu-mors that will be applied in a larger prospective andblinded study in the future. A larger study wouldallow us to characterize the confocal microscopic

appearance of benign bladder tumors to better dis-

REFERENCES

87: 737. 451.

tinguish them from bladder carcinomas. We alsoplan to image more patients with flat lesions such ascarcinoma in situ, an application that may prove tobe the best use of this technology.

CONCLUSIONS

Confocal microscopic imaging of the human bladderusing fluorescein as the contrast agent can be per-formed in vivo and provides real-time imaging of thecellular architecture. In most cases marked differ-ences were visualized among normal urothelium,low grade tumors and high grade tumors. Pendingfurther clinical investigation and technological im-provement, confocal laser endomicroscopy may be-come a useful adjunct to conventional cystoscopy.

ACKNOWLEDGMENTS

Anne Osdoit (Mauna Kea Technologies) providedtechnical support, Hardeep Phull assisted with im-age processing and Thomas D. Wang contributed to

discussions.

1. Botteman MF, Pashos CL, Redaelli A et al: Thehealth economics of bladder cancer: a compre-hensive review of the published literature. Phar-macoeconomics 2003; 21: 1315.

2. Grossman HB, Gomella L, Fradet Y et al: A phaseIII, multicenter comparison of hexaminolevulinatefluorescence cystoscopy and white light cystos-copy for the detection of superficial papillarylesions in patients with bladder cancer. J Urol2007; 178: 62.

3. Lerner SP, Goh AC, Tresser NJ et al: Opticalcoherence tomography as an adjunct to whitelight cystoscopy for intravesical real-time imag-ing and staging of bladder cancer. Urology 2008;72: 133.

4. Pawley JB: Handbook of Biological Confocal Mi-croscopy, 3rd ed. New York: Springer 2006.

5. Minsky M: Microscopy Apparatus. United StatesPatent 301467, 1961.

6. Minsky M: Memoir on inventing the confocalscanning microscope. Scanning 1988; 10: 128.

7. Helmchen F: Miniaturization of fluorescence mi-croscopes using fibre optics. Exp Physiol 2002;

8. Yelin D, Rizvi I, White WM et al: Three-dimen-sional miniature endoscopy. Nature 2006; 443:765.

9. Rouse AR, Kano A, Udovich JA et al: Design anddemonstration of a miniature catheter for a con-focal microendoscope. Appl Opt 2004; 43: 5763.

10. Osdoit A, Genet M, Perchant A et al: In vivofibered confocal reflectance imaging: totally non-invasive morphological cellular imaging broughtto the endoscopist. Presented at the SPIE, SanJose, California 2006.

11. Hsiung PL, Hardy J, Friedland S et al: Detection ofcolonic dysplasia in vivo using a targeted hep-tapeptide and confocal microendoscopy. Nat Med2008; 14: 454.

12. Wang TD, Friedland S, Sahbaie P et al: Functionalimaging of colonic mucosa with a fibered confo-cal microscope for real-time in vivo pathology.Clin Gastroenterol Hepatol 2007; 5: 1300.

13. Kiesslich R, Goetz M and Neurath MF: Confocallaser endomicroscopy for gastrointestinal dis-eases. Gastrointest Endosc Clin N Am 2008; 18:

14. Meining A, Bajbouj M and Schmid RM: Confocalfluorescence microscopy for detection of gastricangiodysplasia. Endoscopy 2007; 39: E145.

15. Kiesslich R, Burg J, Vieth M et al: Confocal laserendoscopy for diagnosing intraepithelial neopla-sias and colorectal cancer in vivo. Gastroenterol-ogy 2004; 127: 706.

16. Meining A and Wallace MB: Endoscopic imagingof angiogenesis in vivo. Gastroenterology 2008;134: 915.

17. Kakeji Y, Yamaguchi S, Yoshida D et al: Devel-opment and assessment of morphologic criteriafor diagnosing gastric cancer using confocal en-domicroscopy: an ex vivo and in vivo study. En-doscopy 2006; 38: 886.

18. D’Hallewin MA, El Khatib S, Leroux A et al:Endoscopic confocal fluorescence microscopy ofnormal and tumor bearing rat bladder. J Urol2005; 174: 736.

19. Boyette LB, Reardon MA, Mirelman AJ et al:Fiberoptic imaging of cavernous nerves in vivo.J Urol 2007; 178: 2694.

20. Sonn GA, Mach KE, Jensen K et al: Fiberedconfocal microscopy of bladder tumors: an ex vivo

study. J Endourol 2009; 23: 197.