~022-5347/96/1565-1597$03.00/0 Vol. 156,1597-1601,November 1996 Printed in U S A .
JOURNAL OF UROLOGY
Copyright 0 1996 by AMERICAN UROL~CICAL ILSSOCUTION, INC.
LASER INDUCED AUTOFLUORESCENCE DIAGNOSIS OF BLADDER CANCER FRANK KOENIG,* FRANCIS J. McGOVERN, ALEX F. ALTHAUSEN, THOMAS F. DEUTSCH KEVIN T. SCHOMACKER
AND
From the Wellman Laboratories of Photomedicine and the Department of Urology, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts
ABSTRACT
Purpose: We assessed the ability of laser induced autofluorescence to differentiate malignant from nonmalignant bladder lesions. Materials and Methods: We studied 53 patients with bladder cancer undergoing mucosal biopsies or transurethral resection of a bladder tumor. A quartz optical fiber was advanced through the working channel of a cystoscope and placed in gentle contact with the bladder. Tissue fluorescence was excited by 337 nm. light pulses (nitrogen laser). One fiber was used for transmission of the excitation and emission (fluorescence) light. An optical multichannel analyzer system was used to record fluorescence spectra of the sites of interest. Results: We analyzed the fluorescence spectra of 114 bladder areas (1carcinoma in situ as well as 28 malignant, 35 inflammatory, 7 dysplastic, 1 squamous metaplastic and 42 normal areas). These lesions included 44 difficult to diagnose suspicious tumors (11malignant and 33 nonmalignant). We developed an algorithm that used the 1385:1455nm. fluorescence ratio to distinguish malignant from nonmalignant lesions, including inflammatory areas. By analyzing the data on all 114 lesions, we noted the sensitivity, specificity, and positive and negative predictive values of this method for differentiating malignant from nonmalignant bladder lesions to be 97,98,93 and 99%, respectively. Conclusions: Under excitation with 337 nm. light a clear differentiation between malignant and nonmalignant bladder tissues can be made using the 13851455 nm. autofluorescence ratio. KEYWORDS:lasers, bladder neoplasms, diagnosis, fluorescence
Transitional cell carcinoma of the bladder is the fourth most common cancer in men and the eighth most common cancer in women in the United States. In 1995 an estimated 50,500 new cases were registered and an estimated 11,200 people died of bladder cancer.1 In approximately 50 to 65% of cases the lesions were superficial at presentation but despite apparently adequate transurethral resection of the tumor the recurrence rate was 55 to 70%.2,3 It is well known that patients with stage T1 tumors have a poorer prognosis than those with stage Ta lesions, and that those with grade 3 disease have a greater chance of tumor recurrence andor progression than those with grade 1 lesion^.^^^ It is also known that approximately 50% of patients with low stage, low grade transitional cell carcinoma have concomitant uothelial dysplasia, either carcinoma in situ or higher differentiated forms of dysplasia.5.6 Up to 80% of the patients with carcinoma in situ found in addition to a low stage, low grade lesion have an associated invasive cancer within 4 Years. Invasion will occur within 4 years in 30% of patients if any other dysplastic lesion is found adjacent to multiple tumors that involve the lamina propria.5 Since early malignant and dysplastic lesions are often barely visible or invislble, random biopsies are recommended in addition to resection of the visible tumor. TOovercome this unsatisfactory situation 2 groups in EUrope are currently using 5-aminolevulinic acid to label early
malignant and dysplastic bladder lesions.7.S Intravesical application of 5-aminolevulinic acid before cystoscopy leads to an increase in protoporphyrin M (a strong photosensitizer) in malignant and in some dysplastic as well as inflammatory lesions of the bladder. Under blue light these lesions fluoresce bright red and, therefore, they are easily detectable during cystoscopy. The sensitivity of 5-aminolevulinic acid induced fluorescencecystoscopy for cancer detection is nearly 100%. However, due to false-positive fluorescent inflammatory lesions the specificity is only 60 to 70%,which may lead to unnecessary biopsies and additional resection of bladder tissue with an increased risk and cost to the patient. Recently, the reflectance and autofluorescence properties of the urothelium have been studied to develop optically based systems for bladder cancer detection. In a reflectance study in vivo measurements of patients with transitional cell carcinoma were performed to distinguish malignant from nonmalignant tissue.9 A diagnostic algorithm based on the values of the slopes for wavelengths of 330 to 370 nm. was used, and had a sensitivity and specificity of 100 and 97%, respectively. Due to the limited number of patients (10) the potential and usefulness of reflectance spectroscopy for bladder cancer diagnosis still must be addressed. Fluorescence from innate tissue fluorophores, that is autofluorescence, can potentially be used to obtain diagnostic information. The intensity and spectral shape of tissue autofluorescence are generally determined by absorption and emission of these fluorophores, as well as by the absorption of blood. The major fluorophores in the bladder wall most likely include collagen and the reduced form of nicotinamide adenine dinucleotide (NADH).lO*11Light with wavelengths of 300 to 365 nm. (ultraviolet light) matches the absorption maxima of these biomolecules and is appropriate for excita-
Accepted for publication May 10, 1996. Supported by Grants N00014-94-1-0927from the Office of Naval Research Contract a d DOE #DE-FG02-91-ER61228from the DeWment of Energy, a d a grant from the Deutsche Akademie der Katurforscher Leopoldina (German foundation). * Requests for re fits: Wellman Laboratories of Photomedicine m L 224), Massactusetta General Hospital, 50 Blossom St., Boston, Massachusetts 02114. 1597
1598
AUTOFLUORESCENCE DIAGNOSIS OF BLADDER CANCER
tion of bladder tissue autofluorescence. The autofluorescence intensity of a tissue layer beneath a nonfluorescent layer can also be affected by the thickness of the nonfluorescent layer. To our knowledge there are only 2 human in vivo autofluorescence studies of bladder tissue to date.11.12 One study used 365 as well as 355 nm. excitation, and found differences in autofluorescence between normal and malignant urothelium, including carcinoma in situ." While the shape of the autofluorescence spectra for carcinoma in situ and transitional cell carcinoma was identical to that of normal bladder tissue, the fluorescence intensity was lower for carcinoma in situ than for normal tissue and even lower for transitional cell carcinoma. Baert et a1 performed autofluorescence measurements using 337 and 405 nm. excitation.12 With 337 nm. excitation light the 460:400 nm. ratio was used to differentiate successfully between normal urothelium and visible papillary bladder tumors. With 405 nm. excitation light the authors found that the autofluorescence intensity for some cases of dysplasia was lower than that of normal tissue. However, these results were found only in a few of all measured dysplastic lesions and, therefore, they are not statistically significant. Inflammatory lesions of the human bladder are difficult to distinguish from flat and micropapillary malignancies by any clinical method. To our knowledge no autofluorescence spectra of inflammatory lesions have been reported. Therefore, we explored the suitability of laser induced autofluorescence for bladder cancer detection by studying the fluorescence spectra of malignant and nonmalignant bladder tissues, including inflammatory lesions. Since the biomolecules NADH and collagen are believed to be important fluorophores of the bladder wall, a nitrogen laser (337 nm.) was chosen as the excitation source. MATERIALS AND METHODS
Fluorescence measurements were performed in 13 women and 40 men with bladder cancer undergoing mucosal biopsies or transurethral resection of the bladder a t initial diagnosis or a t regular bladder tumor followup visits. Average patient age was 67 years (range 47 to 88). The measuring setup is shown in figure 1. Tissue fluorescence was excited by light pulses generated by a nitrogen
laser. The output of the laser (337.1 nm., 3 ns. pulses a t 10 Hz. and 200 @. pulse energies) was coupled via a quartz lens into a 0.6 mm. core diameter quartz optical fiber. Typical energies delivered to tissue were approximately 40 p.J. The fluorescenceemitted from the tissue was transmitted via the same optical fiber and optically coupled to a 0.275 m. polychromator. A fiberoptic connector was used to couple the sterile fiber to the measuring system. The fluorescence spectrum (300 to 800 nm.) was recorded using an intensified 1,024 diode array optical multichannel analyzer system. The intensifier was gated with 100 ns. pulses centered around the 3 ns. laser pulses. Variations in laser energy and fiber COUpling efficiencies were considered by measuring the fluorescence from a cuvette of laser dye of known concentration on the day of the measurements. The multichannel analyzer recorded a complete spectrum with each laser pulse, and 50 spectra were collected and averaged for each measurement. Background spectra were measured daily and subtracted from all recorded tissue spectra. All spectra were also system response corrected by dividing the observed spectra by a system response curve, which was determined by measuring the spectral intensity of a calibrated 1,000 W. quartz halogen lamp. The diode array was spectrally calibrated using a low pressure mercury lamp. As the initial step the bladder was visually inspected thoroughly to detect nodular, sessile or papillary tumors as well as suspicious (hypervascularized, edematous, raised and so forth) lesions of the urothelium. When biopsies or resections were needed for medical purposes the quartz optical fiber was advanced through the working channel of a standard cystoscope and placed in gentle contact with the site of interest. The fluorescence spectrum was then recorded within approximately 5 seconds using the optical multichannel analyzer system. Suspicious areas, visible papillary tumors, including adjacent normal appearing sites, and bladder areas in patients after previous treatment with chemotherapy combined with radiation therapy as well as those treated with bacillus Calmette-Guerin, were included in this study to obtain basic information on different tissue types. All measured areas were biopsied immediately after the fluorescence measurements using a flexible forceps advanced into position with the second working channel of the cystoscope to make sure the biopsy was obtained exactly from the measured point. At the end of the procedure all suspicious areas and visible tumors were removed by standard transurethral resection. All tissue samples were fixed in formalin and submitted for routine pathological analysis. To decrease the potential for interobserver variability fluorescence measurements were performed by only 2 physicians using video camera guidance to ensure that the fiber was placed in gentle contact with the tissue surface. The sensitivity, specificity, and positive and negative predictive values of laser induced autofluorescence for detection of malignant versus nonmalignant urothelium were determined with standard statistical methods. To obtain the significance of fluorescence differences between groups of different tissue type we used a standard Student t test (2-tailed, equal variance) and defined significance at p <0.005.13 RESULTS
As determined by histological classification, there were 35 inflammatory (24 chronic and 11 acute), 42 normal, 1 squamous metaplastic, 7 dysplastic (3 stage D1 and 4 stage D2) Bladder Lesion Optical Multichannel Analyzer and 28 malignant areas, as well as 1 carcinoma in situ FIG. 1. Tissue fluorescence measuring setup. Output from nitro- included in our study. The 28 malignant lesions were stage gen laser is optically coupled to single quartz optical fiber, which is pTa grades 1(15), 2 (7) and 3 (I), stage pT1 grades 2 (1)and passed through working channel of standard cystoscope. Bladder tissue fluorescence, collected with same fiber, is o tically coupled to 3 (l),and stage pT2 grade 3 (3). Overall, 114 biopsies were polychromator and recorded with gated intensifiexdiode array opt,- Obtained from 53 Patients, for an average of 2.2 biopsies per cal multichannel analyzer system. N D , neutral density filter. Trig., Patient (range 1to 5). trigger. After processing the fluorescence spectra of the first 15
1599
AUTOFLUORESCENCE DIAGNOSIS OF BLADDER CANCER
patients some facts were obvious. The autofluorescence intensity of normal urothelium is up to 20 times greater than that of malignant lesions of the human bladder. Inflammatory areas also showed little or weak fluorescence but there were significant spectral differences between nonmalignant tissue and cancer. While inflammatory areas and normal tissue always had 2 fluorescence maxima at 385 and 455 nm., the typical spectrum of malignant lesions showed only 1 maximum at 455 nm. (fig. 2, A). No further statistically significant spectroscopic differences could be found at 300 to 800 nm. (data not shown). As a result of these initial data we chose the integrated fluorescence intensity from 300 to 700 nm. and the ratio of fluorescence intensities at 385 and 455 nm. (13851455)as 2 parameters to distinguish between normal urothelium and the different bladder lesions. The data for all 114 spectra are analyzed in this manner in figure 2, B, and indicate that the 13851455 ratio alone is sufficient to distinguish between nonmalignant and malignant tissue. The table shows this ratio for different bladder tissues. The mean values of the ratio found for inflammation and normal urothelium were significantly different (p <0.0001) from those for malignant lesions. In addition the mean value of the I385:1455 ratio seemed to decrease with progression of disease. Ratios <1.01 indicated malignant and >1.01 nonmalignant tissue. There were 44 macroscopically suspicious (hypervascularized, edematous, raised and so forth) lesions in our study. On pathological analysis 11 areas were malignant and 33 were nonmalignant. Except for 1 false-negative (micropapillary grade 1 tumor) and 2 false-positive (extensive cystitis glandularis) results, all lesions could be classified correctly using the 1385:1455 fluorescence intensity ratio (see table). A total of 36 visually normal appearing sites (normal urothelium confirmed by pathology) adjacent to the visible primary tumor or to suspicious lesions of the bladder was measured and could be classified correctly as nonmalignant with the fluorescence ratio (1385:1455).All 17 measured visible papillary tumors showed fluorescence ratios <1.01. There were 17 measurements performed on nonmalignant urothelium altered by previous treatment with bacillus Calmette-Guerin or radiation, with fluorescence ratios >1.01. To date the sensitivity, specificity, and positive and negative predictive values of the method for bladder cancer detection, using all 114 lesions, were 97,98, 93 and 99%, respectively. A differentiation of dysplastic bladder lesions (types D1 and D2) from normal urothelium was not possible. Furthermore, no significant differencesin autofluorescencecould be detected between different stages (T1 to T3) or grades (1 to 3) of cancer.
A
I
Results of autofluorescencemeasurements Bladder Areas (histologicalresults)
I3851455 Ratio*
Total No'Pts.
Mean-tSE
>1.01 All pts.
Transitional cell Ca Ca in situ Dysplasia Squamous metaplasia Inflammation Normal urothelium Total No. pts.
21 1 0 0
1 0
0.63 2 0 . W 0.61 2.53 2 0.40 2.50
I 1
2
33
0 -
42
30
84
2.29 -C 0.16 2.89 2 0.18
28 1
I 1 35
42 114
Pts. with suspicious (hyperuascularued,edematous, mised and so forth) bladder lesions 9 1 10 Transitional e l l Ca Ca in situ 1 0 1 Dysplasia 0 3 3 Mammation 2 22 24 0 6 6 Normal urothelium Total No. pts. 12 32 44 * The ratio of fluorescence intensities at 385 and 455 nm. (13853455)was chosen as the paramekr for distinguishing between Werent bladder lesions and normal tissue. t p <0.001 compared to normal and inflamed urothelium.
DISCUSSION
Numerous groups have used laser induced autofluorescence to diagnose cancer in different organs, such as the lung, colon, cervix, brain or breast.14-17 However, to our knowledge there have been only 2 studies of in vivo autofluorescence of human bladder tissue to date11.12 and no data on inflammatory bladder lesions have been reported. The similar appearance of inflamed tissue and the early stages of bladder cancer, such as carcinoma in situ, make it difficult for the urologist to distinguish between these lesions. Furthermore, some early malignant lesions are barely visible or invisible under white light. The goal of our study was to determine if changes in tissue autofluorescence accompanying tumor development are sufficient to serve as a new diagnostic tool for detection of early stages of bladder cancer. We were particularly interested in differences in autofluorescence between malignant and inflammatory lesions. Our study demonstrated that there were, indeed, differences in autofluorescence between malignant and nonmalignant tissues, including inflammatory lesions of the human bladder. With 337 nm. excitation there were 2 fluorescence maxima, 1 centered at 385 nm. and 1 at 455 nm. These maxima corresponded to the emission spectra of pure collagen (emission peak approximately 385 nm.) and NADH (emission peak approximately 455 nm.), and we tentatively identify these
6 0 0
+ +
50 300
350
4W
450
SW
550
Wavelength (nm)
WO
650
7W
0
I
1
3
4
5
6
7
Rat0 (l385/1455)
FIG. 2. A, typical fluorescence spectra of human bladder with 2 inflammatory areas and 1 malignant lesion. Note low fluorescence intensities of malignant and inflammatory tissue as well as missing 385 nm. peak t ~ c aof l cancer. AU spectra are corrected for background and system response. fj scatterplot of 2 fluorescence parameters of all 114 measurefiladder areas. Integrated fluorescence intensity (y axis) and ratio offluomscen& intensities at 385 and 455 nm. (xaxis, 1385A455)are plotted for different histological states. TCC,transitional cell -noma. Cis. carcinoma in situ.
1600
AUTOFLUORESCENCE DIAGNOSIS OF BLADDER CANCER
biomolecules as fluorophores of the bladder wall. The fluo- this region must weaken the collagen signal originating from rescence intensity of normal urothelium was always greater deeper layers by filtering this fluorescence. Therefore, we than that of malignant tissue, which agrees with data re- were able to recognize urothelial lesions characterized by an ported in the literature.10-12 While normal areas always increase in cell layers. To confirm our conclusions we found a showed 2 fluorescence maxima, malignant lesions had only tumor-like fluorescence signal of benign cystitis glandularis. 1 maximum centered at 455 nm. A second weak peak at Furthermore, there were no spectroscopic differences be385 nm. occasionally was found. tween well differentiated papillary tumors (grade 1, more We believe that the observed fluorescence changes are than 500 pm.) and muscle invasive dedifferentiated tumors largely due to optical effects of changes in the thickness of the (grade 3, more than 500 Fm.), as well as between normal urothelium as well as possibly to changes in blood supply tissue and dysplastic lesions (types D1 and D2). One would rather than to a changed metabolism of the tumor. During expect such differences in case of different fluorophore conour measurements the bladder was filled with water, thin- centrations due to a changed metabolism along with the ning the wall. The thickness of the urothelium in such a process of dedifferentiation. Detection of dysplastic lesions distended bladder wall is only 80 to 150 pm. It is known that using our method may be possible when there is an accom337 nm. light is able to excite colonic tissue layers as deep as panying hyperproliferation of urothelial cells. Only 1 carcinoma in situ with 5 to 10 cell layers was 500 pm.15 Therefore, we expect that fluorescence from the lamina propria and muscle layers of the bladder wall also is included in our study. The number of cell layers in such excited by a nitrogen laser. NADH is primarily located in the intra-urothelial cancer may vary from 1(or none, focally) to metabolically active layer of muscle cells but it is also found 20 (normal urothelium, 3 to 5 cell layers). The dedifferentiin the other bladder layers. In contrast to the lamina propria ated urothelium is characterized by larger nuclei combined and muscle, there is no collagen in the urothelium. Since with more cells per unit area.22 von Brunn’s nests, which are collagen is predominantly in the lamina propria, the in- rich in cells, are commonly involved.23 Future studies are creased thickness of urothelium could lead to the observed required to determine if these characteristics enable the deloss of the 385 nm. collagen peak in malignant lesions by tection of carcinoma in situ versus normal or inflammatory filtering of the autofluorescence from deeper layers. The in- tissue using the laser induced autofluorescence meascreased blood supply of small malignant lesions could cause urement system described. an additional decrease in fluorescence due to the broad abIt still must be addressed if confocal techniques (meassorption bands of (oxy)hemoglobin between 300 and 600 urements at 1focal point) or measurements on a microscopic nm.18 Our in vivo measurements do not allow us to determine level are able to distinguish different grades of cancer or the relative influence of blood absorption and thickness dif- dysplasia due to different concentrations of fluorophores by ferences. detecting fluorescence changes limited to the urothelium or Only 1 nonvisible micropapillary tumor showed a fluores- to the single cell, respectively. The use of other excitation cence spectrum like that of normal tissue. It is possible that wavelengths matching the absorption maxima of other fluthis spectrum occurred because the urologist pushed the thin orophores (for example tryptophan and porphyrins) may be 0.6 mm. diameter quartz optical fiber a bit too far, which appropriate to separate different grades of dedifferentiation. compressed or penetrated the tissue and led to a fluorescence The advantage of our autofluorescence based diagnostic signal originating from deeper layers. We could oRen dem- system is its ability to differentiate between difficult to diagonstrate this effect when measuring small visible tumors nose malignant and nonmalignant lesions, including benign (data not shown), which suggests that a noncontact imaging inflammatory sites, with a high reliability (see table). In 44 system would probably decrease interobserver variability cases the physician could not decide during cystoscopy if the and improve our method. suspicious area was malignant or nonmalignant. Subsequent Inflammation of the bladder is accompanied by an in- histological classification showed that 33 biopsies were obcreased supply of blood that, due to absorption at the excita- tained from nonmalignant tissue. The use of an autofluorestion wavelength, may lead to the low fluorescence signal from cence based system to diagnose malignant bladder lesions inflamed tissue. However, inflammatory lesions are not gen- will hopefully lead to more specific guided biopsies instead of erally characterized by the degree of urothelial hyperplasia random biopsies, decreasing the number of biopsies and, noted in malignant lesions of the bladder, which could be why therefore, the risk and cost to the patient. the collagen signal at 385 nm. was always detectable for The 1385:1455 ratio of 1.01 used as the parameter to sepinflammatory lesions except in 2 cases. Both exceptions were arate malignant from nonmalignant tissue was chosen emcases of extensive cystitis glandularis, 1of the various forms pirically. TOvalidate the value of 1.01 the methodology must of nonneoplastic urothelial proliferation of the bladder.19 The be applied on a different data set. Since it is not possible t o increased number of cells in such sometimes tumor-like le- scan the entire bladder using a fiber based system, future sions could act as a filter and explain the missing collagen work will be focused on using the autofluorescence data of signal. this study to develop a new noncontact imaging system. One biopsy was taken from a previously treated (transurethral resection) bladder area, which appeared macroscopiCONCLUSIONS cally as a scar. The laser induced autofluorescence measFrom the autofluorescence of bladder tissue excited by a urement showed the greatest fluorescence intensity at 385 nitrogen laser, it is possible to distinguish between malignm. in this study. On pathological analysis the lesion was characterized by normal urothelium and accompanying fibro- nant and nonmalignant areas with a high sensitivity and specificity. Since malignant lesions of the bladder are charsis, which would explain the increased collagen signal. Differences in concentrations of fluorophores between nor- acterized by a weak or missing peak at 385 nm. we are mal and dedifferentiated urothelial cells have been reported particularly able to discriminate malignant and inflammain the literature.20.21 To date there is no evidence that such tory lesions using the ratio of fluorescence intensities at 385 differences can be detected with our method. Using 337 nm. and 455 nm. To our knowledge this is the first in vivo method excitation and the 1385:1455 fluorescence intensity ratio we that differentiates malignant lesions from inflammatory are able to separate malignant areas from benign lesions due sites of the human bladder without use of sensitizing agents. to differences in structure and blood supply of tissue. Our REFERENCES system is expected to measure autofluorescence signals from bladder tissue layers UP to about 500 um. deep. Since there is 1. American Cancer Society: Cancer Facts & Fieures-1995. Atno collagen in the urothelium every increase i n cell layers in lanta: American Cancer Society, Ine., 1995. -
AUTOFLUORESCENCE DIAGNOSIS OF BLADDER CANCER 2. Greene, L. F., Hanash, K. A. and Farrow, G. M.: Benign papilloma or papillary carcinoma of the bladder? J. Urol., 110 205, 1973. 3. Kiemeney, L. A. L. M., Witjes, J. A., Heijbroek, R. P., Verbeek, A. L. M. and Debruyne, F. M. J.: Predictability of recurrent and progressive disease in individual patients with primary superficial bladder cancer. J. Urol., 150:60, 1993. 4. Heney, N. M., Ahmed, S., Flanagan, M. J., Frable, W., Corder, M. P., Hafermann, M. D. and Hawkins, I. R. for National Bladder Cancer Collaborative Group A. Superficial bladder cancer: progression and recurrence. J . Urol., 130 1083, 1983. 5. Althausen, A. F., Prout, G. P., Jr. and Daly, J . J.: Non-invasive papillary carcinoma of the bladder associated with carcinoma in situ. J. Urol., 116 575, 1976. 6. Wolf, H., Olsen, P. R., Fischer, A. and H~jgaard,K.: Urothelial atypia concomitant with primary bladder tumour. Incidence in a consecutive series of 500 unselected patients. Scand. J. Urol. Nephrol., 21: 33, 1987. 7. Kriegmair, M., Baumgartner, R., Kniichel, R., Stepp, H., Hofstadter, F. and Hofstetter, A.: Detection of early bladder cancer by 5-aminolevulinic acid induced porphyrin fluorescence. J. Urol., 156 105, 1996. 8. Jichlinski, P., Mizeret, J., Forrer, M., Wagniere, G., Van den Bergh, H., Schmidlin, F., Graber, P. and Leisinger, H. J.: Superficial bladder tumors. Pathological and clinical review and presentation of a new diagnostic method: fluorescence photodetection of transitional epithelial carcinomas based on protoporphyrin M induction with delta-aminolevulinic acid (5-ALA).Rev. Med. Suisse Romande, 115 233, 1995. 9. Mourant, J. R., Bigio, I. J., Boyer, J., Conn, R. L., Johnson, T. and Shimada, T.: Spectroscopic diagnosis of bladder cancer with elastic light scattering. Lasers Surg. Med., 17: 350, 1995. 10. Schomacker, K. T., Flotte, T. J . and Deutsch, T. F.: Detection of chemically-induced dysplasia in rat urinary bladder with laser-induced fluorescence. In: Advances in Laser and Light Spectroscopy to Diagnose Cancer and Other Diseases. Edited by R. R. Alfano. Proc. SPIE, vol. 2135, pp. 76-79, 1994. 11. D’Hallewin, M. A., Baert, L. and Vanherzeele, H.: Fluorescence imaging of bladder cancer. Acta Urol. Belg., 6 2 49, 1994. 12. Baert, L., Berg, R., Van Damme, B., D’Hallewin, M. A., Johansson, J., Svanberg, K. and Svanberg, S.: Clinical fluorescence diagnosis of human bladder carcinoma following lowdose Photofrin injection. Urology, 41: 322, 1993. 13. Campbell, M. J. and Machin, D.: Probability and decision making: diagnostic tests. In: Medical Statistics, 2nd ed. New York John Wiley & Sons, vol. 2, chapt. 3.2, pp. 3 4 3 5 , 1993. 14. Lam, S., MacAulay, C., Hung, J., LeRiche, J., Profio, A. E. and Palcic, B.: Detection of dysplasia and carcinoma in situ with a lung imaging fluorescence endoscope device. J. Thorac. Car-
1601
diovasc. Surg., 105 1035, 1993. 15. Schomacker. K. T.. Frisoli. J. K.. Compton. C. C.. Flotte, T. J., Richter, J: M., Nishioka,.N. S. .and Deutach, T. F.: Ultraviolet laser-induced fluorescence of colonic tissue: basic biology and diagnostic potential. Lasers Surg. Med., 12 63, 1992. 16. Ramanujam, N., Mitchell, M. F., Mahadevan, A., Thomsen, S., Silva, E. and Richards-Kortum,R.: Fluorescence spectroscopy: a diagnostic tool for cervical intraepithelial neoplasia (CIN). Gynec. Oncol., 5 2 31, 1994. 17. Fink, F., Paul, R. and Pfarrherr, A.: Laser fluorescence spectroscopic experiments for monitoring molecules in brain. Adv. Exp. Med. Biol., 333 233, 1993. 18. Boulnois, J.-L.: Photophysical processes in recent medical laser developments: a review. Laser Med. Sci., 1:47,1986. 19. Young, R. H.: Non-neoplastic epithelial abnormalities and tumorlike lesions. In: Pathology of the Urinary Bladder. Edited by L. M. Roth. New York Churchill Livingstone, vol. 13, pp. 1 3 5 , 1989. 20. Romanenko, A. M. and Mirgorodskaia, L. N.: Histochemical characteristics of oxidation-reduction enzymes in epithelial tumors of the urinary bladder. Vopr. Onkol., 17:43, 1971. 21. Vanderlaan, M., Fong, S. and King, E. B.: Histochemistry of NADH diaphorase and gamma-glutamyltranspptidasein rat bladder tumors. Carcinogenesis, 3 397, 1982. 22. Daly, J. J.: Carcinoma-in-situ of the urothelium. Urol. Clin. N. Amer., 3 87, 1976. 23. Hudson, M. A. and Herr, H. W.: Carcinoma in situ of the bladder. J. Urol., 153 564, 1995. EDITORIAL COMMENT The authors provide a detailed, elegant description of laser induced autofluorescence to differentiate between malignant and inflammatory lesions of the bladder. They demonstrate that the thickness of the urothelial layer is one of the most probable causes to explain the distinction. Here lies also the weakness of the methodology, since dysplasia from normal urothelium or differences in grade and stage could not be demonstrated. The obvious advantage to avoid unnecessary biopsies should be more firmly demonstrated in a larger series in which more than 44 visually suspicious lesions are present in 53 patients with 114 bladder areas analyzed. We hope that a confirmation of the results can be demonstrated in a selective series of primary carcinoma in situ. Louis J . Denis Department of Urology Koningen Elisabethlei Antwerp, Belgium