- Email: [email protected]
Emission Spectra of Colonic Tissue and Endogenous Fluorophores
Emission Spectra of Colonic Tissue and Endogenous Fluorophores BHASKAR BANERJEE, MD,* BRENT MIEDEMA, MD,t HOLAKARE R. CHANDRASEKHAR, PHD:j:
ABSTRACT: Autofluorescence emission spectra of normal, adenomatous, and malignant tissues of the colon were compared to that of known fluorophores to indicate the possible causes of tissue fluorescence. Data were collected from normal mucosa (n = 18), adenomatous polyps (n = 32), and adenocarcinoma (n = 18) of the colon. A range of cellular and extracellular fluorophores (elastin, collagen, flavin adenine dinucleotide, nicotinamide adenine dinucleotide, phenylalanine, pyridoxal 5' phosphate, tryptophan, and tyrosine) were similarly examined using a spectrofluorometer with emission and excitation spectrometers. Emission intensities were plotted against wavelength. Wavelengths of peak emission and the width of each peak at half its maximum intensity were measured. Colonic tissue gave four major emission peaks, the wavelengths of which were independent of tissue histology. Tryptophan and collagen type IV appeared to be responsible for two of the peaks. It is possible that NADH may be the cause of a third emission maxima. KEY INDEXING TERMS: Autofluorescence; Emission; Colon; Polyp; Cancer. [Am J Med Sci 1998;316(3):220226.]
D
espite the advances of flexible endoscopy, the diagnosis of gastrointestinal malignancy that arises in known premalignant disorders is often at a late stage. In conditions such as ulcerative colitis, there is a lack of visual markers for early malignancy and attempts at early diagnosis of neoplasia by surveillance endoscopy have met with limited success,
From the Departments of *Medicine, tSurgery, and +physics & Astronomy, University of Missouri-Columbia and Harry S Tru· man Memorial Veterans Hospital, Columbia, Missouri. Presented in part at the annual meeting of the Central Society for Clinical Research, Chicago, Illinois, September 1996. Correspondence: Bhaskar Banerjee, MD, Division of Gastroen· terology, MA 421, Medical Sciences Building, University of Missouri, Columbia, MO 65203.
220
despite a high economic cost.!,2 It has been estimatE that random biopsies taken blindly at four qua rants and every 10 cm sample less than 0.05% oftl entire surface area of the colon.! Additionally, tl dysplasia is often patchy, which leads to sampliI errors, and histologic interpretation is subject to i ter- and intraobserver variation. 2- 4 Tissue spectroscopy can use endogenous fluor phores (molecules normally present in tissues th fluoresce) to provide valuable objective informati( for certain disease states. Several spectroscopic m dalities have been investigated, including autofl orescence, elastic scattering, and Raman spectre copy, with the greatest amount of work performl with autofluorescence. 5 - 7 Fluorescence refers to brief pulse of light (or electromagnetic radiation) a higher wavelength than the excitation waveleng which is used to excite the tissue. Autofluorescen signifies the absence of any exogenous photoacti agents, such as hematoporphyrin derivatives, th are generally used in photodynamic therapy.8,9 uses low levels of radiation without tissue destn tion and has the potential to diagnose conditio such as cancer in real time. Autofluorescence h been used to distinguish normal from malignant t: sue in several organs, including colon, lung, breal and cervix. 5 ,10,11 In this study, systematic investigation of emissi, spectra in normal, dysplastic, and malignant color tissue was performed. Spectroscopic studies of range of known fluorophores were carried out wi the same apparatus. The aim ofthis work was not provide more data on the ability of autofluorescen spectra to detect neoplastic lesions of the colc which has been reported before, but rather to stu the shape and distribution of the spectral maxir (wavelengths at which emission peaks occurred) an attempt to identify possible endogenous tiss fluorophores. Whether all the emission peaks WE present in all histologic grades and if there was a shift in any emission peak associated with neopla: was investigated. A systematic study of known t sue fluorophores was made and the data concerni the position (wavelength) and width of emissi September 1998 Volume 316 Numbl
Banerjee, Miedema, and Chandrasekha
peaks were compared to help identify possible tissue fluorophores. Materials and Methods Tissue Collection. Tissue was obtained from pa-
tients undergoing routine colonoscopy for clinical indications and surgery for the resection of malignant tumors of the colon. Informed consent was obtained and the study was approved by the institutional review board. Cancerous and normal (matched controls) tissue samples were obtained from sections of colons resected at surgery. Resected colons were washed in normal saline and specimens obtained from the tumor mass and from an area of normalappearing mucosa about 10 cm away from the tumor edge. Polyps (pedunculated and 2-3 mm in diameter) were obtained at colonoscopy using snare and electrocautery on the stalks. Normal mucosa was not sampled during polypectomy because tissue removal with biopsy forceps would induce traumatic hemorrhage, which could affect the spectra. To minimize this, normal mucosa samples were obtained from surgical specimens. All tissues were immediately snap-frozen in liquid nitrogen and stored at -70 C. Normal mucosa, tubular adenomas, and adenocarcinoma of the colon were used to represent normal, dysplastic, and malignant tissue, respectively. Before measurement, tissues were thawed over ice to minimize any structural damage, and moistened with phosphate-buffered saline (PBS) at pH 7.4. Instrumentation and Measurements. A spectrofluorometer with a xenon lamp and excitation and emission spectrometers (RF-5301 PC, Shimatzu, Columbia, MD) was used. The polyps and the mucosal surfaces of the surgical specimens had approximately the same dimensions (3 mm X 3 mm, about 2-mm thick). The specimens were mounted on a glass slide, which was placed on a specially constructed solid tissue sample holder with a black surface, so that the mucosal surface of the specimens faced the excitation spectrometer. The excitation beam was provided by a 150 W xenon lamp, which was incident at 300 to the tissue. The accuracy of the excitation wavelength was ± 1.5 nm, with a wavelength range of 220 to 900 nm. The excitation beam produced a spot size of 10 mm X 1 mm. The slit width was 1.5 mm. Excitation intensity varied with wavelength but was always less than 5 /1W/mm2. Excitation spectra were obtained with emission at 350 nm and excitation from 240 nm to 340 nm. Emission spectra were obtained with excitation from 280 nm to 400 nm (no significant emission was detected by the instrument at excitations beyond this wavelength), at 10 nm increments. The emission was observed from 10 nm above the excitation wavelength to 10 nm less than twice the excitation wavelength. The normalized emission intensity (measured in arbitrary units) was plotted on the y axis and emission wave0
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
length (in nanometers) on the x axis. All data wai digitally stored for subsequent analysis. After spec troscopy, all tissue was fixed in 10% buffered forma lin. Tissue was subsequently embedded in paraffin and sectioned and stained in hematoxylin and eo sir prior to microscopic examination and histologic grading. Histologic analysis was blinded from al spectroscopic data. Data was obtained from H paired samples of normal mucosa and adenocarcino· mas of the colon (all moderately differentiated) Spectra from 35 adenomatous polyps were used (al tubular adenomas with mild to moderate atypia) Data from mixed polyps were not included in thi~ study. Fluorophores. A number of known endogenous flu· orophores suggested by other workers 5 ,1l-17 were in· vestigated and their spectra compared to that of co· Ionic tissue. Type IV collagen was studied becausE it is the type found in the basement membrane oj epithelia and therefore is more likely to be responsi· ble for the observed tissue fluorescence than other types of collagen that may be found in skin, tendon, bone, and cartilage. IS Type IV collagen is found in the basal lamina of epithelia and helps to form a meshwork of fine filaments that support the epithelia and provides a filtration barrier for macromolecules. 19 An acid-soluble form from the human placenta, designated basement-membrane type, was studied. 19 Excitation and emission spectra from elastin (from bovine neck ligament) was similarly measured. Spectral measurements were made from 1.2 mM solutions of ,B-nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD), and from 2-mM solutions of phenylalanine, tryptophan, tyrosine, and pyridoxal 5' phosphate. All chemicals were obtained from Sigma Chemical Company (St. Louis, MO). Results
Spectral data was collected from normal mucosa (n = 18), pedunculated adenomatous polyps (n = 32), and mild to moderately differentiated adenocarcinoma (n = 18). Each pair of normal mucosa and adenocarcinoma samples were obtained from single patients. The polyps were all tubular adenomas. Data from mixed polyps were not included in this study. Representative fluorescence spectra of adenomatous polyps are displayed in Figures 1 and 2. Some emission maxima were seen only at certain excitation wavelengths (eg, peak A was visible at excitation wavelengths of310 nm or less). Obtaining emission spectra by increasing the excitation beam at 10 nm increments ensured detection of all major maxima in the range studied. Four major emission peaks (A to D) were observed in all three tissue types (normal mucosa, adenoma221
Colonic Emission Spectra
-~----
- -I 1
10
other fluorophores in the range of wavelengths st ied (see Table 2). The measurement of fluorescence intensity , not the object of this study, but it was observed t the intensity of peak A increased progressively" dysplasia and carcinoma, whereas the intensit~ peak D decreased with higher grades of tis atypia. Similar changes in peaks Band C were noted. Discussion
---,-,-----,-1 ---
300
350
400
,-~
r450
,
~-
1-----'---T
500
550
-,-I 600
Wavelength in nm
Figure 1. Emission spectrum of an adenomatous polyp with excitation at 310 nm. The intensity of fluorescence is indicated in the y axis (in arbitrary units) with the wavelength of emission in the x axis. Emission peaks A, B, and D are indicated by arrows. Valleys are seen between A and B (at 350 nm) and between B and D (at 420 nm). A third valley is suggested by the depression following peak D at about 540 nm.
tous polyps, and adenocarcinoma), as depicted in Figures 1 and 2. The wavelength at which each emission maxima occurred was recorded for each tissue sample. The mean wavelength for each emission peak for the different histologic grades is given in Table 1. The position of each emission maxima was grouped according to type of tissue (normal mucosa, polyp, carcinoma). For each emission peak (A to D), the distribution of that emission maxima in each tissue type (normal, adenomatous, malignant) was analyzed by one-way analysis of variance; no significant differences were found (Table 1). The position of each emission maxima thus appeared to be unaffected by the histologic grade. Apart from the wavelength of maximum intensity, the width of each peak was evaluated by measuring the full width at half maximum intensity (FWHM). The mean FWHM (in nm) for the peaks A, B, C, and D (± SE) were: 53.8 ± 0.7, 9.5 ± 1.0, 40.4 ± 2.0, and 100.5 ± 6.0, respectively. The emission peaks ofthe fluorophores studied are given in Table 2. In each case, the optimal excitation range (in nm) that produced a given emission wavelength with the corresponding FWHM are given in Table 2. An example of the emission spectrum of tryptophan is given in Figure 3. Elastin produced two peaks: one at 365 nm, seen best when excited at 290 nm, and another at 390 nm, observed with excitation by wavelengths between 310 and 340 nm. Similarly, four emission maxima were produced by collagen type IV (three of which are displayed in Figure 4). Tyrosine and NADH (Figure 5) gave single maxima at 330 nm and 460 nm respectively. One or more emission peaks were observed for each of the
222
Autofluorescence has been used to distinguish I mal from dysplastic and malignant tissue 12 - 14 v a high degree of sensitivity (80%-100%) and sp ficity (92%-99%). Emission spectra can also be u to distinguish inflammatory from normal and ma nant mucosa of the gut. 20 The use of this techni to detect dysplasia and carcinoma has been veri in vivo by using fiberoptic probes to carry excita1 and emission beams through the biopsy chanm endoscopesY This technique has been used to er scopically collect images from surgical specimen localize areas of dysplasia. 15 Previous researcl have concentrated on the diagnostic application this technique, with a lesser emphasis on the po' tial causes of tissue fluorescence. 5,1l-16 A fur1 problem has been the different excitation Wl lengths used by different researchers, which m~ direct comparison difficult. 5,lO,12-15 Some stu have examined normal and dysplastic tissue,5 while others have examined normal, dysplastic, malignant lesions of the colon. 13 ,14 Difference! emission intensities at specific wavelengths werE lected to help identify different histologic gradE tissue, admittedly with high sensitivity and spec
20
T
~ c::
I
:::I
>-
~
:0
~
10 -
300
350
400
450
500
550
Wavelength in nm
Figure 2. Emission spectrum of an adenomatous polyp with tation at 350 nm. The intensity of fluorescence is indicated i y axis (in arbitrary units) with the wavelength of emission i x axis. Emission peaks B, C, and D are indicated by ar Valleys can be seen between peaks Band C (at about 37( and between peaks C and D (at about 420 nm). September 1998 Volume 316 Nun
Banerjee, Miedema, and Chandrasekhar
Table 1. Mean Wavelength at Which Maximal Intensity Occurs for Each Emission Peak (A-D) in Different Colonic Tissues Peak
Tissue
Wavelength
SE
n a c n a c n a c n a c
331.4 331.4 331.0 365.9 364.9 364.5 385.8 384.9 386.3 454.5 451.8 452.9
0.6 0.4 0.7 0.8 0.7 0.6 1.1 0.8 0.5 1.5 1.0 1.3
A B
C D
P Value
0.854 0.306 0.540 0.256
95% CI
Range (nm)
330.1-332.8 330.6-332.1 329.4-332.6 364.1-367.7 363.5-366.4 363.1-365.9 383.3-388.3 383.2-386.7 385.1-387.5 451.3-457.7 449.7-454.0 450.2-455.7
328-336 328-336 326-336 361-368 360-370 361-368 381-391 380-390 384-389 442-460 440-463 448-463
n = normal mucosa; a = adenomatous polyp; c = cancer. Standard error of mean (SE), 95% confidence intervals (CI), and range for the wavelengths in each group are indicated. The distribution of wavelength of each maxima in the three tissue types (eg, An, Aa, and Ac) showed no significant differences (see P values) by one way analysis of variance; thus the position of the four major emission peaks were not influenced by tissue histology.
ity, but without detailed analysis of the spectra, or the use of different excitation wavelengths. 5 ,13-l5 An in vitro study using a range of excitation wavelengths in colonic tissue has been reported, but did not include cancerous tissue. 16 Emission spectra were measured over a wide range of wavelengths (with excitation wavelengths at 10 nm increments) on samples of normal mucosa (n = 4) and adenomatous polyps (n = 11) of the colon. The data was presented in the form of excitation-emission matrices. 16 Excitation-emission matrices are graphic representations of fluorescence intensities as a function of excitation and emission wavelengths and were used to successfully distinguish normal from dysplastic tissues of the colon. 16 The spectral peaks obtained
by exciting normal and dysplastic colonic tissue with a wide range of excitation wavelengths were identified and possible tissue fluorophores discussed. In a further study involving fluorescence microscopy of normal mucosa (n = 10) and dysplastic colonic tissue (six adenomatous polyps of varying grades of epithelial dysplasia), fluorescence due to collagen and cytoplasmic granules was noted.17 Laser excitation at 351 to 365 nm was used and fluorescence was observed by using a series of barrier filters. 17 Although malignant tissue was not examined, greater fluorescence was reported in the cytoplasm of dysplastic cells of colonic adenomas compared to normal mucosal cells. 17 Fewer fluorescent connective tissue fibers were noted in the lamina propria of adenomas than in normal mucosa, with a lower fluorescence inten-
Table 2. Wavelengths of Emission Peaks of Known Fluorophores Studied
Fluorophore Elastin Collagen IV
FAD NADH Phenylalanine
Pyridoxal 5' Phosphate Tryptophan Tyrosine
Optimal Excitation (nm)
Position of Emission Peak (nm)
FWHM (nm)
290 310-340 290-300 310-350 330-350 330-380 290-520 290-400 300 300 350 350 300-350 380-400 290-320 290-350
365 390 345 365 392 430 530 460 335 375 400 445 447 485 330 364
20 103 64 12 28 125 55 89 10 87 10 95 40 24 45 23
'"
~
60
:::>
~
FWHM = full width at half maximum intensity. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
~ :c 40 a;
330 nm
.!:
.~
'"c:
20
CD
E 0 300
350
400
450
500
Wavelength in nm Figure 3. Emission spectrum of tryptophan with excitation at 290 nm. The intensity of fluorescence is displayed in the y axis and the wavelength of emission in the x axis. The peak at 330 nm is indicated by an open arrow. The full width at half maximum intensity (FWHM) of 45 nm is indicated by the horizontal arrows.
223
Colonic Emission Spectra
o
~
365 nm
7
:::I
>.... «S
.::.
:0 ....
/ 5
392 nm
~~430nm
~
300
350
400
450
500
550
600
Wavelength in nm Figure 4. Emission spectrum oftype IV collagen (basement membrane type) with excitation at 350 nm. The intensity of emission is displayed in the y axis in arbitrary units, with the wavelength of emission in the x axis. Three emission peaks are indicated by arrows with their corresponding wavelength of emission.
sity in adenomatous tissue compared to normal mucosa. 17 The ability of fluorescence emission spectra to distinguish normal from dysplastic and malignant tissues of the colon has been reported, but a better understanding of the causes of tissue fluorescence is essential to the development and use of optical diagnostic methods. Most workers have used an emission peak at about 460 nm for such work,5,12-14 even though several other emission peaks can be observed in normal and adenomatous tissue. 16 Tissue fluorescence appears to be both extracellular and intracellularY The cause of fluorescence remains unclear. It has been suggested that the fluorescence at about 460 nm - believed to be from nicotinamide adenine dinucleotide (NADH)-is lower in adenomatous than in normal tissue. 5,12-14 However, cytoplasmic fluorescence in dysplastic epithelia appeared to be more intense than normal mucosal cells on autofluorescence microscopy.17 Other reports have suggested the roles played by a number of fluorophores l4 ,16 without directly measuring the spectra ofthose fluorophores under the same experimental conditions. 16 Fluorophores such as collagen have been implicated l4 ,16 and the spectra measured,16 but there are several types of collagen and the type measured or implicated was not specified. 16 ,17 In one study, the spectra of collagen type I was discussed. 13 Type I collagen is found in skin, bone, tendon, fascia, and capsules of organs 18 and not in the intestinal tissues studied. If collagen contributes to colonic emission spectra, it would most likely be type IV collagen which is found in the basement membrane. 19 It is unlikely that collagen found in skin, tendon, bone, cartilage, and the walls of blood vessels is responsible for colonic emission spectra. 18
224
By comparing the emission spectra from colonic tissue and the fluorophores listed in Table 2, possible causes oftissue fluorescence may be surmised. However the fluorescence peaks may not exactly match the tissue maxima or band positions. Tissue fluorescence due to specific fluorophores may be attenuated due to absorption and scattering, which may cause slight changes in both the position and the line shape of the observed tissue fluorescence. 21 Attenuation of observed fluorescence may be due to absorption by molecules such as oxyhemoglobin (at 280 nm, 350 nm, 420 nm, 540 nm, and 580 nm) which may affect colonic fluorescence spectra. 16 The valleys seen at 350 nm, 420 nm, and 540 nm in this study may be due to oxyhemoglobin. Further changes in position and shape may be caused by similar or overlapping emission maxima from different fluorophores. Emission spectra is also dependent on the pH, temperature, and concentration of fluorophores within the tissue being measured. 22 Local hemorrhage may affect the emission spectra of colonic tissue. Variability was minimized by taking all samples of normal mucosa and tumor from fresh surgical specimens. The polyps were all pedunculated and were excised by using snare and electrocautery on the stalks of polyps, which causes minimal hemorrhage at the superior surfaces of polyps. Removal with hot biopsy forceps, as performed by others,13 would have resulted in substantial local trauma and tissue destruction at the time ofpolypectomy, particularly in sessile polyps. Removal of tissue with biopsy forceps could result in local hemorrhage with consequent attenuation of wavelengths due to absorption by oxyhemoglobin. 16 This was minimized in our study by careful snare polypectomy at
----~
460 nm
10
~
~
'", :::>
~
~
:c
5
~
.?;'in c:
.!!!
.s
0
r----r
300
350
400
450
500
550
600
Wavelength in nm
Figure 5. Emission spectrum of f3-nicotinamide adenine dinucleotide (NADH) with excitation at 350 nm. Intensity of fluorescence is given in the y axis in arbitrary units, with wavelength of emission in the x axis. A single emission peak with a maximum at 460 nm is indicated by the arrow. September 1998 Volume 316 Number 3
Banerjee, Miedema, and Chandrasekha
the stalks of pedunculated polyps. The similarity in the spectra of normal mucosa and tumor samples (obtained at surgery) and polyps (obtained endoscopically) supports this. Contribution from porphyrins is unlikely in this study because porphyrins produce emissions between 600 nm to 750 nm, which was not measured in this study.23 Care was taken in this study to reduce factors that may alter the spectra: tissue was snap-frozen immediately upon collection, thawed over ice to minimize the structural damage that may occur from rapid defrosting, and moistened to physiologic pH. It has been suggested that spectra of tissue in vitro differs from that in vivo. 13 This has been demonstrated in 2 adenomatous polyps and 2 samples of normal colonic mucosa, whose emission intensities were measured over 8 hours at room temperature. The intensity of fluorescence presumed to be from NADH was seen to decease significantly after 2 hours.13 Such decrease of fluorescence emission associated with other fluorophores (FAD, collagen) did not occur.12 This observation was based on a very small sample size and indicated changes in emission intensity and not position or width of emission maxima. 12 Change in emission intensity was not the object of this study. In this work, tissue was immediately snap-frozen and measurements completed within minutes after thawing; at no point was tissue left at room temperature for a long period of time. Therefore any alteration in the spectra due to the in vitro measurement would have been minimal. The four emission maxima (A, B, C, and D) were found in all three histologic grades. There were no specific emission spectra associated with any degree of neoplasia, that is, no spectral bands were associated with dysplasia or indeed cancer. This is in agreement with previous observations. 16 It is the relative contribution of these fluorophores in the different histologic grades that allow normal tissue to be distinguished from dysplastic and malignant tissue. The normal mucosa samples were flat, and the samples of adenocarcinoma had grossly irregular surfaces, while the polyps were not flat, but almost hemispherical. Nevertheless, no substantial differences in the patterns of emission spectra were found between the three. We thus conclude that although no specific fluorophore appears to be associated with any particular histologic grade of colonic tissue (in the range of wavelengths studied), the shape and overall architecture or geometry of colonic tissue does not appear to affect the position or line shape of the emitted fluorescence spectra. Thus, the observed differences in fluorescence intensity are due to the relative concentrations of fluorophores in the different grades of neoplasia. The more densely packed cells with a higher metabolic rate, greater nuclear material, and lower density of connective tissue (eg, THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
collagen) may explain some of the observed changes rather than tissue geometry. Noting these limitations and the attempts madE to minimize the changes in the measured spectra the colonic spectral data can be compared with tha1 of known fluorophores to indicate the possible causeE oftissue fluorescence. It can be said that tryptophar: (Figure 3) is a most likely cause of emission maxima A which occurs at about 331 nm. The FWHM of max· ima A (54 nm) is comparable to that of tryptophan (45 nm). The slight differences may be due to thE factors previously discussed. Tryptophan in solution was studied because undissolved tryptophan has a different emission spectrum. Tryptophan, as a free molecule or as a part of larger molecules such as cellular proteins, may be responsible for this emission peak. It produces strong fluorescence at about 330 nm and is one of the wavelengths found by Richards-Kortum et aP6 to be helpful in distinguishing normal from adenomatous tissues of the colon. There are a number of possibilities for peak B (maxima at 365.1 nm, FWHM 9.5 nm): elastin, collagen, and tyrosine. The 365 nm peak seen with elastin was observed with excitation at 290 nm, which did not reveal any maxima at 365 nm in colonic tissue. Tyrosine produced a band at 364 nm, but with a FWHM that is twice that in peak B. The most likely candidate for peak B is collagen type IV, which produced an emission peak at 365 nm (FWHM of 28 nm) when excited by wavelengths between 310 and 350 nm (Figure 4). This is also found in colonic tissue, which has a comparable FWHM (the difference in FWHM may be due to absorption by other molecules, such as oxyhemoglobin). Collagen is not a single protein but a family of closely related molecules with differences in amino acid composition and sequence. 18 Twelve types of collagen have been identified and are found in organs as diverse as skin, bone, tendon, cartilage, uterus, dentin, vitreous body, and blood vessel walls. 18 The type of collagen discussed by other investigators was not specified. 12 .14,16,17 Collagen type IV or basement-membrane-type collagen is specialized and restricted to the basal lamina of epithelia 18; with laminin and heparan sulphate proteoglycan, it helps to form a meshwork of filaments that supports the epithelia and provides a barrier for the selective filtration of macromolecules. 19 Type IV collagen was thus studied and appeared to be responsible for the emission seen at 365 nm. For maxima C (385.7 nm, with FWHM of 40.4), elastin (peak at 390 nm with FWHM 103 nm), phenylalanine (peak 375 nm , FWHM 87 nm), and collagen type IV (peak at 392 nm with FWHM 28 nm) are possible fluorophores. Of these, collagen (Figure 4) is the more likely candidate, particularly because the tissue emission peak is seen best with excitation between 330 nm and 350 nm. The FWHM of the emission peak from type IV collagen (28 nm) is, how-
225
Colonic Emission Spectra
ever, smaller than that observed for peak C in the colon (40 nm). This may have been due to the effect of other fluorophores or of in vitro measurements. Indeed, this emission peak may be due to an undetermined fluorophore. The main possibility for emission D (453.1. nm, FWHM 100.5 nm) appeared to be NADH (Figure 5). This has been suggested by other workers. 5,12-14 However, the mean position of this band seen in colonic tissues (453.1 nm) is considerably different from that of NADH (460 nm). The wavelengths at which this peak occurred ranged from 440 nm to 463 nm (Table 1). The mean FWHM of emission peak D was 100.5 nm, which was considerably greater than that measured in NADH (89 nm). Allowing for minor changes in fluorescence due to this in vitro study, our data does not fully support the view that emission peak D is solely from NADH. Other fluorophores may be contributing to this band, including phenylalanine, pyridoxal 5' phosphate and collagen type IV (see Table 2). This particular band has been found to fluoresce less intensely in dysplastic than in normal mucosa. 12,13 Attributing all fluorescence at about 460 nm to NADH must be treated with caution. Indirect characterizations can only suggest but not prove the identity of a particular emission peak, particularly in complex materials such as biological tissue. Our work did not indicate any specific maxima due to elastin, FAD, phenylalanine, pyridoxal 5' phosphate, or tyrosine. Some of these may contribute to the emission spectra studied but were not solely responsible for any of the isolated major emission peaks. It must be emphasized that the characterizations described here are, at best, approximations; they are not direct measurements. Many factors are at play, including tissue physiology and the interactions of multiple fluorophores with similar spectra. Further work must be done to elucidate the agents responsible for tissue fluorescence. References 1. Cello JP, Schneiderman DJ. Ulcerative colitis. In: Sleisenger MH, Fordtran JS, eds. Gastrointestinal Disease. Philadelphia: WB Saunders; 1991:1435-77. 2. Collins RH Jr, Feldman M, Fordtran JS. Colon cancer, dysplasia and surveillance in patients with ulcerative colitis: a critical review. N Engl J Med. 1987;316:1654-8. 3. Jensen P, Krogsgaard MR, Christiansen J, Braendstrup 0, Johansen A, Olsen J. Observer variability in the assessment of type and dysplasia of colorecta1 adenomas, analyzed using kappa statistics. Dis Colon Rectum. 1995;38:195-8. 4. Rosenstock E, Farmer RG, Petras R, Sivak MY Jr, Rankin GB, Sullivan BB. Surveillance for colonic carcinoma in ulcerative colitis. Gastroenterology. 1985;89:1342-6.
226
5. Kapadia CR, Cutruzzola WF, O'Brien KM, Stetz ML Enriquez R, Deckelbaum. LI. Laser-induced fluorescence spectroscopy of human colonic mucosa: detection of adenoma tous transformation. Gastroenterology. 1990;90:150-7. 6. Mourant JR, Boyer JD, Johnson TM, Lacey JA, Bigic IJ. Detection of gastrointestinal cancer by elastic scatterinl and absorption spectroscopies with the Los Alamos Optica Biopsy System. SPIE Proc. 1995;2387:210-7. 7. Feld MS, Manoharan R, Salenius J. Detection and charac terization of human tissue lesions with near infrared Ramal spectroscopy. SPIE Proc. 1995;2388:99-104. 8. Bjorkman DJ, Samowitz WS, Brigham EJ. Fluorescenc, localization of early colonic cancer in the rat by hematoporpb yrin derivative. Lasers Surg Med. 1991;11;263-70. 9. Dougherty TJ, Marcus SL. Photodynamic therapy. Eur . Cancer. 1992; 28A:1734-42. 10. Alfano RR, Tang GC, Pradhan A, Lam W, Choy D, Ophe E. Fluorescence spectra from cancerous human breast an lung tissue. IEEE J Quantum Electron. 1987;QE;23:1806
11. 11. Mahadevan A, Mitchell MF, Silva E, et aI. Study of th fluorescence properties of normal and neoplastic human cel vical tissue. Lasers Surg Med. 1993;13:647-55. 12. Cothren RM, Richards-Kortum. R, Sivak MY, et al. GaE trointestinal tissue diagnosis by laser induced tissue diagn( sis at endoscopy. Gastrointest Endosc. 1990; 36: 105-11. 13. Schomacker KT, Frisoli JK, Compton CC, Flotte T~ Richter JM, Nishioka NS, et aI. Ultraviolet laser-induce fluorescence of colonic tissue: basic biology and diagnosti potential. Lasers Surg Med. 1992; 12:63-8. 14. Schomacker KT, Frisoli JK, Compton CC, Flotte T. Richter JM, Deutsh TF, et aI. Ultraviolet laser induce fluorescence of colon polyps. Gastroenterology. 1992; 10: 1155-60. 15. Wang TD, Van Dam J, Crawford JM, Preisinger E..! Wang Y, Feld MS. Fluorescence endoscopic imaging of hl man colonic adenomas. Gastroenterology. 1996; 111:1182 91. 16. Richards-Kortum R, Rava RP, Petras RP, Fitzmauric M, Sivak M, Feld MS. Spectroscopic diagnosis of colon dysplasia. Photochem Photobiol. 1991; 53:777 -86. 17. Romer TJ, Fitzmaurice M, Cothren RM, Richards-Ko; tum. R, Petras R, Sivak MY, Kramer JR. Laser-inducE fluorescence microscopy of normal colon and dysplasia in c· Ionic adenomas: implications for spectroscopic diagnosis. A J Gastroenterol. 1995; 90:81-7. 18. Fawcett DW. A Textbook of Histology. New York: Cha; man & Hall; 1994:133-9. 19. Bailey AJ, Sims TJ, Duance VC, Light ND. Partial cha acterization of a second basement membrane collagen in hman placenta. FEBS. 1979;99:361-6. 20. Banerjee B, Gadde P, Tahiri A, Mazumdar MK. Arg( laser induced fluorescence spectroscopy in gastrointestin malignancy and inflammation. Am J Gastroenterol. 199 87:1335. 21. Keijzer M, Richards-Kortum. R, Jacques SL, Feld M Fluorescence spectroscopy of turbid media: autofluorescen of human aorta. Appl Optics. 1989;28:4286-92. 22. Pesce AJ, Rosen CG, Pasby TL. Fluorescence Spectroscop An Introduction for Biology and Medicine. New York: Marc Dekker; 1971:65-86. 23. Svanberg K, Clemente LP, Clemente MP, Wang Warloe T, Anderson-Engels S. Pharmacokinetic studies 8-aminolevulinic acid-induced protoporphyrin IX build-up some malignant tumors. SPIE Proc. 1995;2387:30-42.
September 1998 Volume 316 Numbe
ABSTRACT: Autofluorescence emission spectra of normal, adenomatous, and malignant tissues of the colon were compared to that of known fluorophores to indicate the possible causes of tissue fluorescence. Data were collected from normal mucosa (n = 18), adenomatous polyps (n = 32), and adenocarcinoma (n = 18) of the colon. A range of cellular and extracellular fluorophores (elastin, collagen, flavin adenine dinucleotide, nicotinamide adenine dinucleotide, phenylalanine, pyridoxal 5' phosphate, tryptophan, and tyrosine) were similarly examined using a spectrofluorometer with emission and excitation spectrometers. Emission intensities were plotted against wavelength. Wavelengths of peak emission and the width of each peak at half its maximum intensity were measured. Colonic tissue gave four major emission peaks, the wavelengths of which were independent of tissue histology. Tryptophan and collagen type IV appeared to be responsible for two of the peaks. It is possible that NADH may be the cause of a third emission maxima. KEY INDEXING TERMS: Autofluorescence; Emission; Colon; Polyp; Cancer. [Am J Med Sci 1998;316(3):220226.]
D
espite the advances of flexible endoscopy, the diagnosis of gastrointestinal malignancy that arises in known premalignant disorders is often at a late stage. In conditions such as ulcerative colitis, there is a lack of visual markers for early malignancy and attempts at early diagnosis of neoplasia by surveillance endoscopy have met with limited success,
From the Departments of *Medicine, tSurgery, and +physics & Astronomy, University of Missouri-Columbia and Harry S Tru· man Memorial Veterans Hospital, Columbia, Missouri. Presented in part at the annual meeting of the Central Society for Clinical Research, Chicago, Illinois, September 1996. Correspondence: Bhaskar Banerjee, MD, Division of Gastroen· terology, MA 421, Medical Sciences Building, University of Missouri, Columbia, MO 65203.
220
despite a high economic cost.!,2 It has been estimatE that random biopsies taken blindly at four qua rants and every 10 cm sample less than 0.05% oftl entire surface area of the colon.! Additionally, tl dysplasia is often patchy, which leads to sampliI errors, and histologic interpretation is subject to i ter- and intraobserver variation. 2- 4 Tissue spectroscopy can use endogenous fluor phores (molecules normally present in tissues th fluoresce) to provide valuable objective informati( for certain disease states. Several spectroscopic m dalities have been investigated, including autofl orescence, elastic scattering, and Raman spectre copy, with the greatest amount of work performl with autofluorescence. 5 - 7 Fluorescence refers to brief pulse of light (or electromagnetic radiation) a higher wavelength than the excitation waveleng which is used to excite the tissue. Autofluorescen signifies the absence of any exogenous photoacti agents, such as hematoporphyrin derivatives, th are generally used in photodynamic therapy.8,9 uses low levels of radiation without tissue destn tion and has the potential to diagnose conditio such as cancer in real time. Autofluorescence h been used to distinguish normal from malignant t: sue in several organs, including colon, lung, breal and cervix. 5 ,10,11 In this study, systematic investigation of emissi, spectra in normal, dysplastic, and malignant color tissue was performed. Spectroscopic studies of range of known fluorophores were carried out wi the same apparatus. The aim ofthis work was not provide more data on the ability of autofluorescen spectra to detect neoplastic lesions of the colc which has been reported before, but rather to stu the shape and distribution of the spectral maxir (wavelengths at which emission peaks occurred) an attempt to identify possible endogenous tiss fluorophores. Whether all the emission peaks WE present in all histologic grades and if there was a shift in any emission peak associated with neopla: was investigated. A systematic study of known t sue fluorophores was made and the data concerni the position (wavelength) and width of emissi September 1998 Volume 316 Numbl
Banerjee, Miedema, and Chandrasekha
peaks were compared to help identify possible tissue fluorophores. Materials and Methods Tissue Collection. Tissue was obtained from pa-
tients undergoing routine colonoscopy for clinical indications and surgery for the resection of malignant tumors of the colon. Informed consent was obtained and the study was approved by the institutional review board. Cancerous and normal (matched controls) tissue samples were obtained from sections of colons resected at surgery. Resected colons were washed in normal saline and specimens obtained from the tumor mass and from an area of normalappearing mucosa about 10 cm away from the tumor edge. Polyps (pedunculated and 2-3 mm in diameter) were obtained at colonoscopy using snare and electrocautery on the stalks. Normal mucosa was not sampled during polypectomy because tissue removal with biopsy forceps would induce traumatic hemorrhage, which could affect the spectra. To minimize this, normal mucosa samples were obtained from surgical specimens. All tissues were immediately snap-frozen in liquid nitrogen and stored at -70 C. Normal mucosa, tubular adenomas, and adenocarcinoma of the colon were used to represent normal, dysplastic, and malignant tissue, respectively. Before measurement, tissues were thawed over ice to minimize any structural damage, and moistened with phosphate-buffered saline (PBS) at pH 7.4. Instrumentation and Measurements. A spectrofluorometer with a xenon lamp and excitation and emission spectrometers (RF-5301 PC, Shimatzu, Columbia, MD) was used. The polyps and the mucosal surfaces of the surgical specimens had approximately the same dimensions (3 mm X 3 mm, about 2-mm thick). The specimens were mounted on a glass slide, which was placed on a specially constructed solid tissue sample holder with a black surface, so that the mucosal surface of the specimens faced the excitation spectrometer. The excitation beam was provided by a 150 W xenon lamp, which was incident at 300 to the tissue. The accuracy of the excitation wavelength was ± 1.5 nm, with a wavelength range of 220 to 900 nm. The excitation beam produced a spot size of 10 mm X 1 mm. The slit width was 1.5 mm. Excitation intensity varied with wavelength but was always less than 5 /1W/mm2. Excitation spectra were obtained with emission at 350 nm and excitation from 240 nm to 340 nm. Emission spectra were obtained with excitation from 280 nm to 400 nm (no significant emission was detected by the instrument at excitations beyond this wavelength), at 10 nm increments. The emission was observed from 10 nm above the excitation wavelength to 10 nm less than twice the excitation wavelength. The normalized emission intensity (measured in arbitrary units) was plotted on the y axis and emission wave0
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
length (in nanometers) on the x axis. All data wai digitally stored for subsequent analysis. After spec troscopy, all tissue was fixed in 10% buffered forma lin. Tissue was subsequently embedded in paraffin and sectioned and stained in hematoxylin and eo sir prior to microscopic examination and histologic grading. Histologic analysis was blinded from al spectroscopic data. Data was obtained from H paired samples of normal mucosa and adenocarcino· mas of the colon (all moderately differentiated) Spectra from 35 adenomatous polyps were used (al tubular adenomas with mild to moderate atypia) Data from mixed polyps were not included in thi~ study. Fluorophores. A number of known endogenous flu· orophores suggested by other workers 5 ,1l-17 were in· vestigated and their spectra compared to that of co· Ionic tissue. Type IV collagen was studied becausE it is the type found in the basement membrane oj epithelia and therefore is more likely to be responsi· ble for the observed tissue fluorescence than other types of collagen that may be found in skin, tendon, bone, and cartilage. IS Type IV collagen is found in the basal lamina of epithelia and helps to form a meshwork of fine filaments that support the epithelia and provides a filtration barrier for macromolecules. 19 An acid-soluble form from the human placenta, designated basement-membrane type, was studied. 19 Excitation and emission spectra from elastin (from bovine neck ligament) was similarly measured. Spectral measurements were made from 1.2 mM solutions of ,B-nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD), and from 2-mM solutions of phenylalanine, tryptophan, tyrosine, and pyridoxal 5' phosphate. All chemicals were obtained from Sigma Chemical Company (St. Louis, MO). Results
Spectral data was collected from normal mucosa (n = 18), pedunculated adenomatous polyps (n = 32), and mild to moderately differentiated adenocarcinoma (n = 18). Each pair of normal mucosa and adenocarcinoma samples were obtained from single patients. The polyps were all tubular adenomas. Data from mixed polyps were not included in this study. Representative fluorescence spectra of adenomatous polyps are displayed in Figures 1 and 2. Some emission maxima were seen only at certain excitation wavelengths (eg, peak A was visible at excitation wavelengths of310 nm or less). Obtaining emission spectra by increasing the excitation beam at 10 nm increments ensured detection of all major maxima in the range studied. Four major emission peaks (A to D) were observed in all three tissue types (normal mucosa, adenoma221
Colonic Emission Spectra
-~----
- -I 1
10
other fluorophores in the range of wavelengths st ied (see Table 2). The measurement of fluorescence intensity , not the object of this study, but it was observed t the intensity of peak A increased progressively" dysplasia and carcinoma, whereas the intensit~ peak D decreased with higher grades of tis atypia. Similar changes in peaks Band C were noted. Discussion
---,-,-----,-1 ---
300
350
400
,-~
r450
,
~-
1-----'---T
500
550
-,-I 600
Wavelength in nm
Figure 1. Emission spectrum of an adenomatous polyp with excitation at 310 nm. The intensity of fluorescence is indicated in the y axis (in arbitrary units) with the wavelength of emission in the x axis. Emission peaks A, B, and D are indicated by arrows. Valleys are seen between A and B (at 350 nm) and between B and D (at 420 nm). A third valley is suggested by the depression following peak D at about 540 nm.
tous polyps, and adenocarcinoma), as depicted in Figures 1 and 2. The wavelength at which each emission maxima occurred was recorded for each tissue sample. The mean wavelength for each emission peak for the different histologic grades is given in Table 1. The position of each emission maxima was grouped according to type of tissue (normal mucosa, polyp, carcinoma). For each emission peak (A to D), the distribution of that emission maxima in each tissue type (normal, adenomatous, malignant) was analyzed by one-way analysis of variance; no significant differences were found (Table 1). The position of each emission maxima thus appeared to be unaffected by the histologic grade. Apart from the wavelength of maximum intensity, the width of each peak was evaluated by measuring the full width at half maximum intensity (FWHM). The mean FWHM (in nm) for the peaks A, B, C, and D (± SE) were: 53.8 ± 0.7, 9.5 ± 1.0, 40.4 ± 2.0, and 100.5 ± 6.0, respectively. The emission peaks ofthe fluorophores studied are given in Table 2. In each case, the optimal excitation range (in nm) that produced a given emission wavelength with the corresponding FWHM are given in Table 2. An example of the emission spectrum of tryptophan is given in Figure 3. Elastin produced two peaks: one at 365 nm, seen best when excited at 290 nm, and another at 390 nm, observed with excitation by wavelengths between 310 and 340 nm. Similarly, four emission maxima were produced by collagen type IV (three of which are displayed in Figure 4). Tyrosine and NADH (Figure 5) gave single maxima at 330 nm and 460 nm respectively. One or more emission peaks were observed for each of the
222
Autofluorescence has been used to distinguish I mal from dysplastic and malignant tissue 12 - 14 v a high degree of sensitivity (80%-100%) and sp ficity (92%-99%). Emission spectra can also be u to distinguish inflammatory from normal and ma nant mucosa of the gut. 20 The use of this techni to detect dysplasia and carcinoma has been veri in vivo by using fiberoptic probes to carry excita1 and emission beams through the biopsy chanm endoscopesY This technique has been used to er scopically collect images from surgical specimen localize areas of dysplasia. 15 Previous researcl have concentrated on the diagnostic application this technique, with a lesser emphasis on the po' tial causes of tissue fluorescence. 5,1l-16 A fur1 problem has been the different excitation Wl lengths used by different researchers, which m~ direct comparison difficult. 5,lO,12-15 Some stu have examined normal and dysplastic tissue,5 while others have examined normal, dysplastic, malignant lesions of the colon. 13 ,14 Difference! emission intensities at specific wavelengths werE lected to help identify different histologic gradE tissue, admittedly with high sensitivity and spec
20
T
~ c::
I
:::I
>-
~
:0
~
10 -
300
350
400
450
500
550
Wavelength in nm
Figure 2. Emission spectrum of an adenomatous polyp with tation at 350 nm. The intensity of fluorescence is indicated i y axis (in arbitrary units) with the wavelength of emission i x axis. Emission peaks B, C, and D are indicated by ar Valleys can be seen between peaks Band C (at about 37( and between peaks C and D (at about 420 nm). September 1998 Volume 316 Nun
Banerjee, Miedema, and Chandrasekhar
Table 1. Mean Wavelength at Which Maximal Intensity Occurs for Each Emission Peak (A-D) in Different Colonic Tissues Peak
Tissue
Wavelength
SE
n a c n a c n a c n a c
331.4 331.4 331.0 365.9 364.9 364.5 385.8 384.9 386.3 454.5 451.8 452.9
0.6 0.4 0.7 0.8 0.7 0.6 1.1 0.8 0.5 1.5 1.0 1.3
A B
C D
P Value
0.854 0.306 0.540 0.256
95% CI
Range (nm)
330.1-332.8 330.6-332.1 329.4-332.6 364.1-367.7 363.5-366.4 363.1-365.9 383.3-388.3 383.2-386.7 385.1-387.5 451.3-457.7 449.7-454.0 450.2-455.7
328-336 328-336 326-336 361-368 360-370 361-368 381-391 380-390 384-389 442-460 440-463 448-463
n = normal mucosa; a = adenomatous polyp; c = cancer. Standard error of mean (SE), 95% confidence intervals (CI), and range for the wavelengths in each group are indicated. The distribution of wavelength of each maxima in the three tissue types (eg, An, Aa, and Ac) showed no significant differences (see P values) by one way analysis of variance; thus the position of the four major emission peaks were not influenced by tissue histology.
ity, but without detailed analysis of the spectra, or the use of different excitation wavelengths. 5 ,13-l5 An in vitro study using a range of excitation wavelengths in colonic tissue has been reported, but did not include cancerous tissue. 16 Emission spectra were measured over a wide range of wavelengths (with excitation wavelengths at 10 nm increments) on samples of normal mucosa (n = 4) and adenomatous polyps (n = 11) of the colon. The data was presented in the form of excitation-emission matrices. 16 Excitation-emission matrices are graphic representations of fluorescence intensities as a function of excitation and emission wavelengths and were used to successfully distinguish normal from dysplastic tissues of the colon. 16 The spectral peaks obtained
by exciting normal and dysplastic colonic tissue with a wide range of excitation wavelengths were identified and possible tissue fluorophores discussed. In a further study involving fluorescence microscopy of normal mucosa (n = 10) and dysplastic colonic tissue (six adenomatous polyps of varying grades of epithelial dysplasia), fluorescence due to collagen and cytoplasmic granules was noted.17 Laser excitation at 351 to 365 nm was used and fluorescence was observed by using a series of barrier filters. 17 Although malignant tissue was not examined, greater fluorescence was reported in the cytoplasm of dysplastic cells of colonic adenomas compared to normal mucosal cells. 17 Fewer fluorescent connective tissue fibers were noted in the lamina propria of adenomas than in normal mucosa, with a lower fluorescence inten-
Table 2. Wavelengths of Emission Peaks of Known Fluorophores Studied
Fluorophore Elastin Collagen IV
FAD NADH Phenylalanine
Pyridoxal 5' Phosphate Tryptophan Tyrosine
Optimal Excitation (nm)
Position of Emission Peak (nm)
FWHM (nm)
290 310-340 290-300 310-350 330-350 330-380 290-520 290-400 300 300 350 350 300-350 380-400 290-320 290-350
365 390 345 365 392 430 530 460 335 375 400 445 447 485 330 364
20 103 64 12 28 125 55 89 10 87 10 95 40 24 45 23
'"
~
60
:::>
~
FWHM = full width at half maximum intensity. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
~ :c 40 a;
330 nm
.!:
.~
'"c:
20
CD
E 0 300
350
400
450
500
Wavelength in nm Figure 3. Emission spectrum of tryptophan with excitation at 290 nm. The intensity of fluorescence is displayed in the y axis and the wavelength of emission in the x axis. The peak at 330 nm is indicated by an open arrow. The full width at half maximum intensity (FWHM) of 45 nm is indicated by the horizontal arrows.
223
Colonic Emission Spectra
o
~
365 nm
7
:::I
>.... «S
.::.
:0 ....
/ 5
392 nm
~~430nm
~
300
350
400
450
500
550
600
Wavelength in nm Figure 4. Emission spectrum oftype IV collagen (basement membrane type) with excitation at 350 nm. The intensity of emission is displayed in the y axis in arbitrary units, with the wavelength of emission in the x axis. Three emission peaks are indicated by arrows with their corresponding wavelength of emission.
sity in adenomatous tissue compared to normal mucosa. 17 The ability of fluorescence emission spectra to distinguish normal from dysplastic and malignant tissues of the colon has been reported, but a better understanding of the causes of tissue fluorescence is essential to the development and use of optical diagnostic methods. Most workers have used an emission peak at about 460 nm for such work,5,12-14 even though several other emission peaks can be observed in normal and adenomatous tissue. 16 Tissue fluorescence appears to be both extracellular and intracellularY The cause of fluorescence remains unclear. It has been suggested that the fluorescence at about 460 nm - believed to be from nicotinamide adenine dinucleotide (NADH)-is lower in adenomatous than in normal tissue. 5,12-14 However, cytoplasmic fluorescence in dysplastic epithelia appeared to be more intense than normal mucosal cells on autofluorescence microscopy.17 Other reports have suggested the roles played by a number of fluorophores l4 ,16 without directly measuring the spectra ofthose fluorophores under the same experimental conditions. 16 Fluorophores such as collagen have been implicated l4 ,16 and the spectra measured,16 but there are several types of collagen and the type measured or implicated was not specified. 16 ,17 In one study, the spectra of collagen type I was discussed. 13 Type I collagen is found in skin, bone, tendon, fascia, and capsules of organs 18 and not in the intestinal tissues studied. If collagen contributes to colonic emission spectra, it would most likely be type IV collagen which is found in the basement membrane. 19 It is unlikely that collagen found in skin, tendon, bone, cartilage, and the walls of blood vessels is responsible for colonic emission spectra. 18
224
By comparing the emission spectra from colonic tissue and the fluorophores listed in Table 2, possible causes oftissue fluorescence may be surmised. However the fluorescence peaks may not exactly match the tissue maxima or band positions. Tissue fluorescence due to specific fluorophores may be attenuated due to absorption and scattering, which may cause slight changes in both the position and the line shape of the observed tissue fluorescence. 21 Attenuation of observed fluorescence may be due to absorption by molecules such as oxyhemoglobin (at 280 nm, 350 nm, 420 nm, 540 nm, and 580 nm) which may affect colonic fluorescence spectra. 16 The valleys seen at 350 nm, 420 nm, and 540 nm in this study may be due to oxyhemoglobin. Further changes in position and shape may be caused by similar or overlapping emission maxima from different fluorophores. Emission spectra is also dependent on the pH, temperature, and concentration of fluorophores within the tissue being measured. 22 Local hemorrhage may affect the emission spectra of colonic tissue. Variability was minimized by taking all samples of normal mucosa and tumor from fresh surgical specimens. The polyps were all pedunculated and were excised by using snare and electrocautery on the stalks of polyps, which causes minimal hemorrhage at the superior surfaces of polyps. Removal with hot biopsy forceps, as performed by others,13 would have resulted in substantial local trauma and tissue destruction at the time ofpolypectomy, particularly in sessile polyps. Removal of tissue with biopsy forceps could result in local hemorrhage with consequent attenuation of wavelengths due to absorption by oxyhemoglobin. 16 This was minimized in our study by careful snare polypectomy at
----~
460 nm
10
~
~
'", :::>
~
~
:c
5
~
.?;'in c:
.!!!
.s
0
r----r
300
350
400
450
500
550
600
Wavelength in nm
Figure 5. Emission spectrum of f3-nicotinamide adenine dinucleotide (NADH) with excitation at 350 nm. Intensity of fluorescence is given in the y axis in arbitrary units, with wavelength of emission in the x axis. A single emission peak with a maximum at 460 nm is indicated by the arrow. September 1998 Volume 316 Number 3
Banerjee, Miedema, and Chandrasekha
the stalks of pedunculated polyps. The similarity in the spectra of normal mucosa and tumor samples (obtained at surgery) and polyps (obtained endoscopically) supports this. Contribution from porphyrins is unlikely in this study because porphyrins produce emissions between 600 nm to 750 nm, which was not measured in this study.23 Care was taken in this study to reduce factors that may alter the spectra: tissue was snap-frozen immediately upon collection, thawed over ice to minimize the structural damage that may occur from rapid defrosting, and moistened to physiologic pH. It has been suggested that spectra of tissue in vitro differs from that in vivo. 13 This has been demonstrated in 2 adenomatous polyps and 2 samples of normal colonic mucosa, whose emission intensities were measured over 8 hours at room temperature. The intensity of fluorescence presumed to be from NADH was seen to decease significantly after 2 hours.13 Such decrease of fluorescence emission associated with other fluorophores (FAD, collagen) did not occur.12 This observation was based on a very small sample size and indicated changes in emission intensity and not position or width of emission maxima. 12 Change in emission intensity was not the object of this study. In this work, tissue was immediately snap-frozen and measurements completed within minutes after thawing; at no point was tissue left at room temperature for a long period of time. Therefore any alteration in the spectra due to the in vitro measurement would have been minimal. The four emission maxima (A, B, C, and D) were found in all three histologic grades. There were no specific emission spectra associated with any degree of neoplasia, that is, no spectral bands were associated with dysplasia or indeed cancer. This is in agreement with previous observations. 16 It is the relative contribution of these fluorophores in the different histologic grades that allow normal tissue to be distinguished from dysplastic and malignant tissue. The normal mucosa samples were flat, and the samples of adenocarcinoma had grossly irregular surfaces, while the polyps were not flat, but almost hemispherical. Nevertheless, no substantial differences in the patterns of emission spectra were found between the three. We thus conclude that although no specific fluorophore appears to be associated with any particular histologic grade of colonic tissue (in the range of wavelengths studied), the shape and overall architecture or geometry of colonic tissue does not appear to affect the position or line shape of the emitted fluorescence spectra. Thus, the observed differences in fluorescence intensity are due to the relative concentrations of fluorophores in the different grades of neoplasia. The more densely packed cells with a higher metabolic rate, greater nuclear material, and lower density of connective tissue (eg, THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
collagen) may explain some of the observed changes rather than tissue geometry. Noting these limitations and the attempts madE to minimize the changes in the measured spectra the colonic spectral data can be compared with tha1 of known fluorophores to indicate the possible causeE oftissue fluorescence. It can be said that tryptophar: (Figure 3) is a most likely cause of emission maxima A which occurs at about 331 nm. The FWHM of max· ima A (54 nm) is comparable to that of tryptophan (45 nm). The slight differences may be due to thE factors previously discussed. Tryptophan in solution was studied because undissolved tryptophan has a different emission spectrum. Tryptophan, as a free molecule or as a part of larger molecules such as cellular proteins, may be responsible for this emission peak. It produces strong fluorescence at about 330 nm and is one of the wavelengths found by Richards-Kortum et aP6 to be helpful in distinguishing normal from adenomatous tissues of the colon. There are a number of possibilities for peak B (maxima at 365.1 nm, FWHM 9.5 nm): elastin, collagen, and tyrosine. The 365 nm peak seen with elastin was observed with excitation at 290 nm, which did not reveal any maxima at 365 nm in colonic tissue. Tyrosine produced a band at 364 nm, but with a FWHM that is twice that in peak B. The most likely candidate for peak B is collagen type IV, which produced an emission peak at 365 nm (FWHM of 28 nm) when excited by wavelengths between 310 and 350 nm (Figure 4). This is also found in colonic tissue, which has a comparable FWHM (the difference in FWHM may be due to absorption by other molecules, such as oxyhemoglobin). Collagen is not a single protein but a family of closely related molecules with differences in amino acid composition and sequence. 18 Twelve types of collagen have been identified and are found in organs as diverse as skin, bone, tendon, cartilage, uterus, dentin, vitreous body, and blood vessel walls. 18 The type of collagen discussed by other investigators was not specified. 12 .14,16,17 Collagen type IV or basement-membrane-type collagen is specialized and restricted to the basal lamina of epithelia 18; with laminin and heparan sulphate proteoglycan, it helps to form a meshwork of filaments that supports the epithelia and provides a barrier for the selective filtration of macromolecules. 19 Type IV collagen was thus studied and appeared to be responsible for the emission seen at 365 nm. For maxima C (385.7 nm, with FWHM of 40.4), elastin (peak at 390 nm with FWHM 103 nm), phenylalanine (peak 375 nm , FWHM 87 nm), and collagen type IV (peak at 392 nm with FWHM 28 nm) are possible fluorophores. Of these, collagen (Figure 4) is the more likely candidate, particularly because the tissue emission peak is seen best with excitation between 330 nm and 350 nm. The FWHM of the emission peak from type IV collagen (28 nm) is, how-
225
Colonic Emission Spectra
ever, smaller than that observed for peak C in the colon (40 nm). This may have been due to the effect of other fluorophores or of in vitro measurements. Indeed, this emission peak may be due to an undetermined fluorophore. The main possibility for emission D (453.1. nm, FWHM 100.5 nm) appeared to be NADH (Figure 5). This has been suggested by other workers. 5,12-14 However, the mean position of this band seen in colonic tissues (453.1 nm) is considerably different from that of NADH (460 nm). The wavelengths at which this peak occurred ranged from 440 nm to 463 nm (Table 1). The mean FWHM of emission peak D was 100.5 nm, which was considerably greater than that measured in NADH (89 nm). Allowing for minor changes in fluorescence due to this in vitro study, our data does not fully support the view that emission peak D is solely from NADH. Other fluorophores may be contributing to this band, including phenylalanine, pyridoxal 5' phosphate and collagen type IV (see Table 2). This particular band has been found to fluoresce less intensely in dysplastic than in normal mucosa. 12,13 Attributing all fluorescence at about 460 nm to NADH must be treated with caution. Indirect characterizations can only suggest but not prove the identity of a particular emission peak, particularly in complex materials such as biological tissue. Our work did not indicate any specific maxima due to elastin, FAD, phenylalanine, pyridoxal 5' phosphate, or tyrosine. Some of these may contribute to the emission spectra studied but were not solely responsible for any of the isolated major emission peaks. It must be emphasized that the characterizations described here are, at best, approximations; they are not direct measurements. Many factors are at play, including tissue physiology and the interactions of multiple fluorophores with similar spectra. Further work must be done to elucidate the agents responsible for tissue fluorescence. References 1. Cello JP, Schneiderman DJ. Ulcerative colitis. In: Sleisenger MH, Fordtran JS, eds. Gastrointestinal Disease. Philadelphia: WB Saunders; 1991:1435-77. 2. Collins RH Jr, Feldman M, Fordtran JS. Colon cancer, dysplasia and surveillance in patients with ulcerative colitis: a critical review. N Engl J Med. 1987;316:1654-8. 3. Jensen P, Krogsgaard MR, Christiansen J, Braendstrup 0, Johansen A, Olsen J. Observer variability in the assessment of type and dysplasia of colorecta1 adenomas, analyzed using kappa statistics. Dis Colon Rectum. 1995;38:195-8. 4. Rosenstock E, Farmer RG, Petras R, Sivak MY Jr, Rankin GB, Sullivan BB. Surveillance for colonic carcinoma in ulcerative colitis. Gastroenterology. 1985;89:1342-6.
226
5. Kapadia CR, Cutruzzola WF, O'Brien KM, Stetz ML Enriquez R, Deckelbaum. LI. Laser-induced fluorescence spectroscopy of human colonic mucosa: detection of adenoma tous transformation. Gastroenterology. 1990;90:150-7. 6. Mourant JR, Boyer JD, Johnson TM, Lacey JA, Bigic IJ. Detection of gastrointestinal cancer by elastic scatterinl and absorption spectroscopies with the Los Alamos Optica Biopsy System. SPIE Proc. 1995;2387:210-7. 7. Feld MS, Manoharan R, Salenius J. Detection and charac terization of human tissue lesions with near infrared Ramal spectroscopy. SPIE Proc. 1995;2388:99-104. 8. Bjorkman DJ, Samowitz WS, Brigham EJ. Fluorescenc, localization of early colonic cancer in the rat by hematoporpb yrin derivative. Lasers Surg Med. 1991;11;263-70. 9. Dougherty TJ, Marcus SL. Photodynamic therapy. Eur . Cancer. 1992; 28A:1734-42. 10. Alfano RR, Tang GC, Pradhan A, Lam W, Choy D, Ophe E. Fluorescence spectra from cancerous human breast an lung tissue. IEEE J Quantum Electron. 1987;QE;23:1806
11. 11. Mahadevan A, Mitchell MF, Silva E, et aI. Study of th fluorescence properties of normal and neoplastic human cel vical tissue. Lasers Surg Med. 1993;13:647-55. 12. Cothren RM, Richards-Kortum. R, Sivak MY, et al. GaE trointestinal tissue diagnosis by laser induced tissue diagn( sis at endoscopy. Gastrointest Endosc. 1990; 36: 105-11. 13. Schomacker KT, Frisoli JK, Compton CC, Flotte T~ Richter JM, Nishioka NS, et aI. Ultraviolet laser-induce fluorescence of colonic tissue: basic biology and diagnosti potential. Lasers Surg Med. 1992; 12:63-8. 14. Schomacker KT, Frisoli JK, Compton CC, Flotte T. Richter JM, Deutsh TF, et aI. Ultraviolet laser induce fluorescence of colon polyps. Gastroenterology. 1992; 10: 1155-60. 15. Wang TD, Van Dam J, Crawford JM, Preisinger E..! Wang Y, Feld MS. Fluorescence endoscopic imaging of hl man colonic adenomas. Gastroenterology. 1996; 111:1182 91. 16. Richards-Kortum R, Rava RP, Petras RP, Fitzmauric M, Sivak M, Feld MS. Spectroscopic diagnosis of colon dysplasia. Photochem Photobiol. 1991; 53:777 -86. 17. Romer TJ, Fitzmaurice M, Cothren RM, Richards-Ko; tum. R, Petras R, Sivak MY, Kramer JR. Laser-inducE fluorescence microscopy of normal colon and dysplasia in c· Ionic adenomas: implications for spectroscopic diagnosis. A J Gastroenterol. 1995; 90:81-7. 18. Fawcett DW. A Textbook of Histology. New York: Cha; man & Hall; 1994:133-9. 19. Bailey AJ, Sims TJ, Duance VC, Light ND. Partial cha acterization of a second basement membrane collagen in hman placenta. FEBS. 1979;99:361-6. 20. Banerjee B, Gadde P, Tahiri A, Mazumdar MK. Arg( laser induced fluorescence spectroscopy in gastrointestin malignancy and inflammation. Am J Gastroenterol. 199 87:1335. 21. Keijzer M, Richards-Kortum. R, Jacques SL, Feld M Fluorescence spectroscopy of turbid media: autofluorescen of human aorta. Appl Optics. 1989;28:4286-92. 22. Pesce AJ, Rosen CG, Pasby TL. Fluorescence Spectroscop An Introduction for Biology and Medicine. New York: Marc Dekker; 1971:65-86. 23. Svanberg K, Clemente LP, Clemente MP, Wang Warloe T, Anderson-Engels S. Pharmacokinetic studies 8-aminolevulinic acid-induced protoporphyrin IX build-up some malignant tumors. SPIE Proc. 1995;2387:30-42.
September 1998 Volume 316 Numbe