MALIGNANT TUMORS FROM BENIGN. TISSUES AND TUMORS. ROBERT R. ALFANO, PH.D., BIDYUT B. DAS, M.S.. Institute for Ultrafast Spectroscopy and ...
143
LIGHT SHEDS LIGHT ON CANCER - DISTINGUISHING MALIGNANT TUMORS FROM BENIGN TISSUES AND TUMORS ROBERT R. ALFANO, PH.D., BIDYUT B. DAS, M.S. Institute for Ultrafast Spectroscopy and Lasers Department of Physics The City College of New York
JOSEPH CLEARY, M.D., ROMULO PRUDENTE, M.D., EDWARD J. CELMER, M.D. Doctors Hospital Division, Beth Israel Medical Center New York, New York
FLUORESCENCE SPECTROSCOPY has been extensively used for a long time by the medical and biological community using fluorescent probes. Researchers obtained fundamental information about conformal changes in muscle and nerves, polarity of the surrounding environment, dynamic conformations of molecules in membranes and secondary structure of DNA, RNA, and other systems.' In the past there has been a search for novel diagnostic approaches to differentiate diseased and normal tissues using optical spectroscopy.2-5 Because extrinsic fluorescent markers interact with native cellular environments, a technique has recently been developed to use the intrinsic fluorescing chromophores present in such biological systems as proteins, nucleic acids, and lipid molecules of the tissues to characterize the physiological states of normal and abnormal systems. Such intrinsic chromophores as flavins and riboflavins are known to fluoresce in visible spectral region6 while protein molecules such as collagen and elastin are found to fluoresce in ultraviolet and visible spectral region.4 In the early 1980s Alfano and co-workers7,8 established that differences exist in fluorescence spectra from normal and tumor rat tissues. They later extended the work to normal and malignant tissues of human breast and lungs and found differences in the spectral profiles.9, 0 In developing a diagnostic method for cancer detection it is essential to separate malignant tumors not only from normal tissues but from benign tissues and tumors. In this paper we detail our work on normal, benign tissues and tumors and malignant tumors from human breast and lungs This work was supported by grants from the SDIO/ONR and Mediscience Technology Company.
Vol. 67, No. 2, March-April 1991
144
144R.R. R.R.
ALFANO AND OTHERS
using ultraviolet fluoresence spectroscopy. We demonstrate that it is possible to separate malignant tumors from benign and normal samples using optical fluorescence spectroscopy-a major breakthrough. This work points to an optical biopsy to diagnose cancer. METHODS AND MATERIALS A lamp based fluorescence spectrometer was used to measure fluorescence from the tissues. Frontal excitation was used to pump tissue samples placed in quartz cells. The samples were excited at 300 nm wavelength, and the resulting emission spectra were recorded from 320 to 580 nm. Excitation spectra at 340 nm and 440 nm emission were recorded for exciting wavelengths from 250 to 325 nm. The full width at half maximum (FWHM) of the exciting beam was about 5 nm. Some 19 malignant (ductal carcinoma), 13 benign tissues, 7 benign tumors (fibroadenoma), and one normal sample from human breast were tested, as were one lung cancer and one normal lung sample. Except for some samples that were later used in a blind study, histopathologic examination was done on all others and the diagnosis was known before the fluorescence experiment was performed. The ratio of intensities at 340 nm to 440 nm were calculated for each sample and compared for differences. This pair of wavelengths (340 and 440 nm) was selected to eliminate the effect of absorption on the fluorescence by blood present in the tissues, important in developing a diagnostic approach for in vivo studies. To have a better insight into what could cause the fluorescence spectra profiles from the tissues, a number of potential fluorophores such as proteins (collagen, elastin, and trypsin), amino acids (tryptophan, tyrosine), and NADPH (a nucleotide) were tested for fluorescence. Except for trypsin, which was tested as a 25% solution with EDTA, all other compounds were tested after solution in water. RESULTS
The fluorescence spectra from 320 nm to 580 nm from benign and cancerous breast tumors under 300 nm excitation are displayed in Figure 1. The difference in spectra shown in the figure permits separation of cancerous from other tissue types. The ratio of intensities at 340 nm to 440 nm were calculated for each sample and compared for differences. This pair of wavelengths (340 and 440 nm) was selected to eliminate the effect of absorption on the fluorescence by the blood present in the tissues. This is possible because the blood absorbances at these two wavelengths are the same.11 The relative difference between the intensities at 340 and 440 nm is prominent in maligBull. N.Y. Acad. Med.
145 145
LIGHT SHEDS LIGHT ON CANCER LIGHT ON CANCER
LIGHT SHEDS
EMISSION SPECTRA 700
Xex
300nm Cancer a\
/
600-
Benign 400
j
a300200100
/
0
320
380
440
500 Emission Wavelength (nm )
560
Fig. 1. Fluorescence spectra from cancer and benign tumor tissues excited at 300 nm wavelength.
nant tumors compared to benign and normal samples. The ratios of intensities at 340 and 440 nm wavelengths were calculated and we found that all cancer ratios were above 10 while benign and normal ratios lie below nine. A
histogram of ratios for malignant, benign, and normal tissues are displayed in Figure 2. The average number for malignant tissue is 15.7 and for benign tissues and tumors is 4.8 while the single normal sample gave a ratio of 4.4. The ratio range for malignant tissue extends from 10 to 20, while that for benign samples is from two to nine. Some of the malignant and benign samples were run for excitation spectra from 250 to 325 nm at emission wavelengths 340 nm and 440 nm. The ratio of these two excitation spectra was calculated for each sample and compared to determine what excitation wavelength yields maximum possible ratio difference between malignant and benign samples. The ratio of intensities at 340 to 440 nm for different excitation wavelengths is shown in Figure 3. The wavelength of 300 nm appears to be the most appropriate excitation wavelength for breast, but wavelengths from 250 to 315 nm can also be used to detect differences. The histogram shown in Figure 2 clearly demonstrates that malignant samples can be separated from normal and benign breast samples using the Vol. 67, No. 2, March-April 1991
146
R.R. ALFANO AND OTHAPR OTHERS
R.R.~~~~~~~~-ALAN ----
146
AN
CANCER Ratio = 15.7 ± 2.75 CEO
Poa
I
I
2
I
I
4
]I
6
I
I1
I
8
10
14
12
16 16
I
I
K
I
18
I
20
-F
BENIGN TUMORS AND TISSUES Ratio = 4.76 ± 1.88
turn.
-F.
£0I I
4
6
8
I
10
I1
I
I
I
14
12
I
I
16
18
I
20
NORMAL Ratio = 4.4 I
0
I
a
I
2
I
I
4
I
6
8
I
I
10
I
9
I
12
I
0
I
14
m
I
a
I
16
I
I
I
18
II
Ia
20
I9
I
22
I(340nm) /I(44Onm) CANCER * BENIGN TISSUES AND TUMORS
0
NORMAL
Fig. 2. A histogram of fluorescence intensity ratio at 340 to 440 nm, excited at 300nm for malignant tumor, benign tissue and tumor, and normal samples. Each box represents one sample. Center of each box matches with the ratio 1(340) / 1(440) of the corresponding sample.
Bull. N.Y. Acad. Med.
147
LIGHT SHEDS LIGHT ON CANCER
LIH
SHD
LIH
NCNE
EXCITATION SPECTRA RATIO
30
r, Its'
I204
-- -- Cancer _
Benign
310
320
'I
340 440
250
260
270
290 300 280 Excitation Wavelength (nm)
Fig. 3. Ratio of fluorescence intensity at 340 nm to that at 440 nm plotted against different excitation wavelengths for malignant and benign tissue samples.
300 nm excitation. To check the accuracy of this diagnostic approach, a blind study was done on several samples without prior knowledge of their morphology. The result was later found to be consistent with the pathological study of those samples. More studies are under way. Although our experiment was done mainly on human breast samples, one sample each of lung cancer and lung normal were also tested. The ratio of intensities in these two cases agrees well with the respective histogram for breast samples displayed in Figure 2. To obtain an idea as to what causes the spectral difference, fluorescence from proteins, amino acids, and nucleotides was recorded. The fluorescence spectra from collagen, elastin, and trypsin are shown in Figure 4a, while that of tryptophan, tyrosine and NADPH are displayed in Figure 4b. As can be seen from these figures, the fluorescence spectrum from tyrosine gives a profile with a maxima that resembles the peak at 340 nm from the tissues. The fluorescence maxima of tryptophan and NADPH are at 360 nm and 460 nm respectively. Both collagen and elastin give broadband fluorescence, while the spectrum given by trypsin could very well be from its tryptophan component.
Vol. 67, No. 2, March-April 1991
*~ ~ .
148
R.R. ALFANO AND OTHERS
EM ISS ION SPECTRA
Xex = 300 nm
exe :-.
.
..-
.
:*.:
:
I. _1
.*~~~~~~~~~~~~~~~~~~~~~~~~~~.
c
U) a
Emission Wavelength (nm) Fig. 4a. Emission wavelength.
spectra of tyrosine, tryptophan, and NADPH excited
at
300
nm
EMISSION SPECTRA
a)
._
-
Emission Wavelength (nm) Fig. 4b. Emission spectra of collagen, elastin, and trypsine excited at 300 nm. Bull. N.Y. Acad. Med.
LIGHT SHEDS LIGHT ON CANCER CANCER
149 149~~~~~~~~~~~~~~~~~~~~
DISCUSSION As shown by data displayed in Figure 2, comparison of two fluorescence intensity peaks at around 340 nm and 480 nm can separate malignant from normal and benign tissues and benign tumors. Since the ratio of the two intensities at 340 nm and 440 nm varies greatly by changing the excitation wavelength around 300 nm (shown in Figure 3), it suggests that the two peaks could not be due to only one fluorophor. 12 Although the exact fluorophores giving rise to these fluorescence peaks are not yet known, our study suggests that tyrosine, tryptophan, or some proteins containing them as components could cause a part of the signature in the spectral region of 340 nm peak. Though the fluorescence peak from the tryptophan solution in our experiment is found at 360 nm, it still cannot be excluded as causing the 340 nm fluorescence peak in the tissues, because the shift of fluorescence maxima is possible in the different tissue pH environments. Similarly, the second fluorescence maximum from the tissues could be due to the nucleotide NADPH or some of its derivatives or due to a combination of NADPH, collagen, and elastin. More studies will have to be done with these components to find any definitive answer. This method of detection not only separates malignant from normal and benign samples but provides a simple technique to be used. Diagnostic systems using the ratio of fluorescence at two wavelengths can be used to measure the state of the tissue in vivo without being affected by body motion. An optical or endonoptic biopsy can measure the characteristics of the tissue or tumor outside or inside the body. SUMMARY
Difference in fluorescence spectra from human malignant and benign tumors, benign and normal breast tissues were measured. Spectral histograms from 40 samples show the diagnostic possibilities of this optical technology. Fluorescence from model fluorophores (nucleotides, amino acids, and proteins) were used to speculate on the sources of marked features of the tissue fluorescence. ACKNOWLEDGMENTS
We thank Dr. Richard Rava for useful discussions and Dr. Arthur Caron for providing us with some tissue samples in the initial studies. We also thank Vol. 67, No. 2, March-April 1991
150
R.R. ALFANO AND OTHERS
Guichen Tang and Asima Pradhan for their useful suggestions during the course of this experiment. REFERENCES 1. Yguerabide, J.: Nanosecond Fluorescence Spectroscopy of Biological Macromolecules and Membranes. In: Fluorescence Techniques in Cell Biology, Thaer, A. A. and Sernetz, M., editors. Berlin, Springer-Verlag, 1973, pp. 311-31. 2. Pringsheim, P.: Fluorescence and Phosphorescence. New York, Interscience, 1949. 3. Birks, J.B.: Photophysics of Aromatic Molecules. New York, Wiley. 1970. 4. Udenfriend, S.: Fluorescence Assay in Biology and Medicine, vol. 1. New York, Academic, 1962; vol. 2, 1969. 5. Long, D.A.: Raman Spectroscopy. New York, McGraw-Hill, 1977. 6. Fasman, G.D., editor: Handbook of Biochemistry and Molecular Biology, 3rd ed. Cleveland, CRC Press, 1975, pp. 205-10. 7. Alfano, R.R., Tata, D., Cordero, J., et
8. 9.
10.
11. 12.
al.: Laser induced fluorescence spectroscopy from native cancerous and normal tissues. IEEE J. Quant. Electr. QE-20: 1507-11, 1984. Alfano, R.R. and Alfano, M.A., Medical diagnostics: a new optical frontier. Photon. Spectra 19:55-60, 1985. Alfano, R.R., Tang, G.C., Pradhan, A., et al.: Fluorescence spectra from cancerous and normal human breast and lung tissues. IEEE J. Quant. Electr. QE-23; 1806-11, 1987. Alfano, R.R., Pradhan, A., Tang, G.C., and Wahl, S.J.: Optical spectroscopic diagnosis of cancer and normal breast tissues. J. Opt. Soc. Am. B, 1015, 1989. Richards-Kortum, R.: Thesis, M.I.T., 1987. Weber, G.: Enunciation of components in complex systems by fluorescence spectro-photometry. Nature 190: 27, 1961.
Bull. N.Y. Acad. Med.
