TIME-RESOLVED LASER-INDUCED FLUORESCENCE SPECTROSCOPY FOR ENHANCED DEMARCATION OF HUMAN ATHEROSCLEROTIC PLAQUES S. ANDERSSON-ENGELSa, and S. SVANBERGaq + aDepartment (Sweden)
J. JOHANSSONa, U. STENRAMb, K. SVANBERGC
(Received February 3,1989;accepted
85 Lund S-221
(Sweden) 85 Lund
Atherosclerosis, fluorescence, spectroscopy, tissue diagnosis.
P. 0. Box
Summary We report on the enhanced demarcation between human atherosclerotic plaques and normal vessel wall obtained using time-resolved detection of laser-induced fluorescence rather than the customary time-integrated monitoring technique. A frequency-doubled mode-locked and cavitydumped continuous wave dye laser was used for picosecond pulse generation at 320 nm, and photon-counting techniques were employed for the timeresolved signal monitoring from human aorta samples in vitro. Implications for imaging fluorescence angioscopy and spectroscopic guidance in laser ablation of plaque are indicated.
1. Introduction During the last few years a large number of reports on the use of laserinduced fluorescence for spectroscopic demarcation of atherosclerotic plaques in human vessels have appeared [ 11. Such studies are interesting because of the potential for spectroscopic guidance in fibre-optic laser ablation of plaques using, for example, excimer lasers. In the original studies [2, 33 green excitation light was used. In our first experiments  we found that the use of shorter wavelengths, such as 337 nm from a nitrogen laser, offered certain advantages over green light. A fluorescence band peaking at
*To whom correspondence should be addressed. loll-1344/90/$3.50
@ Elsevier Sequoia/Printed in The Netherlands
about 390 nm associated with plaque was found. We have recently reported on more detailed studies in which different spectroscopic discrimination functions were tested [ 51. Vessel fluorescence monitoring is complicated by the interplay between fluorophore emission and blood absorption. The sensitivity of the spectral shape to the light collection geometry is reduced by employing a short excitation wavelength for which the tissue penetration depth is minimal. In contrast with previous studies of plaque demarcation, in which the integral fluorescence emission has been detected using, for example, optical multichannel detection techniques, in this paper we report on preliminary studies of vessel fluorescence in the time domain at selected wavelengths using picosecond laser spectroscopic techniques. It is demonstrated that a considerable improvement in plaque demarcation can be obtained using the temporal behaviour of the tissue fluorescence. It is also demonstrated that a far-reaching immunity to blood interference can be obtained using such techniques. Possible implications for imaging fluorescence angioscopy are discussed. Time-resolved fluorescence techniques have been a valuable tool in biophysical studies for a long time . A much improved understanding of the dynamics and function of macromolecules has been obtained. The tumour-localizing agent haematoporphyrin derivative (HPD) in solution has been extensively studied with regard to its temporal behaviour . Tata et al.  used picosecond techniques in studies of malignant tumours, and Deutsch and coworkers  observed improved tumour demarcation in animals injected with haematoporphyrin derivative by employing timeresolved techniques. Such improved tumour demarcation has been verified by us in experiments on rats [lo]. Baraga et al. [ll] studied time-resolved fluorescence in the UV region from human aorta, which they related to tryptophan, collagen and elastin. Since fluorescence spectra from tissue exhibit little structure, it is evident that the additional information obtained from the time domain is of great value. 2. Materials and methods Aortic samples from newly deceased patients exhibiting various stages of atherosclerotic lesions were investigated within 24 h post mortem. Sketches of the samples were drawn before the investigation to be used for the histopathological examination. The investigation of the individual fresh samples lasted from 30 min to 1 h. The samples were investigated in air with the endothelium of the vessel inner wall exposed to the laser beam, which was focused to a spot with a diameter of a fraction of a millimetre. Data were recorded in scans starting in a region of normal vessel wall and passing over a plaque region. The sample was moved on a micrometre-controlled sledge to allow reproducible positioning of the sample in time-resolved recording scans at different fluorescence wavelengths. Decay curves were recorded during 2 min at a count rate of about 1000 Hz.
A mode-locked argon-ion laser (Coherent Radiation CR-12) was used to synchronously pump a coherent radiation dye laser equipped with a cavity dumper. The dye laser provided pulses of 6 ps at 640 nm at a repetition rate of about 3 MHz. The average power of the red light was about 10 mW. The red pulses were frequency doubled to 320 nm in a potassium dideuterium phosphate (KD* P) crystal with an efficiency of frequency doubling of approximately 0.5%. A 180” arrangement with a quartz lens (d = 10 cm, f = 15 cm) was used to collect the backward fluorescence light. The fluorescence light was wavelength selected in a 0.5-m spectrometer together with interference filters, and was detected in a microchannel plate photomultiplier tube (Hamamatsu R 1564 U). The electronics included a starting pulse channel. Suitable signal amplifiers, constant fraction discriminators and a time-to-amplitude converter were employed. Time histograms were built up in a multichannel analyser and data analysis was performed with a program package on an IBM-compatible personal computer. The time response function of the apparatus was measured with scattered light and was found to have a full width at half-maximum (FWHM) of 250 ps. This value was used in the computer deconvolution procedure of the fluorescence signal. A more detailed description of the measurement system and the data analysis routines is given in ref. 10.
3. Results The typical steady state fluorescence structures of normal vessel wall and plaque are shown in Fig. 1 for 320-nm excitation. The dip in the spectra at 420 nm is due to reabsorption of fluorescence light in haemoglobin. Thus the two peaks should not be considered as two independent fluorescence peaks. It is clear from Fig. 1 that the major difference in fluorescence between plaque and normal vessel wall is found around 400 nm. At wave-
Fig. 1. Fluorescence spectra for a plaque region (upper curve) and normal vessel wall. The excitation wavelength was 320 nm. Both curves are spectrally corrected.
lengths above 480 nm, hardly any difference can be seen between the two spectra. The monochromator was set to 400 nm or 480 nm, corresponding to the two wavelengths used in our previous steady state fluorescence studies. The fluorescence intensities at 400 nm and 480 nm are denoted a and c respectively. Time-resolved recordings of sample fluorescence are shown for plaque and normal tissue wall in Fig. 2 using 320~nm excitation. Clear differences in the temporal behaviour can be observed at 400~nm emission. Three different lifetimes of approximately 7 ns, 2 ns and about 300 ps are observed for both plaque and normal vessel. All measured samples contained several areas of diseased as well as healthy tissue. Data from fibrotic plaque, calcified plaque and normal vessel wall based on 42 recordings from three samples are shown in Fig. 3. Here the signal integrated from 5 to 15 ns is divided by the signal obtained from the first 5 ns of the decay. For a fast decay, this ratio obviously has a low value, whereas higher ratios indicate a slower decay. If we first consider measurements of the a signal (400 nm), the time-dependent ratio is a factor of 1.6 higher for the plaque region than for the normal vessel wall. In the case of the c signal (480 nm), however, these ratios are almost equal for plaque and normal vessel wall. Figure 4 presents results for another function, the ratio of the fluorescence at 400 nm (a) integrated over 5 to 15 ns to the fluorescence at 480 nm (c) integrated over 0 to 5 ns. This function was chosen to include both the time-dependent and spectrum-dependent differences between normal vessel wall and atherosclerotic plaque. Results for a scan through a plaque region are shown and a.demarcation improvement from 2.8 to 4.5 is obtained using the temporally-resolved data. Time-integrated data for the a:c ratio are included for comparison.
Time (nsl Fig. 2. Fluorescence plaque. Both curves
decay curves for normal vessel wall and a region with atherosclerotic were recorded at 400 nm. Excitation wavelength was 320 nm.
Fig. 3. Decay ratio presented as the fluorescence intensity integrated from 5 to 15 ns divided by that for the first 5 ns of the decay. Data are given plus or minus one standard deviation for three types of tissue at 400 nm (a) and at 430 nm (c). Fig. 4. A 30-mm scan over a plaque region for two demarcation criteria. The scan starts in calcified plaque, continues into thin plaque and ends in a normal vessel region. Crosses indicate the ratio between steady state fluorescence at 400 nm and that at 480 nm. Circles show slow fluorescence at 400 nm divided by fast fluorescence at 480 nm. Both curves are normalized to unity for normal vessel.
4. Discussion Since the different chromophores in the tissue have different fluorescence lifetimes, they give varying contributions to the fluorescence intensity at various wavelengths; thus characteristic changes in decay times are observed for diseased and non-diseased tissue. There are great advantages in expressing the results of fluorescence measurements as simple ratios of fluorescence intensities, for example, at two different wavelengths. Such ratios almost exclude effects of varying excitation and detection efficiency and sample topology, which have to be carefully taken into account in the case of criteria based on a single intensity. As illustrated above, the plaque demarcation from normal tissue is demonstrated to improve when the temporal information is included. This should have important implications for in uiuo point fibre probe measurements using a picosecond laser or, more realistically, a short-pulse nitrogen laser (approximately 1 ns). The same improvement should be attainable in a fluorescence multicolour imaging system [ 121, which is gateable down to 5 ns. As illustrated in Fig. 3, vessel fluorescence is sufficiently long lived to allow the gated image intensifier to discriminate between fast and slow fluorescence.
It is important to point out that the set of three lifetime components given here could very well be much more complex. Hence the lifetime components should be considered as one possible solution to the evaluation procedure. Barga et al. [ 111 found two long lifetime components for healthy aortic tissue that are in good agreement with our studies, but also two shortlived components that probably correspond to our 300-ps component. However, no extensive conclusions can be drawn from the comparison, since Baraga et al. [ll] used other excitation as well as emission wavelengths and, in addition, snap-frozen samples. In clinical plaque demarcation, the influence of blood reabsorption on the spectroscopic signal is of concern [l, 131. It is interesting to note that plaque can be demarcated from normal vessel wall using time-resolved spectroscopy in a region of strong haemoglobin absorption around 400 nm. Since the blood, by reabsorbing some of the tissue fluorescence at short wavelengths, only affects the intensity of the light and not its decay time, single-wavelength time-resolved data can be considered to be more or less immune to blood interference, an observation of crucial importance for spectroscopic guidance of laser ablation of plaque, The penetration depth in tissue as well as blood is very small for UV light. Both the excitation (320 nm) and the emission (400 nm) light have penetration depths of a few tens of micrometres. Scattering of light in a shallow layer of blood would therefore alter only the early picosecond range of the decay curve, leaving almost the whole curve unaltered. Enhanced demarcation and negligible blood absorption can be obtained by combining time-resolved data from two wavelengths with zero differential blood absorption . This approach could show a small dependence on blood content of the samples, because of the slightly varying reabsorption and scattering cross-sections within even narrow wavelength regions. An empirical determination of suitable filter wavelengths might be a practical approach. Possible residual problems should be compared with the enhancement of demarcation between plaque and normal vessel obtained by combining the wavelength and time-resolved features. Generally speaking, tissue fluorescence diagnostics should benefit greatly from time domain information. Steady state fluorescence at certain wavelength combinations has shown a potential to discriminate between atherosclerotically diseased and normal vessel wall. Information obtained using time-resolved spectroscopy gives additional aspects to laser-induced fluorescence used in the characterization of diseased vessel walls. By combining the two modes of discrimination, a clinically valuable tool in the field of laser angioplasty should be attainable.
Acknowledgments The development of computer software for lifetime evaluation Anders Persson is gratefully acknowledged, as is the general support
Prof. Anders Gustafson. This work was supported Medical Research Council (MFR).
in part by the Swedish
References 1 R. R. Richards-Kortum, A. Metha, T. Kolubayev, C. Cothren, B. Sacks, C. Kittrell, M. S. Feld, N. B. Ratliff, T. Kjellstrom, G. Bordagaray, M. Fitzmaurice and J. Kramer, Spectroscopic diagnosis for control of laser treatment of atherosclerosis, in W. Persson and S. Svanberg (eds.), Laser Spectroscopy VIII, Springer Series in Optical Sciences, Springer, Berlin and Heidelberg, 1987, pp. 366 - 369. 2 C. Kittrell, R. L. Willet, C. de 10s Santos-Pacheo, N. B. Ratliff, J. R. Kramer, E. G. Malk and M. S. Feld, Diagnosis of fibrous arterial atherosclerosis using fluorescence, Appl. Opt., 24 (1985) 2280 - 2281. 3 R. M. Cothren, G. B. Hayes, J. R. Kramer, B. Sachs, C. Kittrell and M. S. Feld, A multifiber catheter with an optical shield for laser angioplasty, Lasers Life Sci., I (1986) 1 - 12. 4 P. S. Andersson-Engels, A. Gustafson, U. Stenram, K. Svanberg and S. Svanberg, Diagnosis of arterial atherosclerosis using laser-induced fluorescence, Lasers Med. Sci., 2 (1987) 261 - 266. 5 S. Andersson, A. Gustafsson, J. Johansson, U. Stenram, K. Svanberg and S. Svanberg, Laser-induced fluorescence used in localizing atherosclerotic lesions, Laser Med. Sci., 4 (1989) 171 - 181. 6 V. S. Letokhov, Ultrashort pulses in studies of biomolecules, in V. S. Letokhov (ed.), Laser Picosecond Spectroscopy and Photochemistry of Biomolecules, Adam Hilger, Bristol and Philadelphia, 1987, pp. 1 - 55. 7 R. Cubeddu, W. F. Keir, R. Ramponi and T. G. Truscott, Photophysical properties of porphyrin-chlorin systems in the presence of surfactants, Photochem. Photobiol., 46 (1987) 633 - 638. 8 D. B. Tata, M. Foresti, J. Cordero, P. Tomashefsky, M. A. Alfano and R. R. Alfano, Fluorescence polarisation spectroscopy and time-resolved fluorescence kinetics of native cancerous and normal rat kidney tissue, Biophys. J., 50 (1986) 463 - 469. 9 T. F. Deutsch et al., personal communication. 10 S. Andersson-Engels, J. Johansson and S. Svanberg, The use of time-resolved fluorescence for diagnosis of atherosclerotic plaque and malignant tumours, Spectrochim. Acta, submitted for publication. 11 J. J. Baraga, P. Taroni, Y. D. Park, K. An, A. Maestri, L. L. Tong, R. P. Rava, C. Kittrell, R. R. Dasari and M. S. Feld, Ultraviolet laser induced fluorescence of human aorta, Spectrochim. Acta, Part A, 45 (1989) 95 - 99. 12 P. S. Andersson, S. Mont&n and S. Svanberg, Multispectral system for medical fluorescence imaging, IEEE J. Quantum Electron., 23 (1987) 1798 - 1805. 13 M. R. Prince, T. F. Deutach, M. M. Mathews-Roth, R. Margolis, J. A. Parrish and A. R. Osteroff, Preferential light absorption in atheromas in oitro - implications for laser angioplasty, J. Clin. Znuest., 78 (1986) 285 - 302.