Biochemical Imaging of Human Atherosclerotic Plaques with Fluorescence Lifetime Angioscopy

Transcription

1 Photochemistry and Photobiology, 2010, 86: Research Note Biochemical Imaging of Human Atherosclerotic Plaques with Fluorescence Lifetime Angioscopy Patrick Thomas 1, Paritosh Pande 1, Fred Clubb 2, Jessie Adame 3 and Javier A. Jo* 1 1 Department of Biomedical Engineering, Texas A&M University, College Station, TX 2 Department of Veterinary Pathobiology, Texas A&M University, College Station, TX 3 Autopsy and Pathology Services, P.A., Kingwood, TX Received 21 August 2009, accepted 16 December 2009, DOI: /j x ABSTRACT A prototype angioscopy system with fluorescence lifetime imaging microscopy (FLIM) capabilities was built and applied for biochemical imaging of human coronary atherosclerotic plaques. The FLIM angioscopy prototype consisted of a thin flexible angioscope suitable for UV-excited autofluorescence imaging, and a FLIM detection system based on a pulse sampling approach. The angioscope was composed of an imaging bundle attached to a gradient index objective lens and surrounded by a ring of illumination fibers (2 mm outer diameter, 50 lm spatial resolution). For FLIM detection based on the pulse sampling approach, a gated-intensified charge-couple device camera (200 ps temporal resolution) was used. Autofluorescence was excited with a pulsed UV laser (337 nm) and FLIM images were acquired at three emission bands ( nm, nm, nm). The system was characterized on standard fluorophores and then used to image postmortem human coronary arteries. The FLIM angioscope allowed us to distinguish elastin-dominant plaques (peak emission at 450 nm, 1.5 ns lifetimes) from collagen-dominant plaques (peak emission at 390 n, 2 3 ns lifetimes) based on their intrinsic fluorescence spectral and lifetime differences. This study demonstrates the potential of FLIM angioscopy for biochemical imaging of human coronary atherosclerotic plaques. INTRODUCTION *Corresponding author javierjo@tamu.edu (Javier A. Jo) Ó 2010TheAuthors. JournalCompilation. TheAmericanSocietyofPhotobiology /10 Sudden thrombosis caused by the rupture of a coronary atherosclerotic plaque is the major cause of fatal sudden cardiac events (1). The plaque biochemical composition is closely related to its vulnerability for rupture (2). Thus, significant effort has been dedicated to developing imaging technologies capable of quantifying plaque biochemical composition (3). One promising approach is the use of the plaque autofluorescence emission as an intrinsic biochemical signature (4,5). The three main sources of plaque autofluorescence (elastin from the media, collagen from fibrotic plaques and caps, and extra- and intracellular lipids) have specific fluorescence spectral and lifetime characteristics (5). Thus, a technology capable of quantifying the arterial wall autofluorescence will have the potential for clinical biochemical imaging of atherosclerosis. Fluorescence lifetime imaging microscopy (FLIM) maps the spatial distribution of fluorescence lifetime in the observed biological sample (6). As FLIM does not depend on absolute intensity images, it provides a more robust alternative than intensity-based fluorescence imaging. Thus, FLIM can provide useful information about biological tissue composition despite tissue heterogeneity and strong optical scattering and could represent a powerful functional imaging modality for clinical applications. However, the clinical potential of FLIM for tissue diagnosis has not been fully explored. In this study, we present the design and implementation of a prototype angioscopy system with FLIM capabilities. The system was first characterized by imaging standard fluorophores and then used to image postmortem human coronary arteries. To our knowledge this is the first attempt of FLIM angioscopy for biochemical imaging of human coronary atherosclerotic plaques. MATERIALS AND METHODS Instrumentation. The FLIM angioscopy prototype is an improvement of a system previously reported (7), and it consists of: (1) a thin flexible angioscope suitable for UV-excited autofluorescence imaging, and (2) a FLIM detection system based on a pulse sampling approach. The custom-made angioscope consisted of a bifurcated probe having one illumination and one imaging bundle. At the common distal leg of the probe, the imaging bundle ( element, lm inner outer diameter; Fujikura, Japan) was cemented to a gradient index objective lens (350 lm diameter, 0.5 NA, 4 mm working distance; GRINTECH GmbH, Germany) and surrounded by a ring of illumination fibers (200 elements, lm inner outer diameter, 0.22 NA, High OH silica silica for UV transmission). At the proximal ends, the imaging bundle was flat polished, while the illumination bundle was terminated on an SMA-905 connector. The angioscope was 2 m in total length and had an outer diameter of 2 mm at the distal common end. For FLIM detection based on the pulse recording approach, a gated-intensified charge-couple device (ICCD) camera (4Picos; Stanford Computer Optics, Berkeley, CA) was used. The ICCD allows gating times as short as 200 ps, delay times from 0 to 80 s with 10 ps resolution, and spectral sensitivity between 200 and 900 nm. The USB connectivity of the 4Picos permitted controlling the ICCD with a 727

2 728 Patrick Thomas et al. laptop computer, making the FLIM system fully portable. Artery autofluorescence was excited using a pulsed nitrogen laser (337 nm, 700 ps pulse width, MNL 205; LTB Lasertechnik, Germany). A digital delay generator was used to synchronize the laser and the ICCD triggering. The fluorescence image relayed at the proximal end of the imaging bundle was focused onto the ICCD through a 20 microscope objective and a 15 cm focal length doublet lens acting as a tube lens in infinity-corrected microscope configuration. A computer-controlled filter wheel inserted into the infinity space allowed acquiring FLIM images at three emission bands: Band 1 centered at 390 nm with a 40 nm bandwidth, Band 2 centered at 450 nm with a 40 nm bandwidth, and Band 3 centered at 550 nm with a 88 nm bandwidth. These bands were chosen based on previous studies showing that arterial fluorescence signal most relevant for plaque characterization is concentrated in these emission bands (8). The temporal resolution of the system was 0.5 ns based on the lifetime measurement of NADH. The spatial resolution of the system measured using an USAF test chart was 50 lm. The acquisition time was s per emission band, depending upon the integration time needed to achieve adequate signal levels. FLIM imaging. The FLIM angioscopy system was first validated using fluorophores with known emission spectra and lifetimes: POPOP [p-phenylenbis(5.-phenyl-2.-oxazol)] ( nm emission, ns lifetime), NADH ( nm emission, ns lifetime) and FAD ( nm emission, 2 3 ns lifetime). Solutions (at 1 mm concentration) of POPOP in methanol, and NADH and FAD in PBS were loaded on quartz capillaries (400 lm diameter) and imaged with the system. The FLIM angioscope was then used for biochemical imaging of postmortem human coronary arteries obtained from eight subjects. The study was carried out with the approval of the Texas A&M University Institutional Review Board. The time between death and FLIM imaging was kept within 48 h. The coronary vessels were opened longitudinally, and the tip of the angioscope was placed at 4 mm perpendicular to the lumen, resulting in a circular field of view of 2 mm in diameter. The imaged artery segments were sent for histopathology analysis. For all measurements, the laser power at the sample was adjusted to 0.2 mw. Data analysis. To estimate the fluorescence lifetime at each pixel of the FLIM images, deconvolution needs to be performed. A computational method for mathematical deconvolution of FLIM images based on the Laguerre expansion technique was recently developed, performing at least two orders of magnitude faster than other algorithms (9,10). In this study, a new online implementation of the Laguerre deconvolution method was used to analyze the FLIM angioscopy data (11). After FLIM data analysis (deconvolution), maps of normalized intensity and lifetime for each emission band were estimated. Maps of Laguerre expansion coefficients (LEC) computed with our deconvolution methods were also generated (10). To establish the correlation between the FLIM signal and the plaque types, the following statistical analysis was performed. Regions of interest (ROI) defined in the FLIM lumen images were correlated with the underlying artery histopathology and assigned to one of three plaque types: (1) Thin plaques with a thickness of <250 lm, Fibrotic plaques with a thickness >250 lm and Thick Cap fibroatheromas (FA) showing a cap thickness >100 lm. FLIM parameters (average values within the ROI of normalized intensity, lifetime and LEC values at the three emission bands) were assigned to the same plaque group. Differences in the FLIM parameters among the plaque types were assessed by one-way ANOVA test, and pairwise comparisons between multiple parameters were evaluated by Student s two-tailed t-test. RESULTS System validation Results from the FLIM system validation with standard fluorophores are shown in Fig. 1. The normalized intensity maps correctly reflected the emission spectrum of the three fluorophores (Fig. 1a): the POPOP capillary showed emission at both the 390 and 450 nm bands, the NADH capillary showed the strongest emission at the 450 nm band and some emission at the 550 nm band, and the FAD capillary showed emission only at the 550 nm band. Normalization of the intensity maps was performed by dividing pixel-by-pixel the intensity value for the given band by the sum of the intensity from the three bands. The lifetime maps show accurate estimation of each fluorophore lifetime characteristics (Fig. 1b): the POPOP capillary showed a homogenous lifetime map with a mean value of ± 0.13 ns, the NADH capillary showed a homogenous lifetime map with a mean value of 0.47 ± 0.03 ns and the FAD capillary showed a homogenous lifetime map with a mean value of 2.28 ± 0.21 ns. The lifetime histogram indicated that the three fluorophores can be identified directly from their estimated lifetime map (Fig. 1c). Figure 1. System validation on standard fluorophores: fluorescence lifetime imaging microscopy imaging of three capillaries filled with POPOP, NADH and FAD. (a) The normalized intensity maps reflect the fluorescence spectra of the three fluorophores. (b) The lifetime map clearly distinguishes each fluorophore in terms of their fluorescence lifetimes. (c) The lifetime histogram shows three distinct peaks corresponding to the intrinsic fluorophore lifetimes (POPOP: 1.41 ± 0.13 ns; NADH: 0.47 ± 0.03 ns; FAD: 2.28 ± 0.21 ns).

3 Photochemistry and Photobiology, 2010, FLIM imaging of atherosclerotic plaques FLIM images from a representative coronary plaque are shown in Fig. 2. The plaque histopathology (Fig. 2a) (corresponding to a perpendicular section cut through the center of the FLIM field of view, dashed line) showed a thick fibrotic (F) collagen-rich plaque in the middle region ofthe arterial section (blue bar), and a thinner (T) elastin- and collagen-rich region toward the bottom side of the section (red bar). Tissue sections underwent Movat pentachrome staining, in which collagen appears as light-blue areas, while elastin appears as dark lines areas. Both the normalized intensity and lifetime maps showed two different regions (Fig. 2b): a middle one (F) correlated with the fibrotic region, and a bottom one (T) correlated with the thinner plaque. The contrast, however, was more dramatic for the 390 nm emission band. The normalized intensity at 390 nm was 60% forthemiddle(f)regionand40%forthebottom(t)region.on the other hand, the normalized intensity at 450 nm was smaller for the middle (F) region (30%) compared with the bottom (T) region (40%). The fluorescence lifetime at 390 nm was longer (3 ns) for the middle (F) region compared with the bottom (T) region (2 ns). On the other hand, the lifetime at 450 nm was ns for both regions. These results are also consistent with arecentstudyoffliminaorticplaquesbyphippset al.,inwhich lifetime values below 400 nm were also shorter for elastin-rich plaques than for collagen-rich plaques (12). The intensity and lifetime maps at 550 nm (data not shown) did not show the two regions. To correlate the artery fluorescence emission with those from purified collagen and elastin extracts (Sigma Aldrich), timeresolved fluorescence spectroscopy of these fluorophores were recorded and analyzed, and their normalized spectrum and lifetimes are also shown(fig. 2c). The fluorescence decaysat each emission wavelength were also deconvolved using the Laguerre method, and average lifetime values were estimated from the deconvolved decays (13). The fluorescence spectra (normalized against the area under the curve) showed that the relative fluorescence intensity of collagen at nm is almost twofold that of elastin; the relative fluorescence emission of elastin at 450 nm is 50% stronger than that of collagen. The fluorescence lifetime of collagen was ns for the entire emission spectrum; while for elastin the lifetime increased from 1.3 ns at nm to 2 ns at 450 nm. The FLIM signal of the thick plaque reflected the fluorescence emission of collagen, while the FLIM signal of the thinner part of the plaque reflected the fluorescence emission of both elastin and collagen. These observations were also confirmed by histopathology. Correlation between FLIM signal and plaque type A total of 64 ROI identified in both the FLIM maps and the histopathology slides were selected and grouped based on histopathology as: Thin plaques (n = 21), Fibrotic plaques (n = 38) and Thick Cap FA (n = 5). Results of the statistical analysis are summarized in Table 1 and Fig. 3. The ANOVA Figure 2. Sample fluorescence lifetime imaging microscopy (FLIM) images of a human coronary atherosclerotic plaque. (a) Plaque histopathology shows a middle collagen-rich fibrotic (F) region and a bottom thin (T) region with significant elastin in the underlying media (Movat pentachrome staining was applied, in which collagen appears as light-blue areas, while elastin appears as dark lines areas). (b) FLIM normalized intensity and lifetime maps shows two different regions: a middle one (F) correlated with the fibrotic region and a bottom one (T) correlated with the thinner plaque. (c) Normalized spectrum and lifetimes of purified collagen and elastin correlate with the fluorescence characteristics of the fibrotic and thin regions, respectively.

4 730 Patrick Thomas et al. Table 1. Results of the ANOVA analysis on the FLIM-derived features. FLIM parameter Band (nm) Thin Fibrotic Thick Cap ANOVA Thin vs Fibrotic Thin vs Thick Cap Fibrotic vs Thick Cap Normalized intensity ± ± ± E) E) E) ± ± ± E) ± ± ± E) E) Lifetime (ns) ± ± ± ± ± ± ± ± ± LEC ± ± ± ± ± ± ± ± ± LEC ± 0.01 )0.07 ± 0.01 )0.09 ± E) E) )0.02 ± 0.01 )0.06 ± 0.01 )0.07 ± )0.05 ± 0.01 )0.07 ± 0.01 )0.05 ± Values are expressed as mean ± SE. FLIM = fluorescence lifetime imaging microscopy analysis on each of the FLIM parameters considered showed that a number of fluorescence features have the potential to distinguish these three plaque types. The pairwise Student s t- test, however, indicated that significant difference was only found between Thin plaques and both Fibrotic and Thick Cap FA plaques, but not between the latter two groups (see last three columns in Table 1). The s of the ANOVA and pairwise tests summarized in Table 1 indicate that the parameters derived from the 390 and 450 nm bands were more relevant than those derived from the 550 nm band. These results also show that both spectral (normalized intensity) and time-resolved (lifetime and LECs) features are both relevant for separating plaque types. For instance, the normalized intensity at the 390 nm band was significantly lower for the Thin plaques compared to the Fibrotic and Thick Cap FA plaques (Fig. 3a). The fluorescence lifetime at the 390 nm band was significantly shorter for the Thin plaques compared with the Fibrotic and Thick Cap FA plaques (Fig. 3b). The LEC coefficients, which offer another way of quantifying the timeresolved properties of the tissue fluorescence, were also significantly different for Thin plaques compared with the Fibrotic and Thick Cap FA plaques (Fig. 3c,d). DISCUSSION Standard white light reflectance angioscopy has been used to detect atherosclerotic plaques based on their color and appearance; however, it provides little information about the plaque composition (3). We propose the use of fluorescence lifetime angioscopy as a complementary modality to image the biochemical composition of atherosclerotic plaques. A FLIM angioscopy system prototype has been developed, and its Figure 3. Fluorescence lifetime imaging microscopy features relevant for distinguishing thin plaques from thick plaques and Thick Cap fibroatheromas (FA): (a) Normalized intensity at 390 nm; (b) lifetime at 390 nm; (c) Laguerre expansion coefficient of order zero (LEC-0) at 390 nm; and (d) Laguerre expansion coefficient of order zero (LEC-1) at 450 nm.

5 Photochemistry and Photobiology, 2010, potential for biochemical imaging of coronary atherosclerotic plaques has been demonstrated. Our results indicate that FLIM images from both Fibrotic plaques and Thick Cap FA reflect the fluorescence characteristic of collagen (which is the dominant fluorophore in these plaques as confirmed by histopathology): their emission is stronger at 390 nm than for the longer wavelengths, and their fluorescence lifetimes are on the order of 1.9 ns. These lifetime values are shorter than those obtained in the purified collagen powder (2.8 ns), indicating the presence of other less dominant fluorophores in these plaques (e.g. elastin and lipids). The FLIM images from the thin plaques, on the other hand, reflect a combination of elastin and collagen fluorescence with elastin being the dominant component: their emission is stronger at 450 nm than for other wavelengths, and their fluorescence lifetimes at 390 nm are 1.5 ns. This is also consistent with the histopathology of the thin plaques, which show some collagen within the lumen and significant elastin content in the underlying media. Results from the statistical analysis on the FLIM-derived features indicate that both spectral and time-resolved characteristics of the plaque fluorescence emission have the potential to distinguish thin lesions from the more advanced Fibrotic and Thick Cap FA lesions. The time-resolved features, however, are less sensitive to intensity artifact (e.g. blood absorption, probeplaque distance variation), which will be difficult to control during intravascular imaging (14). It is also noteworthy to observe that the LECs showed a greater difference among plaque types than the lifetime parameter. This suggests that the LEC can represent a more comprehensive approach for quantifying the time-resolved properties of the plaque fluorescence emission, as found in previous studies (13). Finally, the fluorescence emission above 500 nm was not relevant for distinguishing thin plaques from Fibrotic plaques and Thick Cap FA. However, green autofluorescence has been reported to be relevant to distinguish lipid-rich plaques from fibrotic plaques in previous studies (8). Some major limitations of the FLIM angioscopy system prototype presented here are: (1) the size of the fiber angioscope and (2) the slow acquisition speed. The size of the angioscope can be further reduced by using the imaging bundle for both UV illumination and visible fluorescence transmission. This approach is currently being investigated in our laboratory and will potentially facilitate the fabrication of angioscopes with outer diameters smaller than 1 mm, more suitable for intravascular applications. The acquisition speed in the present prototype is limited by the slow repetition rate of the nitrogen laser and the need to repeat a FLIM measurement for each emission band. We are currently replacing the nitrogen laser by an active Q-switch laser (repetition rates of khz, as opposed to 50 Hz). We are also exploring ways of focusing different fluorescence spectral images onto different regions of the ICCD area, in order to perform the time-resolved detection of all bands simultaneously. These two improvements on the current prototype will theoretically allow video rate FLIM imaging. In addition, blood will play a major role in distorting the autofluorescence signal when measurement is conducted in vivo. Saline flushing (as used in intravascular optical coherence tomography) and video-rate imaging capabilities will both help to reduce or eliminate blood interference. Probe-tissue distance variation could also affect the quantification of the fluorescence spectral and lifetime properties. A side-viewing version of the angioscope would help to minimize artifacts due to probe-tissue distance variations. It is also important to emphasize that timeresolved measurements will be more robust to this type of intensity artifacts, which is one of the main reasons why we are exploring FLIM as opposed to steady-state fluorescence imaging. One main limitation of this study is the absence of lipid-rich plaques in the sample pool. Lipid-rich plaques are the most prone to rupture. Therefore, it will be imperative to extend the scope of this study and include lipid-rich plaques to fully demonstrate the potentials of FLIM angioscopy for detecting rupture-prone atherosclerotic plaques. Nevertheless, this study demonstrates the potentials of FLIM angioscopy for biochemical imaging of human coronary atherosclerotic plaques. Acknowledgements This work was supported by the American Heart Association-Texas Affiliate, Beginning Grant-in-Aid Grant Y and the NIH Grant 1-R21-CA REFERENCES 1. Madjid, M., A. Zarrabi, S. Litovsky, J. T. Willerson and W. Casscells (2004) Finding vulnerable atherosclerotic plaques: Is it worth the effort? ATVB 24(10), Virmani, R., A. P. Burke, A. Farb and F. D. Kolodgie (2006) Pathology of the vulnerable plaque. J. Am. Coll. Cardiol. 47(8 Suppl. C), C13 C Honda, Y. and P. J. Fitzgerald (2008) Frontiers in intravascular imaging technologies. Circulation 117(15), Marcu, L., M. C. Fishbein, J. M. Maarek and W. S. Grundfest (2001) Discrimination of human coronary artery atherosclerotic lipid-rich lesions by time-resolved laser-induced fluorescence spectroscopy. Arterioscler. Thromb. Vasc. Biol. 21(7), Arakawa, K., K. Isoda, T. Ito, K. Nakajima, T. Shibuya and F. Ohsuzu (2002) Fluorescence analysis of biochemical constituents identifies atherosclerotic plaque with a thin fibrous cap. Arterioscler. Thromb. Vasc. Biol. 22(6), Suhling, K., P. M. French and D. Phillips (2005) Time-resolved fluorescence microscopy. Photochem. Photobiol. Sci. 4(1), Elson, D. S., J. A. Jo and L. Marcu (2007) Miniaturized sideviewing imaging probe for fluorescence lifetime imaging (FLIM): Validation with fluorescence dyes, tissue structural proteins and tissue specimens. New J. Phys. 5, Marcu, L., J. A. Jo, Q. Fang, T. Papaioannou, T. Reil, J. H. Qiao, J. D. Baker, J. A. Freischlag and M. C. Fishbein (2009) Detection of rupture-prone atherosclerotic plaques by time-resolved laser-induced fluorescence spectroscopy. Atherosclerosis 204(1), Ramanujan, V. K., J. A. Jo, G. Cantu and B. A. Herman (2008) Spatially resolved fluorescence lifetime mapping of enzyme kinetics in living cells. J. Microsc. 230(Pt 3), Jo, J. A., Q. Fang and L. Marcu (2005) Ultrafast method for the analysis of fluorescence lifetime imaging microscopy data based on the Laguerre expansion technique. IEEE J. Quantum Electron. 11(4), Dabir, A. S., C. A. Trivedi, Y. Ryu, P. Pande and J. A. Jo (2009) Fully automated deconvolution method for on-line analysis of time-resolved fluorescence spectroscopy data based on an iterative Laguerre expansion technique. J. Biomed. Opt. 14(2), Phipps, J., Y. Sun, R. Saroufeem, N. Hatami and L. Marcu (2009) Fluorescence lifetime imaging microscopy for the characterization of atherosclerotic plaques. Proc. Soc. Photo. Opt. Instrum. Eng G. 13. Jo, J. A., Q. Fang, T. Papaioannou, J. D. Baker, A. H. Dorafshar, T. Reil, J. H. Qiao, M. C. Fishbein, J. A. Freischlag and L. Marcu (2006) Laguerre-based method for analysis of time-resolved fluorescence data: Application to in-vivo characterization and diagnosis of atherosclerotic lesions. J. Biomed. Opt. 11(2), Lakowicz, J. R. (2006) Principles of Fluorescence Spectroscopy. Springer, New York.