2.1.

Methods

All RR spectra were collected directly from a region of interest along multiple sites on each specimen. The original RR spectrographs, without subtracting for the baseline of light, were produced using WITec Project 2.10 and ORIGIN 2015 software. The Raman spectra were collected from the four specimens shown in Fig. 1, referred to as FAT, M1, M2, and R1. The numbers superimposed on the specimens indicate the sites where spectra were collected and the fibrous cap thicknesses were examined by standard histopathology (Figs. 1 and 2). The histopathology analysis indicated that arteries had extensive calcification and ossification. Thus measurements were made from the edge of the fibrous cap to the beginning of the calcification. Each sample was placed on the stage of the WITec alpha 300R microconfocal Raman and imaging system as has been described in detail previously.¹⁶ The final spectral resolution was $2{\mathrm{cm}}^{-1}$ in the range of 400 to $3800{\mathrm{cm}}^{-1}$.

Fig. 1

A photograph of the human aorta tissue specimens used in the experiments. Main section “A” shows the VP tissue obtained from NDRI. Four samples cut from specimen “A” are marked as “FAT”, “R1”, “M1”, and “M2” with the numbers indicating the tested and histopathology examined sites.

Fig. 2

Comparisons of normal adventitial arterial wall fat tissue versus three types lesions of atherosclerotic plaques with their RR spectra (column 2), histopathology images (H&E in column 3), cap thicknesses (column 4), and the confocal microscopy photographs (column 5). (a) The image is the histopathology analysis at the test site M2-07. Vulnerable plaque aorta sample analyzed by Columbia University Medical Center, the measurements are only approximate at the regions marked in Fig. 1, the image covered large areas of the artery. (b) Histopathology image of the test site R1-03. (c) Histopathology image of the test site R1-01. The RR spectrum from the seriously calcified atherosclerotic plaques and ossification lesion, the feature vibrational mode is a sharp peak at $964{\mathrm{cm}}^{-1}$ with a weaker band at $1076{\mathrm{cm}}^{-1}$. (d) The image is from confocal microscopy photograph at the tested site human artery fat. (e) The image is from confocal microscopy photograph at the tested site M2-07. (f) The image is from confocal microscopy photograph at the tested site R1-03. (g) The image is from confocal microscopy photograph at the tested site R1-01.

2.2.

Human Tissue Sample

RR spectra of human aorta samples with different states of tissues, including normal fat from the adventitial side arterial wall and three pieces of atherosclerotic plaques lesions, were recorded using a confocal micro-Raman spectrometer. Thirty nine RR spectra were collected from four aortic samples and analyzed. In the RR spectra, a small NIR wing was observed at higher frequencies.

The human aorta specimens exhibiting varying calcific deposits of atherosclerotic vulnerable plaques disease were obtained from the National Disease Research Interchange (NDRI, Philadelphia, Pennsylvania). Some of the tissues exhibited extensive calcification and, in some cases, showed ossification.

The experimental procedures were approved by the City College of the City University of New York, Institutional Review Board (IRB) office. The 12-cm long aorta specimen (Fig. 1, “A”) was dissected longitudinally and then cut into four irregularly shaped $\sim 3\text{-}\mathrm{cm}$ size pieces as shown in Fig. 1 and marked FAT, M1, M2, and R1. The aorta specimen was obtained from a 92-year-old female with hypercholesterolemia who died from respiratory arrest. The patient had a history of hypertension but no history of diabetes. In general, we used about 200 samples of atherosclerotic abdominal aortic tissues and cells. The samples size ranged from microns to centimeters.

2.3.

Pure, High-Density Lipoprotein, and Low-Density Lipoprotein Cholesterols

Pure (99%) cholesterol crystal powder, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) from human plasma in liquid form were purchased from the Sigma-Aldrich Company.

3.1.

Laboratory Grade Samples

Figure 3 shows RR spectra from three materials that will help assign peaks in the spectra from the aortic samples (Fig. 4) to specific compounds. Figure 3(a) shows the RR spectrum of pure cholesterol powder, Fig. 3(b) shows the RR spectrum of fat from the tunica adventitial aorta (Fig. 1, “A” marked as FAT). Figure 3(c) shows the RR spectra of liquid HDL and LDL obtained from human plasma. Figures 3(a)–3(c) highlight the characteristic RR peaks due to cholesterol (located at 704, 1440, and $1674{\mathrm{cm}}^{-1}$), to carotenoids (located at 1007, 1157, and $1517{\mathrm{cm}}^{-1}$), and to the methylene group [$─{\mathrm{CH}}_2─$] vibrational mode and methyl group [$─{\mathrm{CH}}_3$] (located at 2854, 2895, and $2939{\mathrm{cm}}^{-1}$).^28,29

3.2.

Fibrolipid Plaques

The RR spectrum, collected from the sample site numbered M2-07, is shown in Fig. 4(a). Similar RR spectral profiles of M1-02 and R-11 are not shown. A comparison of the spectrum shown in Fig. 4(a) to the RR spectra in Fig. 3 reveals that the Raman peaks common to both spectra are at 1012, 1161, and $1517{\mathrm{cm}}^{-1}$. This suggests an assignment for fatty plaque that is close to that of the lipids in aortic tissue [Figs. 3(b) and 3(c)] and, therefore, is likely to be atherosclerotic aortic tissue.¹¹ Lesions in the sites M2-07, M1-02, and R-11 can be considered as representing an early state of the disease process, a noncalcified atherosclerotic fibrolipid plaque whose RR spectrum contains three distinguishable vibrational modes at 1517, 1161, and $1012{\mathrm{cm}}^{-1}$ arising from intense RR lines of carotenoid molecules. These frequencies have been assigned to vibrational modes of the carotene conjugated polyene backbone, and the $1012\text{-}{\mathrm{cm}}^{-1}$ vibrational mode assigned to aromatic amino acids molecules, tryptophan, and phenylalanine. The weak small peak occurring at $964{\mathrm{cm}}^{-1}$ can be attributed to the initial calcification of early stage plaque lesions.

In these lesion sites, a second set of Raman spectral molecular fingerprints arises from the macromolecules of lipids and lipoproteins.³⁰ These characteristic vibrational modes are at 2854, 2892, and $2934{\mathrm{cm}}^{-1}$, as shown in all three spectra in Fig. 3. Key differences in these peaks are seen in the spectra obtained from normal aortic fat tissue and fibrolipid atherosclerotic plaque tissue.^{11,12,14–16} For example, as shown in Figs. 3(a) and 3(b), in HDL/LDL powder and normal aortic fat, the presence of a sharp methylene ($─{\mathrm{CH}}_2$) peak is observed at $2854{\mathrm{cm}}^{-1}$. Additionally, the characteristic band of spectral peaks observed at 2932 (Fermi resonance of a ${\mathrm{CH}}_3$ mode) and $2895{\mathrm{cm}}^{-1}$ was attributed to the vibrations of methylene (Fermi resonance of a $─{\mathrm{CH}}_2$ mode) groups. However, in noncalcified atherosclerotic fibrolipid plaque tissue [Fig. 4(a)], the Raman spectral peak at $2854{\mathrm{cm}}^{-1}$ is greatly diminished, while the other two distinct Raman spectral peaks of lipid and lipoproteins at 2895 and $2932{\mathrm{cm}}^{-1}$ decayed greatly. The changes in peak intensity and position in the Raman spectral profiles indicate an important finding. These two characteristics provide a signal to track the transformation of the composition and protein conformation in the cells, as well as the changes in the molecular structure of fibrolipid plaque lesion tissues.³¹

3.3.

Vulnerable Atherosclerotic Plaques

Figure 4(b) shows a typical spectrum for another type of atherosclerotic plaque, obtained from the R1-03 site (and similar to the RR spectra of R1-04 and M1-01, which are not shown). The three salient characteristics of Raman spectra in this type are: (1) a large lipid pool that has distinct lipid Raman peaks at 2888, 1678, and $1456{\mathrm{cm}}^{-1}$, with a peak at $964{\mathrm{cm}}^{-1}$ that possibly arises from calcified fragments of extracellular debris in ruptured plaque. (2) The enhanced amide III mode at $1228{\mathrm{cm}}^{-1}$ for these type lesions has no shift in its peak position. The sharp intense stable marker at $758{\mathrm{cm}}^{-1}$ can be interpreted as an independent fingerprint. It represents the breathing mode, which gives the most information about the status of red blood cells, as well as direct measures of the heme group in hemoglobin. It is also a characteristic peak of tryptophan, but it is intense and it changes in intensity with the state of atherosclerotic plaques, suggesting that its origin is from heme groups. (3) The intense resonance enhancements of the Raman spectra were displayed in three groups of peaks: (a) one group at 1131 to $1173{\mathrm{cm}}^{-1}$ and a sharp intense peak at $1131{\mathrm{cm}}^{-1}$ $[\nu (\mathrm{C}─\mathrm{C})\delta (\mathrm{COH})]$ that arose from the fatty acid (i.e., lipid assignment). The significant peak shifts in proteins were observed: $1157{\mathrm{cm}}^{-1}$, at $\nu (\mathrm{C}─\mathrm{C})$, in which the carotenoid was not in normal tissue; $1170{\mathrm{cm}}^{-1}$, where the $\mathrm{C}─\mathrm{H}$ in-plane bending mode of tyrosine (collagen type I) was present; and 1173 and $1174{\mathrm{cm}}^{-1}$, which are the peaks due to tyrosine, phenylalanine, and the $\mathrm{C}─\mathrm{H}$ bend (protein). These spectral profile changes and peak position shifts reflect molecular structure changes corresponding to the difference in fibrous cap thickness. For example, the R1-03 site has a VP cap thickness of $48.68\mu \mathrm{m}$ and the R1-04 site has a similar thickness of $60.96\mu \mathrm{m}$, both of which are thinner than the $65\text{-}\mu \mathrm{m}$ thickness that defines a highly unstable thin cap fibroatheroma (TCFA) plaque. Data points for both of these samples are included in Fig. 5 and R1-03 is shown in Fig. 2. (b) The second group between 1316 to $1374{\mathrm{cm}}^{-1}$ represents several remarkable modes: $1316{\mathrm{cm}}^{-1}$ depicts the G (ring breathing modes of the DNA/RNA bases) $─\mathrm{C}─\mathrm{H}$ deformation (protein) and amide III ($\alpha$-helix), $1370{\mathrm{cm}}^{-1}$ shows the contribution of T, A, G (ring breathing modes of the DNA/RNA, bases-protein), and tryptophan. (c) The third group, between 1554 and $1606{\mathrm{cm}}^{-1}$, contains modes which are marked by the following peaks: the $1554\text{-}{\mathrm{cm}}^{-1}$ peak from the protein amide II, which was derived mainly from the in-plane $\mathrm{N}─\mathrm{H}$ bending, the $1589\text{-}{\mathrm{cm}}^{-1}$ peak indicates the $\mathrm{C}─\mathrm{C}$ stretching and phosphate ($\mathrm{O}─\mathrm{P}─\mathrm{O}$) symmetric stretching vibrational mode, which is a characteristic of nucleic acids, as well as a known mode for tryptophan, and the $1606\text{-}{\mathrm{cm}}^{-1}$ peak that is due to the $\mathrm{C}─\mathrm{O}$ stretching and $\mathrm{C}─\mathrm{C}$ bending in phenylalanine and tyrosine.^32–35

Fig. 5

Intensity changes of the RR peak at the $2895\text{-}{\mathrm{cm}}^{-1}$ vibrational mode plotted as signal-to-noise ratio versus VP cap thickness ($\mu \mathrm{m}$). Two exponential fits to the data are provided illustrating the different trends of Raman signals from smaller cap thickness plaques versus larger cap thickness caps. The two straight lines intercept at $66\mu \mathrm{m}$, very close to the accepted definition for TCFA of $65\mu \mathrm{m}$.

These three distinct groups consist mainly of enhanced Raman peaks for proteins, which suggests that the 532-nm excitation wavelength matched (or closely matched) the molecular absorption wavelengths for compounds in the cells and tissues. For example, the metalloprotein, hemoglobin, has one absorption band at 534 nm. Similarly, the mitochondrial electron transport protein, cytochrome C, has one absorption wavelength at 552 nm (under hypoxia conditions) or the two-photon absorption process under RR conditions that may allow the R1, M1, and M2 sites of the ruptured plaque processes to be seen.³² RR peaks associated with proteins on the molecular level in ruptured fibrolipid plaques tissues were shown and thought to be due to heme proteins, such as the cytochromes that reside in the mitochondria.

These RR spectra from arterial sites may reveal the processes of plaque rupture or thrombosis.

At these positions (R1-03, R1-04, and M1-01), lesions may be classified as advanced in the clinical stage. This is because we found (a) ruptured atherosclerotic plaques with surface erosions, hemorrhages, and thrombus at the test site, as can be seen in the image labeled Fig. 4(b) and column 5 in Fig. 2; (b) the fibrous cap thickness at the tested location on the sample (column 4 of Fig. 2) is a very thin $48.68\mu \mathrm{m}$, as measured by histopathology; and (c) the RR spectrum at tested site R1-03 in Fig. 5 showed enhanced RR peaks from collagen and fibro muscular tissue layers covering the extracellular lipid pool at 1456, 1554, 1640, and $1678{\mathrm{cm}}^{-1}$ indicating that the plaque has progressed to the clinically silent and advanced stage.^14,26

3.4.

Calcified and Ossified Plaques

Figure 4(c) shows a typical RR spectrum for the third set of lesions where calcification has evolved into ossification. In the highly calcified deposits on the atherosclerotic plaques and ossified lesion, the featured vibrational mode is a sharp peak at $964{\mathrm{cm}}^{-1}$ with a weaker band at $1076{\mathrm{cm}}^{-1}$ that arises from the symmetric stretching vibration of ${\nu }_1{\mathrm{PO}}_4^{3-}$, calcium-phosphate stretch band (contains high quantities of cholesterol), and a phosphate symmetric stretching vibration of calcium hydroxyapatite (HA) and quinoid ring in-plane deformation. In addition, the $1078\text{-}{\mathrm{cm}}^{-1}$ band arose from the symmetric stretching vibration of ${\nu }_1{\mathrm{PO}}_4^{3-}$ [phosphate of (HA)-single phase HA], $\nu (\mathrm{C}─\mathrm{C})$ or $\nu (\mathrm{C}─\mathrm{O})$, and phospholipids. In this set of lesions, all the characteristic Raman peaks of lipids and proteins, which appeared in Figs. 4(a) and 4(c), were decayed and diminished. This indicates that, in this state, the main components may be crystalline cholesterol, cholesteryl esters, and phospholipids.^13,15,16,28

3.5.

Fibrous Cap Thickness

The RR spectra in different types of atherosclerotic plaque are clearly correlated to the changes of VP fibrous cap thickness as measured by standard histopathology methods and shown in Figs. 2 and 5. The signal-to-noise ratio of the peak intensity at $2895{\mathrm{cm}}^{-1}$ for two classes of plaques varies exponentially with fibrous cap thickness (Fig. 5), but with different exponential decay rates. The samples with small fibrous cap thicknesses exhibit one rate of exponential decay, while those with thicker caps exhibit a much smaller decay rate. This suggest two different cap morphologies for the two types of plaques since the morphology will determine the opacity of the cap to the RR scattered light from the lipid pool and, therefore, the decay rate. The two decay curves intersect near the cap thickness value of $66\mu \mathrm{m}$. This value is very close to the conventional TCFA definition of $65\mu \mathrm{m}$.^36,37 The result in Fig. 5 is attributed to the transition from ballistic photons propagating through thin caps to diffusive photons propagating through a thicker cap, which occurs at $66\mu \mathrm{m}$.³⁸

Confirmation of the importance of this finding is left for future work but it suggests that RR measurements of cap thickness can be used to detect the presence of a highly unstable TCFA plaque. Figure 2 shows a comparison of normal arterial fat tissue versus three types lesions of atherosclerotic plaques corresponding to their RR spectra, histopathology images (Fig. 2, column 3), fibrous cap thickness, and the confocal microscopy photographs (Fig. 2, columns 4 and 5). Table 1 lists the vibrational assignments of the modes.

Table 1

Human atherosclerosis aortic specimens obtained from NDRI selected RR spectral peak position and assignment of specimen tissues.