Open Access
Add to BibSonomyAbstract
This paper presents a comprehensive computational analysis of the spectral optical response of epithelial tissues labeled with gold nanoparticles. Monte Carlo modeling is employed to simulate nanoparticle-induced changes in reflectance signals and to assess whether labeling can generate sufficient exogenous contrast that can better pinpoint precancer progression. Simulation results suggest that the observed contrast profile is highly dependent on a series of factors including the labeling scheme, optical sensor geometry, and wavelength. It is evident, however, that selection of an optimal labeling and sensing strategy can lead to a significant enhancement of the inherent positive or negative contrast and can improve the diagnostic potential of optical measurements.
©2013 Optical Society of America
1. Introduction
Nanoplatform-based molecular detection has emerged as a potential tool for the early diagnosis of a wide variety of diseases including cancer [1]. The particular detection scheme to be employed is dictated by the physical properties of nanostructures deployed to pathological areas. Optically active metal nanoparticles have unique absorption and scattering characteristics that can significantly change the optical response of target tissues leading to contrast enhancement in reflectance signals. These particles can be rendered cancer-specific through conjugation with biomolecules that have high affinity for biomarkers over-expressed in cancer cells. Indeed, recent experimental studies provide evidence for the feasibility of nanoparticle-enhanced reflectance measurements to differentiate between normal and precancerous tissues [2–6].
Numerous efforts have been made to investigate the spectral absorption and scattering characteristics of metal nanoparticles with respect to their size, shape, and material composition [7–13]. The results obtained provide a map of the resonant optical behavior of these particles and generally serve as a guide for the selection of a particular type that is best suited for the intended application. If nanoparticle-labeled tissues are to be interrogated with techniques that depend on diffuse or multiply scattered light as in [14,15], however, there is also a concurrent need to model photon propagation at the bulk tissue level and to predict the changes in the overall reflectance response of tissues due to addition of nanoparticles. This will pave the way for a quantitative assessment of the diagnostic potential of nanoplatform-based optical contrast enhancement in an optically diffusive setting.
The Monte Carlo (MC) method is a rigorous computational tool that can provide useful insights into photon propagation in tissues labeled with metal nanoparticles. Yet, there exist a fairly limited number of MC studies that address this issue and attempt to clarify how these labeling agents alter the intrinsic reflectance properties of tissues. Lin et al. [16] have employed MC modeling to analyze how tissue response changes with varying nanoshell size and concentration. Their results indicate that a considerable change in reflectance can be observed even with a very small concentration of gold nanoshells. Another study reported by Zagaynova et al. [17] investigates the use of gold nanoshells as contrast agents for optical coherence tomography imaging of skin tissue. The simulations performed confirm the possibility of differential contrast enhancement across superficial skin layers. As a follow-up on this work, Kirillin et al. [18] have carried out MC simulations to better understand the contrasting properties of gold nanoshells and titanium dioxide nanoparticles. Simulation results exhibit good qualitative agreement with experimental images and prove that nanoparticles are directly responsible for elevated contrast observed in the images. More recently, we have presented an MC study to reveal the importance of using a proper form of scattering phase function when modeling photon propagation in nanoparticle-labeled epithelial tissues [19]. Our results indicate that nanoparticle-induced changes in the reflectance response of tissues depend not only on the properties of these particles but also on the optical sensor geometry employed. The results also suggest that an extended spectral analysis of optical signals obtained from labeled tissues is requisite for a complete assessment of the diagnostic potential of nanoparticle-enhanced measurements.
The goal of the research described in this work is to carry out MC simulations and present a systematic investigation of photon propagation in nanoparticle-labeled epithelial tissues. Models for normal tissues, unlabeled precancerous tissues, and precancerous tissues labeled with gold nanospheres are constructed and the spectral reflectance response of these different tissue states is analyzed to quantify the extent of achievable contrast enhancement due to external labeling. Various fiber-based designs with differing optical probing depths are simulated to reveal the effect of the source-detector geometry on the observed contrast profile. It is important to note that this is the first comprehensive computational study of its kind that also lays the groundwork for future research on characterization of any optically diffusive medium with metal nanoparticle inclusions.
2. Methods
2.1 Construction of normal and precancerous tissue models
Most cancers originate in epithelial tissues that cover the inner and outer surfaces of the body [6]. An epithelial tissue is generally described as consisting of two distinct layers with different structural and optical properties. The top layer or the epithelium consists of cells, whereas the underlying stroma is mainly composed of collagen fibers and blood vessels. These tissue constituents act as sources of scattering and absorption and they undergo well-documented changes during precancer progression; morphological and biochemical alterations in proliferating cells lead to increased scattering in the epithelium, degradation of collagen fibers is associated with decreased stromal scattering, and elevated hemoglobin concentration due to angiogenic activity causes increased stromal absorption [20,21].
The modeling approach employed in this study accounted for all the structural and optical properties of normal and precancerous epithelial tissues. As depicted schematically in Fig. 1 , the top epithelial layer was assigned a thickness of 300 μm and the stromal layer underneath was assumed to be semi-infinite. The refractive indices of these tissue layers were both set to 1.35, which has been reported to be a reasonable approximation for the matrix that surrounds tissue constituents [22]. The wavelength-dependent scattering coefficients μs and absorption coefficients μa used to construct normal and precancerous tissue models were based on values reported in the literature [21], and are plotted in Fig. 1 for every 20 nm over the wavelength range of 400 to 700 nm. The descriptive subscripts ‘epi’ and ‘str’ in parts (b)–(e) stand for epithelium and stroma, respectively. Note that precancer progression does not change the absorption properties of the epithelial layer and hence the absorption coefficients for normal and precancerous epithelium overlap. Henyey-Greenstein phase functions with anisotropy factors of gepi = 0.95 and gstr = 0.88 were used to approximate the angular scattering distributions in the epithelial and stromal layers, and these factors were assumed to be independent of wavelength and tissue state [21].
Fig. 1 (a) Schematic depiction of epithelial tissue as a two-layer medium. (b)–(e) Optical coefficients μs and μa used to construct normal and precancerous tissue models. The descriptive subscripts ‘epi’ and ‘str’ stand for epithelium and stroma, respectively. Note that the absorption coefficients for normal and precancerous epithelium have the same values.
2.2 Construction of nanoparticle-labeled precancerous tissue models
When gold nanospheres targeted to cellular cancer biomarkers are delivered to precancerous tissues either topically or systemically, they are expected to accumulate in the top epithelial layer. Note that a complete and essentially uniform epithelial distribution of these particles is possible through administration of permeation enhancers [6]. Herein, the term ‘nanoparticle-labeled’ is used in association with such a tissue state; while the stromal optical properties remain unchanged, the scattering and absorption properties of the epithelium need to be modified to allow for the differential optical effect of exogenous particles.
In the simulations presented here, nanospheres with diameters of 40, 80, or 120 nm were used. The nanospheres were assumed to be embedded in a medium with a refractive index of 1.35 and complex dielectric values for gold were adapted from experimental measurements as previously described [19]. The volume fractions tested were f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%, which correspond to about 1.5 × 1011 to 1.5 × 1012 particles/mL for 40-nm nanospheres, 1.9 × 1010 to 1.9 × 1011 particles/mL for 80-nm nanospheres, and 5.5 × 109 to 5.5 × 1010 particles/mL for 120-nm nanospheres. These concentration ranges are comparable to those reported in other studies and are low enough to ensure independent scattering [3,15–18]. Mie theory was used to compute the optical cross sections and scattering phase functions of single nanospheres for every 20 nm over the wavelength range of 400 to 700 nm. Differential scattering and absorption coefficients for different volume fractions were then computed and simply added to intrinsic tissue coefficients as [19]
where the descriptive subscript ‘np’ stands for nanoparticle and the variables marked with an asterisk represent the modified optical properties of the epithelial layer. Scattering phase functions of labeled epithelium were generated with a weighting scheme that accounted for the relative scattering strengths of unlabeled tissue and exogenous particles. Let pepi and pnp denote the phase functions of unlabeled epithelium and single nanospheres, respectively. The modified phase functions can be expressed as [19]such that the anisotropy factors are related by2.3 MC simulation parameters
It is well established that the detected reflectance spectrum of a multilayered diffusive medium is highly dependent on the probing depth and hence the illumination-collection setup [20]. As illustrated in Fig. 1, the optical sensor geometries employed in this work consisted of a single source fiber and multiple detector fibers positioned at various distances from the source. Each fiber was assigned a diameter of 100 μm and a numerical aperture of 0.11 (in air). The fibers were all assumed to be in contact with the tissue surface and their refractive indices were set to 1.5. The space in between the fibers was index matched to the epithelial layer to suppress any internal reflection of incoming photons. Two configurations were tested to assess the effect of fiber orientation on the observed reflectance profile. In the first configuration, the source and detector fibers were perpendicular to the tissue surface. In the second configuration, the distal ends of a given source-detector fiber pair were tilted toward each other and each fiber made an angle of 30° with respect to the tissue normal. For both configurations tested, the detector fibers were located at center-to-center distances of 150, 300, 500, and 1000 μm from the source fiber.
Details regarding the implementation of the C/C + + MC code used in this study have been described elsewhere [19,21,23]. All the simulations were carried out with 108 input photons. The results presented represent averages over three different simulation runs. Standard errors were also computed and displayed to provide evidence for model convergence. Note that the simulations were performed at the same wavelengths that were used for Mie theory calculations. The results at these sixteen wavelengths were then interpolated using piecewise cubic Hermite polynomials to generate spectral plots with increments of 5 nm. In addition to the total number of reflected photons detected, penetration depth statistics for different source-detector fiber pairs are also reported to provide further insight into photon propagation in nanoparticle-labeled tissues.
3. Results
3.1 Optical properties of nanoparticle-labeled precancerous epithelium
Figure 2 shows the modified scattering coefficients μs,epi*, absorption coefficients μa,epi*, and anisotropy factors gepi* of nanoparticle-labeled precancerous epithelium for different labeling schemes. The results for 40-nm spheres are plotted in the first column [Figs. 2(a)–2(c)], the results for 80-nm nanospheres are plotted in the second column [Figs. 2(d)–2(f)], and the results for 120-nm nanospheres are plotted in the third column [Figs. 2(g)–2(i)]. For each nanosphere size, three different volume fractions are considered: green, blue, and red curves correspond to progressively increasing volume fractions of f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%, respectively. Note that optical properties of unlabeled precancerous epithelium are also plotted in black to enable a direct comparison. It is also important to emphasize that the values of anisotropy factors obtained for labeled epithelium are not sufficient to fully specify the angular scattering probability distributions used in the simulations since the phase functions generated for modeling purposes represent weighted combinations of Henyey-Greenstein and Mie components [19]. They are, however, included to give an overall insight into how the angular scattering properties of tissues change due to addition of nanoparticles.
Fig. 2 Scattering coefficients μs,epi*, absorption coefficients μa,epi*, and anisotropy factors gepi* of nanoparticle-labeled precancerous epithelium for different labeling schemes: (a)–(c) 40-nm nanospheres, (d)–(f) 80-nm nanospheres, and (g)–(i) 120-nm nanospheres. The volume fractions considered in each case are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. Optical properties of unlabeled precancerous epithelium are also plotted to enable a direct comparison.
The results displayed in Fig. 2 indicate that small nanospheres can produce significant changes in the absorption coefficient of tissues. On the other hand, alterations in the scattering coefficient and the anisotropy factor are minimal even for the highest volume fraction considered [Figs. 2(a)–2(c)]. Larger nanospheres can cause extensive changes in scattering properties, whereas the effect of labeling on absorption coefficient becomes less significant as the size of nanospheres increases [Figs. 2(d)–2(i)]. Overall, the most pronounced changes in scattering properties are observed at wavelengths that correspond to the respective scattering cross section maxima of 40-, 80-, and 120-nm nanospheres, and the most pronounced changes in absorption properties occur at wavelengths that correspond to the respective absorption cross section maxima of these particles. The wavelength shift in spectral profiles and the broadening of linewidth with increasing nanosphere size are all manifestations of surface plasmon resonance effects, which are fully accounted for in Mie theory calculations [7–11].
3.2 Spectral reflectance response of nanoparticle-labeled epithelial tissue
Figures 3 through 5 show the spectral reflectance profile of different tissue states when the source and detector fibers are oriented perpendicular to the tissue surface. The results for normal tissue are plotted in dotted black, whereas the results for unlabeled precancerous tissue are plotted in solid black. Colored curves display the results for precancerous tissue labeled with 40-, 80-, or 120-nm gold nanospheres: green, blue, and red curves represent progressively increasing volume fractions of f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%, respectively. In all cases, each figure part corresponds to a different center-to-center source-detector separation (sds). Note that the error bars computed over three simulation runs are plotted only for selected wavelengths to maintain clarity; these error bars provide evidence for the reliability of MC output.
Fig. 3 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 40-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The source and detector fibers are oriented perpendicular to the tissue surface and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
The results shown in Figs. 3 through 5 indicate that when the fibers are oriented perpendicular to the tissue surface, the reflectance intensity of unlabeled precancerous tissue is inherently lower compared to that of normal tissue and hence the inherent diagnostic contrast is negative. Addition of 40- or 80-nm gold nanospheres leads to a further reduction in detected reflectance intensity for all the fiber separations considered; this negative contrast enhancement relative to normal tissue is particularly significant over the first half of the spectral range when the volume fraction is high [Figs. 3 and 4 ].
Fig. 4 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 80-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The source and detector fibers are oriented perpendicular to the tissue surface and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
Similar trends are observed with 120-nm nanospheres when sds = 300, 500, or 1000 μm, but the reflectance response for sds = 150 μm is highly dependent on the wavelength; labeling with 120-nm nanospheres at a volume fraction of f3 = 0.005% causes an enhancement of the inherent negative contrast for wavelengths <~550 nm, whereas the predicted reflectance intensity of labeled precancerous tissue exceeds that of unlabeled precancerous tissue for wavelengths >~550 nm and even approaches that of normal tissue nullifying the inherent contrast at a peak wavelength of ~640 nm [Fig. 5(a) ].
Fig. 5 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 120-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The source and detector fibers are oriented perpendicular to the tissue surface and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
Figures 6 through 8 show the spectral reflectance profile of different tissue states for tilted fibers. The plots in these figures follow the same line styles and the color scheme employed in Figs. 3 through 5. The error bars again provide evidence for model convergence. It is apparent that the inherent diagnostic contrast for sds = 150 and 300 μm is positive; the reflectance intensity of unlabeled precancerous tissue is higher than that obtained for normal tissue. As the fiber separation increases, however, this trend changes; for sds = 500 and 1000 μm, the inherent diagnostic contrast turns out to be negative as in the case of perpendicular fibers. The overall influence of addition of nanospheres is observed to be highly unpredictable. Nanosphere size, volume fraction, distance between the source and detector fibers, and wavelength are all determining factors for the resulting contrast profile.
Fig. 6 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 40-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The distal ends of the source and detector fibers are tilted toward each other and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
When sds = 150 μm, addition of 40-nm nanospheres leads to a decrease in reflectance intensity for wavelengths <~560 nm and this reduces the inherent positive contrast relative to normal tissue. For >~560 nm, addition of 40-nm nanospheres leads to an increase in reflectance intensity and this leads to a positive contrast enhancement [Fig. 6(a)]. When sds = 300 μm, addition of 40-nm nanospheres again results in a rather interesting contrast profile; the highest volume fraction tested causes a significant signal reduction for wavelengths <~600 nm, but no changes are observed for the rest of the spectral range [Fig. 6(b)]. When sds = 500 or 1000 μm, addition of 40-nm nanospheres leads to a consistent signal reduction and a negative contrast enhancement almost over the entire wavelength range [Figs. 6(c)–6(d)].
It is evident that 80-nm nanospheres can act as quite effective contrast enhancers when the fibers are tilted and separated by a distance of sds = 150 μm. As the volume fraction is increased, the intensity increases and the peak of spectral reflectance shifts to longer wavelengths. The level of positive contrast enhancement relative to normal tissue is especially significant at ~600 nm when the volume fraction is f3 = 0.005% [Fig. 7(a) ]. When sds = 300 μm, however, this particular labeling scheme brings the intensity level of precancerous tissue down to that of normal tissue and nullifies the inherent contrast over the first half of the spectral range. Useful positive contrast is obtained only for wavelengths >~580 nm and peaks at ~640 nm [Fig. 7(b)]. For larger fiber separations, addition of 80-nm nanospheres results in a negative contrast that increases with increasing volume fraction [Figs. 7(c)–7(d)].
Fig. 7 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 80-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The distal ends of the source and detector fibers are tilted toward each other and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
A large extent of positive contrast enhancement can be achieved with 120-nm nanospheres when sds = 150 or 300 μm, but this is restricted to wavelengths >~500 nm. As in the case of 80-nm nanospheres, the peak of spectral reflectance shifts to longer wavelengths with increasing volume fraction. It is also important to point at the influence of fiber separation on peak position; a volume fraction of f3 = 0.005% results in a reflectance peak at ~640 nm when sds = 150 μm and this peak shifts further to ~680 nm when sds = 300 μm [Figs. 8(a) –8(b)]. Larger fiber separations are again associated with a signal reduction and hence a negative contrast enhancement.
Fig. 8 Spectral reflectance profile of normal tissue, unlabeled precancerous tissue, and precancerous tissue labeled with 120-nm gold nanospheres. Three different volume fractions considered are f1 = 0.0005%, f2 = 0.001%, and f3 = 0.005%. The distal ends of the source and detector fibers are tilted toward each other and are separated by a center-to-center distance of (a) 150 μm, (b) 300 μm, (c) 500 μm, and (d) 1000 μm.
3.3 Penetration depth statistics
It is also important to monitor the penetration depth profiles of detected photons to assess the effect of labeling on photon propagation characteristics. Penetration depth for a given photon corresponds to the maximum depth in tissue at which the photon undergoes a scattering event. Mean penetration depth is then computed as the average over all detected photons for a given simulation run. The results obtained in this work show that progression of precancer is always associated with an inherent decrease in mean penetration depth regardless of the optical sensor geometry employed. When the fibers are perpendicular to the tissue surface, addition of nanospheres decreases the mean penetration depth for sds = 150 or 300 μm. While this decrease is generally most extensive for 80-nm nanospheres over the first half of the spectral range, 120-nm nanospheres are observed to be more effective in reducing the mean penetration depth over the second half of the spectral range. When sds = 500 or 1000 μm, labeling with 40-nm nanospheres tends to increase the mean penetration depth relative to unlabeled precancerous tissue and labeling with 120-nm nanospheres leads to a consistent decrease in mean penetration depth.
When the fibers are tilted, addition of nanospheres tends to cause a decrease in mean penetration depth for sds = 150, 300, or 500 μm. It is observed that 40- or 80-nm nanospheres are more effective in reducing the mean penetration depth over the first half of the spectral range, whereas the reduction in mean penetration depth is more extensive for 120-nm nanospheres over the second half of the spectral range. When sds = 1000 μm, photons penetrate deeper into tissue compared to the unlabeled case; this difference appears to be most significant for 40-nm nanospheres, especially for short wavelengths.
Table 1 provides sample penetration depth statistics for selected optical sensor geometries, labeling schemes, and wavelengths. The volume fraction considered in all cases is f3 = 0.005%, which corresponds to the highest concentration of nanospheres. Note that the results reported are averages over three simulation runs and the associated standard errors are also included to provide evidence for model convergence. The statistics presented corroborate some general trends alluded to in the preceding paragraphs. Also shown in the table are percentages of detected photons that propagate only in the epithelial layer and do not penetrate the stroma underneath. These percentages indicate that labeling can significantly alter the layer selectivity of a given optical sensor geometry.
Table 1. Penetration depth statistics for selected optical sensor geometries, labeling schemes, and wavelengths. Note that the volume fraction considered in all cases is f3 = 0.005%. The percentages in parentheses indicate the fraction of photons collected from the epithelial layer (< 300 μm).
3.4 Quantification of the extent of achievable contrast enhancement
The results presented so far are based on the assumption that labeling is specific to precancerous tissue and there is no accumulation of nanoparticles in normal epithelium. It is thus worthwhile to investigate any potential influence of such an unwanted accumulation on reflectance signals. Figure 9 presents a quantitative assessment of the extent of nanoparticle-induced contrast enhancement for selected optical sensor geometries, labeling schemes, and wavelengths. The eight cases considered in parts (a)–(h) follow the order given in Table 1 and are characterized by a strong potential for high contrast. Contrast can in general terms be defined as the intensity of precancerous tissue relative to the intensity of normal tissue. In Fig. 9, the black bar represents the intensity of unlabeled precancerous tissue relative to the intensity of unlabeled normal tissue, i.e. the inherent contrast. The red bar represents the intensity of precancerous tissue labeled at a volume fraction of f3 = 0.005% relative to the intensity of unlabeled normal tissue, i.e. the nanoparticle-induced contrast assuming that there is no accumulation of nanoparticles in normal tissue. Finally, the gray bars represent the intensity of precancerous tissue labeled at a volume fraction of f3 = 0.005% relative to the intensity of normal tissue labeled at 5%, 10%, 50%, or 100% of f3; these demonstrate the influence of progressively increasing levels of nanoparticle accumulation in normal tissue. Note that the vertical scales are expressed in a log format such that a positive value corresponds to a positive contrast, while a negative value corresponds to a negative contrast. Relative intensity values are also directly displayed next to the bars to simplify the interpretation of the results. In all cases, the error bars computed over three simulation runs are negligibly small and have been omitted from the plots.
Fig. 9 Assessment of nanoparticle-induced contrast enhancement for selected optical sensor geometries, labeling schemes, and wavelengths as in Table 1: (a) perpendicular fibers, sds = 150 μm, 40-nm nanospheres, wavelength = 520 nm, (b) perpendicular fibers, sds = 150 μm, 80-nm nanospheres, wavelength = 520 nm, (c) perpendicular fibers, sds = 500 μm, 120-nm nanospheres, wavelength = 520 nm, (d) tilted fibers, sds = 300 μm, 40-nm nanospheres, wavelength = 520 nm, (e) tilted fibers, sds = 150 μm, 80-nm nanospheres, wavelength = 600 nm, (f) tilted fibers, sds = 1000 μm, 80-nm nanospheres, wavelength = 560 nm, (g) tilted fibers, sds = 150 μm, 120-nm nanospheres, wavelength = 640 nm, and (h) tilted fibers, sds = 300 μm, 120-nm nanospheres, wavelength = 680 nm. Precancerous tissue, if labeled, has a nanosphere volume fraction of f3 = 0.005%. Normal tissue, if labeled, has a nanosphere volume fraction of 5%, 10%, 50%, or 100% of f3.
The bar plots in Fig. 9 show that the selected combinations of sensor geometries, labeling schemes, and wavelengths are indeed associated with considerable changes in detected reflectance signals. It is also observed that significant contrast enhancement is still achievable with low levels of nanosphere accumulation in normal tissue. If the extent of labeling in normal tissue is increased to 50% or 100% of f3, however, log10(Contrast) values generally tend to approach the zero baseline; this indicates that the diagnostic contrast between normal and precancerous tissues deteriorates and even disappears when nanosphere accumulation is at comparable levels in both tissue types.
4. Discussion
The results presented in this study demonstrate that nanoparticle-induced optical contrast enhancement in epithelial tissues can improve the diagnostic potential of reflectance measurements. As evidenced in Figs. 3 through 8, however, the observed contrast profile is highly dependent on a series of factors including the labeling scheme, optical sensor geometry, and wavelength; an intricate interplay of coincident changes in epithelial scattering and absorption properties can manifest as a significant enhancement of the inherent positive or negative contrast only when an optimum source-detector geometry is employed.
When the source and detector fibers are oriented perpendicular to the tissue surface, the inherent diagnostic contrast is negative. As noted previously in [19,21,23], photons collected from the stromal layer are the main contributors to the detected signal, and the drop in intensity with development of precancer is attributed to decreased scattering and increased absorption in the stroma. When the tissue is labeled, a nanoparticle-induced increase in epithelial absorption appears to have a dominating influence leading to a decrease in reflectance intensity for almost all the cases considered; the valleys around ~520 or 540 nm deepen with increasing volume fraction, consistent with the growing contribution of μa,np to the modified absorption coefficients. The only exception to this trend occurs when a high enough concentration of 120-nm nanospheres significantly alters the scattering properties of the epithelial layer; this may result in an increase in detected intensity if the fibers are placed close to each other.
The main motivation behind using tilted fibers is to enhance sensitivity to the top epithelial layer [19,23]. This is indeed the case for small fiber separations that are effective in isolating photons confined to the epithelium; the inherent diagnostic contrast is then expected to be positive due to increased epithelial scattering associated with development of precancer. When the tissue is labeled, a nanoparticle-induced increase in epithelial scattering may have a dominating influence on detected reflectance. The results obtained for 80- and 120-nm nanospheres point to the possibility of a considerable increase in reflectance intensity when the concentration of added particles is high. Note that the resonance peaks over the second half of the spectral range become more pronounced with increasing volume fraction, consistent with the growing contribution of μs,np and gnp to the modified scattering properties.
Overall, it is evident that it may not always be possible to foresee the effect of addition of nanoparticles since a particular labeling and sensing strategy can lead to a positive, negative, or null contrast depending on the operating wavelength. This is again due to the competing effects of nanoparticle-induced changes in epithelial scattering and absorption properties. Another interesting and perhaps nontrivial consequence is noted when tilted fibers are employed to enhance the inherent positive contrast: the peak of spectral reflectance shifts to longer wavelengths with increasing nanoparticle concentration or increasing fiber separation and optimal contrast between normal and precancerous tissues is not always obtained at wavelengths that correspond to the respective scattering cross section maxima of nanoparticles. Based on these observations, it is apparent that a prior computational analysis is necessary to predict the intensity profile of labeled precancerous tissue relative to normal tissue so that any misinterpretation of diffuse reflectance measurements will be avoided. An added advantage of such an analysis is the ability to monitor the penetration depth profiles of detected photons. As demonstrated by the results presented in Table 1, nanoparticle labeling can significantly alter the probing depth of an optical sensor geometry. It is thus important to estimate the relative contributions of epithelial and stromal signals and to predetermine the origin of the observed contrast.
Predictions from simulations indicate that a positive contrast of up to ~18 or a negative contrast of up to ~0.01 can be obtained when gold nanospheres are assumed to be selectively targeted to cancer cells. This is not an unrealistic assumption since metal nanoparticles can be conjugated to molecules that specifically bind to biomarkers expressed in epithelial precancers [2–6]. Several growth factor receptors and oncoproteins have received attention as possible targets in many diagnostic and prognostic studies. As an example, Sanfilippo et al. [24] have investigated the expression levels of epidermal growth factor receptor (EGFR) in gynecologic tissues. Their results show that while many normal uterine specimens have undetectable levels of EGFR, cancer specimens are generally characterized by significant receptor expression. Yet, other studies by Crow et al. [25] and Seekell et al. [26] suggest that different cell lines exhibit a high degree of variability in receptor expression levels. In some cases, normal cells can also be expected to express detectable levels of the target biomarker. It is hence informative to ease the assumption of selective targeting and to consider the possibility of simultaneous labeling in normal tissues. As evidenced in Fig. 9, high levels of biomarker expression and nanoparticle accumulation in normal tissues can deteriorate the diagnostic contrast. If precancerous tissues are characterized by a twenty-fold or a ten-fold over-expression of the target biomarker, however, nanoparticle-generated positive contrast is still significantly greater than the inherent contrast. Moreover, even just a two-fold over-expression can sometimes be sufficient to obtain an improved negative contrast. These observations highlight the importance of identifying an arsenal of cancer-specific receptors and oncoproteins to which conjugated nanoparticles can be delivered with high selectivity and efficacy.
As a final remark, the quantitative analysis described in this work does not address any potential influence of nanoparticle aggregation. As discussed previously in our antecedent study [19], nanoparticle concentrations employed in the simulations are so low that a rough estimation for 10-μm cells points to a surface coverage of less than 5%; interparticle effects can thus be ignored and the assumption of independent scattering ensues for labeling that predominantly occurs on the cell membrane. Crow et al. [25] further report that even after nanoparticles are taken up by cells via receptor internalization, they remain isolated in individual vesicles as long as their concentrations are kept at low levels. These arguments justify the validity of treating nanoparticles as independent scatterers. Application of nanoparticles at concentrations much higher than those used in this study will inevitably give rise to formation of aggregates. It is then highly likely that collective behavior or plasmon coupling of proximal particles will lead to significant changes in optical cross sections along with a red shift in peak resonance as reported in [2,4,5,27]. Under such conditions, interparticle effects need to be quantified and accounted for in MC simulations. At this point, it is important to reiterate that the spectral red shift observed in the results presented here is solely due to diffusive photon transport and is not akin to what would be caused by plasmon coupling. Receptor dimerization and particle pairing will be another factor to consider when labeling is carried out using nanoparticles with small diameters comparable to the receptor spacing [28]. We plan to assess the effects of aggregation and dimerization as part of our future modeling work.
5. Conclusions
In summary, MC modeling offers a powerful computational framework to predict the spectral optical response of nanoparticle-labeled tissues and to assess whether a particular labeling scheme serves the intended purpose of generating sufficient exogenous contrast that can better pinpoint disease progression. It is expected that the extensive set of simulation results presented here will prove useful as a comprehensive guide to devise strategies for nanoparticle-assisted precancer diagnosis. Even though nanoplatform-based optical contrast enhancement is considered in the context of tissue diagnostics, the modeling scheme described and the concepts discussed are applicable to optical characterization of systems that involve metal nanoparticles embedded in any diffusive medium.
Acknowledgments
The author would like to thank Ashkan Roozbeh and Ali Akbar Shakibaei for their help in performing a small subset of the simulations described in this work.
References and links
1. W. Cai and X. Chen, “Nanoplatforms for targeted molecular imaging in living subjects,” Small 3(11), 1840–1854 (2007). [CrossRef] [PubMed]
2. K. Sokolov, M. Follen, J. Aaron, I. Pavlova, A. Malpica, R. Lotan, and R. Richards-Kortum, “Real-time vital optical imaging of precancer using anti-epidermal growth factor receptor antibodies conjugated to gold nanoparticles,” Cancer Res. 63(9), 1999–2004 (2003). [PubMed]
3. C. Loo, L. Hirsch, M. H. Lee, E. Chang, J. West, N. Halas, and R. Drezek, “Gold nanoshell bioconjugates for molecular imaging in living cells,” Opt. Lett. 30(9), 1012–1014 (2005). [CrossRef] [PubMed]
4. J. Aaron, N. Nitin, K. Travis, S. Kumar, T. Collier, S. Y. Park, M. José-Yacamán, L. Coghlan, M. Follen, R. Richards-Kortum, and K. Sokolov, “Plasmon resonance coupling of metal nanoparticles for molecular imaging of carcinogenesis in vivo,” J. Biomed. Opt. 12(3), 034007 (2007). [CrossRef] [PubMed]
5. A. Wax and K. Sokolov, “Molecular imaging and darkfield microspectroscopy of live cells using gold plasmonic nanoparticles,” Laser Photon. Rev. 3(1-2), 146–158 (2009). [CrossRef]
6. A. Hellebust and R. Richards-Kortum, “Advances in molecular imaging: targeted optical contrast agents for cancer diagnostics,” Nanomedicine (Lond) 7(3), 429–445 (2012). [CrossRef] [PubMed]
7. U. Kreibig and M. Vollmer, Optical Properties of Metal Clusters (Springer, 1995).
8. J. Yguerabide and E. E. Yguerabide, “Light-scattering submicroscopic particles as highly fluorescent analogs and their use as tracer labels in clinical and biological applications,” Anal. Biochem. 262(2), 137–156 (1998). [CrossRef] [PubMed]
9. J. Yguerabide and E. E. Yguerabide, “Light-scattering submicroscopic particles as highly fluorescent analogs and their use as tracer labels in clinical and biological applications,” Anal. Biochem. 262(2), 157–176 (1998). [CrossRef] [PubMed]
10. K. L. Kelly, E. Coronado, L. L. Zhao, and G. C. Schatz, “The optical properties of metal nanoparticles: the influence of size, shape, and dielectric environment,” J. Phys. Chem. B 107(3), 668–677 (2003). [CrossRef]
11. P. K. Jain, K. S. Lee, I. H. El-Sayed, and M. A. El-Sayed, “Calculated absorption and scattering properties of gold nanoparticles of different size, shape, and composition: applications in biological imaging and biomedicine,” J. Phys. Chem. B 110(14), 7238–7248 (2006). [CrossRef] [PubMed]
12. Y. Hu, R. C. Fleming, and R. A. Drezek, “Optical properties of gold-silica-gold multilayer nanoshells,” Opt. Express 16(24), 19579–19591 (2008). [CrossRef] [PubMed]
13. A. Henry, J. M. Bingham, E. Ringe, L. D. Marks, G. C. Schatz, and R. P. Van Duyne, “Correlated structure and optical property studies of plasmonic nanoparticles,” J. Phys. Chem. C 115(19), 9291–9305 (2011). [CrossRef]
14. N. Nitin, D. J. Javier, D. M. Roblyer, and R. Richards-Kortum, “Widefield and high-resolution reflectance imaging of gold and silver nanospheres,” J. Biomed. Opt. 12(5), 051505 (2007). [CrossRef] [PubMed]
15. R. T. Zaman, P. Diagaradjane, J. C. Wang, J. Schwartz, N. Rajaram, K. L. Gill-Sharp, S. H. Cho, H. G. Rylander III, J. D. Payne, S. Krishnan, and J. W. Tunnell, “In vivo detection of gold nanoshells in tumors using diffuse optical spectroscopy,” IEEE J. Sel. Top. Quantum Electron. 13(6), 1715–1720 (2007). [CrossRef]
16. A. W. H. Lin, N. A. Lewinski, J. L. West, N. J. Halas, and R. A. Drezek, “Optically tunable nanoparticle contrast agents for early cancer detection: model-based analysis of gold nanoshells,” J. Biomed. Opt. 10(6), 064035 (2005). [CrossRef] [PubMed]
17. E. V. Zagaynova, M. V. Shirmanova, M. Y. Kirillin, B. N. Khlebtsov, A. G. Orlova, I. V. Balalaeva, M. A. Sirotkina, M. L. Bugrova, P. D. Agrba, and V. A. Kamensky, “Contrasting properties of gold nanoparticles for optical coherence tomography: phantom, in vivo studies and Monte Carlo simulation,” Phys. Med. Biol. 53(18), 4995–5009 (2008). [CrossRef] [PubMed]
18. M. Kirillin, M. Shirmanova, M. Sirotkina, M. Bugrova, B. Khlebtsov, and E. Zagaynova, “Contrasting properties of gold nanoshells and titanium dioxide nanoparticles for optical coherence tomography imaging of skin: Monte Carlo simulations and in vivo study,” J. Biomed. Opt. 14(2), 021017 (2009). [CrossRef] [PubMed]
19. C. Cihan and D. Arifler, “Influence of phase function on modeled optical response of nanoparticle-labeled epithelial tissues,” J. Biomed. Opt. 16(8), 085002 (2011). [CrossRef] [PubMed]
20. J. Q. Brown, K. Vishwanath, G. M. Palmer, and N. Ramanujam, “Advances in quantitative UV-visible spectroscopy for clinical and pre-clinical application in cancer,” Curr. Opin. Biotechnol. 20(1), 119–131 (2009). [CrossRef] [PubMed]
21. D. Arifler, C. MacAulay, M. Follen, and R. Richards-Kortum, “Spatially resolved reflectance spectroscopy for diagnosis of cervical precancer: Monte Carlo modeling and comparison to clinical measurements,” J. Biomed. Opt. 11(6), 064027 (2006). [CrossRef] [PubMed]
22. J. M. Schmitt and G. Kumar, “Turbulent nature of refractive-index variations in biological tissue,” Opt. Lett. 21(16), 1310–1312 (1996). [CrossRef] [PubMed]
23. C. Kortun, Y. R. Hijazi, and D. Arifler, “Combined Monte Carlo and finite-difference time-domain modeling for biophotonic analysis: implications on reflectance-based diagnosis of epithelial precancer,” J. Biomed. Opt. 13(3), 034014 (2008). [CrossRef] [PubMed]
24. J. S. Sanfilippo, S. Miseljic, A. R. Yang, D. L. Doering, R. M. Shaheen, and J. L. Wittliff, “Quantitative analyses of epidermal growth factor receptors, HER-2/neu oncoprotein and cathepsin D in nonmalignant and malignant uteri,” Cancer 77(4), 710–716 (1996). [CrossRef] [PubMed]
25. M. J. Crow, G. Grant, J. M. Provenzale, and A. Wax, “Molecular imaging and quantitative measurement of epidermal growth factor receptor expression in live cancer cells using immunolabeled gold nanoparticles,” AJR Am. J. Roentgenol. 192(4), 1021–1028 (2009). [CrossRef] [PubMed]
26. K. Seekell, M. J. Crow, S. Marinakos, J. Ostrander, A. Chilkoti, and A. Wax, “Hyperspectral molecular imaging of multiple receptors using immunolabeled plasmonic nanoparticles,” J. Biomed. Opt. 16(11), 116003 (2011). [CrossRef] [PubMed]
27. J. Aaron, K. Travis, N. Harrison, and K. Sokolov, “Dynamic imaging of molecular assemblies in live cells based on nanoparticle plasmon resonance coupling,” Nano Lett. 9(10), 3612–3618 (2009). [CrossRef] [PubMed]
28. M. J. Crow, K. Seekell, J. H. Ostrander, and A. Wax, “Monitoring of receptor dimerization using plasmonic coupling of gold nanoparticles,” ACS Nano 5(11), 8532–8540 (2011). [CrossRef] [PubMed]
