| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Cancer Prevention |
Authors' Affiliations: 1 Department of Medicine, Evanston Northwestern Healthcare and 2 Biomedical Engineering Department, Northwestern University, Evanston, Illinois
Requests for reprints: Hemant K. Roy, Feinberg School of Medicine at Northwestern University, Evanston Northwestern Healthcare Research Institute, 1001 University Place, Evanston, IL 60201. Phone: 847-570-2239; Fax: 847-733-5041; E-mail: h-roy{at}northwestern.edu.
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Methods: Enhanced backscattering analysis was done on intestinal mucosa at preneoplastic time points from two experimental models of colorectal cancer: the azoxymethane-treated rat and the multiple intestinal neoplasia (MIN) mouse. Data were analyzed using two previously validated spectral markers: spectral slope and principle components. We then did a pilot study on mucosal biopsies from 63 subjects undergoing screening colonoscopy.
Results: In the azoxymethane-treated rat, when compared with saline-treated controls, significant changes in the enhanced backscattering markers were observed as early as 2 weeks after azoxymethane treatment (before the development of aberrant crypt foci and adenomas). Enhanced backscattering markers continued to progress over time in a manner consonant with future neoplasia. These data were replicated in the preneoplastic MIN mouse mucosa. In humans, spectral slopes in the endoscopically normal cecum, midtransverse colon, and rectum were markedly reduced in patients harboring adenomas when compared with those who were neoplasia free.
Conclusions: We show, for the first time, that enhanced backscattering analysis of an aliquot of uninvolved mucosa has the potential for predicting neoplastic risk throughout the colon in both experimental colorectal cancer models and humans.
Therefore, it seems that colonoscopy will remain the "gold standard" for the foreseeable future. However, resource constraints and potential complications make it impractical to perform colonoscopy on the entire at-risk population (generally considered age, 
50). Thus, identifying patients who are most likely to benefit from colonoscopy is of paramount importance. Many risk stratification techniques exploit the "field effect," the concept that assessment of biomarkers in one area of the colon should be able to determine the likelihood of current/future neoplastic lesions throughout the colon (8, 9). A commonly used clinical example is the identification of the distal adenoma on flexible sigmoidoscopy to predict the occurrence of neoplasia in the proximal colon (10). Other attempts include correlation of rectal aberrant crypt foci (ACF) using chromoendoscopy with colonic adenomas and carcinomas (11). Unfortunately, the performance characteristics of the existing markers remain suboptimal (e.g., the sensitivity and positive predictive value for the ability of flexible sigmoidoscopy to detect advanced proximal lesions are 40% and 6%, respectively; refs. 12, 13).
Thus, it is clear that currently available morphologic markers for the field effect are inadequate for risk stratification. Several lines of evidence suggest that the field effect has the potential of being exquisitely sensitive at identifying patients with colonic neoplasia. Numerous studies have reported that in the histologically normal mucosa of subjects harboring colonic neoplasia, there are profound genetic and epigenetic alterations in the field effect (14, 15). However, detecting these molecular events with a methodology that would be feasible in clinical practice has been challenging.
Biomedical optics represents a powerful means of probing epithelial cellular architecture and has clear applications for colorectal cancer screening given ready access of the distal colonic mucosa via either stand-alone fiber optical probes or endoscope-coupled devises. Our group, in collaboration with others, has pioneered the use of light scattering spectroscopy to detect dysplastic cells in vivo through analysis of nuclear size and chromatin content (16, 17). Furthermore, we have shown that light scattering spectroscopy can identify adenomatous change in the colon (18). However, light scattering spectroscopy is unable to detect the more subtle microarchitectural consequences of the molecular changes in the colonic field. Enhanced backscattering is an optical phenomenon that can provide quantitative information about the nanoscale composition of the epithelium. Enhanced backscattering, otherwise known as coherent backscattering, is a spectacular manifestation of the self-interference of light propagating in a random medium, such as biological tissue, which leads to an enhanced scattering peak in the backward direction (i.e., direction opposite to that of the light beam incident on tissue surface). Although the potential of enhanced backscattering is well established, application of this approach to tissue diagnosis has been stymied by numerous technical problems. These include the extremely narrow widths of the enhanced backscattering peaks in tissue (full width at half maximum 
0.001 degree) and excessive speckle. Indeed, to date, use of enhanced backscattering to interrogate tissue microarchitecture has been technically impossible.
We have recently overcome these hurdles by recording enhanced backscattering under low coherence illumination, which gives rise to >100 times broadened and speckle-free enhanced backscattering peaks (19–21). Moreover, we have shown that this low-coherence enhanced backscattering (LEBS) spectroscopy allows the analysis of tissue nanoarchitecture and microarchitecture. Additionally, LEBS enables accurate depth resolution from a few microns to several hundreds of microns below tissue surface through the analysis of the angular profile of an LEBS peak (which contains information about a wide range of tissue depths simultaneously). This novel feature of LEBS has not been achievable with conventional optical spectroscopy. Depth selectivity has important advantages in analyzing the colonic field effect given the biological heterogeneity of colonocytes in the crypt with the earliest changes in neoplasia occurring towards the base of the crypt (i.e., stem cell compartment; ref. 22). Moreover, depth selectivity of LEBS is required to address one of the most notorious problems in tissue spectroscopy: light absorption by submucosal hemoglobin can obscure endogenous spectral signatures of epithelial cells (23). LEBS spectroscopy solves this problem by enabling selective probing of the most superficial layer of tissue, which maximizes the probability of targeting the epithelium and minimizes the interference from photons penetrating into the submucosa. In the present study, we showed the use of enhanced backscattering to predict neoplastic risk at the earliest points of colon carcinogenesis using the azoxymethane-treated rat model. Furthermore, we duplicated these data in the multiple intestinal neoplasia (MIN) mouse. Finally, we showed the applicability of these results to humans with a pilot clinical trial.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
0.03-0.10 degree), polarized, and delivered onto a sample at 15-degree angle of incidence to prevent the collection of the specular reflection. The instrument allowed varying spatial coherence length Lcs of the incident light from 100 to 200 µm by means of aperture A1 positioned in the Fourier plane of lens system L1/L2 in the light delivery arm of the instrument. The spatial coherence length was confirmed by the double-slit interference (24). The light backscattered by a sample was collected using a Fourier lens L3, a polarizer P2 (oriented along the polarization of the incident light), and an imaging spectrograph SP (Acton Research, Acton, MA), which was positioned in the focal plane of the lens and coupled with a CCD camera (VersArrayxp, Roper Scientific, Trenton, NJ). The lens projected the angular distribution of the backscattered light onto the slit of the spectrograph. Thus, all scattered rays with an identical scattering angle 
were focused into a point on the entrance slit of the spectrometer. Then, the imaging spectrograph dispersed this light according to its wavelength in the direction perpendicular to the slit. Thus, the CCD recorded a matrix of light-scattering intensities as a function of wavelength 
and backscattering angle . In each CCD pixel, collected light was integrated within a certain narrow wavelength band 
around . For each , the width of the band and, thus, temporal coherence length Lct was determined by the width of the spectrograph slit. In the experiments reported below, Lct was fixed at 30 µm ( = 9 nm).
|
1 mm2 were recorded. For each animal, LEBS data were recorded from at least 20 different tissue sites spaced uniformly across the surface of the colon (in case of azoxymethane/saline-treated rats) or small bowel (in case of MIN and control mice).
Human studies. The human studies were approved and done in accordance with the institutional review board at Evanston Northwestern Healthcare. After informed consent, 63 subjects scheduled for screening/surveillance colonoscopy at Evanston Hospital had six biopsies of endoscopically normal mucosa (two each from cecum, midtransverse colon, and rectum) at least 5 cm away from any neoplastic lesion. The exclusion criteria included colitis (either history of or on suspected on present colonoscopy), coagulopathy, or inability to give informed consent. LEBS analysis was done within 1 hour of tissue acquisition by observers blinded to clinical/endoscopic data. For each subject, LEBS data were recorded from 16 tissue sites per each segment (i.e., cecum, midtransverse colon, and rectum).
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
,) collected from rat colonic tissue (Lcs = 140 µm). An enhanced backscattering peak can be clearly identified. Furthermore, this figure illustrates some of the major advantages of LEBS: LEBS peak is >100 times wider than conventional enhanced backscattering, which makes LEBS easy to observe; it is not obstructed by speckle and is recorded for a wide range of wavelengths (400-700 nm) simultaneously.
|
) were obtained from IEBS(,) by integrating over backscattering angle . As shown in Fig. 2B, the LEBS spectra obtained from the preneoplastic and control colonic tissues are distinctly different. As discussed below, the quantitative analysis of such LEBS spectra revealed a number of highly significant spectral markers that were diagnostic for the earliest field changes in colon carcinogenesis.
Optimal depth for identifying field effect alterations in early colon carcinogenesis. As we have previously detailed (16–18), in LEBS, the depth resolution can be achieved by analyzing the angular profile of an enhanced backscattering peak that contains information about a wide range of tissue depths simultaneously (30 to 80 µm for the spatial coherence length used in our experiments). In brief, the periphery of the enhanced backscattering peak (i.e., large backscattering angles: angular region B in Fig. 2A) is primarily contributed by light that does not penetrate deep into tissue and thus probes short tissue depths. On the other hand, the tip of the enhanced backscattering peak ( 0 degree: region A in Fig. 2A) mainly depends on longer light paths, and the spectra evaluated for these small backscattering angles can be used to probe deeper tissue. We have established the dependence of the depth of penetration on the angle for which LEBS spectra are evaluated (18).
First, we identified the optimal depth of penetration for which LEBS markers are the most diagnostic. Because the scattering angles determine the depth of penetration we evaluated a series of angles that corresponded to 30, 50, and 75 µm depths (angles 0.4 degree, 0.2 degree, and 0 degree, respectively). We selected one pre-ACF time point in the azoxymethane-treated rat model (2 weeks after azoxymethane administration). LEBS signals IEBS(,) were recorded from at least 20 tissue sites per animal equally distributed throughout colonic surface. LEBS spectra IEBS() were calculated from these signals as previously discussed above for each tissue depth (19–21).
Spectral behavior of IEBS() depends on the size distribution of light-scattering structures(8, 25). Generally, IEBS() is a declining function of wavelength, and its steepness is related to the relative portion of structures of different sizes. Larger structures that approached micron and supramicron sizes (i.e., cellular organelles, etc.) tend to reduce the steepness of IEBS(), whereas smaller scatterers (sizes as small as 20 nm) tend to increase the steepness of IEBS() over wavelength. To characterize IEBS() with a single variable, we obtained linear fits to IEBS() using linear regression from 530 to 640 nm. The absolute value of the linear coefficient of the fit is referred to as the "LEBS spectral slope" and quantifies the dependence of an LEBS spectrum on wavelength. As shown in Fig. 3, signals recorded from 75-µm depth gave the best distinction between the control and azoxymethane-treated rats. We did not look at deeper depths given that the signals recorded from deeper tissue are affected by hemoglobin absorption. In the following animal studies, we analyzed LEBS spectra obtained from this critical depth.
|
|
99% of the data variance. We searched diagnostic principle components as a linear combination of PC1 and PC2 and identified principle component marker = PC1 + 5PC2 to be the significant. Therefore, LEBS principle component marker can be used as a convenient means to characterize the light scattering data. Once again, we found that LEBS principle component marker was significantly decreased at the 2-week time point (P < 0.02) and continued to progressively decrease over the course of the experiment (P < 0.000001; Fig. 4B). LEBS markers in the MIN mouse. Although the azoxymethane-treated rat model is robust and well validated, to ensure that the changes in LEBS signatures are not model specific, we confirmed the findings in an alternate model of intestinal carcinogenesis, the MIN mouse. This is a genetic model with a germ line mutation in the APC, the initiating mutation in most sporadic colon carcinogenesis. These animals spontaneous develop intestinal adenomas starting at ages 9 to 10 weeks (25). We compared the LEBS signatures obtained from the MIN mice with age-matched negative control C57Bl mice. The control C57Bl mice differ from the MIN mice only in that they harbor a wild-type APC gene. The study involved n = 18 animals (nine MIN and nine control mice). For each animal, LEBS data were recorded for at least 20 different tissue sites spaced uniformly across the surface of the small bowel. We found that the LEBS markers that were significant for early colon carcinogenesis in the azoxymethane-treated rats were also diagnostic for the early preadenoma stage of intestinal neoplasia in the 6-week-old MIN mice. Specifically, we evaluated the intestinal mucosa at week 6 when the mucosa is histologically normal. As shown in Fig. 5, at this preneoplastic time point, there were dramatic alterations in both the LEBS spectral slope (P < 0.01) and LEBS principle component marker (P < 0.01). This highlights the robustness of our findings.
|
As discussed above, the LEBS data were obtained from endoscopically normal rectum, midtransverse colon, and the cecum at least 5 cm away from any neoplastic lesion. Figure 6 shows that there was a significant decrease in the spectral slope obtained from each of these three segments in patients who harbored adenomas somewhere in their colon when compared with those who were neoplasia free (P < 0.01). The magnitude of decrease of spectral slope seemed to be greater if the lesion was located in the same region as the LEBS analysis; however, significant differences were noted from LEBS measurements taken at distant sites (e.g., cecal LEBS spectral slope in patients with rectal adenomas). It needs to be emphasized that the post hoc analysis (effect of distance from neoplasm on spectral slope decrease) with relatively small numbers must be viewed with caution. We point out that the decrease of the spectral slope was consistent with a similar alteration of this marker in the azoxymethane-treated rat and MIN mouse models. These results support the hypothesis that alteration of light scattering; thus, nanoarchitectural/microarchitectural signatures in the uninvolved mucosa in humans (i.e., the field effect) is detectable by LEBS. Therefore, LEBS readings in easily accessible colonoscopically normal mucosa (e.g., in the rectum) have the potential to serve as accurate markers of the risk of neoplasia elsewhere in the colon.
|
1 cm, villous features or high-grade dysplasia). Our findings suggest that the magnitude of alterations in LEBS markers (e.g., LEBS spectral slope) was greater in patients with more advanced lesions compared with simple adenomas (data not shown). However, we need to be circumspect regarding conclusions given the pilot nature of the clinical data.
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
As previously discussed, the exploitation of the field effect is a common strategy in colorectal cancer screening (e.g., assessment of distal adenomas or ACF). To improve sensitivity, others have proposed looking at cellular (apoptosis and proliferation; refs. 26, 27) and biochemical variables (e.g., protein kinase C; ref. 28); however, although somewhat of an improvement, the performance characteristics are still suboptimal for clinical practice. In the present study, our assessment of mucosal nanoarchitectural and microarchitectural markers by means of LEBS far outdid the classic morphologic or biochemical markers. Our approach was not to detect morphologic lesions (polyps) but rather to assess the risk of neoplasia from assessing the visually normal colonic mucosa. There is ample evidence to support the molecular underpinnings of the microarchitectural changes noted in the histologically normal "field." For instance, Chen et al. recently reported that a panel of proto-oncogenes, including cyclooxygenase 2 and osteopontin, were markedly overexpressed in histologically normal mucosa of patients harboring colorectal cancer (14). They also noted this in the preneoplastic MIN mouse and, importantly, the magnitude of proto-oncogenes overexpression was in-between control intestinal epithelium (C57BL/6 mice wild type at APC) and adenomatous tissue, arguing for the relevance of these changes to tumorigenesis (14). Furthermore, work by Cui et al. have noted that another epigenetic event (loss of insulin growth factor II imprinting) was increased in the uninvolved mucosa of patients with who harbored adenomas (15).
Our approach has been to evaluate nanoarchitectural/microarchitectural consequences of these genetic/epigenetic changes to risk stratify for colorectal cancer. Light scattering is determined by the fundamental properties of the scattering particles and thus represents a powerful, practical means of nanoarchitectural/microarchitectural assessment. Although there are a number of technological means for analyzing backscattered information, few have been able to quantitatively assess the nanoscale architecture. Enhanced backscattering is a unique multiple light-scattering phenomenon that encodes data regarding the characteristics of particles at the nanometer to micron scales. In contrast to the other light scattering–based techniques, enhanced backscattering originates from constructive interference of waves traveling in time-reversed paths (i.e., the paths whose first and last points coincide with the last and first points of its time-reversed counterpart, respectively). If light scattered by a random medium such as tissue is recorded as a function of scattering angle, enhanced backscattering manifests itself as a sharp peak centered around the backscattering direction. The scattered light intensity at the peak of an enhanced backscattering cone can be twice the incoherent background intensity outside of the enhanced backscattering cone. The intensity of enhanced backscattering quickly vanishes for angles away from the backward direction (Fig. 1). Our group has overcome these technological hurdles (the speckle and the narrow peak) by the discovery that under low spatially coherent illumination (i.e., spatial coherence length Lcs is much shorter than the transport mean free path for which light traveling in tissue is randomized, which is 1 mm in tissue; refs. 19–21). Thus, our low coherence innovation makes enhanced backscattering analysis of tissue feasible.
We used the alterations of LEBS spectra as a gauge of the microarchitectural consequences of the well-established genetic/epigenetic changes in the field effect. To quantitate these structural alterations, we used several previously validated optical variables. We have previously developed spectral slope as a marker of perturbations at the macromolecular complex to small organelle level (8, 25). As seen in Fig. 4, our studies in the azoxymethane-treated rat model show that LEBS spectral slope was decreased early during carcinogenesis, and this change progressed over time paralleling progression of carcinogenesis. Although the origin of microarchitectural alterations has yet to be identified, we speculate that the decrease in spectral slope may reflect aggregation of macromolecular complexes and other nanoscale intracellular structures, thus decreasing the number of smaller particles. We have to point out that the data acquired through LEBS are extraordinarily rich, and the algorithms that we have currently developed extract a minute portion of the total information. In cancer biology and clinical medicine, one commonly used tool for discerning structure in a complex data set is through principal component analysis. Thus, we also did principal component analysis of LEBS spectra. As shown above, this approach yielded highly significantly altered diagnostic markers in both the azoxymethane-treated rat and the MIN mouse. These variables were highly diagnostic at the earliest pre-ACF and adenoma phases and progressed over time in a manner consonant with neoplastic transformation.
We believe that one of the major reasons for the outstanding sensitivity of LEBS for early colon carcinogenesis is its depth selectivity. The ability to rigorously define the depth of tissue interrogated dramatically improves signal detection by excluding photons that have traveled deeper than the epithelium and also removing the distortion from submucosal hemoglobin. Furthermore, given that the biochemical and functional heterogeneity of colonocytes that is largely determined by the position in the crypt, the depth selectivity allows targeting of the region of the colonic crypt most likely to undergo carcinogenesis, thus enhancing the "signal-to-noise" ratio (29). Our data indicate that the most profound changes during early colon carcinogenesis occurred at 80 µm, which approximately corresponds to the base of the crypt (Fig. 3). Several lines of evidence suggest that the base of the crypt is the location for initiation of colon carcinogenesis (30). For instance, there is compelling evidence that adenomatous transformation (as evident by crypt branching) also starts at the base of the crypt (31). Furthermore, this is the region where the colonic stem cells reside. Emerging evidence suggest that stem cells can accumulate mutations over several decades and thus are initiating cells in colon carcinogenesis (22). In this regard, conditions with high risk of colorectal cancer, such as familial adenomatous polyposis, expansion of the stem cell population has been noted (32).
There are limitations/potential confounders to this work that need to be acknowledged. First, our data processing algorithms do not allow exploitation of the full potential of the data encoded by LEBS. Thus, it is conceivable that there may be LEBS markers that have a greater diagnostic accuracy. Another potential concern is the possibility that we are detecting early ACF or microadenomas. Although we can not completely exclude this possibility, the fact that >90% of tissue sites assessed by LEBS showed abnormalities at pre-ACF and preadenoma time points (e.g., 2 weeks after carcinogen administration in the azoxymethane-treated rat model and 6-week-old MIN mice) strongly argues against the hypothesis that the altered LEBS markers reflect detection of focal dysplasia. In the azoxymethane-treated rat, approximately one half of animals will develop tumors; thus, future studies will be necessary to show that enhanced backscattering markers not only identifies azoxymethane exposure but actually predicts the occurrence of biologically significant neoplasia. In this regard, our human data support the feasibility of developing these prediction rules. One could also argue that the 80-µm depth may not exactly correspond to the location of colonic stem cells. Clearly, our hypothesis that the optimal nature 80-µm depth is related to probing the stem cell compartment is conjecture because stem cells are notoriously difficult to identify. Given the need to do LEBS analysis on fresh, unfixed tissue, simultaneous measurement of LEBS signatures, putative stem cell marker (e.g., Musashi-1) is impossible (30). On the other hand, even if our biologically plausible explanation for the optimal nature of the 80-µm depth is incorrect, this does not alter the major thrust and clinical implications of our LEBS findings. Finally, our human data, whereas compelling, need to be considered pilot data that simply confirm the relevance of LEBS marker abnormalities to human colon carcinogenesis. Future studies will be needed to delineate the power of this novel approach to colorectal cancer screening.
In summary, we showed, for the first time, that a novel optics technology, LEBS spectroscopy was able to accurately risk stratify for colon carcinogenesis in two experimental models of colon carcinogenesis. These light scattering and thus nanoarchitectural/microarchitectural changes in colonic epithelium preceded conventional biomarkers of colon carcinogenesis and progressed over time, mirroring the events in neoplastic transformation. Furthermore, preliminary human data support the clinical relevance. The long-term goal of this work will be to develop a means of risk-stratifying patients for colon carcinogenesis by means of LEBS assessment of easily accessible endoscopically normal mucosa (such as in the rectum) to identify patients harboring neoplastic lesions elsewhere in the colon and optimize decisions regarding an individual colorectal cancer screening modalities and intervals.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Presented in part in abstract form at the 106th Digestive Disease Week Meetings, May 15-19, 2005 in Chicago, Illinois.
Received 7/26/05; revised 10/12/05; accepted 10/24/05.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. J. Chalut, S. Chen, J. D. Finan, M. G. Giacomelli, F. Guilak, K. W. Leong, and A. Wax Label-Free, High-Throughput Measurements of Dynamic Changes in Cell Nuclei Using Angle-Resolved Low Coherence Interferometry Biophys. J., June 15, 2008; 94(12): 4948 - 4956. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. K. Roy, V. Backman, and M. J. Goldberg Colon cancer screening: the good, the bad, and the ugly. Arch Intern Med, November 13, 2006; 166(20): 2177 - 2179. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
