Published ahead of print on October 5, 2006, doi:10.1164/rccm.200605-684OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200605-684OC
In Vivo Imaging of the Bronchial Wall Microstructure Using Fibered Confocal Fluorescence MicroscopyClinique Pneumologique, and Pathology Department, Rouen University Hospital, Rouen; Mauna Kea Technologies; CNRS UMR7033, Université Pierre et Marie Curie, Paris; and INRIA Sophia Antipolis, Sophia Antipolis Cedex, France Correspondence and requests for reprints should be addressed to Luc Thiberville, M.D., Clinique Pneumologique, Hôpital Charles Nicolle–CHU de Rouen, 1 rue de Germont, 76031 Rouen Cedex, France. E-mail: luc.thiberville{at}univ-rouen.fr
Rationale: Fibered confocal fluorescence microscopy (FCFM) is a new technique that produces microscopic imaging of a living tissue through a 1-mm fiberoptic probe that can be introduced into the working channel of the bronchoscope. Objectives: To analyze the microscopic autofluorescence structure of normal and pathologic bronchial mucosae using FCFM during bronchoscopy. Methods: Bronchial FCFM and spectral analyses were performed at 488-nm excitation wavelength on two bronchial specimens ex vivo and in 29 individuals at high risk for lung cancer in vivo. Biopsies of in vivo FCFM-imaged areas were performed using autofluorescence bronchoscopy. Results: Ex vivo and in vivo microscopic and spectral analyses showed that the FCFM signal mainly originates from the elastin component of the basement membrane zone. Five distinct reproducible microscopic patterns were recognized in the normal areas from the trachea down to the more distal respiratory bronchi. In areas of the proximal airways not previously biopsied, one of these patterns was found in 30 of 30 normal epithelia, whereas alterations of the autofluorescence microstructure were observed in 19 of 22 metaplastic or dysplastic samples, five of five carcinomas in situ, and two of two invasive lesions. Disorganization of the fibered network could be found on 9 of 27 preinvasive lesions, compatible with early disruptions of the basement membrane zone. FCFM alterations were also observed in a tracheobronchomegaly syndrome and in a sarcoidosis case. Conclusions: Endoscopic FCFM represents a minimally invasive method to study specific basement membrane alterations associated with premalignant bronchial lesions in vivo. The technique may also be useful to study the bronchial wall remodeling in nonmalignant chronic bronchial diseases.
Key Words: basement membrane • bronchoscopy • fluorescence • microscopy, confocal • precancerous conditions
The bronchial mucosa has many functions in physiology, including the role of physical barrier to the external environment and the maintenance of the normal bronchial wall tissue architecture. As the first barrier exposed to carcinogens, such as tobacco smoke components, the bronchial epithelium is prone to early pathologic alterations associated with or preceding proximal lung cancer (1). These precancerous conditions evolve from basal cell hyperplasia to different grades of epithelial dysplasia and carcinoma in situ (CIS) (1), with evidence of cumulative molecular alterations from one stage to the other (2), and a variable spontaneous evolution over time (3). In vitro studies have shown that these preinvasive changes are associated with early modifications of the underlying matrix, both at the biochemical (4) and histopathologic (5) levels. The bronchial epithelium also appears to be an active participant in tissue remodeling of the reticular basement membrane, especially in different bronchial inflammatory conditions such as chronic obstructive pulmonary disease and asthma (6). Until now, assessment of these early changes relied primarily on gross inspection during an endoscopic procedure and pathologic examination of biopsy samples derived from the macroscopy. Recent endoscopic techniques have been developed to more effectively detect and localize critical, early pathologic changes occurring in the bronchial epithelial and subepithelial regions in vivo. This includes autofluorescence bronchoscopy (7), optical fluorescence and reflectance spectroscopy (8, 9), high-magnification bronchovideoscopy (10), high-frequency endobronchial ultrasound (11), and, more recently, optical coherence tomography (12). Among these techniques, autofluorescence bronchoscopy has been extensively evaluated during the past decade (13). Fluorescence bronchoscopy is based on the observation that premalignant and malignant bronchial mucosae fluoresce less than normal tissue, and thereby allow detection of lesions (e.g., CIS) that may have a normal appearance during conventional white-light bronchoscopy (14). Two recent randomized studies versus conventional bronchoscopy have shown that the technique is able to improve the localization and the diagnosis of high-grade precancerous lesions from two to five times in high-risk individuals (15, 16). However, the technique is hampered by the low specificity of the fluorescence defect, which ranges from 25 to 50% (13). Coupled with autofluorescence bronchoscopy, the use of a method that would allow real-time noninvasive histologic imaging—a principle that is also referred to as "optical biopsy"—may help to ensure higher yield biopsy samples, increase the specificity of the endoscopic technique, and potentially avoid unnecessary biopsy sampling or repeated procedures. Fibered confocal microscopy is a new technique that can be used to image the microscopic structure of a living tissue (17). Fibered confocal microscopy is based on the principle of confocal microscopy, which provides a clear, in-focus image of a thin section within a biological sample, where the microscope's objective is replaced by a flexible fiberoptic miniprobe. In its fluorescence mode (fibered confocal fluorescence microscopy [FCFM]), the technique makes it possible to obtain high-quality images from endogenous or exogenous tissue fluorophores, through a fiberoptic probe of 1-mm diameter or less that can be introduced into the working channel of a flexible bronchoscope. We hypothesized that FCFM could be used to analyze the microscopic autofluorescence structure of normal and pathologic bronchial mucosae in vivo during bronchoscopy. The objectives of the study were as follows: (1) to describe the in vivo autofluorescence microscopic structure of the normal bronchial mucosae from the proximal airways down to the small peripheral bronchi and (2) to study the alterations of the bronchial microstructure in different pathologic conditions, including bronchial preinvasive neoplasia. In the present work, we performed FCFM imaging and biopsies guided by white-light and autofluorescence bronchoscopy. We show that FCFM makes it possible to produce clear microscopic images of the subepithelial lamina reticularis of the bronchial and bronchiolar wall, and that the main endogenous fluorescent signal originates from the elastin component of the bronchial wall. We show that modifications of the lamina reticularis fibered network can be imaged in preinvasive lesions, but also in nonmalignant diseases such as sarcoidosis and Mounier-Kühn syndrome. Some of the results of this study have been previously reported in the form of an abstract at the 16th annual European Respiratory Congress (18).
Ex Vivo Experiment To investigate the origin of the autofluorescence at the microscopic level, an ex vivo experiment was performed on two proximal, fresh, human lobar bronchus specimens obtained from a lobectomy sample. Both resections were performed for peripheral lung cancer. In both cases, the bronchus was longitudinally opened and flattened for direct FCFM imaging. Acriflavine 0.01% (Sigma Chemical Co., St. Louis, MO) was applied to the bronchus and immediately rinsed with saline before fluorescence imaging. The epithelial layer was then mechanically removed using a scalpel, and the unepithelialized bronchial wall was imaged again using FCFM. A sample of the bronchus was sent for pathologic analysis to control the quality of the epithelium removal.
Patients
Bronchoscopy, FCFM Imaging, and Biopsies
F400/S Dual Fibered Confocal Imaging and Spectroscopy
Image Reconstruction and Analysis Real-time image reconstruction was performed during the procedure using the Cellvizio clinical software. Video mosaicing techniques (21) were also used, for the purpose of this article, to provide a more complete image reconstruction of the bronchial area as well as a motion image representation of the FCFM endoscopy (see the online supplement).
FCFM of the Normal Bronchial Wall In Vivo Approximately five different bronchial areas were imaged for each patient. We found that the intensity of the microautofluorescence signal varied from one patient to the other, in direct relation to the Onco-Life intensity image (data not shown). However, real-time imaging of the bronchial wall microautofluorescence was possible in every patient on the normal-appearing part of the tracheobronchial tree. The technique made it possible to record small dynamic sequences of the bronchial microstructure during the time of the medical procedure, corresponding to the motion of the tip of the probe in contact with the bronchial mucosa. An example of such a sequence is shown in Figure 2, where the whole sequence has been reconstructed using video mosaicing techniques (19). The real-time sequence is provided in the online supplement.
Figure 3 displays the autofluorescence microstructure obtained during the FCFM endoscopy corresponding to normal bronchial areas. The bronchial microstructure appears as a fibered network that can be found on the entire surface of the bronchial tree down to the more distal level. Round or ovoid shaped openings, 1- to 400-µm large, were found specifically in the trachea and the cartilaginous bronchi up to the lobar divisions, presumably corresponding to bronchial gland origins (Figures 2 and 3B).
Five patterns could be clearly identified depending on the level imaged in the bronchial tree, as follows:
Ex Vivo Experiment Figure 4 displays the FCFM image of a freshly excised, normal human lobar bronchus after acriflavin nuclear staining, before and after mechanical removal of the epithelial layer. After acriflavin application, the fibrillar image of the bronchial microautofluorescence was replaced by a dense field of regular nuclei (8 µm in diameter) corresponding to the epithelial layer (Figure 4A). FCFM imaging of the bronchial surface after complete mechanical removal of the epithelial layer revealed the usual fibrillar autofluorescence of the bronchial wall (Figure 4B). The histologic examination of the bronchus confirmed the complete and elective removal of the epithelium up to the basement membrane (Figure 4C), showing that the normal bronchial autofluorescence image obtained at 488-nm excitation originates from the more superficial layers of the subepithelial bronchial wall. Identical results were obtained in both ex vivo bronchial samples analyzed.
In Vivo Spectral Analysis of FCFM Images The typical autofluorescence spectrum obtained in vivo from the healthy proximal bronchial mucosa is shown in Figure 5, in comparison with spectra of human elastin powder and collagen I gel. Under 488-nm excitation, all normal bronchus spectra presented an emission band with a peak located around 530 nm, decreasing monotonically to 650 nm (Figure 5A). Although variations in intensity were observed, depending on the corresponding autofluorescence image brightness, spectral shapes were found to be similar from one bronchial site to another in the same patient and from one patient to the other.
The bronchial autofluorescence spectrum appeared to be very similar to that obtained using elastin powder extracted from human lung. Conversely, collagen 1 gel spectral distribution differs significantly from that of elastin and bronchial mucosa (Figure 5B). Moreover, emission intensities of normal tissue and elastin under 488-nm excitation appear to be similar, but one order of magnitude higher than that of collagen 1 (Figure 5A). These results clearly indicate that 488-nm autofluorescence in FCFM images mainly results from the bronchial subepithelial fluorescence emission of elastin fibers.
FCFM of Preinvasive Lesions Twenty-nine patients presented with at least one non–previously biopsied, normal bronchial site that could be used as an internal control for FCFM image analysis (details on the FCFM image analysis method can be found on the online supplement). A total of 103 bronchial biopsy samples and corresponding FCFM images were taken during the study on these 29 patients. From these bronchial sites, 32 had been already sampled during previous endoscopies. Image analyses showed that the FCFM image was significantly altered in previously biopsied sites as compared with non–previously biopsied areas (Table 1). Therefore, FCFM images and corresponding histopathology were analyzed from the 71 non–previously biopsied samples. From these biopsies, 12 had lost the epithelial layer during processing and 59 were informative, including the following: 30 normal epithelia; 2 with basal cell hyperplasia; 13 with regular metaplasia; 5 with mild dysplasia, 1 with moderate dysplasia, and 1 with severe dysplasia; 5 CIS; and 2 invasive lesions. Table 1 displays the FCFM appearance of the normal and pathologic lesions at the microscopic level. All histologically normal epithelial samples exhibited one of the five FCFM patterns described above, which we considered as normal features. In contrast, and in accordance with Onco-Life autofluorescence observations, a dramatic decrease of the FCFM intensity was recorded in precancerous conditions. In addition, a specific pattern of the bronchial wall microstructure could be observed in some precancerous conditions exhibiting a disorganized fibered network (Figure 6A).
This pattern was observed in the close vicinity of one invasive cancer and three CIS, but also in two mild and one moderate dysplastic lesions as well as in three metaplastic lesions (Figures 6A–6D). The abrupt transition between this disorganized pattern and the complete disappearance of the microfluorescence could be observed in one CIS case (Figure 6A). This disorganized pattern or the complete disappearance of the bronchial FCFM microautofluorescence was observed in all but two dysplastic areas and in every CIS or invasive lesion (Table 1). The majority of nondysplastic precancerous lesions displayed a significant decrease of the fluorescence without distinguishable fibered network.
FCFM in Nonmalignant Bronchial Diseases In the patient with sarcoidosis, a "granulomatosis-like" FCFM image could be recorded (Figure 6E) on some bronchial carinas, whereas the macroscopic exploration was normal under white-light and fluorescence bronchoscopy. The corresponding biopsy samples displayed subepithelial granulomas (Figure 6F). In the patient with Mounier-Kühn syndrome, the Onco-Life examination found an intense decrease of the bronchial autofluorescence, except on the areas of the tracheal and main bronchus cartilage rings, whereas FCFM did not show any fibered microautofluorescence along the entire proximal bronchial tree. The corresponding spectral analysis showed a drastic decrease of the overall fluorescence intensity, but the spectral shape appeared similar to that of normal tissue (Figure 5C).
The epithelial basement membrane zone is a specialized area of the bronchial mucosa made up of two recognized component layers: the lamina densa, which is in direct contact with the epithelial cells, and the underlying lamina reticularis (22). The latter is especially pronounced under the respiratory epithelium of the large conducting airways, where it can be several microns thick (22). This basement membrane zone is known to play a crucial role in several pathologic conditions, such as asthma (6), and in the early steps of bronchial malignancies. In this study, we used a new biomedical imaging modality, called FCFM, to record clear and reproducible images of normal and pathologic human bronchial subepithelial connective tissue networks in vivo. Our data show the highly organized nature of this subepithelial region, displaying five different patterns depending on the size of the explored airway from the trachea up to the respiratory bronchioles. The FCFM images produced in our study on large bronchi are very similar to the lamina reticularis images that have been recently obtained from the rat trachea using whole mounts of the airways and fluorescence microscopy (23). As in this experiment, our in vivo approach showed a mat of large fibers mainly oriented along the longitudinal axis of the airways with cross-linked smaller fibers, as well as large openings that are supposed to correspond to the bronchial glands' origins. Our study also provides in vivo direct imaging evidence of the multilayer nature of the bronchial subepithelial connective tissue, made of two sheets of superimposed networks, oriented perpendicularly to each other, compatible with the imaging of the two main components of the basement membrane zone. If confirmed by further studies, this would represent the first in vivo microscopic imaging of the lamina densa and the lamina reticularis in humans. In addition, our in vivo technique makes it possible to take high-resolution images of small airways up to the respiratory bronchioles, which are very recognizable by their unique ringlike helicoidal structure. Our study also contributes to a better understanding of the fluorophores that produce the confocal autoflorescence microscopic images of normal and pathologic bronchial mucosae in vivo. Previous studies have shown that at 488-nm excitation wavelength, the light penetration into the tissue exceeds the thickness of the epithelium and is able to excite the submucosal fluorophores (24, 25). Therefore, the main endogenous fluorophores that could contribute to the bronchial mucosa autofluorescence at this wavelength are the intracellular flavins, which could originate from the epithelial cells (25, 26), and specific cross-links of collagens and elastin (27) present in the subepithelial areas. Indeed, our study shows that the epithelial cell autofluorescence is too weak to allow imaging of the epithelial layer using 488-nm FCFM, a result consistent with the report that the autofluorescence yield is about 10 times higher in the upper part of the submucosa than in the epithelium (24). The fibered nature of the tissue autofluorescence that we could obtain after removal of the bronchial epithelial layer ex vivo, as well as our spectral analysis of bronchial FCFM in vivo, indicate that the fluorescence signal emitted from the proximal airways under 488-nm wavelength excitation mainly originates from the elastin component of the subepithelial region. More surprisingly, the collagens that are supposed to represent the major components of the basement membrane zone do not seem to affect the FCFM image produced at 488 nm. This could in part be explained by the fact that, at 488 nm, the fluorescence yield of collagen is at least one order of magnitude smaller than that of elastin. Few previous studies have addressed the identification of the fluororophores involved in the bronchus autofluorescence emission in humans. These studies are also in accordance with a major role of elastin in the autofluorescence properties of the bronchial mucosae. Using 400-nm excitation wavelength under ex vivo conditions, Kobayashi and colleagues (28) found that the variation of the elastin density observed from histopathologic bronchial analysis is well correlated with the tissue autofluorescence intensity. Glanzman and colleagues (29) performed time-resolved autofluorescence spectroscopy of the bronchial mucosa ex vivo and in vivo during endoscopic procedures. At 406-nm excitation wavelength, one or two predominant fluorophores appeared to be involved, whose fluorescence lifetimes are compatible with elastin and/or collagen. This indicates that a modified FCFM device using shorter wavelengths would make it possible to image the collagen component of the matrix, together with the elastin component. The clinical trial presented in this article was not designed to compare FCFM imaging with other diagnostic tools for premalignant lesions, such as autofluorescence bronchoscopy. Therefore, the place of FCFM in the routine examination of patients at high risk for lung cancer remains to be determined. However, the "normal" subepithelial microfluorescence patterns we described appear to be highly specific for normal epithelium, as these features were found in every non–previously biopsied area corresponding to normal bronchial samples. On the contrary, the autofluorescence microstructure of the premalignant airways differs significantly from the normal bronchus. In this study, the regular fibered pattern was absent in almost every preinvasive lesion, underlining the major influence of epithelial cells on the microstructure of the bronchial matrix. In this regard, the specific disorganization of the fibered network in the vicinity of CIS and some lower grade precancerous lesions is a striking observation, which sheds some light on the origin of the precancerous fluorescence modifications. A number of factors have been hypothesized to explain these alterations, including a reduction in the epithelial cell fluorophore concentrations (26, 27), a modification in the physicochemical microenvironment of tissue, such as oxygenation and pH (31), a change in tissue architecture, such as thickening of the malignant epithelium, and an increase in microvascularization (5, 10). Our data strongly support the hypothesis of an early degradation of the basement membrane components to explain the decrease in autofluorescence in precancerous epithelium. The proteolytic alteration of the extracellular matrix is one of the major steps identified in tumor invasion (32). Recent studies of expression of matrix metalloproteinases and their inhibitors in premalignant bronchial epithelium showed significant alterations as early as in basal cell hyperplasia, as well as extensive disruptions of the basement membranes in areas of dysplasia and CIS (4), which may correspond to the specific disorganization of the subepithelial fibered network that we observed. In addition to the study of the premalignant bronchial wall alterations, the application of FCFM could be extended to nonmalignant bronchial diseases. In this study, we observed the complete disappearance of the bronchial wall fibered connective network in a tracheomegaly syndrome—a pathologic condition related to a defect in the elastic component of the bronchial wall. We also observed a remarkable FCFM aspect in a case of bronchial sarcoidosis. Although still limited, these observations indicate that FCFM could be used to study specific basement membrane remodeling alterations, such as in chronic bronchial inflammations, asthma (33), and chronic obstructive pulmonary disease (6). The FCFM is a simple and minimally invasive procedure that can be performed during a fiberoptic bronchoscopy under local anesthesia. Coupled with the spectral analysis, FCFM provides a qualitative and quantitative method of analysis of the connective tissue microstructure of the airways. This is also the very first described endoscopic technique that makes it possible to analyze very distal small airways up to the respiratory bronchioles. Epithelial nuclear staining using acriflavin shows that FCFM can also produce high-definition images of the epithelial layer that would make it possible to differentiate normal, premalignant, and malignant alterations at the microscopic level. Using nontoxic exogenous fluorophores and appropriate wavelengths, FCFM may become a very powerful technique for in vivo diagnosis of early malignant and premalignant conditions of the bronchial tree by allowing the analysis of both the epithelial and subepithelial layers. Such a technique would also make it possible to precisely study the natural history of the premalignant epithelium, without sampling intervention.
Supported by the French Ministry of Health (Programme Hospitalier de Recherche Clinique 2001) and the French Canceropole Nord Ouest. The F400/S prototype and fiberoptic probes used for the clinical study were kindly provided by Mauna Kea Technologies, Paris, France. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200605-684OC on October 5, 2006
Conflict of Interest Statement: L.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.M-S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. T.V. is an employee of Mauna Kea Technologies. E.P. is an employee of Mauna Kea Technologies. C.C. is an employee of Mauna Kea Technologies. G.B.H. has served as a scientific consultant and patent coauthor with Mauna Kea Technologies since May, 2004. This is an ongoing scientific collaboration and she has received a monthly compensation of Received in original form May 20, 2006; accepted in final form October 5, 2006
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
850 during that time. Mauna Kea Technologies provided the Rouen University Hospital with the prototype device and the fiberoptic probes used during this work, including the software needed for the research project.