Introduction

Histopathological tissue assessment with multidisciplinary correlation [1] remains the gold standard for accurate diagnosis of many lung conditions, including solitary nodules, malignancies, and diffuse lung diseases (DLD). Tissue diagnosis requires lesions to be biopsied, either bronchoscopically or surgically. When undertaking biopsies of DLD or distal lesions at bronchoscopy there is a need for blind transbronchial biopsies, with a 1-12% risk of pneumothorax [2–5] and a 2-9% risk of significant bleeding [3, 6, 7]. Transthoracic biopsy is preferred for peripheral lesions, either percutaneously or via minithoracotomy or thoracoscopy (VATS) [8]. Surgical biopsy requires a general anaesthetic, a longer hospital stay, and a different spectrum of risks [9], particularly as this patient group already has significant respiratory compromise. There are time and cost implications with the preparation and reporting of tissue samples, extracted specimens are susceptible to minor morphological alterations from dehydration and chemical fixation, and biopsying creates trauma and scarring that may hinder subsequent attempts to monitor dysplastic lesions. Microscopic examination of cells from bronchoalveolar lavage (BAL) provides further information, but even this can cause dyspnoea, pleuritic pain [10], and fever/flulike symptoms [11].

White light videobronchoscopy is commonly used for rapid macroscopic examination of the large and medium airways but has a low sensitivity for the detection of small superficial mucosal abnormalities and areas of intraepithelial neoplasia [12]. This has led to the development of modalities such as autofluorescence bronchoscopy (AFB) [13–15], narrow-band imaging (NBI) [16, 17], high-magnification bronchoscopy [18], and optical spectroscopy [19] in order to better recognise tissue abnormalities. Whilst relatively straightforward to perform and highly sensitive, their specificity to differentiate neoplasia from inflammation or dysplasia is insufficient to obviate the need for tissue biopsy, and they are unable to image beyond the proximal bronchial tree or add diagnostic information during the investigation of DLD. Endobronchial ultrasound (EBUS) is an increasingly utilised tool for imaging beyond the mucosal surface. High-frequency (20 MHz) radial ultrasound probes have a resolution of less than 1 mm and can distinguish certain histological characteristics of the airway wall [20] and of peripheral lesions [21, 22], but even at this resolution biopsies are still required for a firm diagnosis to be made.

Technologies with better resolution and contrast have therefore been sought in order to eliminate the need for conventional biopsy by providing histopathological quality real-time images in situ: the so-called “optical biopsy.” Whilst they are still primarily used for research, this review presents the limited but often promising clinical data available for three of these technologies (technical outline in Table 1).

Table 1 A summary of the three main bronchoscopic optical biopsy technologies
Full size table

Optical Coherence Tomography

Optical coherence tomography (OCT) is a probe-based imaging technique that is being explored for several clinical niches, including gastroenterology [23], ophthalmology [24], dermatology [25], and vascular medicine [26, 27]. Analogous to ultrasonography, a catheter tip transmits broadband near-infrared (NIR) light through an optical fibre to detect the reflectance and backscattering properties of the tissue, each of which is influenced by changes in tissue density or refractive index boundaries such as cell membranes or structures [28]. By measuring the echo time delay between transmission and detection using a technique called low-coherence interferometry, an optical depth scan may be recorded. A B-scan may then be performed by translating the optical probe, allowing, for instance, a circumferential transluminal cross-sectional tomographic image to be constructed[29]. Air transmits NIR light, so the technique is immediately suited to the airways and, unlike endobronchial ultrasound (EBUS), there is no requirement for direct contact with tissue, eliminating any potential physical distortion. Whilst current technology provides only 2-3 mm of penetration, this is sufficient for imaging most early bronchogenic carcinomas. As the diffraction limit and coherence length of the light source is a fraction of those of the acoustic waves used in ultrasound, the lateral and axial resolution of OCT can be as good as 5 μm, which is 10-20 times lower than high-frequency ultrasound [21], though commercially 15-μm resolution is more realistic which still enables OCT to image to the level of individual nuclei [30]. It is a relatively inexpensive and portable technology: the collaboration between LightLab Imaging in Boston and Pentax in Tokyo produced a radially scanning 1.5-mm-diameter probe that can be passed down the working channel of a bronchoscope [30]; the Niris imaging system (Imalux Corp, Cleveland, OH) is forward scanning with a larger outer diameter of 2.7 mm [31].

In ex vivo studies OCT shows the layered structure of the bronchial tree, demarcating the mucosa, lamina propria, and the underlying perichondrium and cartilage. The boundary of the epithelium and cartilage is particularly distinct, and it also has sufficient resolution to outline mucous glands and layers of smooth muscle within the submucosa [32]. The structure of the alveoli adjacent to peripheral airways has also been shown [33]. During in vivo OCT with the aforementioned forward-scanning probe, Michel et al. [31] recently noted that these tissue boundaries are destroyed by endobronchial tumours, and during fluorescence bronchoscopic examination by Lam et al. [30], radial OCT enabled accurate measurement of the respiratory epithelium thickness. Breaches of the highly scattering basement membrane allowed differentiation between invasive neoplasia (although not histological subtype), carcinoma in situ, dysplasia, and metaplasia (Fig. 1). However, the resolution is currently inadequate to differentiate between grades of dysplasia.

Fig. 1
figure 1

a, c, e Mild, moderate, and severe dysplasia, respectively, of the bronchial epithelium under OCT imaging with corresponding H&E histological images underneath (b, d, f) (×20 magnification). Worsening dysplasia is demonstrated by increasing thickness of the epithelium (e) above the highly scattering basement membrane (bm). The epithelial nuclei become more visible as black dots with moderate and severe dysplasia (reprinted from [30] with permission from the American Association for Cancer Research)

Full size image

The implementation of swept-source spectral OCT and Fourier domain OCT systems has increased the sensitivity of the technique over the original time domain systems [34, 35]. These developments have led to faster acquisition times so that 3D volumetric data sets may now be recorded in a matter of seconds if the appropriate scanning mechanism is used [36]. The data may then be visualised by rendering transverse images (similar to conventional histopathology), or en-face images, increasing the flexibility of these clinical systems, and the short time required makes large-organ imaging a possibility. OCT is thus emerging as a technique that may aid early bronchoscopic diagnosis of microinvasive pathological lesions, and as it appears to image sufficiently deeply to identify alveolar structures, it may have future potential in the investigation of DLD.

Confocal Laser Fluorescence Endomicroscopy

The recent miniaturisation of the confocal laser scanning microscope (CLSM) now enables in situ imaging of a very thin, focused section of superficial tissue. For in vivo confocal imaging of the gastrointestinal tract, the scanning mirrors and miniature objective lens can be integrated within an endoscope [37]; this arrangement is known as “distal scanning.” Alternatively, the mirrors and lens are housed within an external box on the endoscope stack, thus allowing the confocal laser pulses to be directed into the patient via thousands of optical fibres [38]. This configuration is known as “proximal scanning” or “fibered confocal fluorescence microscopy” [39]. The optical fibres are wrapped in a narrow sheath, enabling the resultant probe to be inserted down the biopsy channel of a conventional endoscope. Probe-based confocal laser endomicroscopy (pCLE) has been adapted for bronchoscopy; the only commercial system being the Cellvizio-Lung® (Mauna Kea Technologies, France) (Fig. 2a). The 1.4-mm-diameter probe shines a 488-nm-wavelength blue excitation laser light at the tissue to be imaged and any radiation between 500 and 650 nm from fluorescing tissue is collected at 12 frames per second. This creates a near histological quality image but only to a tissue depth of approximately 50 μm and with a lateral resolution of 3.5 μm, significantly better than that of OCT. Whilst the field of view is only 600 × 500 μm, if the probe can be slid slowly over the biopsy site, software can “stitch” adjacent images together to create a more representative mosaic of the tissue being analysed [40]. In peripheral optical biopsy the probe is inserted blindly beyond the videobronchoscopic view until it is lodged and a relatively fixed alveolar image is obtained. The potential for tissue mosaicing is therefore greater in the less constraining central airways, but the gentle and smooth manipulation that is required remains ergonomically challenging due to cardiorespiratory and patient movement and the understanding that good imaging requires the sometimes awkward task of perpendicular probe abutment with the tissue.

Fig. 2
figure 2

a The pCLE is easily inserted down the operating channel of a bronchoscope. b Healthy right main bronchus wall under pCLE demonstrating cross-hatching of the basement membrane and a bronchial gland toward the top. c Mosaicing increases the field of view; this shows the elastin scaffold of a microvessel (ellipse) and a septal wall (circle).d An area of moderate emphysema with some reduction in acinar elastin. e Adjacent to a bullous where elastic fibres are particularly sparse. f pCLE images from the lobules of smokers show untethered hyperfluorescent macrophages (MPEG video loops are available in the online supplement)

Full size image

CLSM imaging of most epithelial surfaces within the body requires the topical or intravenous administration of an exogenous fluorescent contrast agent such as fluorescein or acriflavine. However, elastin acts as an endogenous fluorophore and it is abundant both within the basement membrane of the bronchial tree (Fig. 2b) and as a structural connective tissue component of the alveolar walls. Analysis of the spectral emission shape has demonstrated that these structures are responsible for the images created by pCLE without exogenous contrast during bronchoscopy [39]. Thiberville et al. [39] described five distinct lattice arrangements of the connective tissue fibres of the normal basement membrane in separate areas of the bronchial tree, with small openings present down to the lobar divisions, representing bronchial gland openings. These regular structures became disorganised and the intensity of the fluorescing signal diminished when overlying malignant and premalignant epithelium was present. Early work suggests that nonmalignant disease also affects the basement membrane appearance during pCLE [41, 42]; for example, the bronchial carina of a patient with sarcoidosis produced a granulomatosis-like image [39]. To image the respiratory epithelium overlying the basement membrane, a contrast agent must be used: early work suggests that physiological pH cresyl violet applied topically [43] is more effective than intravenous fluorescein [44].

A “distal scanning” confocal endomicroscope prototype was recently constructed [45]. At 6.2-mm diameter without facility for white light imaging, it requires rigid bronchoscopy under general anaesthetic but has been used to image endobronchial mucosal abnormalities of larger airways using intravenous fluorescein. The major advantage of distal scanning devices over pCLE is that the distal optics allow the Z-depth of imaging to be altered during the procedure; in this case 0-200 μm below the surface, from individual ciliated epithelial cells down to submucosal smooth muscle and microvessels.

The thinnest commercially available bronchoscope containing a working channel is the Pentax FB-8 V® with a diameter of 2.7 mm, but pCLE probes can be inserted beyond this to microimage the alveolar sacs [46]. Whilst the blind insertion has similar targeting accuracy to that of bronchoscopic forceps during peripheral biopsy, limited study has shown no risk of pneumothorax [46] and only occasional transient bleeding [47], although there can be transient pain when the probe tip is close to the pleura [46, 47]. The microanatomy of the alveolar entrances, alveolar ducts, and microvessels is clearly shown (Fig. 2c), with cyclical changes in the size and shape of the structure during the breathing cycle evident. It also seems that the autofluorescence increases with age. In healthy volunteers, the visible presence of macrophages during pCLE examination appears to be specific to smokers [46]; absorption of tar is the likely cause [48]. For the many current smokers who undergo bronchoscopy, these bright macrophages act as an additional endogenous fluorophore (Fig. 2f). Early work suggests that the pCLE images of patients with advanced emphysema display fewer, more separated elastic fibres from alveolar septal walls and ducts [47, 49] (Fig. 2d, e); perhaps not surprising when elastin is depleted within alveoli affected by chronic obstructive pulmonary disease (COPD) [50].

Future pCLE work will establish the morphological differences of the elastin scaffold of the acinar structure in health and disease [51] and further the understanding of the influences that airway lesions have upon the basement membrane. Topical contrast agents will be employed to further characterise the epithelium [43, 45] and changes to it with preinvasive lesions. Intravenous contrast agents may help to establish characteristic effects of individual lung diseases on the microvasculature [52], as has been done with CLSM in the gastrointestinal tract [53]. Intravenous fluorescein may help to highlight peripheral pathology [44], or another approach has been to use pCLE at 660 nm and inject topical methylene blue (MB) into the acinus via an extended working channel [54]. The lower frequency is then able to fluoresce the MB. In the future, combinations of CLSM wavelengths may be used to provoke simultaneous autofluorescence from multiple endogenous sources to provide further information [55]. Other recent important biophotonics developments include two-photon-induced fluorescence and second harmonic generation in tissue [56]. These techniques use short pulses of infrared light to induce the signal, which both increases the penetration depth and also effectively allows more diagnostically useful high-energy excitation without using potentially harmful ultraviolet light. So far the technique has been implemented in skin inspection, but a number of research groups are working on the endoscopic implementation [57].

Endocytoscopy

The endocytoscope is a system by which real-time in vivo imaging of tissues at a cellular level can be achieved under white light conditions. It relies upon the principle of contact microscopy, where an image is seen for as long as the probe remains in contact with the tissues, and was first used for microhysteroscopy [58]. It has since been used predominantly as a research tool in the investigation of cancer in the gastrointestinal tract [59–61], and also in the urinary tract [62]. Detailed images of the nuclei allow clear distinctions between normal and malignant cells, with sharp borders between atypical and normal epithelium, differences in cellularity, and differences in the sizes and shapes of the nuclei [63]. The technique seems particularly effective in the colon and oesophagus [64].

Application of a 0.5-1% MB solution to the area of interest allows sharp visualisation of cellular structures down to the level of the nucleolus, allowing near cytological quality images to be viewed without the need for biopsy [64], and once again overcoming the degradation of tissue samples secondary to crushing at the time of biopsy or chemical preparation of histology samples. However, mucous and blood can impair the application of the contrast agent and prevent suitable images from being obtained. This can be partly overcome by the use of mucolytic agents [65].

More recently, endocytoscopy of the airways has been made possible [66] (XEC-300-U/F, Olympus, Tokyo, Japan), and is available either as an integrated system with normal and endocytoscope functions (Fig. 3), or as a dedicated ultrafine probe with a diameter of 3.2 mm that can be used via the biopsy channel of a larger bronchoscope [67]. Both systems have a spatial resolution of 4.2 μm and a magnification of ×570, with fields of view of 400 × 400 μm and 300 × 300 μm, respectively.

Fig. 3
figure 3

An Olympus XEC-300-U bronchoscope with integrated endocytoscope function. Clockwise from operating channel: light guide lens, normal video bronchoscopy objective lens, endocytoscope objective lens, and second light guide lens

Full size image

Unlike pCLE, endocytoscope images are in colour (Fig. 4), but the lack of depth of field (similar to pCLE; up to a maximum of 50 μm) with endocytoscopy essentially offers a two-dimensional view of the tissue surface, whereas pCLE builds a limited three-dimensional digital image of the target tissues, as is demonstrated when simultaneously imaging the walls of multiple alveolar sacs branching off the alveolar duct. The smaller size of the pCLE probe also allows for more distal imaging. Current practice enables imaging of only the elastin fibres with 488-nm pCLE, but by using contrast agents, cellular information may be gleaned to complement the cytological and histological analysis that is proposed with the endocytoscope. Indeed, early work suggests that in vivo endocytoscopic images of endobronchial tumours, sprayed with MB, resemble conventional histology, with an increase in cellular density, polymorphic cells, and an increased nuclear-to-cytoplasm ratio when compared to images of normal mucosa [67, 68].

Fig. 4
figure 4

Endocytoscope images using methylene blue as the contrast agent of a healthy bronchial mucosa and b an RB1/2 entrance endobronchial tumour with a marked heterogeneous epithelial appearance. The tumour’s white light image (c) is shown with the autofluorescence image (d), with malignant area indicated by arrow

Full size image

Conclusion

At present, the advanced imaging modalities discussed here are utilised largely as research tools, but as experience is gained and the techniques become more widely available, there is the real possibility of reliable, in vivo diagnosis of abnormalities in any easily accessible hollow organ. This is especially exciting in the field of respiratory medicine, as current diagnostic methods, such as transbronchial and transthoracic biopsy, entail appreciable risk to subjects whose lung function is often already severely abnormal.

The expense of these new “optical biopsy” technologies and the possible risks of additional time required to perform them may be justified if technological advancements enable them to show impact in certain areas. In the future, if more extensive clinical trials can establish comparable sensitivity and specificity to histopathology, they may be an adjunct to autofluorescence bronchoscopy, enhancing its specificity and ensuring higher-yield endobronchial biopsy samples. They may have a role in the monitoring of dysplastic airway lesions, as the natural history of these is poorly understood and it is appreciated that many will spontaneously regress. Monitoring the natural history of an epithelial lesion after conventional biopsy is difficult because of possible scarring, and there is a potential for entirely removing the lesion within the biopsy hook or possibly inducing spontaneous regression. The patient may be exposed to unnecessary risk with repeated biopsies for a lesion that may never progress to malignancy. Optical biopsy may also eliminate the risks associated with diagnosis of DLDs using transbronchial and surgical biopsies in high-risk patient populations, and then enable repeated alveoscopy to follow the microscopic temporal changes of DLD with or without treatment [69, 70]. These techniques may also improve the speed of diagnosis, allowing earlier treatment and reducing patient anxiety.