1.

Introduction

Detailed endoscopic observation of the esophagus, stomach, and colon has become possible due to advances in magnifying endoscopy and conventional endoscopy.^{1, 2} However, magnified observation at the cellular level remains difficult under endoscopic examination, thus often making histopathologic examination via biopsy a necessity. Laser-scanning confocal microscopy (LCM) provides in-vivo images that are close in quality to histopathologic images. This technology is currently being applied clinically in the field of gastroenterology. Most reports are of fluorescence-type LCM, which require a fluorescent agent. ^{3, 4, 5, 6, 7, 8, 9, 10, 11} Reflectance-type LCM, which does not need fluorescence, is in the investigational stage, and most reports are of in-vitro studies.¹² We compared in-vivo LCM images and histologic images of early gastroesophageal cancer and normal mucosa to assess the usefulness of a newly developed prototype catheter-based reflectance-type LCM system (Mauna Kea Technologies, Paris, France; Fujinon, Saitama, Japan) for diagnosing gastroesophageal cancer.

2.

Materials and Methods

2.1.

Instrument Specifications

We used a prototype LCM system equipped with a pulsed semiconductor laser centered at $685\mathrm{nm}$ . The combination of pulsed illumination ( $15\text{-}\mathrm{ns}$ pulse width, $80\text{-}\mathrm{ns}$ repetition period) with a time-gated detection of the reflected light ( $15\text{-}\mathrm{ns}$ detection window, $40\text{-}\mathrm{ns}$ delay) permits us to overcome the back-reflections onto the proximal optics by means of light travel-time differentiation.¹³ The flexible catheter probe was $2.6\mathrm{mm}$ in outer diameter and $3876\mathrm{mm}$ long (Fig. 1 ). The scanning field was $\mathrm{30,000}\text{pixels}$ —the number of fibers in the bundle—and the frame rate was 12 images per second (Fig. 2 ). An objective with a numerical aperture of 1.2 was placed in contact with the tissue, with a focus of $30\mu \mathrm{m}$ from the objective lens, a lateral resolution of less than $1\mu \mathrm{m}$ , and an observation area $160\mu \mathrm{m}$ in diameter. The catheter probes were connected to the laser scanning unit and introduced under direct endoscopic visualization. LCM was performed after the flexible confocal catheter probe was introduced through the instrument channel of the endoscope (Fig. 3 ). All images of the LCM examinations were inspected, recorded, and stored digitally as real-time video sequences with the use of software on a PowerMac G5 Dual $1.8\text{-}\mathrm{GHz}$ personal computer (Apple Computer, Cupertino, California). In addition, still LCM images were saved for future review.

Fig. 1

Catheter-based reflectance-type laser-scanning confocal microscope (Mauna Kea Technologies, Paris, France; Fujinon, Saitama, Japan).

Fig. 2

Schema of the catheter-based reflectance-type laser-scanning confocal microscopy.

Fig. 3

LCM examination for early gastric cancer under endoscopy.

3.

Results

With the distal tip of the catheter placed gently against the mucosa and the cancer, real-time LCM images of the normal mucosa and of the cancer of the esophagus and stomach were obtained safely and easily, and the influence of slight motion was ignored. Even if water or blood existed in the normal mucosa and cancer, the influence of them was also ignored. The time required for scanning each normal and cancer site ranged between 16 and $390\sec$ .

3.1.

Normal Esophageal Mucosa

In LCM images of normal esophageal mucosa, high-reflectivity spots were observed near the center of honeycomb-like structures of high reflectivity. These high reflectivity spots and structures in the LCM images appeared to correspond to nuclei and cell membranes, respectively, in the histologic images of hematoxylin-eosin-stained sections (Fig. 4 ).

Fig. 4

Images of normal esophageal mucosal: (a) laser-scanning confocal microscopy image, and (b) hematoxylin-eosin-stained tissue from the same specimen.

3.2.

Esophageal Cancer

In LCM images of esophageal cancer, high-reflectivity spots that were considered nuclei were observed. The nucleus-to-cytoplasm (N/C) ratio was much increased, and honeycomb-like structures of high reflectivity, considered cell membranes, were not observed (Fig. 5 ).

Fig. 5

Images of esophageal cancer: (a) laser-scanning confocal microscopy image, and (b) hematoxylin-eosin-stained tissue from the same specimen.

3.3.

Normal Gastric Mucosa

In LCM images of normal gastric mucosa, cell membranes and nuclei were not visualized. However, the crypt cells were arranged like flower petals surrounding the gastric pit (Fig. 6 ).

Fig. 6

Images of normal gastric mucosa: (a) laser-scanning confocal microscopy image, and (b) hematoxylin-eosin-stained tissue from the same specimen.

3.4.

Gastric Cancer

In LCM images of differentiated adenocarcinoma of the stomach, cell membranes were not visualized, and a disorganized configuration of glands with high-reflectivity spots that were considered nuclei was observed (Fig. 7 ). In LCM images of undifferentiated adenocarcinoma, no ductal structure was recognized; only an amorphous structure was seen. Cell membranes and nuclei were not visualized (Fig. 8 ).

Fig. 7

Images of differentiated adenocarcinoma of the stomach: (a) laser-scanning confocal microscopy image, and (b) hematoxylin-eosin-stained tissue from the same specimen.

Fig. 8

Images of undifferentiated adenocarcinoma of the stomach: (a) Laser-scanning confocal microscopy image, and (b) hematoxylin-eosin-stained tissue from the same specimen.

Visualization of nuclei and cell membranes in LCM images in relation to histologic diagnoses is shown in Table 2 . In all normal esophageal mucosa and esophageal cancers, the nuclei were visualized. In nine of the ten (90%) normal esophageal mucosa, cell membranes were visualized, and in five of the ten (50%) esophageal cancers, cell membranes were visualized. In all normal gastric mucosa, nuclei and cell membranes were not visualized, but in ten of the 20 (50%) gastric cancers, nuclei were visualized. The storoma was visulized as high reflectivity. In some case, low-reflectivity spots were also observed in the LCM images, which was considered mucin in goblet cells in the hematoxylin-eosin-stained specimen.

Table 2

Visualization of nuclei and cell membrane in LCM images in relation to histologic diagnoses. Number (and percentage) of samples are shown.

Histologic diagnosis	Nucleus	Cell membrane
Normal esophageal mucosa $(n=10)$	10 (100)	9 (90)
Esophageal cancer $(n=10)$	10 (100)	5 (50)
Normal gastric mucosa $(n=20)$	0 (0)	0 (0)
Gastric cancer: well $(n=10)$	6 (60)	0 (0)
moderately $(n=6)$	4 (66.7)	0 (0)
papillary $(n=1)$	0 (0)	0 (0)
poorly $(n=1)$	0 (0)	0 (0)
signet ring cell $(n=2)$	0 (0)	0 (0)

4.

Discussion

Recent advances in endoscopic technology have afforded high-quality, detailed diagnosis of gastrointestinal diseases. To confirm the presence of malignancy, however, snip biopsy is often performed under endoscopy when endoscopic examination reveals an abnormality. Thus, biopsy is performed for many lesions that are subsequently determined not to be malignant. Histologic analysis of biopsy material remains the gold standard for the final diagnosis of a gastrointestinal lesion. Histologic diagnosis via biopsy involves the following process: formalin fixation of the specimen, cutting the specimen into small columns, paraffin embedding, ultrathin slicing, deparaffinization, staining, glass slide, mounting, and finally light microscopic observation. Moreover, it takes several days to obtain a diagnosis. Also, snip biopsy is associated with bleeding, apparent endoscopic disappearance of cancer cells after biopsy, and artificial ulceration, which make endoscopic treatment, e.g., EMR, ESD, and EAM, difficult. In addition, because of the bleeding, biopsy cannot be easily performed in patients taking anticoagulants.

Being able to accurately image a lesion in vivo at the time of endoscopic examination without biopsy allows for prompt diagnosis and treatment. Fluorescence-type LCM is reported to be a promising tool for in-vivo histopathologic examination during endoscopy, and might overcome the disadvantages associated with conventional biopsy. ^{3, 4, 5, 6, 7, 8, 9, 10, 11} In recent years, there have been several reports describing the ability to obtain an LCM image that corresponds precisely to the histopathologic tissue diagnosis in cases of gastrointestinal tract disease. ^{3, 4, 5, 6, 7, 8, 9, 10, 11, 12} However, many of the reports were based on observations made on excised specimens or with fluorescence-type LCM. We also have previously used probe-based reflectance-type LCM to obtain images that are close to histopathologic specimens in vitro.¹² In the present study, however, we conducted examinations in vivo using catheter-based reflectance-type LCM, which enabled us to insert the microscope through the instrument channel of the endoscope and to capture images at a single depth of $30\mu \mathrm{m}$ below the tissue surface. In our LCM observations of the esophagus, nuclei were detected at both sites of normal mucosa and cancer. In our LCM observations of the stomach, nuclei were not recognized in normal mucosa but were recognized in 50% of cancer sites. Because the slice of LCM was very thin, we assumed that the nuclei in the normal esophageal mucosa were easily visualized because the cells were composed of stratified squamous epithelial cells. Likewise, because the N/C ratio increased, we assumed that the nuclei of the esophageal cancer and gastric cancer were visualized, whereas the nuclei of the normal gastric mucosa were not visualized. Although further prospectives and more studies, e.g., immediate diagnosis of neoplasia versus inflammation, is needed, we have shown that catheter-based reflectance-type LCM can provide images at the cellular level in vivo, suggesting the possibility of immediate cancer diagnosis under endoscopic observation without the need for biopsy.

A fluorescence-type LCM system that uses a catheter was recently developed by Mauna Kea Technologies.^{3, 4, 5, 6, 7, 8} This LCM system has the capability to provide dynamic $(12\text{frames}∕\sec )$ ultrahigh resolution images at the cellular level on a field of view as wide as $260\times 260\mu \mathrm{m}$ with 1.5-lateral and $10\text{-}\mu \mathrm{m}$ axial resolutions, at $60\text{-}\mu \mathrm{m}$ working depth. To overcome the limits of the field of view, an image reconstruction algorithm that uses video mosaicing has been developed.³

One advantage of catheter-based LCM is that it allows the capture of an image during conventional endoscopic examination without changing to a specialized scope. The catheter-based reflectance-type LCM used in this study is of a size and flexibility to pass through the endoscopic instrument channel and to be placed accurately on the mucosa with guidance from the white-light endoscopic image. Furthermore, there is a report of in-vivo acquisition of real-time and dynamic histologic images of the peritoneum, liver, and spleen during a novel, minimally invasive transgastric approach to surgery, termed natural orifice transluminal endoscopic surgery (NOTES).⁴ Compared to reflectance-type LCM, fluorescence-type LCM can provide images with higher signal-to-noise ratios (although a fluorescence agent is needed). The fluorescence-type LCM device can be used for diagnosing cancer,¹¹ visualizing lymphoepithelial lesions in gastric mucosa-associated lymphoid tissue-type lymphoma,⁵ detecting angiodysplasia,⁶ visualizing Helicobacter pylori,⁹ diagnosing lymphocytic colitis,⁷ diagnosing a dysplasia-associated lesional mass or adenoma-like mass in patients with ulcerative colitis,¹⁰ and for functional examinations that provide moving images with visualization of blood flow through microvessels.⁸ Unlike fluorescence-type LCM systems, reflectance-type LCM collects and counts the reflective laser beam and therefore requires no staining process. The fluorescence-type LCM requires some staining to obtain clear images, but the acquired image is of high quality, and the signal-to-noise ratio is better than with the reflectance-type LCM. Further comparison of the two systems is needed, but we believe that these two instruments complement each other.

In summary, this feasibility study shows that catheter-based reflectance-type LCM can be used in clinical practice to provide instant images that correspond well with hematoxylin-eosin-stained microscopic images. Therefore, we expect that this novel method will aid in immediate diagnosis under endoscopy without the need for biopsy.