Magnifying endoscopy in upper gastroenterology for assessing lesions before completing endoscopic removal

Abstract

Any prognosis of gastrointestinal (GI) cancer is closely related to the stage of the disease at diagnosis. Endoscopic submucosal dissection (ESD) and en bloc endoscopic mucosal resection (EMR) have been performed as curative treatments for many early-stage GI lesions in recent years. The technologies have been widely accepted in many Asian countries because they are minimally invasive and supply thorough histopathologic evaluation of the specimens. However, before engaging in endoscopic therapy, an accurate diagnosis is a precondition to effecting the complete cure of the underlying malignancy or carcinoma in situ. For the past few years, many new types of endoscopic techniques, including magnifying endoscopy with narrow-band imaging (ME-NBI), have emerged in many countries because these methods provide a strong indication of early lesions and are very useful in determining treatment options before ESD or EMR. However, to date, there is no comparable classification equivalent to “Kudo’s Pit Pattern Classification in the colon”, for the upper GI, there is still no clear internationally accepted classification system of magnifying endoscopy. Therefore, in order to help unify some viewpoints, here we will review the defining optical imaging characteristics and the current representative classifications of microvascular and microsurface patterns in the upper GI tract under ME-NBI, describe the accurate relationship between them and the pathological diagnosis, and their clinical applications prior to ESD or en bloc EMR. We will also discuss assessing the differentiation and depth of invasion, defying the lateral spread of involvement and targeting biopsy in real time.

Key Words: Magnifying endoscopy with narrow-band imaging, Upper gastroenterology, Assessment, Endoscopic submucosal dissection, Endoscopic mucosal resection

INTRODUCTION

Gastrointestinal (GI) cancer is a major medical and economic burden worldwide. Esophageal and gastric cancers remain a considerable source of morbidity and mortality in Asian countries. For instance, in Linxian, Henan province (China), cancer of the upper GI tract is endemic. Mortality rates for esophageal cancer in Linxian exceed the American average (for white men) one hundredfold[1]. The prognosis of GI cancer is closely related to the stage of disease at diagnosis, and most cases are still detected at advanced stages and result in a relevantly poor outcome[2]. Early detection of these neoplasms or their precursors may be the only chance to reduce this high mortality.

Early GI cancers - such as Barrett’s esophagus (BE) with high-grade dysplasia and early gastric cancer (EGC)-whose invasion is limited to the mucosa or submucosa regardless of the size or the presence of regional lymph-node and distant metastasis[3], confer a survival rate of greater than 90% in 5 years in many centres[4,5].

The screening program for gastric cancer in Japan indicates that 53% of diagnosed gastric cancers are localized lesions. Additionally, the accumulated clinical experience and formal outcome studies have shown that the majority of early-stage neoplastic lesions is localized with a low risk of lymph node metastasis. Recent data from 3261 patients who underwent gastrectomy with meticulous D2-level lymph node dissection over a 30-year period show that lymph node invasion was observed in only 2.7% of mucosal tumors and 18.6% of EGC invading the submucosa[6]. Clinical experience suggests that complete resection of the cancer is possible, and a cure can be achieved as long as the potential for metastatic spread is definitively excluded[7,8].

Based on the above knowledge, the doctors began to try to use endoscopes for local excision with GI early tumors in situ, invading lamina propia or submucosa. More than a decade ago, endoscopic mucosal resection (EMR) technique emerged first in Japan as a critical tool in the management of patients with both high-grade dysplasia and superficial carcinomas[9]. But the indication of EMR is generally limited to mucosal tumors less than 2 cm in size even with the series of improvements that have been most widely used in recent years, such as using a transparent cap-fitted endoscope to suck targeted lesions into the cap and resect them with a snare (EMR-C) or a ligation device (EMR-L). All above EMR technologies are difficult to resect en bloc tumors larger than 2 cm in size, which is required for accurate and reliable pathological examination. However, though some endoscopists adopt piecemeal EMR techniques in order to cure the larger lesions, further investigation has revealed it involves problems such as remnants or high recurrent rates due to incomplete resections[10]. Thus, to overcome the problem of EMR techniques, a recent key issue in the field of therapeutic endoscopy is the development of a new therapeutic strategy for early GI cancers using endoscopic submucosal dissection (ESD). In this procedure, submucosal dissection is carried out by using an electrocautery knife to acquire a single-piece specimen, which is the gold-standard technique for offering en bloc resection of large superficial tumors in the GI tract, especially when R0 resection cannot be performed with other resection techniques. Within only a few years, ESD has become widespread in Asian countries - such as Japan, Korea and China - where there is a large volume of early upper GI lesions that need endoscopic treatment. However, there are hardly any reports about long-term results after ESD, and the procedure involves a much higher complication rate and requires much higher skills[11,12].

The two endoscopic local procedures are increasingly accepted by many patients and doctors mainly because they (1) provide new alternatives for minimal invasiveness; (2) are perhaps the first approximations to true intraluminal resection of superficial malignant GI neoplasms; and (3) yield results that are comparable to surgery. They also result in lower morbidity rates, lower costs and better quality of life than traditional surgery because of tissue preservation. But the difficulty lies in achieving en bloc or R0 resection and getting improved survival that precisely assesses resection margins and the depth of malignant invasion prior to performing EMR or ESD. The lesions with undifferentiated histology, lymphatic or vascular involvement and submucosal invasion were excluded due to possible lymph node metastases[3].

Therefore, a thorough preoperative endoscopic examination is considered necessary for selecting the appropriate therapeutic modality. Due to this requirement, endoscopic equipment has improved markedly with respect to resolution in recent years. However, in 1967, Okuyama et al[13] produced a magnification endoscope for viewing the gastric mucosa. At present, magnification endoscopes have the ability to enlarge the image from 1.5 × to 150 × and produce images that have pixel densities as high as 850 000, allowing the discrimination of objects that are only 10-71 μm in diameter[14]. The newest magnification endoscopes permit magnification without loss of resolution[15]. Nevertheless, it was reported recently that some GI disorders, such as intestinal metaplasia, often appear translucent when observed with magnification endoscopy alone. Thus, the mucosal surface cannot be easily examined without staining[16]. Methylene blue, Lugol’s iodine, and indigo carmine are several topical stains or pigments that have been used in conjunction with magnification endoscopy to improve tissue localization, characterization, or diagnosis during endoscopy[17]. The technique known as magnification chromoendoscopy (MCE) has been applied in a variety of clinical settings and throughout the GI tract for more than 10 years. In addition, other newer technologies, including narrow band imaging (NBI), that have proved particularly helpful during gastrointestinal endoscopic examinations have been developing in recent years. This shows that the two techniques have a similarly high sensitivity for detecting early neoplasia in the upper GI tract[18,19]. However, compared with MCE, the “electronic dyeing endoscopy,” such as NBI, that are based upon the phenomenon that the depth of light penetration depends on its wavelength, are more user-friendly because their filters can be manually enabled and disabled during endoscopy, making it easy to switch them between the standard mode and the “electronic dyeing” mode, and no staining agents are required. Beyond these practical advantages, NBI reveals the superficial capillary network with a high contrast due to absorption of the blue light by hemoglobin, whereas the vascular pattern is often less visible in chromoendoscopy[20]. When magnifying endoscopy is combined with narrow band imaging (ME-NBI), the combination has been shown to enhance visualization of the micromucosal and microcirculatory structure for a more detailed assessment of the early lesions[21].

Hence, in many institutions, especially in Japan, MCE or the ME-NBI technique has been extensively included in standardized procedure and is performed in addition to conventional white-light endoscopy prior to ESD or EMR[22]. For the colorectum, “Kudo’s Pit Pattern Classification” has begun to be widely adopted by many endoscopists because it appeared valuable in the histological prediction from the observation of five-types pit patterns by MCE or NBI-although the microvascular observation is helpful as well[23]. In the upper GI, despite numerous studies from investigators around the world and especially in some Asian countries, there is still no consistent classification diagnosis system for ME-NBI before the endoscopic removal of esophageal and gastric lesions; each medical institution tends to adopt its own classification[24-36]. Therefore, here we will comprehensively review the literature in recent years on the main characteristics of microsurface (MS) and microvascular patterns, introduce their classifications that have become relatively popular in some Asian countries under ME-NBI, describe the accurate relationship between them, the pathological diagnosis for early lesions in the upper GI tract, and their clinical utility in ESD or en bloc EMR. We do this to help build consensus on observation flowcharts of ME-NBI and to help endoscopists recognize tthe classification of early upper GI lesions more clearly so that they can select the most appropriate therapeutic intervention.

DEFINING OPTICAL IMAGING CHARACTERISTICS VISUALIZED UNDER MAGNIFYING ENDOSCOPY WITH NARROW BAND IMAGING IN UPPER GASTROINTESTINAL

In general, the doctor inspects the patient first under white-light endoscopy without magnification. He then slowly moves the scope, washes the tissue well, and pays special attention to areas containing slight differences. The key endoscopic finding by using white light (WL) has been reported to be a change of color (slight redness) and pallid mucosa[37]. However, the margin is difficult to identify by conventional WL. Then, the NBI model was employed to make it easier to detect the change in colors and structure of the mucosa. Moreover, with magnification, the microvascular (MV) pattern and MS pattern can be evaluated. So, what will be seen under ME-NBI if the cancerous lesion is suspected within the area?

Esophagus

Brownish area: A brownish area can often be recognized by NBI observation as distinct boundaries are formed between the tumor lesion and normal epithelium (Figure 1)[38]. An intraepithelial papillary capillary loop (IPCL) appears as brown dots under NBI-enhanced observation. For example, in the esophagus, if the lesion appears brownish under magnifying NBI observation, it will predict the possibility of mucosal squamous-cell carcinoma as a result of assessing the morphologic changes in the IPCL. The brownish areas in the esophagus visualized by NBI generally correspond to the Lugol chromoendoscopy displayed the lesions as unstained areas[39].

Figure 1 The carcinoma visualized in esophagus. A: Carcinoma in esophagus is difficult to identify by conventional white light; B: Carcinoma in esophagus can be easily recognized by narrow-band imaging (NBI) as well-demarcated brownish area; C: Intraepithelial papillary capillary loop can be observed by magnifying endoscopy with NBI at the edge of the tumor.

Intraepithelial papillary capillary loop: It is well known that angiogenesis plays a critical role in the transition from premalignant to malignant lesions. Consequently, early detection and diagnosis based on morphological changes to the microvessels are crucial[40]. Superficial blood vessels in the esophageal mucosa consist of branching vessels and IPCL. However, in some cases, only the former can be observed under the WL that extend to the horizontal plane and exist immediately above the muscularis mucosa while IPCL that rises perpendicularly from a branching vessel can be observed through ME-NBI (Figure 2)[41]. In these cases, Muto et al[43] have reported that a well-demarkated brownish area or an area of scattered brownish dots under NBI is connected with the proliferation of IPCL. This is a useful indicator for early esophageal squamous-cell carcinoma or high-grade intraepithelial neoplasia.

Figure 2 The superficial blood vessels in the squamous esophagus (from Inoue et al[42], with permission, making a little change for the original graph), the intraepithelial papillary capillary loop rises from the branching vessel and terminates in a diffuse.

Stomach

Besides the MV architecture, the imaging characteristics of the MS structure of mucosa the so-called pit or crypt patterns can be obtained by ME-NBI in the stomach (Figure 3).

Figure 3 Some typical microvascular and microsurface imaging characteristics visualized in stomach under magnifying endoscopy with narrow band imaging. A: Fine network pattern, mostly corresponding to well-differentiated adenocarcinoma (0-IIc, gastric); B: Corkscrew pattern, mostly corresponding to the poorly-differentiated adenocarcinoma (0-IIc, gastric); C: Intrastructural irregular vessel, enclosed in villous or papillary fine mucosal structure, had irregular shape characters such as dilation, heterogeneity, abrupt caliber or tortuousness (0-IIb, gastric); D: Regular white opaque substance (WOS), that shows well-organized and symmetrical distribution with a regular reticular pattern and obscures the subepithelial microvascular (MV) pattern (0-IIa adenoma, gastric); E: Irregular WOS, that is present within the cancerous epithelium with an irregular speckled pattern and makes the subepithelial MV pattern cannot be clearly visualized (0-IIa cancer, gastric); F: Light blue crest, defined as a fine, blue-white line on the crests of the epithelial surface in the gastric mucosa may be a distinctive endoscopic finding associated with the presence of histological intestinal metaplasia.

Subepithelial capillary network and collecting venule: By ME-NBI, the subepithelial capillary network (SECN) and the collecting venule (CV) can be clearly visualized. A polygonal-shaped subepithelial capillary loop surrounding each pit forms a network in a regular arrangement, and this capillary network drains into a CV. SECN and CV are basic anatomical components for analysis of the MV architecture. The SECN shows two distinct patterns depending on the region of the normal stomach being imaged: The body mucosa demonstrates a regular honeycomb-like SECN pattern with a CV, whereas the gastric antrum shows a coil-shaped SECN but the CVs are rarely observed. This might be because the CVs in the antral mucosa are relatively deeper from the surface epithelium than those of the gastric body mucosa[44].

For the abnormal stomach, there are two characteristics of MV that can be identified by ME-NBI: the first is a relatively regular “fine network pattern” (Figure 3A), which is more likely to be observed in well-differentiated adenocarcinoma and appears as mesh and abundant microvessels connected with each other; the second is a “corkscrew pattern” (Figure 3B) as with isolated and tortuous microvessels, which often represents the low density of MV and corresponds to poorly-differentiated, depressed (0-IIc), early gastric adenocarcinoma[45].

Intrastructural irregular vessel (ISIV) (Figure 3C) also has an irregular MV pattern but often appears in the superficial flat gastric lesion (0-IIb) as well as the marginal flat area of an elevated or a depressed lesion. This is a cancerous indication. Differing from the fine network pattern and corkscrew pattern shown in the areas where fine mucosal structure (FMS) disappear or are unclear in 0-IIc gastric lesions, the ISIVs are found enclosed in villous or papillary FMSs and have characteristics of dilation, heterogeneity, abrupt caliber or tortuousness of shape[46].

Microsurface: Applying ME-NBI is helpful for clearly visualizing not only some of the MV characteristics introduced above but also the gastric mucosal MS structures, namely pit or crypt opening patterns. The MS structures include the FMS in a normal stomach as well as the irregular or loss of pit pattern that occurs with early gastric carcinomatous lesions.

Although it is necessary to assess a neoplasm in the stomach by the MV and MS patterns simultaneously, it is sometimes impossible to visualize the subepithelial MV pattern on account of overcurtaining by the white opaque substance (WOS). In most adenomatous lesions, the WOS is frequently observed more clearly under ME-NBI than WL and is speculated to be some intracellular component within the neoplastic epithelium of the intervening part between the crypts, obscuring the morphology of the subepithelial MV and causing difficulty in assessing the MV pattern. In such cases, rather than assessing the MV pattern, the morphology of the WOS could be an alternative new optical microstructure sign for distinguishing adenomas from adenocarcinomas. Yao et al[47] reported that only about 6% of the WOS was found in IIb and IIc lesions. For 0-IIa type neoplasms, the WOS was more frequently visualized in adenomas (78%) than in carcinomas (43%) and showed a well-organized and symmetrical distribution of the dense WOS of a regular reticular/maze-like/speckled pattern (Regular WOS) (Figure 3D) within adenomas (100%), but showed a disorganized and asymmetrical distribution of the fine WOS of irregular reticular/speckled pattern (Irregular WOS) (Figure 3E) within carcinomas (83%). That is to say, the regular WOS is characteristic of adenomas, whereas its irregular distribution is characteristic for carcinomas.

Similar to the WOS, the light blue crest (LBC) (Figure 3F) is another characteristic optical microstructure under ME-NBI caused by the dense reflection of 400 to 430 nm short-wavelength light at the ciliated tissue. The LBC is defined as a fine, blue-white line on the crests of the epithelial surface/gyri, just at the edge of crypts. It has been suggested that the appearance of the LBC on the epithelial surface of the gastric mucosa may be a distinctive endoscopic finding associated with the presence of histological intestinal metaplasia in high sensitivity (89%), high specificity (93%), and high accuracy (91%)[48]. The LBC was also demonstrated to have a significant association with gastric atrophy and a high occurrence of gastric cancer[49].

As noted above, the strategies for diagnosing upper GI lesions by ME-NBI are specific to different organs. Under a magnifying endoscope, an esophageal neoplasia could be diagnosed solely according to the findings from the MV pattern, namely IPCL, because the esophageal squamous epithelium does not show FMS. In contrast, a gastric neoplasia could be diagnosed with the findings of the MV pattern as well as the MS pattern[50,51]. Of course, sometimes the WOS or LBC is more useful for the diagnosis.

CURRENT REPRESENTATIVE CLASSIFICATIONS OF MICROVASCULAR AND MICROSURFACE PATTERNS IN THE UPPER GASTROINTESTINAL UNDER MAGNIFYING ENDOSCOPY WITH NARROW BAND IMAGING

Classifications of intraepithelial papillary capillary loops in the esophagus

IPCLs beneath the basement membrane of the esophageal squamous epithelium can be observed by ME-NBI. It has been shown that identifying IPCL changes is very important in predicating early lesions of the esophagus. Regarding the classifications of IPCLs, there have been several systems adopted by different researchers[26-28], but in Japan, Inoue’s classification and Arima’s classification of IPCLs have been relatively popular.

Inoue’s classification of intraepithelial papillary capillary loop: Under NBI, the IPCLs are easily recognized as brown spots, and the normal patterns appear as a smooth-running, small-diameter capillary vessel in the normal epithelium. The abnormal shapes appear as four typical changes: Dilation, tortuous weaving, irregular caliber and form variation. Inoue et al [24,52] classed them into five types and several subtypes from type I to type V-N as below (Table 1 and Figure 4). IPCLs in type I is no different from the normal pattern. IPCLs in type II has one or two different characteristics: elongation and/or dilation is often seen. IPCLs in type III have no or few differences from the normal pattern, but this type differs from type I mainly in the features of color changes under NBI and iodine staining. Under NBI, the lesions of type I and type II often show no change or negligible change, but the types between type III and type V-N appear brownish. In addition, type I and type II lesions are often positively stained with iodine while the types from type III to type V-N are negatively stained. IPCLs in type IV appear to have two or three of the four abnormal characteristic changes. IPCLs in type V-1 demonstrate all the four typical changes. IPCLs in type V-2 are elongated on the base of the four shapes and only keeping part of the original IPCL. IPCLs in type V-3 are further degraded and run on a horizontal plane. As for type V-N, the most remarkable feature is the appearance of new tumor vessels.

Table 1 Inoue’s classification of intra-papillary capillary loop in esophagus.

Typing	IPCL	Iodine staining	Under NBI	Pathological assessing	Treatment
Type I	Smooth running small diameter capillary vessel with no difference from the normal pattern	Stained		Normal epithelium
Type II	Elongation and/or dilation capillary is often seen	Slightly stained		Esophagitis or re-generative tissue
Type III	No or minimal change from the normal	Unstained	Brownish	HGIEN	Further follow-up
Type IV	Showing two or three of four patterns among dilation, meandering, caliber changes and different shapes	Unstained	Brownish	HGIEN or m1 carcinoma in situ	ESD/en bloc EMR
Type V	Demonstrating all four characteristic changes: dilation, tortuous weaving, irregular caliber and form variation	Unstained	Brownish	M1 carcinoma in situ
Type VI	Elongation basing on the shapes of type V IPCL , keeping IPCL partly	Unstained	Brownish	M2 carcinoma in situ
Type VII	Destructing dramatically and running on horizontal plane	Unstained	Brownish	M3-Sm1deeper carcinoma	Relatively indicated for ESD/EMR
Type VIII	New tumor vessel appear	Unstained	Brownish	Sm2 deep carcinoma	Surgery, chemorado-therapy

IPCL: Intraepithelial papillary capillary loop; NBI: Narrow-band imaging; ESD: Endoscopic submucosal dissection; EMR: Endoscopic mucosal resection; HGIEN: High-grade intraepithelial neoplasia.

Figure 4 The case examples of Inoue’s intraepithelial papillary capillary loop classification from type I to type V-N. NBI: Narrow-band imaging.

According to the grade of the changes of IPCL, the depth of invasion can be assessed. Type I mainly appears in normal epithelium. Type II corresponds to inflammatory changes or regenerative tissue. Type III often reflects low-grade intraepithelial neoplasia. Type IV is linked to with high-grade intraepithelial neoplasia (HIN) or M1 carcinoma in situ. Type V-1 is definitively diagnosed as M1 carcinoma in situ. The appearance of Type V-2 strongly suggests m2 carcinoma. Type V-3 often indicates m3 to sm1 deep lesions. Type V-N is often associated with sm2 invasion cancer. In short, type I to type V-1 demonstrate the characterization for flat lesions while type V-1 to type V-N reflect invasive cancers.

With treatment, lesions of type III IPCLs need further follow-up, and type IV to type V-2 should be considered for ESD or en bloc ESD. Type V-3 lesions are thought to be an indication for ESD or EMR because of the depth of invasion ranges between m3 and sm1. A complete biopsy should be applied before deciding on a treatment strategy. For type V-N, it is taken for granted that the surgical treatment or chemoradiotherapy should be recommended to counteract the significantly increasing risk of lymph node metastasis.

Arima’s classification of intraepithelial papillary capillary loop: In 2005, Arima et al[25] reported another classification of the microvasculature of esophageal IPCLs under magnifying endoscopy. The microvascular patterns are categorized into four types (Figure 5). The thin, liner capillaries in subepithelial papillae are recognized as type I, resembling the shapes in normal mucosa. The vessels of type II become distended and dilated in subepithelial papillae, and the structure of capillaries is preserved. Most of them are usually found in lesions with inflammatory changes and are also associated with intraepithelial neoplasia. Spiral vessels with an irregular caliber and crushed vessels with red spots are characteristics of type III, which are often seen in m1 or m2 cancers. Type IV usually appears to be irregularly multilayered, irregularly branched, reticular vessels with an irregular caliber as generally observed in cancers with an m3 invasion or deeper. Avascular areas as well as stretched vessels are seen in cancers with downward growth. In addition, reticular vessels are commonly seen in poorly differentiated cancers, and the size of a vascular area surrounded by distended vessels is related to the depth of tumor invasion.

Figure 5 The morphology of Arima’s intraepithelial papillary capillary loop classification in esophagus. LIN: Low-grade intraepithelial neoplasia; HIN: High-grade intraepithelial neoplasia.

Comparing to the above two classification systems on the morphologic changes of IPCL and predicting the depth of the tumor invasion, it can be argued that type I of Arima’s classification partly corresponds to type I-type III of Inoue’s classification. Furthermore, type II of Arima’s classification partly corresponds to Inoue’s type IV, Arima’s type III partly to Inoue’s type V-1 or V-2, and Arima’s type IV partly to Inoue’s type V-3 or V-N. However, the two systems do not always have such clear corresponding links. The invasion depth diagnosis by Inoue’s classification is possible for most lesions, and the correct ratio is about 78%[24,52]. By contrast, when using Arima’s type III and type IV classifications as diagnostic criteria for HIN and cancers, the rate of differential diagnosis goes up to 99%[25]. Recently, it has been reported[53] that some flat areas are not able to be predicted by Inoue’s classification. However, combining the two classification systems could result in greater accuracy of the preoperative diagnosis, which is proved by the pathological diagnosis after ESD. Therefore, it is recommended for clinical endoscopists using Inoue’s classification and Arima’s classification together to make an invasion depth diagnosis of esophageal cancer under ME-NBI.

Stomach

As for the MS of the stomach, in 1978, Sakaki et al[54] described the gastric pit appearances under magnifying endoscopy and classified them into five types: (1) foveolar pattern; (2) foveo-intermediate pattern (FIP); (3) foveolo-sulciform pattern; (4) sulciform pattern; and (5) mesh pattern. Although “Sakaki’s classification” is still currently the most widely adopted classification by many Japanese endoscopists, not all gastric pathological changes can be expressed by this system because it is not consistent with structural changes under some pathological conditions[55], which were found to have round and long elliptical gastric pits. The width of the FIP band seems to be related to the severity of atrophic gastritis, and the FIP is considered to indicate the position of the atrophic border.

Therefore, in 2002, Yagi et al[56] first reported a new modified classification system named the “A-B classification system,” which is useful to describe typical micromucosal structures related to the development of Helicobacter pylori (H. pylori) gastritis. They classified the morphological changes in the glandular structure and microvascular architecture obtained by WL magnifying endoscopy into four types: (1)type Z-0: Gastric round pits resembling pinholes surrounded by a regular arrangement of collecting venules with SECN forming a network; (2) type Z-1: Irregular true capillaries but no collecting venules observed; (3) type Z-2: White gastric pits and sulci with neither collecting venules nor true capillaries being seen; and (4) type Z-3: Dilated pits with surrounding redness. Type Z-0 specifically indicated the H. pylori-negative mucosa and differed significantly from types Z-1, Z-2 and Z-3 with regard to the grade of inflammation, activity and presence of H. pylori.

More recently, with the development of brand new optical techniques, such as ME-NBI, which can clearly visualize not only the glandular structure but also the mucosal microvascular architecture in units as small as the capillary, the prior diagnostic classification system seemed less able to meet clinical needs, especially for early diagnosing of premalignant lesions and assessing the relationship between microvessel patterns, pit patterns and histological patterns ahead of endoscopic en bloc resection. In recent years, many researchers modified the above classifications but varied individually[45,57-61], and there is still no set of consistent classification guidelines. Nonetheless, the key characteristic findings of all the current classifications for ME-NBI with respect to early gastric carcinomatous lesions are based on the types of abnormal MV patterns and irregular MS patterns. Among these, the representative diagnostic system is advocated by Yagi et al[62], who established a flowchart for ME-NBI diagnosis in early gastric cancerous lesions as below: first, the “white zone” should be imaged, which is Yagi’s term for the border of the uniform or heterogeneous papillae in the mucosal MS structure that appears as a bold white line. Next, microvessels should be observed. A regular MV pattern means the microvessels appear regular in shape and arrangement and look like closed or open loops of uniform size caliber. An irregular MV pattern means the microvessels appear irregular in shape and arrangement, looking like tortuous or irregular branches of various sizes or abnormal caliber[47]. Then, according to the white zone, the MV pattern, the WOS, and the LBC, the histological imaging of entire mucosa should be done. (1) Fine network patterns and loop patterns are mostly associated with well- or moderately-differentiated adenocarinoma; (2) irregular MV patterns, namely ISIVs, enclosed in villous or papillary FMSs can often be observed in IIb gastric cancerous lesions; (3) corkscrew patterns or wavy microvessels mostly correspond to the poorly-differentiated adenocarcinoma; (4) regular WOSs often appear in IIa gastric adenoma lesions while irregular WOSs often present in IIa gastric cancerous lesions; and (5) LBC is mostly connected with intestinal metaplasia[47,48,63].

As a matter of course, with regard to the classification of early gastric lesions under ME-NBI, more in-depth studies are needed to address the more morphologically-complex microstructures of the stomach relative to the other parts of the GI system. Some features described previously are not general enough to apply to each lesion, and the number of cases in the studies is limited as well. At present, it is reasonable to use ME-NBI as a supplementary diagnostic tool to normal endoscopy with chromoend-oscopy in the stomach before deciding on therapy strategies. The current strategies require new additions and some modifications.

Barrett’s esophagus

BE is thought to be a complication of longstanding gastroesophageal reflux and a condition of the distal esophagus where normal squamous lining is replaced by columnar epithelium containing specialized intestinal metaplasia (SIM), which has the tremendous potential for developing esophageal adenocarcinoma with generally poor prognoses and a median survival rate of less than one year. Short BE is defined as < 3 cm and long BE as ≥ 3 cm[64].

Using ME-NBI allows clear visualization of micro-mucosal and vascular patterns in BE. Now, depending on which targeted biopsy technique can be performed, improved distinction of nondysplastic SIM from HIN is possible. Recently, several pieces of literature[16,33,65,66] have reported their own classification systems, of which the principal features are summarized as follows: SIM is characterized by the mixing of villous, tubular and linear patterns with mostly regular arrangements and having regular vascular patterns or appearing as long, branching vessels in a flat mucosa. In addition, absent microstructural patterns also have a very high correlation to and predictive power for SIM. HIN is characterized by irregular/disrupted microstructural and irregular microva-scular patterns, and the frequency of abnormalities shows a significant rise with increasing grades of dysplasia.

USEFULNESS OF MAGNIFYING ENDOSCOPY WITH NARROW BAND IMAGING PRIOR TO ENDOSCOPIC SUBMUCOSAL DISSECTION OR EN BLOC ENDOSCOPIC MUCOSAL RESECTION

To assess the differentiation and depth of invasion

Criteria for endoscopic submucosal dissection/endoscopic mucosal resection: Only some differentiation and invasion limited to sm1 lesions should be considered for endoscopic removal. Nowadays, in Asian countries, one of the widely adopted guidelines for ESD or en bloc EMR is that the histology of the tissues must be intramucosal, well-differentiated, early carcinoma, and the minute invasion of submucosal lesions must be limited to sm1-namely, with a depth less than 200 μm in the squamous epithelium of the esophagus and less than 500 μm in the stomach. If the lesion is recognized as undifferentiated, surgery should be recommended[6,67].

Japan’s data show that the five-year cancer-specific survival rates of EGC limited to the mucosa and submucosa are 99% and 96%, respectively[67]. In other words, en bloc endoscopic treatment should be mainly applied to some category 0 superficial GI neoplastic lesions with the invasion limited to the mucosa or submucosa. These are divided into three subtypes according to the “Paris classification”: 0-I include Ip and Is, referring to polypoid pedunculated and sessile respectively; 0-II are non-polypoid and non-exca vated, and they are further subdivided into 0-IIa for slightly elevated lesions, 0-IIb for completely flat lesions, and 0-IIc for slightly depressed lesions; 0-III are non-polypoid with an ulcer (Figure 6, left side)[68]. In order to get a more precise evaluation for choosing the appropriate therapy, endoscopists classify early GI cancer into the following subdivisions according to the depth of invasion: M1, carcinoma with questionable invasion carcinoma limited to the epithelium; m2, cancer invasion to the lamina propria; m3, cancer infiltration into the muscularis mucosa; sm1, to the upper third of submucosa; sm2, to the middle third; and sm3, to the lower third. (Figure 6, right side)[69]. The distribution of subtypes in category 0 differs in the esophagus and stomach. As an example, the respective proportions of subtypes 0-I and 0-IIc are 16% and 45% in the squamous epithelium of the esophagus, and they are 17% and 78% in the glandular epithelium of the stomach, respectively[70].

Figure 6 The Paris classification of early lesion of gastrointestinal tract (A) and the depth of tumor infiltration (B). ep: Epithelium; lmp: Lamina propria; mm: Muscularis mucosa; sm: Submucosa; mp:Muscularis propria.

Presently, the most critical factor in the decision of whether to perform ESD or en bloc EMR is the probability of unexpected lymph node metastasis. Studies have shown that early cancer without lymphovascular involvement could be cured by endoscopic removal. Intromucosal, moderately- or well-differentiated early carcinomas that have been proved do not have submucosal lymphovascular involvement. In contrast, poorly differentiated squamous-cell carcinoma, adenocarcinoma and/or signet-ring cell carcinoma have a high incidence of lymph node metastasis. M1 and m2 carcinomas have no metastasis, whereas less than 10% of m3 carcinomas and about 15%-20% of sm1 have lymph node metastasis. The risk increases to more than 50% of sm2 and sm3 carcinomas[7,67,71,72]. Therefore, before performing ESD or EMR, an accurate histological evaluation of the resected specimens is essential to avoid recurrence.

Magnified images obtained with the ME-NBI system could be a useful, non-invasive method of histologically predicting for early lesions in clinical practice, especially with regard to the IPCL pattern in the esophagus and MV and MS patterns in the stomach, based on which alone usually could help us perform a successful endoscopic therapy. Many researchers focused on the relations between the ME-NBI classifications categories with the characteristics of the histopathological types. For example, regarding the depth of superficial esophageal cancer, the accuracy rate of diagnosis is about 83.3%, according to the Inoue’s classification of IPCL[73]. And in the stomach, differentiated-type adenocarcinomas are mainly observed as fine-network patterns in about 15.7% of cases or loop patterns in about 83.8% of cases. Undifferentiated-type lesions are primarily characterized by the corkscrew pattern in approximately 58.8% of cases[57]. For HIN of BE, without the need for staining, the ME-NBI images have a sensitivity of 94% and a specificity of 76% as well as a positively predictive value of 64% and a negatively predictive value of 98%[65].

Comparing the diagnostic accuracy of ME-NBI and endoscopic ultrasonography (EUS) for estimating the depth of invasion of early cancers before removal, some endoscopists conclude that the overall accuracy of ME-NBI is a little higher than EUS, but the difference is not statistically significant. However, ME-NBI is at least as accurate as EUS for preoperative locoregional staging of early cancers. On the other hand, EUS can be used for observing lymph nodes, but the diagnostic capability of EUS for lymph nodes is less reliable, which can affect therapy-related decisions before ESD. Regarding this point, a consensus is still required. For some cases that are difficult to diagnose, it is even necessary to combine two stool tests with computed tomography before ESD or en bloc EMR[74-76].

To define the margin and size of involvement

Criteria for endoscopic submucosal dissection/endoscopic mucosal resection: For differentiated lesions, a size of ≤ 2 cm in diameter is an indication for EMR; a size of ≤ 3 cm of mucosal cancer with ulcers or sm1 submucosal cancers, and any size of mucosal cancer without ulcer are indication for ESD.

In Japan and a few other Asian countries, another current guideline for ESD or en bloc EMR regarding well-differentiated lesions is based on data relating the size of the early lesion and the rate of lymph node metastasis. For mucosal cancers with ulcers or sm1 submucosal cancers, lesions that are 3 cm or smaller present a negligible risk of venous or lymphatic involvement. These are indications for ESD. For larger lesions, surgery should be recommended. For lesions confined to the mucosa but without ulcers, the risk of lymph node metastasis is not affected by the size of the tumor, so there is no consensus on a maximal size, although circumferential lesions in the esophagus are usually avoided because of the potential for strictures. Because a 2 cm diameter is the upper limit for resection by EMR in one piece, if the lesions simultaneously meet the conditions of ESD and are not more than 2 cm large, these should also be reasonable indications for en bloc EMR treatment because this technique is easier than ESD[6,67,77,78].

Therefore, prior to endoscopic treatment, it is absolutely necessary to accurately identify the full lateral spread of the margins of the lesion, which leads to the determination of the lesion’s final size and contributes to the next step of making well-reasoned treatment decisions. In the upper GI, en bloc endoscpic removal needs to be carried out 2 mm outside the margin outlined by the spots. This is the key to ensuring that the complete R0 resection has a negative margin for the tumor cells and that the risk of local recurrence is reduced.

ME-NBI allows a more detailed observation of the mucosal changes of microstructures and microvessel patterns of GI carcinoma and is extremely useful, not only for identifying EGC itself, but also for differentiating the borders of cancerous tumors from background non-cancerous mucosa. By ME-NBI, the following points can help determine precise horizontal margins in clinical practice[44]: (1) recognize a demarcation line by the difference between an irregular MV or MS pattern and the surrounding regular normal mucosa. This has been proven to correspond to the tumor margins determined by histopathological examination; and (2) pay close attention to the areas disappearing from the regular SECN pattern as well as the appearance of an ISIV pattern. Sometimes, WOSs are helpful for identifying tumor margins that have not been determined. Also, LBC is a specific indicator for tumors derived from intestinal metaplasia by ME-NBI.

However, for IIb flat reddened lesions that have the same color as the surrounding normal mucosa, it is still occasionally difficult to detect the margins. On the other hand, accurate marking of tumors by ME-NBI also relies on an operator’s skill. Therefore, in order to improve the accuracy rate of marking margins, many endoscopists combine ME-NBI with conventional chromoendoscopy. For example, Lugol’s solution can dramatically outline the boundaries of a squamous cell esophageal cancer in the esophagus. Although one recent article has concluded that tumor margins can be identified more clearly by ME-NBI than by indigocarmine chromoendoscopy in the stomach[79], it is likely that in the majority of cases, a combination of these two methods prior to ESD or EMR will ensure there are no residual lesions.

To perform a target biopsy in real time

Before local endoscopic en bloc resection, the histopathologic diagnosis is very important for making therapy decisions. For a surveillance biopsy to detect early tumors, multiple random biopsies under conventional WL endoscopy are quite time-consuming and may miss a small lesion[80]. For example, for monitoring BE so far, the present recommended strategy is to perform random four-quadrant biopsies at every 2 cm. However, this approach is still prone to sampling errors, inconsistent histopathological interpretations, and delays in diagnosis[81].

Ultimately, the higher-accuracy pathological diagnosis as well as the ultrarapid in vivo diagnosis would be preferred in clinical practice[82]. It has been reported that chromoendoscopy could provide a good validity score for early cancer targeted biopsies[83]. However, it still has its limitations, including spending time lost in spraying and washing out the dye. Moreover, some dyes-such as methylene blue-might induce DNA damage in columnar cell-lined mucosa[84].

To this end, in recent years, many researchers suggest using the ME-NBI technique as an “optical biopsy” to better target biopsies in real time. Because this approach can provide better details of mucosa MV and MS patterns that significantly correlate with pathological diagnosis, it has the potential to reduce the need for histological examination of mucosal biopsy specimens[47,49,85,86]. Additionally, some endoscopists even think that ME-NBI can sometimes be substituted for a biopsy before endoscopic therapy because a biopsy might only focus on some suspected, poorly-differentiated lesions under magnifying endoscopy. However, to date, ME-NBI cannot always replace biopsies for histological assessment. In addition, ESD or en bloc EMR can supply specimens that are resected in one piece and provide more accurate histopathological diagnosis for determining whether the patient should receive an operation or other treatments[36,87-89].

CONCLUSION

In conclusion, ME-NBI is a very promising endoscopic technique that can clearly reveal detailed micromorphological differences corresponding to histology and provide some information about layer, origin, size, and extramural extension of GI early lesions. All of these benefits may augment the endoscopic R0 resection of early cancers in the GI tract and help guide targeted biopsies in the surveillance of certain high-risk conditions[19]. To some extent, ME-NBI has now become an indispensable tool in ultra-rapid in vivo diagnosis and immediate clinical decision-making, such as when performing ESD or EMR.

In this topic review, most representative references come from the experience of Japanese endoscopists because Japan remains the country with the most ESD cases reported around the world by far. Outside Asia, more recently, techniques such as magnification, NBI and ESD have been increasingly used although viewpoints differ between Eastern and Western cultures, especially regarding extending indications for ESD, the classifications of MV and MS under ME-NBI in the upper GI tract, and partly substituting EUS or biopsy with ME-NBI. However, current data is limited, and we would need long-term outcome data to unify some assessments in order to conduct multicenter trials to develop clear, internationally accepted classification systems. This system review was intended to make a small contribution to some of the aforementioned debates.

Additionally, besides ME-NBI, it is necessary to combine various endoscopic techniques including EUS and chromoendoscopy in some difficult cases before en bloc endoscopic resection. It is important to emphasize here that the first step should always be to look carefully for the suspected area by conventional WL endoscopy before switching to the ME-NBI model.

ACKNOWLEDGEMENTS

The authors thank for Professor Inoue H from Showa University Northern Yokohama Hospital, Japan, for his help in supplying the image of superficial blood vessels in the squamous esophagus (Figure 2) in this paper.