Factors influencing lateral margin diagnosis challenges in Barrett’s esophageal cancer: a bicenter retrospective study in Japan

Article information

Clin Endosc. 2024;.ce.2024.068

Publication date (electronic) : 2024 November 11

doi : https://doi.org/10.5946/ce.2024.068

Ippei Tanaka^,1,2, Shuhei Unno³, Kazuki Yamamoto², Yoshitaka Nawata¹, Kimihiro Igarashi¹, Tomoki Matsuda¹, Dai Hirasawa¹

¹Department of Gastroenterology, Sendai Kousei Hospital, Miyagi, Japan

²Digestive Diseases Center, Showa University Koto Toyosu Hospital, Tokyo, Japan

³Department of Gastroenterology, Seirei Hamamatsu General Hospital, Shizuoka, Japan

Correspondence: Ippei Tanaka Digestive Diseases Center, Showa University Koto Toyosu Hospital, 5-1-38 Toyosu, Koto-ku, Tokyo 135-8577, Japan E-mail: ippeitanaka777@gmail.com

Received 2024 March 28; Revised 2024 May 27; Accepted 2024 June 4.

Abstract

Graphical abstract

INTRODUCTION

The incidence of Barrett’s esophageal adenocarcinoma (BEA) has rapidly increased in Western countries over the past few decades, accounting for more than 50% of all esophageal cancers.1,2 The main reasons for this increase are the rising prevalence of gastroesophageal reflux disease (GERD) and obesity, coupled with the declining prevalence of Helicobacter pylori infection.3-5 While squamous cell carcinoma of the esophagus remains predominant in East Asia, including Japan, the incidence of BEA has shown an increasing trend due to increased efforts in GERD and H. pylori eradication.6,7 The incidence of BEA per 100,000 person-years in a certain prefecture in Japan was reported to have increased from 0.2 to 0.8 over a 30-year period between 1885 and 2014, and there is a possibility of a further increase in the future.8

Recent advances in endoscopic technology have increased the opportunities for early stage esophageal cancer detection, leading to more patients being treated with endoscopy, the primary therapeutic approach for early stage cancer.9,10 In endoscopic treatment, en bloc resection is essential to avoid piecemeal resection, as the latter is considered a risk factor for local recurrence.11 The key to en bloc resection is to accurately diagnose the lateral margin of the cancer.

BEA, especially when it occurs in the long-segment Barrett’s esophagus (LSBE), often presents challenges in determining the lateral margin of the lesion because of chronic inflammation.12 However, the characteristics of cases that are difficult to diagnose in terms of the lateral margin have not yet been clearly elucidated. Therefore, this study aimed to clarify the clinical and pathological features and underlying causes of BEA with unclear demarcation.

METHODS

Study design and patients

This retrospective study was conducted at two institutions in Japan, Sendai Kousei Hospital and Seirei Hamamatsu General Hospital. Data were collected from patients diagnosed with BEA who underwent endoscopic submucosal dissection (ESD) or surgery between January 2010 and August 2022. The inclusion criteria were diagnosis of BEA based on histopathological and endoscopic examinations. The exclusion criteria included lesions with suspected invasion into or beyond muscle layer after diagnostic work-up (≥T2 carcinoma), lesions completely covered by squamous mucosa, and insufficient endoscopic images suitable for review (e.g., those with blood, mucus, or out of focus images). A flow diagram of the patient enrollment process is illustrated in Figure 1.

Fig. 1.

Flow diagram of the patient enrollment process. ER, endoscopic resection; BEA, Barrett’s esophageal adenocarcinoma.

Definition of tumor group classification

All endoscopic images of BEA included in this study were retrospectively reviewed. Two expert endoscopists, who had experience with more than 100 cases of endoscopic treatment for esophageal cancer, reevaluated all images and classified them based on magnifying narrow-band imaging (M-NBI) findings. Lesions with clearly recognizable circumferential demarcations on M-NBI images were categorized into the clear demarcation group (CD group). Conversely, lesions in which the demarcation was unclear, even in a part of the lesion, were classified into the unclear demarcation group (UD group). Figure 2 shows the typical images of lesions in the CD and UD groups. Furthermore, this study aimed to focus on diagnosing demarcation between cancerous and non-cancerous areas within the glandular epithelium. Thus, the diagnosis of the lateral margin of lesions was strictly limited to the glandular epithelium, excluding the area adjacent to the squamous epithelium.

Fig. 2.

(A) Endoscopic image of BEA in the CD group. (B) Magnifying endoscopy with narrow-band imaging show clear demarcation of the lesion. (C) Endoscopic image of BEA in the UD group. (D) Magnifying endoscopy with narrow-band imaging show mild irregularity of the lesion and unclear demarcation of the lesion. BEA, Barrett’s esophageal adenocarcinoma; CD, clear demarcation; UD, unclear demarcation.

Pathological evaluation

All specimens were fixed in formalin, embedded in paraffin, and stained with hematoxylin and eosin. Pathological evaluation was performed by two expert pathologists specialized in esophageal cancer diagnosis. Histological diagnoses including macroscopic type, tumor size, tumor differentiation, depth of invasion, lymphovascular invasion, and positive vertical and horizontal resection margins (vertical margin [VM] and horizontal margin [HM]) were assessed according to the guidelines for the management of esophageal cancer.13

In this study, pathological structures were meticulously evaluated. First, a slide displaying the demarcation of the lesion was selected as a representative section. Subsequently, the demarcation area was examined (Fig. 3A). In both cancerous and non-cancerous areas, the following parameters were measured in the five glandular ducts: length and width of the foveolar structure and the width of marginal crypt epithelium (MCE) (Fig. 3B). The average values of these measurements in five glandular ducts were calculated to ensure accuracy, as a precise vertical section is not always obtained.

Fig. 3.

Evaluation of the pathological structures. (A) A representative slide displaying the demarcation of the lesion. Black arrow indicates the demarcation between cancerous and non-cancerous area (hematoxylin and eosin stain, ×40). (B) This figure specifically illustrates the length and width of foveolar structures and the width of marginal crypt epithelium.

RESULTS

Clinicopathological features of BEA lesions and patients’ background

The study included 67 (58 male and 9 female) patients with 68 lesions. Patients’ characteristics are summarized in Table 1. The median age of the patients was 72 years (IQR, 66–80 years). ESD was performed for 63 lesions (92.6%), whereas surgical intervention was performed for five lesions (7.4%). PPIs were administered to 28 (41.8%) patients. Hiatal hernia and reflux esophagitis were observed in 57 (85.1%) and 13 (19.4%) patients, respectively. Analysis of the background mucosa revealed the prevalence of SSBE in 45 (67.2%) patients and LSBE in 22 (32.8%) patients.

Table 1.

Patient characteristics

The clinicopathological features of patients with BEA are presented in Table 2. The macroscopic types consisted of 35 protruding (51.5%), 26 depressed (38.2%), and seven flat (10.3%) lesions. Regarding the depth of invasion, 58 lesions were diagnosed as mucosal (85.3%) and 10 as submucosal (14.7%). Lymphovascular invasion was observed in 12 (17.6%) patients, with six patients showing positive HM (8.8%) and two showing positive VM (2.9%).

Table 2.

Lesion characteristics

Risk factors for unclear demarcation line

Univariate analysis revealed LSBE (odds ratio [OR], 8.44; 95% confidence interval [CI], 2.64–27.00; p<0.001) and macroscopic type 0–Ⅱb lesions (OR, 8.43; 95% CI, 1.36–52.06; p=0.046) as significant risk factors for the UD group. Multivariate analysis was performed including these two factors, which revealed LSBE as the sole significant risk factor in the UD group (OR, 12.17; 95% CI, 2.84–47.6; p=0.001) (Table 3).

Table 3.

Multivariate analysis for the risk factor of Barrett’s esophageal adenocarcinoma with unclear demarcation

Evaluation of pathological structures between cancerous and surrounding mucosa

To elucidate the causes of indistinct demarcation, we conducted a detailed assessment of the pathological foveolar structures in both cancerous and background mucosa (Table 4). In the CD group, the length and width of the foveolar structures in the cancerous mucosa was significantly shorter than that in background mucosa (135 µm vs. 190 µm and 75 µm vs. 131 µm, p=0.03 and p<0.001). In contrast, these measurements did not show significant differences in the UD group (149 µm vs. 118 µm and 90 µm vs. 130 µm, p=0.77 and p=0.06). However, in both CD and UD groups, the width of MCE in cancerous area was significantly shorter than that in the background mucosa (CD group; 24 µm vs. 40 µm and 26 µm vs. 34 µm, p<0.05 and p<0.05). Figure 4 shows the difference in the width of MCE between cancerous and non-cancerous areas endoscopically and pathologically in the UD group.

Table 4.

The evaluation of pathological structures

Fig. 4.

A representative case from the unclear demarcation group. (A) The reddish depressed lesion are seen at the anterior side of the esophagogastric junction. (B) Magnifying endoscopy with narrow-band imaging show mild irregular mucosal pattern, and the demarcation of the lesion is unclear. However, there is a distinct difference in the width of marginal crypt epithelium (MCE) between the mucosa on the upper and lower part of this image; the MCE appears narrower on the upper part and wider on the lower, which corresponds to the cancerous mucosa and non-cancerous mucosa, respectively. (C) Pathological findings reveal significant difference in the width of MCE between cancerous and non-cancerous area. Black arrows indicate the edge of MCE, and thus the area sandwiched by black arrows shows the width of MCE.

DISCUSSION

This study revealed three important findings. First, LSBE was identified as a risk factor for unclear demarcation in BEA. Second, in the UD group, no significant differences in the length and width of the foveolar structures was observed between cancerous and background mucosa, suggesting that similarity in foveolar structures contribute to unclear demarcation. Third, the width of MCE in cancerous mucosa was significantly shorter than that in background mucosa, even in the UD group, indicating its potential as an important diagnostic tool.

In 2017, the Japan Esophageal Society introduced a magnified endoscopic classification of Barrett's esophageal neoplasms.14 This classification system has gained widespread use in Japan owing to its simplicity and high diagnostic accuracy. One notable feature of this classification is that, unlike magnifying endoscopy simple diagnostic algorithm for early gastric cancer (MESDA-G), which is a magnified endoscopic classification for gastric cancer, it does not require a clear demarcation of lesions.15 This suggests the a certain number of BEA cases exhibit unclear demarcations. Unclear demarcations in BEA are frequently encountered in clinical practice. In our study, approximately 30% of all BEA cases exhibited unclear demarcations. Therefore, it is important to elucidate the characteristics of BEA with unclear demarcations.

This study revealed that LSBE is a risk factor for BEA with unclear demarcation. Probst et al.12 reported that the R0 resection rate for BEA was significantly lower in LSBE than that in SSBE (70.4% vs. 90.0%, p=0.029), a finding that aligns with our results. The higher incidence of elevated-type BEA in SSBE and flat or depressed lesions in LSBE may have influenced the results.16,17 However, we believe the more significant cause is related to the nature of Barrett’s mucosa. Barrett’s mucosa is an epithelium that develops in the esophagus due to chronic reflux of gastric acid juice. Consequently, it is often accompanied by chronic inflammatory changes, particularly evident in LSBE. Inflamed mucosa occasionally shows mild atypia, making diagnosis of the lateral margin of BEA extremely difficult.18 However, some previous studies have reported that R0 resection rates in ESD cases were comparable between SSBE and LSBE.19,20 This could be attributed to the practice of setting wider margins during ESD procedures, even when the actual demarcation is unclear, thereby facilitating R0 resection. In the present study, despite the high prevalence of BEA with unclear demarcations, setting a wider margin around the lesion appeared to result in a lower percentage of positive horizontal margins.

The causes of unclear demarcation were revealed in this study by assessing the pathological structures. In BEA with clear demarcation, significant differences in the length and width of the foveolar structures were observed between cancerous and background mucosa. However, in BEA with unclear demarcations, these structures exhibited nearly identical patterns in cancerous and background mucosa. This similarity in pathological foveolar structures is considered a factor contributing to the difficulty in diagnosing demarcations. As previously mentioned, it is possible that background non-neoplastic mucosa is accompanied by atypical changes due to chronic inflammation.18 Alternatively, this phenomenon may result from the development of low-grade dysplasia. Low-grade dysplasia and atypical mucosa can arise not only in Barrett's mucosa but also in ulcerative colitis (UC), which is characterized by chronic inflammatory changes. A previous study suggested difficulties in endoscopic recognition of tumors that develop in UC.21 Considering these findings, the results of the present study can be considered reasonable.

Through the examination of pathological structures, another significant finding emerged, which could potentially be deemed as one of the most important findings of this study. Regardless of whether the lesions were well demarcated or indistinct, MCE was noticeably narrower in cancerous mucosa than in non-neoplastic surrounding mucosa. This finding was confirmed in the present study. MCE is a term that holds familiarity within the field of pathology and endoscopy, particularly in Japan, where the endoscopic diagnosis of gastric cancer has become prevalent.22 It is sometimes endoscopically referred to as the “white zone” denoting a white, band-like area surrounding the foveolus or duct.23 Yagi et al.23 also reported that the “white zone” observed endoscopically corresponds to the MCE observed pathologically. According to our results, the width of MCE could imply the presence of cancer when narrow or suggest non-neoplastic lesion when wide. Thus, “short MCE” may be an important endoscopic finding to diagnose BEA. To utilize this finding in clinical practice, the following approach can be applied. initially, when suspicious lesions indicating cancer are identified, follow the algorithm of magnified endoscopic classification for BEA and assess mucosal and vascular regularity or irregularity to diagnose the lateral margin of the lesion.14 However, in cases where it is challenging to diagnose an area as cancerous or non-cancerous based on the criteria, “short MCE” can be a valuable asset. A narrow width of MCE could be indicative of cancer, whereas a wider width could be non-neoplastic. This new approach, combined with the Japanese BEA classification (Fig. 5) can potentially lead to a more accurate diagnosis of Barrett’s esophageal lesions. Therefore, a prospective validation study is required to confirm these findings.

Fig. 5.

Newly proposed magnified endoscopic classification of Barrett's esophageal neoplasms, incorporating the results of this study. GTV, green thick vessels; m-FP, modified flat pattern; MCE, marginal crypt epithelium.

This is the first study to elucidate the cause of unclear margins in BEA based on the evaluation of pathological structures.

However, this study has several limitations. First, this was a retrospective study with a relatively small sample size, particularly for LSBE cases. Second, this study did not include patients diagnosed with low-grade dysplasia. In Asia, especially Japan, pathologists are not accustomed to diagnosing low-grade dysplasia, and even the official guidelines for esophageal cancer management do not specify the diagnosis of low-grade dysplasia.13 We hope that future research will address the diagnostic disparities surrounding dysplasia in Japan and Western countries. Third, the distinction between well-defined and indistinct demarcations was determined by two endoscopy specialists, which may have introduced subjectivity into the assessment. Fourth, a one-to-one comparison between the endoscopic images and pathological findings was not performed in this study.

In conclusion, careful attention should be paid to diagnose the lateral margin of BEA arising from the LSBE. Furthermore, the width of MCE, which is endoscopically recognized as the “white zone,” could potentially serve as a significant factor in the diagnosis of the lateral margins of BEA. Further prospective studies are necessary to assess whether the width of MCE is useful in the diagnosis of the lateral margin of BEA in clinical practice.

Article information Continued