First reported by Norman Barrett, Barrett’s esophagus (BE) can be simply defined as the presence of columnar epithelium in the esophagus (1). Until 1976, Paull et al. (2) classified the presence of columnar epithelium within the esophagus based on histiological subtypes: Gastric fundic type, junctional type, and specialized type with intestinal metaplasia (IM).

IM refers to the replacement of normal mucosal epithelium with the intestinal epithelium outside the intestine, implying the transformation of a well-differentiated and mature tissue into another differentiated mature tissue under abnormal conditions. IM is common in the stomach and esophagus, but it can also occur in the gallbladder, bile ducts, uterus, bladder, pelvis, ureter, and urethra (3–9).

In recent years, there has been much controversy regarding the role of IM in the diagnosis of BE. We hereby summarize the significance of IM and the evidence supporting the fact that the presence of IM is either mandatory or not to establish a diagnosis of BE. IM in BE is often missed, and several other cells can mimic goblet cells. We cover the classification, manifestation and detection rate of IM under different endoscopic techniques, and methods to accurately identify IM in BE based on endoscopic and pathologic findings. Finally, we also discuss the principles regarding the monitoring and management of BE with and without IM.

IM is considered a precancerous lesion and is closely related to the occurrence of cancer. According to the study of Watanabe et al (10), the presence of IM is associated with the degree of severity of the Barrett’s mucosa, regardless of the size of the tumor. Therefore, they speculated that IM might be an epiphenomenon of BE extension rather than a reflection of the tumor’s origin. However, differentiated mature intestinal epithelial cells can acquire additional mutations, then develop into dysplasia and cancer (11). There is also evidence that IM is associated with a higher frequency of cancer-related mutations compared with nongoblet cell metaplasia (12). Similar to the significance of IM in the stomach (13), although there was evidence that tumor cells may not directly originate from goblet cells (11, 14). It has been undoubtedly suggested that the BE’s mucosa is more prone to neoplastic transformation. In the replacement of the normal esophageal stratified squamous epithelium with columnar epithelium, IM is the only type that is clearly prone to malignant transformation (15).

The possible etiologies for BE include the transcommitment of stem cells or progenitor cells, transdifferentiation of differentiated cells, the expansion of residual embryonic cells located at the squamous-columnar junction (SCJ), and the differentiation of circulating bone marrow cells (16–20). It was not until 2017 that Jiang et al. (21) simulated a BE mouse model with real goblet cells. They described a novel transitional columnar epithelium with distinct basal progenitor cells (p63 + KRT5 + KRT7 +) at the SCJ. After the expression of the caudal-type homeobox transcription factor-2 (CDX2), these progenitor cells were differentiated into intestinal-like epithelium, including goblet cells, resulting in BE with IM. Moreover, such epithelium was found to be localized at the subject’s gastroesophageal junction (GEJ). Moreover, Jin et al. (11) proposed the SPEM model of esophageal IM to explain the multiformity of the stem cells involved in esophageal IM. The differentiated cells and stem cells in this model could be transformed into SPEM-like cells (TFF2+ MUC6+); persistent inflammatory injury eventually led to IM, dysplasia, and carcinogenesis of the SPEM.

Replacement of the distal esophageal squamous epithelium by metaplastic columnar epithelium forms the pathological basis for BE. Whether IM must be present or whether the length of the columnar mucosa needs to be larger than 1 cm are the main controversial points. The American Gastroenterological Association (AGA), American College of Gastroenterology (ACG), and European Society of Gastrointestinal Endoscopy (ESGE) all believe that IM is necessary for the diagnosis of BE, while the British Society of Gastroenterology (BSG), Asia-Pacific Working Group (APWG) and Benign Barrett’s and Cancer Taskforce consensus group (BOB CAT) all reckon that IM is not needed for the diagnosis of BE (15, 22–26). However, BOB CAT also stressed that the existence of IM should be noted when diagnosing BE (23) ( Table 1 ).

The main basis supporting the fact that IM is mandatory for the diagnosis of BE includes the following: a large number of population-based cohort studies have shown that the risk of esophageal adenocarcinoma is much lower in patients with columnar metaplasia without IM compared to those with IM (27). Compared with columnar metaplasia without goblet cells, tumorigenesis is most commonly seen in columnar metaplasia with goblet cells (12, 28). In-depth evaluation of esophageal biopsy specimens before EMR and esophageal resection specimens after EMR found that IM was found in the columnar esophagus of all the patients with adenocarcinoma (29). Furthermore, some studies have demonstrated that it might be unwise to diagnose a disease that has a negative impact on insurance status and quality of life in non-IM-CLE (Columnar metaplasia without IM) patients (30, 31). The use of IM as a prerequisite for the diagnosis of BE was conducive to the adoption of a more cost-effective approach to the care of BE patients (26).

Evidence supporting the fact that IM is not necessary for diagnosing BE includes the following: a single endoscopic examination and a small number of biopsy samples were not sufficient to rule out IM (32, 33). In the study of Jankowski et al. (34), all the patients with BE who did not have IM at the beginning of the study developed IM during follow-up. Related retrospective studies have reported that patients with or without IM had a similar risk of developing cancer (35). Studies carried out on columnar mucosa resected endoscopically in early cancer subjects have confirmed that cancer may also occur in non-IM columnar epithelium (36). There was also evidence that columnar epithelium without goblet cells may contain similar molecular abnormalities to columnar epithelium with goblet cells (28, 37–39). Lavery et al. (40) reconstructed the cloned ancestor of EAC and provided direct genetic evidence for the malignant tendency of metaplastic columnar epithelium without goblet cells. This is the first direct demonstration that the clonal expansion and precancerous progression of BE are not limited to the metaplastic columnar epithelium with goblet cells.

With or without IM, it was uncertain whether columnar mucosa with a length smaller than 1 cm at the GEJ would progress to cancer. Current research has also depicted that the rate of carcinogenesis in this group of patients is significantly lower than in those with BE (Columnar mucosa more than 1 cm). Thus, most guidelines explicitly require that the diagnosis of BE must include a columnar mucosa exceeding 1 cm at the GEJ (22, 24–26) ( Table 2 ). Although BOB CAT did not specify that the columnar mucosa at the GEJ must be more than 1 cm to establish a diagnosis of BE. BOB CAT also indicated that an irregular and ≤ 1 cm lesion at the GEJ may be a natural phenomenon (26). Besides, according to a recent prospective multicenter cohort study, patients with BE lesions smaller than 1 cm will not develop high-grade dysplasia (HGD) or esophageal cancer within 5 years of endoscopic examination (41).

In Japan, the 1 cm criteria regarding the morphological changes of BE’s mucous membrane is not mandatory, and such lesions are called ultra-short-segment BE (USSBE). The true meaning of cardiac cancer refers to cancer that occurs in the intestinal metaplastic area of the anatomical cardia or esophagogastric junction (42). By definition, the anatomical position of the cardia and USSBE basically coincided. IM can occur in the cardia and USSBE, and the cardiac mucosa with IM is considered a precancerous lesion of cardiac carcinoma. Over the past few years, with the increasing incidence of cardiac cancer, more attention has been paid to USSBE. Nonetheless, due to the above reasons, the concept of USSBE is not widely used and accepted. In a large sample study in 2016, the annual cancer rate of USSBE was only 0.01% (compared with 0.22% and 0.03% for the long and short segments of BE, respectively) (43). A related etiological investigation also pointed out that BE lesion’s length exceeding 1 cm was not the defining factor (44). However, IM is not required in the definition of USSBE in these studies. If we investigate USSBE with IM exclusively, there may be new results. Monitoring may be crucial for these patients’ prognosis, albeit further research is required to confirm this assumption. Given the high incidence of USSBE in the population, this issue should be taken into account when calculating and comparing the incidence of BE.

Conventional endoscopy has not been satisfactory in the diagnosis of BE, especially in the detection of precancerous BE lesions. Biopsy of BE lesions is routinely performed from four quadrants superior to the GEJ at intervals of 1-2 cm. According to previous studies, only 10–79% of doctors take biopsy specimens according to this protocol. The higher the length of BE lesions, the lower the application rate of the four-quadrant biopsy principle (45–47). Also, obtaining multiple biopsy specimens and pathological interpretation both imply a significant financial burden (48). BE with IM is a precancerous lesion, which has more significant clinical significance (49). Relevant investigations conducted in Japan suggested that similar to the most common site of occurrence of esophageal adenocarcinoma (EAC), IM often appeared in the 0–3 o’clock area (50). This finding indicates that this area should be the focus of attention and that biopsy specimens taken from this area may increase the detection rate of IM. With the advancement of endoscopic techniques, there are more means for endoscopic detection of IM ( Table 3 , Figure 1 ).

The presence of goblet cells in the CLM is the pathological diagnosis of IM in BE. However, some non-goblet columnar epithelial cells can mimic goblet cells and impair the diagnosis of IM. Panarelli et al. (77)delineated the characteristics of these cells and their differences from goblet cells. Damaged foveolar epithelial cells containing a copious amount of cytoplasmic mucin can mimic the globular structure of goblet cells and are therefore known as “pseudo-goblet cells,” which can be distinguished from goblet cells due to their pink cytoplasm. A foveal cell that produces enough acid mucin to cause the blue discoloration of the cytoplasm can mimic the blue tone of goblet cells and become a “columnar blue cell.” These cells are cylindrical and can be distinguished from goblet cells due to their difference in shape. The multilayered epithelium is composed of immature-appearing squamoid cells and superficial clusters of columnar cells containing acid mucin. The multilayered epithelium can also simulate goblet cells. Furthermore, all three cell types simulating goblet cells continually fill the surface epithelium diffusely. In contrast, goblet cells are typically individually dispersed in the foveal epithelium background, which is a distinctive feature.

Immunohistochemical staining is commonplace nowadays. Alkaline blue, high iron diamine, Alcian blue, periodic acid Schiff (PAS), CDX2, MUC 2, Heppar1, etc. can all stain goblet cells. However, numerous studies have shown that these types of stains can also stain non-goblet cells, entailing a decrease in specificity (78–86). Although the expression of some immuno-labels in non-goblet columnar epithelium may suggest that IM is more likely to be found elsewhere, it is still not possible to identify goblet cells and non-goblet cells. As a result, the relationship between the staining characteristics of non-goblet epithelium and the risk of dysplasia and/or cancer remains unclear as of yet (78).

Besides, Maskacci et al. (87) demonstrated the combined use of descriptive endoscopy, the suspected esophageal metaplasia range of the Prague standard, and diagnostic charts all conveniently improved the consistency of the CLE interpretation of the esophageal biopsy, which resulted in improved consistency in the diagnosis of IM.

Although there is a difference in the need for IM in BE diagnosis, only some of the guidelines are specified in the relationship between BE monitoring and treatment strategies and IM ( Table 2 ). To optimize the detection rate of IM, ACG (24) suggested that patients who were suspected of having BE (but lacked IM) should undergo a repeat endoscopy within 1 to 2 years, patients with suspected BE should have at least eight random biopsies, patients having short (1–2 cm) segments of suspected BE in which eight biopsies may be unfeasible should have at least four biopsies per cm of circumferential BE, and one biopsy per cm in tongues with BE. BSG (22) suggested that surveillance should consider the presence of IM and the length of BE. For BE, without IM or dysplasia and BE shorter than 3 cm, repeated endoscopic biopsies are recommended to confirm the diagnosis, and if no IM is confirmed repeatedly, de-surveillance is indicated. For patients with BE shorter than 3 cm and accompanied with IM, endoscopic monitoring should be performed every 3–5 years. For patients with BE ≥ 3 cm, endoscopy should be performed every 2–3 years.

The treatment of BE mainly includes drug therapy, anti-reflux surgery, esophagectomy, endoscopic treatment, etc. The main advantage of drug treatment and anti-reflux surgery is that they can alleviate the symptoms of reflux. For simple BE, with or without IM, drug therapy is recommended (23, 24), mainly to bring reflux symptoms under control. If the reflux symptoms are severe and the curative effect of drugs is lacking, anti-reflux surgery can also be considered. Routine endoscopic treatment of BE without dysplasia is not recommended unless the length of BE exceeds 10 cm (26) or is accompanied by nodules (24).

Some breakthroughs in the drug therapy of BE have been made. Initial studies suggest that proton pump inhibitors (PPIs) have a protective effect on the malignant progression of BE (87–89). However, a case-control study accompanied with the follow-up of 9883 patients with BE for up to 10 years showed that the use of PPIs increased the risk of developing EAC or HGD (90). But, a recent large-scale randomized controlled trial of patients with BE showed that high doses of PPIs protect against a composite endpoint of all-cause mortality, EAC, and HGD (34). The study also showed a coupled treatment of PPIs and Aspirin was more effective than PPIs alone, and high doses of PPI (40 mg twice daily) combined with aspirin were the best combination. Aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) have been reported to reduce the risk of malignant progression of BE, and studies have shown that aspirin is more effective than nonsteroidal anti-inflammatory drugs in reducing the risk of malignant progression of BE (69). Recently, Huo et al. (91) provided a theoretical basis for this. Their work showed that aspirin exerts a unique anti-BE protective effect by acting on IKKβ, inhibiting the action of the nuclear factor-k-gene binding (NF-kB) pathway and inhibiting the expression of CDX2. Also, some auxiliary drugs that are often used in the treatment of gastroesophageal reflux disease (GERD), including medicines that promote gastric motility, neutralize bile acids, and inhibit transient relaxation of the lower esophageal sphincter, which reduces the progression of GERD to BE at varying degrees (92–96) ( Table 4 ).

Surgery is still the primary method employed in anti-reflux treatment. Furthermore, laparoscopy gave rise to numerous surgical techniques. The surgical methods mainly include gastric fundus folding, magnetic sphincter augmentation (MSA), electrical stimulation of the lower esophageal sphincter (LES), and bariatric surgery. These procedures all carry merits and shortcomings, but the difficulty of the operation and related complications remain a matter of concern (97–104) ( Table 5 ).

The field of endoscopy is constantly evolving. Examples of endoscopic anti-reflux surgery include trans-oral incision-free fundoplication (TIF), radiofrequency ablation (Stretta), anti-reflux mucosectomy (ARMS), and endoscopic injection or implantation (105–112) ( Table 6 ).

Generally, surgery or endoscopic treatment is only indicated in BE that is more than 10 cm in length or is accompanied by nodules, intraepithelial neoplasia, or cancer (24, 26). Esophagectomy can achieve complete eradication, but the quality of life is seriously impaired, and its mortality rate is high. There are many kinds of endoscopic therapy, including endoscopic mucosal resection (EMR), endoscopic submucosal dissection (ESD), radiofrequency ablation (RFA), argon plasma coagulation (APC), contact cryo-balloon focal ablation system (CbFAS), and so on ( Table 7 ). Compared with other methods, endoscopic treatment possesses the merits of high safety, conserving the normal function of the esophagus and eradicating lesions simultaneously.

Patients suffering from esophageal intramucosal cancer who have undergone EMR and esophagectomy had a similar long-term recurrence rate and mortality (118). RFA is a mature endoscopic technique for the eradication of IM with dysplasia (116). EMR and RFA are usually combined in the treatment of early tumors of BE. According to a recent study (113), the focal EMR followed by RFA has a similar efficiency level as EMR but is safer than EMR in patients with HGD/EAC. However, due to the decreased access to specimens, the uncertainty of the damage to the anatomical layer, and high costs, RFA is not commonplace in Asia. For larger lesions, EMR can only be segmented and resected multiple times. The Japanese view is that the single resection of ESD is the preferred choice regardless of the size of the lesion due to the blind resection line in patients suffering from adenocarcinoma (119). However, due to the long-term learning curve and the high rate of adverse events, the use of ESD in the West is still limited. A study has shown that the complete remission rate from neoplasia of ESD is the same as that of EMR at 3 months. ESD can potentially cause severe adverse events (two cases of perforation) because of its long operation time (114). However, according to the results of a multicenter study from the West, ESD is effective and safe and can achieve a good level of proficiency after approximately 30 operations (120). A recent meta-analysis also shows its effectiveness and safety (115).

The remission rate of APC in the treatment of BE ranged from 64.9%-94.7%. Like previous endoscopic treatments, esophageal stricture was still the most commonly observed complication (121). To prevent postoperative esophageal stricture, hybrid APC was developed based on APC, that is, submucosal fluid injection before APC treatment to reduce the depth of coagulation which in turn reduces the incidence of esophageal stricture (122). In vitro animal experiments showed that the coagulation depth of hybrid APC was shallower than that of conventional APC (123). In the esophageal wall level analysis, the two methods were consistent in the injury of the epithelium, lamina propria, and muscularis mucosae. The number of submucosal injury cases by conventional APC was more significant, and only conventional APC damaged the proper muscle layer. In humans, it was found that the common observation remission rate of hybrid APC for BE was 96%, the pathological observation remission rate was 78%, and the postoperative stenosis rate was only at 2% (122).

Recently, CbFAS has been developed for esophageal mucosal ablation, which freezes the target mucosa to -85 degrees Celsius to achieve ablation of the BE mucosa (124). According to a recent prospective trial (117), the overall 1-year complete eradication of all dysplasia (CE-D) and complete eradication of intestinal metaplasia rates of BE patients with neoplasia was 95% and 88%, respectively. However, the CE-D rate of ultra-long BE, with length ≥ 8 cm, was only at 67%. The main complications were esophageal stricture (9.7%) and upper gastrointestinal bleeding without treatment (2.4%). Based on the current data, CbFAS can be used in the primary or emergency treatment of BE-related tumors and IM, but there is a lack of large-scale clinical research data to evaluate this technique.

This review summarizes BE and IM from the aspects of the definition, endoscopic recognition, pathology, diagnosis, and treatment. It is believed that there will be more detection and treatment in the future.

LZ and BS drafted this review. XZ and QW assisted in table making and sketching. SL and GL edited the grammar, text format, and framework, etc. TL put forward some essential suggestions on the content of the article. ML provided the direction and ideas of writing and made repeated revisions and guidance in the whole process of writing. All authors contributed to the article and approved the submitted version.

This research was supported by the NFSC (No. 81672458 and No. 81972996), Sichuan provincial Science and Technology Department (No. 2017JQ0052), and the Science and Technology Strategic Cooperation Programs of Luzhou Municipal People’s Government and Southwest Medical University (No. 2016LZXNYD-T06). Talent development project of The Affiliated Hospital of Southwest Medical University.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.