Surveillance and Treatment of Barrett’s Esophagus: Results from a Portuguese Tertiary Center Open Access

Barrett’s esophagus (BE) is a premalignant condition in which the normal esophageal stratified squamous epithelium is substituted by columnar epithelium with intestinal metaplasia [1‒5] in a length of at least ≥1 cm proximally to the gastroesophageal junction. The estimated prevalence of BE is around 1% in the general population but can reach as high as 7% in patients with gastroesophageal reflux disease. Portugal is a low-prevalence country of BE [2, 6, 7].

BE is an established precursor to esophageal adenocarcinoma (EAC) [1, 2, 8]. BE screening is based on the assumption that its early diagnosis would lead to effective endoscopic surveillance, early detection and management of neoplastic lesions [1, 8, 9]. The recommended intervals for surveillance and management strategies vary depending on the presence and grade of dysplasia. For non-dysplastic BE, routine endoscopic eradication therapy (EET) is not recommended as the risk of progression to neoplasia is extremely low [1, 2].

Guidelines recommend endoscopic resection of all visible neoplastic lesions followed by complete eradication of the remaining Barrett epithelium to reduce the risk of recurrent neoplasia. Recommendations are not consensual for flat low-grade dysplasia (LGD) among different societies. The European Society for Gastrointestinal Endoscopy (ESGE) recommends EET for LGD on at least two separate endoscopies, both confirmed by a second pathologist [3]. However, all societies state that surveillance of confirmed and persistent LGD is also an acceptable alternative in expert centers. For flat high-grade dysplasia (HGD), current guidelines recommend EET as the standard of care [1, 2].

Endoscopic resection of visible lesions can be performed with band or cap assisted endoscopic mucosal resection (EMR) or by endoscopic submucosal dissection (ESD). However, EMR is the most frequently used resection technique [1]. BE patients with intramucosal EAC or with submucosal EAC with low-risk features should also be offered endoscopic resection [9].

Although there are multiple available techniques for ablation therapy, radiofrequency ablation (RFA) remains the most widely used procedure due to its efficacy and safety profile [1]. Treatment is usually applied every 2–3 months [10] and usually requires 3 to 4 sessions for complete eradication [4]. Other effective ablation techniques include cryoablation and argon-plasma coagulation (APC) [2]. There is no standard definition of complete eradication of intestinal metaplasia (CE-IM), that is variably described in the literature as neo-squamous esophagus without any visible BE [3, 5, 10, 11].

Although EET provides high rates of success, there is still a risk of recurrent IM and/or dysplasia in some patients [1, 7]. Hence, ESGE guidelines recommend that patients treated for LGD should undergo endoscopic evaluation 1, 3 and 5 years after the last treatment, while surveillance for treated HGD/EAC should be performed 1, 2, 3, 4, 5, 7, and 10 years after the last treatment [6, 9]. At present, the leading approach for recurrent BE following EET detected during surveillance is to pursue further EET [7].We aimed to evaluate the outcomes of endoscopic resection and ablation of BE patients in a tertiary center, in a low-prevalence BE country.

A retrospective cohort study was performed in a tertiary center. Our center is a referral center for the treatment of BE in the North of Portugal. Referral criteria include ultralong-segment BE, dysplasia/neoplasia (flat or visible), and patients for ablation after resection of lesions in other centers. Endoscopies are preferentially performed by experienced endoscopists using high-definition endoscopes with near-focus technology and chromoendoscopy. Patients with the diagnosis of flat dysplasia have their pathological specimens reviewed by a second experienced pathologist in BE.

The pathological database was searched for “columnar epithelium,” “dysplasia,” and “adenocarcinoma” in esophageal specimens collected between January 2014 and December 2023. One hundred sixty-nine consecutive patients were included. Eighty patients were excluded: 37 patients with neoplasia not related to BE; 23 patients did not fulfill diagnostic criteria for BE; 11 patients with gastric heterotopia; 9 patients with BE and advanced neoplasia that were directly referred to surgery or palliative care.

Data collection was performed through analysis of electronic medical records and patient charts. Patient demographic characteristics were collected along with the following clinical, endoscopic, and pathological characteristics: start of follow-up at our institution, smoking habits, symptoms of chronic gastroesophageal reflux disease, body mass index, family history of BE or esophageal neoplasia, reason for referral, Prague classification at index endoscopy in our institution, pathology findings at index esophagogastroduodenoscopy and during follow-up, treatment of lesions, ablation of Barrett’s mucosa, AE of endoscopic resection and ablation, recurrence of intestinal metaplasia and neoplasia.

Endoscopic management (surveillance, treatment of lesions, and ablation) was undertaken according to the ESGE guidelines. The endoscopic characteristics were examined during esophagogastroduodenoscopy, focusing on the length of Barrett’s mucosa, presence of visible abnormalities on white light and chromoendoscopy, and signs of esophagitis. Biopsies of columnar mucosa in the esophagus were taken following Seattle biopsy protocol.

If deemed endoscopically resectable, lesions referred to our center or identified during surveillance were treated by band-assisted EMR or ESD – the choice of technique was made by the endoscopist, taking into account lesion size, morphology and pathology findings. Ablation of Barrett’s mucosa was preferentially performed using band-assisted EMR until 2013 and RFA from 2014 onwards, when this technique was introduced in our center.

BE was defined as columnar epithelium (with histopathologically confirmed intestinal metaplasia) that extends in the esophagus for at least 1 cm proximally to the proximal end of the gastric folds.

Response to ablation was defined as eradication ≥50% of initial Barrett’s length after 3 sessions of ablation. Classic CE-IM was defined as neo-squamous epithelium in the esophagus without any columnar mucosa. Clinical CE-IM was defined as neo-squamous epithelium in the esophagus, with residual diminute (<1 cm) islands of columnar mucosa.

All statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) software (version 29.0; IBM Corp., Chicago, IL, USA). Categorical variables are presented as frequency and/or proportion. Continuous variables are presented as means and standard deviations or as medians and interquartile ranges (IQR), according to the dispersion curve.

We included 89 patients with BE that underwent endoscopic surveillance or treatment between January 2004 and September 2024, with 80% being male (n = 71) and a mean age at referral to our center of 59 years (±14) (shown in Table 1). Eighteen patients had a diagnosis of BE at our institution. The indications for endoscopy were: heartburn (n = 4), findings on imaging studies (n = 3), surveillance of gastric lymphoma (n = 3), re-evaluation of grade C/D peptic esophagitis (n = 2), surveillance of hereditary risk syndromes (n = 2), surveillance post-gastric surgery (n = 2), anemia (n = 1), epigastric pain (n = 1).

Thirty-five patients were referred for flat dysplasia, of which: 22 had no flat dysplasia/lesion on index endoscopy (2 had a visible lesion and 1 flat dysplasia diagnosed on follow-up); 7 underwent ablation (4 for second flat LGD on referral, 2 for flat HGD on referral, and 1 for second flat LGD on index endoscopy); 6 had a visible lesion identified on index endoscopy. Nineteen patients were referred for resection of lesion, of which 2 did not have any lesion on index endoscopy and remained as non-dysplastic BE throughout follow-up.

A total of 37 lesions were identified (shown in Table 2). Seven lesions were first identified at index endoscopy (6 patients referred for flat dysplasia and 1 for long-segment BE). For primary lesions detected during surveillance (n = 6), the median time to detection was 28 months (IQR 22–50), and for metachronous lesions (n = 7) it was 5 months (IQR 3–12).

Twenty-eight lesions were treated with EMR, with a curative rate of 86% (24/28; 95% confidence interval [CI] 67%–96%). The four non-curative resections were of: 1 adenocarcinoma with submucosal invasion (pT1b) resected in piecemeal, later treated with surgery; 1 intramucosal adenocarcinoma (pT1a) with dysplasia in the horizontal margin, treated with second EMR; 1 lesion with HGD with positive horizontal margin treated with second EMR that complicated with perforation; and 1 intramucosal adenocarcinoma (pT1a) with positive margins for dysplasia and repeated dysplastic recurrence after EMR.

Six lesions were treated with ESD, with a curative rate of 67% (4/6; 95% CI: 22%–96%). The two non-curative resections were of: 1 adenocarcinoma with deep submucosal invasion (pT1bSm2) that was treated with surgery; and 1 adenosquamous intramucosal carcinoma (pT1a m3) that was considered to be a non-curative resection, given the aggressive nature of this entity, the young age of the patient and the ultra-long BE segment (C14M14), and the patient was proposed surgery.

An AE rate of 17% (1/6; 95% CI: 0%–64%) for ESD was observed: 1 stricture in a patient with previous EMR. An AE rate of 11% (3/28; 95% CI: 2%–28%) for EMR was observed: 2 strictures (7%, 95% CI: 1%–24%) – scars of 60% and 75% of the esophageal circumference, before the era of prophylactic steroids – and 1 perforation (4%, 95% CI: 0%–18%). All strictures were successfully treated with endoscopic dilation (median of 2 sessions, IQR 1–3). The perforation followed the EMR of a 21-mm dysplastic lesion, whose pathological evaluation showed an underlying leiomyoma. Due to the common finding of a double muscularis mucosae layer in patients with BE, it was not possible to conclude with certainty if the leiomyoma originated from the muscularis mucosae or muscularis propria, but ultimately its presence was considered a major contributor for the perforation. The perforation was successfully treated with stent, and the patient was submitted to a third EMR uneventfully.

Thirty-four patients had indication for Barrett’s mucosa eradication. Two patients had complete eradication of Barrett’s mucosa during EMR of lesions, with no need of further ablation sessions. Of the remaining 32 patients (shown in Table 3), the preferred first-line technique was RFA (69% of patients). Three patients are still undergoing eradication (1 with RFA and 2 with APC). A 9% global AE rate was noted (95% CI: 2%–24%) – 3 patients with strictures that were successfully treated with endoscopic dilation (1 session each). All 3 AE followed band-assisted EMR, resulting in an AE rate of 38% for EMR (95% CI: 9%–76%). There were no reports of AE following RFA (0%; 95% CI: 0%–15%).

Thirty-three patients have completed first-line of ablation, presenting a response rate was of 97% (n = 32; 95% CI: 84%–100%). Only 1 patient presented refractory BE – a patient with non-cirrhotic portal hypertension, esophageal varices and Barrett’s length of C6M6, who had a visible lesion resected during variceal elastic band ligation, and underwent RFA after elastic band ligation, with a final length of C0M4.

Of the 31 patients that completed the ablation program, the classic CE-IM was 61% (n = 19; 95% CI: 42%–78%) and the clinical CE-IM was 94% (n = 29; 95% CI: 79%–99%). The median number of ablation sessions until clinical CE-IM was 3 (IQR 1–5). Ten patients finished the ablation program with diminute islands; none of them developed significant pathology on remnant columnar mucosa (namely dysplasia/neoplasia) during follow-up (range 5–91 months).

Only 1 patient presented recurrence of intestinal metaplasia after successful eradication, which resulted from noncompliance to proton-pump inhibitor therapy, and was submitted to band-assisted EMR given the short length of the Barrett’s mucosa, being currently under surveillance.

Patients were followed for a median of 3 years (IQR 1–6): 59 patients are still on follow-up; 16 patients were discharged to their general practitioner; 9 patients died of non-BE-related causes during follow-up; 4 patients were referred for surgery (3 lesions not meeting curative criteria after ER and 1 lesion in buried Barrett); 1 patient was lost to follow-up. Only 1 patient died of BE-related neoplasia (1%; 95% CI: 0%–6%) – a patient that underwent successful Barrett’s mucosa eradication after endoscopic resection of an intramucosal adenocarcinoma, was diagnosed with buried Barrett’s neoplasia 8 years after eradication, which was treated with surgery, and 2 years after surgery would develop metastatic disease and eventually die.

The incidence of BE has been rising due to an increase in endoscopy volume and growing awareness among endoscopists. However, studies indicate that only one-third of patients with long-segment BE have been diagnosed and, therefore, only a minority is under surveillance for early detection of dysplasia and EAC [1]. This is probably aggravated in low-prevalence countries as physicians might not be as aware of the importance of case finding in high-risk patient groups, as well as when to refer patients for surveillance and treatment.

In this study, we evaluated the results of BE surveillance of a tertiary center in a low-prevalence country. Our study was limited by its retrospective nature, the relatively small sample size and short median follow-up period. These factors could potentially affect the robustness of the results, particularly in drawing conclusions about long-term outcomes or rare events. On the other hand, the occurrence of otherwise uncommon events on small numbers could overestimate the rate of such events.

The long study period could also hamper the interpretation of some results. The inclusion period of pathology specimens from 2014 to 2023 was chosen to maximize the sample size and follow-up period, resulting in the inclusion of patients starting follow-up as far back as 2004. We recognize that the extended timeframe of this study encompasses significant technological and clinical practice developments that could potentially introduce variability in the outcomes. The development of high-definition imaging and magnification may have improved lesion delineation, thus limiting the extension of endoscopic resection with EMR and reducing the rate of strictures. In addition to this, the use of prophylactic corticosteroids in large scars is only more recently routinely recommended. Furthermore, the introduction of RFA improved not only the ablation rate, as well as the stricture rate.

Compared to previous reports (Table 4), we highlight a lower CE-IM rate after EET. ER of any detected lesion is recommended due to a higher likelihood of malignancy. EMR has proven to be safe in various studies. Previous cohort studies [12, 13] have reported lower rates of perforation (0.0–0.1%), bleeding (1.2–1.4%), and stricture (1.0–1.3%) than our study. However, a recent meta-analysis of ESD and EMR for BE neoplasia (19 studies on EMR) reported pooled rates of 0.1%, 0.4–1.0%, and 8% for perforation, bleeding and stricture following resection with EMR, respectively [14]. Our numbers are similar to these, with the exception of the perforation rate, which was considerably higher in our cohort. Special consideration must be given to the fact that we included a lower volume of resected lesions when compared to other larger studies. The occurrence of a rare perforation (likely caused by the presence of underlying leiomyoma) on a small sample contributed significantly to the perforation rate. This AE might have been prevented if the lesion had been rejected by ESD, but neither did the optic diagnosis demanded this technique nor there was suspicion of a subepithelial lesion.

While our study includes few patients who underwent ESD for resection of BE lesions, the safety and efficacy results of this technique in our cohort are comparable to previous studies. A meta-analysis of 11 studies (524 lesions) reported a pooled curative resection rate of 64.9%, 1.8% bleeding rate, 1.5% perforation rate, and an 11.6% stricture rate [15]. Another recent meta-analysis (23 studies on ESD) reported a pooled curative resection rate of 65%, 1% intraprocedural bleeding, 1% delayed bleeding, 1% perforation rate, and 10% stricture rate [14]. A retrospective analysis of ESD results in BE in Korea (18 patients) reported curative resection in 61% of patients, perforation in 5%, and no strictures [21]. A comparison study of EMR and ESD in western centers (243 patients), reported stricture in 16% of ESD patients and zero perforations [22]. ESD has emerged as an alternative for en bloc resection of larger lesions, although it is associated with longer procedural times, higher rate of AE and being technically more challenging than EMR.

Following resection of any visible lesions or in the presence of dysplastic BE, remaining BE mucosal ablation is recommended. Although the entire BE segment can be resected with complete EMR or ESD, given the current evidence, several guidelines recommend RFA as the first-line therapy for ablation of flat dysplasia or following resection of visible lesions [23].

Despite the great success of EET, there is still a subset of patients who fail to achieve CE-IM and CE-D. Our classic CE-IM of 61% is inferior to previously reported rates of 77.4–88.2% [18‒20, 24]. Orman et al. [20] reported lower rates of CE-IM in retrospective studies than in prospective studies. Furthermore, the cited studies evaluated ablation results after RFA. Although RFA has been the preferential endoscopic modality in our center since 2014, 25% of our patients underwent ablation with EMR, which may have had a negative effect on our results. However, clinical CE-IM was high (94%) and patients with remnant diminute islands did not develop significant pathology on columnar mucosa. Therefore, the presence of diminute <1 cm islands on eradicated BE is not synonymous with MI, as described in another series [19], and in our cohort the clinical impact of such islands was null.

BE eradication with EMR is associated with a high AE rate. The post-EMR stricture rate of 38% in our cohort is similar to a previous study reporting strictures in 43.9%, although our overall AE rate was not as high the one reported [16]. Even though our study only included a few patients treated with EMR for complete eradication of BE mucosa, these results support the current evidence pointing that, despite its efficacy, the AE rate is not negligible, especially when compared to the first-line option. In fact, we had no AE from RFA in our cohort. A systematic review with meta-analysis of AE after RFA estimated a pooled rate of overall AE of 8.8% (95% CI: 6.5%–11.9%), reporting a trend toward higher adverse events in prospective studies/clinical trials compared to retrospective studies [17]. The retrospective nature of our study may explain the absence of post-procedure chest pain or transient dysphagia following RFA, but major AE such as strictures requiring dilation would not be overlooked.

Despite RFA being the preferred therapy for ablation, APC was successful in the 2 cases where it was applied. However, we must emphasize that this is not enough to draw conclusions regarding its use in BE mucosa eradication. Considering the uneven depth of ablation and the increased risk of stricture development, perforation and buried glands, its use should be limited to short-segment BE. Nonetheless, novel Hybrid APC (combining APC with submucosal injection prior to coagulation) is likely to decrease the risk of AE and, therefore, seems to be a promising method [25].

Although there is insufficient data on refractory BE and lack of a standard definition, studies have shown that it affects between 4% and 14% of patients [7]. In our cohort, only 1 patient (3%) presented refractory BE. The specificities of this patient might have played an important role in the eradication therapy of the BE mucosa. The optimal management strategy for these patients remains unclear. While wide-field salvage EMR of the entire BE has shown high eradication rate, it also carries a major risk of esophageal stricture and perforation, especially with increasing length of the affected segment. APC is not currently recommended as a first-line treatment of BE due to an excessively high rate of recurrence of metaplasia and dysplasia (up to one-third). However, considering its availability in most endoscopic units and endoscopists’ expertise with the device, it might be helpful in treating residual disease limited to the gastroesophageal junction or isolated islands. Finally, cryoballoon, which is a non-contact method using liquid nitrous oxide, is being studied with good results [7]; however, this modality is still not available at our center.

Several studies have also shown that intestinal metaplasia recurrence and progression to neoplasia is still possible after successful EET [6]. Large prospective cohorts and meta-analysis suggest a recurrence rate of approximately 8%–10% per patient/year of follow-up [23, 26], while three large systematic reviews and meta-analyses confirmed true recurrence rates between 7% and 13% [7]. In our cohort, only 1 (3%) patient presented recurrence of intestinal metaplasia due to noncompliance with PPI therapy. However, this result must be interpreted with caution due to the low number of patients and the short follow-up period.

Because the risk of recurrence of neoplasia persists even after successful EET, long-term surveillance and lifelong PPI therapy are required for all BE patients [9]. However, van Munster et al. [27] reported other-cause mortality risk to be more than 40 times higher than EAC-related mortality, and information on other comorbidities should help guide decision-making regarding cessation of follow-up. We reported a 1% mortality rate from BE-related neoplasia, but mortality from other causes was 10-fold higher. Nonetheless, a longer follow-up period is required to recognize the long-term BE-related mortality risk after EET. Indeed, the post-endoscopy esophageal cancer rate is a likely quality metric that should be studied in future research [5]. Moreover, there are other gaps that should be addressed in the future regarding EET: the impact of artificial intelligence on lesion delineation, stricture prevention and management, the role of hybrid APC and cryoballoon, the definition of refractory BE and its management, the clinical impact of diminute <1 cm islands, a standardized definition of CE-IM and optimal surveillance intervals after CE.

In conclusion, this is the first study reporting the results of a surveillance program of BE in a Portuguese center. Despite the low BE prevalence in our country, reflected in the low number of patients followed in our center, our results are mostly comparable to those published in literature, and underscore the safety and efficacy of our surveillance program.

This study was approved by the Ethical Committee of our institution in 2019 (CES 375/019). Patients signed written informed consent for the endoscopic procedures.

Diogo Libânio and Mário Dinis-Ribeiro were members of the journal’s editorial board at the time of submission.

This study was not supported by any sponsor or funder.

A.C.V collected and analyzed data, performed research, and wrote the paper; M.P. performed research and wrote the paper; D.L., M.D.-R., and I.M.S. critically revised the paper.

Ana Clara Vasconcelos and Marco Pereira share first co-authorship.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.