Time-dependent burden of reflux-related endoscopic pathology after sleeve gastrectomy: Clinical value of surveillance endoscopy in a 520-patient cohort

Abstract

BACKGROUND

Sleeve gastrectomy is a widely performed procedure; however, postoperative gastroesophageal reflux disease (GERD) and reflux-related mucosal injury remain clinically persistent concerns. The value and timing of postoperative surveillance endoscopy are controversial, particularly because symptoms may not fully reflect mucosal pathology.

AIM

To evaluate time-dependent, reflux-related, clinical and endoscopic pathology after sleeve gastrectomy and to assess the clinical utility of postoperative surveillance endoscopy.

METHODS

This retrospective cohort study included adults who underwent sleeve gastrectomy and both pre- and postoperative upper gastrointestinal endoscopy. Among 560 patients screened, 520 were included in the final cohort; after excluding those with preoperative Barrett’s, 502 were eligible for de novo Barrett’s esophagus analysis. The primary endpoint was strict clinically significant postoperative endoscopic pathology, defined as Los Angeles grade B (LA-B) or higher erosive esophagitis, histology-confirmed Barrett’s, de novo Barrett’s, or esophageal stricture. A broader clinically actionable endpoint was analyzed separately.

RESULTS

The mean follow-up was 52.8 ± 20.9 months. GERD symptoms increased from 26.5% to 43.3%, proton pump inhibitor use increased from 22.5% to 50.6%, erosive esophagitis increased from 16.0% to 29.6%, LA-B or higher esophagitis increased from 5.8% to 13.1%, and Barrett’s increased from 3.5% to 10.0%. De novo Barrett’s esophagus occurred in 34/502 (6.8%) patients. Strict clinically significant pathology was present in 108/520 (20.8%) patients, and broader clinically actionable pathology was present in 198/520 (38.1%). Treatment changes occurred in 172/520 (33.1%) patients.

CONCLUSION

Reflux-related clinical/endoscopic findings increase after sleeve gastrectomy, especially at longer follow-up intervals, thus supporting risk-adapted endoscopic surveillance, particularly beyond 3-5 years, but do not establish requirements for annual endoscopy.

Key Words: Sleeve gastrectomy; Gastroesophageal reflux disease; Erosive esophagitis; Barrett’s esophagus; Surveillance endoscopy; Bariatric surgery

Core Tip: Sleeve gastrectomy was associated with increased reflux-related clinical and endoscopic findings in this 520-patient cohort with paired pre- and postoperative endoscopy data. De novo Barrett’s esophagus was detected in 34/502 (6.8%) eligible patients. Strict clinically significant postoperative endoscopic pathology was greater at longer follow-up intervals, reaching 53/175 (30.3%) beyond 60 months, whereas treatment change reached 90/175 (51.4%) within the same interval. These findings support structured, risk-adapted endoscopic surveillance, particularly beyond 3-5 years, but do not establish the need for universal annual endoscopy.

INTRODUCTION

Sleeve gastrectomy has become one of the predominant bariatric procedures performed worldwide due to its technical simplicity, favorable metabolic effects, and durable weight loss efficacy[1]. Randomized trials have reported substantial long-term weight loss after sleeve gastrectomy, although reflux outcomes may be less favorable than those after Roux-en-Y gastric bypass in some patients[2,3].

Postoperative gastroesophageal reflux disease (GERD) is clinically relevant because it involves both symptoms and objective mucosal injury. The proposed mechanisms include increased intragastric pressure, altered sleeve geometry, disruption of the angle of His, reduced compliance, hiatal hernia, and impaired function of the esophagogastric junction[4-13]. These mechanisms may contribute to erosive esophagitis, bile reflux, Barrett’s esophagus, and/or strictures.

Systematic reviews and endoscopic cohort studies have reported increased reflux burden and Barrett’s esophagus after sleeve gastrectomy[4-7,14-20]. Current bariatric and endoscopic guidance recognizes the role of upper endoscopy before and after bariatric surgery, while guidelines for Barrett’s esophagus emphasize histological confirmation and risk-stratified surveillance once Barrett’s esophagus is diagnosed[21-28]. The unresolved clinical question is whether symptom-based follow-up alone is sufficient after sleeve gastrectomy, or whether structured endoscopic surveillance adds clinically actionable information.

Accordingly, this study aimed to evaluate pre- to postoperative changes in reflux-related clinical and endoscopic findings after sleeve gastrectomy, to determine the prevalence of de novo Barrett’s esophagus, and to assess whether strict clinically significant endoscopic pathology, broader clinically actionable pathology, and treatment changes are greater at longer follow-up intervals.

MATERIALS AND METHODS

Study design and setting

The present investigation was a retrospective, single-center cohort study involving adults who underwent sleeve gastrectomy and structured pre- and postoperative upper gastrointestinal endoscopic assessments. This study was approved by the Ethics Committee of Mersin City Training and Research Hospital (Mersin, Türkiye) on May 20, 2026 (No. 2026/251).

Patient selection

In total, 560 patient records were screened. Fifteen and 25 patients were excluded because of missing preoperative and postoperative endoscopy data, respectively. As such, the final cohort comprised 520 patients. For de novo Barrett’s esophagus analysis, 18 patients with preoperative Barrett’s were excluded, leaving data from 502 eligible for analysis.

Definitions and outcomes

GERD symptoms, proton pump inhibitor (PPI) use, bile reflux, hiatal hernia, erosive esophagitis, Los Angeles (LA) grade, Barrett’s esophagus, histology-confirmed Barrett’s esophagus, esophageal stricture, and treatment changes after postoperative endoscopy were analyzed and recorded. De novo Barrett’s was defined as postoperative Barrett’s esophagus in patients without preoperative Barrett’s.

The primary endpoint was strict clinically significant postoperative endoscopic pathology, defined as LA grade B (LA-B) or higher erosive esophagitis, histology-confirmed Barrett’s, de novo Barrett’s esophagus, or esophageal stricture. The dataset-defined broader endpoint was labeled as broader clinically actionable postoperative pathology.

Postoperative endoscopy indication and timing

Postoperative endoscopy in the analytical cohort was performed within a real-world postoperative follow-up framework, including protocol-based surveillance and clinically indicated evaluation, when applicable. Examinations were categorized according to the postoperative interval at which they were performed: 24-36 months, 36-60 months, or > 60 months. Symptom status was analyzed separately and was not used as the sole criterion for inclusion. Therefore, these follow-up groups should be interpreted as time-stratified postoperative surveillance intervals, rather than as serial repeated endoscopic examinations in individual patients.

Diagnosis of Barrett’s esophagus and biopsy documentation

Barrett’s esophagus was analyzed using dataset fields for endoscopic Barrett’s, histologically confirmed Barrett’s, and intestinal metaplasia confirmation. For the strict primary endpoint and analyses of histologically confirmed Barrett’s esophagus, Barrett’s required endoscopic suspicion with histological confirmation of intestinal metaplasia. A standardized Seattle biopsy protocol was not consistently documented in all records; this limitation is acknowledged in the Discussion section.

Treatment change definition

Treatment changes after postoperative endoscopy were analyzed as binary outcomes. These changes may have included the initiation or escalation of PPI therapy, planned repeat endoscopy or histological surveillance, gastroenterology referral, further clinical evaluation, or surgical/endoscopic referral, when appropriate. The exact type, intensity, and timing of treatment changes were not available for all patients and could not be inferred.

Statistical analysis

Continuous variables are expressed as mean ± SD or median with interquartile range. Categorical variables are expressed n (%). Paired preoperative and postoperative categorical variables were compared using McNemar’s exact test. Follow-up strata were compared using the χ² or Fisher’s exact tests, as appropriate, with ordered trend testing. Multivariable logistic regression was used to analyze predictors of strict clinically significant postoperative pathology. Preoperative Barrett’s esophagus was removed from this model due to sparse data instability. Because Firth logistic regression was not available, a parsimonious standard logistic regression model was used. A parsimonious model was also used for de novo Barrett’s esophagus because only 34 events were available. Odds ratio (OR), corresponding 95% confidence interval (CI), and two-sided P values are reported. All statistical tests were two-sided and differences with P < 0.05 were considered statistically significant.

Missing variables

The dataset did not include postoperative body mass index (BMI), percentage of total weight loss, percentage of excess weight loss, pH monitoring, impedance testing, manometry, validated GERD questionnaire scores, detailed sleeve morphology, exact treatment change details, dysplasia outcomes, or cancer outcomes. The availability of study variables is summarized in Supplementary Table 1.

RESULTS

Patient selection and baseline characteristics

Of 560 patient records screened, 40 were excluded: 15 with missing preoperative endoscopy and 25 with missing postoperative endoscopy data. The final cohort included 520 patients, 502 of whom were eligible for de novo Barrett’s esophagus analysis. The mean (± SD) age was 45.9 ± 10.6 years, 332/520 (63.8%) patients were female, the mean preoperative BMI was 43.8 ± 4.5 kg/m², and the mean follow-up was 52.8 ± 20.9 months. Preoperative GERD symptoms were present in 138/520 (26.5%) patients, PPI use in 117/520 (22.5%), erosive esophagitis in 83/520 (16.0%), LA-B or higher esophagitis in 30/520 (5.8%), and Barrett’s esophagus in 18/520 (3.5%) (Figure 1 and Table 1).

Figure 1 Study flow diagram. The exclusion criteria are shown in a separate box. The de novo Barrett’s esophagus cohort excluded patients with preoperative Barrett’s esophagus.

Table 1 Baseline demographic and clinical characteristics.

Characteristic	Overall (n = 520), n/N (%)
Age, years	45.9 ± 10.6
Preoperative BMI, kg/m²	43.8 ± 4.5
Follow-up duration, months	52.8 ± 20.9
Female sex	332/520 (63.8)
Smoking status	114/520 (21.9)
Diabetes mellitus	145/520 (27.9)
Hypertension	184/520 (35.4)
Obstructive sleep apnea	105/520 (20.2)
Preoperative GERD symptoms	138/520 (26.5)
Preoperative PPI use	117/520 (22.5)
Preoperative bile reflux	29/520 (5.6)
Preoperative hiatal hernia	110/520 (21.2)
Preoperative erosive esophagitis	83/520 (16.0)
Preoperative LA-B-or-higher esophagitis	30/520 (5.8)
Preoperative Barrett’s esophagus	18/520 (3.5)
Follow-up group: 24-36 months	144/520 (27.7)
Follow-up group: 36-60 months	201/520 (38.7)
Follow-up group: > 60 months	175/520 (33.7)
Postoperative GERD symptoms	225/520 (43.3)
Postoperative PPI use	263/520 (50.6)
Postoperative bile reflux	74/520 (14.2)
Postoperative hiatal hernia	100/520 (19.2)
Postoperative erosive esophagitis	154/520 (29.6)
Postoperative LA-B-or-higher esophagitis	68/520 (13.1)
Postoperative Barrett’s esophagus	52/520 (10.0)
Histology-confirmed Barrett’s esophagus	52/520 (10.0)
Esophageal stricture	14/520 (2.7)
Treatment change after postoperative endoscopy	172/520 (33.1)
De novo Barrett’s esophagus	34/502 (6.8)

BMI: Body mass index; GERD: Gastroesophageal reflux disease; PPI: Proton pump inhibitor; LA-B: Los Angeles grade B.

Preoperative vs postoperative findings

GERD symptoms increased from 26.5% to 43.3% (P < 0.001), PPI use from 22.5% to 50.6% (P < 0.001), bile reflux from 5.6% to 14.2% (P < 0.001), erosive esophagitis from 16.0% to 29.6% (P < 0.001), LA-B or higher esophagitis from 5.8% to 13.1% (P < 0.001), and Barrett’s esophagus from 3.5% to 10.0% (P < 0.001). The prevalence of hiatal hernia decreased from 21.2% to 19.2% (P = 0.407) (Table 2 and Figure 2).

Figure 2 Preoperative vs postoperative reflux-related findings. GERD: Gastroesophageal reflux disease; PPI: Proton pump inhibitor; LA-B+: Los Angeles grade B or higher.

Table 2 Preoperative vs postoperative reflux-related findings.

Finding	Preoperative, n/N (%)	Postoperative, n/N (%)	Absolute change, percentage points	McNemar P value
GERD symptoms	138/520 (26.5)	225/520 (43.3)	16.7	< 0.001
PPI use	117/520 (22.5)	263/520 (50.6)	28.1	< 0.001
Bile reflux	29/520 (5.6)	74/520 (14.2)	8.7	< 0.001
Hiatal hernia	110/520 (21.2)	100/520 (19.2)	-1.9	0.407
Erosive esophagitis	83/520 (16.0)	154/520 (29.6)	13.7	< 0.001
LA-B-or-higher esophagitis	30/520 (5.8)	68/520 (13.1)	7.3	< 0.001
Barrett’s esophagus	18/520 (3.5)	52/520 (10.0)	6.5	< 0.001

GERD: Gastroesophageal reflux disease; PPI: Proton pump inhibitor; LA-B: Los Angeles grade B.

Time-stratified outcomes and regression

Strict clinically significant postoperative endoscopic pathology occurred in 108/520 (20.8%) patients and increased from 18/144 (12.5%) at 24-36 months to 37/201 (18.4%) at 36-60 months and 53/175 (30.3%) at > 60 months. Broader clinically actionable postoperative pathology occurred in 198/520 (38.1%) patients and increased from 29/144 (20.1%) to 67/201 (33.3%) and 102/175 (58.3%), respectively. De novo Barrett’s occurred in 34/502 (6.8%) patients, increasing from 3/138 (2.2%) to 14/195 (7.2%) and 17/169 (10.1%) across the follow-up groups, respectively (P = 0.023; trend P = 0.007). Treatment changes occurred in 172/520 (33.1%) patients and increased from 25/144 (17.4%) to 57/201 (28.4%) and 90/175 (51.4%), respectively (P < 0.001; trend P < 0.001) (Table 3; Figures 3 and 4). Predictors of treatment changes after postoperative endoscopy are presented in Supplementary Table 2.

Figure 3 Strict clinically significant endoscopic pathology and treatment changes by follow-up duration.

Figure 4 De novo Barrett’s esophagus by follow-up duration using corrected eligible denominators: 3/138 (2.2%), 14/195 (7.2%), and 17/169 (10.1%).

Table 3 Time-dependent postoperative outcomes.

Outcome	24-36 months, n/N (%)	36-60 months, n/N (%)	> 60 months, n/N (%)	Overall P value	Trend P value
Postoperative GERD symptoms	60/144 (41.7)	88/201 (43.8)	77/175 (44.0)	0.900	0.684
Postoperative PPI use	60/144 (41.7)	97/201 (48.3)	106/175 (60.6)	0.002	< 0.001
Postoperative bile reflux	21/144 (14.6)	25/201 (12.4)	28/175 (16.0)	0.609	0.677
Postoperative hiatal hernia	13/144 (9.0)	37/201 (18.4)	50/175 (28.6)	< 0.001	< 0.001
Postoperative erosive esophagitis	35/144 (24.3)	57/201 (28.4)	62/175 (35.4)	0.085	0.028
Postoperative LA-B+ esophagitis	11/144 (7.6)	21/201 (10.4)	36/175 (20.6)	0.001	< 0.001
Postoperative Barrett’s esophagus	9/144 (6.2)	20/201 (10.0)	23/175 (13.1)	0.124	0.041
De novo Barrett’s esophagus	3/138 (2.2)	14/195 (7.2)	17/169 (10.1)	0.023	0.007
Esophageal stricture	1/144 (0.7)	4/201 (2.0)	9/175 (5.1)	0.037	0.013
Strict clinically significant postoperative endoscopic pathology	18/144 (12.5)	37/201 (18.4)	53/175 (30.3)	< 0.001	< 0.001
Broader clinically actionable postoperative pathology	29/144 (20.1)	67/201 (33.3)	102/175 (58.3)	< 0.001	< 0.001
Treatment change after postoperative endoscopy	25/144 (17.4)	57/201 (28.4)	90/175 (51.4)	< 0.001	< 0.001

GERD: Gastroesophageal reflux disease; PPI: Proton pump inhibitor; LA-B: Los Angeles grade B.

After removing preoperative Barrett’s due to sparse data instability, the strict endpoint model included age, female vs male sex, preoperative BMI, smoking, preoperative GERD symptoms, preoperative LA-B-or-higher esophagitis, and the follow-up group (Table 4 and Figure 5). Follow-up of 36-60 months was of borderline statistical significance compared with 24-36 months (OR 1.94, 95%CI: 1.00-3.77; P = 0.050), whereas follow-up > 60 months was independently associated with strict pathology (OR 4.65, 95%CI: 2.40-9.03; P < 0.001). The de novo Barrett’s esophagus model remained parsimonious because only 34 events were available (Table 5).

Figure 5 Forest plot of multivariable predictors of strict clinically significant postoperative endoscopic pathology after the removal of preoperative Barrett’s esophagus because of sparse data instability. The reference group for the follow-up duration was 24-36 months. GERD: Gastroesophageal reflux disease; BMI: Body mass index; LA-B+: Los Angeles grade B or higher.

Table 4 Multivariable predictors of strict clinically significant postoperative endoscopic pathology.

Variable	OR	95%CI	P value
Age, per year	1.03	1.01-1.06	0.007
Female sex vs male	0.62	0.38-1.00	0.051
Preoperative BMI, per kg/m²	0.98	0.93-1.04	0.496
Smoking status: Yes vs no	3.10	1.81-5.31	< 0.001
Preoperative GERD symptoms: Yes vs no	5.17	3.17-8.44	< 0.001
Preoperative LA-B-or-higher esophagitis: Yes vs no	1.68	0.67-4.21	0.272
Follow-up 36-60 months vs 24-36 months	1.94	1.00-3.77	0.050
Follow-up > 60 months vs 24-36 months	4.65	2.40-9.03	< 0.001

BMI: Body mass index; GERD: Gastroesophageal reflux disease; LA-B: Los Angeles grade B; OR: Odds ratio; CI: Confidence interval.

Table 5 Multivariable predictors of de novo Barrett’s esophagus.

Variable	OR	95%CI	P value
Age, per year	1.04	1.01-1.08	0.021
Female sex vs male	0.33	0.16-0.71	0.004
Smoking status: Yes vs no	2.88	1.32-6.26	0.008
Preoperative GERD symptoms: Yes vs no	2.41	1.14-5.09	0.021
Follow-up 36-60 months vs 24-36 months	4.13	1.14-15.02	0.031
Follow-up > 60 months vs 24-36 months	6.73	1.86-24.36	0.004

GERD: Gastroesophageal reflux disease; OR: Odds ratio; CI: Confidence interval.

Sensitivity analyses

Sensitivity analyses for the broader endpoint showed clinically actionable pathology in 190/502 (37.8%) patients after excluding preoperative Barrett’s, 169/375 (45.1%) among those with postoperative endoscopy at > 36 months, and 102/175 (58.3%) among those followed up for > 60 months. De novo Barrett’s occurred in 31/363 (8.5%) patients after > 36 months, and in 17/169 (10.1%) after 60 months. A broader clinically actionable pathology was present in 70/295 asymptomatic (23.7%) and 128/225 symptomatic (56.9%) patients (Supplementary Table 3).

DISCUSSION

This study revealed that reflux-related pathology after sleeve gastrectomy is common and observed more frequently at longer postoperative follow-up intervals. Postoperative GERD symptoms were present in 43.3% of patients, PPI use in 50.6%, erosive esophagitis in 29.6%, LA-B-or-higher esophagitis in 13.1%, postoperative Barrett’s esophagus in 10.0%, and de novo Barrett’s esophagus in 34/502 (6.8%). Strict clinically significant pathology was observed in 53/175 (30.3%) patients beyond 60 months, and treatment changes were observed in 90/175 (51.4%) during the same interval.

These findings are consistent with randomized trials and meta-analyses reporting that reflux outcomes after sleeve gastrectomy may be less favorable than those after Roux-en-Y gastric bypass[2-6]. The postoperative GERD symptom rate of 43.3% and the PPI use rate of 50.6% in our cohort were within the broad pattern of increased reflux burden reported internationally[4-7,17-20,26-28]. Our erosive esophagitis rate of 29.6% and LA-B-or-higher esophagitis rate of 13.1% also align with endoscopic series reporting objective mucosal injury after sleeve gastrectomy[8-11,15,16,20].

Barrett’s esophagus findings merit particular attention. Our postoperative Barrett’s esophagus rate of 10.0% and de novo rate of 6.8% are broadly consistent with systematic reviews and long-term endoscopic cohort studies that reported Barrett’s after sleeve gastrectomy[8,14-16]. Causality cannot be inferred; however, the increase from 2.2% at 24-36 months to 10.1% after 60 months supports the concern that reflux-related mucosal changes may become more apparent over time.

The Barrett’s esophagus rate observed in this cohort provides a clinically useful benchmark for long-term, post-sleeve follow-up. Our postoperative Barrett’s esophagus rate of 10.0% and corrected de novo Barrett’s esophagus rate of 34/502 (6.8%) are close to the range reported in systematic reviews and long-term endoscopic surveillance cohorts, including the meta-analysis by Qumseya et al[14] and the endoscopy-based series by Felsenreich et al[8], Genco et al[15] and Sebastianelli et al[16]. These comparisons should be interpreted cautiously because published estimates vary according to follow-up duration, biopsy protocol, whether asymptomatic patients underwent endoscopy, and whether Barrett’s esophagus required histological confirmation. Nevertheless, our findings are directionally consistent with the broader literature suggesting that Barrett’s esophagus after sleeve gastrectomy is not restricted to patients with severe symptoms, and may become more apparent only several years after surgery.

Time-stratified and management-related findings place the present cohort in a clinically relevant context. Strict clinically significant postoperative endoscopic pathology was present in 53/175 (30.3%) patients beyond 60 months, and treatment changes occurred in 90/175 (51.4%) within the same interval. In addition, broader clinically actionable postoperative pathology was detected in 70/295 asymptomatic patients (23.7%), supporting the concern that symptom-based follow-up alone may miss a meaningful subset of patients with mucosal injury. However, these data should not be interpreted as evidence that all asymptomatic patients require a universal annual endoscopy. Rather, consistent with bariatric endoscopy guidance and Barrett’s surveillance principles, our findings support structured, risk-adapted surveillance, particularly beyond 3-5 years and in patients with clinical or endoscopic risk factors.

The question of postoperative endoscopic surveillance is not unique to sleeve gastrectomy. Patients undergoing Roux-en-Y gastric bypass and other bariatric procedures may require endoscopic evaluation for procedure-specific symptoms, marginal ulceration, anemia, dysphagia, abdominal pain, and other postoperative concerns[21,22]. However, the reflux-Barrett pathway is particularly relevant after sleeve gastrectomy because sleeve anatomy may increase reflux exposure and because Barrett’s surveillance principles become clinically relevant once intestinal metaplasia is identified[14,21-25]. Thus, surveillance after sleeve gastrectomy is not simply a generic postoperative endoscopy issue: It focuses on detecting reflux-related mucosal injury and identifying patients who may benefit from medical, endoscopic, gastroenterological, and/or surgical management.

This study has several strengths, including a 520-patient cohort, paired pre- and postoperative endoscopy, corrected de novo Barrett denominators, time-stratified follow-up, separation of strict and broader endpoints, sensitivity analyses, and treatment changes as a clinically meaningful management endpoint. Limitations of this study include its retrospective, single-center design, possible selection bias, binary symptom recording, absence of validated GERD scores, no pH/impedance/manometry, no postoperative weight loss outcomes, no sleeve morphology assessment, no exact treatment-change details, no dysplasia or cancer outcomes, limited de novo Barrett event count, and possible residual confounding. The absence of objective reflux testing is an important limitation. The dataset did not include pH monitoring, impedance testing, or manometry; therefore, this analysis could not distinguish acidic from non-acidic reflux, quantify reflux exposure, or evaluate the physiology of the esophagogastric junction. Data regarding postoperative weight loss outcomes, including postoperative BMI, percentage of total weight loss, and percentage of excess weight loss, were also unavailable. Consequently, the relationship between postoperative weight loss response and reflux-related endoscopic pathology could not be assessed. In addition, a standardized Seattle biopsy protocol was not consistently documented, and treatment change was available only as a binary variable without consistent details regarding medication adjustment, planned surveillance, referral, or procedural intervention.

CONCLUSION

Sleeve gastrectomy was associated with increased postoperative reflux-related clinical and endoscopic findings. Strict clinically significant postoperative endoscopic pathology occurred in 20.8%, broader clinically actionable postoperative pathology occurred in 38.1%, and de novo Barrett’s esophagus occurred in 34/502 (6.8%) eligible patients. These findings support structured, risk-adapted endoscopic surveillance, particularly beyond 3-5 years. Therefore, prospective studies are needed to define optimal timing and patient selection. Results of the present study, however, do not justify a blanket recommendation for universal annual endoscopy.