Functional dyspepsia and risk of Barrett’s esophagus and dysplasia: A propensity-matched multicenter retrospective cohort study

Abstract

BACKGROUND

Functional dyspepsia (FD) and gastroesophageal reflux disease (GERD) exhibit substantial symptom overlap (40%-50%), and up to one-third of patients diagnosed with FD demonstrate pathologic acid exposure on objective testing. While GERD is a well-established risk factor for Barrett’s esophagus (BE), the association between FD diagnosis and BE risk remains unexplored. Understanding this relationship is critical because FD is typically diagnosed symptomatically without routine objective reflux testing in clinical practice, potentially masking undiagnosed GERD or identifying shared pathophysiologic mechanisms. We hypothesized that patients diagnosed with FD demonstrate an elevated risk for BE and Barrett’s-associated dysplasia compared to matched controls.

AIM

To investigate the association between FD diagnosis and the risk of BE and dysplasia.

METHODS

This retrospective cohort study utilized the TriNetX Global Collaborative Network database. Patients diagnosed with FD (International Classification of Diseases, 10^th Revision code K30) between January 2006 and December 2022 were identified, and propensity-score matched 1:1 with controls based on demographics, comorbidities, and GERD diagnosis. The primary outcomes were the development of BE and Barrett’s-associated dysplasia. Cox proportional hazards regression analysis was performed to calculate hazard ratios (HR) with 95% confidence intervals (CI). A sensitivity analysis was conducted to include patients with documented esophagoduodenoscopy during the study period.

RESULTS

After PSM, 384267 patients were included in each cohort. Patients with FD demonstrated significantly higher risk for BE (HR: 1.900, 95%CI: 1.813-1.991), Barrett's associated dysplasia (HR: 3.568, 95%CI: 3.064-4.155), and esophageal cancer (HR: 1.649, 95%CI: 1.450-1.875). Sensitivity analysis confirmed the increased risk of BE and dysplasia but not esophageal cancer.

CONCLUSION

Patients diagnosed with FD demonstrate substantially elevated risk for BE and dysplasia. This association, whether reflecting undiagnosed GERD, shared pathophysiology, or both, identifies FD patients as a higher-risk population warranting consideration of objective reflux testing and selective endoscopic evaluation in appropriate clinical contexts.

Key Words: Functional dyspepsia; Barrett’s esophagus; Gastroesophageal reflux disease; Esophageal adenocarcinoma; Dysplasia; Risk stratification; Propensity score matching; Diagnostic overlap; Acid reflux

Core Tip: This large-scale cohort study of over 3.6 million patients demonstrates that individuals diagnosed with functional dyspepsia (FD) have a substantially elevated risk for Barrett's esophagus and dysplasia compared to matched controls. Interestingly, this increased risk persisted even after adjusting for documented gastroesophageal reflux disease (GERD). Whether this reflects true pathophysiologic overlap, diagnostic misclassification (especially considering overlap in presentation for FD and gastroesophageal reflux disease), or shared predisposing factors, patients diagnosed with FD represent a higher-risk population warranting reconsideration of current diagnostic strategies, including objective reflux testing and risk-stratified endoscopic evaluation.

INTRODUCTION

Functional dyspepsia (FD) is a clinical syndrome that presents with epigastric pain, burning, postprandial fullness, early satiety, bloating, belching, and/or nausea. This presentation can mimic several gastrointestinal disorders[1]. These symptoms are generally thought to be due to delayed gastric emptying, impaired accommodation of the proximal stomach, abnormal gastric acid secretion, visceral hypersensitivity, and psychological factors[2]. Most dyspepsia cases do not have an underlying structural explanation and are attributed to FD[3]. The global prevalence of FD is around 8.4% with 95% confidence interval (CI) between 7.4%-9.5%. In the United States, the reported prevalence is 8.9%, with higher prevalence noted among females[4].

Barrett’s esophagus (BE) is a condition where the esophageal mucosa changes from typical squamous epithelium to columnar intestinal-type epithelium. It can be a precursor to esophageal cancer (EC). BE is strongly associated with a history of gastroesophageal reflux disease (GERD)[5]. Impaired gastric emptying/accommodation associated with FD has been shown to increase gastroesophageal reflux, which has prompted the consideration of BE risk in those with a diagnosis of FD[6]. While evidence of association between GERD and esophageal metaplasia/dysplasia is abundant, data about the association between FD and risk of developing esophageal metaplasia/dysplasia are lacking. Thus, we performed a propensity-matched network analysis to evaluate this association.

MATERIALS AND METHODS

We used TriNetX, a multi-center research network that provides real-time access to medical record data from healthcare organizations, to perform a retrospective cohort study. TriNetX provides de-identified data that is Health Insurance Portability and Accountability Act compliant. The Western Institutional Review Board has granted TriNetX a waiver of review and informed consent. Patients were identified based on International Classification of Diseases, 10^th Revision (ICD-10) codes for diagnoses, and Current Procedural Terminology (CPT) codes were used to retrieve procedure data. Of note, this approach aligns with real-world FD labeling, which is often a clinical diagnosis, rather than strictly Rome-defined FD.

Study design and participants

Patients aged ≥ 18 were enrolled between January 1, 2006, and December 31, 2022. Patients with a history of BE or EC before the study period were excluded. Patients were stratified into two cohorts: One with patients diagnosed with FD and the other with patients without a diagnosis of FD.

Study definitions and outcomes

We used the diagnosis of BE as our primary outcome, with Barrett’s associated dysplasia and EC as secondary outcomes.

Statistical analysis

We used TriNetX’s built-in statistical capabilities to conduct our analyses. Propensity matching was used to balance the cohorts for the following potential confounders: Age, gender, race, ethnicity, elevated body mass index, smoking, alcohol misuse, GERD, and proton pump inhibitor (PPI) use at baseline. The 1:1 propensity score matching was conducted using greedy nearest-neighbour matching with a caliper width of 0.1 pooled SDs. Covariate balance was evaluated using a standardized mean difference (SMD) threshold of 0.1. Outcomes were evaluated using the Cox proportional hazards model and Kaplan-Meier analysis in R’s Survival package v3.2-3.

Sensitivity analysis

Given that GERD and BE can be asymptomatic and thus remain undetected, we conducted a sensitivity analysis limited to patients with documented upper gastrointestinal endoscopic evaluations following the index. This approach sought to test the robustness of our findings by minimizing the impact of undetected cases.

RESULTS

Before propensity matching, the FD cohort (n = 487706) and the control cohort (n = 1384584) showed significant differences in baseline characteristics (Table 1). The FD cohort was older on average (49.2 years vs 45.0 years; SMD = 0.258), had a higher proportion of females (62.0% vs 59.2%; SMD = 0.057), and a greater prevalence of GERD (33.6% vs 7.9%; SMD = 0.670), overweight/obesity (15.4% vs 6.7%; SMD = 0.279), and nicotine dependence (9.2% vs 5.2%; SMD = 0.157). PPI use was also much higher in the FD cohort (43.7% vs 9.5%; SMD = 0.838). After 1:1 propensity matching, each cohort had 384267 patients. All characteristics were balanced with an SMD < 0.1 (Figure 1). The mean ± SD duration of follow-up was 5.3 ± 4.4 years for the FD cohort and 9.5 ± 6.9 years for the control cohort.

Figure 1  Propensity score density graph (before and after matching).

Table 1 Baseline characteristics of the study population, n (%).

Characteristics	Before propensity matching patients			After propensity matching patients
	Functional dyspepsia group (n = 487706)	Control group (n = 1384584)	SMD	Functional dyspepsia group (n = 384267)	Control group (n = 384267)	SMD
Age at index, mean ± SD	49.2 ± 17.1	45.0 ± 15.3	0.258	48.2 ± 17.1	47.5 ± 15.2	0.042
Sex
Female	302207 (62.0)	819436 (59.2)	0.057	241980 (63.0)	238544 (62.1)	0.018
Male	170462 (35.0)	565077 (40.8)	0.121	142239 (37.0)	145652 (37.9)	0.018
Race
Black or African American	49777 (10.2)	162797 (11.8)	0.050	39892 (10.4)	44297 (11.5)	0.037
White	238838 (49.0)	842091 (60.8)	0.240	199026 (51.8)	192434 (50.1)	0.034
American Indian or Alaskan Native	1271 (0.3)	2808 (0.2)	0.012	897 (0.2)	1205 (0.3)	0.015
Native Hawaiian or other Pacific Islander	2554 (0.5)	2096 (0.2)	0.064	1079 (0.3)	1243 (0.3)	0.008
Asian	24991 (5.1)	34167 (2.5)	0.139	18235 (4.7)	19283 (5.0)	0.013
Ethnicity
Hispanic or Latino	39106 (8.0)	127065 (9.2)	0.041	31226 (8.1)	28391 (7.4)	0.028
Not Hispanic or Latino	231905 (47.6)	814939 (58.9)	0.228	189279 (49.3)	183993 (47.9)	0.028
Diagnosis
Alcohol-related disorders	14068 (2.9)	23579 (1.7)	0.079	9577 (2.5)	10826 (2.8)	0.020
Overweight/obesity	74934 (15.4)	92963 (6.7)	0.279	46874 (12.2)	49200 (12.8)	0.018
Nicotine dependence	44907 (9.2)	71676 (5.2)	0.157	29655 (7.7)	33806 (8.8)	0.039
Gastro-esophageal reflux disease	164072 (33.6)	109050 (7.9)	0.670	92927 (24.2)	93552 (24.3)	0.004
Medication
Proton pump inhibitors	213001 (43.7)	131767 (9.5)	0.838	123576 (32.2)	119484 (31.1)	0.023

SMD: Standard mean difference.

Outcomes

The FD cohort had a significantly higher risk of BE [hazard ratios (HR) = 1.900, 95%CI: 1.813-1.991, Log-rank P < 0.001] and Barrett's associated dysplasia (HR = 3.568, 95%CI: 3.064-4.155, Log-rank P < 0.001) during follow-up (Figure 2 and Table 2). The risk of developing EC was also higher in the FD cohort (HR = 1.649, 95%CI: 1.450-1.875, Log-rank P < 0.001). The risk of BE (HR = 1.414, 95%CI: 1.342-1.490, Log-rank P < 0.001) and Barrett's associated dysplasia (HR = 2.181, 95%CI: 1.854-2.566, Log-rank P < 0.001) remained significantly higher in the FD cohort in sensitivity analysis. However, the rate of EC (HR = 1.051, 95%CI: 0.889-1.241, Log-rank P = 0.562) was not significantly different between the two cohorts.

Figure 2 Kaplan-Meier curve for Barrett’s esophagus. FD: Functional dyspepsia; BE: Barrett’s esophagus.

Table 2 Primary and secondary outcomes in the two cohorts, n (%).

Outcome	Patients with outcome in the FD cohort (n = 384267)	Patients with outcome in the control cohort (n = 384267)	Hazard ratio (95%CI)	Log rank P value
Barrett’s esophagus	4105 (1.07)	3790 (0.99)	1.900 (1.813-1.991)	< 0.001
Barrett’s-associated dysplasia	454 (0.12)	349 (0.09)	3.568 (3.064-4.155)	< 0.001
Esophageal cancer	524 (0.14)	516 (0.13)	1.649 (1.450-1.875)	< 0.001

FD: Functional dyspepsia; CI: Confidence interval.

DISCUSSION

FD is generally considered a benign condition[1]. Endoscopic evaluation often shows non-specific gastritis or normal mucosa. Helicobacter pylori testing and eradication are recommended in all patients. Rome IV criteria classify FD into postprandial distress syndrome (PDS), characterized by postprandial fullness and early satiety, and epigastric pain syndrome (EPS), defined by epigastric discomfort without prandial association[6]. Lifestyle modifications and PPIs remain the mainstay of treatment, with agents like amitriptyline and prokinetics showing benefits in refractory cases[7-9]. PPIs have proven to be more effective in EPS, whereas prokinetics have a favorable impact in PDS cases[10]. Current guidelines generally advise against routine gastroscopy in patients younger than 60 years with dyspepsia, even in the presence of alarm features, citing low diagnostic yield relative to cost and procedural risk in this population[9,11,12]. There is growing evidence about the overlap of FD and GERD, with PDS the predominant FD type in these syndromes[13].

BE is a precancerous lesion that can progress to esophageal adenocarcinoma. Several risk factors, including male sex, advanced age, Caucasian ethnicity, significant family history, GERD, tobacco use, and obesity, are known to increase the likelihood of BE[14]. Despite the evidence of FD and GERD overlap, literature exploring the association between FD and the development of BE is scarce. Our large, multicenter study found a significant association between FD and both BE and Barrett’s-associated dysplasia. This effect was observed despite adjusting for GERD and baseline PPI use via propensity matching, and remained unchanged in the sensitivity analysis. Risk of EC was not statistically significant in the sensitivity analysis. This could be explained by detection bias, where people undergoing EGD had premalignant lesions identified and managed earlier, leading to a decrease in EC cases. Alternatively, the decrease in statistical power from decreased sample size may also have contributed to the observed results.

As opposed to the well-established GERD-BE association, the exploration of FD’s potential association with the likelihood of BE, independently or in the context of FD-GERD overlap, in the current study is a novel addition to the literature. The underlying pathophysiological mechanisms are unclear. A possible explanation is increased acid exposure time due to the impaired gastric accommodation and delayed gastric emptying seen in FD[6]. Also, FD and BE may possibly share unknown predisposing factors such as autonomic nervous system dysregulation, genetic predisposition, and environmental influences that alter gut physiology. Moreover, some of the results can be explained by undiagnosed GERD cases in the FD cohort, as the availability of objective reflux testing data was limited. Regardless, our results indicate that FD may be a potential contributor to BE risk, either mechanistically or through overlap and misclassification with GERD.

Importantly, our findings carry clinical relevance in identifying higher-risk populations among patients labeled with FD. Our large, multicenter, propensity-matched analysis demonstrates a measurable prevalence of BE and Barrett’s-associated dysplasia among patients coded with FD. While these risks may substantially reflect diagnostic overlap or misclassification with GERD, this very overlap underscores the clinical relevance of our findings: Patients labeled with FD may include a subgroup in whom objective reflux testing and selective endoscopic evaluation can identify reflux-related pathology at rates higher than previously appreciated. This highlights the need to refine risk stratification strategies in FD, with future work focusing on identifying high-risk subpopulations - particularly those with additional risk factors such as older age, male sex, obesity, chronic symptoms, or features suggestive of reflux - who are most likely to benefit from targeted diagnostic evaluation. Follow-up studies can assess BE risk in EPS and PDS sub-groups for a more comprehensive characterization of this association. Additionally, further research is needed to explore the pathophysiology underlying this relationship and guide targeted therapy.

The strengths of our study include the large sample size and propensity score matching to adjust for confounding. However, our study is limited by its retrospective observational design and inability to establish causality. It relies on accurate diagnostic coding and cannot stratify results by different FD subtypes. Although GERD was included in propensity score matching, it is important to note that some patients with a diagnosis of FD may have had undiagnosed or overlapping GERD. While diagnoses are often established following objective reflux testing, such testing data are not available through TriNetX. Further prospective studies are needed to validate these findings and identify specific FD subpopulations at risk for BE.

CONCLUSION

In conclusion, our study demonstrated a significant association between the diagnosis of FD and increased risk of BE and Barrett’s-associated dysplasia. Given that diagnostic overlap with GERD is likely substantial in real-world FD coding, our findings suggest that patients labeled with FD represent a higher-risk population that may warrant consideration for objective reflux testing and risk-stratified endoscopic evaluation, particularly in those with additional risk factors for BE. These results support further study into identifying specific high-risk subgroups within the FD population who would benefit most from targeted diagnostic strategies.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the West Virginia Clinical and Translational Science Institute for providing access to the TriNetX research network.