Introduction

Barrett's esophagus (BE) is a chronic condition that is characterized by the replacement of the normal squamous epithelial lining of the lower esophagus with specialized columnar epithelium containing goblet cells1. This metaplastic change is considered a precursor of esophageal adenocarcinoma (EAC). The incidence of EAC in patients with BE is much higher than that in the general population2,3. Studies on time trends have demonstrated that the prevalence of BE and incidence of EAC are increasing3,4. The pathogenesis of BE remains poorly understood but is thought to involve a complex interplay of genetic, environmental, and lifestyle factors. Gastroesophageal reflux disease (GERD), central obesity, advanced age, male sex, tobacco use, and Caucasian race are well-known risk factors for BE development5,6,7,8. Because patients with BE have a high risk of developing EAC, timely identification and vigilant monitoring of BE are important for preempting the onset of EAC and ensuring a favorable prognosis. Hence, a thorough analysis of the risk factors linked to the onset of BE is imperative to advance our understanding of its pathogenesis, as well as to formulate precise prevention and intervention strategies.

Overweight/obesity is a medical condition characterized by the excessive accumulation of adipose tissue, manifesting as a metabolically unhealthy state that is associated with an elevated risk of developing a spectrum of diseases9. Several studies have shown that individuals with overweight/obesity or increased waist circumference have a higher risk of developing BE5,8. In a study have found no association between BE incidence and overweight/obesity10. The discordant results indicate an intricate interplay between overweight/obesity and the risk of BE. The metabolically unhealthy (MU) phenotype, defined as the presence of insulin resistance, hypertension, and dyslipidemia, is commonly observed in individuals with overweight/obesity11. Notably, emerging evidence indicates that MU is also closely associated with chronic inflammation and increased risk of developing several diseases, including BE12,13.

Recently, the concept of metabolically healthy overweight/obesity (MHO) has been proposed to describe those who have increased adiposity but do not have traditional cardiometabolic risk factors, such as hypertension, type 2 diabetes mellitus, and dyslipidemia, in contrast to individuals with metabolically unhealthy overweight/obesity (MUO)11,14. Existing evidence suggests that MHO represents a benign condition with a reduced risk of developing diseases, although conflicting findings have also been reported15,16,17,18. However, to date, the association between the body mass index (BMI)-metabolic status phenotype and development of BE has not been investigated. Therefore, in the present study, we evaluated the relationship of overweight/obesity and MU with the incidence of BE by using biobank data from the UK.

Methods

Study population

The UK Biobank, an ongoing national cohort study, recruited more than 500,000 individuals aged 40–70 years from 2006 to 2010 across 22 assessment centers in England, Wales, and Scotland. At baseline, the participants completed self-reported touchscreen surveys, interviews, and physical assessments, and data on demographics, lifestyle, health factors, and anthropometrics were collected. Biological samples were also collected for analysis19. All participants provided informed consent. The study protocol was approved by the U.K. North West Multicenter Research Ethics Committee. Data access for this study was approved by the UK Biobank Access Committee (Application 73873). This study was conducted in accordance with the ethical principles of the Declaration of Helsinki.

Among the 502,419 participants, those with a history of BE or esophageal cancer (n = 175), and gastric cancer (n = 177) were excluded. We further excluded 97,985 participants without metabolic indicators (n = 224,205) and missing BMI data (n = 1574). A total of 402,510 participants were included in the final analysis (Fig. 1).

Fig. 1
figure 1

Flow diagram of the study subjects.

Full size image

Data collection and measurements

During participant recruitment, a detailed touchscreen questionnaire was used to collect information on medical history, lifestyle habits, and sociodemographic characteristics. The questionnaire included a self-reported medical history of diseases such as hypertension (I10–13 and I15), type 2 diabetes mellitus (E10–14), and gastroesophageal reflux disease (K21), which were assessed based on self-reported information or International Classification of Diseases-10 (ICD-10) codes. Smoking and alcohol consumption habits were categorized as 'never' or 'ever.' Annual household income was divided into four categories: (1) less than €18,000; (2) €18,000 to €30,999; (3) €31,000 to €51,999; and (4) more than €52,000. Physical activity was measured in metabolic equivalent task (MET) minutes per week and categorized into three groups (< 600, 600–3000, and > 3000 min/week) based on walking, moderate activity, and vigorous activity. Trained staff performed physical measurements of the participants. Height and weight were measured using a Seca 202 height meter while the participants were barefoot. BMI was calculated as weight (kg) divided by height squared (m2). Blood pressure was measured using an Omron 705 IT electronic sphygmomanometer.

After overnight fasting, venous blood samples were collected to measure hemoglobin, creatinine, total protein, albumin, glucose, total cholesterol, triglycerides, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and high-sensitivity C-reactive protein (hs-CRP) levels. Laboratory parameters were measured using established protocols and standardized equipment from Beckman Coulter (UK). Further details on the methodology and specific protocols can be obtained from the UK Biobank website (https://biobank.ctsu.ox.ac.uk/showcase).

Assessment of metabolically unhealthy and overweight/obesity

In this study, "metabolically healthy (MH)" was defined as the absence of all of the following metabolic indicators20: (1) High blood pressure (≥ 130/85 mmHg), diagnosis of hypertension, or use of anti-hypertensive medications; (2) Elevated fasting blood sugar (≥ 125 mg/dL), diagnosis of type 2 diabetes, or use of anti-diabetes medications; (3) High triglycerides (≥ 150 mg/dL) or use of fibrate medications; (4) HDL-C levels ≤ 40 mg/dL for men and ≤ 50 mg/dL for women, or use of lipid-lowering medications. Conversely, "metabolically unhealthy (MU)" status was defined as having at least one of the above risk factors. Overweight/obesity was defined based on the World Health Organization’s BMI cut-off of ≥ 25.0 kg/m2. The validity of these definitions was further supported by the study results. As shown in Fig. 2, the risk of incident BE increased significantly in participants with one or more metabolic indicators and BMI ≥ 25.0 kg/m2. Consequently, the participants were categorized into four different groups: (1) metabolically healthy non-overweight/obesity (MHNO, BMI < 25.0 kg/m2 without MU), (2) metabolically unhealthy non-overweight/obesity (MUNO, BMI < 25.0 kg/m2 with MU), (3) metabolically healthy overweight/obesity (MHO, BMI ≥ 25.0 kg/m2 without MU), and (4) metabolically unhealthy overweight/obesity (MUO, BMI ≥ 25.0 kg/m2 with MU).

Fig. 2
figure 2

Hazard ratios for the occurrence of Barrett’s esophagus according to according to body mass index and number of metabolic factors. In adjusted model, covariates including age, sex, race, smoking, alcohol habits, and history of gastroesphageal reflux disease. No, number; CI, confidence interval.

Full size image

Outcome assessment

The primary outcome of interest was BE occurrence as defined by the 10th revision of the International Classification of Diseases (ICD-10) code (K22.7). Outcome assessments were performed from enrollment to final follow-up. The follow-up period was determined based on data availability in the UK Biobank and defined as the period from the date of study enrollment to the date of the last available data for each participant, which was February 28, 2021, for participants from England and Scotland and February 28, 2018, for participants from Wales.

Statistical analysis

The baseline characteristics were summarized using descriptive statistics (mean, standard deviation [SD], count, percentage, median, and inter-quartile range [IQR]). Differences between continuous and categorical covariates according to the BMI (≥ 25 vs. < 25 kg/m2) and metabolic status were compared using ANOVA, chi-square test, or Kruskal–Wallis test. The cumulative incidence of BE according to the BMI and metabolic status was estimated using the Kaplan–Meier method and compared using the log-rank test. Survival time was defined as the time from enrollment to the onset of BE. The scaled Schoenfeld residual method was used to verify the proportional hazards assumption21.

The risk of BE development was assessed using multivariate Cox proportional hazard regression models: Model 1 (crude risk) without adjustment; Model 2 was adjusted for age, sex, race, smoking, and alcohol consumption; and Model 3 was further adjusted for history of GERD in addition to the Model 2 covariates. We evaluated the proportional hazards assumption by examining log (−log [survival]) plots of the survival function. In Model 4, we performed a separate analysis using a logistic regression model with inverse probability of treatment weighting (IPTW) derived from the baseline covariates included in Model 3 to address potential selection bias and confounding22. After IPTW adjustment, the maximum pairwise standardized difference for any variable was < 0.1, indicating good covariate balance (Table S1). The results of multivariate Cox proportional hazards regression and IPTW models are presented as hazard ratios (HRs) and 95% confidence intervals (CIs). Differences in the risk of BE among the four groups were further assessed using the Bonferroni method23. We also performed a sensitivity analysis using a BMI cutoff value of 30.0 kg/m2 to test the robustness of the primary results. All the statistical analyses were performed using Stata version 14.2 (StataCorp, College Station, TX, USA). Statistical significance was set at p < 0.05.

Results

Demographic and clinical characteristics

Baseline characteristics according to the BMI and/or metabolic status are shown in Table 1. Of the 402,510 participants, 39,281 (9.7%), 92,000 (22.9%), 25,297 (6.3%), and 245,932 (61.1%) were classified as MHNO, MUNO, MHO, and MUO, respectively. The mean age of the participants was 56.6 years and 53.6% were men. The prevalence of hypertension, type 2 diabetes, and GERD was 33.5%, 5.6%, and 7.3%, respectively. Regardless of the BMI, individuals with MU were older and had significantly higher blood pressure, unfavorable lipid profiles, and higher levels of inflammatory markers than those with MH. These differences were more pronounced in the MUO group compared to the other group. However, the differences in the baseline characteristics between the MUNO and MHO groups were not significant.

Table 1 Baseline characteristics of study patients in four phenotypes classified by BMI and/or metabolic status.
Full size table

Body mass index, metabolically unhealthy, and risk of Barrett’s esophagus

First, we assessed the independent association of BMI and metabolic status with the risk of BE development. During 5,374,032.7 person-years of follow up, BE occurred in 6,195 (1.5%) individuals. The overall incidence rate was 1.2 per 1000 person-years (Table 2). As expected, the incidence of BE was significantly higher in individuals with a BMI ≥ 25 kg/m2 and those with MU. Multivariable Cox proportional analyses revealed a 1.35-fold increased risk (HR 1.35; 95% CI 1.31–1.48) for individuals with BMI ≥ 25 kg/m2 and 1.43-fold increased risk (HR 1.43; 95% CI 1.31–1.56) for those with MU (Model 3). These associations remained consistent even after adjusting for potential confounding factors using IPTW (Model 4).

Table 2 Hazard ratios for the incidence of Barrett’s esophagus according to the body mass index category and metabolic status.
Full size table

Risk of Barrett’s esophagus according to the presence of the four metabolic phenotypes

We further analyzed the risk of incident BE according to a BMI cutoff value of 25.0 kg/m2 and/or metabolically unhealthy. The incidence rates of BE in the MHNO, MUNO, MHO, and MUO groups were 0.5, 0.9, 0.9, and 1.4 per 1000 person-years, respectively (Table 3). The MUO group had a significantly higher incidence rate than the other groups (all p-values < 0.001). Interestingly, the incidence rates of BE in the MUNO and MHO groups were similar (p-value = 0.91). Similarly, the cumulative incidence rate of BE was significantly higher in the MUO group than in the other groups (all p < 0.001) (Fig. 3). However, the incidence rates of BE in the MUNO and MHO groups were not significantly different (p-value = 0.88). In multivariate Cox regression analysis, the MUO group had a 2.62–fold (95% CI 2.32–2.96) increased risk of BE development compared to the MHNO group (Model 1). This risk remained significantly elevated at 1.88-fold (95% CI 1.67–2.13) even after adjusting for confounding factors (Model 3). Both the MUNO and MHO groups exhibited similar risks of BE development compared to the MHNO group, with HRs of 1.73 (95% CI 1.51–1.97) and 1.78 (95% CI 1.51–2.09), respectively (Model 1). After adjustment for confounding factors, the risk of incident BE was 1.45-fold (95% CI 1.27–1.66) and 1.57-fold (95% CI 1.33–1.84) higher in the MUNO and MHO groups, respectively (Model 3). These findings were consistent with the analysis using IPTW to minimize the influence of confounding factors (Model 4). Intergroup comparison with multiple Bonferroni corrections also revealed that the risk of BE development was consistently higher in individuals with MUO than in the other groups (all p < 0.001) (Table 4). Importantly, the risk of BE development did not differ significantly between the MHO and MUNO groups: (HR, 1.08; 95% CI 0.95–1.22; corrected p = 1.000). These results indicate that a higher BMI and metabolic unhealthy status independently increase the risk of BE, and their combined effect is synergistic.

Table 3 Hazard ratios for the incidence of Barrett’s esophagus among 4 metabolic subtypes classified by the BMI cut off 25.0 kg/m2 and/or metabolically unhealthy.
Full size table
Fig. 3
figure 3

Cumulative incidence rate for Barrett’s esophagus according to metabolic phenotypes. MHNO, metabolically healthy non-overweight/obesity; MUNO, metabolically unhealthy non-overweight/obesity; MHO, metabolically healthy overweight/obesity; MUO, metabolically unhealthy overweight/obesity.

Full size image
Table 4 Statistical adjustments with multiple comparisons for risk of Barrett’s esophagus according to the body mass index category and metabolic status.
Full size table

Sensitivity analysis

To assess the robustness of our primary findings, we conducted several sensitivity analyses. First, we used a different BMI cutoff value (≥ 30.0 kg/m2), finding that individuals with BMI ≥ 30.0 kg/m2 exhibited a 1.11-fold increased risk of BE development (HR 1.11; 95% CI 1.05–1.18) (Table S2). The risk of BE was significantly higher in the MUNO, MHO, and MUO groups than in the MHNO group (Table S3). Furthermore, after statistical adjustment using Bonferroni method, MUNO and MHO demonstrated a similar risk of BE development (HR 1.12; 95% CI 0.88–1.43; corrected p = 1.000) (Table S4). We also defined obesity using the waist-to-hip ratio (WHR) and found consistent results (Tables S5, S6). Additionally, we examined differences according to sex and BMI-metabolic status phenotype (Table S7). The results of these sensitivity analyses consistently supported our primary findings, emphasizing the synergistic effect of higher BMI and metabolic unhealthy status on the risk of incident BE.

Discussion

The present study investigated the association among overweight/obesity, MU status, and incidence of BE using data from the UK Biobank database. The major findings were as follows: (1) Study participants were categorized into four BMI-metabolic status phenotypes, including the MHNO (9.8%), MUNO (22.9%), MHO (6.3%), and MUO (61.1%), which were compared for the risk of developing BE during follow-up. (2) Individuals who were overweight/obese had a significantly greater risk of developing BE in both the MHO and MUO groups than in the MHNO group. (3) Interestingly, individuals in the MUNO group were also at an increased risk of BE compared with those in the MHNO group, and those in the MUO group had the highest risk of developing BE, indicating that overweight/obesity and MU both synergistically contribute to the occurrence of BE.

According to the 2016 World Health Organization statistics, being overweight affects up to 1.9 billion adults aged 18 years and older, with over 650 million individuals classified as obese24. This surge in the global prevalence of obesity is closely associated with MU and has emerged as a pressing public health issue. Overweight/obesity, especially central obesity, has been reported to be a risk factor for the development of BE5,8,25. The association between overweight/obesity and BE is multifaceted, involving mechanisms related to GERD, hormonal signaling, chronic inflammation, microbial dysbiosis, and inadequate immune response5,26,27,28,29. Central obesity amplifies intragastric pressure, disturbs normal sphincter function, and delays gastric emptying, culminating in a higher propensity for GERD and increased risk of BE5,26,28. Moreover, adipose tissue, particularly visceral fat located in the abdomen, secretes various hormones and inflammatory substances. These hormones, including leptin and adiponectin, can affect the function of the lower esophageal sphincter and contribute to the development of GERD or BE27,30,31. These findings suggest a potential indirect pathway connecting increasing adiposity to the development of BE29. In contrast, some studies have shown that overweight/obesity has an unclear role in BE development in patients with GERD10. Although various theories have suggested that overweight/obesity may increase the risk of BE, it is still difficult to draw reliable conclusions. In particular, the relationship between BMI, but not central obesity, and the risk of BE remains controversial. In this study using data from the UK Biobank database, individuals with overweight/obesity (both BMI ≥ 25 kg/m2 and ≥ 30 kg/m2) exhibited a higher risk of developing BE than those with normal weight.

The components required for identifying MU are not clearly defined yet, and vary among studies. Regarding the contribution of metabolic status in determining the risk of BE, this study found that the presence of even a single metabolic risk factor was associated with an increased incidence of BE. Metabolic syndrome is a cluster of common pathologies, and we defined MU status as the presence of one or more metabolic risk factors. Interestingly, the risk of BE was significantly higher in individuals with MUNO than in those with MHNO. Moreover, the highest risk of developing BE was observed in the MUO group, further emphasizing that MU has a substantial impact on BE and should not be overlooked.

In general, overweight/obesity and impaired metabolism are tightly linked phenotypical traits, and MU is highly prevalent in obese individuals32. Although previous studies have shown that central obesity is related to the occurrence of BE, the role of MU, independent of overweight/obesity, in the development of BE remains unclear. A notable finding of our study was that MU was significantly associated with an increased risk of developing BE, independent of being overweight/obese. Inflammation emerges as a plausible reason for this finding, with its potential involvement in both the initiation and advancement of BE, by contributing to its pathogenesis and progression13,33,34. Given the detrimental effects of inflammation on cholesterol metabolism, insulin resistance, and vascular remodeling, MU may reflect a preclinical hyper-inflammatory state that predisposes individuals to BE35,36. This notion is supported by our observation that the presence of MU was associated with a higher hs-CRP level, a marker of systemic inflammation, in both individuals with BMI < 25.0 kg/m2 and those with BMI ≥ 25 kg/m2. However, the precise nature of this association has not been explored, necessitating further research to elucidate the underlying mechanisms.

While overweight/obesity and metabolic risk factors commonly exhibit interconnectedness, they may also coexist independently in certain individuals, and their clinical implications can differ based on the BMI-metabolic status phenotypes. Notably, a distinct subtype of overweight/obesity known as MHO has received significant attention due to its relatively favorable clinical outcomes regarding obesity-related diseases, such as cardiovascular disease37. Furthermore, not all individuals with a lean physique exhibit a healthy metabolic profile, and the concept of MUNO has been proposed to explain the heterogeneous nature of overweight/obesity37. Individuals with MHO and MUNO exhibit insulin sensitivity, lipid pathways, and inflammatory profiles that are opposite to those predicted for the overweight/obesity status. Although there is sufficient evidence that MHO represents a relatively benign state with an attenuated risk of developing various illnesses, discordant research outcomes have been reported15,16,17. This suggests the need for a more comprehensive investigation into how MU and overweight/obesity distinctly impact various diseases. However, little is known about the BMI-metabolic status phenotypes in populations with BE. Intriguingly, the present study demonstrated that the risk of BE was significantly greater in individuals with MUNO and MHO than in those with MHNO, suggesting that in the presence of overweight/obesity, metabolic health may not be a benign phenotype with respect to the development of BE.

The results of our study underscore the importance of metabolic health and maintaining an optimal body weight to mitigate the risk of developing BE. Both metabolic disorders and obesity can increase the risk of developing BE. Hence, proactive initiatives targeting weight reduction, consistent physical activity, and adherence to a nutritious diet are advisable as preventive measures against the onset of BE. Given that body weight, exercise, and diet modification can positively influence metabolic health38, the beneficial effect may also extend to individuals who are not overweight or obese. We further evaluated the independent association of BMI and metabolic status with the risk of BE, adjusting for confounding factors including Total metabolic equivalent task (MET) in addition to model 3 (data not shown). The results remained consistent with our initial finding, therefore, maintaining physical activity is highly beneficial to prevent BE.

Our study has some limitations. First, we could not confirm a causal relationship between MU and BE because of the observational nature of the study. Second, the primary outcome of our study was based on ICD codes. Thus, issues regarding diagnostic accuracy and overestimation or underestimation of BE incidence can occur in such a large-scale population-based study. However, the incidence rate of BE reported in this study (6,195/402,510, 1.5%) does not contradict prior estimates of BE incidence (average estimate of 0.5–2% in the general population)4,39,40. Third, BMI and metabolic conditions can change overtime, potentially influencing the risk of developing BE in different directions. The present study did not analyze possible changes in the BMI-metabolic status phenotypes during the follow-up period owing to the unavailability of relevant data. It is challenging to clearly distinguish whether the MHO group has independent characteristics or if they are a group at high risk of progressing to MUO. However, this is a cross-sectional study, we can observe the participants' status at the time of study enrollment. Fourth, MU is a complex condition comprising diverse clinical manifestations such as increased blood pressure, insulin resistance, dyslipidemia, diet and systemic inflammatory conditions. However, we could not independently assess the specific contributions of these components to BE development. These components may act together to promote an inflammatory environment that predisposes individuals to BE. Fifth, additional analyses based on certain epidemiologic factors may be required. The UK Biobank includes individuals from diverse ethnic backgrounds, and BMI classifications for Asians differ from those of other populations. Although we conducted a subgroup analysis according to ethnicity, the results were inconclusive (data not shown). This may be due to the challenge of obtaining detailed ethnic information within groups that include Asians and other ethnicities, as well as the low incidence of BE within these groups. Additionally, while the overall trend shows similar risk increases in both sexes (Table S7), the significant interaction suggests that the magnitude of these effects differs by sex (data not shown). This indicates that although both males and females experience an increased risk of BE with weight and metabolic abnormalities, the extent of this increase is influenced by the interaction between sex and phenotypes. These findings highlight the importance of considering sex-specific factors in the analysis and interpretation of BE risk factors. Further research is needed to confirm these findings and to explore the biological or behavioral reasons behind these sex and ethnic differences. Finally, the study population was limited to individuals in the UK, and the UK Biobank study has been criticized for "healthy volunteer" selection bias41. Therefore, our findings may not be generalizable to other populations.

In conclusion, our results showed that both overweight/obesity and MU are independently associated with an increased risk of BE. Furthermore, the risk of BE was significantly higher in individuals with both overweight/obesity and MU than in those with either overweight/obesity or MU alone. Our findings provide valuable insight into the complex relationship between overweight/obesity and MU, leading to the development of BE, and may help to guide future efforts to prevent and manage BE.