Background: Gastro-oesophageal reflux disease complications may reflect imbalances between protective and injurious factors. Through its effects on cell growth, leptin may influence oesophageal mucosal homeostasis.
Aims: To determine whether leptin receptors are present in the oesophagus, and whether serum or gastric leptin levels are associated with oesophageal inflammation and metaplasia.
Methods: From patients referred for upper endoscopy, biopsies were obtained from the stomach and distal oesophagus, and serum samples were collected. Patients were classified as having normal, inflamed or Barrett’s oesophagus. Quantitative immunohistochemistry was performed on representative sections, and leptin levels in plasma and gastric biopsy samples were determined by specific immunoassay.
Results: Of 269 individuals enrolled, 105 were Helicobacter pylori-negative. Of the 88 patients with complete oesophageal biopsies, 44 were normal, 24 were inflamed and 20 were Barrett’s oesophagus. Receptors for leptin were highly expressed on oesophageal epithelial cells, with similar density and staining pattern in all three conditions, and plasma and antral leptin levels did not differ significantly. Patients with Barrett’s had significantly (p = 0.01) higher fundic leptin levels (median 202 (interquartile range 123–333) pg/mg) compared with normal (126 (78–221) pg/mg) or inflamed (114 (76–195) pg/mg) oesophagus. In multivariate analysis, for every twofold increase in fundic leptin, the odds of having Barrett’s was 3.4 times (95% CI 1.5 to 7.6) higher compared with having a normal oesophagus.
Conclusions: Leptin receptor expression on oesophageal epithelial cells provides a pathway for leptin-mediated signal transduction. Variation in gastric leptin production could contribute to differential oesophageal healing and metaplasia progression.
Statistics from Altmetric.com
Gastro-oesophageal reflux disease has become a highly prevalent condition in Western populations.1 Although heartburn is a hallmark of gastro-oesophageal reflux disease, its presence alone is not predictive of oesophageal mucosal injury. Gastro-oesophageal reflux disease-associated complications (oesophagitis, Barrett’s oesophagus and oesophageal adenocarcinoma) result from an imbalance between exposure to injurious elements such as acid, pepsin and bile, and oesophageal epithelial protective mechanisms.2
Oesophageal mucosal injury in gastro-oesophageal reflux disease patients is not directly associated with increased gastric acid production,3 but the duration of oesophageal exposure to refluxed gastric contents contributes to the development of gastro-oesophageal reflux disease sequelae.4 Although oesophageal epithelial proliferation and restitution are critical in limiting erosions following acid-pepsin exposure,5 proliferation leading to a metaplastic epithelium increases oncogenic risk. Although salivary epidermal growth factor (EGF) stimulates oesophageal repair,6 the roles of other luminal peptides in oesophageal epithelial homeostasis have not been examined.
Leptin, the 167 amino acid hormone product of the (ob) gene,7 is produced primarily by adipocytes, and plays a role in appetite regulation as well as in energy homeostasis.8 Both leptin and its receptor (ob-R) are expressed by the gastric epithelium.9 Leptin administration inhibits gastric ulcer formation in rats,10 and stimulates growth of oesophageal adenocarcinoma cells11 12; gastric leptin thus may contribute to both mucosal homeostasis and abnormal proliferation. The relationship of leptin to oesophageal pathophysiology has not been studied.
We hypothesised that leptin of gastric origin may participate in the maintenance of the normal (non-inflamed) oesophageal mucosa, or of the more acid-resistant Barrett’s epithelium. The aims of this study were to evaluate whether leptin receptors are present in human oesophageal mucosa in healthy and in diseased tissue, and to determine whether circulating or gastric leptin levels are associated with particular oesophageal pathology. Since gastric colonisation with Helicobacter pylori is inversely associated with gastro-oesophageal reflux disease and its sequelae, and because H pylori status may affect gastric leptin synthesis,13 14 we examined these questions in H pylori-negative persons, since such individuals are at highest risk for developing oesophageal diseases.15–18
Using a convenience sampling method, we prospectively recruited adults ⩾18 years of age undergoing routine upper endoscopy for any indication at the ambulatory endoscopy unit at the New York campus of the VA New York Harbor Healthcare System. Participants were excluded if they had a known coagulopathy, had a history of oesophageal or gastric varices, had prior gastric surgery or if they had been treated with corticosteroids, or other immunomodulating drugs in the month prior to enrolment. To avoid confounding associated with H pylori status, those enrolled individuals with positive results for H pylori by histology, culture, serology or rapid urease testing also were excluded from further analysis. Patients were considered to be H pylori negative if both serological and tissue-based tests were negative. The Institutional Review Board approved the study protocol, and written informed consent was obtained from all participants.
Clinical evaluation and specimen collection
All patients presented after a 12 h overnight fast. Prior to the scheduled endoscopy, a doctor conducted a pre-procedure evaluation including history and physical examination. Demographic and clinical information was collected via a standardised questionnaire administered by trained interviewers at the time of study entry. Participants self-reported ethnic designation as White non-Hispanic, Black non-Hispanic, Hispanic or Asian. Height and weight were recorded for each participant, and body mass index (BMI) was calculated as weight in kilograms divided by height in metres2. Gastro-oesophageal reflux disease was defined as the presence of heartburn or regurgitation,19 occurring at least weekly in the 4 weeks prior to enrolment. Phlebotomy was performed and 15 ml of blood was collected at the time of endoscopy.
A complete endoscopic evaluation was performed in standard fashion with the Olympus GIF-130 or GIF-160 videoendoscope (Olympus America, Melville, NY) after intravenous administration of meperidine and midazolam. Endoscopic oesophagitis was graded according to the LA classification.20 Immediately upon entry into the stomach, approximately 15 ml of gastric secretions were collected into a suction trap for pH measurement. The endoscopic aspect of the Sydney–Houston system was used to describe gastric inflammation.21 Two biopsies were obtained with standard biopsy forceps from the antrum for rapid urease testing (Hpfast, GI Supply, Camp Hill, PA). Four biopsies each were obtained from the gastric antrum, fundus and distal oesophagus, 2 cm above the squamocolumnar junction.
One biopsy specimen from each of the three sites (antrum, fundus and oesophagus) was fixed in 10% formalin, embedded in paraffin and consecutively sectioned at 5 μm for haematoxylin and other histological staining. A single experienced gastrointestinal pathologist (Z.P.) who was blinded to the clinical diagnoses interpreted all specimens. The presence of H pylori was evaluated using the Warthin–Starry technique as well as immunostaining techniques. Oesophageal biopsies were evaluated and scored on a scale of 0 to 3+ for (1) hyperplasia of the basal layer; (2) elongation of the papillae; (3) dilatation of the papillary vascular spaces; (4) intraepithelial infiltrate of eosinophils; (5) neutrophils; and (6) mucosal erosion, as described.22 23 By adding the scores for each of the six evaluated parameters, the composite histological score for each biopsy sample was determined out of a possible 18 points. Oesophageal epithelial inflammation was considered to be present if an acute polymorphonuclear or a mixed polymorphonuclear and mononuclear cell infiltrate was present, along with epithelial erosion or ulceration.24 Barrett’s oesophagus was confirmed histologically by detection of specialised intestinal metaplasia, as described.22 The alcian blue/periodic acid–Schiff (AB-PAS) stain was used to confirm the presence of acid mucin-containing goblet cells in intestinal metaplasia.
Specimens from four representative subjects each with normal, inflamed or Barrett’s oesophagus were used in the immunohistochemistry experiments. Three serial 5 μm sections were obtained from each representative formalin-fixed paraffin-embedded specimen. In brief, samples from each group were simultaneously processed and all experiments were repeated at least twice. The specimens were deparaffinised in stepwise fashion with graded alcohol. Endogenous peroxidase activity was removed and the sections were incubated in a humidified chamber for 30 min with a 1:300 dilution of mouse monoclonal antibody raised against the extracellular domain of the human leptin receptor (R&D Systems, Minneapolis MN), which is present in both long (signalling) and short (non-signalling) forms of the receptor.25 26 Sections were incubated with a peroxidase-labelled polymer conjugated to goat anti-mouse antibody for 30 min, followed by incubation with 3′,3′-diaminobenzidine (DAB) chromogen solution for 10 min (DakoCytomation, Carpinteria, CA). Tissue from the gastric fundus served as a positive control27 for both leptin receptor and leptin immunoreactivity. Omission of the primary antibody was used as a negative control, since there was no signal. Sections were viewed with a Nikon ECLIPSE E600 microscope (Nikon, Japan) using 10×, 20× and 40× objective lenses, and images were acquired with a SPOT INSIGHT™ digital colour camera, model 3.2.0 (Sterling Heights, MI). Quantification of immunoreactivity was performed on digitally captured colour images saved as tiff files and analysed using ImageJ 1.37 (http:/rsb.info.nih.gov/ij).28 After subtracting the non-white background colour, the haematoxylin component was separated from the DAB component using colour deconvolution as described29 (fig 1A–C). Calibration of optical density was performed using a Kodak photographic 21-step grey-scale tablet with a density range of 0.05–3.05. For each specimen, the lowest and the highest mean optical density (MOD) values measured from eight manually identified positively stained cells selected from the DAB component (with visual comparison with the original image) was used to set the threshold manually for whole sample staining density quantification (fig 1D). The percentage area stained and MOD was measured for each image, and comparison of samples from the three oesophageal groups was made using one-way analysis of variance (ANOVA).
Tissue and plasma leptin determination
Biopsy samples measuring 8 mm3 were rinsed in phosphate-buffered saline (PBS) buffer and homogenised. An ELSIA (R&D Systems) was used to measure leptin in the homogenates and concentrations were normalised, based on biopsy protein (pg/mg protein). As extensively described,30 31 the same assay was used to measure fasting serum leptin levels. As indicated by the manufacturers, the well-to-well variation of the assay is 3.0–3.2% and the minimum detectable level is 7.8 pg/ml. All tests were done at least in duplicate.
Endoscopic and histological results allowed for classification of patients into one of three categories for further analysis: no oesophageal pathology (normal), inflamed or Barrett’s oesophagus. Normal epithelium was defined as the absence of visible erosions or salmon-coloured mucosa on endoscopy and no basal cell hyperplasia, elongated papillae or cellular infiltrates on histological examination. Inflamed epithelium was considered present if there were cellular infiltrates as previously described, with or without visible erosions on endoscopy based on the LA classification.20 Barrett’s oesophagus was defined as salmon-coloured epithelium of any length observed at the distal oesophagus on endoscopy and confirmed as specialised intestinal epithelium on histological examination. Integrated gastric leptin was determined by calculating the mean antral and fundic leptin concentrations. In analyses of clinical gastro-oesophageal reflux disease, the American College of Gastroenterology criteria19 were used. Continuous variables were compared using the ANOVA method, Kruskal–Wallis or Mann–Whitney U test, as appropriate. Data are expressed as mean (SD), or medians and interquartile range (IQR; 25th–75th percentile). Categorical variables were compared using the χ2 test with Yates correction or Fisher exact test. Spearman correlation coefficients were calculated for leptin–BMI and leptin–MOD. Multivariate regression analysis was performed on log-transformed data controlling for possible confounders (age, ethnicity, BMI, proton pump inhibitor (PPI) use and gastric pH). Statistical analysis was performed using SPSS software version 11.0.4 for Macintosh (SPSS Inc., Chicago, IL), and a two-tailed p-value of <0.05 was considered statistically significant.
Patient demographic and clinical characteristics
Among the 269 subjects enrolled, 164 (61%) tested positive for H pylori (fig 2) and biopsies were adequate for analysis in 133. Of the remaining 105 H pylori-negative individuals, oesophageal biopsy specimens were not adequate in 17, leaving 88 subjects for this study. Among these, 44 were found to have no oesophageal pathology (normal), 24 had inflamed epithelium and 20 had Barrett’s epithelium. The normal oesophagus group also had no evidence of gastric ulcers or masses. In the inflamed oesophagus group, one patient had LA grade A, and one patient had LA grade D oesophagitis by endoscopy. The prevalence of Barrett’s epithelium (13.6% among the 269 enrolled subjects and 22.7% among the 88 subjects studied) was consistent with other published reports.32 33 The mean length of Barrett’s was 3.8 (2.8) cm. Compared with the 22.7% prevalence of Barrett’s among H pylori-negative subjects, only 7.5% of H pylori-positive subjects had Barrett’s (odds ratio (OR) = 0.28 (0.12–0.62), p = 0.002]. As planned, the H pylori-positive individuals were excluded from further analysis.
The most common indication for endoscopy was haem-positive stool in the normal group, iron deficiency anaemia in the inflamed oesophagus group and evaluation of dysplasia in the Barrett’s group. The three groups did not differ significantly in age, BMI, prevalence of heartburn or gastric juice pH (table 1). Compared with the subjects with normal or inflamed oesophagus, significantly more individuals with Barrett’s epithelium were Caucasian and used PPIs. After excluding all patients receiving PPI therapy, the median gastric pH values did not differ significantly between normal, inflamed or Barrett’s oesophagus; 2.0 (1.0–5.0), 2.0 (1.0–2.5) and 1.0 (1.0–2.5), respectively, p = 0.24. As expected, the median histological score for inflammation was significantly lower in subjects with normal compared with those with inflamed oesophagus (p = 0.004) and Barrett’s oesophagus (p = 0.013), consistent with the spectrum of injury (pairwise ANOVA comparisons).
Leptin and leptin receptor expression
As expected,9 leptin receptor expression was observed in chief and parietal cells within gastric fundus specimens (fig 3B), which served as positive controls for further studies. In the oesophageal epithelium, there was immunoreactivity for the leptin receptor in both superficial and basal layers (fig 1), with enhanced intensity towards the superficial layer, and a diffuse staining pattern in both apical and basolateral membranes. Quantitative immunohistochemistry (fig 3D–F) was performed to compare normal oesophagus, inflamed oesophagus and Barrett’s oesophagus for leptin receptor expression. The mean area of the evaluated specimens, measured in square pixels, did not vary significantly between the three groups, according to oesophageal status (normal, 1 486 984 (107 167); inflamed, 1 377 981 (212 304); Barrett’s, 1 279 999 (261 594) (p = 0.42)). There was a trend towards a higher mean percentage area showing leptin receptor positivity in the Barrett’s epithelium samples (11% (2%)) compared with normal epithelium (5% (2%)) and inflamed epithelium (4% (1%)) (fig 3D–F), although the difference was not statistically significant (p = 0.19). The leptin receptor immunohistochemical MOD did not differ significantly (p = 0.67) between the normal (0.68 (0.04)), inflamed (0.68 (0.08)) and Barrett’s epithelium (0.76 (0.08)). In total, these results indicated that among the three conditions evaluated, the leptin receptor is expressed in all layers of the oesophagus with similar density. In parallel studies, using leptin-producing gastric tissue as a positive control (fig 3A), we examined whether the oesophageal epithelium was a source of leptin. There was no immunoreactivity to indicate leptin protein presence in either normal, inflamed or Barrett’s oesophagus. This result demonstrated that although leptin receptors are expressed on the apical and basolateral surfaces of the oesophagus, their ligand, leptin, originates elsewhere.
Plasma leptin, heartburn and oesophageal pathology
Since the risk of gastro-oesophageal reflux disease has been linked to obesity,34 and adipose tissue fat stores are reflected by plasma leptin,35 we next addressed whether plasma leptin might be related to either symptomatic or histological evidence of reflux disease. As expected,35 in the 88 study subjects, plasma leptin was significantly correlated with BMI (r = 0.59, p<0.0001). The median leptin level (pg/ml) trended non-significantly higher among PPI users (5562 (IQR 1921–8467)) than among those not using PPIs (2355 (IQR 1195–5770); (p = 0.06)), Mann–Whitney U test. Plasma leptin levels were not associated with the presence, severity or duration of heartburn symptoms. The distribution of plasma leptin also was not significantly different between the subjects with normal (median 5136 (IQR 1653–7103) pg/ml), inflamed (3155 (IQR 1969–7057) pg/ml) or Barrett’s oesophagus (3130 (IQR 742–8142) pg/ml; (p = 0.67)) (fig 4). From these results, we concluded that although, as anticipated, plasma leptin reflected fat stores, levels did not correlate with symptoms or pathology related to gastro-oesophageal reflux disease.
Gastric leptin, heartburn and oesophageal pathology
Since leptin receptors are present in the distal oesophagus but there is no endogenous leptin production, we addressed whether levels of leptin of gastric origin were correlated with gastro-oesophageal reflux disease and its sequelae. Although the presence of small amounts of leptin has been detected in gastric secretions,36 we were not able to measure basal leptin reproducibly in gastric juice. Since most leptin-producing cells in the stomach are located in oxyntic glands,9 we compared leptin levels in an oxyntic gland region (fundus), and also examined the pyloric gland region (antrum) as a control. Antral leptin was marginally correlated with BMI (r = 0.23, p = 0.024), whereas fundic leptin showed a stronger but still moderate correlation with BMI (r = 0.38, p<0.001); neither antral nor fundic leptin levels differed significantly according to PPI use (table 2).
The presence, severity and duration of heartburn symptoms were not associated with gastric fundus or antrum leptin levels. However, the integrated gastric leptin (fundus and antrum) was strongly correlated with mean optical density for leptin receptor expression in the oesophagus (r = 0.66, p = 0.039) (fig 5). There were no significant differences in antral leptin levels in subjects with normal, inflamed or Barrett’s oesophagus (p = 0.46) (fig 4). However, the patients with Barrett’s oesophagus had significantly higher fundic leptin (pg/mg) levels (median 202 (IQR 123–333)) compared with the normal (126 (IQR 78–221)) or inflamed oesophagus groups (114 (IQR 76–195)), p = 0.01 (fig 4). This difference persisted when the data were adjusted for patient age, ethnicity, BMI, PPI use and gastric pH (table 3). The goodness-of-fit test indicated an acceptable model, with χ2 = 143.8, p = 0.67. These data may be interpreted as showing that for every twofold increase in fundic leptin, the odds of having Barrett’s oesophagus was 3.4 times (95% CI 1.5 to 7.6) higher compared with having a normal oesophagus (table 3). From these results, we concluded that gastric leptin levels are correlated with oesophageal leptin receptor expression (fig 5), and that fundic leptin levels correlate with risk of Barrett’s oesophagus.
The refluxing of gastric contents into the oesophagus that occurs in all individuals37 38 can damage the oesophageal epithelium and potentially cause symptoms and long-term injury. However, the consequences of oesophageal exposure to acid and pepsin also depend on the defences of the mucosal epithelium,39 including the intercellular tight junctions restricting acid diffusion into the intercellular space.40 Compromise of these junctions can influence the intercellular and cytosolic buffering capacity of the epithelial tissue and consequent cell viability.41 Epithelial repair in the oesophagus depends primarily on cellular replication,42 stimulated by growth factors binding to basal cell receptors after they gain access to the intercellular space via compromised tight junctions.43 Salivary EGF and other peptide growth factors present in the oesophageal lumen have been postulated to play roles in repair of oesophagitis and Barrett’s metaplasia, but studies have yielded contradictory results.5 6 44 45
In the present study, we addressed the association of distal oesophageal mucosal pathology with levels of leptin, a peptide hormone involved in diverse cellular and metabolic functions that influence energy homeostasis and tissue repair.8 46 Chief cells in the gastric mucosa, among many other sites, secrete leptin.9 36 47 Leptin receptors appear in all gastrointestinal organs between 7 and 9 weeks of gestation during fetal development, and mucosal expression persists in the normal adult oesophagus,48 stomach,9 small bowel49 and colon.50
Since leptin is also a growth factor for gastric,51 small bowel,52 colonic53 54 and oesophageal epithelial cells in culture,11 it may have similar proliferative physiological roles in vivo. That both endogenous leptin and exogenously administered leptin protect rats against mucosal injury induced by non-steroidal anti-inflammatory drugs (NSAIDs) or alcohol55 56 supports the hypothesis that leptin may play a protective role for the gastrointestinal mucosa. Despite these cytoprotective effects,55 56 leptin can also induce neoplastic cell proliferation11 57; these seemingly disparate findings suggest that the physiological role of leptin may depend on the context of other host or environmental factors.
We now show that leptin receptors are present in all layers of the epithelium of the adult distal oesophagus, regardless of whether the tissues were normal, inflamed or had columnar cell metaplasia. These findings suggest that differential leptin receptor expression is not a major factor associated with the development of the observed pathology, but indicate that oesophageal cells, whether normal or injured, may have a pathway to transduce signals from leptin molecules present in the oesophageal lumen. Since current antibodies do not distinguish between the long (signalling) and short (non-signalling) forms of the leptin receptor, future studies will need to address this question.
We considered that the systemic effects of leptin may extend to the oesophageal epithelium. Plasma leptin levels reflect fat stores,13 35 and positive associations have been reported between BMI and both Barrett’s oesophagus58 and oesophageal adenocarcinoma.59 60 Most of the subjects we studied were overweight with no significant BMI differences between the three test groups, nor significant plasma leptin differences. As such, our findings did not support any association between plasma leptin levels and mucosal conditions in the distal oesophagus. Since leptin is produced and secreted in the stomach, and has been found in gastric refluxant,36 we next addressed whether differential gastric epithelial leptin levels correlate with oesophageal tissue status. To minimise bias we excluded H pylori-positive subjects in the current study since we have found that antral leptin levels tend to be lower in H pylori-positive persons compared with those who are H pylori-negative; 58 (25–157) pg/ml versus 94 (58–178) pg/ml; p = 0.04.61 The patients with Barrett’s oesophagus had significantly higher gastric fundus leptin levels than those with either normal or inflamed epithelium, suggesting that the combination of exposure of the distal oesophagus to refluxed acid and high leptin could predispose to mucosal proliferation. The associations we observed for gastric leptin do not reflect BMI, since BMI values were similar in the three groups, but rather represent local effects. The notion of leptin acting both systemically and locally is consistent with data showing dissociation between gastric mucosal and plasma leptin levels.13 14 The data suggest a model in which leptin-induced cellular proliferation may help repair oesophageal epithelial inflammation, or may lead to progression of Barrett’s oesophagus, depending on host context62 (fig 6).
An alternative possibility is that gastric oxyntic leptin levels are elevated secondary to the development of Barrett’s metaplasia, independent of BMI. This explanation would require that Barrett’s epithelium: (1) is a source of leptin; (2) stimulates the fundic epithelium to produce more leptin; or (3) tissue events in Barrett’s and fundic epithelium both are reflective of independent primary processes. While our cross-sectional study was not designed to address these possibilities, that leptin was not detected in the Barrett’s mucosa provides evidence against the first possibility. Favouring the second possibility is that nitric oxide synthase overexpression in Barrett’s epithelium63 increases nitric oxide, a known stimulant of chief cells,9 64 the primary sources of gastric leptin. Paracrine communications between the oesophageal and gastric epithelia should be further explored.
Our findings are limited by study placement at a veterans medical centre where most of the evaluated patients were older men receiving multiple medications, limiting generalisability, and use of a cross-sectional study design that limits the inferences that can be reached. The majority of our Barrett’s patients were Caucasian, consistent with the epidemiology of the disease in the USA.65 The three groups were not completely matched, since significantly more Barrett’s patients were receiving PPI therapy. Furthermore, the majority of subjects in the inflammation group had microscopic inflammation as opposed to macroscopic oesophagitis. The strengths of the study include the collection of detailed demographic and clinical data from a prospectively enrolled cohort, ascertainment of H pylori status using multiple criteria which improves sensitivity66 and diminishes confounding due to false-negative assays, biopsy of all patients regardless of clinical or endoscopic findings, use of quantitative immunohistochemical techniques and the inclusion of a group with neither inflammation nor Barrett’s oesophagus.
In conclusion, our study indicates that leptin receptors are expressed in the distal oesophageal epithelium, regardless of disease status, implying responsiveness to leptin-mediated signal transduction. That Barrett’s metaplasia and oesophageal inflammation were differentially associated with gastric leptin levels is consistent with a model in which oesophageal pathology develops along distinct inflammatory (oesophagitis) or proliferative (Barrett’s) pathways (fig 6) as proposed by Fass and Ofman,67 rather than along a progression from inflammation to metaplasia. Other studies also support a bifurcated pathogenetic pathway following acid reflux.68 69 In any event, our results suggest that modulation of gastric leptin may play a role in the progression to Barrett’s metaplasia.
Supported in part by K23CA107123 and R01GM63270 from the National Institutes of Health, and by a Clinical Investigator Research Grant from the American College of Gastroenterology. We thank the NYU School of Medicine Division of Gastroenterology, its fellows, and nurses for their support.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.