Prevalence of Barrett's esophagus in systemic sclerosis

Authors


Abstract

Objective

The esophagus is the most commonly affected gastrointestinal area in systemic sclerosis (SSc). Patients with SSc frequently develop gastroesophageal reflux, esophageal injury, and sometimes, intestinal metaplasia, or Barrett's esophagus (BE), which may increase the risk of esophageal adenocarcinoma. This study sought to determine the prevalence of BE and esophageal adenocarcinoma in a cohort of SSc patients.

Methods

One hundred ten SSc patients who were receiving long-term treatment with proton-pump inhibitors (PPIs) were assessed systematically by esophageal manometry and endoscopy. Esophageal biopsies were performed when macroscopic abnormalities were detected, and BE was diagnosed histologically.

Results

Among the 110 patients, 14 had BE (12.7%). None of the patients with BE had adenocarcinoma, but 3 of the 14 patients (21%) had dysplasia on esophageal biopsy. Similar proportions of patients with and without BE had abnormal peristalsis and decreased lower esophageal sphincter pressure. No association between BE and other disease characteristics was demonstrated.

Conclusion

In this study, 12.7% of SSc patients who had been treated with PPIs for long periods had BE, similar to the prevalence in patients with gastroesophageal reflux disease. Although none of the patients had esophageal adenocarcinoma, patients with BE should be followed up closely, particularly patients with dysplasic BE. Long-term prospective studies are warranted to determine the phenotype of SSc patients at high risk of developing dysplasia or esophageal adenocarcinoma.

Systemic sclerosis (SSc) is a connective tissue disease characterized by early vascular involvement, which culminates in excessive deposition of collagen in the skin and internal organs. The gastrointestinal tract is the second most commonly involved site (1). The most frequently involved area of this tract is the esophagus, affecting 75–90% of patients with SSc (1–4). Damage is prominent in the distal two-thirds of the esophagus, resulting in abnormal esophageal motility. Manometry reveals a dysfunctioning lower esophageal sphincter and diminution or abolition of esophageal peristalsis, with poor clearance of acidic material following reflux (3). This increases the time of contact between acid and the esophageal mucosa, thus creating a predisposition to esophageal injury.

The pathophysiologic mechanism of esophageal involvement is thought to depend on early vascular impairment, as indicated by the relationship between aperistalsis and Raynaud's phenomenon (5), followed by neurologic abnormalities that lead to muscular disturbance (3, 6). Regardless of its etiology, chronic gastroesophageal reflux can be complicated by a columnar-lined epithelium with goblet cells, known as Barrett's esophagus (BE). This complication affects between 5% and 15% of patients with chronic gastroesophageal reflux disease (GERD) (7, 8). Furthermore, BE increases the risk of esophageal adenocarcinoma (9).

Only a few studies have determined the prevalence of BE and esophageal adenocarcinoma in SSc, and the results are contradictory (10, 11). The aims of our present cross-sectional study were to use systematic endoscopy to investigate the prevalence of BE and esophageal adenocarcinoma, and to try to identify a disease phenotype associated with these lesions.

PATIENTS AND METHODS

All patients who were hospitalized consecutively with a diagnosis of SSc in accordance with the classification criteria established by the American College of Rheumatology (formerly, the American Rheumatism Association) (12) and who were followed up systematically during a 2-year period were included. All patients gave their informed consent for all procedures. The following clinical data were collected: age, sex, cutaneous SSc subtype according to the definition by LeRoy et al (13), duration of disease (date of first non-Raynaud's symptom), duration of Raynaud's phenomenon, digital ulceration, and prostacyclin use. Pulmonary fibrosis was diagnosed on the basis of a computed tomography scan and abnormal results on respiratory function tests (forced vital capacity, and diffusing capacity for carbon monoxide divided by alveolar volume, with abnormal defined as <75% of predicted values) (14). Pulmonary arterial hypertension was defined as a pulmonary arterial pressure higher than 40 mm Hg at rest on Doppler echocardiography. The following biologic tests were carried out: standard blood tests, immunofluorescence on HEp-2 cells for antinuclear and anticentromere antibodies, and counterimmunoelectrophoresis for anti–topoisomerase I and anti-RNP.

Gastroesophageal involvement was screened by manometry to analyze lower esophageal sphincter pressure, with a normal cutoff level of 10 mm Hg (14 cm H2O). Patients were divided into 3 groups according to peristaltic wave propagation: aperistalsis (<10% of waves propagated), abnormal propagation (<50% of waves propagated), and normal propagation. Manometry was performed after a 3-day washout period to eliminate drugs that may interfere with esophageal motility (cisapride, metoclopramide, calcium channel blockers). All included patients underwent an endoscopy with a Fujinon EG 200FP or 450FP endoscope (Fujinon, Saitama, Japan). In more than 80% of patients, the endoscopy was performed after administration of general anesthesia (propofol), which is standard procedure for all patients in the gastroenterology department.

When macroscopic lesions suggestive of BE, esophagitis, or strictures were detected, several biopsy samples (an average of 4) were collected from each patient. BE was defined by the finding of intestinal architecture in the lower esophagus displaying a villiform columnar-lined mucosa with goblet cells; these features of BE were identified by classic coloration on hematoxylin and eosin staining. Long-segment BE, which is suspected to confer a higher risk of adenocarcinoma, was defined by the presence of lesions >3 cm. We did not take the fundic metaplasia into account. All patients with a normal macroscopic aspect on endoscopy were considered to be free of metaplasia.

In the final 12 months of followup, all patients who underwent endoscopy were tested for Helicobacter pylori. The method for detection of H pylori was an enzyme-linked immunosorbent assay (Helico Premier; Meridian Diagnostic, Cincinnati, OH).

The chi-square test and the Mann-Whitney U test were used to compare data. P values less than 0.05 were considered significant.

RESULTS

One hundred ten patients were included in the study. All patients had been treated with proton-pump inhibitors (PPIs) since receiving the diagnosis of SSc or since the initial commercial marketing of omeprazole in 1989, and had also been treated, when necessary, with prokinetic agents. The characteristics of the SSc patients are summarized in Table 1. Fourteen patients (12.7%) were diagnosed with BE on the basis of histologic findings (Figure 1A). Macroscopic findings were suggestive of BE in 24 patients, but the diagnosis was confirmed in only 14 of the patients. Intestinal metaplasia was found in several biopsy sites in 8 of the 14 patients with BE. None of these patients developed adenocarcinoma. However, 3 of the 14 patients with BE (21%) had dysplasia on intestinal metaplasia; among these patients, 2 had low-grade dysplasia (one-fifth of biopsy sites) and 1 had high-grade dysplasia (high grade in one-fourth [Figure 1B] and low grade in two-fourths).

Table 1. Characteristics of systemic sclerosis (SSc) patients with and without Barrett's esophagus (BE)*
 All SSc patients (n = 110)SSc patients with BE (n = 14)SSc patients without BE (n = 96)
  • *

    Except where indicated otherwise, values are the no. (%) of patients. PPI = proton-pump inhibitor; CT = computed tomography.

  • Abnormal forced vital capacity (FVC) defined as <75% of normal value.

  • Abnormal diffusing capacity for carbon monoxide divided by alveolar volume (DLCO/VA) defined as <75% of normal value.

Age, mean ± SD years58 ± 1253 ± 1359 ± 12
Sex, female/male93/1712/281/15
Tobacco use404
Alcohol use000
PPI dose110 (100)14 (100)96 (100)
 Mean mg/day313327
 20 mg/day65 (59)5 (36)60 (62)
 >20 mg/day45 (41)9 (64)36 (37)
Nonsteroidal antiinflammatory drugs14 (13)3 (21)11 (11)
Platelet antiaggregating33 (30)3 (21)30 (31)
Prednisone <10 mg per day44 (40)6 (43)38 (40)
Calcium channel blockers95 (86)12 (86)83 (86)
Angiotensin-converting enzyme inhibitors35 (32)6 (43)29 (30)
Diffuse/limited cutaneous form of SSc45 (41)/65 (59)6 (43)/8 (57)39 (41)/57 (59)
SSc duration, mean ± SD years9 ± 810 ± 99 ± 8
Raynaud's phenomenon duration, mean ± SD years13 ± 1113 ± 1113 ± 11
Pulmonary arterial hypertension18 (16)3 (21)15 (16)
Pulmonary fibrosis on CT scan55 (50)8 (57)47 (49)
Positive antinuclear antibodies79 (72)12 (86)67 (70)
Positive anti–topoisomerase I antibodies32 (29)4 (29)28 (29)
Positive anticentromere antibodies19 (17)3 (21)16 (17)
Decreased FVC27 (25)4 (29)23 (24)
Decreased DLCO/VA44 (40)7 (50)37 (39)
Figure 1.

Histologic pattern of Barrett's esophagus (BE) occurring in 2 patients. A, Biopsy sample of the distal esophagus showing BE without dysplasia. B, Distal esophageal mucosa showing BE with high-grade dysplasia; the intestinal epithelium contains cells with markedly enlarged nuclei. These atypia involve not only the glands, but also the mucosal surface. (Hematoxylin and eosin stained; original magnification × 800.)

Of the patients with dysplasia, all 3 were women who had a limited cutaneous form of SSc and a long duration of Raynaud's phenomenon (duration of 37 years, 21 years, and 15 years). These 3 women had recurrent digital ulcerations, and in 2 of the 3, the ulcerations were treated with prostacyclin. Esophageal manometry showed abnormal propagation in all 3 women and aperistalsis in 2 of them. Sphincter pressure in the lower esophagus was markedly decreased, with levels below 12 cm H2O in all 3 patients.

The manometry results for the whole population are shown in Table 2. In patients with BE compared with those without BE, there was a trend toward a higher frequency of aperistalsis (57% versus 34%, respectively; P not significant [NS]), a lower wave propagation (93% versus 70%, respectively; P NS), and a higher frequency of abnormal (<14 cm H2O) lower esophageal sphincter pressure (64% versus 55%, respectively; P NS) (Table 2). We found a trend toward a higher frequency of abnormal manometry results (abnormal peristalsis or abnormal sphincter pressure in the lower esophagus) in the BE group (100%) than in the non-BE group (76%) (P = 0.09). Patients with BE were significantly less likely to be H pylori–positive than were patients without BE (10% versus 42.5%, respectively; P < 0.05).

Table 2. Results of gastroesophageal assessments in systemic sclerosis (SSc) patients with and without Barrett's esophagus (BE)*
 All SSc patients (n = 110)SSc patients with BE (n = 14)SSc patients without BE (n = 96)
  • *

    Values are the no. (%) of patients.

BE14 (12.7)14 (100)
 Short-segment (<3 cm)13 (12)13 (93)
 Long-segment (>3 cm)1 (1)1 (7)
Dysplasia3 (3)3 (21)0
Esophageal adenocarcinoma000
Manometric aperistalsis41 (37)8 (57)33 (34)
Manometric abnormal peristalsis80 (73)13 (93)67 (70)
Manometric decreased lower esophageal sphincter pressure (<14 cm H2O)62 (56)9 (64)53 (55)
Helicobacter pylori–positive on serology18/50 (36)1/10 (10)17/40 (42.5)

We found no significant difference in the clinical and biologic characteristics between patients with BE and those without BE (Table 1). Notably, patients with and without BE had a similar mean duration of disease (mean ± SD 10 ± 9 years versus 9 ± 8 years, respectively) and mean duration of Raynaud's phenomenon (both 13 ± 11 years), and were equally likely to have the limited cutaneous (57% versus 59%, respectively) or diffuse (43% versus 41%, respectively) forms of SSc.

DISCUSSION

In this series of 110 SSc patients who underwent long-term treatment with PPIs and were analyzed by systematic endoscopy, the prevalence of BE was 12.7%. Since the initial description of an association between BE and SSc by Cameron and Payne in 1978 (15), numerous case reports (16–28) and 3 retrospective series (29–31) (Table 3) have been published. In the earliest retrospective series, the prevalence of BE was estimated to be 37% among patients who were assessed by endoscopy following reports of clinical gastrointestinal symptoms (30), and 16% among patients referred to a gastroenterology department (31). SSc patients who undergo endoscopy because of digestive symptoms may represent the subgroup with the most severe esophageal involvement, thus increasing the prevalence of BE. In the other retrospective series, systematic esophageal manometry and endoscopy were performed in 43 patients, with BE detected in only 1 patient (prevalence of 2%) (29). The discrepancy in the prevalence between the latter study and our results could be related mainly to the fact that the study by Marie et al was not designed to determine the prevalence of BE, but rather to investigate the link between esophageal involvement and pulmonary involvement in SSc.

Table 3. Studies of Barrett's esophagus (BE) in systemic sclerosis (SSc) patients
Study design, author, year (ref.)No. (%) of SSc patients with BENo. (%) of SSc patients with BE and adenocarcinoma
Retrospective
 Recht et al, 1988 (30)10 (37)2
 Weston et al, 1998 (31)9 (16)
 Marie et al, 2001 (29)1 (2)
Cross-sectional
 Present study14 (13)
Case report
 Cameron and Payne, 1978 (15)2
 Agha and Dabich, 1985 (16)2
 Anderson and Seymour, 1987 (17)3
 Chang et al, 1991 (18)1
 Dill, 1983 (19)1
 Halpert et al, 1983 (20)22
 Ilan et al, 1996 (21)11
 Lambert et al, 1987 (22)1
 McKinley and Sherlock, 1984 (23)11
 Maekawa and Nogami, 1993 (24)11
 Navon et al, 1991 (25)1
 Niv et al, 1988 (26)11
 Sprung and Gibb, 1985 (27)3
 Szigeti et al, 2001 (28)5
Total598 (14)

Abnormalities of esophagus motility leading to BE can occur in asymptomatic patients (2, 4, 32); 30–40% of patients with esophageal lesions are asymptomatic (1). Indeed, we did not take the digestive symptoms of our patients into account because they were receiving PPIs, which would thus mask gastrointestinal symptoms. Long-term systematic treatment with PPIs that may also influence the prevalence of BE has been proven to be effective in ameliorating symptoms and macroscopic lesions when erosive esophagitis occurs in GERD (33), and to be effective in reducing symptoms of esophagitis in SSc patients by decreasing the time with pH under 4, as demonstrated by pH-metry (34). However, PPIs have less influence on biliary reflux, and therefore this may account for the generation of BE in our patients.

GERD is a recognized risk factor for BE (30). In our cross-sectional series, the frequency of BE in patients with SSc seemed to be similar to that in patients with GERD, 5–15% of whom are thought to develop BE (7, 8). The typical BE risk factors associated with intestinal metaplasia include male sex, white race, obesity, and alcohol or tobacco use (35). In contrast, 85% of our SSc patients were female, all had a body mass index <30 kg/m2, and the majority did not consume tobacco or alcohol. Although nonsteroidal antiinflammatory drugs are known to decrease the evolution of BE until dysplasia, only 21% of our patients with BE were currently receiving these drugs; therefore, this will need further investigation. It is possible that BE associated with SSc could be a particular entity with a different pathophysiologic mechanism.

The only common risk factor between SSc and GERD patients with BE is the H pylori status. The presence of H pylori is believed to protect against GERD and associated long-segment BE (36–38). In our series, BE patients were less likely to be H pylori–positive than were non-BE patients (10% versus 42.5%, respectively; P < 0.05), confirming the potential protective role of H pylori. The prevalence of H pylori was lower in our SSc population than that demonstrated in previous studies (39); however, the H pylori test was not similar and H pylori prevalence differs according to age and sociocultural conditions (mostly in childhood).

The incidence of esophageal adenocarcinoma has not been clearly established in GERD patients, whereas the risk of carcinoma is elevated in patients with Barrett's metaplasia (9, 40, 41). In fact, the incidence of esophageal adenocarcinoma is estimated to be 0.5% per BE patient-year (35). Strategies for the management of patients with GERD have thus been proposed, suggesting that patients with chronic GERD symptoms should be assessed for detection of BE, and those patients with established BE should be screened by endoscopy every 2 or 3 years, every year if low-grade dysplasia occurs, and every 3 months in the case of high-grade dysplasia (7).

The findings of one study suggested that 8 (14%) of 59 SSc patients with BE had developed esophageal adenocarcinoma (35). However, that study took into account only the patients with the most severe symptoms; therefore, because of publication bias, those findings cannot be extrapolated to the whole SSc population. Among 680 SSc patients with a mean followup of 12 years, Segel et al (11) found only 1 case of adenocarcinoma; the prevalence of BE was not provided in that study. Our results are in accordance with the findings of Segel et al, since none of our patients with SSc (mean followup of 9 years) had esophageal adenocarcinoma. However, given the low incidence of esophageal adenocarcinoma and the number of patients screened, we cannot draw any conclusion about this risk.

Although SSc patients with BE exhibit strikingly different characteristics from those observed in the typical patient with BE, 3 of the 14 patients in our study developed dysplasia, suggesting that BE associated with SSc can evolve in the same way as classic BE. Our patient with high-grade dysplasia had been treated with plasma argon, showing that followup of BE can lead to a specific treatment. Other endoscopic techniques have also been validated for premalignant lesion treatment (42).

Our results did not highlight a specific manometric pattern associated with BE. Manometry is a highly sensitive method for the detection of esophageal injury in SSc (43); this is emphasized by our finding that 80% of the SSc patients screened and 100% of the SSc patients with BE had at least 1 abnormality. However, manometry is less accurate in patients with severe esophageal involvement, such as can occur in patients with BE.

The fact that patients with BE were more frequently receiving a high dose of PPIs (9 of 14, or 64%) than were patients without BE (36 of 96, or 37%) (P NS, because of low effectiveness) may suggest that BE patients could have more severe or more symptomatic esophageal involvement. However, the cross-sectional and observational methods of our study limit our ability to investigate this aspect.

It has previously been suggested that esophageal abnormalities could be more frequently associated with the limited cutaneous subtype of SSc than with the diffuse cutaneous subtype (10). However, other data suggest that it can occur regardless of the cutaneous subtype and regardless of disease duration (44, 45).

In this cross-sectional study of 110 patients with a mean SSc disease duration of 9 years who were systematically treated with PPIs, we found a prevalence of BE of 12.7% and no adenocarcinoma. This prevalence of BE in SSc patients is similar to that found in patients with GERD (7, 8). An increased risk of adenocarcinoma has been demonstrated in GERD patients with BE, and recommendations for detection and followup of these patients are available (7). Whether the same recommendations (i.e., use of periodic endoscopic assessments) should be applied to SSc patients remains to be further validated in long-term prospective studies.

Ancillary