SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

Background : The length of Barrett's oesophagus seems to correlate well with indicators of severe gastro-oesophageal reflux disease. However, it remains unknown whether prior acid suppressive therapy affects the length of newly diagnosed Barrett's oesophagus.

Methods : A retrospective analysis of a well-characterized large cohort of patients with Barrett's oesophagus diagnosed between 1981 and 2000.

Aim : To compare the length of Barrett's oesophagus between patients who received acid suppressive therapy prior to their diagnosis to those who did not receive such therapy. Pharmacy records were obtained from Department of the Veterans Affairs computerized records and prospectively collected research records. We further examined the association between prior use of acid suppressive therapy and the length of Barrett's oesophagus in correlation analyses, as well as multivariate linear regression analyses while adjusting for differences in year of diagnosis, age, gender, ethnicity, and the presence of intestinal metaplasia of the gastric cardia.

Results : There were 340 patients with Barrett's oesophagus first diagnosed between 1981 and 2000. The average length of Barrett's oesophagus at the time of first diagnosis was 4.4 cm (range: 0.5–16). Of all patients, 139 (41%) had prior use of histamine-2 receptor antagonists, or proton-pump inhibitors (41 used both), and 201 (59%) used neither prior to the diagnosis of Barrett's oesophagus. The mean length of Barrett's oesophagus was significantly shorter in patients with prior use of proton-pump inhibitors (3.4 cm) or proton-pump inhibitors and histamine-2 receptor antagonists (3.1 cm) when compared to those with none of these medications (4.8 cm). In the multivariate linear regression model, the prior use of proton-pump inhibitors or either proton-pump inhibitors or histamine-2 receptor antagonists was an independent predictor of shorter length of Barrett's oesophagus (P = 0.0396).

Conclusions : The use of acid suppressive therapy among patients is associated with a reduction in the eventual length of newly diagnosed Barrett's oesophagus with gastro-oesophageal reflux disease. This finding is independent of the year of diagnosis or demographic features of patients. Further studies are required to confirm this finding.


Background

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

Barrett's oesophagus (BE) is conventionally defined as a change in the lining of the tubular oesophagus from normal squamous to specialized intestinal epithelium. The length of BE as seen endoscopically varies from ultrashort (< 1 cm), short segments (< 3 cm), or long segment (≥ 3 cm).1 The length of Barrett's segments has been linked to the risk of oesophageal carcinoma. Longer segments of Barrett's carry a greater risk of developing dysplasia and oesophageal adenocarcinoma.2–4

The length of BE might be related to the severity of the underlying gastro-oesophageal disease.5 For example, it was reported that the amount and duration of oesophageal acid exposure correlate well with the length of BE.6–7 Similarly, the prevalence and size of hiatus hernia is higher among patients with longer segments of BE when compared to subjects with short segment BE or no BE.5, 8

Barrett's oesophagus is an acquired condition and is thought to develop in adulthood as a result of faulty healing of oesophageal mucosa damaged by gastro-oesphageal reflux. The exact time of onset of BE is unknown in the great majority of cases. Only the year of onset of symptomatic reflux can be determined. Several studies reported that there is no or little change in the length of Barrett's segments once patients are diagnosed and are being subjected to repeated follow-up.9 Recent studies suggested a small but measurable decrease in the length of BE in patients treated with high-dose proton-pump inhibitor therapy.10–13 However, no study has examined the effect of antisecretory therapy prior to the diagnosis of BE on the length of Barrett's segments in gastro-oesophageal reflux disease (GERD) patients with newly diagnosed BE.

At the Southern Arizona Department of the Veterans Affairs (VA) Healthcare System, a single experienced endoscopist has been responsible for performing endoscopy and collecting information on prior therapy on newly referred patients with diagnosed or suspected BE over the past 20 years. In this study, we analyse the potential effect of prior antisecretory therapy on the length of Barrett's segments among patients diagnosed between 1981 and 2000.

Methods

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

This is a retrospective analysis of a prospectively characterized cohort of patients with BE diagnosed between 1981 and 2000 at the Southern Arizona VA Healthcare System. BE was defined in all patients by abnormal distal oesophageal lining with intestinal metaplasia (IM) by biopsy on two endoscopies at least 6 months apart. For this study, manual review of all research records was performed. The length of BE in centimetres was determined by measuring from the proximal margin of continuous columnar appearing epithelium in the oesophagus to the end of the tubular oesophagus or the proximal margin of hiatal hernia folds without excess air insufflation.14 The presence of islands of squamous epithelium was not considered in the definition of BE length. The absence of significant oesophageal erosions was required and as two endoscopic examinations were required for BE diagnosis, this condition was fulfilled in all cases.

The following variables were available in the Barrett's cohort on all patients included in this analysis: date of birth, date of first diagnosis with BE, gender, race (white, black, Hispanic, other), the presence of IM of the gastric cardia, and the length of BE at the year of the diagnosis.

In 1994, the pharmacy at the Southern Arizona VA Healthcare System started a computerized database in which information of prescription and dispension of medications have been maintained. Using the VA pharmacy database, we abstracted prescription and dispension data for any H2RA (cimetidine, ranitidine, famotidine, nizatidine), PPI (omeprazole, lansoprazole, rabeprazole; the only PPIs available at this VA) among veterans with newly diagnosed BE. Prior to 1994, and for non-VA patients, the information on the medications was obtained from the research records. The duration of non-overlapping episodes of use for each category of medication was calculated as the number of days for which a medication was prescribed. Patients were categorized as users of PPI, H2RA, or either depending on the presence of at least one prescription of these medications. A sensitivity analysis was conducted in which users of antisecretory therapy were defined by the presence of prescriptions for PPI or H2RA for a minimum of 6 months. Non-users of antisecretory therapy were defined by the absence of any prescription for PPI or H2RA.

Statistical analyses

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

Descriptive and summary statistics were conducted and presented as mean values (and s.d.), median, and proportions. We examined the frequency and the proportions of patients using H2RA, PPI, or either therapy prior to the recorded time of BE diagnosis. To assess whether the proportions of patients with reported use of these medications has increased or decreased during the study period a Cochrane-Armitage test for linear trend was calculated.

To examine the association between the use of antisecretory therapy and the length of newly diagnosed BE, we conducted two-tailed t-tests to compare the mean length of BE according to the receipt of PPI, H2RA, or either therapy prior to Barrett's diagnosis. These comparisons were conducted against the rest of the study population, and also against non-users of any prior antisecretory therapy. We also conducted anova tests to compare the mean length of BE among three mutually exclusive group of patients who received only PPI, only H2RA, or a combination of these either at the same or at different times.

Among patients who used antisecretory medication, we conducted correlation analyses between the length of BE in centimetres and the duration of use of PPI, H2RA, or either therapy prior to BE diagnosis. Spearman's correlation coefficient and the corresponding P-values were calculated.

Multivariate linear regression analyses were performed to examine the association between the length of BE at the time of diagnosis (outcome variable) and the use of acid suppressive therapy (independent variable) while adjusting for several covariates including year of diagnosis, demographic features, and IM of the gastric cardia.

The study was approved by the Institutional Review Board at the Southern Arizona VA Healthcare System.

Results

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

There were 340 patients with documented BE who were first diagnosed between 1981 and 2000 (Table 1). The majority (81%) were veterans who were seen at the VA Medical Center. The mean age at the time of diagnosis was 61 years. Most patients were White (84%), 28 were Hispanic (8%), three African-American (1%), four native American (1%) and 20 (6%) in whom race could not be defined. The majority (95%) were men. Overall, there were 26 cases that had concomitant IM of the gastric cardia (8%).

Table 1.  Patients with newly diagnosed Barrett's oesophagus (BE) between 1981 and 2000 (total 340): demographic features, mean length of BE, and the prior use of antisecretory therapy
Time periodNMean age (s.d.)Mean BE length (s.d.)Median BE lengthWhite (%)Prior PPI use (%)Prior H2RA use (%)Prior PPI or H2RA use (%)
1981–19855462 (10.3)6.1 (3.8)5.047 (87.0)025 (46.3)25 (46.3)
1986–19908459.7 (12.7)5.2 (3.3)5.070 (83.3)2 (2.4)17 (20.2)19 (22.6)
1991–199512162.1 (11.2)3.7 (2.6)3.0103 (85.1)35 (28.9)36 (29.8)51 (42.2)
1996–20008160.6 (12.3)3.6 (2.9)2.065 (80.3)39 (48.2)26 (32.1)44 (54.3)
Total34061.0 (11.3)4.4 (3.0)4.0285 (83.8)76 (22.3)104 (30.6)139 (40.9)

Seventy-six (22%) patients had used PPI prior to BE diagnosis, 104 (31%) had prior use of H2RA, 41 patients received both medications either at the same or different times, thus 139 (41%) had prior use of either medication, and 201 (59%) had neither medication. The proportion of those with prior use of antisecretory therapy increased over time (Table 1). This trend was statistically significant for PPI (P-value for linear trend <0.001) but not for H2RA (P = 0.41). Twenty-one patients used PPI, and 53 used H2RA for >6 months. Of those patients, 6, 14, and one used PPI for 6–12 months, 1–5 years, and longer than 5 years, and 13, 27, and 13 used H2RA for 6–12 months, 1–5 years, and longer than 5 years. The mean duration (s.d.) of using PPI, H2RA, or either prior to the diagnosis of BE were 6740 (1036), 248 (516), and 592 (970) days.

Overall, the mean length of BE was 4.4 cm with a median of 4.0 cm (range: 0.5–16). The mean length (s.d.) of BE among 76 patients who received PPI before diagnosis was 3.2 cm (2.6) when compared with 4.8 cm (3.3) among 264 patients who did not receive prior PPI therapy (received either H2RA or neither H2RA or PPI), P < 0.0001. The mean length of Barrett's at the time of diagnosis among 139 patients who received either H2RA or PPI prior to BE diagnosis was 3.8 cm (2.9) when compared with 4.4 cm (3.3) among 210 patients who used neither PPI or H2RA prior to BE diagnosis, P = 0.003. Similar results in direction and magnitude were found in analyses restricted to veteran patients (n = 276); data not shown.

Table 2 shows an unadjusted comparison among four mutually exclusive groups of patients: those received PPI only (10%), H2RA only (19%), both PPI and H2RA (12%), or none (59%). The mean length of BE at the time of diagnosis was significantly shorter among patients who used PPI than those who used H2RA, while the longest average length of BE was observed in non-users of PPI or H2RA.

Table 2.  The length of newly diagnosed Barrett's oesophagus (BE) in groups of patients categorized based on prior use of antisecretory therapy (n = 139)
 PPI onlyH2RA onlyBoth PPI and H2RANeither H2RA nor PPIP-value (anova)
  1. P-value for anova = 0.0023.

Number of patients (%)35 (10)63 (19)41 (12)201 (59) 
Mean length of BE in cm (s.d.)3.4 (2.1)4.6 (3.0)3.1 (2.9)4.8 (3.3)0.0023
Mean difference in BE length from non-users of antisecretory therapy (P-value for t-test)−1.4 (P = 0.0013)−0.2 (P = 0.5529)−1.7 (P = 0.002)  
Correlation between duration of prior use and length of BE coefficient (P-value)−0.12 (P = 0.31)−0.13 (P = 0.22)−0.17 (P = 0.06)  

Among patients who received antisecretory medications (PPI, H2RA, or either), the relationship between the duration of prior medication use (days) and the length of BE (cm) was examined in a correlation analysis among patients with BE. In general, there was no duration–response relationship. In 76 patients who used PPI (r = −0.12, P = 0.31), or in 92 patients who used H2RA prior to diagnosis (r = −0.13, P = 0.22) (Table 2).

We examined the effect of prior therapy with PPI, H2RA, or either, on the length of BE in three separate multivariate linear regression that adjusted for the year of diagnosis, age at time of diagnosis, gender, race, and the presence of cardiac IM. In all models, gender, race, and the presence of cardiac IM were non-significant predictors of the length of BE, and therefore were removed from the final models. Table 3 illustrates the results of the final model that examined the prior use of either an H2RA or PPI on the length of BE. The prior use of H2RA or PPI was significantly associated with decreased length of BE. The more recent time of diagnosis was significantly associated with shorter BE segments. This variable modelled as four categories of years provided better fitting of the model than that of a continuous variable. The model indicates that each of the prior use of PPI or H2RA and the more recent year of BE diagnosis were independent predictors of shorter length of BE at the time of diagnosis.

Table 3.  Multivariable linear regression analysis of the prior use of antisecretory therapy on the length of Barrett's oesophagus (BE, in cm)
VariableParameter estimateStandard errorP-value
  1. Patients were diagnosed between 1981 and 2000 (n = 340).

  2. * The parameter estimate is interpreted as the adjusted unit change (0.8 cm less) in the outcome (length in cm) per unit change in the predictor (i.e. the use of PPI or H2RA).

Use of PPI or H2RA prior to BE diagnosis−0.836*0.3440.0157
Period of BE diagnosis
 1981–19852.3080.534<0.0001
 1986–19901.3380.4850.0062
 1991–1995−0.0360.4690.9340
 1996–2000Reference  
Age at BE diagnosis0.0060.0150.6009

We conducted a sensitivity analysis in which the prior use of PPI, or H2RA was defined by use that exceeds 6 months. According to this definition, 21 patients had prior PPI use, 53 had prior H2RA use, and 64 had either H2RA or PPI use. Among patients with PPI or H2RA use that exceeded 6 months, the length of BE was significantly shorter than those who used these medications for <6 months (3.8 vs. 4.7 cm, P = 0.004). In general, the associations described above persisted in direction and increased in magnitude. Among PPI users the mean length of BE was 2.2 (1.6) when compared with 4.6 (3.2) in the rest of patients, P < 0.0001. In the multivariable regression model that included age, time of diagnosis, and use of either PPI or H2RA, the latter variable was a significant predictor of shorter BE at the time of diagnosis (Table 4). In a separate model that controlled for the time of diagnosis and age at diagnosis, the prior use of PPI was a significant predictor of shorter BE at the time of diagnosis (parameter estimate: −1.64, S.E.: 0.71, P-value: 0.02).

Table 4.  A sensitivity analysis in which antisecretory therapy is considered only if used for more than 6 months
VariableParameter estimateStandard errorP-value
  1. The effect of prior use of antisecretory therapy on the length of Barrett's oesophagus (BE, in cm) among patients diagnosed between 1981 and 2000 (n = 340) is examined using a multivariable linear regression analysis.

Use of PPI or H2RA prior to BE diagnosis−0.0160. 4320.0193
Period of BE diagnosis
 1981–19852.2920.535<0.0001
 1986–19901.3640.4840.0062
 1991–1995−0.0060.4380.9340
 1996–2000Reference  
Age at BE diagnosis0.0080.0140.5675

Discussion

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References

This study suggests that prior acid suppressive therapy is inversely related to the length of newly diagnosed BE among patients with GERD. The prior use of PPI or H2RA therapy was associated with statistically significant shorter segments of BE among patients newly diagnosed with this condition. The effect of prior antisecretory therapy was independent of the year of BE diagnosis or demographic features. The absence of a duration–length correlation does not support a medication effect, however, our retrospective methods are limited in ascertaining the duration of medication use.

There are limitations to this study related to its retrospective design and the sources of pharmacy data. The ascertainment of therapy might have been incomplete for patients who prefer to use non-VA pharmacies as their source of medications. Given the significantly reduced charges for medications, especially PPI, in VA (up to 10-fold difference when compared with non-VA), it is unlikely that many patients resorted to this option. This study also preceded the availability of OTC PPI. Another limitation is related to the use of prescription and dispension data as surrogate markers for drug intake. Adherence to medications could vary and is unknown in this study. Thirdly, the VA pharmacy at the Southern Arizona VA Healthcare System began keeping electronic records in 1994; prior to that the Barrett's research chart was reviewed for prospectively collected information on patients’ antireflux medications. However, these three limitations are likely to result in non-differential errors rather than a systematic bias to explain why patients with longer BE were found to be taking less medication prior to diagnosis. For example, there is no clear reason why the length of BE should affect patients’ use of VA pharmacy, or the adherence to antisecretory therapy.

This study is not a population-based study. Approximately 20% of cases were referred after diagnosis rather than newly diagnosed by the senior author. Therefore, a referral bias might have accounted for only a part of the observed increase in the frequency of short segment BE over the course of the study. However, if the analysis is restricted to veteran patients (81% of the study population) who were mostly diagnosed by the senior author, the same findings persisted (data not shown). Changes in the diagnostic criteria of BE have occurred throughout the duration of the study. However, the fact that a single endoscopist recorded the measurement prospectively throughout the study period minimizes bias related to observer variability. All cases have been measured in centimetres rather than labelled as long or short BE as previously defined;9 the latter descriptions being more liable to be interpreted differently depending on the prevailing definitions of BE. This served to minimize the effect of changes in the diagnostic criteria for BE. Moreover, the first cases of BE < 3 cm in this series were recorded in 1983, long before the widespread recognition of this entity. Nevertheless, variations in diagnosis and recording practices could have still occurred.

Our study also has several strengths. First, this large sample of BE consisted of all newly diagnosed consecutive patients seen by the same investigator over two decades. Secondly, strict and consistent criteria were used throughout the study period to make the formal diagnosis of BE. This was based on positive histopathology on two consecutive occasions as well as endoscopic findings. Furthermore, the fact that a single endoscopist recorded the measurement throughout the study period minimizes bias related to observer variability. Thirdly, several analyses were conducted to adjust for the possible confounding by the time of diagnosis; which could reflect medication prescription practices, and for demographic features, which could also be associated with difference in health care seeking behaviour. The significant inverse association persisted in multivariate regression analyses that adjusted for several strata of time of diagnosis as well as important demographic features. However, residual confounding might have been present, as we did not control for potentially important variables such as the duration and severity of GERD symptoms, the presence or length of hiatus hernia, or the use of other medications.

It is controversial whether or not in the individual patients there is progression or regression between the different disorders attributed to GERD (non-erosive disease, erosive oesophagitis, oesophageal strictures, BE). There are several epidemiological and clinical features that suggest lack of significant progression from one form of GERD to the other.15–17 Those who develop the most severe form of disease (e.g. erosive oesophagitis or BE) usually present to medical attention with these conditions, rather than presenting with no gross oesophageal damage and gradually developing erosive oesophagitis that finally culminates in BE. However, the information supporting this scheme mostly predates the use of PPIs. The use of profound acid suppression such as that produced by PPIs has the potential of modifying the clinical course of GERD. For example, acid suppression has also been shown to effectively prevent the recurrence of erosive oesophagitis and/or oesophageal strictures.18–19 Anecdotal observations also indicate that the incidence of new oesophageal peptic strictures is declining in recent times. In a study of patients with normalization of intraoesophageal pH, acid suppression has been shown to decrease markers of cellular proliferation, e.g. proliferating cell nuclear antigen (PCNA) and increase markers of differentiation, e.g. villin expression in BE tissue. In the same study, patients with persistently pathological intraoesophageal acid reflux had no change in PCNA or villin expression.20

If validated, the inverse relation between antireflux therapy and the length of BE counters the concept of antisecretory therapy increasing the risk of oesophageal adenocarcinoma; there being at least a trend for longer BE being associated with a greater likelihood of cancer.4 It also provides a rationale for aggressive therapy of GERD. However, future studies are needed to confirm these findings.

References

  1. Top of page
  2. Summary
  3. Background
  4. Methods
  5. Statistical analyses
  6. Results
  7. Discussion
  8. Acknowledgement
  9. References
  • 1
    Sharma P, Morales TG, Sampliner RE. Short segment Barrett's esophagus – the need for standardization of the definition and of endoscopic criteria. Am J Gastroenterol 1998; 93: 10336.
  • 2
    Weston AP, Badr AS, Hassanein RS. Prospective multivariate analysis of clinical, endoscopic, and histological factors predictive of the development of Barrett's multifocal high-grade dysplasia or adenocarcinoma. Am J Gastroenterol 1999; 94: 34139.
    Direct Link:
  • 3
    Weston AP, Badr AS, Hassanein RS. Prospective multivariate analysis of factors predictive of complete regression of Barrett's esophagus. Am J Gastroenterol 1999; 94: 34206.
    Direct Link:
  • 4
    Avidan B, Sonnenberg A, Schnell TG, Chejfec G, Metz A, Sontag SJ. Hiatal hernia size, Barrett's length, and severity of acid reflux are all risk factors for esophageal adenocarcinoma. Am J Gastroenterol 2002; 97: 19306.
    Direct Link:
  • 5
    Avidan B, Sonnenberg A, Schnell TG, Sontag SJ. Hiatal hernia and acid reflux frequency predict presence and length of Barrett's esophagus. Dig Dis Sci 2002; 47: 25664.
  • 6
    Fass R, Hell RW, Garewal HS, et al. Correlation of oesophageal acid exposure with Barrett's oesophagus length. Gut 2001; 48: 3103.
  • 7
    Oberg S, DeMeester TR, Peters JH, et al. The extent of Barrett's esophagus depends on the status of the lower esophageal sphincter and the degree of esophageal acid exposure. J Thorac Cardiovasc Surg 1999; 117: 57280.
  • 8
    Cameron AJ. Barrett's esophagus: prevalence and size of hiatus hernia. Am J Gastroenterol 1999; 94: 20549.
    Direct Link:
  • 9
    Sampliner RE, Garewal HS, Fennerty MB, Aickin M. Lack of impact of therapy on extent of Barrett's esophagus in 67 patients. Dig Dis Sci 1990; 35: 936.
  • 10
    Gore S, Healey CJ, Sutton R, et al. Regression of columnar lined (Barrett's) oesophagus with continuous omeprazole therapy. Aliment Pharmacol Ther 1993; 6: 63228.
  • 11
    Sampliner RE. Effect of up to three years of high dose lansoprazole on Barrett's. Am J Gastroenterol 1994; 89: 184448.
  • 12
    Malesci A, Savarino V, Ventilin P, et al. Partial regression of Barrett's esophagus by long term therapy with high dose omeprazole. Gastrointest Endosc 1996; 44: 7005.
  • 13
    Peters FTM, Ganesh S, Kuipers E, et al. Endoscopic regression of Barrett's oesophagus during omeprazole treatment: a randomized double blind study. Gut 1999; 45: 48994.
  • 14
    Sampliner RE. Detecting, measuring, and managing Barrett's esophagus: the complete endoscopist's dilemma. Gastrointest Endosc 1997; 45: 53335.
  • 15
    El-Serag HB, Sonnenberg A. Associations between different forms of gastroesophageal reflux disease. Gut 1997; 41: 59499.
  • 16
    Sonnenberg A, El-Serag HB. Epidemiology of gastroesophageal reflux disease. In: Büchler, MW, Farthmann, EH, eds. Progress in Surgery. Basel: Karger, 1997; 23: 2036.
  • 17
    Ollyo JB, Monnier P, Frontolliet C, Savary M. The natural history, prevalence, and incidence of reflux esophagitis. Gullet 1993; 3: 310.
  • 18
    Vigneri S, Termini R, Leandro G, et al. A comparison of five maintenance therapies for reflux esophagitis. N Engl J Med 1995; 333: 110610.
  • 19
    Marks RD, Richter JE, Rizzo J, et al. Omeprazole versus H2-receptor antagonists in treating patients with peptic stricture and esophagitis. Gastroenterology 1994; 106: 90715.
  • 20
    Ouatu-Lascar R, Fitzgerald RC, Triadafilopoulos G. Differentiation and proliferation in Barrett's esophagus and the effects of acid suppression. Gastroenterology 1999; 117: 32735.