Validation of the Prague C&M classification of Barrett’s esophagus in clinical practice

• Introduction
• Methods
• Setting
• Patient selection
• Study design and randomization
• Endoscopists
• Endoscopic procedure
• Outcome parameters
• Statistical analysis
• Results
• Baseline characteristics
• Barrett’s esophagus and hiatal hernia length
• Landmarks
• Subanalyses for Barrett’s length and endoscopic expertise
• Discussion
• References

Background and study aims: The Prague C&M classification for Barrett’s esophagus has found widespread acceptance but has only been validated by Barrett’s experts scoring video sequences. To date, validation has been lacking for its application in routine practice during real-time endoscopy. The aim of this study was to evaluate agreement between Barrett’s experts and community hospital endoscopists when using this classification to describe Barrett’s esophagus and hiatal hernia length during real-time endoscopy.

Patients and methods: Patients underwent two consecutive endoscopies performed by different endoscopists. The study was performed in two cohorts: one cohort was seen by Barrett’s experts and the other cohort by community hospital endoscopists. Landmarks were recorded according to the Prague classification. Outcomes were interobserver agreement (assessed with intraclass correlation coefficient [ICC]), absolute agreement, and relative agreement.

Results: A total of 187 patients were included, with median extent of C3M5 (IQR C1 – 7 M4 – 9) for Barrett’s esophagus and 3 cm (IQR 2 – 5) for hiatal hernia length. ICC was 0.91 (95 % confidence interval [CI] 0.88 – 0.93) for maximum length, 0.92 (95 %CI 0.90 – 0.94) for circumferential extent, and 0.59 (95 %CI 0.49 – 0.68) for hiatal hernia length. Absolute agreement within ≤ 1 cm was 74 % (95 %CI 68 – 80) for circumference, 68 % (95 %CI 62 – 75) for length, and 63 % (95 %CI 56 – 70) for hiatal hernia length. Relative agreement was 91 % for Barrett’s esophagus and 80 % for hiatal hernia length. Barrett’s experts and community hospital endoscopists showed no differences in agreement. Shorter Barrett’s segments (≤ 5 cm) had lower agreement compared with longer segments (> 5 cm).

Conclusions: Agreement was good for Barrett’s esophagus and reasonable for hiatal hernia length. These findings strengthen the value of the Prague C&M classification to describe Barrett’s esophagus and hiatal hernia length. Although absolute agreement during real-time endoscopy was high, one should anticipate that Barrett’s values may vary by 1 – 2 cm between two endoscopies.

Introduction

Barrett’s esophagus is a condition of the distal esophagus in which the normal squamous epithelium has been replaced by columnar-lined epithelium containing intestinal metaplasia on biopsy [1]. This condition is a premalignant condition in which esophageal adenocarcinoma can arise through a multi-step transition sequence [2]. Because of this risk, patients with Barrett’s esophagus are offered periodic surveillance endoscopies with biopsies [1] [3] [4].

In addition to assessing the Barrett’s segment for neoplastic lesions, it is of clinical relevance to adequately document the Barrett’s length in a reproducible manner during surveillance endoscopies. Previous studies have suggested that longer Barrett’s segments are at higher risk of developing neoplasia [5] [6] [7] [8] [9]. Assessment and registration of Barrett’s length might therefore be helpful in the future for cancer risk stratification, for example by lengthening or shortening the surveillance intervals according to the length of the Barrett’s segment. Furthermore, the Barrett’s length is also important when planning endoscopic therapy, as well as for a more objective evaluation of the response to endoscopic therapy.

Recently, the Prague C&M classification has been introduced as an easy and uniform way to describe the length of the Barrett’s segment during endoscopy [10]. In this classification, the circumferential extent (C) and maximum length (M) of the Barrett’s epithelium are described in cm. Since its introduction, the Prague C&M classification has found widespread clinical application. However, this classification has only been validated through the scoring of video sequences by endoscopists with a special interest in Barrett’s esophagus. To date, there are no available data validating the use of the classification during real-time endoscopy in routine practice.

The aim of this study was to evaluate, in a post hoc analysis, the agreement of Barrett’s esophagus and hiatal hernia length according to the Prague C&M classification by Barrett’s expert endoscopists and community hospital endoscopists assessed in real time during two consecutive endoscopic procedures.

Methods

Setting

A post hoc analysis was performed by analyzing the data from two previously published randomized crossover trials comparing standard video endoscopy (SVE) with endoscopic tri-modal imaging (ETMI; an endoscopy system combining high-resolution endoscopy, autofluorescence endoscopy, and narrow-band imaging) for the detection of neoplasia in Barrett’s esophagus [11] [12]. The design of the studies required that all patients underwent two endoscopies on two separate occasions by different endoscopists.

The first study enrolled 87 patients with Barrett’s esophagus who were referred for endoscopic work-up of endoscopically inconspicuous high grade intraepithelial neoplasia (HGIN) or early cancer. The procedures in this study were performed at five international tertiary referral centers by eight endoscopists with expertise in imaging and therapy of Barrett’s neoplasia.

The second study included 100 patients with a confirmed histological diagnosis of low grade intraepithelial neoplasia (LGIN). These procedures were performed at eight community hospitals in the Amsterdam region by nine endoscopists without specific expertise in Barrett’s esophagus imaging.

Both trials were approved by the Institutional Review Boards of the participating centers (ISRCTN68328077 and ISRCTN91816824).

Patient selection

Inclusion criteria for both trials were: 1) age ≥ 18 years; 2) prior diagnosis of Barrett’s esophagus, defined as the presence of columnar-lined epithelium in the tubular esophagus with specialized intestinal metaplasia on histological examination; 3) written informed consent. The study population in the tertiary referral centers consisted of patients with HGIN or early cancer (amenable for endoscopic therapy) in a Barrett’s segment extent of at least C ≥ 2M ≥ 2 or C < 2M ≥ 4 (Group I). The study population in the community hospitals consisted of patients with any Barrett’s length with a prior diagnosis of LGIN confirmed by an expert gastrointestinal pathologist (Group II).

Exclusion criteria for both trials were: 1) presence of an advanced neoplastic lesion at the first endoscopy that precluded a 6-week delay in intervention (i. e. the minimum interval between the two crossover endoscopies); 2) presence of active erosive esophagitis grade B or worse according to the Los Angeles classification of erosive esophagitis at the first endoscopy; 3) presence of conditions precluding histological sampling of the esophagus (e. g. esophageal varices, coagulation disorders, anticoagulant therapy).

Study design and randomization

All patients underwent two consecutive upper endoscopies. One procedure was performed using SVE (GIF-140, GIF-160 or GIFQ-160, Olympus, Hamburg, Germany) and the other procedure used ETMI (XGIF-Q240 /260FZ, Olympus, Tokyo, Japan); all of the endoscopes had 5-cm interval markings on the shaft. The two endoscopic procedures were separated by an interval of 6 – 16 weeks and were performed by two different endoscopists who were blinded to the results of any previous endoscopies (i. e. study endoscopies as well as any previous endoscopy before the study). Endoscopists were assigned to the first procedure before randomization to SVE or ETMI. Randomization of the technique was performed at the first endoscopy by opening a sealed opaque envelope. The second procedure was automatically performed by another endoscopist using the alternative endoscopy technique.

Endoscopists

Prior to the start of the study, all endoscopists received a training DVD containing instructions on how to use the Prague C&M classification. Endoscopists did not receive explicit instructions regarding the point at which the location of the landmarks should be recorded (i. e. when advancing or withdrawing the endoscope). Also no explicit instructions were given on the deflation of the stomach to record the location of the gastric folds.

Patients in Group I were seen by eight endoscopists with extensive expertise in endoscopic imaging and treatment of Barrett’s esophagus (Barrett’s experts). These eight expert endoscopists formed four fixed pairs. Patients in Group II were seen by nine different endoscopists without extensive expertise in Barrett’s esophagus (community hospital endoscopists). No fixed pairs of community hospital endoscopists were formed.

Endoscopic procedure

In all procedures, patients were sedated according to the standard protocol of the participating centers, which was mostly intravenous midazolam supplemented with fentanyl or pethidine if necessary. According to the Prague C&M classification, distance from the incisors to the diaphragmatic pinch, the top of the gastric folds, and the circumferential extent and maximum length of the Barrett’s segment were prospectively recorded at both endoscopic procedures for each patient. During ETMI, these data were collected in the high resolution white light mode. The length of the Barrett’s segment was the distance between the top of the gastric folds and the most proximal circumferential Barrett’s extent (C) and between the top of the gastric folds and the most proximal (i. e. maximum) Barrett’s length (M). Hiatal hernia length was the distance between the diaphragmatic pinch and the top of the gastric folds.

Outcome parameters

The primary outcome parameter was the agreement of Barrett’s circumferential extent and maximum length and hiatal hernia length, defined as:

• interobserver agreement for circumference, length, and hiatal hernia length

• absolute agreement for circumference, length, and hiatal hernia length

• and relative agreement for circumference, length, and hiatal hernia length.

The secondary outcome parameters were the interobserver agreement and absolute agreement for the different landmarks (diaphragmatic pinch, top of the gastric folds, most proximal Barrett’s circumferential extent, and most proximal Barrett’s maximum length) measured as the distance from the incisors in cm.

Subanalyses of data were performed to find differences in interobserver agreement, absolute agreement, and relative agreement for Barrett’s extent between shorter Barrett’s segments (mean length ≤ 5 cm) and longer Barrett’s segments (mean length > 5 cm), and for differences in interobserver agreement, absolute agreement, and relative agreement for circumference, length, and hiatal hernia length between Barrett’s experts and community hospital endoscopists.

Statistical analysis

Statistical analysis was performed using SPSS software version 16.0.2 for Windows (SPSS Inc., Chicago, Illinois, USA). For descriptive statistics, the mean and SD were used for variables with a normal distribution, and the median and interquartile range (IQR) were used for variables with a skewed distribution. Categorical data were compared using the chi-squared test (the Fisher’s exact test was used when expected cell values were too small) or were tested for trend. Continuous data that were normally distributed were compared using the (paired) t test and analysis of variance when multiple variables were tested. Continuous data that were not normally distributed were compared using the Mann–Whitney U test. Due to multiple testing, the Bonferroni correction was applied and a P value of ≤ 0.01 was considered to be significant.

Interobserver agreement was assessed using the intraclass correlation coefficient (ICC), with an ICC value of 1.00 for perfect agreement and an ICC value of 0 for poor agreement. Comparison of the ICC between two groups was performed with the two-sample independent t test on OpenEpi statistics (http://www.openepi.com/OE2.3 /menu/openEpiMenu.htm). ICC is mathematically defined as the ratio of the variance between the raters (i. e. the consecutive endoscopists) and the total variance of the study population. Thus, with a given variance between endoscopists and a small variance in Barrett’s length between patients, the ICC will be low, while the same variance between endoscopists in a group of patients with a wider range of Barrett’s lengths will result in a higher ICC. Therefore we also report the absolute agreement and the relative agreement.

Absolute agreement for each variable (circumferential extent, maximum length, and hiatal hernia length) was calculated for 0 cm discrepancy, ≤ 1 cm discrepancy, and ≤ 2 cm discrepancy between the two endoscopies. Confidence intervals (CIs) and differences with their CIs for absolute agreement were calculated with the Confidence Interval Analysis package [13]. As the ICC value may overestimate the agreement because of the wide range of Barrett’s values and hiatal hernia length within the patient population, and as absolute agreement is not related to the length of the Barrett’s segment, the relative agreement was created in an attempt to overcome these drawbacks. Relative agreement was calculated for each variable (circumferential extent, maximum length, and hiatal hernia length) by dividing the concordance between both endoscopies by the total length of the variable (mean of both endoscopies) for each patient. Concordance was calculated as:

([length on first endoscopy + length on second endoscopy] – difference between the endoscopies) ÷ 2

For example if the first endoscopy resulted in M = 6 cm, and the second in M = 4 cm, the difference between the two endoscopies would be 2 cm and concordance would be calculated as ([6 cm + 4 cm] – 2 cm) ÷ 2 = 4 cm. Relative agreement in this case would result in 4 cm (concordance) ÷ 5 cm (mean of both endoscopies). In cases where both endoscopies showed a length of 0 cm for a variable, a relative agreement of 100 % was set.

Finally, Bland – Altman plots were performed for each variable to visualize the relation between the mean length of each variable and its difference [14].

Results

Baseline characteristics

Baseline characteristics are shown in [Table 1]. Compared with Group I, Group II included slightly younger patients, with a shorter Barrett’s segment and lower grades of neoplasia, all of which reflected the selection criteria of the study.

Barrett’s expert endoscopists performed a mean of 22 procedures (range 15 – 38) and community hospital endoscopists perfomed a mean of 21 procedures (range 11 – 37).

Barrett’s esophagus and hiatal hernia length

Overall interobserver agreement and absolute agreement defined as a discrepancy of 0 cm, ≤ 1 cm, and ≤ 2 cm between the two endoscopies are shown in [Table 2]. Overall interobserver agreement was almost perfect for circumferential extent and maximum length of Barrett’s esophagus, and moderate for hiatal hernia length. Overall absolute agreement, defined as a discrepancy of 0 cm, varied between 29 % and 39 %, increasing to 63 % – 74 % and 84 % – 91 % for a discrepancy of ≤ 1 cm an  ≤ 2 cm, respectively ([Table 2]). Overall median relative agreement was 91 % (IQR 75 % – 100 %) for circumferential extent, 91 % (IQR 82 % – 100 %) for maximum length, and 80 % (IQR 67 % – 100 %) for hiatal hernia length. Bland – Altman plots for the three variables are shown in [Fig. 1].

No systematic differences in the mean Barrett’s C&M length or hiatal hernia length were seen when comparing the first and second endoscopic procedures (P = 0.473, 0.550, and 0.754 for Barrett’s C&M length, and hiatal hernia length, respectively) or when comparing the procedures performed with SVE and ETMI (P = 0.049, 0.069, and 0.210, respectively). Also, no difference in the mean Barrett’s C&M length and the hiatal hernia length were seen when comparing the first 50 procedures with the last 50 procedures (P = 0.054, 0.057, and 0.343, respectively ). Finally, no systematic differences were seen in Group I between the four pairs of endoscopists (P = 0.378, 0.151, and 0.027 for Barrett’s C&M length and for hiatal hernia length, respectively) or in Group II between the nine observers (P = 0.907, 0.974, and 0.837, respectively).

Landmarks

Interobserver agreement and absolute agreement for landmark location are shown in [Table 3]. Interobserver agreement for the location of the circumferential extent, maximum length, and top of the gastric folds was almost perfect, and for the diaphragmatic pinch location the agreement was substantial. Absolute agreement, defined as a discrepancy of 0 cm was 26 % – 32 %; when defined as a discrepancy of ≤ 1 cm absolute agreement was 56 % – 75 %, and when defined as a discrepancy of ≤ 2 cm it was 79 % – 95 % ([Table 3]).

Subanalyses for Barrett’s length and endoscopic expertise

For shorter Barrett’s segments (mean length ≤ 5 cm), the interobserver agreement for Barrett’s C&M length was significantly lower compared with longer Barrett’s segments (mean length > 5 cm) ([Table 4]). Relative agreement for the maximum Barrett’s length was significantly lower in shorter Barrett’s segments compared with longer Barrett’s segments: 86 % (IQR 80 % – 100 %) and 93 % (IQR 88 % – 100 %), respectively (P = 0.002). In contrast, when testing for trend, absolute agreement for maximum Barrett’s length was higher in shorter Barrett’s segments ([Table 5]).

Comparison of endoscopists showed no differences in interobserver agreement between expert endoscopists (Group I) and community hospital endoscopists (Group II) for Barrett’s C&M length and hiatal hernia length ([Table 6]). Absolute and relative agreement between these two groups were also not significantly different.

Discussion

In this study, the use of the Prague C&M classification during real-time endoscopy was associated with good agreement (i. e. interobserver agreement, absolute agreement, and relative agreement) for Barrett’s circumference and length and a reasonable agreement for hiatal hernia length. No differences in agreement between expert Barrett’s endoscopists and community hospital endoscopists were observed. These findings strengthen the value of the Prague C&M classification to describe Barrett’s segment length in clinical practice.

Interobserver agreement (i. e. ICC) during real-time endoscopy was found to be almost perfect (ICC 0.91 – 0.92) for Barrett’s C&M length and moderate for hiatal hernia length (ICC 0.59). These results are surprisingly similar to previous studies showing ICC 0.92 – 0.94 for Barrett’s esophagus and ICC 0.44 for hiatal hernia length [10] [15]. In these previous studies, however, video sequences were scored either by endoscopists with a special interest in Barrett’s esophagus or by Asian rather than Western endoscopists [10] [15]. The present study included Western endoscopists with and without special expertise in Barrett’s esophagus and was performed during real-time consecutive endoscopies at different endoscopy sessions. The assessing endoscopists were also the ones who actually performed the endoscopy; they controlled the amount of air insufflation during endoscopy (which is especially important when defining the position of the upper end of the gastric folds) and decided when to record the location of the landmarks (either when advancing or withdrawing the endoscope). Furthermore, the assessment of the location of landmarks was based on the 5-cm interval markings on the endoscope rather than the 1-cm intervals used in video sequences.

Given the abovementioned factors, a lower interobserver agreement for the C&M classification was expected compared with the standardized evaluation of video sequences. Although these results seem to support the validity of the C&M criteria, one could argue that the ICC may not be the correct method for the assessment of interobserver agreement of the C&M classification, as mentioned in the methods section. The ICC value in this study may be artificially high due to the wide range of Barrett’s lengths in the included patients (circumference range 0 – 18 cm, length range 1 – 18 cm).

The hypothesis that the ICC in this study may overestimate the agreement of the Barrett’s measurements is supported by the absolute agreement, which was low for Barrett’s C&M length values when 0 cm discrepancy was allowed. In only 39 % (circumferential extent) and 31 % (maximum length) of the patients did the two endoscopists completely agree on the values. The absolute agreement increased to 74 % and 68 %, respectively, when a discrepancy of up to 1 cm was allowed, and to 89 % and 91 % for a discrepancy of ≤ 2 cm. These percentages are only slightly lower than those found in the original validation study using video sequence scoring: 38 % and 53 % for 0 cm discrepancy, 82 % and 88 % for ≤ 1 cm discrepancy, and 95 % and 97 % for ≤ 2 cm discrepancy [10]. Absolute agreement during real-time endoscopy is thus high but it should be anticipated that Barrett’s values may vary by 1 – 2 cm between two different endoscopies.

From a clinical perspective, a difference of 1 – 2 cm between two endoscopists may be less relevant when evaluating long Barrett’s segments than shorter segments. The limits of agreement of the Bland – Altman plots do not increase with increasing length of the variable (Barrett’s C&M length or hiatal hernia length). In addition, absolute agreement is not related to the length of the Barrett’s segment. The relative agreement was calculated by dividing the concordance between the two consecutive endoscopies by the mean length of the variable of interest (i. e. Barrett’s C&M length or hiatal hernia length) for each patient. In accordance with the interobserver agreement and absolute agreement, the relative agreement was also high for Barrett’s values (91 %) and slightly lower for hiatal hernia length (80 %). Previous studies have shown that hiatal hernia length varies considerably over time [16]. This phenomenon might account for the relatively lower agreement in hiatal hernia length found in the study. Additionally, respiratory movements and the amount of insufflation also have an impact on the length of the measured hiatal hernia length.

When comparing short Barrett’s segments (≤ 5 cm) with longer Barrett’s segments (> 5 cm) ambiguous results were found: long Barrett’s segments had a significantly higher ICC but a lower absolute agreement ([Table 4] and [Table 5]). The relatively higher ICC value probably reflects the difference in variance of Barrett’s length in both groups as explained above: a higher ICC was found for long Barrett’s segments, which had a wider variance (6 – 18 cm), whereas a lower ICC was found for shorter Barrett’s segments, which had a smaller variance (1 – 5 cm). The relative agreement, however, supports the ICC value as it also shows a significantly higher relative agreement in maximum length for longer Barrett’s segments. Therefore it seems that endoscopists agree less on C&M length values in shorter Barrett’s segments.

When comparing Barrett’s experts and community hospital endoscopists, no differences were found in interobserver agreement, absolute agreement or relative agreement despite the differences in patient population between the groups. In Group I, patients were included with Barrett’s C&M length of > 2 cm, whereas Group II consisted of patients with any length of Barrett’s esophagus. This may have influenced the ICC of the community hospital endoscopists in two ways: ICC may have been overestimated, as Group II had a wider range of Barrett’s length; or the ICC may have been underestimated, as small Barrett’s segments have shown to have a lower interobserver agreement [10] [15]. Nevertheless, the results suggest that endoscopists without specific Barrett’s expertise are also able to use the Prague C&M classification adequately.

This study is not without limitations. First, as mentioned above, the variance of Barrett’s length in the study population has an impact on the calculated values of the ICC. Second, the patient populations evaluated by Barrett’s experts and community hospital endoscopists were not completely comparable. Third, there was a low number of patients with short Barrett’s segments, with only three patients included with a Barrett’s segment of < 1 cm. This was partly caused by the inclusion criterion of C&M length being > 2 cm in Group I. Another possible contributing factor might be that the inclusion of neoplastic Barrett’s segments (LGIN or HGIN and/or cancer) may have resulted in longer Barrett’s segments being included in the study as the length is known to correlate with risk of neoplasia. Another potential limitation is possible systematic differences. Although no systematic differences were found, patient numbers were low. Finally, we cannot exclude the possibility that endoscopists, especially the community hospital endoscopists, may have been more careful than usual with their measurements because they were aware that they were taking part in a clinical trial. However, endoscopists were not aware during the procedures that the Barrett’s landmarks (C&M classification) would be used for interobserver evaluation as this was a post hoc analysis.

This study also has several strengths. First, endoscopies were performed consecutively at separate endoscopy sessions and were assessed independently by endoscopists who were blinded to the previous Barrett’s values. Second, endoscopies were performed under standardized conditions with the same case record forms, sedation, and endoscopy circumstances. Third, the study included homogeneous cohorts of endoscopists with similar endoscopy expertise at the two consecutive endoscopies. Fourth, expert Barrett’s endoscopists were compared with community hospital endoscopists (although no direct comparison was possible because Barrett’s experts and community hospital endoscopists assessed a different group of patients). Finally, endoscopic assessment of the Barrett’s C&M length landmarks was performed in real-time rather than from video sequences.

These results show that the Prague C&M classification has a good agreement during real-time endoscopy when applied by expert and community hospital endoscopists, indicating its usefulness in clinical practice, not only in tertiary referral centers with extensive expertise in Barrett’s imaging, but also among community endoscopists. In our opinion, assessment and registration of the Barrett’s length prompts the endoscopist to inspect and observe the Barrett’s segment more carefully. In addition, the Barrett’s length is important when performing endoscopic therapy as it is necessary to plan and evaluate the therapy as well. In the future, assessment and registration of Barrett’s length might be helpful for cancer risk stratification, for example by lengthening or shortening surveillance intervals according to the length of the Barrett’s segments. In this respect it is important to stress the finding that absolute agreement in Barrett’s esophagus length was only high when a difference of ≤ 2 cm was accepted.

In conclusion, application of the Prague C&M classification during real-time endoscopy showed good agreement for assessing circumferential extent and maximum Barrett’s length and reasonable agreement for assessing hiatal hernia length. Agreement for shorter Barrett’s segments ( ≤ 5 cm) seems to be lower. No differences in agreement were observed between Barrett’s experts and community hospital endoscopists. These findings strengthen the value of the Prague C&M classification in clinical practice to describe the length of the Barrett’s segment.

Competing interests: The project was financially supported by an unrestricted grant from AstraZeneca Netherlands.

Acknowledgements

The authors thank L.C. Baak, C. Böhmer, A.H. Naber, C.Y. Ponsioen, M.B. Wallace, K.K. Wang, and V. Subramanian for their participation as endoscopists.

• References

• 1 Wang KK, Sampliner RE. Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett’s esophagus. Am J Gastroenterol 2008; 103: 788-797
  CrossrefPubMedSearch in Google Scholar
• 2 Hameeteman W, Tytgat GN, Houthoff HJ et al. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989; 96: 1249-1256
  CrossrefPubMedSearch in Google Scholar
• 3 Hirota WK, Zuckerman MJ, Adler DG et al. ASGE guideline: the role of endoscopy in the surveillance of premalignant conditions of the upper GI tract. Gastrointest Endosc 2006; 63: 570-580
  CrossrefPubMedSearch in Google Scholar
• 4 Spechler SJ, Sharma P, Souza RF et al. American Gastroenterological Association technical review on the management of Barrett’s esophagus. Gastroenterology 2011; 140: e18-e52
  PubMedSearch in Google Scholar
• 5 Iftikhar SY, James PD, Steele RJ et al. Length of Barrett’s oesophagus: an important factor in the development of dysplasia and adenocarcinoma. Gut 1992; 33: 1155-1158
  CrossrefPubMedSearch in Google Scholar
• 6 Menke-Pluymers MB, Hop WC, Dees J. The Rotterdam Esophageal Tumor Study Group et al. Risk factors for the development of an adenocarcinoma in columnar-lined (Barrett) esophagus. Cancer 1993; 72: 1155-1158
  CrossrefPubMedSearch in Google Scholar
• 7 Avidan B, Sonnenberg A, Schnell TG et al. Hiatal hernia size, Barrett’s length, and severity of acid reflux are all risk factors for esophageal adenocarcinoma. Am J Gastroenterol 2002; 97: 1930-1936
  CrossrefPubMedSearch in Google Scholar
• 8 Weston AP, Sharma P, Mathur S et al. Risk stratification of Barrett’s esophagus: updated prospective multivariate analysis. Am J Gastroenterol 2004; 99: 1657-1666
  CrossrefPubMedSearch in Google Scholar
• 9 Anandasabapathy S, Jhamb J, Davila M et al. Clinical and endoscopic factors predict higher pathologic grades of Barrett dysplasia. Cancer 2007; 109: 668-674
  CrossrefPubMedSearch in Google Scholar
• 10 Sharma P, Dent J, Armstrong D et al. The development and validation of an endoscopic grading system for Barrett’s esophagus: the Prague C&M criteria. Gastroenterology 2006; 131: 1392-1399
  CrossrefPubMedSearch in Google Scholar
• 11 Curvers WL, Herrero LA, Wallace MB et al. Endoscopic tri-modal imaging is more effective than standard endoscopy in identifying early-stage neoplasia in Barrett’s esophagus. Gastroenterology 2010; 139: 1106-1114
  CrossrefPubMedSearch in Google Scholar
• 12 Curvers WL, Van Vilsteren FG, Baak LC et al. Endoscopic trimodal imaging versus standard video endoscopy for detection of early Barrett’s neoplasia: a multicenter, randomized, crossover study in general practice. Gastrointest Endosc 2010; 73: 195-203
  PubMedSearch in Google Scholar
• 13 Gardner MJ. Confidence Interval Analysis. London: British Medical Journal 1989;
  PubMedSearch in Google Scholar
• 14 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307-310
  PubMedSearch in Google Scholar
• 15 Lee YC, Cook MB, Bhatia S et al. Interobserver reliability in the endoscopic diagnosis and grading of Barrett’s esophagus: an Asian multinational study. Endoscopy 2010; 42: 699-704
  Thieme ConnectPubMedSearch in Google Scholar
• 16 Bredenoord AJ, Weusten BL, Timmer R et al. Intermittent spatial separation of diaphragm and lower esophageal sphincter favors acidic and weakly acidic reflux. Gastroenterology 2006; 130: 334-340
  CrossrefPubMedSearch in Google Scholar

Corresponding author

• References

• 1 Wang KK, Sampliner RE. Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett’s esophagus. Am J Gastroenterol 2008; 103: 788-797
  CrossrefPubMedSearch in Google Scholar
• 2 Hameeteman W, Tytgat GN, Houthoff HJ et al. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989; 96: 1249-1256
  CrossrefPubMedSearch in Google Scholar
• 3 Hirota WK, Zuckerman MJ, Adler DG et al. ASGE guideline: the role of endoscopy in the surveillance of premalignant conditions of the upper GI tract. Gastrointest Endosc 2006; 63: 570-580
  CrossrefPubMedSearch in Google Scholar
• 4 Spechler SJ, Sharma P, Souza RF et al. American Gastroenterological Association technical review on the management of Barrett’s esophagus. Gastroenterology 2011; 140: e18-e52
  PubMedSearch in Google Scholar
• 5 Iftikhar SY, James PD, Steele RJ et al. Length of Barrett’s oesophagus: an important factor in the development of dysplasia and adenocarcinoma. Gut 1992; 33: 1155-1158
  CrossrefPubMedSearch in Google Scholar
• 6 Menke-Pluymers MB, Hop WC, Dees J. The Rotterdam Esophageal Tumor Study Group et al. Risk factors for the development of an adenocarcinoma in columnar-lined (Barrett) esophagus. Cancer 1993; 72: 1155-1158
  CrossrefPubMedSearch in Google Scholar
• 7 Avidan B, Sonnenberg A, Schnell TG et al. Hiatal hernia size, Barrett’s length, and severity of acid reflux are all risk factors for esophageal adenocarcinoma. Am J Gastroenterol 2002; 97: 1930-1936
  CrossrefPubMedSearch in Google Scholar
• 8 Weston AP, Sharma P, Mathur S et al. Risk stratification of Barrett’s esophagus: updated prospective multivariate analysis. Am J Gastroenterol 2004; 99: 1657-1666
  CrossrefPubMedSearch in Google Scholar
• 9 Anandasabapathy S, Jhamb J, Davila M et al. Clinical and endoscopic factors predict higher pathologic grades of Barrett dysplasia. Cancer 2007; 109: 668-674
  CrossrefPubMedSearch in Google Scholar
• 10 Sharma P, Dent J, Armstrong D et al. The development and validation of an endoscopic grading system for Barrett’s esophagus: the Prague C&M criteria. Gastroenterology 2006; 131: 1392-1399
  CrossrefPubMedSearch in Google Scholar
• 11 Curvers WL, Herrero LA, Wallace MB et al. Endoscopic tri-modal imaging is more effective than standard endoscopy in identifying early-stage neoplasia in Barrett’s esophagus. Gastroenterology 2010; 139: 1106-1114
  CrossrefPubMedSearch in Google Scholar
• 12 Curvers WL, Van Vilsteren FG, Baak LC et al. Endoscopic trimodal imaging versus standard video endoscopy for detection of early Barrett’s neoplasia: a multicenter, randomized, crossover study in general practice. Gastrointest Endosc 2010; 73: 195-203
  PubMedSearch in Google Scholar
• 13 Gardner MJ. Confidence Interval Analysis. London: British Medical Journal 1989;
  PubMedSearch in Google Scholar
• 14 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307-310
  PubMedSearch in Google Scholar
• 15 Lee YC, Cook MB, Bhatia S et al. Interobserver reliability in the endoscopic diagnosis and grading of Barrett’s esophagus: an Asian multinational study. Endoscopy 2010; 42: 699-704
  Thieme ConnectPubMedSearch in Google Scholar
• 16 Bredenoord AJ, Weusten BL, Timmer R et al. Intermittent spatial separation of diaphragm and lower esophageal sphincter favors acidic and weakly acidic reflux. Gastroenterology 2006; 130: 334-340
  CrossrefPubMedSearch in Google Scholar