Methodology:
Patients with non-dysplastic Barrett's esophagus (BE) undergoing surveillance and
patients meeting criteria for screening for BE with columnar-lined esophagus detected at
endoscopy will be recruited in this multicenter randomized controlled trial.
All patients will undergo upper endoscopy using high-definition white light endoscopy and
electronic chromoendoscopy (NBI/FICE). The findings (detection of visible
lesions/abnormal mucosal or vascular pattern) with electronic chromoendoscopy will be
recorded.
Patients will be randomized to receive sampling using either the Seattle biopsy protocol
or WATS3D. The Seattle protocol will entail obtaining random biopsies every 2 cm in a
4-quadrant fashion. Biopsies will be taken from any visible lesion (no matter how subtle)
and will be submitted in separate jars and excluded from the comparative analysis of the
study. The WATS3D procedure will be conducted using a standardized protocol. Patients
will then receive sampling using the technique to which they were not assigned to by
randomization. This allows for all patients to receive standard of care (sampling using
the Seattle biopsy protocol), and to determine concordance between the two sampling
techniques acknowledging that one sampling technique could potentially affect the results
of other sampling technique.
In patients with discordant results (WATS3D positive and random biopsy negative), repeat
upper endoscopy with sampling using the Seattle biopsy protocol will be performed in an
attempt to confirm the findings noted at WATS3D. This will be performed in the
surveillance and screening population. Biopsy specimens will be submitted for
histopathologic examination for routine clinical care. For the purpose of this study, all
biopsy specimens will be sent to the Cleveland Clinic for interpretation by a central GI
pathologist who will be blinded to patient details and to reading by the WATS3D
pathologist.
Study Duration:
Interventions performed as described above will be performed during one endoscopic
procedure. The endpoints in this study will include progression to dysplasia, cancer,
need for endoscopic eradication therapy, and death.
Planned enrollment period: 2.5 years. Planned duration of the study: 3 years.
Study Centers:
To maximize the generalizability of results, this randomized controlled trial will be
conducted across different practice settings that include tertiary care centers, closed
healthcare networks and large community practices at approximately 14 sites.
Anticipated Number of Participants:
2298 (see Statistical Methodology below)
1982 to compare diagnostic yield of dysplasia between the two approaches in BE
patients undergoing surveillance
316 to compare diagnostic yield of intestinal metaplasia between the two approaches
in patients undergoing screening for BE
Statistical Methodology:
For a prospective randomized trial in detecting LGD, HGD or EAC comparing the Seattle
protocol, the current standard practice, to wide-area transepithelial sampling (WATS3D)
where all positive results for dysplasia or EAC with WATS3D sampling not detected on
Seattle biopsy protocol at index endoscopy will require confirmation by repeat sampling
using the Seattle biopsy protocol at repeat endoscopy, the following sample sizes were
calculated assuming α=0.05 and power of 0.8. All calculations are completed using PROC
POWER in SAS 9.4 for Fisher's exact test unless otherwise noted.
For the primary aim (surveillance population), the investigators will compare the
proportion of positive results between the Seattle protocol and WATS3D where all positive
results for dysplasia or EAC with WATS3D sampling not detected on Seattle biopsy protocol
at index endoscopy will require confirmation by repeat sampling using the Seattle biopsy
protocol at repeat endoscopy with either Pearson's chi-square test or Fisher's exact test
for proportions. Positive findings of dysplasia or EAC on WATS not detected on Seattle
biopsy protocol at index endoscopy or subsequent repeat endoscopy with sampling using the
Seattle biopsy protocol will not be considered as dysplasia or EAC detected by WATS for
the primary analysis. Previous pooled estimates indicated a 2.5% detection rate with
Seattle protocol and 4.9% with WATS3D, resulting in a need for 828 participants for each
arm (a total of 1656). However, the investigators are restricting to WATS3D cases which
are confirmed with target biopsy or Seattle protocol at the initial visit or a follow-up
sample when results are discordant. Assuming a 95% confirmation rate, 991 participants
are need for each arm to compare a 2.5% detection rate for Seattle protocol and 4.655%
for WATS3D (a total of 1982).
For the primary aim (screening population), the investigators will compare the proportion
of positive results between techniques to which patients are randomized with Fisher's
exact test for proportions. Previous pooled estimates indicated a 40% intestinal
metaplasia detection rate with Seattle protocol and 60% with WATS3D. However, the
investigators are restricting to WATS3D cases which are confirmed with target biopsy or
Seattle protocol at the initial visit or a follow-up sample when results are discordant.
Assuming a 95% confirmation rate, 158 participants are need for each arm to compare a 40%
detection rate for Seattle protocol and 57% for WATS3D (a total of 316).
The investigators will conduct one interim analysis using Lan and DeMet's alpha spending
function approximating O'Brien-Fleming boundaries. Based the sample size estimates above,
the investigators expect the interim analysis will occur after the data for 991
participants (50% enrollment) has been collected. This interim analysis will be
restricted to the primary aim in the surveillance population. No interim analysis will be
conducted to address the primary aim in the screening population. In the case where study
enrollment does not stop early due to efficacy, all participating physicians will be
blinded to the results of this interim analysis.
As a secondary analysis, the investigators will use results from both the randomized
procedure and the follow-up procedure with the protocol to which patients were not
randomized to perform a paired analysis of the concordance between the two diagnostic
tests. Although an important analysis for determining whether WATS3D could be used
instead of the Seattle protocol, due to the risk of the initial biopsy affecting these
results of the follow up procedure this is a secondary analysis. Given the sample size of
1982 for the primary outcome, there is greater than 99.9% power to detect a difference in
the paired results between proposed discordant proportions of 0.3575% for Seattle
protocol positive and WATS3D negative to 3.11% for WATS3D positive and Seattle protocol
negative using McNemar's test for paired nominal data. These discordant proportion
estimates are derived from prior pooled data for dysplasia detection rates of 4.9% for
WATS3D alone and 2.5% for Seattle protocol alone, with the discordant proportions of test
results derived from data reported in Vennalaganti assuming that only 80% of WATS3D
positive and Seattle protocol negative cases are verified with targeted follow-up
biopsies to confirm the diagnosis.
The comparison of yield of dysplasia will also be conducted based on expert pathology
review for biopsy and WATS3D specimens. The overall proportion of cases with visible
lesions with their associated pathology results will be recorded and will not be a part
of the analyses comparing the diagnostic yield of dysplasia between the Seattle biopsy
protocol with WATS3D.
The primary outcome comparing diagnostic yield of any dysplasia and intestinal metaplasia
between the two randomized arms will use either Pearson's chi-square or Fisher's exact
test for portions. The secondary analysis for paired data will use McNemar's test and
will exclude cases that were WATS3D positive but could not be confirmed by a concurrent
(Seattle protocol) or follow-up endoscopic biopsies using the Seattle biopsy protocol. A
sensitivity analysis will be completed which includes WATS3D positive cases that did not
have diagnostic confirmation with biopsy. Comparison for the detection of intestinal
metaplasia between the two strategies will use McNemar's test for paired data including
all cases.