Introduction
Barrett’s esophagus (BO) is a change in the cells of esophageal tissue and a precursor
to esophageal cancer. BO carries a risk of cancer 30 to 150 times greater than that
for an age-matched population without BO, and affects 375,000 people in the UK [1 ]
[2 ]. Esophageal cancer is the fifth most common cancer in the UK and carries a poor
5-year survival rate of < 5 % [3 ]
[4 ]. Current guidelines recommend regular surveillance (between 2 – 5 years according
to length of Barrett’s) to detect and remove early dysplasia, [5 ] as patients diagnosed with esophageal cancer within a surveillance program have
earlier-stage disease and longer survival times [6 ]
[7 ].
The current gold-standard technique for detecting dysplasia in surveillance programs
involves quadrantic esophageal biopsies every 2 cm during endoscopy (Seattle protocol).
This is necessary as only 13 % of dysplastic changes appear as visible nodules [8 ]. However, up to 40 % of early cancers are missed, as despite multiple biopsies,
only a small proportion of the total surface area of the esophagus is sampled [9 ]. In practice, only 50 % of surveillance endoscopies are estimated to adhere to this
method as it is laborious and time-consuming, poorly tolerated by patients, and expensive
to process biopsies [10 ]. Research to improve surveillance effectiveness and test performance has been recognized
as a priority [11 ].
Currently, the key limitation to Barrett’s surveillance is the macroscopic invisibility
of Barrett’s dysplasia. Methods proposed to enhance dysplasia visibility, thus permitting
targeted biopsies, include dye sprays, [12 ]
[13 ]
[14 ] electronic image intensifiers built into the endoscope namely: narrow band Imaging,
[15 ] Fujinon intelligent color enhancement [16 ] and autofluorescence imaging [17 ]. However, evidence focuses on high-risk populations where dysplasia has already
been identified, or in the follow-up of patients treated endoscopically for dysplasia.
It is lacking in a surveillance population, where dysplasia prevalence is significantly
lower.
When sprayed on Barrett’s mucosa, acetic acid (AA) highlights specific surface patterns
including dysplasia and cancer, which can be readily recognized by the trained endoscopist
[18 ]. After application of AA, dysplastic tissue appears red against the surrounding
Barrett’s making it easily visible ([Fig. 1 ]) [19 ]. This process is reversible, causes no damage to tissues and can be repeated. Two
large cohort studies have demonstrated the efficacy of AA for detection of Barrett’s
dysplasia in a high-risk population, reporting sensitivities of 90 % to 95 % and specificities
of 75 % to 85 % [20 ]
[21 ]. Further research suggested that the number of biopsies could be substantially reduced
using AA-targeted biopsies instead of quadrantic biopsies, reducing pathology costs
by 97 % [22 ]. A recent retrospective study suggested that the AA method may detect more dysplasia
than protocol-guided biopsies in a Barrett’s surveillance population; 4.3 % vs 1.4 %,
however, that study had a number of limitations in that it did not control for endoscopist,
was retrospective and liable to confounding [23 ].
Fig.1 a Barrettʼs neoplasia seen in white light. b Same area after acetic acid dye-spray – note early loss of acetowhitening.
The 2 methods need to be compared in an RCT. However, prior to a definitive study,
a feasibility study is required to establish whether a cross-over design involving
2 endoscopies per participant is acceptable, by assessing recruitment and retention
rates and exploration through qualitative interviews. It is also important to ascertain
the efficacy of training clinicians to use the technique, and their perceptions of
a change in the endoscopic method. Estimates of diagnostic accuracy for both groups
are required to power a definitive trial.
Patients and methods
Research question
Is a trial to investigate the diagnostic accuracy of AAC in a Barrett’s surveillance
population feasible and acceptable to patients and endoscopists?
Objectives
This feasibility study will investigate: (1) the outcomes of a novel training program
for AA-guided dysplasia detection for endoscopists; (2) feasibility of recruiting
200 Barrettʼs surveillance patients in 18 months; (3) eligibility rate of screened
patients; (4) recruitment rate of eligible patients; (5) the rate of attendance for
second endoscopy (retention rate) and reasons for non-attendance; (6) participant
acceptability of the study design and acceptability of a change in surveillance technique;
(7) the difference in dysplasia detection rates between AA endoscopy and standard
endoscopic practice to inform the power calculation for a definitive study; (8) AE
types and frequencies; and (9) perceptions, values and opinions of endoscopists of
changing surveillance technique in practice.
Design
A mixed-methods feasibility study will be conducted with 3 components: (1) training
and assessment of endoscopists in the AA (intervention) and Seattle biopsy (current
standard) technique; (2) multicenter randomized, cross-over tandem endoscopy trial
examining diagnostic accuracy of 2 methods of detecting cellular changes in patients
with BO; and (3) telephone interviews to gather qualitative data from patients and
endoscopists regarding study design and acceptability of implementation of the AA
technique as standard care.
Endoscopist training
Validation of image and video library: Images and videos of AA chromoendoscopy in
normal and dysplastic Barrett’s validated by 2 international expert endoscopists will
form a training and assessment library. The accuracy and inter-observer agreement
between experts will be calculated and used to set the pass mark for the assessment
tool.
Web-based training and assessment: The web-based training tool will be developed to
provide a robust model suitable for training endoscopists for the definitive trial.
Baseline knowledge will be assessed, followed by a tutorial of images and videos from
the library to demonstrate the AA appearances of normal, dysplastic and cancerous
Barrett’s, including the Portsmouth AA classification. A series of practice questions
for a diagnosis (normal or dysplastic Barrett’s) will be administered. Endoscopist
trainees will be given immediate feedback including a detailed description of what
was in the image or video to facilitate improved knowledge.
Face- to-face training: All endoscopists will attend an interactive workshop, including
lecture-based training and interactive discussion with experts, followed by observing
live procedures over a video-link. Trainees will then repeat the assessment tool.
Once endoscopists achieve adequate accuracy scores, defined as achieving scores within
5 % of experts, they will perform an endoscopy list under the direct supervision of
an expert in AA at their own institutions.
Participants
Participants will be recruited from 6 centers in the UK, i. e. Portsmouth, Leicester,
Gloucestershire, Western Sussex, Brighton and Sussex, and Royal Bournemouth and Christchurch
trusts. The study design summarized in [Fig. 2 ] and [Fig. 3 ] demonstrates the schedule of enrollment, interventions and assessments. Recruitment
commenced in July 2015 at Portsmouth Hospitals NHS Trust, the lead trial site.
Fig. 2 Study design for participants.
Fig. 3 Study timeline.
Eligibility criteria
Inclusion criteria
aged 18 years or above
biopsy (histologically) proven Barrett’s metaplasia
at least 2-cm length of Barrett’s esophagus
informed consent
Exclusion criteria
significant oesophagitis
known or prior esophageal cancer or esophageal dysplasia
previous endoscopic therapy for dysplasia
known allergy to acetic acid
previous inclusion in the study
Recruitment
Cross-over tandem endoscopy study
Patients scheduled to attend surveillance gastroscopy endoscopy will receive a letter
4 to 6 weeks before the visit, followed by a letter of invitation and patient information
sheet for interested individuals.
Thirty minutes prior to gastroscopy, potential participants will discuss the study
with the research team and provide informed consent.
Qualitative interviews
Patients
For qualitative data collection, a probabilistic sample size calculation was not appropriate.
Instead, a purposive sampling framework to include male and female participants of
all ages, and across the study sites will be selected for interview. Data collection
will cease when saturation of data has been achieved or no new insights are forthcoming.
In this way, we will be confident that the data will truly reflect the views and perspectives
of the range of study participants
On enrollment into the endoscopy study, participants will be asked whether they would
consent to being contacted to take part in a telephone interview up to 4 weeks after
the endoscopies are completed, and written information will be given to them. Patients
who do not wish to take part in the endoscopy study will be asked for their consent
to be interviewed about their experience of hearing about the study and why they decided
not to participate.
Endoscopists
Prior to the endoscopy training program, endoscopists will be asked to take part in
telephone interviews at the completion of the trial (minimum of 8 interviews).
Randomization and blinding
Participants will be randomized using permuted blocks, stratified by center in a 1:1
ratio, to initial endoscopy of either the standard procedure (Seattle protocol mapping
biopsies) or AA endoscopy (targeted biopsies). The secure randomization website, https://www.sealedenvelope.com will be used. Patients will be unable to detect whether acetic acid had been used
and thus will be blinded to the procedure method. The endoscopist will be blinded
to the histological results of the previous endoscopy for an individual participant.
It will not be possible to blind the histopathologist
as the Seattle protocol produces a greater number of biopsies for processing.
Study intervention
Study participants will receive an endoscopy using both methods, at an interval of
4 to 10 weeks, with the initial endoscopy method determined by randomization allocation.
The endoscopies will be performed with high-definition processors (Olympus Lucera
Elite or Spectrum and Pentax EPKi7000) and high-definition gastroscopes (Olympus HQ290
or H260 and Pentax i10 series) without use of electronic chromoendoscopy or alternative
dye sprays. All study endoscopies will be videoed and 10 % of non-dysplastic cases
reviewed by the study team, in addition to all cases of dysplasia.
Acetic acid method
Following a pre-endoscopy mucosal cleansing drink, the Barrett’s segment will be inspected
using standard white light and visible abnormalities noted. A spray catheter will
be inserted through the biopsy channel of the endoscope and AA 2.5 % sprayed onto
the Barrett’s mucosa. Excess fluid will be removed and mucosa inspected for abnormalities,
which will be biopsied. Each biopsy will be sent to the laboratory in an individual
cassette. Areas for biopsy will be identified using the Portsmouth Barrett’s Acetic
Acid Classification (PREDICT Classification) [24 ], and the tissue will be categorized into non-dysplastic, dysplastic or cancer. If
no visible neoplasia is seen, no biopsies will be taken.
Control method
The control arm involves a standard gastroscopy according to the Seattle protocol
of quadrantic biopsies taken every 2 cm, in addition to biopsies from any visible
abnormalities.
Histological methods
Biopsies will be analyzed and reported by accredited NHS hospital laboratories at
each site according to clinical pathology accreditation (CPA) ISO 15189 quality control
procedures. Specimens where a histological diagnosis of dysplasia is suspected will
be examined by 2 expert gastrointestinal pathologists, and a third opinion will be
sought in case of disagreement.
Measurement of outcomes
Training assessment
Sensitivity, specificity, positive predictive value and negative predictive value
will be calculated for all image and video assessments performed by the endoscopists.
Kappa scores will be used to assess interobserver agreement. Endoscopists will be
asked how confident they would be to switch practice at each stage of the training
process (pre-online training, post-online training, post-training day). Following
training, they will be asked what the components of an AA training program should
be.
Study process
Information will be gathered from trial documentation, including numbers of patients
recorded on screening logs, consent logs and withdrawal forms (e. g. detection of
cancer on first endoscopy or patient-led withdrawal prior to second endoscopy).
Endoscopic and histology outcomes
Endoscopy process outcomes include duration of endoscopy, use of sedation and level
of patient discomfort using a 10-point scale (1 = no discomfort, 10 = severe discomfort).
Diagnostic outcomes include the length of Barrett’s observed, islands of Barrett’s
reported according to the Prague criteria, presence of endoscopically visible inflammation,
visible abnormalities requiring targeted biopsies (recorded as per Paris classification
and Prague criteria), number of biopsies performed, and endoscopic diagnosis (metaplasia,
dysplasia or cancer) [25 ]
[26 ]. Dysplasia will be categorized as low- or high-grade [5 ].
Study visit schedule
Baseline information includes: demographics, educational achievement and occupation,
lifestyle (smoking, alcohol), history of Barrett’s (date of first diagnosis, previous
histology), current medications (including aspirin, clopidogrel, non-steroidal anti-inflammatory
medicines, proton-pump inhibitors) and significant medical history.
The first endoscopy will be performed, followed by a second endoscopy after a 6-week
interval (– 2/+ 4 week window), permitting the esophagus to heal.
During endoscopy, data will be collected concerning sedation (dose and type administered),
procedure duration, characteristics of Barrett’s, number of biopsies and patient discomfort.
Histology results from both endoscopies will be released as per standard care as soon
as possible after the second endoscopy has been completed.
Safety reporting
AEs will be collected during the procedure and in the following 24 hours. Data collected
include a description of the event, date/time of onset, intensity, relatedness to
the procedure, whether the event was categorized as a serious adverse event (SAE),
action taken, date of resolution (if resolved) and final outcome of the event [27 ].
Sample size
Based on the historical cohort studies and wide consultation within the British Society
of Gastroenterology research committee and consultation with experts in the field,
it was estimated that 200 patients would be reasonable for recruitment to enable the
reproducibility and generalizability of the data to be established. The purpose of
this study is not to produce statistically significant data in itself but to establish
likely event rates and effect size to inform the power calculation for the definitive
study. Within the target sample size of 200 patients, the intention is to recruit
at least 30 patients per site over 6 centers.
Data management
Data will be managed by the Gloucester Clinical Trials Unit (GCTU), using a Structured
Query Language database (Transact-SQL 2008, Microsoft, USA), including an electronic
audit trail.
Statistical analysis
The study will be reported in accordance with the Consolidated Standards of Reporting
Trials statement and ICH Guidelines for Good Clinical Practice.
There are no planned interim analyses. All participants will be included in analysis.
Participants who do not attend the second endoscopy will have their first endoscopy
data included with analysis of the appropriate treatment group. Analyses will be performed
using Stata (College Station, Texas). As this is a feasibility study and was not powered
to be able to test for differences between the 2 endoscopy methods, formal statistical
testing is not appropriate. Descriptive statistics will be used for each arm in order
to inform a further definitive trial. The proportion of patients not attending the
second endoscopy will be calculated, and for purposes of the feasibility analysis,
the results of the endoscopy will be included in the descriptive statistics for each
group. This will allow collection of preliminary data to inform the sample size calculation
for a definitive trial, and therefore, all data collected will be included to provide
better information.
Study process indicators
These include: (1) total participant recruitment, by site and month of study; (2)
screening success rate (eligible patients/patients screened); (3) consent rate (patients
recruited/eligible patients); (4) rate of attendance to first and second endoscopies;
and (5) rate of withdrawal from study, and reasons where provided.
Neoplasia yield
Rates of dysplasia detection, (endoscopies with dysplasia/total endoscopies), will
be calculated for both methods of endoscopy. True histology is defined as the combination
of results from both endoscopies (targeted + mapping biopsies). The dysplasia detection
rate for each technique is defined as patients with biopsy confirmed dysplasia by
each technique/total patients with dysplasia.
Additional endoscopy outcomes
These will be presented according to the endoscopy technique. Descriptive statistics
(proportions, means, standard deviations, confidence intervals) will describe the
following: duration of gastroscopy; use of sedation; endoscopic diagnosis (metaplasia,
dysplasia or cancer); Barrett’s length; presence of Barrett’s islands; visible abnormalities
leading to targeted biopsies during all procedures and total number of biopsies; suspected
diagnosis (metaplasia, dysplasia); patient discomfort (10-point scale).
Safety
The frequency and severity of AEs of interest, grouped by type, namely chest infection/aspiration,
bleeding and other complications considered to be related to the procedure will be
reported.
Qualitative methods
Patient participants: an introductory letter, contact details and a photograph of
the lead qualitative researcher will be provided prior to the telephone interview.
Informed consent will be taken prior to the interview. Telephone interviews will take
place a minimum of 2 days, but no longer than a month after the end of study participation
and last approximately 30 minutes. Topics include experiences of study participation,
views of the impact for patients if the AA technique became standard practice, and
facilitators and barriers to recruitment and retention for a definitive trial. Following
the interview, brief field notes will be taken to record any immediate reflections
and emerging themes.
Endoscopist participants: At least 8 endoscopists who had participated in delivering
the ABBA trial will be recruited to take part in telephone interviews at the completion
of the trial. Topics for the interview include concerns regarding the procedure prior
to training, experience of training, perception of usefulness and suitability of the
AA technique and feasibility study methods for the definitive trial.
Data analysis
Participants will be identified by a unique number in the database and audio recordings.
Audio recordings will be transcribed and entered into NVivo 10 (QSR International,
Melbourne, Australia).
Data will be analyzed through a process of thematic analysis using the Framework approach.
Categories will be independently coded by 2 researchers, and a member of the steering
group will read a sample of transcripts to generate a preliminary framework without
knowledge of the original researchers’ list. In case of disagreement, a solution will
be sought to clarify the meaning of a code/theme developed. A deductive approach to
map themes from the 2 groups of participants (patients and endoscopists) will enable
us to compare and contrast similarities and differences of attitudes, perceptions
and experience.
Public patient involvement
Patients, carers and the public from Heartburn Cancer UK and the University of Portsmouth
School of Health Sciences and Social Work ENGAGE group were involved
in the grant application, study design including interview schedules, and will be
involved in study delivery.
Ethics
Favorable ethical approval for this study was given by the National Research Ethics
Service (NRES) Committee, South Central – Berkshire, reference 15/SC/0085.
Because the study involves 2 endoscopies, participants will be exposed to the risk
of experiencing endoscopy-related AEs twice. However, the overall risk of such events
is very small, and the benefit:risk of having 2 methods of detecting potential dysplasia
may outweigh the small increased risk of AEs. Secondly, there is a delay of a maximum
of 6 weeks for histology to become available due to the requirement for blinding of
the second endoscopy. However, in the case that cancer is detected during either endoscopy,
samples will be prioritized, immediate referral for treatment made, and the participant
withdrawn so as not to delay treatment.
Funding
The ABBA study is funded by National Institute of Health Research (NIHR) Research
for Patient Benefit (RFPB) program (Project Number PB-PG-1013-32045).
Dissemination
The results of this feasibility study will be disseminated to all groups who are able
to offer input and suggestions for further work, i. e. patients with Barrett’s and
patient support groups, gastroenterologists, pathologists, and those involved in managing
the workflow for the Barrett’s surveillance program within the British Society of
Gastroenterology (BSG).
