Background
Cytosponge diagnostic test for Barrett’s oESophagus
Study Ref n# | Publication Year | Study type | Setting | BE length | Sensitivity % (95% CI) | Specificity % (95% CI) |
---|---|---|---|---|---|---|
Pilot [14] | 2008 | Cohort | 2ary care | ≥C1 | 78.0 (64.0–89.0) | 94.0 (87.0–98.0) |
BEST1 [12] | 2010 | Prospective | 1ary care | ≥C1 | 73.3 (44.9–92.2) | 93.8 (91.3–95.8) |
≥C2 | 90.0 (55.5–99.7) | 93.5 (90.9–95.5) | ||||
BEST2 [13] | 2014 | Case:control | 2ary care | ≥C1 | 79.5 (75.9–82.9) | 92.4 (89.5–94.7) |
≥C2 | 83.9 (80.0–87.3) | |||||
≥C3 | 87.2 (83.0–90.6) | |||||
CASE1 [15] | 2015 | Cohort | 2ary care | ≥C1 or ≥ M3 ≥C3 | 95.4 (86.9–98.9) 96.8 (83.7–99.5) | N/A |
Rationale
Objectives
Objective | Endpoint | Usual care arm | Intervention arm |
---|---|---|---|
Primary objectives | |||
1. To compare histologically confirmed BE diagnosis between intervention and control | BE diagnosis within 12 months of joining the study (excluding BE found on random 12 month research endoscopy that will occur following 12 month snapshot). | Anonymised data aggregated by sex and age group from: - GP databases - Confirmed by upper GI endoscopy (biopsy result) as recorded in the GP record within 12 months | Anonymised data aggregated by sex and age group from: - GP databases - Confirmed by upper GI endoscopy (biopsy result) as recorded in the GP record within 12 months In addition, for patients with Cytosponge™ -TFF3 test: - Endoscopy record and pathology results |
Secondary objectives | |||
(i) To evaluate the cost of the Cytosponge™-TFF3 test versus usual care Objective (ii) To evaluate the cost-effectiveness of the Cytosponge™-TFF3 test versus usual care | (i) Mean cost per patient receiving the Cytosponge™-TFF3 test versus usual care. Costs to include costs of diagnosis using the Cytosponge™-TFF3 test, endoscopies and biopsies, endotherapy, oesophagectomy, medications, and follow-up in primary and secondary care. (ii) Incremental cost per QALY gained of the Cytosponge™-TFF3 test versus usual care | (i) Volume of resource use (endoscopies and biopsies, endotherapy, oesophagectomy, medications, and follow-up in primary and secondary care) from patient records. Unit costs (of each item of resource use) from published sources. (ii) Calculation of incremental cost per QALY gained to be based on a pre-existing model, supplemented with new data from the Trial. | (i) Volume of resource use (Cytosponges™, endoscopies and biopsies, endotherapy, oesophagectomy, medications, and follow-up in primary and secondary care) from patient records. Unit costs (of each item of resource use) from published sources. (ii) Calculation of incremental cost per QALY gained to be based on a pre-existing model, supplemented with new data from the trial. |
To assess the diagnostic accuracy of the Cytosponge™ in primary care | Positive Predictive Value (PPV), Negative Predictive Value (NPV) in relation to the length of BE | N/A | - PPV: proportion of TFF3 positive results confirmed to have BE by endoscopy - NPV: proportion of TFF3 negative cases confirmed to not have BE by endoscopy (10% endoscopy) |
To assess diagnostic performance of Cytosponge™ in detecting severity for BE | Score of BE severity based on BE biopsy results | N/A | Endoscopy reports for participants |
To report on the sampling adequacy | Inadequacy rate (same as BEST1 and BEST2) | N/A | CRF to capture: - Sample sufficient to generate result - Proportion of Cytosponge™ samples with < 5 and < 1 columnar cells (minimal standard) |
To confirm the endoscopy referral rate in the intervention arm | Proportion positive out of all adequate TFF3 tests and out of all patients swallowing a Cytosponge™ at least once | N/A | Cytosponge™-TFF3 test results |
To report on patient acceptability for Cytosponge™ | (i) Willingness: proportion of patients offered Cytosponge™ test who accept (ii) Cytosponge™ swallowing failures (iii) Increased and decreased cancer worry due to procedure and results (iv) Long term emotional or physical harm caused by procedure (v) Test experience (vi) Willingness to have repeat procedure | N/A | (i) Number of patients invited vs those consenting to Cytosponge™-TFF3 test (ii) Number of patients who fail to swallow and number of attempts Acceptability measures at baseline: (iii) STAI-6 (iv) Perceived risk of oesophageal cancer Acceptability measures at day 7–14: (iii) Perceived risk of oesophageal cancer (iv) STAI-6 (v) A visual analogue scale to rate experience (v-vi) the Inventory to Assess Patient Satisfaction, (iii – v) up to 30 qualitative patient interviews |
To assess physician/nurse acceptability of the Cytosponge™ | Experience and acceptability of Cytosponge™: administration, skills, reliability, side effects, user information | N/A | Qualitative interviews of clinical staff |
To report on the safety of the Cytosponge™ in primary care | Any ADE/ARs reported by patients up to 7 days post swallowing | N/A | Contact card given in case of ADE/SADE and 7-day telephone call |
(i) To understand how much BE is missed in current management of patients (ii) To compare undiagnosed BE in general population vs those who have been tested with Cytosponge™-TFF3 | - BE at 12 months - PPV for endoscopy referral, i.e. Cytosponge™ vs current GP criteria for referring for an endoscopy to look for BE | 12 month endoscopy for 10% of patients not requiring a clinically indicated endoscopy in time period of the study | Confirmatory endoscopy for patients with positive result and endoscopy findings from patients with negative result who accept research endoscopy at 12 months. |
To assess prevalence of benign oesophageal conditions | Prevalence of oesophageal conditions aside from BE in primary care population consulting with reflux symptoms | Endoscopy findings in 10% patients endoscoped | Endoscopy findings in 10% patients endoscoped and on Cytosponge™ test (via pathology assessment) |
Epidemiology: (i + ii) To confirm the prevalence (and incidence) of BE in both arms (iii) To confirm the prevalence (and incidence) of OC diagnosis (by stage) in both arms (iv) To confirm the prevalence (and incidence) diagnosis cancers of the gastric cardia (by stage) in both arms (v) To produce a model to predict the reduction in EAC related mortality from this strategy | (i) Diagnosis of BE (ii) Diagnosis BE with dysplasia (iii) Diagnosis of oesophageal cancer (OC) + stage at diagnosis (iv) Diagnosis of cancer of the gastric cardia + stage at diagnosis (v) Percentage of expected reduction in EAC mortality based on prevalence of BE if Cytosponge™ test introduced | (i-iv) Aggregate data from GP databases | (i-iv) Aggregate data from GP databases (i-iv) Endoscopy data from 10% endoscopy offered at 12 months (i-iv) Cytosponge™ patients: - Cytosponge™ findings - Endoscopy findings |
10% endoscopy invitation across both arms: (i) Acceptability of endoscopy (ii) Perceptions around Cytosponge™ use and reliability (iii) Number of BE diagnoses 12 months after negative TFF3 tests | (i) Comparisons between acceptance of invitation to endoscopy compared to Cytosponge™ test (ii) Proportion of patients with Cytosponge™ test who take up invitation to endoscopy (iii) Number of BE diagnosis in TFF3 negative patients in intervention arm | (i) 10% endoscopy: uptake of invitation to endoscopy | (i) 10% endoscopy: uptake of invitation to endoscopy for all participants who have not received the CytospongeP™ P (excluding ineligible patient and non-attendees for Cytosponge™) (i) CytospongeP™ Ptest invitation uptake (ii) Endoscopy uptake amongst patients with previous Cytosponge™ test (iii) BE diagnosis amongst patients with previous Cytosponge™ test |
Objective | |||
Longer-term objectives | |||
Epidemiology: For up to 10 years, to confirm the prevalence (and incidence): (i + ii) of BE in both arms (iii) of OC diagnosis (by stage) in both arms (iv) of cancers of the gastric cardia (by stage) in both arms (iv) To undertake modelling to predict the reduction in EAC related mortality from this strategy | (i) Diagnosis of BE (ii) Diagnosis of BE with dysplasia (iii) Diagnosis of OC + stage at diagnosis (iv) Diagnosis of cancer of the gastric cardia + stage at diagnosis (v) Percentage of expected reduction in EAC mortality based on prevalence of BE if Cytosponge™ test introduced | (i-iv) Anonymised data from cancer registry flagging- conducted anonymously via novel encryption method (v) Based on BE prevalence, prevalence of BE with dysplasia, flagging with the cancer registry, ONS and HES datasets | (i-iv) Anonymised data from cancer registry flagging- conducted anonymously via novel encryption method (v) Based on BE prevalence, prevalence of BE with dysplasia, flagging with the cancer registry, ONS and HES datasets |
Research and Development (including in future studies) | Genetic and biochemical risk factors for disease progression (germline and somatic variants and other biomarkers) including targeted, exome level and whole genome sequencing. | 10% patients who have endoscopy- surplus material from biopsies | - Surplus Cytosponge™ material - Saliva samples (for TFF3 positive patients only) - Surplus endoscopy biopsies |
Primary objectives
Secondary objectives
Methods
Design
Study setting
Participants
Inclusion criteria | Exclusion criteria for | ||
---|---|---|---|
BEST3 data collection | Cytosponge procedure | Upper GI endoscopy | |
• Male and female • Aged ≥50 • Records of ≥6 months of prescription for acid-suppressant medication in the last year | • Recorded regular prescriptions of NSAIDs • Recorded upper GI endoscopy in the previous 5 years as identified from the practice database • Recorded diagnosis of a current or previous oro-pharynx, oesophageal or gastro-oesophageal tumour • Recorded diagnosis of BE • Unable to attend the GP surgery • Deemed not fit enough by their GP, including lacking capacity | • Meeting the guidelines for an urgent endoscopy referral according to NICE guidelines • Recorded diagnosis of an oro-pharynx, oesophageal or gastro-oesophageal tumour (T2 staging and above), or symptoms of dysphagia • Difficulty in swallowing due to a known cerebrovascular accident or neurological disorder • Recorded oesophageal varices, cirrhosis of the liver • Inability to temporarily discontinue anti-thrombotic medication prior to procedure • Having eaten and drank within the preceding 4 h • Received prior surgical intervention to the oesophagus • Known pregnancy • Lacking capacity to provide informed consent | • Upper GI endoscopy during the study period • Severe hypertension (e.g. systolic > 200 diastolic > 100) • Myocardial infarction or any cardiac event within the previous 6 months • Cerebrovascular event or other neurological disorder where swallowing has been affected within the previous 6 months • Any previous treatment such as Photodynamic therapy (PDT) or Radio Frequency Ablation (RFA) to the oesophagus • Anticoagulation therapy/ medication on day of procedure (warfarin, heparin or tinzaparin) according to local guidelines • Other medical condition: low platelets or blood abnormalities that may cause excessive bleeding post procedure • Eaten or drank within the previous 6 h • Preference for sedation and has not brought anyone to accompany them at home. Follow local guidelines • Known pregnancy • Lacking capacity to provide informed consent |
Number of participants
Randomisation
-
Stratum 1 (small practices): 50–60
-
Stratum 2 (medium practices): 61–74
-
Stratum 3 (large practices): 75–100.
Dates and duration of the trial
Consent procedures
Consent at practice level (opt in)
Introductory letter provided to patients about use of anonymous data (BEST3 data collection)
Written consent for BEST3 intervention and endoscopy (opt in written consent)
BEST3 introductory letter and anonymous data collection
BEST3 introductory letter
Aggregate data collection for BEST3 trial
Baseline data extract | 12 month follow up extract | ||
---|---|---|---|
Time point/period | Variable | Time point/period | Variable |
Baseline | Sex | Baseline | Sex |
Age | Age | ||
Obesity records | Baseline and 12 months (where available) | Obesity records | |
Smoking status | Smoking status | ||
Alcohol consumption | Alcohol consumption | ||
Previous 12 months | PPI / H2RA prescriptions | Previous 12 months | PPI / H2RA prescriptions |
Other prescription medication: Aspirin, COX2i, antibiotics for H. pylori eradication | Other prescription medication: Aspirin, COX2i, antibiotics for H. Pylori eradication | ||
Heartburn and / or GERD related symptoms | Heartburn and / or GERD related symptoms | ||
Number of GP and practice nurse visits at home and at the practice | |||
Number of endoscopy or GI referrals | |||
Diagnosis of BE | |||
Diagnosis of EAC or pre-malignant conditions | |||
Diagnosis of benign oesophageal conditions | |||
Records on any upper GI specific procedures, e.g. endotherapies or oesophagectomies | |||
Anonymised endoscopy reports including pathology reports for BE and OC diagnosis; and benign conditions via a tick box (EoE, candida, inflammation, ulcer slough, squamous dysplasia, herpes, other) | |||
Type of referral: emergency via A&E, 2 week wait / urgent, routine and in or out patient (either form GP records or endoscopy report) | |||
Number of biopsies (from endoscopy reports) | |||
Anonymised letters from upper GI consultants |
Long term follow-up
BEST3 intervention: Cytosponge™-TFF3 test
Patient invitation to BEST3 intervention
Cytosponge™ clinic procedure
Storage and analysis of Cytosponge™ samples
Cytosponge™-TFF3 results
Endoscopies
Invitation for endoscopies - intervention arm: TFF3-positive patients
Invitation for research endoscopies - 10% of patients who do not require diagnostic endoscopy (all arms)
Endoscopy procedures
-
For all endoscopies where BE is found, biopsies will be collected (in all 4 quadrants) every 2 cm according to surveillance guidelines. In addition, endoscopically-suspicious areas will be targeted for biopsies.
-
A further two biopsies from the gastro-oesophageal junction (GOJ) will be collected (below the z-line) for research purposes from both BE positive and patients with a TFF3 negative test.
-
All biopsy samples will be processed and analysed by the local pathologist according to standard clinical practice including for benign conditions.
-
Study participants will be informed about endoscopy findings in the usual way via a letter from the gastroenterologist copied to their GP.
-
Endoscopic images will be requested of every GOJ and newly diagnosed BE to try to exclude misdiagnosed hiatus hernias and intestinal metaplasia (IM) at normal appearing GOJ.
Diagnosis of BE
-
Diagnosis by the endoscopist or gastroenterologist following BSG guidelines
-
∘ > 3 cm likely correct;
-
∘ < 3 cm more suspect unless biopsy with IM;
-
-
Confirmed by study pathologist or gastroenterologist
-
∘ >C1 or >M3 +IM on biopsy;
-
-
IM on biopsy
-
∘ any length.
-
Score | BE severity |
---|---|
0 | Pathology report not available |
1 | Intestinal metaplasia (IM) on biopsy and endoscopic findings not seen in categories below |
2 | C1 or C0 M3 + IM |
3 | C2 or more, C0 M4 or more +IM |
4 | C3 or more |
5 | Low grade dysplasia (LGD) |
6 | High grade dysplasia (HGD) or T1a cancer |
Acceptability measures- intervention group only
Patient acceptability measures
Intervention acceptability for patients and health care professionals
Patients
Healthcare professionals
Involvement of public and patient involvement representatives
Statistics and data analysis
Sample size calculations
Statistical analysis
Primary endpoint
Economic evaluation
Primary endpoint
Pilot phase and 6-month milestone review
Pilot study findings | Amendments to study protocol |
---|---|
Female to male ratio slightly higher than 50:50 overall | If the proportion of females to males consistently exceeds 55:45 within the overall cohort, the study team may institute a 50:50 split for females: males in line with known BE prevalence, at the discretion of the Trial Statistician. |
Cytosponge™ appointment uptake < 50% | Patients will receive a total of one reminder in the form of a letter, phone call or text message. |
Time required for practice and patient recruitment might take longer for each practice than anticipated | Practices recruited in the latter stages of the trial may adopt a 6 month follow-up period to allow timely completion of study activities with a simulation tool used to ensure parity across the datasets. |
Milestone review findings | Amendments to study protocol | Details |
---|---|---|
Cytosponge appointment uptake 27%: Substantial impact on sample size as number of practices would have to be increased to ~ 200 | 1) Sample size amended - Using uptake = 27% - including patients with false negative Cytosponge test diagnosed with BE during 12-month follow-up: (1–0.85)*0.6% 2) Additional individual randomisation arm added to reduce the sample size | - Individual randomisation sample size (without adjusting for cluster randomisation): 6764 - Variance inflation factor (VIF) for confirmed and projected sample sizes are 3.72 and 4.5, respectively. 1 individually randomised participant would therefore be equivalent to 3.72 or 4.5 cluster randomised participants - Practices already commenced set up on the cluster randomisation design allowed to continue to randomise in a cluster fashion - Sample size will be adjusted depending on number of patients in cluster randomised group |
Due to small number of smaller practices, stratification by both area and practice size has resulted in imbalances in arm allocations for some areas | Stratification by practice size will not be taken into account during randomisation but in the analysis instead | - To simplify randomisation and avoid any further imbalances for remaining cluster randomisation practices - Analysis: the primary analysis will be a stratified test of proportions taking into account the variation inflation within each stratum. |