Clinical Gastroenterology and Hepatology 2022;20:2763–2771

Optimized Surveillance Intervals Following Endoscopic

Eradication of Dysplastic Barrett’s Esophagus:
An International Cohort Study
Allon Kahn,* Julia Crook,‡ Michael G. Heckman,‡ Mikolaj A. Wieczorek,‡
Sarmed Sami,k Diana Snyder,* Siddharth Agarwal,¶ Jose Santiago,#
Jacobo Ortiz Fernandez-Sordo,# W. Keith Tan,** Ramona Lansing,¶
Kenneth K. Wang,¶ Krish Ragunath,‡‡ Massimiliano DiPietro,** Herbert Wolfsen,§
Francisco Ramirez,* David Fleischer,* Cadman L. Leggett,¶ and Prasad G. Iyer¶
*Division of Gastroenterology and Hepatology, Mayo Clinic, Scottsdale, Arizona; ‡Department of Quantitative Health Sciences,
**Division of Gastroenterology and Hepatology, Mayo Clinic, Jacksonville, Florida; §University College London, London, United
Kingdom; ||Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, Minnesota; ¶Nottingham University Hospitals
NHS Trust, University of Nottingham, Nottingham, United Kingdom; #MRC Cancer Unit, University of Cambridge, Cambridge,
United Kingdom; ‡‡Curtin University, Perth, Australia

BACKGROUND & AIMS: Recommended surveillance intervals after complete eradication of intestinal metaplasia (CE-
IM) after endoscopic eradication therapy (EET) are largely not evidence-based. Using recur-
rence rates in a multicenter international Barrett’s esophagus (BE) CE-IM cohort, we aimed to
generate optimal intervals for surveillance.

METHODS: Patients with dysplastic BE undergoing EET and achieving CE-IM from prospectively maintained
databases at 5 tertiary-care centers in the United States and the United Kingdom were included.
The cumulative incidence of recurrence was estimated, accounting for the unknown date of
actual recurrence that lies between the dates of current and previous endoscopy. This cumu-
lative incidence of recurrence subsequently was used to estimate the proportion of patients
with undetected recurrence for various surveillance intervals over 5 years. Intervals were
selected that minimized recurrences remaining undetected for more than 6 months. Actual
patterns of post–CE-IM follow-up evaluation are described.

RESULTS: A total of 498 patients (with baseline low-grade dysplasia, 115 patients; high-grade dysplasia
[HGD], 288 patients; and intramucosal adenocarcinoma [IMCa], 95 patients) were included. Any
recurrence occurred in 27.1% and dysplastic recurrence occurred in 8.4% over a median of 2.6
years of follow-up evaluation. For pre-ablation HGD/IMCa, intervals of 6, 12, 18, and 24 months,
and then annually, resulted in no patients with dysplastic recurrence undetected for more than
6 months, comparable with current guideline recommendations despite a 33% reduction in the
number of surveillance endoscopies. For pre-ablation low-grade dysplasia, intervals of 1, 2, and
4 years balanced endoscopic burden and undetected recurrence risk.

CONCLUSIONS: Lengthening post–CE-IM surveillance intervals would reduce the endoscopic burden after CE-IM
with comparable rates of recurrent HGD/IMCa. Future guidelines should consider reduced
surveillance frequency.

Keywords: Esophagus; Barrett’s Esophagus; Adenocarcinoma; Surveillance.

Abbreviations used in this paper: ACG, American College of Gastroen-

terology; BE, Barrett’s esophagus; CE-IM, complete eradication of intes-
tinal metaplasia; EAC, esophageal adenocarcinoma; EET, endoscopic Most current article
eradication therapy; EGD, esophagogastroduodenoscopy; GEJ, gastro-
esophageal junction; HGD, high-grade dysplasia; IMCa, intramucosal © 2022 by the AGA Institute
adenocarcinoma; LGD, low-grade dysplasia; NDBE, nondysplastic Bar- 1542-3565/$36.00
rett’s esophagus. https://doi.org/10.1016/j.cgh.2022.02.043
2764 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

ndoscopic eradication therapy (EET) is recom-

E mended by all society guidelines for the treatment
of Barrett’s esophagus (BE) with dysplasia or intra-
What You Need to Know
Background
mucosal adenocarcinoma (IMCa).1–3 After complete
eradication of intestinal metaplasia (CE-IM), patients Surveillance after endoscopic eradication of
undergo endoscopic surveillance at intervals largely dysplasia or intramucosal adenocarcinoma in Bar-
based on expert opinion.1,3–5 These intervals do not ac- rett’s esophagus is performed in frequent intervals
count for the reduction in cancer risk conferred by EET,6 and is not evidence-based.
resulting in a substantial number of procedures, partic- Findings
ularly in the 2 years after CE-IM. Cumulative recurrence incidence after successful
The timeline of recurrence after CE-IM varies across endoscopic therapy in a large international cohort
studies. Recurrence incidence is reported as higher in was modeled to determine optimal surveillance
the first year,7 second year,8 and remaining constant paradigms. For baseline high-grade dysplasia or
over time.9 Furthermore, although recurrent IM after intramucosal adenocarcinoma, optimal intervals
EET is not uncommon (8%–10% per year10), most re- were 6, 12, 18, and 24 months, and then annually.
currences are nondysplastic.9 The rate of dysplastic For baseline low-grade dysplasia, optimal intervals
recurrence is low at 2% per year.10–12 Thus, evidence- were 1, 2, and 4 years.
based guidelines for surveillance after CE-IM are
essential. Implications for patient care
A single prior study used national registry data and Surveillance intervals likely can be lengthened safely
proposed lengthening of surveillance intervals.13 How- after successful endoscopic therapy of Barrett’s
ever, this study was limited by many nondysplastic BE esophagus–related neoplasia, allowing for a sub-
(NDBE) patients undergoing ablation, lacked granularity stantial reduction in the number of endoscopic sur-
regarding actual timing of patient follow-up evaluation, veillance procedures.
and did not account for the delay between time of
recurrence and its detection.
We aimed to evaluate the cumulative incidence of occurred elsewhere, records were reviewed and included
recurrence after CE-IM in a large cohort of patients in the analysis.
treated with EET for BE-associated neoplasia, charac- We also endeavored to understand patterns of in-
terize their patterns of follow-up evaluation, and esti- terval follow-up evaluation. To reduce practice-related
mate the rate and duration of undetected recurrence for variability, we limited this analysis to the US sites that
various surveillance paradigms, aimed at guiding the function as a single integrated health care system. Re-
selection of optimal surveillance intervals. sults of each individual endoscopy and corresponding
histologic findings after CE-IM were recorded to char-
acterize patterns of post-EET follow-up evaluation. The
Methods UK data were used for all analyses in the study aside
from this follow-up timing analysis.
Patient Selection and Data Collection
Endoscopic Therapy and Surveillance
Adult patients with BE-low-grade dysplasia (LGD),
high-grade dysplasia (HGD), or IMCa undergoing EET Endoscopists with expertise in BE-associated
and achieving CE-IM between March 2004 and May 2017 neoplasia performed all procedures. Using high-
were identified from prospectively maintained databases definition white-light and narrow-band imaging, any
at 5 tertiary-referral academic centers with expertise in visible abnormality was resected. Radiofrequency
the management of BE. Three were located in the United ablation then was administered every 3 months until
States (Mayo Clinics in Minnesota, Arizona, and Florida) CE-IM was achieved. Patients were maintained on
and 2 in the United Kingdom (Nottingham and Cam- twice-daily proton pump inhibitor therapy during
bridge University). All diagnoses of dysplasia or IMCa treatment.
were confirmed by expert gastrointestinal pathologists. After CE-IM, endoscopic surveillance was recom-
Patients without endoscopic follow-up evaluation, preg- mended for LGD at 6 months, 12 months, and yearly
nant patients, and those with esophageal varices or a thereafter. For HGD/IMCa, surveillance was recom-
history of advanced esophageal adenocarcinoma (EAC) mended at 3, 6, 9, 12, 18, and 24 months, and yearly
were excluded. thereafter. During surveillance, any visible abnormal-
We previously published a study examining the ity was sampled separately. Biopsy specimens also
recurrence timeline and location using this database.9 In were obtained in 4 quadrants every 1 to 2 cm from
the present study, only patients with HGD/IMCa or LGD the gastroesophageal junction (GEJ) and tubular
at baseline were included. Clinical and demographic data esophagus encompassing the pretreatment BE
were abstracted manually. When follow-up evaluation segment.
December 2022 Optimized Surveillance Intervals After BE Ablation 2765

Definitions and Outcomes recurrences, including NDBE, generally were treated.

Because treatment lowers recurrence risk, failure to
CE-IM was defined as 2 endoscopic sessions at least 3 censor these NDBE recurrences in the dysplastic recur-
months apart during which no visible or histologic evi- rence analysis would result in cumulative incidence es-
dence of BE was found on biopsy specimens of the GEJ timates that are biased too low. Thus, analyses of
and tubular esophagus. Nondysplastic recurrence dysplastic and cancer recurrence assumed that the risk
included the finding of IM in biopsy specimens of the GEJ. of dysplastic/cancer recurrence was independent of
Dysplastic recurrence encompassed any recurrence with nondysplastic recurrence.
LGD, HGD, or IMCa. There were 3 separate clinical end
points: (1) any recurrence, (2) dysplastic recurrence, and Metrics for Comparison of Surveillance
(3) cancer recurrence. Scenarios

We focused on the initial 5 years after CE-IM for LGD

Statistical Analysis
patients and on the initial 2 years after CE-IM for HGD/
IMCa patients because this is the period of most frequent
A detailed description of the statistical analyses is
surveillance. The 2 primary metrics were the proportion
provided in the Supplementary Methods. All analyses
of patients with a wait of more than 3 or more than
were performed separately according to baseline histol-
6 months between the recurrence date and the date of
ogy (LGD or HGD/IMCa). To estimate the cumulative
endoscopy at which recurrence was detected. A range of
incidences of any recurrence, dysplastic recurrence, and
plausible surveillance scenarios were considered,
cancer recurrence for 5 years after CE-IM, we first esti-
including the American College of Gastroenterology
mated crude cumulative incidences using the
guidelines, and those proposed by Cotton et al.7,13
Kaplan–Meier method. Censoring occurred on the date of
The 2 primary metrics were calculated as follows for
the last esophagogastroduodenoscopy (EGD) for patients
each surveillance scenario based on the cumulative
without recurrence and recurrence was considered to
incidence estimates for any recurrence, dysplastic
occur on the date of the EGD at which it was detected.
recurrence, or cancer recurrence. For each targeted
This results in cumulative incidence estimates that are
follow-up EGD time point, the cumulative incidence of
biased too low.
the given outcome (any recurrence, dysplastic recur-
We used the following strategy to address this: Those
rence, or cancer recurrence) was estimated at both the
with any recurrence were each included in N different
prior targeted follow-up EGD time point (eg, CE-IM,
instances in a modified data set, where N equals the
6 mo, 1 y) as well as the targeted time point minus
difference between the date of the EGD at which recur-
either 3 months or 6 months. The latter estimate sub-
rence was detected (recurrence EGD) and the date of the
sequently was subtracted from the former, yielding a
previous EGD; each N instance was given a weight equal
separate difference value for each targeted follow-up
to 1 divided by N. For each instance, times to recurrence
EGD time point. These differences then were summed
after CE-IM were assigned uniformly distributed values
to calculate the proportion of patients with a wait of
across the range of dates between the previous EGD to
more than 3 months (or >6 months) between the date of
the recurrence EGD. This directly accounted for the un-
recurrence and the date of recurrence EGD under the
known true date of recurrence by allowing for recur-
given surveillance scenario. In baseline LGD patients, for
rence to occur on each possible intervening date, all of
surveillance scenarios that did not include a 5-year
which were weighted equally. We then re-estimated cu-
follow-up time point, an additional difference value
mulative recurrence incidence via the Kaplan–Meier
was included as the difference in cumulative incidences
method using the modified data set, accounting for the
of recurrence at the 5-year time point and the final
correlation between the N instances for each patient, and
follow-up time point. A visual illustration of this calcu-
assuming right-censoring. We opted for the current
lation is shown in Supplementary Figure 3.
approach over the traditional Kaplan–Meier method with
interval censoring because it directly accounts for the
possibility of recurrence on each day. There was strong Results
agreement in cumulative incidence estimates for our
method and traditional Kaplan–Meier estimates using Patient Characteristics and Clinical Outcomes
interval censoring (Supplementary Figures 1 and 2).
The cumulative incidences of dysplastic and cancer A total of 498 patients (420 US patients, 78 UK pa-
recurrence then were estimated for baseline HGD/IMCa tients) were included. Their baseline characteristics are
patients. Censoring occurred at either of the following shown in Table 1. After a median follow-up period of 2.6
dates: last EGD (for patients without recurrence), non- years (range, 0.2–10.6 y), recurrence was observed in
dysplastic recurrence (when analyzing dysplastic recur- 135 patients (27.1%), of whom 42 (8.4%) had a
rence), or noncancer recurrence (when analyzing cancer dysplastic recurrence. The location and timing of recur-
recurrence). This censoring scheme was used because rence in this population have been reported previously.9
2766 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

Table 1. Patient Demographics, Clinical Characteristics, and Outcomes

Characteristics LGD HGD IMCa

N 115 288 95
Age, y, median (IQR) 67 (62–72) 68 (61–74) 68 (62–74)
Male sex, n (%) 96 (83.5) 242 (84.0) 87 (91.6)
Length of BE, cm, median (IQR) 5 (3–7) 4 (2–6) 3 (1–6)
Follow-up period, mo, median (IQR) 29.2 (13.8–52.1) 23.7 (12.0–37.8) 21.9 (12.2–42.1)
Recurrence
Nondysplastic 18 56 19
LGD 0 15 3
HGD 0 8 2
Cancer 3 7 4

BE, Barrett’s esophagus; HGD, high-grade dysplasia; IMCa, intramucosal adenocarcinoma; IQR, interquartile range; LGD, low-grade dysplasia.

Recurrence Incidence and Histology shown in Figure 2 and Table 2; the cumulative incidence of
any recurrence was 10.8% (95% CI, 4.9%–16.3%) at 2
A summary of clinical outcomes is presented in years and 23.3% (95% CI, 12.6%–32.6%) at 5 years.
Figure 1. For the 383 patients with baseline HGD/IMCa, Table 2 further describes the incidence of recurrence
recurrence occurred in 114 (29.7%), of which 39 in relation to baseline pretreatment histology. In total, 14
(10.2%) were dysplastic (Table 1). Cumulative incidence patients developed EAC after successful endoscopic
estimates for all recurrences and dysplastic recurrences eradication, 10 of whom had either LGD or HGD before
after CE-IM are shown in Figure 2 and Table 2 for the ablation. Eight of 14 (57.1%) recurrent EACs were IMCa,
baseline HGD/IMCa subgroup. The cumulative incidence while the remaining 6 cases were more deeply invasive
of any recurrence was 25.3% (95% CI, 21.1%–30.2%) at and required either chemotherapy/radiation, surgery, or
2 years and 40.2% (9% CI, 32.9%–46.8%) at 5 years. The a combination thereof. A single patient died as a conse-
cumulative incidence of dysplastic recurrence (LGD or quence of recurrent EAC.
higher grade) was 8.9% (95% CI, 5.9%–11.8%) at 2
years and 16.4% (10.2%–22.2%) at 5 years.
In baseline LGD (N ¼ 115), recurrence occurred in 21 Intervals of Actual Follow-Up Evaluation
(18.3%) patients, of which 18 (15.7%) were nondysplastic
and only 3 (2.7%) were dysplastic (Table 1). In these pa- The specific dates of each surveillance endoscopy
tients, the cumulative incidences of any recurrence is were tabulated for all US patients and frequency data are

Figure 1. Flow chart of

patient outcomes. HGD,
high-grade dysplasia;
IMCa, intramucosal
adenocarcinoma; LGD,
low-grade dysplasia.
December 2022 Optimized Surveillance Intervals After BE Ablation 2767

curves separately for LGD and HGD/IMCa patients

(Table 3). Because of the small number of LGD patients
experiencing dysplastic recurrence, it was not examined
separately for this subgroup. For baseline LGD patients,
metrics were calculated for a 5-year time frame of
follow-up evaluation after CE-IM. For patients with
baseline HGD/IMCa, all metrics were assessed only for
the first 2 years of follow-up evaluation because this time
frame is associated with the most intensive surveillance
frequency.
Each surveillance paradigm shows a trade-off be-
tween the number of endoscopies and the duration of
undetected recurrence. Based on an assessment of these
parameters, for baseline HGD/IMCa we recommend a
strategy of 6-, 12-, 18-, and 24-month surveillance after
CE-IM, and annually thereafter. Although the strategy
advocated by Cotton et al13 would result in fewer
dysplastic recurrences undetected for more than
3 months, it would increase the proportion of recurrent
HGD/EAC left undetected for more than 6 months. This
proposed strategy would reduce the number of endo-
scopic surveillance sessions by 33% compared with
those advocated by the American College of Gastroen-
terology guidelines. For LGD, we recommend a strategy
of 1-, 2-, and 4-year surveillance in the first 5 years.
Compared with the strategy suggested by Cotton et al,13
this would result in fewer dysplastic and cancer re-
Figure 2. Cumulative incidence of recurrence for patients currences undetected for more than 6 months while still
with low-grade dysplasia (LGD) or high-grade dysplasia
(HGD)/intramucosal adenocarcinoma (IMCa) at baseline. reducing the number of endoscopic surveillance sessions
Kaplan–Meier curves representing the cumulative incidence by 48% compared with current guidelines. In both
of recurrence after complete eradication of intestinal meta- groups, the limited number of patients and recurrence
plasia in patients with (A) LGD or (B) HGD/IMCa at baseline outcomes beyond 5 years did not allow for further
before endoscopic eradication therapy. (A) Cumulative inci- extrapolation beyond this time frame.
dence of any recurrence (dysplastic or nondysplastic) is
shown as a solid blue line, with blue dotted lines denoting
95% CIs. (B) cumulative incidence of dysplastic recurrence is
shown as a solid red line, with red dotted lines denoting 95% Discussion
CI. Additional crude cumulative incidence estimates are
shown, assuming recurrence occurred at the time of detec- After successful EET for dysplastic BE, currently
tion (solid black line) and during before the endoscopic ses- recommended endoscopic surveillance is frequent and
sion (dotted black line)
dysplastic recurrence is uncommon, leading some to
suggest intervals should be lengthened. In this study, we
shown in Figure 3. These are shown as time between
analyzed a large multicenter international cohort of CE-
sessions rather than absolute time from CE-IM because
IM patients treated for dysplastic BE to generate recur-
subsequent surveillance recommendations would have
rence estimates, model surveillance paradigms, and
been issued based on the actual date of endoscopy. The
suggest optimal surveillance intervals. This study
median time from CE-IM (defined as 2 consecutive en-
modeled the duration of undetected recurrence. We
doscopies with no BE on histology) to first endoscopic
propose lengthening surveillance for HGD/IMCa after
surveillance was 365 days (range, 74–1111 d) for LGD
CE-IM to 6, 12, 18, and 24 months, and then annually,
and 183 days (range, 70–1000 d) in HGD/IMCa. The time
and for LGD we recommend a strategy of 1-, 2-, and
frames between subsequent surveillance sessions varied
4-year surveillance in the first 5 years. These amended
widely, but most commonly occurred at either 6- or
strategies would reduce the number of endoscopies by
12-month intervals.
33% to 48%, substantially reduce cost, and lead to
comparable clinical outcomes. These findings have a
Assessment of Metrics for Comparison of direct and substantial impact on the management of BE
Surveillance Scenarios patients successfully completing EET.
Current evidence to support lengthening of post-EET
The 2 primary metrics for comparison of surveillance surveillance is limited to very few studies.13,14 The only
scenarios were calculated based on cumulative incidence prior study to explicitly examine surveillance intervals
2768 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

Table 2. Cumulative Incidence of Any Recurrence (95% CI) After CE-IM According to Baseline Histology

Length of time after CE-IM

Baseline Rate/100
histology N Person-years Recurrence 1y 2y 5y person-years

LGD 115 359.6 All 7.7 (2.7–12.4) 10.8 (4.9–16.3) 23.3 (12.6–32.6) 5.8
Dysplastic 0.9 (0.0–2.7) 0.9 (0.0–2.7) 3.1 (0.0–7.6) 0.8
EAC 0.9 (0.0–2.7) 0.9 (0.0–2.7) 3.1 (0.0–7.6) 0.8
HGD/IMCa 383 913.1 All 19.4 (15.5–23.2) 25.8 (21.1–30.2) 40.2 (32.9–46.8) 12.5
Dysplastic 7.2 (4.6–9.7) 8.9 (5.9–11.8) 16.4 (10.2–22.2) 4.3
EAC 4.2 (2.2–6.2) 4.9 (2.6–7.0) 9.5 (4.5–14.3) 1.2

CE-IM, complete eradication of intestinal metaplasia; EAC, esophageal adenocarcinoma; HGD, high-grade dysplasia; IMCa, intramucosal adenocarcinoma; LGD,
low-grade dysplasia.

after CE-IM13 used registry data to build models and term follow-up evaluation.14 These data provide
select the optimal surveillance paradigm. The in- compelling evidence for the re-examination of currently
vestigators constructed a model to yield a 0.1% per-visit recommended surveillance intervals.
rate of invasive EAC (approximate risk of serious Our study differs from the one by Cotton et al13 in
complication from a surveillance endoscopy), corre- several important ways. Our data were derived directly
sponding to a 2.9% neoplastic recurrence rate. The from patient medical records, allowing for granularity
optimal surveillance intervals for LGD was at 1 and with respect to confirmation of dates, histology, and
3 years after CE-IM, and for HGD/IMCa at 3, 6, and patient outcomes. Importantly, we used our recurrence
12 months, and annually thereafter. These intervals data to generate metrics for the duration of undetected
would reduce the number of surveillance endoscopies recurrence, rather than overall observed neoplastic
over 5 years by 38%, while still meeting the specified recurrence. This was done to provide a range of out-
neoplastic recurrence threshold. A Dutch study also comes across modeled surveillance paradigms and allow
retrospectively examined centralized data on BE sur- clinicians and guideline authors to weigh these factors in
veillance after EET and showed that frequent vs annual selecting the optimal surveillance timing. Although the
surveillance in the first year after CE-IM did not affect precise biology of recurrent BE after CE-IM is not known,
the outcome of dysplastic recurrence in short- or long- data drawn from surveillance of treatment-naïve

Figure 3. Actual follow-up intervals by visit after complete eradication of intestinal metaplasia (CE-IM). (A) Low-grade dysplasia
at baseline. (B) High-grade dysplasia/intramucosal adenocarcinoma at baseline. IQR, interquartile range.
December 2022 Optimized Surveillance Intervals After BE Ablation 2769

Table 3. Summary of Recurrence Metrics in Different Surveillance Paradigms

6 mo, 1, 2, 3, 4, 5 y 1, 3 y
Metric (ACG) 1, 2, 3, 4, 5 y 3, 6 mo, 2, 4 y 1, 2, 4 y (Cotton)

LGD pre-ablation
Sessions over 5 years if no recurrence, 6 5 4 3 2
Patients who have any recurrence that is not 15.8 17.9 16.4 20.0 21.1
detected until >3 mo, %
Patients who have any recurrence that is not 7.7 12.4 14.2 16.0 18.6
detected until >6 mo, %
HGD/IMCa pre-ablation
Sessions patient will have >2 years if no recurrence, n 6 5 4 4 3
Patients who have any recurrence that is not 3.5 7.2 8.8 14.2 15.8
detected until >3 mo, %
Patients who have any recurrence that is not 0.0 0.0 3.9 0.0 3.9
detected until >6 mo, %
Patients who have dysplastic recurrence that is not 0.9 2.5 2.9 4.7 5.1
detected until >3 mo, %
Patients who have dysplastic recurrence that is not 0.0 0.0 1.1 0.0 1.1
detected until >6 mo, %
Patients who have cancer recurrence that is not 0.3 1.0 1.1 1.6 1.7
detected until >3 mo, %
Patients who have cancer recurrence that is not 0.0 0.0 0.4 0.0 0.4
detected until >6 mo, %

ACG, American College of Gastroenterology; HGD, high-grade dysplasia; IMCa, intramucosal adenocarcinoma; LGD, low-grade dysplasia.

dysplastic BE can provide some insight. The risk of EAC counseling on surveillance intervals, it is evident that
has been estimated at 6% to 7% annually in HGD and at recommendations are not strictly followed by patients.
0.5% to 1% in LGD.15,16 What degree of time, then, can Potential reasons for this include patient factors (eg,
pass safely before detection of recurrent BE or dysplasia? socioeconomic barriers, personal events, and so forth)
In a study of 79 patients with HGD undergoing intensive and system factors (eg, scheduling delays, lack of
surveillance (every 3 months for 1 year, decreased only if centralized BE patient tracking, and so forth).
HGD regressed) for a mean of 7.3 years, no single patient Indeed, this large real-world cohort with less frequent
died from complications of esophageal cancer.17 Endo- surveillance than currently recommended clearly shows
scopic therapy for HGD or IMCa also typically is con- the implications of this approach. Of 498 patients, only
ducted at 3-month intervals. It therefore may be 6 patients (1.2%) developed invasive EAC requiring
reasonable to presume that a dysplastic recurrence left surgery and/or chemoradiation. Furthermore, only 1 of
undetected for 3 months would be unlikely to result in these patients (0.2% of total patients) died from com-
advanced or metastatic EAC. plications of metastatic EAC and another died from un-
Another important difference is the use of a definition known causes. Thus, despite HGD/IMCa patients having
of CE-IM as 2 consecutive endoscopic sessions. Sampling more than 60% fewer than recommended endoscopies in
error is critical to avoiding misclassification bias when the first 2 years after CE-IM and LGD patients having
assigning definitions of eradication and recurrence. 28% fewer endoscopies than recommended in 5 years
Recent studies of recurrence after CE-IM have conducted after CE-IM, the clinical outcomes were acceptable
subset analyses modifying this definition from a single to overall. The EAC incidence in this study was 1.2 per 100
2 endoscopic sessions and shown no substantial change person-years for pre-ablation HGD/EAC and 0.8 for pre-
in their findings.7,8 However, the number of patients in ablation LGD. This is higher than the incidence rates of
each of these studies was relatively modest when 0.6 and 0.51, respectively, seen in patients enrolled in the
compared with the Cotton et al13 modeling study and AIM Dysplasia randomized trial6 and likely is attribut-
this effect may have had a greater impact on the early able to lower adherence to follow-up recommendations
recurrence estimates. A meta-analysis of 1973 patients outside of a trial protocol.
showed that recurrence was higher in the first year for The selection of a best surveillance paradigm thus is
CE-IM defined as a single endoscopy, suggesting this inherently complex and multidimensional. Is there a
impacts timing of detection more than recurrence rate tolerable level of invasive EAC during surveillance? It is
itself.18 clear from this study and others referenced that invasive
Importantly, this study also examined the actual EAC can occur even in a relatively stringent surveillance
follow-up intervals of patients with dysplastic BE or program. Our data provide an opportunity to weigh the
IMCa after successful CE-IM. Despite extensive duration of undetected recurrence against the number of
2770 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

endoscopic sessions in a real-world scenario balancing a baseline histology and involve patients in the complex
33% to 48% reduction in endoscopic evaluations against decision making that governs their resources and risk.
reasonable outcomes.
We acknowledge certain limitations. By virtue of the
Supplementary Methods
retrospective design, treatment protocols could have
varied between endoscopists. However, given the
expertise of treating endoscopists in BE EET, this is un- Note: To access the supplementary material accom-
likely. Furthermore, the observed recurrence incidence panying this article, visit the online version of Clinical
was comparable with those reported in large randomized Gastroenterology and Hepatology at www.cghjournal.org,
trials.19,20 The study was performed in expert centers in and at https://doi.org/10.1016/j.cgh.2022.02.043.
2 countries, which may limit global generalizability to
other populations. The modeling analysis requires the References
assumption that rate of dysplastic recurrence was inde- 1. Shaheen NJ, Falk GW, Iyer PG, et al. ACG clinical guideline:
pendent of prior nondysplastic recurrence. A recent diagnosis and management of Barrett’s esophagus. Am J
study supported this notion because dysplastic recur- Gastroenterol 2016;111:30–50; quiz 51.
rence incidence was not significantly different between 2. Standards of Practice Committee, Wani S, Qumseya B, et al.
those who did or did not develop GEJ nondysplastic Endoscopic eradication therapy for patients with Barrett’s
esophagus-associated dysplasia and intramucosal cancer.
recurrence.21 Nevertheless, it is conceivable that the
Gastrointest Endosc 2018;87:907–931 e9.
relationship between nondysplastic and subsequent
3. Fitzgerald RC, di Pietro M, Ragunath K, et al. British Society of
dysplastic recurrence could impact these estimates. We
Gastroenterology guidelines on the diagnosis and management
acknowledge that the clinical significance of nondys- of Barrett’s oesophagus. Gut 2014;63:7–42.
plastic recurrence after CE-IM is a matter of some con- 4. American Gastroenterological Association, Spechler SJ,
troversy. However, we deliberately included Sharma P, et al. American Gastroenterological Association
nondysplastic recurrence as an outcome in this analysis medical position statement on the management of Barrett’s
for several reasons. Our pre-ablation concept of carci- esophagus. Gastroenterology 2011;140:1084–1091.
nogenesis assumes that all dysplastic recurrences arise 5. ASGE Standards of Practice Committee, Qumseya B, Sultan S,
from nondysplastic ones and we endoscopically treat all et al. ASGE guideline on screening and surveillance of Barrett’s
recurrences. Therefore, the precise natural history of esophagus. Gastrointest Endosc 2019;90:335–359 e2.
untreated NDBE after CE-IM is not known and progres- 6. Shaheen NJ, Overholt BF, Sampliner RE, et al. Durability of
sion rates in treatment-naïve NDBE do not provide a radiofrequency ablation in Barrett’s esophagus with dysplasia.
valid comparison. An analysis that removes nondys- Gastroenterology 2011;141:460–468.
plastic recurrences and creates estimates solely on the 7. Cotton CC, Wolf WA, Overholt BF, et al. Late recurrence of
basis of dysplastic recurrence incidence would not be a Barrett’s esophagus after complete eradication of intestinal
metaplasia is rare: final report from ablation in intestinal meta-
valid statistical model for the outcome of interest.
plasia containing dysplasia trial. Gastroenterology 2017;
Follow-up data were based on return to the treating fa-
153:681–688 e2.
cility. Patients could have undergone unreported sur-
8. Wani S, Han S, Kushnir V, et al. Recurrence is rare following
veillance at other facilities, although clinical complete eradication of intestinal metaplasia in patients with
documentation was reviewed carefully to detect this. Barrett’s esophagus and peaks at 18 months. Clin Gastroenterol
Finally, the number of patients with LGD, and by exten- Hepatol 2020;18:2609–2617 e2.
sion the number of recurrences in this group, was rela- 9. Sami SS, Ravindran A, Kahn A, et al. Timeline and location of
tively small and therefore limited predictive estimates recurrence following successful ablation in Barrett’s oesoph-
based on dysplastic recurrence. agus: an international multicentre study. Gut 2019;
In conclusion, in this study of dysplastic BE patients 68:1379–1385.
achieving CE-IM we used recurrence data to model par- 10. Krishnamoorthi R, Singh S, Ragunathan K, et al. Risk of recur-
adigms for more efficient endoscopic surveillance. rence of Barrett’s esophagus after successful endoscopic
Despite suboptimal adherence to recommended surveil- therapy. Gastrointest Endosc 2016;83:1090–1106 e3.
lance timing, patient outcomes generally were good. 11. Fujii-Lau LL, Cinnor B, Shaheen N, et al. Recurrence of intestinal
Furthermore, we suggest strategies that substantially metaplasia and early neoplasia after endoscopic eradication
therapy for Barrett’s esophagus: a systematic review and meta-
reduce the number of endoscopies with only a modest
analysis. Endosc Int Open 2017;5:E430–E449.
increase in the proportion of patients with undetected
12. Tan MC, Kanthasamy KA, Yeh AG, et al. Factors associated with
recurrence at more than 6 months, in line with a recent
recurrence of Barrett’s esophagus after radiofrequency ablation.
modeling study suggesting similarly expanded strategies. Clin Gastroenterol Hepatol 2019;17:65–72 e5.
These data are another important step forward in 13. Cotton CC, Haidry R, Thrift AP, et al. Development of evidence-
crafting evidence-based guidelines for post–CE-IM sur- based surveillance intervals after radiofrequency ablation of
veillance, they provide additional metrics to assess Barrett’s esophagus. Gastroenterology 2018;155:316–326.e6.
various surveillance paradigms and incorporate an un- 14. van Munster S, Nieuwenhuis E, Weusten B, et al. Long-term
derstanding of real-world follow-up evaluation. Further outcomes after endoscopic treatment for Barrett’s neoplasia
studies are needed to personalize strategies beyond with radiofrequency ablation þ/- endoscopic resection: results
December 2022 Optimized Surveillance Intervals After BE Ablation 2771

from the national Dutch database in a 10-year period. Gut 2022;

71:265–276. Reprint requests
Address requests for reprints to: Prasad G. Iyer, MD, MSc, Division of
15. Rastogi A, Puli S, El-Serag HB, et al. Incidence of esophageal Gastroenterology and Hepatology, Mayo Clinic, 200 1st Street SW, Rochester,
adenocarcinoma in patients with Barrett’s esophagus and high-grade Minnesota 55905. e-mail: [email protected]; fax: (507) 255-7612.
dysplasia: a meta-analysis. Gastrointest Endosc 2008;67:394–398.
CRediT Authorship Contributions
16. Singh S, Manickam P, Amin AV, et al. Incidence of esophageal Allon Kahn, MD (Data curation: Lead; Writing – original draft: Lead; Writing
adenocarcinoma in Barrett’s esophagus with low-grade – review & editing: Lead)
dysplasia: a systematic review and meta-analysis. Gastrointest Julia Crook, PhD (Data curation: Equal)
Michael Heckman, MS (Data curation: Equal)
Endosc 2014;79:897–909 e4; quiz 83 e1, 83 e3. Mikolaj Wieczorek, BS (Data curation: Equal)
17. Schnell TG, Sontag SJ, Chejfec G, et al. Long-term nonsurgical Sarmed Sami, MBChB, PhD (Data curation: Equal)
Diana Snyder, MD (Data curation: Equal)
management of Barrett’s esophagus with high-grade dysplasia. Siddharth Agarwal, MBBS (Data curation: Equal)
Gastroenterology 2001;120:1607–1619. Jose Santiago, MD (Data curation: Equal)
Jacobo Ortiz Fernandez-Sordo, MD (Data curation: Equal)
18. Sawas T, Iyer PG, Alsawas M, et al. Higher rate of Barrett’s W. Keith Tan, MD (Data curation: Equal)
detection in the first year after successful endoscopic therapy: Ramona Lansing, RN (Data curation: Equal)
meta-analysis. Am J Gastroenterol 2018;113:959–971. Kenneth K. Wang, MD (Data curation: Equal)
Krish Ragunath, MD (Data curation: Equal)
19. Phoa KN, Pouw RE, Bisschops R, et al. Multimodality endo- Massimiliano DiPietro, MD (Data curation: Equal)
scopic eradication for neoplastic Barrett oesophagus: results of Herbert Wolfsen, MD (Data curation: Equal)
Francisco Ramirez, MD (Data curation: Equal)
an European multicentre study (EURO-II). Gut 2016;65:555–562. David Fleischer, MD (Data curation: Equal)
20. Pouw RE, Klaver E, Phoa KN, et al. Radiofrequency ablation for Cadman L. Leggett, MD (Data curation: Equal)
low-grade dysplasia in Barrett’s esophagus: long-term outcome Prasad G. Iyer, MD MS (Supervision: Lead; Writing – original draft:
Supporting; Writing – review & editing: Supporting)
of a randomized trial. Gastrointest Endosc 2020;92:569–574.
21. Solfisburg QS, Sami SS, Gabre J, et al. Clinical significance of Conflicts of interest
recurrent gastroesophageal junction intestinal metaplasia after This author discloses the following: Prasad Iyer has received research
funding from Exact Sciences and Pentax Medical, and has consulted for
endoscopic eradication of Barrett’s esophagus. Gastrointest Medtronic, Ambu, and Symple Surgical. The remaining authors disclose no
Endosc 2020. conflicts.
2771.e1 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

Supplementary Methods occurred, all of which were given equal weight). We then
estimated the cumulative incidence of recurrence using
the Kaplan–Meier method using this modified data set,
Statistical Analysis accounting for the correlation between the N instances for
each patient, and assuming right censoring. This is
All statistical analyses were performed separately perhaps more simply illustrated with an example. Take a
according to baseline histology (LGD or HGD/IMCa) and patient with no recurrence detected on the previous EGD
were conducted using R version 4.0.3 (Boston, MA). To that was 100 days after CE-IM, and recurrence detected on
estimate the cumulative incidences of the separate out- the current EGD that occurred 50 days after the previous
comes of any recurrence (both baseline LGD and HGD/ EGD (ie, 150 days after CE-IM). This patient would be
IMCa groups), dysplastic recurrence (baseline HGD/IMCa included in the modified data set in 50 different instances
patients only), and cancer recurrence (baseline HGD/ (150 minus 100), with each of these instances given a
IMCa patients only) for 5 years after CE-IM, we first weight equal to 1/50 ¼ 0.02. These 50 different instances
estimated the crude cumulative incidences using the then would be assigned 50 different time to recurrence
Kaplan–Meier method, in which censoring occurred on values – 101, 102, 103, .., 149, and 150.
the date of the last EGD for patients without recurrence, The cumulative incidences of dysplastic recurrence
and recurrence was considered to occur on the date of and cancer recurrence were estimated in the same way
the EGD at which it was detected. These cumulative in baseline HGD/IMCa patients; censoring occurred at
incidence estimates are crude because the true date of either the date of the last EGD (for patients without any
recurrence was most likely not on the exact date of the recurrence), the date of nondysplastic recurrence in
current EGD, but instead at some unknown time point analysis of the dysplastic recurrence outcome, or the
between the date of the EGD at which recurrence was date of noncancer recurrence in analysis of the cancer
detected and the day after the date of the previous EGD recurrence outcome. This censoring scheme was used
(at which no recurrence was detected). Therefore, the because recurrences, even nondysplastic recurrences,
resulting cumulative crude incidence estimates were generally were treated and therefore any postrecurrence
biased too low. EGD was not reflective of the initial EET therapy. In
We used the following strategy to address this issue. addition, failure to censor nondysplastic recurrences at
When considering the any recurrence outcome, for pa- the time of nondysplastic recurrence in analysis of the
tients who developed recurrence, these patients were dysplastic recurrence outcome would result in cumula-
each included in N different instances in a modified data tive incidence estimates that were biased too low
set, where N is equal to the difference in days between because risk of subsequent recurrence would be lowered
the date of the EGD at which recurrence was detected by the secondary treatment. As such, analyses of
and the date of the previous EGD (at which no recur- dysplastic recurrence and cancer recurrence made the
rence was detected); each of these N instances was given assumptions that the risk of dysplastic recurrence was
a weight equal to 1 divided by N in the analysis. For each independent of nondysplastic recurrence, and that the
of these N instances, times to recurrence after CE-IM risk of cancer recurrence was independent of noncancer
were assigned as values that were distributed uni- recurrence.
formly across the range of the dates from the day after The timing of all individual surveillance EGDs was
the previous EGD to the EGD at which recurrence was collected for US patients. The distributions of the times
detected, thereby directly accounting for the unknown between follow-up visits were summarized in histo-
true date of recurrence (by allowing for a date of grams to examine how they compared with recom-
recurrence to occur on each possible date it could have mended surveillance guidelines over the first 2 years.
December 2022 Optimized Surveillance Intervals After BE Ablation 2771.e2

Supplementary Figure 1. Cumulative incidence of any

recurrence after complete eradication of intestinal metaplasia
for patients with low-grade dysplasia at baseline according to
our method of estimation and that of traditional interval
censoring using the Kaplan–Meier method.

Supplementary Figure 2. Cumulative incidence of dysplastic

recurrence of complete eradication of intestinal metaplasia
for patients with high-grade dysplasia or intramucosal
adenocarcinoma at baseline according to our method of
estimation and that of traditional interval censoring using the
Kaplan–Meier method.
2771.e3 Kahn et al Clinical Gastroenterology and Hepatology Vol. 20, No. 12

Supplementary Figure 3. Graphic illustration of the calculation of the proportion of patients for whom there would be a wait of
more than 3 months between the date of recurrence and the data from the esophagogastroduodenoscopy (EGD) at which the
recurrence was detected, using an example of the American College of Gastroenterology (ACG) surveillance scheme (6 mo,
and 1, 2, 3, 4, 5 y) in baseline low-grade dysplasia (LGD) patients. The solid blue line represents the cumulative incidence of
any recurrence after complete eradication of intestinal metaplasia in baseline LGD patients. The proportion of interest was
calculated by subtracting cumulative incidences from 3 months and 0 months (ie, Diff 1), from 9 months and 6 months (ie, Diff
2), from 1.75 years and 1 year (ie, Diff 3), from 2.75 years and 2 years (ie, Diff 4), from 3.75 years and 3 years (ie, Diff 5), and from
4.75 years and 4 years (ie, Diff 6). The sum of the 6 Diff values then was calculated and was the proportion of patients for whom
there would be a wait of more than 3 months between the date of recurrence and the data of EGD at which the recurrence was
detected for the ACG surveillance scheme in baseline LGD patients.