Introduction
High risk T1 esophageal adenocarcinoma (EAC) is defined as: cancer invading the mucosa
(T1a) with the presence of poor tumor differentiation or lymphovascular invasion (LVI);
or cancer invading the submucosa (T1b) with or without these high risk features. The
conventional approach for managing high risk T1N0M0 EAC has been surgical resection,
involving esophagectomy and lymphadenectomy, to remove the cancer and any potential
lymph node metastasis (LNM). However, esophagectomy carries considerable mortality
(up to 6%) and morbidity rates (1.7%–49.5%), and may result in lifelong functional
complaints, even in high volume centers [1 ]
[2 ]
[3 ].
Recent advancements in endoscopic techniques have facilitated radical endoscopic resection
(ER) of early esophageal cancers, even high risk T1 EAC, using methods like endoscopic
submucosal dissection (ESD), which has become even more efficient when combined with
traction techniques [4 ]
[5 ]. The optimal post-ER management for high risk T1 EAC is still debated owing to the
uncertain but increased risk of LNM [6 ]. LNM rates of 0–46% for T1 EAC have prompted guidelines to recommend additional
esophagectomy with lymph node resection [7 ]
[8 ]
[9 ]; however, a small number of studies exploring endoscopic surveillance as an alternative
post-ER approach have demonstrated its feasibility and safety for selected patients
with favorable tumor characteristics (<500-µm invasion [sm1], no LVI, and well-to-moderately
differentiated), particularly for those at high risk of surgical complications [4 ]
[10 ]
[11 ]. These endoscopy-focused publications report lower LNM rates for T1b EAC (0–16%),
compared with earlier surgical series, although their small cohort sizes and mostly
retrospective designs may have introduced bias. Further research is needed to clarify
LNM risk and management outcomes.
This study aimed to assess the outcomes in a larger cohort of patients who, following
radical ER for T1 EAC with at least one high risk feature, either underwent surgical
resection or entered endoscopic surveillance.
Methods
Study design
This was a retrospective multicenter study involving 11 tertiary referral centers
in Europe and Australia who were collaborating on large scale studies on early Barrett’s
neoplasia management. The institutional review board (IRB) of the Amsterdam University
Medical Centers declared that the study registry was not subject to the Medical Research
Involving Human Subjects act, waiving the need for formal ethical review and patient
consent. Each participating center’s IRB reviewed and approved the protocol.
Study population
We identified patients who underwent ER for T1 EAC with at least one high risk feature
between January 2008 and December 2019. Cases were mostly extracted from existing
databases, although one center conducted a manual search. Notably, not all centers
had initiated endoscopic mucosal resection (EMR) and ESD procedures by 2008.
Patients were included if they had a tumor-negative vertical (deep) resection margin
(R0v). Tumor extension at the horizontal (lateral) margin was not considered an exclusion
criterion, provided a radical ER had been performed. We categorized cases into the
following three histologic risk groups.
High risk T1a EAC (HR-T1a ): mucosal EAC with poor-to-no differentiation (G3–4) and/or LVI
Low risk T1b EAC (LR-T1b ): submucosal EAC with superficial invasion (<500 µm; sm1), well-to-moderately differentiated
(G1–2) and no LVI
High risk T1b EAC (HR-T1b ): submucosal EAC with deep invasion (≥500 µm; sm2–3), and/or G3–4, and/or LVI.
Exclusion criteria were: (i) tumor-positive (R1v) or inconclusive vertical resection
margin; (ii) residual/metachronous tumor ineligible for endoscopic retreatment present
at the first endoscopy following ER; (iii) baseline metastatic disease; (iv) prior
EAC treatment; (v) use of chemo-/radiotherapy; (vi) no follow-up or management initiated;
(vii) follow-up data unavailable.
This study did not include patients from the prospective PREFER study (NCT03222635),
or Dutch patients from prior studies on this topic [12 ]
[13 ]. There is overlap with the prior cohorts in the studies of Graham et al. [14 ] (n = 8), and Benech et al. [15 ] and Doumbe-Mandengue et al.
[16 ] (n = 10), although our cohort has a longer follow-up period.
Endoscopic resection
ERs were conducted using cap- or band-assisted EMR techniques, or ESD by endoscopists
with experience in managing Barrett’s neoplasia ([Fig. 1 ]).
Fig. 1 Endoscopic images of a high risk T1b lesion treated by endoscopic resection (ER) in
an 81-year old patient with significant co-morbidity who entered endoscopic surveillance
post-ER showing: a a Paris type 0-Is lesion of 25 mm in diameter within Barrett’s esophagus, delineated
with electrocoagulation marks prior to endoscopic submucosal dissection (ESD); b the mucosal incision at the oral side of the lesion; c signs of deep submucosal invasion found during submucosal dissection; d,e the wound after radical ER of the lesion, which histologically was a radically resected
(R0), poorly differentiated (G3) esophageal adenocarcinoma, invading over 2 mm into
the submucosa (sm3), with signs of lymphovascular invasion; f the healed ESD scar with squamous mucosa at the restaging endoscopy 8 weeks after
ESD.
Pathology assessment
ER specimens were assessed by experienced gastrointestinal pathologists adhering to
the seventh edition of the International Union Against Cancer (UICC) TNM classification
[17 ]. For T1a tumors, a distinction was made between those invading the lamina propria
(m2) and the muscularis mucosae (m3). T1b tumors were categorized by depth of submucosal
invasion: <500 µm (sm1) or ≥500 µm (sm2–3). The ER was considered radical if the vertical
margin was tumor-free (R0v). In this study, we re-evaluated the endoscopy and pathology
reports of all cases with surgically staged T1 disease initially marked as ER R0v.
Staging examinations
During the inclusion period (2008–2019), staging and follow-up protocols for high
risk T1 EAC varied and included endoscopies (with or without endoscopic ultrasound
[EUS]) and computed tomography (CT) or positron emission tomography (PET)-CT scans
based on physician preference.
Post-ER management
Additional surgery
Surgical strategies, including minimally invasive and open thoracolaparoscopic esophagectomies,
were chosen based on the tumor location and surgeon’s preference, with lymph node
resection documented in most cases. Following surgery, a new TNM staging was determined.
Endoscopic surveillance
Conducted at the original ER center, surveillance endoscopies were scheduled at the
treating physician’s discretion and included imaging (EUS, CT, and/or PET-CT) as needed.
Study end points
The primary end point was the risk of LNM and distant metastasis during follow-up.
Secondary end points were the rates of local recurrence requiring surgery in those
under endoscopic surveillance, and disease-specific, other-cause, and overall mortality
during follow-up.
Data collection
Research fellows (M.D.) or nurses entered baseline and follow-up data on standardized
forms in a joint online database (Castor EDC), with each institution maintaining a
patient identification file. Missing data and illogical values were completed and
corrected where possible. The database closed on 25 July 2023, with all authors reviewing
and approving the final data.
Statistical analysis
Statistical analysis was performed using IBM SPSS Statistics version 28.0.1.1 and
R version 4.4.2. Descriptive statistics included mean (SD) for normally distributed
variables, and median with interquartile range (IQR) for non-normal variables. Categorical
variables are presented as counts with percentages. Exact 95%CIs for proportions were
calculated using the exact binomial test in R to account for non-normal distributions.
To compare subcohorts, in SPSS, the independent samples t test was used for continuous variables, and the chi-squared test for categorical
variables, or Fisher’s exact test when expected cell counts were <5. All tests were
two sided with a significance level of 0.05.
Follow-up duration was calculated from the initial ER to the last hospital contact,
metastatic event, or death. Endoscopic follow-up was calculated from the initial ER
to the last endoscopy. Kaplan–Meier was used for survival analysis, and the log-rank
test was used to test for differences between subcohorts. The annual risk for recurrent
disease was calculated by dividing the number of metastatic cases by the total follow-up
time in years.
Results
Patient cohort
Between January 2008 and December 2019, 242 patients underwent staging ER for high
risk T1 EAC in Barrett’s esophagus, with 106 meeting the inclusion criteria (86 men;
mean age at time of ER 70 years [SD 11]). Baseline endoscopic characteristics of the
included patients are presented in [Table 1 ]. Details of exclusions are shown in [Fig. 2 ]. The excluded cases included six surgical cases that were initially categorized
and included as ER R0v but, owing to evidence of (residual) T1 EAC in the surgical
specimen, were reassessed, with there being clear arguments to register them as R1v
or inconclusive, leading to their eventual exclusion (Table 1s , see online-only Supplementary material).
Table 1 Baseline endoscopic characteristics of the 106 included patients.
HR-T1a n = 43
LR-T1b n = 27
HR-T1b n = 36
EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; HR-T1a,
intramucosal EAC with poor/no differentiation and/or lymphovascular invasion; HR-T1b,
submucosal EAC with ≥500-µm invasion, poor/no differentiation, and/or lymphovascular
invasion; IQR, interquartile range; LR-T1b, submucosal EAC with <500-µm invasion,
well/moderately differentiated, and no lymphovascular invasion.
1 Missing: HR-T1a, n = 2 (5%); HR-T1b, n = 2 (6%).
2 Missing: HR-T1a, n = 1 (2%); HR-T1b, n = 2 (6%).
3 Missing: HR-T1b, n = 1 (3%).
4 Missing: HR-T1a, n = 1 (2%); HR-T1b, n = 1 (3%).
5 Missing: HR-T1a, n = 7 (16%); LR-T1b, n = 11 (41%); HR-T1b, n = 16 (43%).
Barrett’s length, median (IQR), cm
1 (0–3)
3 (0–5)
1 (0–5)
4 (2–7)
5 (1–8)
4 (1–6)
Tumor location3
0
0
0
4 (9%)
5 (19%)
5 (14%)
30 (70%)
18 (67%)
20 (56%)
9 (21%)
4 (15%)
10 (28%)
Endoscopic resection techniques
30 (70%)
18 (67%)
14 (39%)
#Multiband mucosectomy
26
15
14
#Endoscopic cap resection
2
1
0
#EMR technique unreported
2
2
0
13 (30%)
9 (33%)
22 (61%)
Tumor infiltration depth
4 (9%)
–
–
33 (77%)
–
–
6 (14%)
–
–
–
27 (100%)
11 (31%)
–
0
25 (69%)
–
0
0
Tumor differentiation grade4
2 (5%)
5 (19%)
5 (14%)
6 (15%)
22 (81%)
15 (42%)
31 (72%)
0
15 (42%)
3 (7%)
0
0
Lymphovascular tumor invasion
17 (40%)
–
9 (25%)
26 (61%)
27 (100%)
27 (75%)
Tumor diameter, mean (SD), mm5
15 (± 8)
27 (± 23)
20 (± 11)
Fig. 2 Flow of patient selection. EAC, esophageal adenocarcinoma; ER, endoscopic resection;
FU, follow-up; G, tumor differentiation grade; HR-T1a, intramucosal EAC with poor/no
differentiation and/or lymphovascular invasion; HR-T1b, submucosal EAC with ≥500-µm
invasion, poor/no differentiation and/or lymphovascular invasion; LR-T1b, submucosal
EAC with <500-µm invasion, well/moderately differentiated and no lymphovascular invasion;
LVI, lymphovascular invasion; R1v, tumor-positive vertical ER margin.
Of the 106 patients, 26 (25%) underwent additional surgical resection following ER,
while 80 (75%) entered endoscopic surveillance as they were deemed unfit for surgery
(n = 31), or on the basis of the patient’s preference (n = 13) or local guidelines
for low risk T1b (n = 20), with the reason unknown for 16 patients. Patients in the
endoscopic surveillance group were older (mean [SD] age 71 [9] years) than those who
underwent additional surgery (64 [11] years; P < 0.001). Endoscopic surveillance patients were also more frequently diagnosed with
HR-T1a and LR-T1b (P < 0.001), with no significant difference in their American Society of Anesthesiologists
(ASA) classification (P = 0.82).
Surgical treatment after endoscopic resection
There were 26 patients (HR-T1a, n = 9; LR-T1b, n = 1; HR-T1b, n = 16) who underwent
esophagectomy at a median (IQR) of 2 (1–3) months after ER. The esophagectomies were
of the following types: minimally invasive thoracolaparoscopic, n = 14 (54%); open
transthoracic, n = 6 (23%); open transhiatal, n = 2 (8%); and minimally invasive transhiatal,
n = 1 (4%), with three unknown. Surgical morbidity was 65% (95%CI 49%–83%; n = 17),
with infection being most common (n = 13; 50%), followed by anastomotic leakage (n
= 3; 12%). The 30-day mortality was 0%.
In the esophagectomy specimens, invasive (residual) intraluminal cancer was found
in five patients (19%; 95%CI 7%–39%): mucosal, n = 4; submucosal, n = 1. Nodal disease
was found in two patients (8%; 95%CI 1%–25%), each with one positive lymph node, with
a median (IQR) of 22 (17–29) nodes resected. Post-surgical staging showed T0N0M0 (n
= 19), T1N0M0 (n = 5), and T0N1M0 (n = 2). Follow-up after surgery was a median (IQR)
of 47 (32–79) months.
Endoscopic surveillance after endoscopic resection
There were 80 patients (HR-T1a, n = 34; LR-T1b, n = 26; HR-T1b, n = 20) who entered
endoscopic surveillance, with a median (IQR) of 7 (4–11) endoscopies performed over
a median (IQR) of 41 (20–59) months. EUS, CT, and PET-CT were infrequently performed
(median 0; IQR 0–1). The median (IQR) total follow-up was 46 (25–59) months.
Metastatic disease
During follow-up
There were 7/106 patients (7%, 95%CI 3%–13%) who developed LNM and/or distant metastasis
during follow-up, diagnosed after a median (IQR) of 29 (12–38) months after the initial
ER. The main patient outcomes are shown in [Fig. 3 ]. In the surgical group, two patients showed LNM at 9 and 28 months post-esophagectomy,
with one also having distant metastasis. In the endoscopic surveillance group, four
patients developed LNM, with one simultaneously diagnosed with distant metastasis;
the remaining patient developed distant metastasis after 38 months. The overall rate
of LNM was 6% (95%CI 2%–12%), and of LNM and/or distant metastasis 7% (95%CI 3%–13%)
over a median (IQR) of 47 (27–63) months of follow-up. The characteristics of these
patients and the subsequent treatment of their metastatic disease are displayed in
Table 2s .
Fig. 3 Flow diagram depicting the main follow-up outcomes. DM, distant metastasis; EAC, esophageal
adenocarcinoma; FU, follow-up; LNM, lymph node metastasis. * Includes two cases diagnosed
in the esophagectomy specimen.
Overall risk
Considering metastatic events following immediate surgery as well as during follow-up,
metastatic rates were: for HR-T1a, 9% (4/43; 95%CI 3%–22%), with an annual risk during
follow-up of 2.2% (95%CI 0.6%–5.6%); for LR-T1b, 4% (1/27; 95%CI 0.1%–19%), with an
annual risk of 0.9% (95%CI 0.02%–4.9%); for HR-T1b, 11% (4/36; 95%CI 3%–26%), with
an annual risk of 3.9% (95%CI 1.0%–9.6%) ([Table 2 ]).
Table 2 Outcomes of patients during follow-up categorized by risk group.
Patients
All (n = 106)
HR-T1a (n = 43)
LR-T1b (n = 27)
HR-T1b (n = 36)
HR-T1a, intramucosal EAC with poor/no differentiation and/or lymphovascular invasion;
HR-T1b, submucosal EAC with ≥500-µm invasion, poor/no differentiation, and/or lymphovascular
invasion; IQR, interquartile range; NA, not applicable; LR-T1b, submucosal EAC with
<500-µm invasion, well/moderately differentiated, and no lymphovascular invasion.
1 After initial endoscopic resection.
2 One case diagnosed in the esophagectomy specimen in the HR-T1a group and in the HR-T1b
group.
Duration of follow-up, median (IQR), months1
47 (27–63)
52 (38–65)
50 (29–51)
36 (23–51)
Diagnosis of metastatic disease, n (%) [95%CI]2
9 (8) [4–16]
4 (9) [3–22]
1 (4) [0.1–19]
4 (11) [3–26]
Annual risk of metastasis during follow-up (95%CI), %
2.2 (1.0–4.2)
2.2 (0.6–5.6)
0.9 (0.02–4.9)
3.9 (1.0–9.6)
Time to metastasis, median (IQR), months2
29 (12–38)
12 (NA)
17 (NA)
33 (NA)
Disease-specific death during follow-up, n (%) [95%CI]
5 (5) [2–11]
3 (7) [1–19]
1 (4) [0.1–19]
1 (3) [0.1–15]
Local intraluminal recurrence
Of the 80 patients in the endoscopic surveillance group, two (3%; 95%CI 0.3%–9%) required
esophagectomy owing to intraluminal recurrence during follow-up exceeding re-ER limits,
with post-surgical diagnoses of T1N0M0 and T3N0M0.
Mortality
During follow-up, 5/106 patients died from EAC-related causes (one in the surgery
group [4%; 95%CI 0.1%–20%] and four in the endoscopic surveillance group [5%; 95%CI
1%–12%]), resulting in a disease-specific mortality rate of 5% (95%CI 2%–11%).
Of the 106 patients, 16 died of non-EAC-related causes (one in the surgery group [4%;
95%CI 0.1%–20%] and 15 in the endoscopic surveillance group [19%; 95%CI 11%–29%]).
Other-cause mortality was 15% (95%CI 9%–23%), with an overall mortality rate of 20%
(95%CI 13%–29%).
Kaplan–Meier analysis ([Fig. 4 ]) suggested a potential trend toward better overall survival in the surgery group
following ER (log-rank test, P = 0.07), considering all causes of death. Disease-specific mortality showed no significant
difference between the two treatment subcohorts (P = 0.80).
Fig. 4 Kaplan–Meier survival curves for : a overall survival, showing no significant difference between the two subgroups (log-rank
test; P = 0.07); b disease-specific deaths, showing no significant difference between the two subgroups
(log-rank test, P = 0.80). ER, initial endoscopic resection procedure.
Discussion
We conducted an international multicenter retrospective cohort study of 106 patients
who underwent radical ER for high risk T1 EAC. In this cohort, EMR was performed more
frequently (58%) than ESD (42%), with ESD becoming more common later in the study
timeframe (2008–2019). The study encompassed outcomes from 26 patients who underwent
additional surgery and 80 patients who entered endoscopic surveillance. Our cohort
presents a considerable number of patients with relatively long follow-up periods.
Within our study cohort, 29 patients had follow-up exceeding 5 years, and 10 more
than 8 years, contributing to a cumulative 440 person-years of follow-up.
Our findings suggest low metastatic rates during follow-up. Notably, no significant
difference in overall mortality rates was observed between the surgical and endoscopic
surveillance groups, suggesting that additional surgery as post-ER management does
not offer a survival advantage over conservative endoscopic management; however, the
surgical group in our cohort was relatively small, and our results were not adjusted
for age, pre-existing co-morbidities, or tumor stage.
These findings align with the previous literature, which shows that esophagectomy
is not always a definitive curative approach for high risk T1 EAC, and metastatic
disease can still occur. Westerterp et al. analyzed 120 T1 EAC patients undergoing
esophagectomy (with nodal involvement in 19/120), without chemo-/radiotherapy, revealing
18 cases of recurrent disease over a median 44 months of follow-up, resulting in 10
deaths [18 ]. Molena et al. reported that among 23 T1b EAC patients undergoing esophagectomy
(nodal involvement in 6/23 patients), with a median 37 months of follow-up, one patient
died of systemic recurrence [19 ]. Schölvinck et al. found two recurrences among 25 patients with high risk T1b EAC
who underwent esophagectomy (nodal involvement in 5/25 patients), over a 49-months
follow-up period, both fatal [11 ].
In addition, in our cohort, surgery did not appear to improve disease-specific mortality,
aligning with the findings of Otaki et al. [20 ]. In their large multicenter study involving 141 T1b EAC patients, 68 underwent esophagectomy
and 73 were managed endoscopically, with no correlation found between surgery and
improved disease-free survival. Importantly, both studies lacked standardized follow-up
protocols, limiting their conclusions. While in our cohort, overall survival seemingly
favored the surgical group ([Fig. 4 ]), the older age of the endoscopic surveillance group limits direct comparison, as
does the absence of a standardized follow-up regimen.
The existing literature suggests that submucosal tumor invasion is likely associated
with increased metastatic risks, with HR-T1b tumors carrying a higher risk than LR-T1b
tumors. However, our data, albeit with limited case numbers, reveal low annual metastatic
rates across all three risk groups, ranging from 0.8%–3.1% ([Table 2 ]).
Regarding T1a EAC specifically, nonsurgical management has long been relatively consensual
owing to the assumption of a very low to nonexistent risk of LNM (<1%) [10 ]
[11 ]
[21 ]
[22 ]; however, studies specifically addressing HR-T1a are limited. Nieuwenhuis et al.
reported a surprisingly high annual risk of LNM (6.9%, 95%CI 3%–15%) in their cohort
of 25 HR-T1a patients undergoing ER (R0v) and surveyed for 35 (IQR 22–53) months [13 ]. Benech et al. included nine HR-T1a patients undergoing ER (R0v/R1v) who showed
no metastatic disease during 35 (IQR 24–61) months of follow-up [15 ].
In contrast, our larger HR-T1a subgroup (n = 43) with longer follow-up (median [IQR]
52
[37–65] months) exhibited an annual LNM risk of 2.2% (95%CI 0.6%–5.6%). Surprisingly,
the
metastatic rate during follow-up stood at 9% (95%CI 3%–22%), which exceeded our expectations.
It is possible that the limited number of cases included and the lack of histopathology
review
might account for this, although this data, like the study by Nieuwenhuis et al.,
suggests
that mucosal cancers with high risk features potentially carry a higher risk for metastasis
than previously assumed.
Regarding metastatic risk of LR-T1b EAC, previous endoscopic cohort studies focusing
on the long-term outcomes of this patient group have reported rates ranging between
0 and 2% [11 ]
[21 ]
[23 ]
[24 ]. Our present analysis echoes these findings, demonstrating a similarly low annual
risk of 0.9% (95%CI 0.02%–4.9%) within this patient group. Although the observed metastatic
rate during follow-up might appear relatively high at 4% (95%CI 0.1%–19%), this assessment
is most likely owing to the small sample size (n = 27).
Concurrently, despite their limited cohort sizes and retrospective nature, an increasing
number of endoscopy-focused studies report relatively low metastatic rates for HR-T1b,
ranging from 0 to 16% [10 ]
[11 ]
[21 ]
[23 ]
[24 ]. Our findings regarding this patient subgroup align with these recent studies, displaying
an annual risk of 3.9% (95%CI 1.0%–9.6%), which, although relatively low, exceeds
the annual risks observed in our HR-T1a and LR-T1b subcohorts, as anticipated.
The metastatic rate of 11% (95%CI 3%–26%), based on a small sample size (n = 36),
also falls within the anticipated range reported in the endoscopy-focused studies.
Gotink et al. recently published a cohort study comprising 248 T1b EAC patients who
underwent ER and/or surgery, assessing LNM presence in surgical resection specimens
and during clinical follow-up [12 ]. In their cohort, one-third of patients experienced metastases within 5 years. Their
scoring system, considering submucosal invasion depth, LVI, and tumor size, estimates
a possible high metastatic risk of between 5.9% and 70.1% for T1b EAC. While we do
advocate for a personalized risk model to advance personalized care, there are important
limitations to their study design, such as the retrospective design covering mostly
historical cases (1986–2016), handling of samples (lack of additional slide preparation
in surgical specimens and no additional immunohistochemical staining). Moreover, the
model relies predominantly on surgical data and may not be directly applicable to
patients who undergo ER. Therefore, using these data for therapeutic decision-making
is, in our opinion, not appropriate without external validation of the model in endoscopically
treated patients.
The risk of metastatic disease in high risk T1 EAC has been reported to be as high
as 46% in the literature [7 ]
[8 ]
[9 ]. Our rates, aligned with recent endoscopy-focused studies and involving extended
follow-up durations, fall within the lower end of this spectrum, indicating low annual
recurrence rates during follow-up. The discrepancy between surgical and endoscopy-focused
studies may be attributed to differences in handling and processing surgical specimens
versus ER specimens for pathologic diagnosis. Surgical specimens are cut at wider
intervals, while ER specimens are fully embedded, raising the risk of underdiagnosis
in surgical specimens. Additionally, advances in endoscopic imaging nowadays allow
detection of more subtle high risk T1 lesions, which are then treated with ER. These
subtle high risk T1 lesions may have different malignant potential compared with the
more prominent high risk T1 lesions historically treated with surgery.
Furthermore, within our study, the extensive reassessment of 11 surgically staged
T1 EAC cases initially registered as R0v at prior ER revealed the majority (6/11 cases)
were mislabeled as R0v ER in the local registries. An in-depth reassessment of each
case highlighted diverse reasons to reclassify as R1v or inconclusive, including pathology
reports that were unable to confirm tumor-free vertical margins, endoscopy reports
that indicated a metachronous lesion, and incomplete lifting during the ER, so preventing
radical resection. These findings challenge previous retrospective studies that failed
to thoroughly investigate such cases, thereby potentially missing misclassifications
that might have contributed to higher reported rates of LNM.
This study has several key limitations. First, its retrospective design introduces
potential selection and information biases, compromising the robustness of our findings.
Second, there were no standardized baseline staging or follow-up protocols. Inconsistent
use of EUS for LNM screening, coupled with infrequent imaging to assess for distant
metastasis prior to initiating follow-up, could mean some patients already had baseline
metastatic disease. Similarly, follow-up metastatic disease may have been undetected
because of the low frequencies of follow-up imaging and EUS. Incorporating more rigorous
follow-up visits with increased imaging examinations could potentially have identified
metastatic disease at earlier, curable stages. Third, although the overall study population
was substantial, the smaller subgroup sizes limit comprehensive comparative or predictive
assessments for metastatic risk. Moreover, patients who underwent direct surgery without
prior ER were not included, potentially skewing metastatic risk. Fourth, central pathology
review was performed on only five selected cases (Table 1s ). Finally, the retrospective nature of our study limited our ability to stratify
mortality by pre-existing clinical factors.
The strengths of our study encompass its large multicenter cohort derived from 11
tertiary referral centers (Table 3s ), making it, to our knowledge, center-wise the largest study on T1 EAC metastatic
risk after ER. By including both surgical and endoscopic surveillance patients, it
reflects real-world clinical practices, where nonsurgical candidates often receive
endoscopic surveillance, enhancing the study's applicability. Additionally, the study
uniquely focused on a well-defined cohort of high risk T1 EAC patients who underwent
radical ER, excluding R1v and inconclusive resections after thorough re-examination
of doubtful surgically staged T1 cases. The extensive median follow-up durations of
47 and 46 months strengthen the validity of our findings. While extended follow-up
could potentially alter metastatic rates, this seems improbable given the median (IQR)
post-ER time to diagnosis of metastatic disease was 29 (12–38) months. Prior studies
also indicate metastases appear within 2 years post-ER [12 ]
[25 ].
In summary, our study underscores the feasibility of a conservative organ-preserving
endoscopic surveillance approach after radical ER for high risk T1 EAC as a valid
alternative to surgical resection, in selected patients without baseline signs of
residual cancer or metastatic disease. Our findings emphasize the need for re-evaluation
of existing tumor risk factors to enhance risk stratification. The annual metastasis
risks were low, but not negligible, across all three risk groups, which is consistent
with recent endoscopy-focused studies, which have shown low metastatic incidences
in high risk T1 EAC. Nonetheless, robust prospective data with standardized protocols
and prolonged follow-up (PREFER study; NCT03222635) are requisite to ascertain the
optimal management strategy and refine guidelines for the treatment of individuals
with high risk T1 EAC.
