Articles

Neoadjuvant chemoradiotherapy followed by active

surveillance versus standard surgery for oesophageal cancer
(SANO trial): a multicentre, stepped-wedge,
cluster-randomised, non-inferiority, phase 3 trial
Berend J van der Wilk*, Ben M Eyck*, Bas P L Wijnhoven, Sjoerd M Lagarde, Camiel Rosman, Bo J Noordman, Maria J Valkema, Tanya M Bisseling,
Peter-Paul L O Coene, Marc J van Det, Jan Willem T Dekker, Jolanda M van Dieren, Michail Doukas, Stijn van Esser, W Edward Fiets,
Henk H Hartgrink, Joos Heisterkamp, I Lisanne Holster, Bastiaan Klarenbeek, David van Klaveren, Eva Kouw, Ewout A Kouwenhoven,
Misha D Luyer, Bianca Mostert, Grard A P Nieuwenhuijzen, Liekele E Oostenbrug, Jean-Pierre Pierie, Johanna W van Sandick, Meindert N Sosef,
Manon C W Spaander, Roelf Valkema, Edwin S van der Zaag, Ewout W Steyerberg, J Jan B van Lanschot, SANO Study Group†

Summary
Background A substantial proportion of individuals with oesophageal cancer have a pathological complete response Lancet Oncol 2025
after neoadjuvant chemoradiotherapy and oesophagectomy. We aimed to investigate whether active surveillance Published Online
could be an alternative for individuals with a clinical complete response after neoadjuvant chemoradiotherapy. March 17, 2025
https://doi.org/10.1016/
S1470-2045(25)00027-0
Methods We performed a multicentre, stepped-wedge, cluster-randomised, non-inferiority, phase 3 trial in 12 Dutch
See Online/Comment
hospitals. Individuals with locally advanced oesophageal cancer and a clinical complete response after neoadjuvant https://doi.org/10.1016/
chemoradiotherapy (ie, no tumour detected with endoscopic biopsies, ultrasound, and PET-CT) underwent active S1470-2045(25)00072-5
surveillance or standard surgery (ie, oesophagectomy within 2 weeks after reaching clinical complete response). *Joint first authors
There were no inclusion restrictions regarding comorbidities or performance status, but participants had carcinoma, †Members of the SANO Study
were age 18 years or older, and were treated with curative intent. Randomisation of hospitals was performed using Group are listed in the appendix
computer-generated sequences without stratification methods, after an initial phase of all hospitals performing (pp 26–28)
standard surgery. The primary endpoint was overall survival, analysed according to a modified intention-to-treat Department of Surgery
(B J van der Wilk MD PhD,
principle (allowing crossover at time of clinical complete response) and an intention-to-treat principle. Non-
B M Eyck MD PhD,
inferiority was defined as 2-year survival rate for active surveillance of 15% or less below that for standard surgery. Prof B P L Wijnhoven MD PhD,
The trial was registered within the Netherlands Trial Register, NTR-6803, and the inclusion phase has been S M Lagarde MD PhD,
completed. B J Noordman MD PhD,
M J Valkema MD PhD,
Prof J J B van Lanschot MD PhD),
Findings Between Nov 8, 2017, and Jan 17, 2021, 1115 individuals were screened, of whom 309 were included. Department of Pathology
198 underwent active surveillance and 111 underwent standard surgery. 242 (78%) participants were male and (M Doukas MD PhD),
67 (22%) were female. Median follow-up was 38 months (IQR 32–48). 2-year overall survival for active Department of Public Health
(D van Klaveren PhD),
surveillance (74% [95% CI 69–78]) was non-inferior to standard surgery (71% [62–78]) after modified intention-to-
Department of Medical
treat analysis (one-sided 95% boundary: 7% lower). It remained non-inferior in the intention-to-treat Oncology (B Mostert MD PhD),
analysis (75% [68–80] vs 70% [63–77], one-sided 95% boundary: 6% lower). There were no significant differences in Department of
overall survival according to modified intention-to-treat analysis (hazard ratio 1·14, two-sided 95% CI 0·74–1·78) or Gastroenterology and
Hepatology
intention-to-treat analysis (0·83, 0·53–1·31). The frequency of postoperative complications and postoperative (Prof M C W Spaander MD PhD),
mortality after standard surgery or postponed surgery after active surveillance was similar between groups. and Department of Radiology
and Nuclear Medicine
Interpretation Overall survival after active surveillance for oesophageal cancer was non-inferior compared with (R Valkema MD PhD), Erasmus
MC Cancer Institute, University
standard surgery after 2 years. For the long-term efficacy of active surveillance, extended follow-up is required. The Medical Centre, Rotterdam,
results of the present trial could be used for patient counselling and shared decision making. Netherlands; Department of
Surgery (Prof C Rosman
Funding Dutch Cancer Society (KWF) and Netherlands Organisation for Health Research and Development (ZonMw). MD PhD, B Klarenbeek MD PhD),
Department of
Gastroenterology and
Copyright © 2025 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and Hepatology
similar technologies. (T M Bisseling MD PhD),
Radboud University Medical
Centre, Nijmegen,
Introduction complete response and 30–40% will develop early distant Netherlands; Department of
Neoadjuvant chemoradiotherapy followed by surgery is a metastases,1 raising questions about the benefit of Surgery
standard treatment for locally advanced oesophageal standard surgery in all individuals after neoadjuvant (P-P L O Coene MD PhD),
cancer.1 After trimodality treatment, nearly a quarter of chemoradiotherapy. Consequently, active surveillance Department of
Gastroenterology
patients with adenocarcinoma and nearly half of patients has been proposed as an alternative for standard (I L Holster MD PhD), Maasstad
with squamous cell carcinoma have a pathological surgery.2 Active surveillance involves frequent clinical Hospital, Rotterdam,

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 1

 Articles

Netherlands; Department of
Surgery (M J van Det MD PhD, Research in context
E A Kouwenhoven MD PhD), ZGT
Hospital, Almelo, Netherlands; Evidence before this study together, there is no consensus on whether active surveillance
Department of Surgery A substantial fraction of individuals with oesophageal cancer is a valid alternative treatment for individuals with locally
(J W T Dekker MD PhD, have a pathological complete response after neoadjuvant advanced oesophageal cancer and clinical complete response
S van Esser MD PhD), Reinier de
Graaf Gasthuis, Delft,
chemoradiotherapy and oesophagectomy. Consequently, after neoadjuvant chemoradiotherapy. The accuracy of clinical
Netherlands; Department of organ-sparing active surveillance strategies for individuals with response evaluations was assessed prospectively in the
Gastrointestinal Oncology a clinical complete response (ie, no detected residual tumour preSANO trial. The results of this trial justified proceeding with
(J M van Dieren MD PhD), after neoadjuvant chemoradiotherapy using clinical response a phase 3 trial comparing overall survival after active
Department of Surgical
Oncology
evaluations) have been proposed by several groups worldwide. surveillance or standard surgery in such individuals.
(J W van Sandick MD PhD), The A systematic review and meta-analysis was performed in 2020
Added value of this study
Netherlands Cancer Institute, comparing overall survival after active surveillance versus
Amsterdam, Netherlands; Oesophagectomy is a major surgical procedure associated with
standard surgery following neoadjuvant chemoradiotherapy.
Department of Medical a risk of substantial postoperative complications in the majority
Oncology (W E Fiets MD PhD),
Keywords used were “esophageal cancer”, “surgery”, and “active
of patients, in-hospital mortality, and a decrease in health-
Department of Surgery surveillance”, and relevant variations thereof, and results were
related quality of life. To our knowledge, this study is the first
(Prof J-P Pierie MD PhD), not restricted to studies in English. Databases screened until
Medical Centre Leeuwarden, large, phase 3 trial comparing overall survival after active
Feb 12, 2020, were Embase, MEDLINE, Web of Science, Scopus,
Leeuwarden, Netherlands; surveillance versus standard surgery for individuals with locally
Department of Surgery, Leiden
and Cochrane Central Register of Trials. Research was deemed
advanced oesophageal cancer.
University Medical Centre, eligible when individuals were included with clinical complete
Leiden, Netherlands response after chemoradiotherapy, when both individuals Implications of all the available evidence
(H H Hartgrink MD PhD);
undergoing active surveillance and standard surgery were This study shows that after a minimum follow-up of 2 years,
Department of Surgery,
Elisabeth Tweesteden Hospital, included and when overall survival was compared between overall survival is non-inferior after active surveillance compared
Tilburg, Netherlands two groups. In this meta-analysis, seven studies were included. with standard surgery, and it results in better health-related
(J Heisterkamp MD PhD); All except one retrospective study concluded that overall quality of life. After reaching clinical complete response,
Department of
survival was similar between treatments. The validity of this approximately half of participants benefitted from active
Gastroenterology and
Hepatology (E Kouw MD), meta-analysis was limited by poor methodological quality of surveillance, avoiding oesophagectomy. Active surveillance
Department of Surgery the included studies and heterogeneity between studies. More could be introduced as an alternative approach, using the
(E S van der Zaag MD PhD), Gelre recently, the safety of postponed oesophagectomy was present data, in patient counselling and shared decision making.
Hospital, Apeldoorn,
questioned due to results reported in the NeoRes II trial. Taken
Netherlands; Department of
Surgery, Catharina Hospital,
Eindhoven, Netherlands
(Prof M D Luyer MD PhD, response evaluations to assess tumour response after Furthermore, undetected locoregional regrowth during
G A P Nieuwenhuijzen MD PhD); neoadjuvant chemoradiotherapy, with surgery active surveillance could result in development of
Department of
Gastroenterology and
performed only in individuals with locoregional distant metastases or non-radical resections, potentially
Hepatology regrowth in the absence of distant metastases. In the jeopardising overall survival.5 Lastly, frequent clinical
(L E Oostenbrug MD PhD), preSANO trial, the accuracy of clinical response response evaluations could cause a physical and mental
Department of Surgery, evaluations after neoadjuvant chemoradiotherapy for burden for individuals during the active surveillance
Zuyderland Medical Centre,
Heerlen, Netherlands
individuals with locally advanced adenocarcinoma or period. A meta-analysis of several retrospective cohort
(M N Sosef MD PhD); squamous cell carcinoma of the oesophagus was studies and one small randomised controlled trial
Department of Biomedical assessed prospectively.3 90% of substantial residual suggested that overall survival after active surveillance is
Data Sciences, Leiden tumours could be detected using a combination of similar to that after oesophagectomy.2 However, these
University Medical Centre,
Leiden, Netherlands
endosonography with fine-needle aspiration of suspected studies were small and probably affected by selection
(Prof E W Steyerberg PhD) lymph nodes, endoscopy with bite-on-bite biopsies, and bias. We aimed to compare overall survival
Correspondence to: PET-CT. Active surveillance could avoid unbeneficial for individuals with locally advanced oesophageal
Dr Berend J van der Wilk, surgery in individuals with a persisting clinical complete cancer and clinical complete response after
Department of Surgery, Erasmus response (ie, no detected residual tumour during clinical neoadjuvant chemoradiotherapy undergoing active
MC Cancer Institute, University
Medical Centre,
response evaluation) after neoadjuvant chemo­ surveillance versus standard surgery, in a multicentre,
Rotterdam 3015GD, Netherlands radiotherapy. Individuals who develop distant metastases non-inferiority trial.
b.vanderwilk@erasmusmc.nl during active surveillance that were undetected at initial
See Online for appendix staging might also benefit from active surveillance. The Methods
benefits of avoiding oesophagectomy are clear, given Study design and participants
that approximately 60% of patients develop postoperative We performed a multicentre, stepped-wedge, cluster-
complications and many experience decreased health- randomised, non-inferiority, phase 3 trial comparing
related quality of life.4 overall survival for individuals with locally advanced
Possible harms of active surveillance include the oesophageal cancer and clinical complete response after
potentially increased risk for postoperative neoadjuvant chemoradiotherapy undergoing active
complications after postponing oesophagectomy. surveillance versus standard surgery in 12 Dutch

2 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0

 Articles

hospitals all performing high-volume oesophageal not had any influence on the analysis of the data or on
surgery. The trial protocol of the Surgery As Needed for the writing of the report.
Oesophageal Cancer (SANO) trial has been published
previously.6 During the course of the study, amendments Randomisation and masking
were made involving the study design and analysis plan, For surgical interventions, it has been described
which were described in an update (May 17, 2021).7 The extensively that traditionally randomised controlled
study is reported according to the CONSORT guidelines, trials are associated with numerous challenges that can
including extensions for non-inferiority trials and lead to the trial failing (eg, due to low accrual).11 In the
stepped-wedge, cluster-randomised trials and complies specific case of randomising between surgical versus
with the Declaration of Helsinki and guidelines for good non-surgical treatment, randomisation at the patient
clinical practice.8,9 The study protocol was approved on level was found not to be feasible, and subsequently,
Nov 1, 2017, by the medical ethics committee of Erasmus trials have been terminated prematurely
MC (MEC-2017-392), and patients provided written, (eg, NCT02551458).12 In interview data, reasons for not
informed consent. The study was registered in wanting to be randomly assigned included not truly
the Netherlands Trial Register, NTR-6803, and the understanding the concept of randomisation and a
inclusion phase has been completed. computer making an impactful decision on their
Participants aged at least 18 years old with locally treatment.12 To avoid premature closure, but to retain
advanced carcinoma of the oesophagus or interpretation of a causal effect in our trial, a stepped-
oesophagogastric junction who were scheduled for wedge cluster randomisation method was chosen. In
neoadjuvant chemoradiotherapy according to CROSS this way, the timing of switching from the control group
regimen10 followed by surgery were eligible. There were to the intervention group was randomised, using
no inclusion restrictions regarding comorbidities or computer-generated sequences without stratifi­ cation
performance status, but participants had carcinoma, methods. Randomisation was performed by an
were aged 18 years or older, and were treated with independent third party (Clinical Trial Centre, Erasmus
curative intent. Neoadjuvant chemoradiotherapy MC Rotterdam, Netherlands). No masking was
consisted of five weekly cycles of intravenous paclitaxel performed. Consequently, participants presenting in the
(50 mg/m²) and carboplatin (area under the curve hospital knew at time of consenting which treatment
2 mg/mL per min). Concurrently, a total dose of they would undergo. We deliberately chose the stepped-
41·4 Gy external beam radiotherapy was given in wedge cluster randomisation design instead of, for
23 fractions of 1·8 Gy for 5 days per week, starting on the example, parallel cluster randomisation or a cluster-
day of the first course of chemotherapy. The exact randomised crossover design. In this way, all
radiotherapy protocol of CROSS was described in the participating hospitals were able to gain experience in
original publication.10 Because CROSS has been standard performing clinical response evaluations and active
treatment in the Netherlands since 2012, no specific surveillance treatment, and fewer logistical problems
radiotherapy quality assurance programme was were expected compared with a parallel cluster
implemented in this trial. Specific exclusion criteria randomisation design. At the start of the trial, all
comprised non-fluorodeoxyglucose-avid primary participating hospitals performed standard surgery.
tumours at baseline PET-CT scan (determined by nuclear Clusters of two participating hospitals switched to active
physician) or initial treatment with endoscopic resection. surveillance approximately every 3 months, with
The accuracy of clinical response evaluations was adjustment based on actual inclusion rates to avoid
considered sufficient based on the results of the unequal study groups due to inclusion varations over
preSANO trial, after the introduction of bite-on-bite time. Actual inclusion periods and details are shown in
biopsies.3 Therefore, study procedures and inclusion the appendix (p 2). As reported in our protocol, the first
criteria of participants undergoing standard surgery in cluster of hospitals that was switched to active
the present study are based on, and identical to, the surveillance was predefined to consist of the initiating
study procedures of the preSANO trial. Given the hospital and either one of two hospitals (determined by
stepped-wedge trial design, in which all hospitals randomisation) with extensive experience in performing
performed standard surgery at the initiation of the trial clinical response evaluations, due to participation in
(described later), and to avoid unnecessary potential risk preSANO. Eventually, all clusters were randomly
from exposure to an experimental treatment, participants assigned to active surveillance until completion of the
from the preSANO-trial were also included in the target sample size.
standard surgery group of the present SANO trial, as
predefined in our study protocols and statistical analysis Procedures
plan (NCT05953181).6 The Dutch patient organisation The first clinical response evaluation was planned
for patients with oesophageal or gastric cancer (SPKS) 6 weeks after neoadjuvant chemoradiotherapy,
had an advisory role in this trial, especially for the design consisting of endoscopy with at least four bite-on-bite
of the trial and the choice for the outcomes. They have biopsies. If no vital tumour cells were detected, as

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 3

 Articles

determined with haematoxylin and eosin staining and of clinical interest: proportion of non-curable regrowths,
deeper sectioning with pankeratin staining, participants proportion of participants with R1 or R2 resections,
were planned for a second clinical response evaluation proportion of postoperative mortality or hospital stay
12 weeks after neoadjuvant chemoradiotherapy. The longer than 60 days, and proportion of participants with
second clinical response evaluation consisted of distant metastases. Stopping rules were tested after
PET-CT scan, followed by endoscopic ultrasound with every ten events. Information on gender was obtained
fine-needle aspiration of suspected lymph nodes (as via electronic medical records.
determined by endoscopist and nuclear physician) and
by endoscopy with at least four bite-on-bite biopsies. Outcomes
Biopsies were taken from the original primary tumour The primary endpoint was overall survival, defined as
site and from any suspicious lesion. the interval between reaching clinical complete
Absence of distant metastases and of locoregional response (ie, date of biopsies taken at second clinical
vital tumour cells during both clinical response response evaluation, 12 weeks after neoadjuvant
evaluations defined a clinical complete response. Only chemo­ radiotherapy) and date of all-cause mortality,
biopsies with uncertain outcomes or with high-grade censored at last follow-up. The primary outcome was
dysplasia were revised centrally, or by a second non-inferiority at 2 years, but the results for the entire
pathologist in the participating centre. In case of follow-up period were also provided, according to the
discordant results, the specimen was reviewed by a statistical analysis plan. Participants had a minimal
third independent expert gastrointestinal pathologist follow-up of 2 years from reaching clinical complete
and a consensus diagnosis was to be reached if at least response (initially a minimal follow-up of 3 years from
two pathologists agreed. Participants with major primary diagnosis, as explained and elaborated on in
protocol deviations, as defined in the statistical analysis our published update of the SANO protocol).7 Secondary
plan, were not considered to have a clinical complete outcomes were disease-free survival (defined as interval
response. When locoregional vital tumour cells were between date of clinical complete response and
detected during clinical response evaluation, diagnosis of irresectable recurrent disease; amended
participants without distant metastases on PET-CT from progression-free survival with approval of study
underwent oesophagectomy within 2 weeks. board on Sept 13, 2022), recurrent locoregional disease
Oesophagectomy was also performed in case of high- after oesophagectomy, distant metastases, and all-cause
grade dysplasia or endoscopic no-pass. mortality, censored at last follow-up. Other secondary
Depending on their cluster and time of recruitment, outcomes were operative and pathological outcomes
participants were assigned to active surveillance or (ie, proportion of participants in the active surveillance
standard surgery. Participants undergoing active group who did not undergo surgery, rate of clinical
surveillance underwent subsequent clinical response irresectability, rate of microscopically irradical
evaluations equal to the second evaluation, occurring resections (R0 resections), postoperative complications,
every 3 months in year 1, every 4 months in year 2, and postoperative 90-day mortality), health-related
every 6 months in year 3, and annually in years 4–5. quality of life, and cost-effectiveness of active
Participants underwent oesophagectomy only in case of surveillance. To sufficiently report all relevant data and
pathologically confirmed or highly suspected nuances concerning health-related quality of life, we
locoregional regrowth without signs of distant describe only global health-related quality of life in the
metastases, or in case the endoscope could not be present study. Global health-related quality of life was
traversed during clinical response evaluations. Surgical chosen, and confirmed by the patient organisation for
approach including access, technique, and nodal patients with oesophageal or gastric cancer (SPKS), as
dissection was dependent on tumour location and the most important objective measure to be reported in
surgeons’ preference. Participants undergoing standard the present manuscript. Other health-related quality-of-
surgery underwent postoperative follow-up with life domains and the cost-effectiveness analysis will be
PET-CT at 16 months and 30 months after completion reported separately.
of neoadjuvant chemoradiotherapy (corresponding to
clinical response evaluations 6 and 9 for participants Statistical analysis
undergoing active surveillance) to compare distant All analyses were performed according to the predefined
dissemination rate. Individuals who refused to statistical analysis plan. The last update was performed
participate in the study underwent standard treatment on June 20, 2023. Prespecified stopping rules were
according to current clinical guidelines in checked using the χ² test or Fisher’s exact test.
the Netherlands: neoadjuvant chemoradiotherapy The anticipated 2-year survival from the date of
followed by surgery after 6–8 weeks. reaching clinical complete response for individuals
Prespecified stopping rules were tested during undergoing standard surgery was 75%, based on a
interim analyses by an independent data safety previously published meta-analysis.2 Therefore,
monitoring board. Stopping rules were based on events 224 participants with clinical complete response had to

4 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0

 Articles

be included to determine non-inferiority of overall clinical tumour category, clinical nodal category, WHO
survival for active surveillance compared with standard performance score, calendar time, and cluster effect.
surgery with a power of 80%, one-sided (as is usual for The assumption for proportional hazards was tested
non-inferiority studies) significance level of 5%, and using Schoenfeld residuals. In case this assumption
time-independent intracentre correlation coefficient could not be met, delta-residual mean survival time was
of 0·02 (ie, no cluster autocorrelation). It was assumed compared in addition to Cox models analysis. Kaplan–
that 34% of participants would have clinical complete Meier estimates were not adjusted for confounders and
response, 20% would cross over from the assigned clustering and were not used for expressing numerical
treatment, and 29 participants could be included from estimates. Distant dissemination rates were compared
the preSANO trial. At baseline, these calculations using the χ² test. R0 resection rates were compared
translated to a required recruitment of 740 participants. using multivariable binary logistic regression adjusted
Non-inferiority of active surveillance was defined as for age, sex, histological tumour subtype, histological
overall survival of 15% or less below the anticipated grading, clinical tumour category, clinical nodal
75% 2-year survival after standard surgery, which was category, and WHO performance score.
based on postoperative mortality and hypothesised For non-inferiority studies, intention-to-treat analysis
improved health-related quality of life for active increases the risk of falsely claiming non-inferiority if
surveillance. This margin was further supported by a non-adherence to one of the study arms occurs.14 As
discrete choice experiment from our group on described in the study protocol, a crossover rate of
participants’ preferences.13 approximately 20% was expected. To avoid the risk of
Active surveillance was defined to be non-inferior if falsely claiming non-inferiority, participants were
the one-sided 95% confidence boundary of the analysed following a modified intention-to-treat
difference in 2-year overall survival estimates between principle, accepting crossover at time of reaching
active surveillance and standard surgery was below the clinical complete response (eg, a participant was
non-inferiority margin of 15%. Cox mixed-effects included in the surgery group, but refused surgery).
regression, based on data censored at 2 years, was used This was prespecified in our publicly available statistical
to adjust the 2-year survival estimates for age, sex, analysis plan. Predefined sensitivity analyses were also
histological tumour subtype, histological grading, performed according to intention-to-treat and per-
clinical tumour category, clinical nodal category, WHO protocol analyses. Moreover, modified intention-to-treat
performance score, calendar time, and cluster effect. analyses were performed after stratification according
Calendar time was defined as the number of months to histological subtype (prespecified analysis) and
between the first patient reaching a clinical complete modified intention-to-treat and intention-to-treat
response at second clinical response evaluation and the analyses were performed after exclusion of the
patient of interest reaching a clinical complete preSANO participants (post-hoc analysis).
response. In this way, we adjusted for calendar time Global health-related quality of life was assessed at
during both the preSANO-trial and the present SANO- baseline and at 3, 6, 9, 12, 16, 20, and 24 months after
trial. The (adjusted) difference in 2-year survival completion of neoadjuvant chemoradiotherapy using
estimates between active surveillance and standard the European Organisation for Research and Treatment
surgery was derived from the Cox mixed-effects model, of Cancer (EORTC) quality of life questionnaire
using the Breslow estimate as the baseline hazard (QLQ-C30). EORTC scores were computed according to
(ie, the difference between the mean predicted 2-year the EORTC manual. Repeated measurement analysis
survival if all participants underwent active surveillance with generalised estimated equations, adjusted for age,
and the mean predicted 2-year survival if all participants sex, histologic subtype of the tumour, histological
underwent standard surgery). To derive confidence grading, clinical T category, clinical N category, WHO
boundaries for the adjusted difference in 2-year survival performance, and hospital of inclusion, was used to
estimates, regression coefficients were resampled evaluate within-group and between-group differences.
100 000 times from a multivariate normal distribution Individual participants were treated as the clustering
with the original regression coefficients as mean and variable. To assess trends over time, the follow-up
their covariance as variance. The 95% quantile of the period was fitted with predefined timepoints in the
differences in 2-year survival estimates, based on analysis. Cohen’s d effect sizes were calculated to assess
resampled regression coefficients, was used as the clinical relevance of differences.
95% boundary. For all analyses, p<0·05 was considered to indicate
Median follow-up was determined using the reverse statistical significance. For global health-related quality
Kaplan–Meier method. Overall survival and disease- of life, Cohen’s d greater than 0·50 was considered
free survival for the entire follow-up were visualised clinical relevant. All hazard ratios (HRs) and odds
with the Kaplan–Meier method and compared using ratios (ORs) are reported with the standard surgery
mixed-effects Cox regression models adjusted for age, group as reference. Statistical analyses were performed
sex, histological tumour subtype, histological grading, using R version 4.3.0.

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 5

 Articles

1115 screened

306 not included

270 eligible but declined participation
36 eligible but participation not offered

809 provided informed consent

33 excluded
30 did not undergo chemoradiotherapy
3 no FDG-avid primary tumour

776 enrolled

15 excluded
6 withdrew consent
5 died
4 metastases before clinical response evaluation

761 underwent first clinical response evaluation

291 excluded
232 locoregional regrowth
40 no pass during OGD
6 requested immediate surgery
3 protocol violations
2 died
2 withdrew consent
2 metastases
4 could not undergo clinical response evaluation

470 underwent second clinical response evaluation

196 excluded
115 locoregional regrowth
40 metastases
19 no pass during OGD
17 suspected disease without histological evidence
5 could not undergo clinical response evaluation

274 had clinical complete response

35 from preSANO trial with clinical complete

response included in standard surgery group

153 assigned to standard surgery (intention-to-treat 156 assigned to active surveillance (intention-to-treat
population) population)

42 crossed over (including 7 from preSANO trial)

111 underwent standard surgery (modified intention- 198 underwent active surveillance (modified
to-treat population) intention-to-treat population)

Figure 1: Trial profile

Reasons for declining participation could not be recorded during the trial because of privacy regulation rules. FDG=fluorodeoxyglucose.
OGD=oesophagogastroduodenoscopy.

6 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0

 Articles

Role of the funding source

Active Standard
The funders of the study had no role in study design, surveillance surgery
data collection, data analysis, data interpretation, or (n=198) (n=111)
writing of the report. Age, years 69 (63–74) 68 (61–73)
Gender
Results Female 42 (21%) 25 (23%)
From Nov 8, 2017, to Jan 17, 2021, 1115 individuals were Male 156 (79%) 86 (77%)
screened, 809 individuals provided informed consent, Histology
and 776 participants were included in the study. The Adenocarcinoma 147 (74%) 84 (76%)
reasons for declining participation could not be recorded
Squamous cell carcinoma 47 (24%) 23 (21%)
during the trial because of privacy regulation rules. After
Other 4 (2%) 4 (4%)
the first and second clinical response evaluation,
Tumour differentiation grade*
274 (35%) participants had a clinical complete response
G1 33 (17%) 19 (17%)
and were assigned to active surveillance (n=156) or
G2 97 (49%) 51 (46%)
standard surgery (n=118). After inclusion of 35 patients
G3 65 (33%) 37 (33%)
from the preSANO trial in the standard surgery group,
Gx 3 (2%) 4 (4%)
seven patients crossed over from standard surgery and
Tumour stage*
were included in the active surveillance group and
IB 12 (6%) 9 (8%)
28 patients were included in the standard surgery group
IIA 14 (7%) 6 (5%)
(instead of 29 as stated in our protocol and in the
IIB 61 (31%) 40 (36%)
Methods, due to one extra patient with protocol violation;
appendix p 3). In addition, 35 patients from in the IIIA 57 (29%) 27 (24%)

standard surgery group in SANO crossed over to the IIIB 32 (16%) 13 (12%)

active surveillance group. This resulted in a modified IIIC 5 (3%) 4 (4%)

intention-to-treat population of 198 participants Could not be determined 17 (9%) 12 (11%)
undergoing active surveillance and 111 participants Clinical T category*
undergoing standard surgery (figure 1). Baseline cT1 2 (1%) 1 (1%)
characteristics were well balanced between groups cT2 43 (22%) 24 (22%)
among participants with clinical complete response cT3 140 (71%) 76 (68%)
(table 1), as well as between groups among all participants, cT4 3 (2%) 3 (3%)
and between participants who crossed over to active cTx 10 (5%) 7 (6%)
surveillance compared with those undergoing standard Clinical N category*
surgery (appendix pp 15–18). 242 (78%) participants were cN0 83 (42%) 49 (44%)
male and 67 (22%) were female. No data were collected cN1 72 (36%) 37 (33%)
on race or ethnicity. Information on inclusion rates cN2 31 (16%) 16 (14%)
per participating hospital and corresponding principal cN3 4 (2%) 1 (1%)
investigators is reported in the appendix (p 25). cNx 8 (4%) 8 (7%)
Subsequent anticancer treatment consisted of nivolumab WHO performance status†
for four participants in the active surveillance group, 0 131 (66%) 68 (61%)
according to newly introduced national guidelines during 1 54 (27%) 32 (29%)
the study. 2 2 (1%) 0
According to modified intention-to-treat analysis, 3 1 (1%) 0
overall median follow-up from the time of clinical Missing 10 (5%) 11 (10%)
complete response was 38 months (IQR 32 to 48); it was Tumour location‡
34 months (30 to 40) for participants undergoing active Proximal third oesophagus 3 (2%) 4 (4%)
surveillance and 50 months (40 to 60) for participants Middle third oesophagus 24 (12%) 16 (14%)
undergoing standard surgery. Median overall survival Distal third oesophagus 120 (61%) 64 (58%)
was 43 months (95% CI 39 to not reached [NR]) for Oesophagogastric junction 51 (26%) 27 (24%)
participants undergoing active surveillance versus
Data are median (IQR) or n (%). Data on race or ethnicity were not collected.
53 months (40 to NR) for participants undergoing *cTNM staging is reported according to the 7th edition of the Union for
standard surgery. Overall survival at 2 years was 74% International Cancer Control staging manual; T staging was done with endoscopic
(95% CI 69–78) for participants undergoing active ultrasonography or CT, and clinical nodal staging was based on endoscopic
ultrasonography. †Assessed on a five-point scale, with a higher number indicating
surveillance and 71% (62–78) for participants undergoing
greater disability. ‡Determined by endoscopy.
standard surgery, which was 3% higher (one-sided
95% boundary: 7% lower) for active surveillance, and was Table 1: Baseline characteristics of patients included in the modified
below the predefined non-inferiority margin of 15% intention-to-treat analysis
worse overall survival. Overall survival was also

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 7

 Articles

non-inferior according to intention-to-treat analysis (75% chemoradiotherapy). The other 11 were detected at one of
[95% CI 68–80] vs 70% [63–77], one-sided 95% boundary: the fifth (12 months after completion of neoadjuvant
6% lower). Taking into account the entire follow-up, chemoradiotherapy) to ninth clinical response
50 participants died in the standard surgery group and evaluations (30 months after completion of neoadjuvant
79 participants died in the active surveillance group in chemoradiotherapy). 69 (35%) of 198 participants had
the modified intention-to-treat analysis (HR 1·14, persistent clinical complete response (including
95% CI 0·74 to 1·78, p=0·55; figure 2A). In the intention- 44 participants with an adenocarcinoma) and
to-treat analysis, 59 participants died in the standard 96 (48%) of 198 participants developed isolated
surgery group and 56 died in the active surveillance locoregional regrowth (appendix pp 11, 20). Of these
group (0·83, 0·53 to 1·31, p=0·42; figure 2B). Inspection 96 participants, 82 (85%) with isolated locoregional
of Schoenfeld residuals suggested that proportional regrowth were detected at one of the the third to fifth
hazards could not be assumed for the modified intention- clinical response evaluations. After eight clinical
to-treat analysis. Therefore, we compared additional response evaluations (ie, 24 months after completion of
delta-residual mean survival times, which showed no
differences in overall survival (estimate 0·95, Active Standard
95% CI –1·55 to 3·46, p=0·46). Other sensitivity analyses surveillance surgery
(according to per-protocol analysis, according to histology, (n=83) (n=101)
and in the post-hoc analysis after exclusion of preSANO Any complication 68 (82%) 85 (84%)
participants) and subgroup analyses showed no Anastomotic leakage 18 (22%) 27 (27%)
statistically significant differences in survival between Severity of anastomotic leakage
active surveillance and standard surgery and supported Subclinical, spontaneous recovery 2 (2%) 3 (3%)
non-inferiority at 2 years for active surveillance compared Subclinical, requiring surgery 1 (1%) 0
with standard surgery (appendix pp 4–10, 19). Clinical, spontaneous recovery 10 (12%) 15 (15%)
During active surveillance, 33 (17%) of 198 participants Clinical, requiring surgery 5 (6%) 9 (9%)
developed distant metastases without undergoing Pulmonary complications
surgery. In 22 of these participants, the metastases were Any 39 (47%) 64 (63%)
detected at the third clinical response evaluation Pneumonia 20 (24%) 29 (29%)
(ie, 6 months after completion of neoadjuvant Respiratory failure requiring 2 (2%) 5 (5%)
reintubation
Cardiac complications
A
100 Any 28 (34%) 44 (44%)
I
Dysrhythmia requiring intervention 11 (13%) 20 (20%)
Overall survival (%)

II II
75 II I III
I I II IIIIIIIIIIIII
I III III
I III II
III I IIIIII I I IIIII I II
Vocal cord outcome
I II III I
IIIII I IIIII I II
50 Normal vocal cord 71 (86%) 94 (93%)
Vocal cord dysfunction, unilateral 3 (4%) 3 (3%)
25 Active surveillance
Standard surgery HR 1·14 (95% CI 0·74–1·78); p=0·55 Vocal cord dysfunction, bilateral 2 (2%) 1 (1%)
0 Unknown vocal cord dysfunction 7 (8%) 3 (3%)
0 6 12 18 24 30 36
Number at risk Thromboembolic complications
(number censored)
Pulmonary embolus 0 2 (2%)
Active surveillance 198 (0) 191 (0) 178 (0) 164 (1) 144 (6) 95 (38) 45 (79)
Standard surgery 111 (0) 105 (0) 97 (0) 85 (0) 78 (0) 68 (7) 60 (10) Adverse events from clinical response evaluations
PET-CT 0 0
B Endosonography with fine-needle 1 (1%) 0
100 aspiration
I
Overall survival (%)

75
II II
II I III
I II IIIIIIIIIIIII
Endoscopy with biopsies 0 0
I
I III III I III
II III I I
I IIII I IIIIIIIIIII I III I III
II I III I IIII I III Chylothorax, requiring TPN 3 (4%) 10 (10%)
50
Chylothorax, requiring surgery 0 1 (1%)
25 Multi-organ failure 1 (1%) 1 (1%)
HR 0·83 (95% CI 0·53–1·31); p=0·42 Length of ICU stay, days 2 (1–2) 2 (1–3)
0
0 6 12 18 24 30 36 Length of hospital stay, days 10 (8–17) 11 (8–17)
Time since clinically complete response (months) 30-day mortality 1 (1%) 3 (3%)
Number at risk
(number censored) 90-day mortality 3 (4%) 5 (5%)
Active surveillance 156 (0) 150 (0) 139 (0) 130 (1) 114 (6) 68 (38) 28 (72)
Standard surgery 153 (0) 146 (0) 136 (0) 119 (0) 108 (0) 95 (7) 77 (17) Data are n (%) or median (IQR). Percentages represent the occurrence of
complications, as part of the total. TPN=total parenteral nutrition. ICU=intensive
care unit.
Figure 2: Kaplan–Meier estimates of overall survival according to modified intention-to-treat analysis (A) and
intention-to-treat analysis (B) Table 2: Postoperative complications and serious adverse events from
Shaded areas around the curves show 95% CIs. Vertical dashes denote censored patients. CIs were not adjusted for clinical response evaluations of patients undergoing oesophagectomy
confounders and should not be used for hypothesis testing. HRs are from adjusted analyses. HR=hazard ratio.

8 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0

 Articles

neoadjuvant chemoradiotherapy), 90 (94%) of 100

96 participants with isolated locoregional regrowth were

Overall survival (%)

I

detected. 86 participants were planned for surgery and 75

II II II I II I I III I
83 participants eventually underwent oesophagectomy 50
IIIIIIIIIIIIIII I II
II
I
I IIII I IIIIIIIII III
I II I
IIIII I IIII I II

after a median time of 5·9 months (range 3·3–33·8)

25 Active surveillance
after reaching clinical complete response. Micro­ HR 1·25 (95% CI 0·83–1·89); p=0·29
Standard surgery
scopically and macroscopically irradical resection rates 0
0 6 12 18 24 30 36
were two (2%) of 83 and one (1%) of 83 in the active
Time since clinically complete response (months)
surveillance group, respectively, versus two (2%) of 101 Number at risk
(number censored)
and none of 101 in standard surgery group, respectively. Active surveillance 198 (0) 173 (0) 156 (0) 130 (0) 119 (5) 81 (30) 38 (65)
Operative and pathological outcomes are summarised Standard surgery 111 (0) 102 (0) 87 (1) 73 (1) 68 (1) 60 (7) 53 (10)
in the appendix (pp 21–22). Of participants undergoing
oesophagectomy, 68 (82%) of 83 in the active surveillance Figure 3: Kaplan–Meier estimates of disease-free survival according to modified intention-to-treat analysis
Shaded areas around the curves show 95% CIs. Vertical dashes denote censored patients. CIs were not adjusted for
group and 85 (84%) of 101 in the standard surgery group confounders and should not be used for hypothesis testing. The HR is from an adjusted analysis. HR=hazard ratio.
had at least one postoperative complication (table 2).
The frequency of postoperative complications was disease-free survival and distant-dissemination rate.
similar between groups. Three (4%) of 83 participants Operative outcomes and postoperative complications
died within 90 days after postponed surgery and were similar after postponed surgery for locoregional
five (5%) of 101 participants died within 90 days after regrowth compared with standard surgery. Furthermore,
standard surgery. participants undergoing active surveillance had better
Median disease-free survival was 35 months short-term global health-related quality of life.
(95% CI 28–NR) for participants undergoing active Two groups of participants appeared to benefit from
surveillance and 49 months (31–NR) for participants active surveillance: participants with persistent clinical
undergoing standard surgery (HR 1·25, 95% CI complete response and those who were diagnosed with
0·83–1·89, p=0·29; figure 3). At the time of second distant metastases early during follow-up, before
follow-up PET-CT scan (30 months after completion of oesophagectomy. If we assume that metastases detected
neoadjuvant chemoradiotherapy), 76 (43%) of within 6 months after completion of neoadjuvant
176 participants in the active surveillance group had chemoradiotherapy were already subclinically present at
distant metastases versus 30 (34%) of 87 participants in diagnosis, 91 of 198 participants reaching clinical
the standard surgery group (OR 1·45, 95% CI 0·85–2·48, complete response undergoing active surveillance had
p=0·18). been spared unbeneficial oesophagectomy.
Global health-related quality-of-life questionnaires The safety of postponed oesophagectomy after chemo­
were returned by 43–79% of participants at various radiotherapy has previously been questioned. Salvage
timepoints during follow-up (appendix p 23). oesophagectomy following definitive chemoradiotherapy
Participants who underwent active surveillance with radiotherapy doses of 50 Gy or higher is associated
(including the participants with postponed oeso­ with increased complication rates.5 It is important,
phagectomy in case of locoregional regrowth), had however, to differentiate between salvage surgery in
statistically significantly better global health-related selected participants after definitive chemo­radiotherapy
quality of life than participants who underwent standard versus neoadjuvant chemoradiotherapy followed by
surgery at 6 months (increase of 10·4 [95% CI 4·1–16.5], active surveillance, in which frequent clinical response
p=0·0010, Cohen’s d=0·58) and 9 months (increase evaluations are performed and postponed
of 8·5 [2·0–15.0], p=0·0099, Cohen’s d=0·47) after oesophagectomy is offered to all participants with
completion of neoadjuvant chemoradiotherapy isolated locoregional regrowth. Previously, in the
(appendix p 12). From 12 months, no statistically ARTDECO trial, it was reported that increasing
significant differences were seen between groups. radiotherapy doses up to 60 Gy did not improve
According to intention-to-treat analyses, no statistically locoregional tumour control or overall survival.15
significant differences were seen for disease-free Therefore, to decrease the risk for an increased
survival, for short-term global health-related quality of complication rate after postponed oesophagectomy, the
life, nor for distant-dissemination rate (appendix relatively mild CROSS regimen with a radiotherapy
pp 13–14, 24). dose of 41·4 Gy was chosen for active surveillance. In
the present study, R0 resection rates were identical in
Discussion the two groups and, although not statistically tested, the
After a minimum follow-up of 2 years, overall survival frequency of postoperative complications was similar.
after active surveillance was non-inferior compared with Thus, postponed oesophagectomy after neoadjuvant
standard surgery for participants with clinical complete chemoradiotherapy was not associated with increased
response after neoadjuvant chemoradiotherapy. There surgical risk in the present trial. This finding is in line
were no statistically significant differences in with the results of the randomised NeoRes II trial,

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 9

 Articles

comparing surgery after 6 weeks versus surgery after with CTLA-4 inhibition is associated with improved
12 weeks following neoadjuvant chemoradiotherapy.16 disease-free survival and increased rates of pathological
Additionally, the NeoRes II trial showed no statistical complete response, respectively, in oesophageal cancer.20
differences in overall survival when surgery was delayed Application of immunotherapy during active surveillance
from 6 weeks to 12 weeks after neoadjuvant might decrease the risk for locoregional regrowth and
chemoradiotherapy. However, in a non-predefined development of metastases, and this approach is
post-hoc subgroup analysis, participants with major currently being investigated in the SANO-3 trial
residual disease (tumour regression grade 4) were (NCT05491616). Another strategy could be to intensify
suggested to have inferior overall survival when surgery the systemic component of neoadjuvant treatment by
was done at 12 weeks instead of 6 weeks. This analysis adding induction or consolidation chemotherapy or
consisted of only 46 participants in total, and therefore PD-1 inhibition to the CROSS regimen. However, before
no definitive conclusions can be drawn. Previous another regimen can be used in terms of active
retrospective studies and a meta-analysis on time-to- surveillance, more data should be available on the rate of
surgery after neoadjuvant chemoradiotherapy have complete response after surgery and the safety of
consistently showed that postponement of oeso­ performing postponed oesophagectomy.
phagectomy does not negatively affect overall survival.17 In the present study, we assumed that metastases
Although these studies are affected by selection bias, we detected within 6 months were already present
feel that the relatively low level of evidence provided by subclinically, and that participants were spared futile
post-hoc analyses of the NeoRes II trial does not disprove oesophagectomy because metastatic disease strongly
these previously published data. In our view, the present dictates participants’ prognosis. Of course, if possibilities
strategy with clinical response evaluations at 6 weeks for surgical treatment of oligometastatic disease expand
and 12 weeks is safe and beneficial for the whole group in the future, a more personalised surgical approach is
(including those with a clinical non-complete response) warranted. Although disease-free survival and distant-
compared with a strategy in which all participants dissemination rate were not statistically different
undergo oesophagectomy at 6 weeks. between the two treatment groups, both appeared to be
It has been argued that active surveillance should only numerically worse in the active surveillance group. This
be explored in participants with squamous cell finding might indicate an increased risk for distant
carcinoma because this histological subtype has the dissemination during active surveillance. In rectal
highest chance of a pathological complete response. cancer, studies from the past year suggest that
We decided to include both participants with participants with locoregional regrowth after clinical
adenocarcinoma and participants with squamous cell complete response are indeed at increased risk for
carcinoma because neoadjuvant chemoradiotherapy is distant dissemination.21 However, follow-up in the active
standard of care for both histological subtypes, and both surveillance group of the present study included more
adenocarcinoma and squamous cell carcinoma have a frequent response evaluations, which might have
substantial chance of a pathological complete response. resulted in earlier detection of asymptomatic recurrent
In the precursor preSANO trial, the risk of false-negative disease. Given the impact of oesophagectomy on
clinical response evaluations was similar for both participants’ health-related quality of life, even a slightly
tumour types. Moreover, if participants have a clinical worse long-term overall survival might be acceptable for
complete response, it is unlikely that the efficacy of certain subgroups of participants. This is in line with
active surveillance is different for one of the two tumour participants’ own preferences, as described previously.13
types. The results of the present study support the Extension of the follow-up period is needed to assess
suggestion that active surveillance is also suitable for long-term efficacy of active surveillance.
participants with adenocarcinoma, given that two-thirds Some limitations of the present study should be
of participants (n=44) with persistent clinical complete addressed. Although the groups seemed well balanced,
response had adenocarcinoma, which was 30% of all institution-level randomisation is susceptible to selection
included participants with adenocarcinoma undergoing bias, in contrast to patient-level randomisation. To
active surveillance (n=147), and overall survival was provide the highest level of evidence and to reduce risk
similar after inclusion of 231 participants with for premature termination of the study, pragmatic
adenocarcinoma, reaching sufficient power to make cluster randomisation following a stepped-wedge design
robust conclusions. was used in the present study. In this way, participants
The number of participants who initially had clinical already knew what they were consenting to at time of
complete response, but who developed locoregional inclusion. Although we implemented some inclusion
regrowth requiring oesophagectomy, was substantial. rules in this trial to avoid selection bias (eg, the
New strategies to improve accuracy of clinical response prohibition for participating centres to refer their own
evaluations (eg, by using Cytosponge or adding participants to other participating centres due to
circulating tumour DNA) and to increase complete treatment preference), participants might have known
response rates should be investigated.18,19 PD-1 and PD-1 in advance which hospitals offered the treatment of their

10 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0

 Articles

choice and based their first hospital visit on this data curation, and writing—review and editing. CR, TMB, P-PLOC,
MJvD, JWTD, JMvD, MD, SvE, WEF, HHH, JH, ILH, BK, EK, EAK,
preference. To provide insight into whether selection MDL, BM, GAPN, LEO, J-PP, JWvS, MNS, MCWS, RV, and ESvdZ: data
bias might have occurred after all, despite the previously acquisition and writing—review and editing. DvK: conceptualisation,
mentioned measures, we reported the number of data analysis, and writing—review and editing. EWS: conceptualisation,
participants who refused participation in the trial. These supervision, and writing—review and editing.
numbers indicate that, although the risk is still present, Declaration of interests
selection bias did not occur on a large scale. Low response BM has a consultancy role for Lilly, Servier, BMS, Amgen, and
AstraZeneca and receives research funding from BMS, Pfizer, and Sanofi.
rates on the health-related quality-of-life questionnaires EAK has a consultancy role for Intuitive Surgical. MJvD has a training and
could have resulted in selection bias as well. Therefore, proctoring role for Intuitive Surgical. BPLW receives research funding
we adjusted for prognostic factors at baseline, to correct from BMS and Medtronic, and has particpated on the data safety and
for this potential selection bias. monitoring for the Neo-Aegis trial. MDL receives funding from Galvani
and Medtronic. All other authors declare no competing interests.
As described in our protocol and statistical analysis
plan, a group of participants from the preSANO trial Data sharing
Individual participant data (eg, for individual patient data meta-analyses)
were included in the control group of the present SANO will be available on request to the corresponding author, with interested
trial. This strategy introduced a heterogeneity between parties. Particular data on all of the individual participant data collected
the two treatment groups. We believe that this effect was during the trial, after de-identification, will be shared on agreement.
marginal, because the study procedures for participants The study protocol has been published and the statistical analysis plan
and analytical code will be available on request, immediately after
undergoing preSANO and SANO were identical and only publication.
a modest part of the standard surgery group consisted of
Acknowledgments
preSANO participants. It was considered undesirable to We thank the Dutch Cancer Society (KWF) and Netherlands
expose a larger group of participants than statistically Organisation for Health Research and Development (ZonMw) for their
necessary to an experimental treatment. Adding financial support in the present study. Furthermore, we thank all
participants for taking part in our trial. We thank Liesbeth Timmermans
participants from the preSANO trial decreased the
from the patient organisation (SPKS) for her advisory efforts, all nurse
number of participants required to be randomly assigned. practitioners from participating centres for their clinical assistance, the
We decided to perform a post-hoc sensitivity analysis data managers for their support in storing our data safely, and Xing Gao
after exclusion of the preSANO participants from survival for her administrative assistance.
analysis. These results were consistent with the primary References
analysis and conclusions of the present study. Of note, 1 Eyck BM, van Lanschot JJB, Hulshof MCCM, et al. Ten-year
outcome of neoadjuvant chemoradiotherapy plus surgery for
participants in this study were selected for having a esophageal cancer: the randomized controlled CROSS trial.
favourable response during two clinical response J Clin Oncol 2021; 39: 1995–2004.
evaluations. Therefore, the results cannot be extrapolated 2 van der Wilk BJ, Eyck BM, Hofstetter WL, et al. Chemoradiotherapy
followed by active surveillance versus standard esophagectomy for
to participants with oesophageal cancer who have not esophageal cancer: a systematic review and individual patient data
undergone proper clinical response evaluations. Lastly, meta-analysis. Ann Surg 2022; 275: 467–76.
four participants in the active surveillance group received 3 Noordman BJ, Spaander MCW, Valkema R, et al. Detection of
residual disease after neoadjuvant chemoradiotherapy for
nivolumab after oesophagectomy, according to newly oesophageal cancer (preSANO): a prospective multicentre,
introduced national guidelines during the study, which diagnostic cohort study. Lancet Oncol 2018; 19: 965–74.
could have had minimal effect on the results. 4 Low DE, Kuppusamy MK, Alderson D, et al. Benchmarking
complications associated with esophagectomy. Ann Surg 2019;
In conclusion, participants undergoing active 269: 291–98.
surveillance after clinical complete response had non- 5 Chidambaram S, Owen R, Sgromo B, et al. Delayed surgical
inferior 2-year overall survival compared with those intervention after chemoradiotherapy in esophageal cancer: (DICE)
study. Ann Surg 2023; 278: 701–08.
undergoing standard surgery. However, longer follow-up
6 Noordman BJ, Wijnhoven BPL, Lagarde SM, et al. Neoadjuvant
is needed to assess the robustness of this strategy. Of chemoradiotherapy plus surgery versus active surveillance for
participants with clinical complete response undergoing oesophageal cancer: a stepped-wedge cluster randomised trial.
BMC Cancer 2018; 18: 142.
active surveillance, almost half were spared unbeneficial
7 Eyck BM, van der Wilk BJ, Noordman BJ, et al. Updated protocol of
oesophagectomy, resulting in improved short-term the SANO trial: a stepped-wedge cluster randomised trial
health-related quality of life. Postponed oesophagectomy comparing surgery with active surveillance after neoadjuvant
was not associated with increased morbidity or mortality. chemoradiotherapy for oesophageal cancer. Trials 2021; 22: 345.
8 Hemming K, Taljaard M, McKenzie JE, et al. Reporting of stepped
Although longer-term outcome is awaited, the present wedge cluster randomised trials: extension of the CONSORT 2010
trial provides the framework for patient counselling, statement with explanation and elaboration. BMJ 2018; 363: k1614.
shared decision making, and developing clinical response 9 Piaggio G, Elbourne DR, Pocock SJ, Evans SJ, Altman DG.
Reporting of noninferiority and equivalence randomized trials:
evaluation protocols. extension of the CONSORT 2010 statement. JAMA 2012;
Contributors 308: 2594–604.
BJvdW and BME: conceptualisation, data curation, data analysis, 10 van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative
funding acquisition, accessed and verified the data, and writing— chemoradiotherapy for esophageal or junctional cancer.
original draft. BPLW, SML, and JJBvL: conceptualisation, funding N Engl J Med 2012; 366: 2074–84.
acquisition, data acquisition, supervision, and writing—review and 11 McCulloch P, Taylor I, Sasako M, Lovett B, Griffin D. Randomised
editing. BJN: conceptualisation, data curation, data analysis, funding trials in surgery: problems and possible solutions. BMJ 2002;
acquisition, and writing—review and editing. MJV: data acquisition, 324: 1448–51.

www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0 11

 Articles

12 Blazeby JM, Strong S, Donovan JL, et al. Feasibility RCT of 18 Azad TD, Chaudhuri AA, Fang P, et al. Circulating tumor DNA
definitive chemoradiotherapy or chemotherapy and surgery for analysis for detection of minimal residual disease after
oesophageal squamous cell cancer. Br J Cancer 2014; 111: 234–40. chemoradiotherapy for localized esophageal cancer. Gastroenterology
13 Noordman BJ, de Bekker-Grob EW, Coene PPLO, et al. Patients’ 2020; 158: 494–505.e6.
preferences for treatment after neoadjuvant chemoradiotherapy for 19 Jones CM, O’Connor H, O’Donovan M, et al. Use of a non-
oesophageal cancer. Br J Surg 2018; 105: 1630–38. endoscopic immunocytological device (Cytosponge™) for post
14 Mo Y, Lim C, Watson JA, White NJ, Cooper BS. Non-adherence in chemoradiotherapy surveillance in patients with oesophageal
non-inferiority trials: pitfalls and recommendations. BMJ 2020; cancer in the UK (CYTOFLOC): a multicentre feasibility study.
370: m2215. EClinicalMedicine 2022; 53: 101664.
15 Hulshof MCCM, Geijsen ED, Rozema T, et al. Randomized study 20 Kelly RJ, Ajani JA, Kuzdzal J, et al. Adjuvant nivolumab in resected
on dose escalation in definitive chemoradiation for patients with esophageal or gastroesophageal junction cancer. N Engl J Med 2021;
locally advanced esophageal cancer (ARTDECO study). J Clin Oncol 384: 1191–203.
2021; 39: 2816–24. 21 Fernandez LM, São Julião GP, Renehan AG, et al. The risk of
16 Nilsson K, Klevebro F, Sunde B, et al. Oncological outcomes of distant metastases in patients with clinical complete response
standard versus prolonged time to surgery after neoadjuvant managed by watch and wait after neoadjuvant therapy for rectal
chemoradiotherapy for oesophageal cancer in the multicentre, cancer: the influence of local regrowth in the international watch
randomised, controlled NeoRes II trial. Ann Oncol 2023; and wait database. Dis Colon Rectum 2023; 66: 41–49.
34: 1015–24.
17 Karthyarth MN, Mathew A, Ramachandra D, et al. Early versus
delayed surgery following neoadjuvant chemoradiation for
esophageal cancer: a systematic review and meta-analysis.
Esophagus 2023; 20: 390–401.

12 www.thelancet.com/oncology Published online March 17, 2025 https://doi.org/10.1016/S1470-2045(25)00027-0