Articles

Neoadjuvant chemoradiotherapy plus surgery versus

surgery alone for oesophageal or junctional cancer (CROSS):
long-term results of a randomised controlled trial
Joel Shapiro, J Jan B van Lanschot, Maarten C C M Hulshof, Pieter van Hagen, Mark I van Berge Henegouwen, Bas P L Wijnhoven,
Hanneke W M van Laarhoven, Grard A P Nieuwenhuijzen, Geke A P Hospers, Johannes J Bonenkamp, Miguel A Cuesta, Reinoud J B Blaisse,
Olivier R C Busch, Fiebo J W ten Kate, Geert-Jan M Creemers, Cornelis J A Punt, John Th M Plukker, Henk M W Verheul, Ernst J Spillenaar Bilgen,
Herman van Dekken, Maurice J C van der Sangen, Tom Rozema, Katharina Biermann, Jannet C Beukema, Anna H M Piet, Caroline M van Rij,
Janny G Reinders, Hugo W Tilanus, Ewout W Steyerberg, Ate van der Gaast, for the CROSS study group

Summary
Background Initial results of the ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study (CROSS) Lancet Oncol 2015
comparing neoadjuvant chemoradiotherapy plus surgery versus surgery alone in patients with squamous cell Published Online
carcinoma and adenocarcinoma of the oesophagus or oesophagogastric junction showed a significant increase in August 6, 2015
http://dx.doi.org/10.1016/
5-year overall survival in favour of the neoadjuvant chemoradiotherapy plus surgery group after a median of
S1470-2045(15)00040-6
45 months’ follow-up. In this Article, we report the long-term results after a minimum follow-up of 5 years.
See Online/Comment
http://dx.doi.org/10.1016/
Methods Patients with clinically resectable, locally advanced cancer of the oesophagus or oesophagogastric junction S1470-2045(15)00127-8
(clinical stage T1N1M0 or T2–3N0–1M0, according to the TNM cancer staging system, sixth edition) were randomly Department of Surgery
assigned in a 1:1 ratio with permuted blocks of four or six to receive either weekly administration of five cycles of (J Shapiro MD,
neoadjuvant chemoradiotherapy (intravenous carboplatin [AUC 2 mg/mL per min] and intravenous paclitaxel Prof J J B van Lanschot MD,
P van Hagen MD,
[50 mg/m² of body-surface area] for 23 days) with concurrent radiotherapy (41·4 Gy, given in 23 fractions of 1·8 Gy on B P L Wijnhoven MD,
5 days per week) followed by surgery, or surgery alone. The primary endpoint was overall survival, analysed by Prof H W Tilanus MD),
intention-to-treat. No adverse event data were collected beyond those noted in the initial report of the trial. This trial Department of Pathology
is registered with the Netherlands Trial Register, number NTR487, and has been completed. (Prof F J W ten Kate MD,
H van Dekken MD,
K Biermann MD), Department of
Findings Between March 30, 2004, and Dec 2, 2008, 368 patients from eight participating centres (five academic Radiation Oncology
centres and three large non-academic teaching hospitals) in the Netherlands were enrolled into this study and (C M van Rij MD), Department
of Public Health
randomly assigned to the two treatment groups: 180 to surgery plus neoadjuvant chemoradiotherapy and 188 to
(Prof E W Steyerberg PhD), and
surgery alone. Two patients in the neoadjuvant chemoradiotherapy group withdrew consent, so a total of 366 patients Department of Medical
were analysed (178 in the neoadjuvant chemoradiotherapy plus surgery group and 188 in the surgery alone group). Oncology (A van der Gaast MD),
Of 171 patients who received any neoadjuvant chemoradiotherapy in this group, 162 (95%) were able to complete the Erasmus MC—University
Medical Centre Rotterdam,
entire neoadjuvant chemoradiotherapy regimen. After a median follow-up for surviving patients of 84·1 months
Rotterdam, Netherlands;
(range 61·1–116·8, IQR 70·7–96·6), median overall survival was 48·6 months (95% CI 32·1–65·1) in the neoadjuvant Department of Surgery
chemoradiotherapy plus surgery group and 24·0 months (14·2–33·7) in the surgery alone group (HR 0·68 [95% CI (Prof J J B van Lanschot,
0·53–0·88]; log-rank p=0·003). Median overall survival for patients with squamous cell carcinomas was 81·6 months M I van Berge Henegouwen MD),
Department of Radiation
(95% CI 47·2–116·0) in the neoadjuvant chemoradiotherapy plus surgery group and 21·1 months (15·4–26·7) in the
Oncology (M C C M Hulshof MD,
surgery alone group (HR 0·48 [95% CI 0·28–0·83]; log-rank p=0·008); for patients with adenocarcinomas, it was Prof O R C Busch MD),
43·2 months (24·9–61·4) in the neoadjuvant chemoradiotherapy plus surgery group and 27·1 months (13·0–41·2) Department of Medical
in the surgery alone group (HR 0·73 [95% CI 0·55–0·98]; log-rank p=0·038). Oncology
(Prof H W M van Laarhoven MD,
Prof C J A Punt MD), and
Interpretation Long-term follow-up confirms the overall survival benefits for neoadjuvant chemoradiotherapy when Department of Pathology
added to surgery in patients with resectable oesophageal or oesophagogastric junctional cancer. This improvement is (Prof F J W ten Kate), Academic
clinically relevant for both squamous cell carcinoma and adenocarcinoma subtypes. Therefore, neoadjuvant Medical Centre, Amsterdam,
Netherlands; Department of
chemoradiotherapy according to the CROSS trial followed by surgical resection should be regarded as a standard of Surgery
care for patients with resectable locally advanced oesophageal or oesophagogastric junctional cancer. (G A P Nieuwenhuijzen MD),
Department of Medical
Funding Dutch Cancer Foundation (KWF Kankerbestrijding). Oncology (G-J M Creemers MD),
and Department of Radiation
Oncology
Introduction resectability, long-term locoregional control, and overall (M J C van der Sangen MD),
Oesophageal cancer is an aggressive disease, survival, through the addition of chemotherapy, Catharina Hospital, Eindhoven,
characterised by a high degree of locoregional and radiotherapy, or both, to surgery, in a neoadjuvant or Netherlands; Department of
Medical Oncology
distant recurrence after primary surgical resection and adjuvant setting.2–5 However, many studies have not (Prof G A P Hospers MD),
poor 5-year overall survival that rarely exceeds 40%.1–3 shown a significant long-term survival benefit of such Department of Surgery
Much effort has been put into improving tumour approaches.6,7 (Prof J T M Plukker MD), and

www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6 1

 Articles

Department of Radiation
Oncology (J C Beukema MD), Research in context
University Medical Centre
Groningen, Groningen, Evidence before this study as standard of care in some countries for patients with
Netherlands; Department of Based on the extensive meta-analysis by Sjoquist and oesophageal and junctional cancer. This perspective is mainly
Surgery (J J Bonenkamp MD), colleagues in 2011, at the initiation of the CROSS trial in the consequence of the results of the MAGIC trial, which
Department of Medical
Oncology (Prof C J A Punt), and
2004, the results from four previous randomised trials compared perioperative chemotherapy—consisting of
Department of Radiation comparing neoadjuvant concurrent chemoradiotherapy plus epirubicin, cisplatin, and infused fluorouracil—plus surgery
Oncology (T Rozema MD), surgery to surgery alone had been reported. Chemotherapy in versus surgery alone. However, only a few included patients
Radboud University Nijmegen these trials consisted of cisplatin and fluorouracil (and also had distal oesophageal cancers (14%) or junctional cancers
Medical Centre, Nijmegen,
Netherlands; Department of
vinblastine in one trial), with a total concurrent radiation (12%), which raises questions about the applicability of these
Surgery (Prof M A Cuesta MD), dose ranging from 40 to 45 Gy. However, these trials included results for oesophageal and junctional cancers. Furthermore,
Department of Medical only small numbers of patients and showed opposing results. the MAGIC trial, which was published in 2006 after a
Oncology Our previous non-randomised phase 2 feasibility trial tested a minimum follow-up of less than 2 years, has not yet reported
(Prof H M W Verheul MD), and
Department of Radiation
regimen of weekly administrations of carboplatin and its long-term results, which makes it unclear whether or not
Oncology (A H M Piet MD), VU paclitaxel with 41·4 Gy concurrent radiotherapy and showed the initially reported survival benefit of perioperative
Medical Centre, Amsterdam, a radical resection percentage of 100%, with low chemotherapy is sustained at long-term follow-up. The
Netherlands; Department of treatment-related toxicity. These promising short-term ongoing Japanese randomised NExT trial (JCOG1109) and the
Medical Oncology
(R J B Blaisse MD) and
results provided the rationale to assess this CROSS Irish randomised Neo-AEGIS trial
Department of Surgery neoadjuvant chemoradiotherapy regimen in a subsequent (ICORG 10-14; NCT01726452) will probably provide more
(E J Spillenaar Bilgen MD), randomised phase 3 trial. definitive evidence about the optimum preoperative or
Rijnstate Hospital, Arnhem,
perioperative treatment for oesophageal squamous cell
Netherlands; Department of Added value of this study
Pathology, St Lucas Andreas carcinoma and adenocarcinoma, respectively. Future research
At long-term follow-up, the CROSS trial has now shown that
Hospital, Amsterdam, should focus on more personalised treatment strategies, such
treatment of locally advanced oesophageal or junctional cancer
Netherlands (H van Dekken); as watchful waiting protocols after neoadjuvant therapy, in
Verbeeten Institute, Tilburg, with carboplatin, paclitaxel, and concurrent radiotherapy
which surgery is offered only to those patients in whom
Netherlands (T Rozema); and followed by surgery significantly improves 5-year overall and
Arnhem Radiotherapeutic locoregional disease is detected (in the absence of signs of
progression-free survival, compared with treatment with
Institute ARTI, Arnhem, distant dissemination). Additionally, newer, more effective
Netherlands (J G Reinders MD)
surgery alone.
combinations of systemic agents need to undergo further
Correspondence to: Implications of all the available evidence study, such as the addition of targeted therapy to existing
Dr Joel Shapiro, Department of chemoradiotherapeutic treatment regimens.
Despite the favourable results of the initial CROSS trial,
Surgery, Erasmus MC – University
Medical Centre, PO Box 2040, preoperative or perioperative chemotherapy is still regarded
3000 CA Rotterdam,
Netherlands
j.shapiro@erasmusmc.nl The randomised controlled ChemoRadiotherapy for Methods
Oesophageal cancer followed by Surgery Study (CROSS) Study design and participants
trial8 compared neoadjuvant chemoradiotherapy plus Full details of patients’ eligibility criteria and the
surgery versus surgery alone. The trial enrolled 368 patients procedures of this open-label, multicentre, randomised
between March 30, 2004, and Dec 2, 2008, from eight controlled trial have been reported previously.8,9 In brief,
Dutch participating centres (five academic centres and eligible patients were aged 75 years or younger; had
three large non-academic teaching hospitals). Initial results adequate haematological, renal, hepatic, and pulmonary
were published in 2012 after a minimum follow-up of function; and a WHO performance score of 2 or better,
24 months (median follow-up 45 months [range 25·2–80·9, without a past or present history of other malignancy.
IQR 32·6–60·6]). We recorded an absolute benefit in 5-year Only patients with locally advanced (clinical stage
overall survival in favour of the multimodality group. The T1N1M0 or clinical stage T2–3N0–1M0, according to the
neoadjuvant chemoradiotherapy regimen was completed Union for International Cancer Control [UICC] TNM
by 162 (95%) of 171 patients who received any neoadjuvant cancer staging, 6th edition10), histologically proven, and
chemoradiotherapy, with a low occurrence of grade 3 or potentially curable squamous cell carcinoma or
adverse events for this setting (29 [17%] of 171 patients). adenocarcinoma of the oesophagus or oesophagogastric
Furthermore, a microscopically radical resection (ie, no junction (ie, tumours involving both the cardia and the
vital tumour present at <1 mm from the proximal, distal, or oesophagus on endoscopy) were eligible for inclusion.
circumferential resection margins) was achieved in 148 The main exclusion criteria were past or current history
(92%) of 161 patients in the multimodality group, compared of malignancy other than the oesophageal malignancy,
with 112 (69%) of 162 in the surgery alone group (p<0·001). previous chemotherapy and/or radiotherapy, and weight
In this Article, we investigate the consistency of longer- loss of more than 10% of the original bodyweight. The
term results with our previous findings and analyse institutional review board at each participating centre
secondary endpoints, such as progression-free survival approved the study protocol. All patients provided
and disease recurrence patterns. written informed consent.

2 www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6

 Articles

Randomisation and masking located well below the level of the carina, either a
Patients were randomly assigned 1:1 to each treatment transthoracic approach with two-field lymph node
group, and were stratified according to histological tumour dissection or a transhiatal approach was used, depending
type (adenocarcinoma vs squamous cell carcinoma), on both patient characteristics and local preferences. For
treatment centre, clinical nodal status (cN0 vs cN1), and carcinomas involving the oesophagogastric junction, a
WHO performance score (WHO-0 vs WHO-1 vs WHO-2). transhiatal oesophageal resection was preferred. In both
Randomisation was done centrally at the Clinical Trial approaches, an upper abdominal lymphadenectomy,
Center at Erasmus MC (Rotterdam, the Netherlands), by including resection of nodes along the hepatic artery,
computer-generated randomisation lists for each stratum, splenic artery, and left gastric artery, was done.
with random permuted block sizes of four or six. For TNM classification, tumour grading, and stage
grouping, the sixth edition of the UICC TNM cancer
Procedures staging was used.10 Proximal, distal, and circumferential
All patients underwent pretreatment staging, including resection margins were assessed. Microscopically radical
upper gastrointestinal endoscopy with histological biopsy resection (R0) was defined as a tumour-free resection
and endoscopic ultrasonography; CT scan of the neck, margin of at least 1 mm.
chest, and upper abdomen; and external ultrasonography During the first year after treatment completion,
of the neck, with fine-needle aspiration of suspected patients were seen every 3 months. In the second year,
lymph nodes on indication. follow-up took place every 6 months, and annually
For patients assigned to receive neoadjuvant chemo- thereafter until 5 years after treatment. Additional interim
radiotherapy, carboplatin (AUC 2 mg/mL per min) and visits were scheduled if complaints such as renewed
paclitaxel (50 mg/m² of body-surface area) were dysphagia and unexplained weight loss or pain arose
administered intravenously for five cycles, starting on before the next scheduled visit. Diagnostic investigations
days 1, 8, 15, 22, and 29. A total concurrent radiation dose were only undertaken as necessary during follow-up. No
of 41·4 Gy was given in 23 fractions of 1·8 Gy, on 5 days per data for adverse events were collected beyond the initial
week (excluding weekends), starting on the first day of the report of this trial.8
first chemotherapy cycle. The total duration of neoadjuvant
treatment was 23 days (5 days per week in weeks 1, 2, 3, Outcomes
and 4, then 3 days in week 5). If on days 8, 15, 22, or 29 the The primary endpoint was overall survival, which was
white blood cell count was lower than 1·0 × 10⁹ cells per L calculated from the date of randomisation to the date of
or the platelet count was lower than 50 × 10⁹ per L, admini- all-cause death or to the last day of follow-up. Secondary
stration of neoadjuvant chemoradiotherapy was delayed by endpoints included progression-free survival and
1 week until recovery above these thresholds. Furthermore, progression-free interval. Progression-free survival was
in case of mucositis with oral ulcers or protracted vomiting defined as the interval between randomisation and the
despite antiemetic premedication, neoadjuvant chemo- earliest occurrence of disease progression resulting in
radiotherapy was delayed by 1 week. Further chemotherapy primary (or peroperative) irresectability of disease, loco-
was withheld in case of febrile neutropenia (defined as a regional recurrence (after completion of therapy), distant
neutrophil count <0·5 × 10⁹ cells per L and a body dissemination (during or after completion of treatment),
temperature >38·5°C), persistent creatinine clearance of or death from any cause. This definition for progression-
less than 50% of the pretreatment level, symptomatic free survival was taken from the modified STEEP criteria
cardiac arrhythmia or atrioventricular block (with the for neoadjuvant treatment trials.11,12 The last day of follow-
exception of first-degree atrioventricular block), or other up for progression-free survival varied, depending on the
major organ toxicity at grade 3 or worse (with the exception most recent (scheduled) contact with the patient.
of oesophagitis). During neoadjuvant chemoradiotherapy, Progression-free interval was similar to progression-free
laboratory tests (including complete blood cell counts and survival, with the difference that treatment-related deaths
serum creatinine measurement) were done on a weekly and non-oesophageal cancer-related deaths were not
basis, whereas radiological assessments were done only on counted as events. Locoregional progression was defined
indication. All patients in the neoadjuvant chemo- as either progression of locoregional disease during
radiotherapy plus surgery group were included into the treatment (resulting in irresectability) or as locoregional
intention-to-treat analysis, irrespective of total dose of recurrence after completion of treatment. Locoregional
neoadjuvant chemoradiotherapy received. sites included the mediastinum, the supraclavicular
Patients in the surgery alone group were operated on as region, and the coeliac trunk region. Distant progression
soon as possible, whereas those in the neoadjuvant was defined as occurrence of disseminated disease, either
chemoradiotherapy plus surgery group were preferably during or after completion of treatment. Distant disease
operated on within 4–6 weeks after completion of included cervical and (para-aortic) lymph node
chemoradiotherapy. For carcinomas at or above the level dissemination below the level of the pancreas, malignant
of the carina, a transthoracic oesophageal resection with pleural effusions, peritoneal carcinomatosis, and further
two-field lymph node dissection was done. For carcinomas haematogenous (organ) dissemination.

www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6 3

 Articles

Neoadjuvant Surgery alone

837 patients assessed for oesophageal or
chemoradiotherapy (n=188)
oesophagogastric junction cancer
plus surgery
(n=178)

469 excluded Age, years 60 (55–67) 60 (53–66)

Sex
Women 44 (25%) 36 (19%)
368 patients enrolled and randomly assigned
Men 134 (75%) 152 (81%)
Tumour histology
Squamous cell carcinoma 41 (23%) 43 (23%)
180 assigned to neoadjuvant chemoradiotherapy 188 assigned to surgery alone Adenocarcinoma 134 (75%) 141 (75%)
and surgery
Could not be established 3 (2%) 4 (2%)
Tumour length, cm 4 (3–6) 4 (3–6)
2 withdrew consent Tumour location
Proximal third oesophagus 4 (2%) 4 (2%)
Middle third oesophagus 25 (14%) 24 (13%)
166 received neoadjuvant chemoradiotherapy 187* underwent surgery
and surgery 162 underwent resection Distal third oesophagus 104 (58%) 107 (57%)
5 received chemoradiotherapy only Oesophagogastric junction 39 (22%) 49 (26%)
2 underwent surgery only
5 received neither chemoradiotherapy Missing data 6 (3%) 4 (2%)
nor surgery Clinical tumour (cT) stage
161 underwent resection
cT1 1 (1%) 1 (1%)
cT2 26 (15%) 35 (19%)
178 were included in the analysis 188 were included in the analysis cT3 150 (84%) 147 (78%)
cT4 0 1 (1%)
Figure 1: Trial profile Could not be established 1 (1%) 4 (2%)
*One patient had disease progression in waiting period to surgery. Clinical nodal (cN) stage
cN0 59 (33%) 58 (31%)
cN1 116 (65%) 120 (64%)
Statistical analysis
Could not be established 3 (2%) 10 (5%)
Data were analysed according to an intention-to-treat
WHO performance score
principle. To detect a difference of 6 months in median
overall survival (22 months in the neoadjuvant 0 144 (81%) 163 (87%)

chemoradiotherapy plus surgery group vs 16 months in 1 34 (19%) 25 (13%)

the surgery alone group, according to a two-sided test Data are median (IQR) or n (%).
with α level 0·05 and β level 0·80), we calculated that we
needed to enrol at least 175 patients in each treatment Table 1: Baseline characteristics

group. The statistical significance level was set to 0·05.

We used the Kaplan-Meier method to estimate overall afterwards. Statistical analysis was done by JS and EWS,
and progression-free survival, with the log-rank test to using SPSS version 21.0. This trial is registered with the
ascertain significance. We used univariable and multi- Netherlands Trial Register, number NTR487.
variable Cox proportional hazards models to establish the
effect of neoadjuvant chemoradiotherapy in subgroups, Role of the funding source
adjusting for baseline covariates.8 Univariable Cox The funder of the study had no role in the study design,
regression modelling was used to analyse differences in data collection, data analysis, data interpretation, or
progression-free interval between treatment groups, writing of the report. The corresponding author had full
expressed as hazard ratios (HRs). Follow-up time was access to all the data in the study and had final
divided to study the temporal distribution of disease responsibility for the decision to submit for publication.
progression. Three separate analyses were done,
including follow-up until 6 months, 12 months, and Results
24 months after randomisation. Progression was defined 368 patients from eight participating centres (five
as locoregional or distant. Patients in whom both types of academic centres and three large non-academic teaching
disease progression occurred had events scored in both hospitals) in the Netherlands were enrolled in the study.
categories. In the scoring of disease progression in one 180 patients were randomly assigned to the neoadjuvant
category, disease progression in the other category and chemoradiotherapy plus surgery group (of whom two
death without progression were censored. For each later withdrew consent and were not included in the
timepoint, we compared the number of events between analysis), and 188 were randomly assigned to the surgery
treatment groups, before the cutoff timepoint and alone group (figure 1). Baseline characteristics were well

4 www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6

 Articles

balanced between the two treatment groups (table 1).

A
One patient in the surgery alone group was originally 100 Neoadjuvant chemoradiotherapy
misclassified as not having received a resection. Surgery alone
However, in the present update, we discovered that this 90
patient had undergone a resection abroad. This
misclassification had no effect on current or previous 80
analyses because of the intention-to-treat principle.
Furthermore, a patient in the neoadjuvant chemo- 70
radiotherapy plus surgery group moved abroad and was
therefore lost to follow-up 73 months after randomisation. 60

Overall survival (%)

Of 171 patients who received any neoadjuvant
chemoradiotherapy in this group, 162 (95%) were able to 50
complete the entire neoadjuvant chemoradiotherapy
regimen. 13 (8%) of 171 patients had grade 3 or worse 40
haematological toxicity and 18 (11%) had grade 3 or
worse non-haematological toxicity. The most common 30
grade 3 or worse toxicities were leucopenia in 11 (6%) of
171 patients, anorexia in nine (5%), and fatigue in five 20

(3%).8 In the neoadjuvant chemoradiotherapy plus

surgery group, 161 (90%) of 178 patients underwent 10

resection, compared with 162 (86%) of 188 in the surgery Log-rank p=0·003
0
alone group. The proportion of patients with transhiatal 0 12 24 36 48 60 72 84 96
resections was similar between both treatment groups Number at risk
(72 [45%] of 161 in the neoadjuvant chemoradiotherapy Neoadjuvant chemo- 178 145 119 103 91 83 59 40 22
radiotherapy plus surgery
plus surgery group and 72 [44%] of 162 in the surgery Surgery alone 188 131 94 83 70 62 42 28 14
alone group; χ²=0·01, p=0·96). Total 366 276 213 186 161 145 101 68 36
The final day of follow-up was Dec 31, 2013, guaranteeing
a minimum potential follow-up of 60 months for all B
included patients. At the time of this analysis, median 100 SCC, neoadjuvant chemoradiotherapy plus surgery
AC, neoadjuvant chemoradiotherapy plus surgery
follow-up for surviving patients was 84·1 months (range SCC, surgery alone
61·1–116·8; IQR 70·7–96·6). Of the 366 analysed patients, 90 AC, surgery alone
126 were still alive at the final analysis (73 [41%] of
178 patients in the neoadjuvant chemoradiotherapy plus 80

surgery group and 53 [28%] of 188 in the surgery alone

70
group). These results correspond with 19 and 21 additional
deaths, respectively, since the last follow-up of the original
60
Overall survival (%)

publication.8 Median overall survival was 48·6 months

(95% CI 32·1–65·1) in the neoadjuvant chemoradiotherapy
50
plus surgery group and 24·0 months (14·2–33·7) in the
surgery alone group (HR 0·68 [95% CI 0·53–0·88];
40
figure 2A). Median overall survival for patients with
squamous cell carcinomas was 81·6 months
30
(95% CI 47·2–116·0) in the neoadjuvant chemoradio-
therapy plus surgery group and 21·1 months (15·4–26·7)
20
in the surgery alone group (HR 0·48 [95% CI 0·28–0·83]),
and median overall survival for patients with 10
adenocarcinomas was 43·2 months (24·9–61·4) in the SCC: log-rank p=0·008
AC: log-rank p=0·038
neoadjuvant chemoradiotherapy plus surgery group and 0
27·1 months (13·0–41·2) in the surgery alone group 0 12 24 36 48 60 72 84 96
(HR 0·73 [95% CI 0·55–0·98]; figure 2B). Overall survival Follow-up (months)
was 81% (95% CI 76–86) at 1 year, 67% (60–74) at 2 years,
Number at risk
SCC, neoadjuvant chemo- 41 35 30 28 26 25 17 11 6
Figure 2: Overall survival radiotherapy plus surgery
(A) By treatment group. (B) By treatment group and histological tumour type; SCC, surgery alone 43 29 19 17 16 13 9 5 4
four patients in the surgery alone group and three in the neoadjuvant AC, neoadjuvant chemo- 134 107 87 73 64 58 42 29 16
radiotherapy plus surgery
chemoradiotherapy plus surgery group were excluded from this analysis because
AC, surgery alone 141 99 73 64 53 47 32 23 10
their histological tumour type could not be ascertained. SCC=squamous cell Total 359 270 209 182 158 143 100 68 36
carcinoma. AC=adenocarcinoma.

www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6 5

 Articles

Neoadjuvant Surgery alone Interaction Univariable analysis Multivariable analysis

chemoradiotherapy (n=188) p value
plus surgery (n=178)
HR (95% CI) p value aHR (95% CI) p value
All patients 105 (59%) 135 (72%) 0·078 0·68 (0·53–0·88) 0·003 0·69 (0·53–0·89) 0·004
Sex 0·451
Women 25 (14%) 24 (13%) 0·83 (0·47–1·45) 0·502 0·85 (0·48–1·50) 0·570
Men 80 (45%) 111 (59%) 0·65 (0·49–0·86) 0·003 0·66 (0·49–0·88) 0·004
Tumour histology 0·207
Squamous cell 21 (12%) 32 (17%) 0·48 (0·28–0·83) 0·009 0·46 (0·26–0·79) 0·005
carcinoma
Adenocarcinoma 81 (46%) 101 (54%) 0·73 (0·55–0·98) 0·037 0·75 (0·56–1·01) 0·059
Clinical nodal (cN) stage 0·170
cN0 27 (15%) 42 (22%) 0·50 (0·31–0·80) 0·004 0·49 (0·30–0·80) 0·004
cN1 77 (43%) 85 (45%) 0·81 (0·59–1·10) 0·176 0·83 (0·61–1·13) 0·237
WHO performance score 0·729
0 84 (47%) 117 (62%) 0·66 (0·50–0·88) 0·004 0·67 (0·51–0·90) 0·006
1 21 (12%) 18 (10%) 0·75 (0·40–1·41) 0·367 0·79 (0·41–1·51) 0·473

Data are n (%), unless otherwise indicated. Multivariable analysis included the following baseline characteristics: sex, tumour histology, clinical lymph node (N) stage, and
WHO performance score. Clinical N stage was based on endoscopic ultrasonography, CT, or 18F-fluorodeoxyglucose PET (in which cN0=no nodes suspected or positive, and
cN1=at least one node suspected or positive). WHO performance status:14 grade 0=able to carry out all normal activity without restrictions, grade 1=restricted in physically
strenuous activity but ambulatory and able to do light work. HR=hazard ratio (nCRT plus surgery vs surgery alone). aHR=adjusted hazard ratio.

Table 2: Univariable and multivariable HRs for all-cause mortality, according to subgroup characteristics

58% (51–65) at 3 years, and 47% (39–54) at 5 years in the surgery alone group (HR 0·64 [95% CI 0·49–0·82];
neoadjuvant chemoradiotherapy plus surgery group, figure 3A). Median progression-free survival for patients
compared with 70% (63–76), 50% (43–57), 44% (37–51), with squamous cell carcinomas was 74·7 months
and 33% (26–40), respectively, in the surgery alone group (95% CI 55·1–94·4) in the neoadjuvant chemoradiotherapy
(HR 0·57 [95% CI 0·37–0·88] at 1 year, 0·59 [0·43–0·82] plus surgery group and 11·6 months (4·4–18·8) in the
at 2 years, 0·65 [0·49–0·88] at 3 years, and 0·67 [0·51–0·87] surgery alone group (HR 0·48 [95% CI 0·28–0·82];
at 5 years). During follow-up, 16 patients died from figure 3B). Median progression-free survival for patients
treatment-related causes (ie, during neoadjuvant chemo- with adenocarcinomas was 29·9 months (95% CI
radiotherapy or during postoperative hospital stay), of 15·9–43·9) in the neoadjuvant chemoradiotherapy plus
whom nine were in the neoadjuvant chemoradiotherapy surgery group and 17·7 months (11·9–23·5) in the surgery
plus surgery group and seven in the surgery alone group. alone group (HR 0·69 [95% CI 0·52–0·92]; figure 3B).
23 patients died from non-disease-related causes beyond Progression-free survival in the neoadjuvant chemo-
the first 90 days postoperatively (13 in the neoadjuvant radiotherapy plus surgery group was 71% (95% CI 65–78)
chemoradiotherapy plus surgery group and ten in the at 1 year, 60% (52–67) at 2 years, 51% (43–58) at 3 years,
surgery alone group). and 44% (37–52) at 5 years, compared with 54% (47–61),
The estimated number of patients who need to be treated 41% (34–48), 35% (28–42), and 27% (21–33), respectively,
to prevent one additional death at 5 years was 7·1 (95% CI in the surgery alone group (HR 0·55 [95% CI 0·39–0·77]
4·6–13·2).13 The overall survival benefit of neoadjuvant at 1 year, 0·57 [0·42–0·77] at 2 years, 0·62 [0·47–0·82] at
chemoradiotherapy plus surgery was generally confirmed 3 years, and 0·61 [0·47–0·78] at 5 years).
across subgroups (table 2). The concordance of the The estimated number of patients who need to be treated
multivariable model for overall survival in all patients was to prevent one additional disease progression at 5 years
0·584.15 The proportionality of hazards assumption for the was 6·1 (95% CI 4·2–10·0).13 The progression-free survival
main analysis was not violated (χ²=0·77, p=0·38).16 benefit of neoadjuvant chemoradiotherapy plus surgery
See Online for appendix Of the 366 analysed patients, 116 were alive and disease was generally confirmed across subgroups (appendix p 3).
free (eventually without evidence of recurrent disease) at We studied the progression-free intervals, in addition
final analysis: 69 (39%) of 178 patients in the neoadjuvant to progression-free survival, to focus in more detail on
chemoradiotherapy plus surgery group and 47 (25%) of recurrence patterns in both treatment groups. From
188 patients in the surgery alone group. Median randomisation, 211 patients showed disease progression
progression-free survival was 37·7 months (table 3). In the neoadjuvant chemoradiotherapy plus
(95% CI 23·7–51·8) in the neoadjuvant chemoradiotherapy surgery group, 87 patients had disease progression, of
plus surgery group and 16·2 months (10·7–21·7) in the whom 39 had locoregional progression and 70 had

6 www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6

 Articles

distant progression (22 patients had both locoregional

A
and distant progression). In the surgery alone group, 100 Neoadjuvant chemoradiotherapy
124 patients had disease progression, of whom 72 had Surgery alone
locoregional progression and 90 had distant progression 90
(38 patients had both). Disease progression during
treatment (causing adjustment from curative to palliative 80
treatment intent) occurred in 17 patients in the
neoadjuvant chemoradiotherapy plus surgery group and 70
in 26 patients in the surgery alone group.

Progression-free survival (%)

Compared with patients in the surgery alone group, 60
those in the neoadjuvant chemoradiotherapy plus surgery
group had significantly less locoregional progression and 50
significantly less distant progression (table 3). The
reduction in locoregional progression was already 40
apparent during the first 6 months of follow-up and
remained significant after the first 24 months of follow- 30
up (appendix p 5). This finding indicates that the effect of
reduction in locoregional progression continued for an 20
extended period after randomisation. The reduction in
distant progression was also already recorded during the 10
first 6 months of follow-up and remained significant Log-rank p=0·000217
during the first 24 months of follow-up but not thereafter 0
0 12 24 36 48 60 72 84
(appendix p 5), which suggests that the reduction in
Number at risk
distant progression mainly occurred within the first Neoadjuvant chemo- 178 127 106 90 82 74 31 14
24 months after randomisation. radiotherapy plus surgery
Surgery alone 188 102 77 66 55 50 16 6
Total 366 229 183 156 137 124 47 20
Discussion
These long-term results, after a median follow-up for B
surviving patients of 84 months, confirm the initially 100 SCC, neoadjuvant chemoradiotherapy plus surgery
AC, neoadjuvant chemoradiotherapy plus surgery
reported survival benefit for neoadjuvant chemoradio- SCC, surgery alone
therapy plus surgery compared with surgery alone. The 90 AC, surgery alone
improvement in distant disease control occurred within
80
the first 2 years after initiation of treatment, whereas the
improvement in locoregional control continued for a
70
longer period. These findings further support the clinical
Progression-free survival (%)

value of this multimodality treatment strategy.

60
The overall survival benefit and the progression-free
survival benefit were confirmed for both histological
50
subtypes of oesophagal or oesophagogastric junctional
cancer and for other clinically relevant subgroups.
40
Although univariable and multivariable hazard ratios for
individual subgroups were reported for informative
30
purposes, no significant interactions in treatment effect
were identified for any of the subgroups, which means
20
that differences in treatment effect between subgroups
could well have arisen by chance, and the overall 10
treatment effect should be regarded as valid for all SCC: log-rank p=0·006
AC: log-rank p=0·010
considered subgroups. In other words, no clear evidence 0
exists to support the assumption that the adjusted overall 0 12 24 36 48 60 72 84
treatment effect of neoadjuvant chemoradiotherapy does Follow-up (months)
not also apply to patients with adenocarcinoma. We
Number at risk
SCC, neoadjuvant chemo- 41 28 28 25 25 24 12 5
Figure 3: Progression-free survival radiotherapy plus surgery
(A) By treatment group. (B) By treatment group and histological tumour type; SCC, surgery alone 43 31 15 14 13 12 3 1
AC, neoadjuvant chemo- 134 97 76 63 57 50 19 9
four patients in the surgery alone group and three in the neoadjuvant
radiotherapy plus surgery
chemoradiotherapy plus surgery group were excluded from this analysis because AC, surgery alone 141 79 61 51 41 38 13 5
histological tumour type could not be ascertained. SCC=squamous cell Total 359 225 180 153 136 124 47 20
carcinoma. AC=adenocarcinoma.

www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6 7

 Articles

early reduction in distant disease progression, without

Neoadjuvant Surgery alone HR (95% CI) p value
chemoradiotherapy plus (n=188) evidence of a reduction beyond the first 24 months,
surgery (n=178) supports a direct systemic effect of this neoadjuvant
Locoregional progression 39 (22%) 72 (38%) 0·45 (0·30–0·66) <0·0001 chemotherapy regimen. The reduction in distant disease
Distant progression 70 (39%) 90 (48%) 0·63 (0·46–0·87) 0·0040 progression achieved by this neoadjuvant chemoradio-
Overall progression 87 (49%) 124 (66%) 0·58 (0·44–0·76) <0·0001 therapy regimen is similar to reductions achieved with
more protracted (and more toxic) perioperative
Data are n (%), unless otherwise indicated. Comparison between treatment groups was based on univariable cause- chemotherapy regimens.4,5
specific Cox regression modelling of progression-free intervals. Deaths from non-disease-related causes were censored.
Overall progression was defined as either locoregional progression or distant progression. Patients with both
Results from this trial might not be readily extrapolated
locoregional disease progression and distant disease progression (22 patients in the neoadjuvant chemoradiotherapy to patients with poorer performance status, older
plus surgery group and 38 in the surgery alone group) were counted in both locoregional progression and distant patients, or those with tumours located in the proximal
progression categories. HR=hazard ratio.
or middle oesophagus, because of the relative scarcity of
Table 3: Patients with locoregional or distant progression in the two treatment groups patients in these categories. The value of this treatment
regimen will need to be confirmed for these patients in
future follow-up studies.
therefore conclude that both patients with squamous Despite recent advances in curative treatment of
cell carcinoma and those with adenocarcinoma benefit oesophageal or junctional cancers, the benefit of (neo)
significantly from the CROSS multimodality treatment adjuvant treatment is generally quite limited and a
regimen. definitive statement on the optimum perioperative
The addition of neoadjuvant chemoradiotherapy to treatment in terms of survival is still absent. A recent
primary surgery significantly improved locoregional meta-analysis suggested a (non-significant) advantage
disease control. Notably, the largest reported trials with of neoadjuvant chemoradiotherapy over neoadjuvant
neoadjuvant chemotherapy only showed limited chemotherapy alone in both a direct comparison
improvement in rates of R0 resection, pathologically (HR 0·77 [95% CI 0·53–1·12]) and in an indirect
complete response, and locoregional recurrence. comparison (0·88 [0·76–1·01]).7 The ongoing Japanese
Furthermore, two small randomised trials17,18 comparing randomised NExT trial (JCOG1109)21 and the Irish
neoadjuvant chemotherapy plus surgery to neoadjuvant randomised Neo-AEGIS trial (ICORG 10-14)22 will
chemoradiotherapy plus surgery both reported similar hopefully provide more definitive evidence on the best
R0 resection rates between treatment groups, but possible perioperative treatment for squamous cell
significantly higher pathologically complete response carcinoma and adenocarcinoma, respectively. Unless
rates and lower locoregional recurrence rates in the convincing results to the contrary become available,
neoadjuvant chemoradiotherapy plus surgery groups. strong evidence from the CROSS trial continues to
Therefore, the results from these trials point towards support neoadjuvant chemoradiotherapy as a standard
neoadjuvant radiotherapy combined with sensitising of care for both squamous cell carcinoma and
chemotherapy rather than neoadjuvant chemotherapy adenocarcinoma of the oesophagus or oesophagogastric
alone as the likely cause of improved locoregional junction.
control, as achieved in the CROSS trial. In conclusion, chemoradiotherapy according to the
In the CROSS trial, not only locoregional control, but CROSS regimen improves long-term overall and
also distant disease control improved significantly in the progression-free survival in patients with oesophageal and
neoadjuvant chemoradiotherapy plus surgery group. junctional cancer. This improvement is statistically
Theoretically, several potential explanations exist for this significant and clinically relevant for both squamous cell
improved distant disease control. First, if fewer loco- carcinoma and adenocarcinoma subtypes. Neoadjuvant
regional recurrences occur, then possibly less distant chemoradiotherapy according to CROSS followed by
dissemination develops from these locoregional surgical resection should be viewed as a standard of care
recurrences. Second, effective treatment of the primary for patients with resectable locally advanced oesophageal
tumour in the presence of disseminated disease has or junctional cancer.
been reported to prolong survival in some cancer Contributors
types.19,20 Therefore, a mechanism by which improved JS, JJBvL, MCCMH, PvH, MIvBH, BPLW, HWMvL, GAPN, GAPH, JJB,
locoregional control might improve distant disease MAC, RJBB, ORCB, FJWtK, G-JMC, CJAP, JTMP, HMWV, EJSB, HvD,
MJCvdS, TR, KB, JCB, AHMP, CMvR, JGR, HWT, EWS, and AvdG
control could be merely control of the primary tumour designed the study, and collected and interpreted the data. JS and EWS
itself, thereby removing a presently unknown stimulus analysed the data. JS and JJBvL wrote the first draft of the report. All
for disseminated tumour outgrowth. A third explanation authors approved the final version of the report.
is that improved distant disease control could be caused Declaration of interests
by a direct systemic effect of chemotherapy. In the JJBvL has received grants from the Dutch Cancer Foundation (KWF
present study, we recorded a significant reduction in Kankerbestrijding) during the conduct of the study, and grants from the
Dutch Cancer Foundation (KWF Kankerbestrijding), the Coolsingel
distant disease progression already within the first Stichting, and the Erasmus MC/MRace fund, outside the submitted
6 months after randomisation (appendix p 5). Such an work. The other authors declare no competing interests.

8 www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6

 Articles

Acknowledgments 10 Sobin LH, Wittekind C, et al. TNM classification of malignant

This study was funded by the Dutch Cancer Foundation tumors, 6th edn. New York: Wiley-Liss, 2002.
(KWF Kankerbestrijding). We thank the data managers from the 11 Hudis CA, Barlow WE, Costantino JP, et al. Proposal for
Integraal Kankercentrum Nederland (IKNL) for their dedicated standardized definitions for efficacy end points in adjuvant breast
prospective data collection, and Caspar Looman and Daan Nieboer cancer trials: the STEEP system. J Clin Oncol 2007; 25: 2127–32.
from the Department of Public Health, Erasmus MC—University 12 Fumagalli D, Bedard PL, Nahleh Z, et al. A common language in
Medical Centre (Rotterdam, Netherlands) for their assistance with neoadjuvant breast cancer clinical trials: proposals for standard
statistical analysis and interpretation of the data. definitions and endpoints. Lancet Oncol 2012; 13: e240–48.
13 Altman DG, Andersen PK. Calculating the number needed to treat for
References trials where the outcome is time to an event. BMJ 1999; 319: 1492–5.
1 Hulscher JBF, van Sandick JW, de Boer AGEM, et al. Extended 14 Oken MM, Creech RH, Tormey DC, et al. Toxicity and response
transthoracic resection compared with limited transhiatal resection criteria of the Eastern Cooperative Oncology Group.
for adenocarcinoma of the esophagus. N Engl J Med 2002; Am J Clin Oncol 1982; 5: 649–55.
347: 1662–69.
15 Harrell FE Jr, Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating
2 Kelsen DP, Winter KA, Gunderson LL, et al. Long-term results of the yield of medical tests. JAMA 1982; 247: 2543–46.
RTOG trial 8911 (USA Intergroup 113): a random assignment trial
comparison of chemotherapy followed by surgery compared with 16 Grambsch PM, Therneau TM. Proportional hazards tests and
surgery alone for esophageal cancer. J Clin Oncol 2007; 25: 3719–25. diagnostics based on weighted residuals. Biometrika 1994; 81: 515–26.
3 Allum WH, Stenning SP, Bancewicz J, Clark PI, Langley RE. 17 Stahl M, Walz MK, Stuschke M, et al. Phase III comparison of
Long-term results of a randomized trial of surgery with or without preoperative chemotherapy compared with chemoradiotherapy in
preoperative chemotherapy in esophageal cancer. J Clin Oncol 2009; patients with locally advanced adenocarcinoma of the
27: 5062–67. esophagogastric junction. J Clin Oncol 2009; 27: 851–56.
4 Cunningham D, Allum WH, Stenning SP, et al. Perioperative 18 Burmeister BH, Thomas JM, Burmeister EA, et al. Is concurrent
chemotherapy versus surgery alone for resectable gastroesophageal radiation therapy required in patients receiving preoperative
cancer. N Engl J Med 2006; 355: 11–20. chemotherapy for adenocarcinoma of the oesophagus? A
randomised phase II trial. Eur J Cancer 2011; 47: 354–60.
5 Ychou M, Boige V, Pignon JP, et al. Perioperative chemotherapy
compared with surgery alone for resectable gastroesophageal 19 Flanigan RC, Salmon SE, Blumenstein BA, et al. Nephrectomy
adenocarcinoma: an FNCLCC and FFCD multicenter phase III followed by interferon alfa-2b compared with interferon alfa-2b alone
trial. J Clin Oncol 2011; 29: 1715–21. for metastatic renal-cell cancer. N Engl J Med 2001; 345: 1655–59.
6 Gebski V, Burmeister B, Smithers BM, et al. Survival benefits from 20 Rapiti E, Verkooijen HM, Vlastos G, et al. Complete excision of
neoadjuvant chemoradiotherapy or chemotherapy in oesophageal primary breast tumor improves survival of patients with metastatic
carcinoma: a meta-analysis. Lancet Oncol 2007; 8: 226–34. breast cancer at diagnosis. J Clin Oncol 2006; 24: 2743–49.
7 Sjoquist KM, Burmeister BH, Smithers BM, et al. Survival after 21 Nakamura K, Kato K, Igaki H, et al. Three-arm phase III trial
neoadjuvant chemotherapy or chemoradiotherapy for resectable comparing cisplatin plus 5-FU (CF) versus docetaxel, cisplatin plus
oesophageal carcinoma: an updated meta-analysis. Lancet Oncol 2011; 5-FU (DCF) versus radiotherapy with CF (CF-RT) as preoperative
12: 681–92. therapy for locally advanced esophageal cancer (JCOG1109, NExT
study). Jpn J Clin Oncol 2013; 43: 752–55.
8 van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative
chemoradiotherapy for esophageal or junctional cancer. 22 Keegan N, Keane F, Cuffe S, Cunningham M, Ravi N, Lee G.
N Engl J Med 2012; 366: 2074–84. ICORG 10-14: Neo-AEGIS: a randomized clinical trial of
neoadjuvant and adjuvant chemotherapy (modified MAGIC
9 van Heijl M, van Lanschot JJB, Koppert LB, et al. Neoadjuvant regimen) versus neoadjuvant chemoradiation (CROSS protocol) in
chemoradiation followed by surgery versus surgery alone for adenocarcinoma of the esophagus and esophagogastric junction.
patients with adenocarcinoma or squamous cell carcinoma of the Proc Am Soc Clin Oncol 2014; 32 (5 suppl): abstr TPS4145.
esophagus (CROSS). BMC Surg 2008; 8: 21.

www.thelancet.com/oncology Published online August 6, 2015 http://dx.doi.org/10.1016/S1470-2045(15)00040-6 9