Japanese Journal of Clinical Oncology, 2015, 45(11) 1023–1028
doi: 10.1093/jjco/hyv119
Advance Access Publication Date: 5 August 2015
Original Article
Original Article
Systematic review and network meta-analysis:
neoadjuvant chemoradiotherapy for locoregional
esophageal cancer
Ta-Chen Huang1,2, Chih-Hung Hsu1,2, Chia-Chi Lin1,2, and Yu-Kang Tu3,*
1
Department of Oncology, National Taiwan University Hospital, Taipei City, 2Graduate Institute of Oncology, National
Taiwan University, College of Medicine, Taipei City, and 3Institute of Epidemiology and Preventive Medicine, College
of Public Health, National Taiwan University, Taipei City, Taiwan
*For reprints and all correspondence: Yu-Kang Tu, Institute of Epidemiology and Preventive Medicine, College of Public
Health, National Taiwan University, 5F, No 17, Xuzhou Road, Taipei City 100, Taiwan. E-mail:[email protected]
Received 10 March 2015; Accepted 7 July 2015
Abstract
Objective: Neoadjuvant chemoradiotherapy improves survival in patients with locoregional
esophageal cancer. This study compares the efficacy of two common regimens, paclitaxel plus platinum and platinum plus 5-fluorouracil, based on overall survival.
Methods: We performed a systematic review and network meta-analysis of randomized trials comparing paclitaxel plus platinum-neoadjuvant chemoradiotherapy or and platinum plus 5-fluorouracilneoadjuvant chemoradiotherapy with surgery alone. The outcome was the hazard ratios for death in
the entire population and the two major histologic subgroups, squamous cell carcinoma and adenocarcinoma.
Results: Ten clinical trials were included. Compared with surgery alone, the hazard ratios [95% credible interval (CrI)] in the entire, squamous cell carcinoma, and adenocarcinoma population were 0.63
(0.50–0.80), 0.50 (0.36–0.71) and 0.74 (0.54–1.01) for paclitaxel plus platinum, and 0.79 (0.68–0.92),
0.82 (0.67–1.01) and 0.81 (0.63–1.05) for platinum plus 5-fluorouracil, respectively. When paclitaxel
plus platinum was compared with platinum plus 5-fluorouracil, the hazard ratios (95% CrI) in the entire, squamous cell carcinoma, and adenocarcinoma population were 0.80 (0.60–1.06), 0.61 (0.41–
0.91) and 0.91 (0.61–1.36), respectively. The probability of paclitaxel plus platinum being ranked
the optimal treatment for the entire, squamous cell carcinoma, and adenocarcinoma population
was 94.2, 99.1 and 67.6%, respectively.
Conclusions: Neoadjuvant chemoradiotherapy with paclitaxel plus platinum regimen seemed to be
a better treatment than platinum plus 5-fluorouracil regimen for locoregional esophageal cancer,
especially for squamous cell carcinoma.
Key words: esophageal cancer, paclitaxel, chemoradiation, neoadjuvant therapy
Introduction
Esophageal cancer is the sixth leading cause of cancer death worldwide (http://globocan.iarc.fr/Pages/fact_sheets_cancer.aspx). The mortality rate is nearly as high as the incidence rate in most regions of the
world, especially in endemic areas such as China, the Middle East,
Eastern Africa and regions of South America. More than half of the
esophageal cancer cases reported worldwide have been locally advanced at diagnosis. The main treatments for locally advanced
esophageal cancer include surgery, definitive concurrent chemoradiotherapy (CRT) and neoadjuvant CRT or chemotherapy followed
© The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]
1023
1024
Neoadjuvant CRT for esophageal cancer
by surgery. The preferred treatment typically depends on the facilities
available and the expertise of the medical team.
Since the 1980s, several randomized clinical trials have been conducted to assess the efficacy of neoadjuvant CRT in treating esophageal cancer. However, most of these trials do not possess sufficient
statistical power to yield a definitive conclusion. A recent metaanalysis (1) examined 13 randomized clinical trials of neoadjuvant
CRT and revealed that the hazard ratios (HRs) for death were 0.78
[95% confidence interval (CI): 0.70–0.88], 0.80 (95% CI: 0.63–
0.93), and 0.75 (0.59–0.95) for the entire population, squamous cell
carcinoma (SCC) population, and adenocarcinoma (AC) population,
respectively, when neoadjuvant CRT was compared with surgery
alone. Therefore, neoadjuvant CRT followed by surgery is generally
accepted as a standard treatment for locoregional esophageal cancer.
All the clinical trials included in the meta-analysis conducted by
Sjoquist et al. involved the use of platinum-based regimens, primarily
cisplatin in conjunction with 5-fluorouracil (PF), and only two trials
involved the use of paclitaxel and platinum (TP). One TP study (2)
in which SCC patients were enrolled exclusively, indicated the significant benefit of neoadjuvant CRT; specifically, the HR was 0.55 (95%
CI: 0.36–0.84), which is substantially lower than those reported by
other studies. The HR reported in another TP study, the CROSS
study (3), was 0.67 (95% CI: 0.50–0.88), which is also lower than
that obtained using most other cisplatin-based regimens (Table 1).
In CROSS study, the HR in the SCC subgroup was only 0.42 (95%
CI: 0.24–0.79). These data suggest that various neoadjuvant CRT
regimens might engender distinct treatment benefits in subgroups of
esophageal cancer patients. Therefore, the purpose of this study was
to systematically review the efficacy of the two major neoadjuvant
CRT regimens, PF and TP, and compare it with that of surgery
alone in treating locoregional esophageal cancer.
Patients and methods
Search strategy and selection criteria
We followed the Preferred Reporting Items for Systematic Reviews and
Meta-Analyses guidelines (http://www.prisma-statement.org/) for performing a systematic review. We searched PubMed, Embase, ClinicalTrials.gov, the abstract database of annual meetings held by the
American Society of Clinical Oncology, the Cochrane Collaboration’s
Central Register of Controlled Clinical Trials and Cochrane Systematic
Reviews for randomized controlled trials in which neoadjuvant CRT
has been used to treat locoregional esophageal cancer and related reviews and meta-analyses conducted until the end of December 2013
without language restrictions (Supplementary S1, online version).
We also searched the references of published systematic reviews on
the treatment of esophageal cancers.
In our meta-analysis, the target population was locoregional
esophageal cancer patients who received either neoadjuvant CRT followed by surgery or surgery alone. We compared the overall survival
among patients who received TP-neoadjuvant CRT followed by surgery, PF-neoadjuvant CRT followed by surgery and surgery alone.
Table 1. Summary of randomized controlled trials of neoadjuvant CRT followed by surgery versus surgery alone in esophageal cancer
First author,
published year
Enrollment
year
Protocol: chemotherapy/radiotherapy
Patient number
(CRT+S:S)
Clinical N0 (%)
Adjusted HRs of overall
survival (95% CI)
Lv, 2010
1997–2004
SCC: 80:80
44.4
SCC: 0.55 (0.36–0.84)
Hagen, 2012
2004–08
SCC: 41:43
AC: 134:141
32.0
SCC: 0.42 (0.23–0.79)
AC: 0.74 (0.54–1.02)
Apinop, 1994
1986–92
SCC: 35:34
na
SCC: 0.80 (0.48–1.34)
Le Prise, 1994
1988–na
SCC: 41:45
na
SCC: 0.85 (0.50–1.46)
Walsh, 1995
1990–na
SCC: 29:32
na
SCC: 0.74 (0.46–1.18)
Walsh, 1996
1990–95
AC: 58:55
na
AC: 0.58 (0.38–0.88)
Burmeister, 2005
1994–2000
SCC: 45:50
AC: 80:78
84.8
SCC: 0.69 (0.42–1.15)
AC: 0.94 (0.66–1.34)
Lee, 2004
1999–na
SCC: 51:50
35.6
SCC: 0.88 (0.48∼1.62)
Tepper, 2008
1997–2000
SCC: 7:7
AC: 23:19
75
Mariette, 2014
2000–09
Paclitaxel 135 mg/m2 on Days 1 and 22
Cisplatin 20 mg/m2 on Days 1–3 and 22–25
40 Gy/20 fx, concurrent
Paclitaxel 50 mg/m2 weekly for 5 weeks
Carboplatin AUC = 2 weekly for 5 weeks
41.4 Gy/23 fx, concurrent
Cisplatin 100 mg/m2 Days 1 and 22
5-FU 1000 mg/m2 Days 1–4 and 22–25
40 Gy/20 fx, concurrent
Cisplatin 100 mg/m2 Days 1 and 22
5-FU 600 mg/m2 Days 1–4 and 22–25
20 Gy/10 fx, sequential
Cisplatin 75 mg/m2 on Day 7 for two cycles
5-FU 15 mg/kg on Days 1–5 for two cycles
40 Gy/15 fx, concurrent
Cisplatin 75 mg/m2 on Day 7 for two cycles
5-FU 15 mg/kg on Days 1–5 for two cycles
40 Gy/15 fx, concurrent
Cisplatin 80 mg/m2 on Day 1 for one cycles
5-FU 800 mg/m2 on Days 2–5 for one cycle
35 Gy/15 fx, concurrent
Cisplatin 60 mg/m2 on Day 1 for two cycles
5-FU 1000 mg/m2 on Day 3–5 for two cycles
45.6 Gy/38 fx, 2fx/day, concurrent
Cisplatin 100 mg/m2 on Day 1 for two cycles
5-FU 1000 mg/m2 on Days 1–4 for two cycles
50.4 Gy/28 fx, 1.8 fx/day, concurrent
Cisplatin 75 mg/m2 on Day 1 for 2 cycles
5-FU 800 mg/m2 on Days 1–4 for two cycles
45 Gy/25 fx, concurrent
SCC: 67:70
AC: 30:27
72.3
0.35 (0.18∼0.68)
SCC: 0.98 (0.65∼1.49)
AC: 1.13 (0.59∼2.15)
na, not available; CRT+S, chemoradiotherapy followed by surgery; S, surgery; SCC, squamous cell carcinoma; AC, adenocarcinoma; Clinical N0(%), the
percentage of clinical stage of N0; 95% CI, 95% confidence interval; HR, hazard ratio.
Jpn J Clin Oncol, 2015, Vol. 45, No. 11
Studies that have involved the use of other chemotherapy regimens
were excluded.
Data extraction and assessment for risk of bias
Two investigators (T.-C.H. and C.-H.H.) independently reviewed the
full manuscripts of eligible studies that were publicly available. The information extracted included patient characteristics, enrollment year,
year of publication, treatment protocols, clinical nodal status before
treatment and overall survival. For reports on the same trial documented at various follow-up periods and reviews or meta-analyses, the
most recent information was used. Risks of bias in individual studies
were assessed independently by the same reviewers by using the
Cochrane risk of bias method (Supplementary Table S2, online version). Disagreement was resolved by conducting a joint review of
the manuscript to reach a consensus.
Data synthesis and analysis
The primary outcome in our analysis was overall survival. We used
adjusted HRs as the effect size measure in the network meta-analysis.
Figure 1. Literature search and selection.
1025
We first conducted a pooled analysis irrespective of the type of
esophageal cancer and subsequently undertook subgroup analyses
of 2 histological types of esophageal cancer: SCC and AC. The subgroup data were obtained from either the original articles or previous
meta-analyses.
We used the Bayesian network meta-analysis models, as described
by Lu and Ades (4–6). All Bayesian analyses were undertaken using
the software package WinBUGS, Version 1.4.3 (MRC Biostatistics
Unit, Cambridge, UK) with non-informative uniform and normal
prior distributions. As the number of studies involved in the network
was small, the variance of the random effect would be sensitive to the
specification of prior distribution, so we therefore chose to undertake
fixed effect analysis. Three sets of starting values were used to initiate
the model, yielding 30 000 iterations (10 000 per chain), and thereby
obtain the posterior distributions of model parameters with 5000
burn-ins and a thinning interval of 50 for each chain. The convergence
of iterations was assessed using the Gelman–Rubin–Brooks statistic.
In addition, we conducted traditional fixed effect pairwise metaanalyses that compared neoadjuvant CRT with surgery alone.
1026
Neoadjuvant CRT for esophageal cancer
Results
We identified 797 studies during the literature search process based on
a review of titles and abstracts (Fig. 1). After the initial screening, we
retrieved the full text of 15 potentially eligible articles for detailed assessment, and 10 studies (Table 1) were included in the final
meta-analysis. Of the 10 studies included, TP-neoadjuvant CRT was
compared with surgery alone in two studies, one of which examined
only SCC (2) and the other examined both SCC and AC (3).
PF-neoadjuvant CRT was compared with surgery alone in eight studies, four of which examined SCC only (7–10), one of which investigated AC only (11), and three of which examined both histologic
subgroups (12–14). CALGB9781 (13) was an early-terminated clinical trial with 56 enrolled patients, provided no histologic subgroup
data; therefore, it was excluded from the subgroup analysis. A total
of 1452 patients were randomized in the network. Seven hundred
ninety-three patients were included in the subgroup analysis of SCC,
with 244 from TP studies and 549 from PF studies. A total of 603 patients were included in the subgroup analysis of AC, with 275 from a
TP study and 328 from PF studies.
van Hagen et al. (3) published their study in 2012 and included
histologic subgroup data. Mariette et al. provided their histologic subgroup data for our meta-analysis before publishing. The overall HRs
published in the most recent meta-analysis (1) were the most current
and, therefore, were adopted for assessing the other eight studies.
We summarized the effect sizes of the Bayesian fixed effects
models by using forest plots (Fig. 2). Compared with surgery alone,
TP- and PF-neoadjuvant CRT improved overall survival based on
the results of pooled analysis [HR = 0.63, 95% credible interval
(CrI): 0.50–0.80, and HR = 0.79, 95% CrI: 0.68–0.92, respectively].
In the subgroup analyses, TP- and PF-neoadjuvant CRT improved
overall survival in SCC (HR = 0.50, 95% CrI: 0.36–0.71, and HR =
0.82, 95% CrI: 0.67–1.01, respectively) and AC patients (HR = 0.74,
95% CrI: 0.54–1.01, and HR = 0.81, 95% CrI: 0.63–1.05, respectively). Because no direct comparisons between TP- and PF-neoadjuvant
CRT were performed, no concerns regarding inconsistency existed.
In the indirect comparison conducted using Bayesian network
meta-analysis, TP-neoadjuvant CRT resulted in superior overall survival compared with PF-neoadjuvant CRT in the entire population
(HR = 0.80, 95% CrI: 0.60–1.06), SCC population (HR = 0.61,
95% CrI: 0.41–0.91) and AC population (HR = 0.91, 95% CrI:
0.61–1.36). Regarding treatment ranking, the probability of
TP-neoadjuvant CRT being ranked the optimal treatment among
the 3 treatments was 94.2, 99.1 and 67.6% for the entire population,
SCC population and AC population, respectively (Fig. 3).
Figure 2. Pooled hazard ratios for death in the whole population and the subgroups. The subtotal line shows the number of patients in the subgroup, the hazard ratio
(HR) with 95% confidence interval (CI), I-squared (I 2) and P-value of the heterogeneity test by the traditional fixed effect meta-analysis for the pair of treatments in the
subgroup. Network metaanalysis shows the HR and 95% credible interval (CrI) from the network meta-analysis.
Jpn J Clin Oncol, 2015, Vol. 45, No. 11
Figure 3. The probability to be the best treatment for TP-CRT, PF-CRT and
surgery alone, in the whole study population, squamous cell carcinoma
subgroup, and adenocarcinoma subgroup. Y-axis is the likelihood. SCC,
squamous cell carcinoma; AC, adenocarcinoma; TP-CRT, paclitaxel and
platinum-based concurrent chemoradiotherapy; PF-CRT, platinum and 5fluorouracil-based concurrent chemoradiotherapy.
Table 2. Summary of the starting year, staging system and the
technique of radiotherapy of the enrolled trials
First author
published year
Starting
year
Staging
systemsa
Radiotherapy
Surgeryb
Lv, 2010
Hagen, 2012
Apinop, 1994
Le Prise, 1994
Walsh, 1995
Walsh, 1996
Burmeister, 2005
Lee, 2004
Tepper, 2008
Mariette, 2014
1997
2004
1986
1988
1990
1990
1994
1999
1997
2000
C
C
A
A
B
B
C
C
C
C
2D
3D
2D
2D
2D→3D
2D→3D
2D
2D
2D
3D
B
A
na
na
na
A
A
B
B
B
2D, two dimensional; 3D, three dimensional; na, not available.
a
Staging system: A, AJCC version 1 or 2; B, AJCC version 3, 4, 5 or 6 without
routine CT scan; C, version 3, 4, 5 or 6 with routine CT scan.
b
Surgery: A, transhiatal and transthoracic; B, exclusive transthoracic.
1027
TP-neoadjuvant CRT seems to be a better treatment especially for
SCC, exhibiting a 99.1% likelihood of being superior to other
treatments.
Our analysis had several limitations. Significant heterogeneity between the TP and PF studies existed. The TP studies were performed
primarily between 1997 and 2008, whereas the PF studies were conducted from 1986 to 2009. Staging systems, the technology for staging, and radiotherapy and surgical procedures have advanced over
time (Table 2). In addition, in three studies conducted by Burmeister
et al. (12), Mariette et al. (14) and Tepper et al. (13), more patients
with negative regional lymph nodes were enrolled compared with
those included in other studies. We tried to use publication years as
a covariate to adjust for potential bias in our network meta-analyses
(Supplementary Table S3, online version), as trials on TP were
published much later than most trials on PF; however, results regarding to the relative efficacy of TP and PF remained unaffected.
Meta-regression conducted based on lymph node status is not feasible
because numerous early studies in which CT scans were not performed
routinely have involved missing data. Therefore, our meta-analysis
might not have sufficient power to demonstrate the influence of
these heterogeneities.
No prospective head-to-head studies comparing TP versus PF as
neoadjuvant CRT in locoregional esophageal cancer have been conducted so far. Several studies were reported in a recent Gastrointestinal
Cancer Symposium, suggesting that TP-CRT compared with PF-CRT
did not yield an improved efficacy in esophageal cancer patients (16–
18). However, these studies were limited by the retrospective study designs and the relatively small patient numbers.
Conclusion
Our analysis, based on an indirect comparison of two treatment regimens using network meta-analysis of published Phase III trials, demonstrates a potential superior survival benefit of TP over that of
PF in neoadjuvant CRT administered to treat locoregional esophageal
cancer, especially SCC, despite our inability to eliminate all potential
bias. A randomized Phase III clinical trial of neoadjuvant CRT, in
which TP and PF are compared, to treat locoregional esophageal
cancer, especially SCC, is warranted to verify the conclusion.
Supplementary data
Supplementary data are available at http://www.jjco.oxfordjournals.org.
Discussion
When the CROSS study (3) was published in 2012 and reported the
superior survival benefit of TP-neoadjuvant CRT over that achieved
using surgery alone, whether paclitaxel plus carboplatin should become the standard regimen in neoadjuvant CRT for treating locoregional esophageal cancer became a subject of debate (15). However,
no direct comparison between TP and PF in neoadjuvant CRT has
been conducted; consequently, the optimal neoadjuvant CRT regimen
has not been determined.
We conducted network meta-analysis to determine whether TP is
superior to the traditional regimen, PF, based on overall survival in the
treatment of locoregional esophageal cancer. We included clinical
trials in which either TP- or PF-neoadjuvant CRT followed by surgery
was compared with surgery alone. Our results suggested that TP
provides an overall survival advantage over PF, especially in the
SCC population. TP-neoadjuvant CRT reduces mortality by 40% in
SCC patients compared with that reduced using PF. Therefore,
Acknowledgements
The authors are grateful to Christophe Mariette for kindly providing updated
results and subgroup information from the FFCD 9901, which was published
after our analysis.
Conflict of interest statement
None declared.
References
1. Sjoquist KM, Burmeister BH, Smithers BM, et al. Survival after neoadjuvant
chemotherapy or chemoradiotherapy for resectable oesophageal carcinoma: an updated meta-analysis. Lancet Oncol 2011;12:681–92.
2. Lv J, Cao XF, Zhu B, Ji L, Tao L, Wang DD. Long-term efficacy of perioperative chemoradiotherapy on esophageal squamous cell carcinoma.
World J Gastroenterol 2010;16:1649–54.
1028
Neoadjuvant CRT for esophageal cancer
3. van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative
chemoradiotherapy for esophageal or junctional cancer. N Engl J Med
2012;366:2074–84.
4. Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med 2004;23:3105–24.
5. Ades AE, Sculpher M, Sutton A, et al. Bayesian methods for evidence synthesis in cost-effectiveness analysis. Pharmacoeconomics 2006;24:1–19.
6. Sutton A, Ades AE, Cooper N, Abrams K. Use of indirect and mixed treatment comparisons for technology assessment. Pharmacoeconomics
2008;26:753–67.
7. Apinop C, Puttisak P, Preecha N. A prospective study of combined therapy
in esophageal cancer. Hepatogastroenterology 1994;41:391–3.
8. le Prise E, Etienne PL, Meunier B, et al. A randomized study of chemotherapy, radiation therapy, and surgery versus surgery for localized squamous
cell carcinoma of the esophagus. Cancer 1994;73:1779–84.
9. Lee JL, Park SI, Kim SB, et al. A single institutional phase III trial of preoperative chemotherapy with hyperfractionation radiotherapy plus surgery
versus surgery alone for resectable esophageal squamous cell carcinoma.
Ann Oncol 2004;15:947–54.
10. Walsh TN. The role of multimodality therapy in improving survival: a prospective randomized trial. In: Predicting, defining and improving outcomes
for oesophageal carcinoma [MD thesis]. Dublin: Trinity College, University
of Dublin 1995;124–50.
11. Walsh TN, Noonan N, Hollywood D, Kelly A, Keeling N, Hennessy TP.
A comparison of multimodal therapy and surgery for esophageal adenocarcinoma. N Engl J Med 1996;335:462–7.
12. Burmeister BH, Smithers BM, Gebski V, et al. Surgery alone versus
chemoradiotherapy followed by surgery for resectable cancer of the
oesophagus: a randomised controlled phase III trial. Lancet Oncol
2005;6:659–68.
13. Tepper J, Krasna MJ, Niedzwiecki D, et al. Phase III trial of trimodality
therapy with cisplatin, fluorouracil, radiotherapy, and surgery compared
with surgery alone for esophageal cancer: CALGB 9781. J Clin Oncol
2008;26:1086–92.
14. Mariette C, Dahan L, Mornex F, et al. Surgery alone versus chemoradiotherapy followed by surgery for Stage I and II esophageal cancer: final
analysis of randomized controlled Phase III Trial FFCD 9901. J Clin
Oncol 2014;32:2416–22.
15. Adenis A, Mirabel X, Mariette C. Is preoperative chemoradiation with paclitaxel and carboplatin a new standard of treatment for esophageal cancer?
Int J Radiat Oncol Biol Phys 2013;86:16–7.
16. Honing J, Smit J, Muijs C, et al. A comparison of carboplatin with paclitaxel and cisplatinum with 5-fluorouracil in definitive chemoradiotherapy in
esophageal cancer patients. J Clin Oncol 2014;32(Suppl 3; abstr 104).
17. Schellenberg D, Peixoto R, Lee A, Lim HJ. A province-wide retrospective
comparison of pathologic response and progression-free survival in locally
advanced esophagus cancer patients treated with either cisplatin-5FU (CF)
or carboplatin-paclitaxel (CP) concurrent with radiation (XRT) with or
without surgery. J Clin Oncol 2014;32(Suppl 3; abstr 107).
18. Thomay AA, Su S, Friedant AJ, et al. Freedom from recurrence after induction cisplatin/5-FU/RT versus carboplatin/paclitaxel/RT in patients with
esophageal cancer. J Clin Oncol 2014;32(Suppl 3; abstr 126).