Vous êtes sur la page 1sur 12

Articles

Chemotherapy and radiotherapy in nasopharyngeal


carcinoma: an update of the MAC-NPC meta-analysis
Pierre Blanchard, Anne Lee, Sophie Marguet, Julie Leclercq, Wai Tong Ng, Jun Ma, Anthony T C Chan, Pei-Yu Huang, Ellen Benhamou, Guopei Zhu,
Daniel T T Chua, Yong Chen, Hai-Qiang Mai, Dora L W Kwong, Shie Lee Cheah, James Moon, Yuk Tung, Kwan-Hwa Chi, George Fountzilas,
Li Zhang, Edwin Pun Hui, Tai-Xiang Lu, Jean Bourhis, Jean Pierre Pignon, on behalf of the MAC-NPC Collaborative Group*

Summary
Background A previous individual patient data meta-analysis by the Meta-Analysis of Chemotherapy in Nasopharynx Lancet Oncol 2015; 16: 64555
Carcinoma (MAC-NPC) collaborative group to assess the addition of chemotherapy to radiotherapy showed that it Published Online
improves overall survival in nasopharyngeal carcinoma. This benet was restricted to patients receiving concomitant May 7, 2015
http://dx.doi.org/10.1016/
chemotherapy and radiotherapy. The aim of this study was to update the meta-analysis, include recent trials, and to
S1470-2045(15)70126-9
analyse separately the benet of concomitant plus adjuvant chemotherapy.
*Listed in the appendix
Gustave-Roussy, Villejuif,
Methods We searched PubMed, Web of Science, Cochrane Controlled Trials meta-register, ClinicalTrials.gov, and France (P Blanchard MD,
meeting proceedings to identify published or unpublished randomised trials assessing radiotherapy with or without S Marguet MSc, J Leclercq MSc,
chemotherapy in patients with non-metastatic nasopharyngeal carcinoma and obtained updated data for previously E Benhamou MD,
analysed studies. The primary endpoint of interest was overall survival. All trial results were combined and analysed J P Pignon MD); Centre for
Research in Epidemiology and
using a xed-eects model. The statistical analysis plan was pre-specied in a protocol. All data were analysed on an Population Health, INSERM
intention-to-treat basis. U1018, Paris-Saclay
University, Villejuif, France
(P Blanchard,
Findings We analysed data from 19 trials and 4806 patients. Median follow-up was 77 years (IQR 62119). We found
E Benhamou, J P Pignon);
that the addition of chemotherapy to radiotherapy signicantly improved overall survival (hazard ratio [HR] 079, 95% CI Pamela Youde Nethersole
073086, p<00001; absolute benet at 5 years 63%, 95% CI 3591). The interaction between treatment eect Eastern Hospital, Hong Kong,
(benet of chemotherapy) on overall survival and the timing of chemotherapy was signicant (p=001) in favour of China (Prof A Lee MD,
W T Ng MD); Sun Yat-sen
concomitant plus adjuvant chemotherapy (HR 065, 056076) and concomitant without adjuvant chemotherapy (080,
University Cancer Center,
070093) but not adjuvant chemotherapy alone (087, 068112) or induction chemotherapy alone (096, 080116). State Key Laboratory of
The benet of the addition of chemotherapy was consistent for all endpoints analysed (all p<00001): progression-free Oncology in South China,
survival (HR 075, 95% CI 069081), locoregional control (073, 064083), distant control (067, 059075), and Collaborative Innovation
Center for Cancer Medicine,
cancer mortality (076, 069084).
Guangzhou, PR China
(Prof J Ma MD, P-Y Huang MD,
Interpretation Our results conrm that the addition of concomitant chemotherapy to radiotherapy signicantly Y Chen MD); Partner State Key
improves survival in patients with locoregionally advanced nasopharyngeal carcinoma. To our knowledge, this is the Laboratory of Oncology in
South China, Sir YK Pao Centre
rst analysis that examines the eect of concomitant chemotherapy with and without adjuvant chemotherapy as
for Cancer, The Chinese
distinct groups. Further studies on the specic benets of adjuvant chemotherapy after concomitant chemoradiotherapy University of Hong Kong,
are needed. Hong Kong, China
(Prof A T C Chan MD,
E P Hui MD); Fudan University
Funding French Ministry of Health (Programme dactions intgres de recherche VADS), Ligue Nationale Contre le Shanghai Cancer Center,
Cancer, and Sano-Aventis. Shanghai, PR China
(G Zhu MD); Hong Kong
Introduction 1753 patients, and combined results from trials of Sanatorium & Hospital, Hong
Kong, China (D T T Chua MD);
Nasopharyngeal carcinoma is distinct from other head concomitant plus adjuvant chemotherapy and con- Sun Yat-sen University Cancer
and neck carcinomas; it has a specic geographical comitant chemotherapy alone. Since those publications, Center, Guangzhou, PR China
distribution, is associated with the Epstein-Barr virus, additional trials have been done, including replications (H-Q Mai MD, Prof L Zhang MD,
has an aggressive natural locoregional history, and has a of the INT-0099 trial, allowing a study of the interaction Prof T-X Lu MD); Queen Mary
Hospital, Hong Kong, China
high risk of distant metastases.1 Nevertheless, high between the timing of chemotherapy and the eect on (D L W Kwong MD); National
proportions of patients are cured with standard therapy, various endpoints in more detail. The aim of this study Cancer Centre, Singapore,
even in cases of locoregionally advanced disease. was to update the meta-analysis, include recent trials, Singapore (S L Cheah MD);
Radiotherapy is the cornerstone of initial treatment due and to analyse separately the benet of concomitant plus SWOG Statistical Center,
Seattle, WA, USA
to the radiosensitive behaviour of nasopharyngeal adjuvant chemotherapy. (J Moon MSc); Tuen Mun
carcinoma and its deep-seated location. The landmark Hospital, Hong Kong, China
Intergroup 0099 (INT-0099) trial2 and the rst Meta- Methods (Y Tung MD); Shin Kong Wu
Analysis of Chemotherapy in Nasopharynx Carcinoma Selection criteria and search strategy Ho-Su, Memorial Hospital,
Tapei, Taiwan (K-H Chi MD);
(MAC-NPC)3 showed that there was an overall survival This updated meta-analysis was done according to a pre- Aristotle University of
benet related to concomitant chemotherapy. However, specied protocol. To be eligible, trials had to compare Thessaloniki School of
this meta-analysis included only eight trials and radiotherapy alone with radiotherapy plus chemotherapy, Medicine, Thessaloniki, Greece

www.thelancet.com/oncology Vol 16 June 2015 645


Articles

(Prof G Fountzilas MD); and


Centre Hospitalier Chemotherapy Control OE Variance HR (95% CI)
(number of deaths/ (number of deaths/
Universitaire Vaudois,
number entered) number entered)
Lausanne, Switzerland
(Prof J Bourhis MD) Induction
Correspondence to: PWH-8812 15/37 13/40 18 69
Dr Jean Pierre Pignon, Ligue AOCOA13 54/167 55/167 03 272
National Contre le Cancer
VUMCA-8914 94/171 93/168 02 467
plateforme de mta-analyse en
oncologie, Service de Japan-9115 17/40 20/40 25 92
Biostatistique et dEpidmiologie, NPC00823 12/34 14/31 28 63
Gustave-Roussy, HeCOG24 29/72 29/72 01 145
114 rue Edouard Vaillant, Subtotal 221/521 224/518 41 1109 096 (080116)
94805 Villejuif, France Adjuvant
jean-pierre.pignon@
TCOG-9416 52/80 53/78 13 262
gustaveroussy.fr
QMH-95Adj18 24/54 24/55 08 120
QMH-95Adj+18 19/57 25/56 46 109
See Online for appendix
Guangzhou 200619 26/251 34/257 36 150
Subtotal 121/442 136/446 86 641 087 (068112)
For the protocol see
Concomitant
http://bit.ly/1apnS9J
PWHQEH-9417 92/174 101/176 104 480
QMH-95Conc18 25/56 24/55 00 122
QMH-95Conc+18 19/57 24/54 49 106
VUMCA-95 (unpublished) 111/256 116/253 64 566
Guangzhou 200125 21/59 31/56 78 127
Guangzhou 20020226 73/204 81/204 22 385
Guangzhou 200327 9/116 26/114 94 87
Subtotal 350/922 403/912 410 1873 080 (070093)
Concomitant + adjuvant
INT-00992 59/97 79/96 228 330
SQNP0120 60/111 72/110 124 324
NPC-990121 69/172 93/176 128 405
NPC-9902CF22 22/51 18/42 03 99
NPC-9902AF22 15/44 29/52 76 109
Guangzhou 20020128 52/158 65/158 109 290
Subtotal 277/633 356/634 667 1557 065 (056076)

Total 969/2518 1119/2510 1205 5180 079 (073086)

Test for heterogeneity: p=0087 I=30% 015 275

Test for interaction: p=0012 Chemotherapy Control


Residual heterogeneity: p=036 better better
Chemotherapy eect: p<00001

Figure 1: Forest plots for overall survival with hazard ratios by timing of chemotherapy
Within each timing category, trials are ordered according to accrual period from the oldest to the most recent. The centre of each square is the HR for individual trial
comparison with the corresponding horizontal line showing the 95% CI. The size of the square is proportional to the number of deaths from the trial. The centre of
the open diamonds is the HR for dierent timings of chemotherapy and the extremities are the 95% CIs. The broken line and centre of black diamonds represent the
overall pooled HR with the extremities of the diamond showing the 95% CIs. PWH=Prince of Wales Hospital. AOCOA=Asian-Oceanian Clinical Oncology Association.
VUMCA=International Nasopharynx Cancer Study Group (cavum). NPC=nasopharyngeal carcinoma. HeCOG=Hellenic Cooperative Oncology Group. TCOG=Taiwan
Cooperative Oncology Group. QMH=Queen Mary Hospital (2 2 design, counted twice in the analysis). PWHQEH=Prince of Wales Hospital, Queen Elizabeth Hospital.
INT-0099=SWOG (Southwest Oncology Group)-coordinated Intergroup trial, also known as SWOG 8892. SQNP=Singapore Naso-Pharynx. CF=conventional
fractionation. AF=accelerated fractionation. O-E=observedexpected. HR=hazard ratio. Adj=radiotherapy versus radiotherapy plus adjuvant chemotherapy.
Adj+=radiotherapy plus concomitant chemotherapy versus radiotherapy plus concomitant plus adjuvant chemotherapy. Conc=radiotherapy versus radiotherapy plus
concomitant chemotherapy. Conc+=radiotherapy plus adjuvant chemotherapy versus radiotherapy plus adjuvant plus concomitant chemotherapy.

or to compare a treatment strategy with one carcinoma. Trials were eligible if at least 60 patients had
chemotherapy timing (ie, radiotherapy plus concomitant been included (30 patients per group for trials with more
chemotherapy, radiotherapy plus induction chemo- than two groups) and if all patients had undergone
therapy, or radiotherapy plus adjuvant chemotherapy) potentially curative locoregional treatment. Accrual had
with the same treatment strategy plus chemotherapy at to be completed before Dec 31, 2010.
another timing. They had to be randomised and include Both published and unpublished trials meeting these
patients with untreated non-metastatic nasopharyngeal criteria were eligible. We searched for trials in electronic

646 www.thelancet.com/oncology Vol 16 June 2015


Articles

Chemotherapy Control O-E Variance HR (95% CI)


(number of deaths/ (number of deaths/
number entered) number entered)

Induction
PWH-8812 17/37 17/40 01 84
AOCOA13 100/167 108/167 65 519
VUMCA-8914 102/171 114/168 150 534
Japan-9115 19/40 24/40 38 107
NPC00823 12/34 14/31 29 63
HeCOG24 31/72 34/72 25 162
Subtotal 281/521 311/518 309 1469 081 (069095)
Adjuvant
TCOG-9416 53/80 55/78 43 269
QMH-95Adj18 27/54 31/55 30 145
QMH-95Adj+18 25/57 28/56 26 132
Guangzhou 200619 41/251 57/257 78 245
Subtotal 146/442 171/446 177 790 080 (064100)
Concomitant
PWHQEH-9417 101/174 108/176 86 521
QMH-95Conc18 28/56 31/55 42 146
QMH-95Conc+18 25/57 27/54 33 128
VUMCA-95 (unpublished) 138/256 146/253 107 707
Guangzhou 200125 26/59 33/56 73 144
Guangzhou 20020226 90/204 100/204 40 475
Guangzhou 200327 15/116 32/114 99 117
Subtotal 423/922 477/912 480 2239 081 (071092)
Concomitant + adjuvant
INT-00992 61/97 81/96 277 328
SQNP0120 64/111 75/110 115 343
NPC-990121 77/172 104/176 179 451
NPC-9902CF22 23/51 21/42 09 109
NPC-9902AF22 19/44 31/52 76 124
Guangzhou 20020128 56/158 73/158 138 320
Subtotal 300/633 385/634 794 1674 062 (053072)

Total 1150/2518 1344/2510 1759 6172 075 (069081)

Test for heterogeneity: p=030 I=12% 02 20

Test for interaction: p=0041 Chemotherapy Control


Residual heterogeneity: p=036 better better
Chemotherapy eect: p<00001

Figure 2: Forest plots for progression-free survival with hazard ratios by timing of chemotherapy
Within each timing category, trials are ordered according to accrual period from the oldest to the most recent. The centre of each square is the HR for individual trial
comparison with the corresponding horizontal line showing the 95% CI. The size of the square is proportional to the number of relapses or deaths from the trial. The
centre of the open diamonds is the HR for dierent timings of chemotherapy and the extremities are the 95% CIs. The broken line and centre of black diamonds
represent the overall pooled HR with the extremities of the diamond showing the 95% CIs. PWH=Prince of Wales Hospital. AOCOA=Asian-Oceanian Clinical Oncology
Association. VUMCA=International Nasopharynx Cancer Study Group (cavum). NPC=nasopharyngeal carcinoma. HeCOG=Hellenic Cooperative Oncology Group.
TCOG=Taiwan Cooperative Oncology Group. QMH=Queen Mary Hospital (22 design, counted twice in the analysis). PWHQEH=Prince of Wales Hospital, Queen
Elizabeth Hospital. INT-0099=SWOG (Southwest Oncology Group)-coordinated Intergroup trial, also known as SWOG 8892. SQNP=Singapore Naso-Pharynx.
CF=conventional fractionation. AF=accelerated fractionation. O-E=observedexpected. HR=hazard ratio. Adj=radiotherapy versus radiotherapy plus adjuvant
chemotherapy. Adj+=radiotherapy plus concomitant chemotherapy versus radiotherapy plus concomitant plus adjuvant chemotherapy. Conc=radiotherapy versus
radiotherapy plus concomitant chemotherapy. Conc+=radiotherapy plus adjuvant chemotherapy versus radiotherapy plus adjuvant plus concomitant chemotherapy.

publication databases, trial registries, and meeting allocation, dates of failures and death, treatment details,
proceedings (details in appendix). and acute and late toxicities for each trial. Information was
updated for previously analysed trials whenever possible.
Individual patient data collection We checked the data according to a standard procedure,
We contacted the study investigators and requested and compared with the trial protocol and published
individual patient data for patient and tumour reports. Missing values and discrepancies were discussed
characteristics, date of randomisation and treatment group with the trialists. Randomisation validity was assessed by

www.thelancet.com/oncology Vol 16 June 2015 647


Articles

A B
100 Absolute dierence Absolute dierence
at 5 years (95% CI) 900% at 10 years (95% CI)
+25% (42 to 92) +77% (16 to 170)
754% 894%
80 Absolute dierence 757%
at 10 years (95% CI) 649%
738% 573% 11% (89 to 67) 724%
Overall survival (%)

60 Absolute dierence
480%
at 5 years (95% CI) 572%
548% +33% (38 to 104)
40 469%

20
Experimental
Control

0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Number of deaths/ Years 02 Years 35 Years 610 Years >10 Years 02 Years 35 Years 610 Years >10
person-years
Experimental 119/856 77/827 25/537 0/36 46/833 40/699 18/586 17/403
Control 129/856 78/766 17/516 0/49 44/855 52/704 25/525 15/348

C D
100 Absolute dierence Absolute dierence
870% at 10 years (95% CI) at 10 years (95% CI)
+82% (26 to 138) 866% +139% (79 to 199)
80 824% 704% 789% 705%
587%
570%
Overall survival (%)

60 651%
Absolute dierence Absolute dierence 581%
at 5 years (95% CI) 505% at 5 years (95% CI)
40 +53% (08 to 98) +124% (70 to 178)
431%

20

0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Time from randomisation (years) Time from randomisation (years)
Number of deaths/ Years 02 Years 35 Years 610 Years >10 Years 02 Years 35 Years 610 Years >10
person-years
Experimental 116/1714 144/1968 63/1790 27/662 84/1188 102/1460 67/1477 24/420
Control 157/1654 145/1841 81/1592 20/513 135/1151 129/1226 72/1171 20/281

Figure 3: Survival curves for overall survival in trials investigating (A) induction, (B) adjuvant, (C) concomitant, and (D) concomitant plus adjuvant chemotherapy

checking the patterns of treatment allocation and the locoregional failure and vice versa. If both a locoregional
balance of baseline characteristics between treatment failure and a distant failure occurred at the same time,
groups. Follow-up of patients was also compared between patients were considered as having an event for distant
treatment groups.4 Each trial was reanalysed and the failure only. Living patients without an event corresponding
analyses were sent to the trialists for validation. The to any endpoint were censored at the date of their last
quality of the additional trials was assessed and no major follow-up. Non-cancer deaths were dened as deaths from
bias was identied (appendix). known causes other than nasopharyngeal carcinoma for
patients without progression. Cancer deaths included
Outcomes deaths from any cause with previous progression and
The primary endpoint was overall survival, dened as the deaths from nasopharyngeal carcinoma or unknown
time from randomisation until death from any cause. The cause.
secondary endpoints were progression-free survival,
locoregional and distant failure, and cancer and non- Statistical analysis
cancer mortality. Progression-free survival was dened as We analysed all the data on an intention-to-treat basis. We
the time from randomisation to rst progression estimated median follow-up with the reverse Kaplan-Meier
(locoregional or distant) or death from any cause. Patients method.5 Analyses were stratied by trial. Individual and
with a distant failure as a rst event were censored for overall pooled hazard ratios (HRs) with 95% CIs were

648 www.thelancet.com/oncology Vol 16 June 2015


Articles

A B
100 Absolute dierence Absolute dierence
at 5 years (95% CI) at 5 years (95% CI)
+77% (13 to 141) +61% (06 to 128)
780%
80 719%
Progression-free survival (%)

749% 624%
601%
60 658%
469%
543%
512% 390%
40
392%
361%

20
Experimental
Control

0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Number of events/ Years 02 Years 35 Years 610 Years >10 Years 02 Years 35 Years 610 Years >10
person-years
Experimental 201/776 61/722 18/495 1/33 97/782 19/614 15/529 15/360
Control 251/712 53/561 7/422 0/42 111/770 29/583 17/445 14/297

C D
100 Absolute dierence Absolute dierence
at 5 years (95% CI) at 5 years (95% CI)
+66% (19 to 113) +124% (68 to 180)
777%
80 742%
Progression-free survival (%)

611% 622%
686%
60 518% 532%
630%
545%
498%
40 443%
385%

20

0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
Time from randomisation (years) Time from randomisation (years)
Number of events/ Years 02 Years 35 Years 610 Years >10 Years 02 Years 35 Years 610 Years >10
person-years
Experimental 234/1577 114/1728 53/1259 22/581 141/1125 96/1306 44/1323 19/388
Control 285/1492 118/1528 60/1371 14/446 237/1007 80/987 54/1014 14/241

Figure 4: Survival curves for progression-free survival in trials investigating (A) induction, (B) adjuvant, (C) concomitant, and (D) concomitant plus
adjuvant chemotherapy

calculated with a xed-eect model using the log-rank were estimated for control and experimental groups using
expected number of events and variance.6 A similar model annual death rates and HRs, and absolute benet at 5 years
was used to estimate odds ratios (ORs) for the comparison with 95% CI was calculated.8
of toxicity between groups, and incidences of toxicity in the We did subset analyses to study the interaction between
experimental group were calculated using the incidence in treatment eect and trial level characteristics, using a test
the control group and the OR.7 heterogeneity tests and of heterogeneity among the dierent groups of trials.
the I statistic were used to investigate the overall Residual heterogeneity within trial subgroups was
heterogeneity between trials.8,9 The use of a random-eects computed by subtracting the statistic of the
model was planned in the case of important and heterogeneity test between groups from the statistic of
unexplained heterogeneity. With 4500 patients it would be the overall heterogeneity test.10 Predened subsets were
possible to detect, with a power exceeding 90%, an absolute the timing of randomised chemotherapy (adjuvant [after
improvement in survival from 40% to 45% at 5 years (two- radiotherapy] vs induction [before radiotherapy] vs
sided log-rank test with an alpha of 5%). Cancer mortality concomitant [during radiotherapy] vs concomitant plus
was calculated indirectly by subtracting the log-rank adjuvant), chemotherapy drug, trial size, and method of
statistic for non-cancer mortality from the log-rank statistic randomisation. The interaction between treatment eect
for mortality from all causes.8 Stratied survival curves and patient subgroups (according to age, sex, performance

www.thelancet.com/oncology Vol 16 June 2015 649


Articles

Concomitant chemotherapy (seven comparisons Concomitant plus adjuvant chemotherapy (six


and 1834 patients) comparisons and 1267 patients)
Type of chemotherapy
Cisplatin Two comparisons (580 patients) All comparisons
Carboplatin or oxaliplatin Two comparisons (523 patients) None
Non-platin Three comparisons (731 patients) None
Other timing of chemotherapy
None Four comparisons (806 patients) All comparisons
Induction chemotherapy Two comparisons (917 patients) None
Adjuvant chemotherapy One comparison (111 patients) None
Sensitivity analysis excluding trials with another OS 071 (057089); PFS 072 (059089) NA
timing of chemotherapy in both groups*
Histology (WHO classication)
1 127 (7%) 49 (4%)
23 1701 (93%) 1218 (96%)
Missing 6 (<1%) 0 (0%)
Sensitivity analysis excluding patients with OS 081 (070094); PFS 081 (070093) OS 067 (057079); PFS 065 (055076)
WHO type 1 histology*
Overall stage
Stage II 359 (20%) 6 (<1%)
Stage III 784 (43%) 671 (53%)
Stage IV 691 (38%) 590 (47%)
Sensitivity analysis (unplanned) without OS 080 (069093); PFS 082 (071094) OS 065 (056077); PFS 062 (054073)
patients with stage II disease*
Length of follow-up
Median (range) 79 years (58141) 102 years (62168)
Overall survival in control group
At 2 years 825% 789%
At 5 years 645% 581%
At 10 years 495% 431%

Data are n (%) unless otherwise stated. NA=not applicable. OS=overall survival. PFS=progression-free survival. *Hazard ratio (95% CI). Compared with 080 (070093) in
the overall analysis. Compared with 081 (071092) in the overall analysis. Compared with 065 (056076) in the overall analysis. Compared with 062 (053072) in
the overall analysis.

Table 1: Main dierences between trials investigating concomitant chemotherapy and those investigating concomitant plus adjuvant chemotherapy

status, and overall stage) was estimated directly in a single Results


Cox model stratied by trial and containing treatment Eight trials (1753 patients)2,1218 were included in our
eect, covariate (eg, age) eect, and treatmentcovariate previous meta-analysis. This study includes 4806 patients
interaction (one-stage model method).11 Sensitivity from 19 trials (appendix), including one unpublished trial
analyses were done after the exclusion of trials including (VUMCA-95: International Nasopharynx Cancer Study
less than 100 patients, trials including two dierent Group, NCT00180973). Median follow-up was 77 years
chemotherapy timings where only one was randomised, (IQR 62119). Only two trials12,19 (585 patients) had a
trials with a median follow-up shorter than 5 years, median follow-up shorter than 5 years, and seven
outliers, and patients with WHO type 1 cancer. trials2,1618,2022 (1681 patients) had a median follow-up longer
All p values were two-sided. We used SAS software, than 10 years. Updated follow-up or additional data on
version 9.3. toxicity were obtained for all but three trials included in the
previous meta-analysis.12,13,15 One 2 2 design trial
Role of the funding source (222 patients)18 was counted twice (four comparisons,
The funding sources had no role in study design, data appendix) and another22 was split into two comparisons,
collection, data analysis, data interpretation, or writing of leading to a total of 23 comparisons and 5028 patients. Six
the report. The submission of the paper for publication comparisons (1039 patients)1215,23,24 investigated induction
was decided by the MAC-NPC Collaborative Group. PB, therapy, including one (77 patients)12 with the addition of
SM, JL, and JPP had access to the raw data. The adjuvant chemotherapy in the treatment group and two
corresponding author had full access to all of the data (209 patients)23,24 with the addition of concomitant chemo-
and the nal responsibility to submit for publication. therapy in both groups. Four comparisons (888 patients,

650 www.thelancet.com/oncology Vol 16 June 2015


Articles

one trial with two comparisons)16,18,19 investigated adjuvant A


chemotherapy, including two (621 patients)18,19 with 100 941% Absolute dierence
addition of concomitant chemotherapy in both groups. at 10 years (95% CI)
+99% (24 to 174)
Seven comparisons (1834 patients, one trial with two 791%
comparisons; including the VUMCA-95 trial),17,18,2527 80 858%

investigated concomitant chemotherapy, including one 653%


(111 patients)18 with addition of adjuvant chemotherapy in 697%

Overall survival (%)


60
both groups and two (917 patients; including the
Absolute dierence 554%
VUMCA-95 trial)26 with addition of induction chemo- at 5 years (95% CI)
therapy in both groups. Six comparisons (1267 patients, 40
+94% (33 to 155)
one trial with two comparisons)2,2022,28 investigated
concomitant plus adjuvant chemotherapy.
Most patients were male (3734 [74%] of 5020) and 20
Experimental
younger than 50 years of age (3149 [63%] of 5020), with a Control
performance status of 0 to 1 (4031 [98%] of 4113). 4493 (89%)
of 5028 had a stage III or IV cancer and 4759 (96%) of 0
0 1 2 3 4 5 6 7 8 9 10 11 12
4957 had a WHO histological type 2 or 3 cancer (appendix).
Overall survival and progression-free survival analyses Number of deaths/ Years 02 Years 35 Years 610 Years >10
person-years
were based on 2088 (42%) deaths and 2494 (50%) events, Experimental 24/791 60/1013 42/1128 21/509
respectively, out of 5028 patients. The causes of death Control 57/754 63/913 44/993 18/412
and the types of progression events are provided in the
B
appendix. The addition of chemotherapy to radiotherapy 100 Absolute dierence
improved overall survival (HR 079, 95% CI 073086; at 5 years (95% CI)
p<00001) and progression-free survival (HR 075, 837% +101% (34 to 168)
95% CI 069081; p<00001; gures 1 and 2). There was 80
699%
an absolute benet in overall survival at 5 years of 63%
Progression-free survival (%)

735%
(95% CI 3591). Low heterogeneity was observed 594%
among trials for overall survival (p=0087, I=30%). 60
598%
Heterogeneity was mainly attributable to the timing of
499%
chemotherapy (pinteraction=0012) and no heterogeneity
40
remained after taking this into account (p=036). The
interaction between chemotherapy timing and chemo-
therapy eect was also signicant for progression-free 20
survival (pinteraction=0041). Subset analyses showed that
trials investigating concomitant chemotherapy or
concomitant plus adjuvant chemotherapy were the only 0
subsets to show a signicant improvement in both 0 1 2 3 4 5 6 7 8 9 10 11 12
overall survival and progression-free survival. Survival Time from randomisation (years)
curves assessing the eect of chemotherapy timing on Number of events/ Years 02 Years 35 Years 610 Years >10
overall survival and progression-free survival are shown person-years
Experimental 66/748 55/906 32/1015 17/451
in gures 3 and 4. An unplanned sensitivity analysis Control 107/688 54/785 31/845 12/360
showed that the benet of the addition of chemotherapy
on overall survival was not dierent between the trials Figure 5: Survival curves for (A) overall survival and (B) progression-free survival in trials investigating
concomitant chemotherapy without addition of induction or adjuvant chemotherapy in both groups
included in the rst meta-analysis2,1218 (HR 083, 95% CI
073095) and the new trials included in this update1927,29
(including the VUMCA-95 trial) (HR 076, 95% CI group than in the concomitant plus adjuvant
067085; pinteraction=028). chemotherapy group, and there was a higher proportion
Table 1 summarises the main dierences between of patients with stage II disease in the concomitant
trials investigating concomitant chemotherapy and chemotherapy group. Exclusion of stage II patients in an
concomitant plus adjuvant chemotherapy. All unplanned sensitivity analysis led to results similar to
comparisons investigating concomitant plus adjuvant the main analysis (table 1). Three comparisons investi-
chemotherapy used cisplatin-based chemotherapy, but gating concomitant chemotherapy added induction
among those investigating concomitant chemotherapy, (including the VUMCA-95 trial)26 or adjuvant18 chemo-
two used cisplatin,17,27 two carboplatin26 or oxaliplatin,25 therapy in both groups (control groups were induction
and three non-platinum-based chemotherapy (including chemotherapy and radiotherapy, and radiotherapy plus
the VUMCA-95 trial).18 Overall survival in the control adjuvant chemotherapy, respectively). A planned sensi-
group was higher in the concomitant chemotherapy tivity analysis excluding these comparisons increased

www.thelancet.com/oncology Vol 16 June 2015 651


Articles

100 chemotherapy-related toxicity (HR without this trial 114,


Experimental cancer deaths
Control cancer deaths 95% CI 088148; p=033). This sensitivity analysis was
Experimental non-cancer deaths done to assess the robustness of the results because this
Control non-cancer deaths
80 trial was the only one to show a signicant excess of non-
Absolute dierence cancer death and used an outdated induction chemo-
at 5 years (95% CI) therapy regimen (bleomycin, epirubicin, cisplatin).
Risk of death (%)

60 +19% (05 to 33)


Absolute dierence Table 2 summarises the results for all endpoints.
at 5 years (95% CI) No interaction was observed between treatment eect
73% (103 to 43)
on overall survival and choice of chemotherapy drug
40 354%
(pinteraction=036), trial size (pinteraction=015), method of
randomisation (pinteraction=038), patient age (pinteraction=066),
281% patient sex (pinteraction=084), performance status
20 172%

130%
(pinteraction=028), or tumour stage (pinteraction=041) (appendix).
48%
27% Because of the variation in the risk of death in the control
15% 29%
0 group according to age and tumour stage, the 5-year
0 1 2 3 4 5 7
6 8 absolute benet of chemotherapy on overall survival was
Time from randomisation (years)
larger for older patients and patients with advanced stages.
Number of deaths/ Years 02 Years 35 Years >6 For progression-free survival, the p value for the trend
person-years
Experimental non-cancer deaths 55/3863 27/4031 66/4685
between treatment eect and increasing patient age was
Control non-cancer deaths 30/3789 14/3575 58/3927 010, with a HR for patients younger than 50 years of 072
Experimental cancer deaths 264 274 130 (95% CI 065080), of 081 (070095) for patients aged
Control cancer deaths 361 326 139
between 50 and 59 years, and 084 (070101) for patients
Figure 6: Survival curves for cancer and non-cancer deaths aged 60 years or older.
Sensitivity analyses (appendix) that excluded trials with
overall survival and progression-free survival for fewer than 100 patients, a median follow-up shorter than
concomitant chemotherapy (overall survival: HR 071, 5 years, patients with WHO type 1 cancer, and patients
95% CI 057089; progression-free survival: 072, with stage III disease (unplanned analysis) led to similar
059089; gure 5, appendix), of the same magnitude results from the main analysis. The dierence of
as the concomitant plus adjuvant group. Risk reduction treatment eect on overall survival between trials of pure
was higher with concomitant plus adjuvant chemo- concomitant chemotherapy or concomitant plus adjuvant
therapy, but the dierence with concomitant chemotherapy and the subsets of trials investigating
chemotherapy only was not statistically signicant induction chemotherapy and adjuvant chemotherapy
(overlap of the CIs; appendix). was enhanced (appendix). An interaction between
Out of 5028 patients, 910 (18%) had locoregional failure chemotherapy eect and chemotherapy timing for
and 1115 (22%) had distant failure. Chemotherapy overall survival and progression-free survival was not
reduced the risk of locoregional failure (HR 073, 95% CI signicant after exclusion of two trials2,27 (423 patients)
064083; p<00001) and distant failure (HR 067, that were outliers for overall survival (pinteraction=010) and
95% CI 059075; p<00001; appendix). Heterogeneity one trial2 (193 patients) that was a progression-free
among trials was observed for distant failure rate survival outlier (pinteraction=039).
(p=0024, I=41%), but not for locoregional failure rate Table 3 describes the acute and late severe toxicities
(p=048, I=0%). After excluding two outliers that were observed in these studies. Acute toxicities were
(306 patients),2,18 the heterogeneity was no longer consistent with those expected for the cisplatin-based
signicant (p=012, I=28%). No evidence of interaction chemotherapy regimens. Results for the interaction of
between chemotherapy timing and chemotherapy eect toxicity with the timing of chemotherapy are summarised
was observed for distant failure rate (pinteraction=018) or in the appendix. Concomitant plus adjuvant chemo-
locoregional failure rate (pinteraction=0054). Survival curves therapy was associated with the highest frequency of
are presented in the appendix. Cancer and non-cancer acute toxicities. Among the late toxicities, which were
mortality analyses were based on only 4312 patients mainly related to radiotherapy, only cranial nerve palsy
because the cause of death was missing for three and hearing decit were increased by chemotherapy.
trials.2,25,26 1494 (35%) cancer deaths and 250 (6%) non-
cancer deaths occurred (appendix). The use of Discussion
chemotherapy was associated with a decrease in cancer- This updated individual patient data meta-analysis of the
related deaths (HR 076, 95% CI 069084; p<00001) role of chemotherapy in nasopharyngeal carcinoma
(gure 6). Chemotherapy was not associated with an conrms the benets associated with the use of
increase in non-cancer mortality (HR 127, 95% CI chemotherapy in addition to radiotherapy, including
099164; p=0056), and this did not change after signicant and clinically relevant improvements in
excluding one trial14 (339 patients) that had a high level of overall survival and progression-free survival, and

652 www.thelancet.com/oncology Vol 16 June 2015


Articles

Overall survival Progression-free survival Locoregional control Distant control Cancer death* Non-cancer death*
Induction 096 (080116) 081 (069095) 084 (066107) 062 (048079) 089 (073109) 185 (105329)
Adjuvant 087 (068112) 080 (064100) 061 (041092) 080 (059109) 084 (064110) 108 (059195)
Concomitant 080 (070093) 081 (071092) 082 (067101) 074 (061090) 074 (062089) 120 (077187)
Concomitant plus adjuvant 065 (056076) 062 (053072) 054 (041071) 056 (045070) 063 (052077) 119 (077185)
Overall 079 (073086) 075 (069081) 073 (064083) 067 (059075) 076 (069084) 127 (099164)
Overall test p<00001 p<00001 p<00001 p<00001 p<00001 p=0056
Interaction test (timing treatment eect) p=0012 p=0041 p=0054 p=018 p=0084 p=055
Residual heterogeneity test p=036 p=062 p=078 p=0031 p=054 p=025

Data are HR (95% CI) or p value. *Analyses based on 20 comparisons (4312 patients) because the cause of death was missing for three trials.
2,25,26
No dierence (HR 091, 95% CI 039215) was seen in a sensitivity
analysis, excluding one trial (339 patients)14 that was a clear outlier.

Table 2: Summary of the results overall, and by chemotherapy timing, for all endpoints

Availability Incidence of toxicity OR (95% CI) Ecacy Heterogeneity p value for interaction with
p value chemotherapy timing
Number Number Chemo- Control I p value
of trials* of patients therapy
(comparisons)
Acute toxicity
Anaemia 15 (19) 4059 43% 15% 295 (211412) <00001 48% 0011 0011
Neutropenia 15 (19) 4028 257% 49% 671 (553814) <00001 84% <00001 00032
Mucositis 14 (18) 3870 406% 312% 151 (131173) <00001 40% 0042 000053
Cutaneous 13 (17) 3838 127% 110% 118 (097144) 010 54% 00039 0040
Nausea and vomiting 13 (17) 3585 122% 53% 249 (197313) <00001 60% 000076 <00001
Thrombocytopenia 14 (18) 3737 30% 15% 206 (139306) 000034 0% 092 015
Kidney failure 12 (16) 3542 02% 01% 194 (091414) 0086 0% 100 039
Neurotoxicity 11 (14) 2998 02% 01% 165 (073375) 023 0% 100 069
Hearing loss 11 (15) 3037 29% 13% 228 (146355) 000029 37% 0076 000019
Weight loss 9 (12) 2350 144% 82% 188 (144245) <00001 55% 0011 000048
Febrile neutropenia 8 (11) 1995 30% 23% 130 (079216) 031 40% 0082 0059
Late toxicity
Bone necrosis 10 (14) 2404 05% 04% 117 (051266) 071 0% 100 087
Visual decit 9 (13) 2324 17% 13% 128 (069238) 044 0% 099 060
Brainstem or spinal cord damage 9 (13) 2298 07% 05% 125 (057274) 058 0% 094 049
Symptomatic temporal lobe necrosis (yes or no) 9 (13) 2266 19% 21% 091 (052160) 075 0% 096 100
Xerostomia 9 (12) 2030 51% 36% 145 (095221) 0087 0% 097 050
Cranial nerve palsy 9 (13) 2013 114% 87% 135 (100182) 0052 19% 025 021
Hearing decit 9 (13) 2009 209% 151% 149 (118187) 000068 20% 024 058
Cutaneous brosis 7 (7) 1643 26% 21% 125 (067232) 048 0% 086 068
Trismus 7 (10) 1686 15% 12% 126 (062260) 052 0% 100 085

OR=odds ratio. *Only trials having more than 60% of patients with available data were used for the analyses. Computed rates (see Methods). p values are summarised from appendix. Only patients with a
follow-up longer than 6 months were included in the analyses.

Table 3: Toxicity data

reductions in locoregional failure, distant failure, and possible to create a specic group of trials for the analysis
nasopharyngeal carcinoma-related mortality. The ndings of concomitant plus adjuvant chemotherapy
further support the use of concomitant chemotherapy; administration separate from the concomitant only
there is a signicant association between chemotherapy regimen. The results of the INT-0099 trial was validated
timing and overall and progression-free survival in favour in four independent trials (ve comparisons in the meta-
of concomitant administration. Compared with our analysis), although with a smaller magnitude than in
previous meta-analysis,3 this study has analysed more INT-0099. However, this analysis does not completely
trials and patients and has added new data (toxicities). answer the question whether there is a benet of the
Given the characteristics of the additional trials, it was adjuvant phase in the concomitant setting.

www.thelancet.com/oncology Vol 16 June 2015 653


Articles

The strengths of this meta-analysis are its size and its use of locoregional and distant failures. This strategy has been
of individual patient data, which allowed detailed checking compared with concomitant and adjuvant chemotherapy
and re-analysis of each trial that was subsequently validated in the Hong Kong nasopharyngeal carcinoma study group
by the trial investigator. At the time of this update, seven (HKNPCSG 0501) trial with similar results31 to the
trials had a follow-up longer than 10 years. The large administration of concomitant plus adjuvant chemo-
number of patients allowed for subgroup and subset therapy, and is now considered a potential option in the
analyses to be done with adequate power while respecting 2014 National Comprehensive Cancer Network guide-
the intention-to-treat principle. Sensitivity analyses show lines.32 The use of modern induction chemotherapy
that the results are robust (appendix). The major limitations regimens comprising a taxane or gemcitabine in head and
of this study are the heterogeneity of trial designs and neck squamous cell carcinoma is also being assessed in
chemotherapy regimens, and the use of outdated clinical trials and might change treatment options in the
radiotherapy (two dimensional) in more than three- future.33
quarters of the patients with data available. The analysis of In conclusion, this individual patient data meta-analysis
long-term toxicities was restricted due to the quality of the conrms the benets associated with the addition of
data and the low number of events. Data on dose intensity chemotherapy to radiotherapy in nasopharyngeal car-
and cumulative dose of cisplatin were not collected so an cinoma; the greatest benet was found in the groups with
analysis of the eect of chemotherapy dose on outcome concomitant administration. The benets of the addition of
could not be done. induction or adjuvant chemotherapy in the context of
The benet of the concomitant plus adjuvant chemo- concomitant chemo-radiation still need further assessment.
therapy schedule for all tumour-related outcomes is the Contributors
greatest compared with other treatment modalities, PB, JB, AL, WTN, and JPP designed and supervised the study. JPP, PB,
suggesting that this regimen has the greatest ecacy. AL, and WTN obtained funding. JPP, PB, WTN, and T-XL searched and
selected the trials. SM, JL, PB, and JPP participated in data collection
However, dierences between trials assessing concomitant and checking. SM, JL, and JPP did the statistical analyses. PB, JB, AL,
chemotherapy and those assessing concomitant plus SM, JL, WTN, and JPP wrote the draft, with revisions from the other
adjuvant chemotherapy (table 1) prevent an unbiased investigators. All authors had full access to all the data and analyses and,
comparison between these two treatment schedules. after consultation with the collaborators, had nal responsibility for the
decision to submit for publication. Steering committee members
When trials assessing only concomitant chemotherapy (appendix) revised the protocol, and contributed to the identication and
are analysed, the HR for overall survival is 071 (95% CI selection of the trials. Most steering committee members took part in an
057089) and the dierence between these schedules investigator meeting where preliminary results were presented and
(concomitant chemotherapy vs concomitant plus adjuvant discussed; they revised the manuscript. Trialists (appendix) contributed
to the study by providing data, replying to the secretariat queries, and
chemotherapy), which was already not signicant, validating the re-analysis of their trial. Most trialists took part in an
decreases further. This nding suggests that these two investigator meeting; they revised the manuscript.
schedules are very close in terms of benet. Declaration of interests
Compliance with chemotherapy or radiotherapy can JPP reports grants from Ligue National Contre le Cancer, grants from
potentially aect treatment outcome. A correlation was French Ministry of Health (PAIR-VADS), and support for the
shown between the number of adjuvant cycles of investigator meeting from Sano-Aventis, during the conduct of the
study. EB reports personal fees from General Electric for an
chemotherapy and the occurrence of distant relapse in an epidemiological study, outside the submitted work. The other authors
exploratory analysis of a randomised trial.30 In a disease declare no competing interests.
where more than half of the progression events are distant Acknowledgments
failures, reducing the occurrence of metastatic relapse is a This research was funded by grants from French Ministry of Health
major goal. But given the limited compliance to adjuvant (Programme dactions intgres de recherche VADS), and Ligue
Nationale Contre le Cancer. The INT0099 (SWOG 8892) trial was
chemotherapy after concomitant chemoradiotherapy
supported by US NIH/NCI grants CA180888 and CA180819. The
(between half and three-quarters of the patients in this HeCOG trial was supported by Hellenic Cooperative Oncology Group
meta-analysis received the three planned chemotherapy research grant HE R_5G. The investigator meeting was organised with
cycles), and the uncertain benet and real risks of adjuvant the help of Sano-Aventis. We thank Franoise Delassus for her
secretarial assistance.
treatment, selecting patients at risk for distant relapse
could be an option to individualise therapy in the adjuvant References
1 Lee AWM, Lin JC, Ng WT. Current management of nasopharyngeal
phase. The concentration of circulating Epstein-Barr virus cancer. Semin Radiat Oncol 2012; 22: 23344.
DNA is an attractive biomarker. Persistence of detectable 2 Al-sarraf BM, Leblanc M, Giri PS, et al. Chemoradiotherpy versus
circulating Epstein-Barr virus DNA after local therapy is radiotherapy in patients with advanced nasopharyngeal cancer:
phase III randomized Intergroup Study 0099. J Clin Oncol 1998;
associated with a high risk of tumour recurrence.1 Large- 16: 131017.
scale trials are exploring this biomarker-driven strategy to 3 Baujat B, Audry H, Bourhis J, et al. Chemotherapy in locally
personalise the decision for adjuvant chemotherapy advanced nasopharyngeal carcinoma: an individual patient data
meta-analysis of eight randomized trials and 1753 patients.
(NCT00370890 and NCT02135042). Another option is to Int J Radiat Oncol Biol Phys 2006; 64: 4756.
give chemotherapy before local treatment, because 4 Stewart LA, Clarke MJ. Practical methodology of meta-analyses
induction chemotherapy is generally better tolerated than (overviews) using updated individual patient data: Cochrane
Working Group. Stat Med 1995; 14: 205779.
adjuvant chemotherapy and allows the potential reduction

654 www.thelancet.com/oncology Vol 16 June 2015


Articles

5 Schemper M, Smith TL. A note on quantifying follow-up in studies 21 Lee AW, Tung SY, Chua DT, et al. Randomized trial of radiotherapy
of failure time. Control Clin Trials 1996; 17: 34346. plus concurrent-adjuvant chemotherapy vs radiotherapy alone for
6 Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during regionally advanced nasopharyngeal carcinoma. J Natl Cancer Inst
and after myocardial infarction: an overview of the randomized 2010; 102: 118898.
trials. Prog Cardiovasc Dis 1985; 27: 33571. 22 Lee AW, Tung SY, Chan AT, et al. A randomized trial on addition of
7 Stewart LA, Parmar MKB. Meta-analysis of the literature or of concurrent-adjuvant chemotherapy and/or accelerated fractionation
individual patient data: is there a dierence? Lancet 1993; for locally-advanced nasopharyngeal carcinoma. Radiother Oncol
341: 41822. 2011; 98: 1522.
8 Early Breast Cancer Trialists Collaborative Group. Systemic 23 Hui EP, Ma BB, Leung SF, et al. Randomized phase II trial of
treatment of early breast cancer by hormonal, cytotoxic, or immune concurrent cisplatin-radiotherapy with or without neoadjuvant
therapy. 133 randomised trials involving 31,000 recurrences and docetaxel and cisplatin in advanced nasopharyngeal carcinoma.
24,000 deaths among 75,000 women. Lancet 1992; 339: 115. J Clin Oncol 2009; 27: 24249.
9 Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta- 24 Fountzilas G, Ciuleanu E, Bobos M, et al. Induction chemotherapy
analysis. Stat Med 2002; 21: 153958. followed by concomitant radiotherapy and weekly cisplatin versus
the same concomitant chemoradiotherapy in patients with
10 Non-small Cell Lung Cancer Collaborative Group. Chemotherapy in
nasopharyngeal carcinoma: a randomized phase II study conducted
non-small cell lung cancer: a meta-analysis using updated data on
by the Hellenic Cooperative Oncology Group (HeCOG). Ann Oncol
individual patients from 52 randomised clinical trials. BMJ 1995;
2012; 23: 42735.
311: 899909.
25 Wu X, Huang PY, Peng P, et al. Long-term follow-up of a phase III
11 Fisher DJ, Copas A, Tierney JF, Parmar MKB. A critical review of study comparing radiotherapy with or without weekly oxaliplatin for
methods for the assessment of patient-level interactions in locoregionally advanced nasopharyngeal carcinoma. Ann Oncol
individual participant data meta-analysis of randomized trials, and 2013; 24: 213136.
guidance for practitioners. J Clin Epidemiol 2011; 64: 94967.
26 Huang P-Y, Cao K-J, Guo X, et al. A randomized trial of induction
12 Chan AT, Teo PM, Leung TW, et al. A prospective randomized study chemotherapy plus concurrent chemoradiotherapy versus induction
of chemotherapy adjunctive to denitive radiotherapy in advanced chemotherapy plus radiotherapy for locoregionally advanced
nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1995; nasopharyngeal carcinoma. Oral Oncol 2012; 48: 103844.
33: 56977.
27 Chen Q-Y, Wen Y-F, Guo L, et al. Concurrent chemoradiotherapy vs
13 Chua DT, Sham JS, Choy DD, et al. Preliminary report of the Asian- radiotherapy alone in stage II nasopharyngeal carcinoma: phase III
oceanian clinical oncology association randomized trial comparing randomized trial. J Natl Cancer Inst 2011; 103: 176170.
cisplatin and epirubicin followed by radiotherapy versus
28 Chen Y, Sun Y, Liang S-B, et al. Progress report of a randomized
radiotherapy alone in the treatment of patients with locoregionally
trial comparing long-term survival and late toxicity of concurrent
advanced nasopharyngeal carcinom. Cancer 1998; 83: 227083.
chemoradiotherapy with adjuvant chemotherapy versus
14 International Nasopharynx Cancer Study Group: VUMCA I Trial. radiotherapy alone in patients with stage III to IVB nasopharyngeal
Preliminary results of a rendomized trial comparing neoadjuvant carcinoma from endemic regions of China. Cancer 2013;
chemotherapy (cisplatin, epirubicin, bleomycin) plus radiotherapy 119: 223038.
vs. radiotherapy alone in stage IV (> or = N2, M0) undierentiated 29 Chen Y, Liu M-Z, Liang S-B, et al. Preliminary results of a
nasopharyngeal carcinoma: a positive eect on progression. prospective randomized trial comparing concurrent
Int J Radiat Oncol Biol Phys 1996; 35: 46369. chemoradiotherapy plus adjuvant chemotherapy with radiotherapy
15 Hareyama M, Sakata K, Shirato H, et al. A prospective, randomized alone in patients with locoregionally advanced nasopharyngeal
trial comparing neoadjuvant chemotherapy with radiotherapy alone carcinoma in endemic regions of China. Int J Radiat Oncol Biol Phys
in patients with advanced nasopharyngeal carcinoma. Cancer 2002; 2008; 71: 135664.
94: 221723. 30 Lee AWM, Tung SY, Ngan RKC, et al. Factors contributing to the
16 Chi K-H, Chang Y-C, Guo W-Y, et al. A phase III study of adjuvant ecacy of concurrent-adjuvant chemotherapy for locoregionally
chemotherapy in advanced nasopharyngeal carcinoma patients. advanced nasopharyngeal carcinoma: combined analyses of
Int J Radiat Oncol Biol Phys 2002; 52: 123844. NPC-9901 and NPC-9902 Trials. Eur J Cancer 2011; 47: 65666.
17 Chan AT, Leung S, Ngan RK, et al. Overall survival after concurrent 31 Lee AW, Ngan RK, Tung SY, et al. Preliminary results of trial NPC-
cisplatin-radiotherapy compared with radiotherapy alone in 0501 evaluating the therapeutic gain by changing from concurrent-
locoregionally advanced nasopharyngeal carcinoma. adjuvant to induction-concurrent chemoradiotherapy, changing
J Natl Cancer Inst 2005; 97: 53639. from uorouracil to capecitabine, and changing from conventional
18 Kwong DL, Sham JS, Au GK, et al. Concurrent and adjuvant to accelerated radiotherapy fractionation in patients with
chemotherapy for nasopharyngeal carcinoma: a factorial study. locoregionally advanced nasopharyngeal carcinoma. Cancer 2015;
J Clin Oncol 2004; 22: 264353. 121: 132838.
19 Chen L, Hu C-S, Chen X-Z, et al. Concurrent chemoradiotherapy 32 National Comprehensive Cancer Network guidelines. 2014. http://
plus adjuvant chemotherapy versus concurrent chemoradiotherapy www.nccn.org/professionals/physician_gls/pdf (accessed April 23,
alone in patients with locoregionally advanced nasopharyngeal 2015).
carcinoma: a phase 3 multicentre randomised controlled trial. 33 Blanchard P, Bourhis J, Lacas B, et al. Taxane-cisplatin-uorouracil
Lancet Oncol 2012; 13: 16371. as induction chemotherapy in locally advanced head and neck
20 Wee J, Tan EH, Tai BC, et al. Randomized trial of radiotherapy cancers: an individual patient data meta-analysis of the meta-
versus concurrent chemoradiotherapy followed by adjuvant analysis of chemotherapy in head and neck cancer group.
chemotherapy in patients with American Joint Committee on J Clin Oncol 2013; 31: 285460.
Cancer/International Union against cancer stage III and IV
nasopharyngeal cancer of the endemic variety. J Clin Oncol 2005;
23: 673038.

www.thelancet.com/oncology Vol 16 June 2015 655


Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.