1. Benjamin Lacas, François Janot, Jean Bourhis,
Méta-analyses de chimiothérapie dans les cancers des VADS: actualisation
Jean-Pierre Pignon, Pierre Blanchard, Anne Lee, Laureen Majed, Sophie Marguet,
Claire Petit, Cécile Landais, Julie Leclercq, Béranger Luéza, Federico Rotolo,
Benjamin Lacas, François Janot, Jean Bourhis,
Service de biostatistique, dpts de radiothérapie et chirurgie cervicofaciale,
Plateforme LNCC Méta-analyse en Oncologie, INSERM U1018 CESP, Hong-Kong NPC Collaborative Group, CHU vaudois de Lausanne.

2. Head and neck meta-analyses
Meta-analysis
Initial (no. trials, pts) Publication
Update 1 (no. trials, pts) Publication
Update 2 (no. trials, pts) Publication
MACH-NC
(CT, HNSCC)
*,$
(63,10 741)
Lancet 2000 $
(87, 16 485)
Rad Oncol 2009/2011
£,**,$$
(102, 19 325)
ESMO 2016
MAC-NPC
(CT, NPC)
< (8, 1 753) Red 2006
£,$$ (19, 4 806) Lancet Oncol 2015
 Initiated in
MARCH
(RT, HNSCC)
(15, 6 515) Lancet 2006
£,% (30, 11 140) ECCO ESTRO 2014
Final analysis and manuscript
* MA on larynx preservation  ; $ MA sequential vs. concomitant
£ Effect at 10 years, data on toxicity and compliance ;
% Postoperative trials eligible, MA standard radiotherapy (RT) + CT vs. modified fractionation RT
** JCO 2013 TPF vs. PF; $$ Trials of Timing 1 ± Timing 2 eligible 2

3. Meta-analysis of chemotherapy in nasopharynx carcinoma
Standard MA (Blanchard et al. Lancet Oncol 2015)
Network MA (Ribassin-Majed et al. JCO 2016)
Surrogate analysis (Rotolo et al. JNCI 2016) 3

4. MAC-NPC : Material for network and standard MA
20 trials, patients
26 comparisons
median follow-up = 7,4 years
Accrual < 2011
MA standard: 4 groups
1. RT vs. induction CT (IC) + RT
CRT vs. IC-CRT
2. RT vs. adjuvant CT (AC) + RT
CRT + CRT-AC
3. RT vs. concomitant CT (C) + RT
IC-RT vs. IC-CRT
4. RT vs. CRT-AC
145 sur 157 HR (92,3%) 4

5. Standard meta-analysis: summary results
Overall Survival
Progression-Free Survival
Loco-regional Control
Induction*
 0.96 [0.80;1.16]
0.81 [0.69;0.95]
0.84 [0.66;1.07]
Adjuvant*
 0.87 [0.68;1.12]
0.80 [0.64;1.00]
0.61 [0.41;0.93]
Concomitant*
0.80 [0.70;0.93]
0.81 [0.71;0.92]
0.82 [0.67;1.01]
Concomitant and adjuvant*
 0.65 [0.56;0.76]
0.62 [0.53;0.72]
0.54 [0.41;0.71]
Overall*
 0.79 [0.73;0.86]
0.75 [0.69;0.81]
0.73 [0.64;0.83]
Overall test**
<0.0001 
<0.0001
Interaction test (between timing and treatment effect)**
 0.01
0.04
0.05
Residual heterogeneity test**
0.36 
0.62
0.78
* Hazard ratio [95% confidence interval]
** p-value
* Hazard ratio [95% confidence interval]; ** p-value; *** toxicity related to one trial, disappears after exclusion of this trial (new HR: 0.91 [0.39;2.15]) 5

6. Standard meta-analysis of CT for non metastatic NPC: Conclusions
Overall benefit of the addition of chemotherapy on OS (~6% at 5-years), PFS, locoregional and distant control, and cancer death
Superiority of concomitant (~9-12% at 5-years) over induction or adjuvant
Comparison between concomitant +/- adjuvant (or induction) deserves further studies 6

7. Network meta-analysis : Methods
Individual data to compute hazard ratio from log-rank test, stratified by trial
Frequentist approach with random effect model in case of heterogeneity
Overall survival (OS) main endpoint
Heterogeneity and inconsistency were assessed by a global Cochran Q statistic.
P-score (P-s) used to rank treatments = percent of certainty to be the best treatment
145 sur 157 HR (92,3%) 7

8. Second First Radiotherapy (RT) Induction CT-RT (IC-RT) IC-CRT
Conc-CT-RT (CRT)
Induction CT-RT (IC-RT)
IC-CRT
CRT-AC
IC-CRT-AC
RT-Adjuvant CT (AC-RT)
Radiotherapy (RT) 8
145 sur 157 HR (92,3%)

9. Network meta-analysis in patients with non metastatic NPC: conclusion
The addition of adjuvant chemotherapy (CT) to concomitant CT-radiotherapy (CRT) achieved the highest survival benefit and consistent improvement for all endpoints
The addition of induction chemotherapy to CRT achieved the highest effect on distant control
Regimens with more CT were associated with increased risk of acute toxicity
Results of recent trials on induction CT will clarify the role of the timing of CT 9

10. MAC-NPC: surrogate analysis (Rotolo et al JNCI 2016)
Statistical methods
Individual level: rank correlation ρ between surrogate and OS estimated from the bivariate distribution
Trial level: coefficient of determination R2 between treatment effects (log hazard ratios) on surrogate and OS, estimated from a linear regression
Results and conclusion
PFS and DMFS* are valid surrogate endpoints for OS in randomized CT trials of patients with LANPC
PFS can be observed earlier than DMFS and showed more robust results
145 sur 157 HR (92,3%)
*DMFS = Distant Metastasis-Free Survival 10

11. Meta-analysis of chemotherapy in head en neck cancer
Preliminary results:
Standard MA (Blanchard et al. ESMO 2016)
Network MA (Petit et al. ECCO 2017) 11

12. Locoregional treatment (LRT ) vs. LRT + CT: Overall Survival
patients and deaths (69%)
Median follow-up: 6.7 years
Accrual between 1965 and 2010
Trials on concomitant CT 12

13. Summary for overall and progression-free survival
Overall Survival
Progression-Free Survival
LRT vs LRT + CT
0.89 [0.86;0.92]
0.87 [0.84;0.90]
Concomitant CT
0.83 [0.79;0.87]
0.80 [0.76;0.83]
Induction CT
0.97 [0.91;1.03]
0.98 [0.92;1.04]
Adjuvant CT
1.02 [0.92;1.13]
0.99 [0.89;1.10]
Interaction
p<0.0001
p <
Induction (+/- adj.) vs concomitant CT-RT = direct comparison (8 trial, patients)
0.84 [0.74;0.95]
0.85
[0.75;0.96]
Indirect
comparison 13

14. Second update of the MACH-NC data base : additional trials and long term follow-up : conclusion
No change: concomitant CT > induction and adjuvant CT; superiority of platin alone, and 5FU plus platin (PF) compared to the other concomitant chemotherapy
Older (>70) and frail (PS >2) patients might not benefit (or benefit less) from CT
For induction chemotherapy, PF is superior to the absence of CT
No conclusion for TPF as results are heterogeneous (some trial with major toxicity and data collection still ongoing) 14

15. Network meta-analysis: ECCO 2017
117 trials, corresponding to 150 comparisons from MACH-NC and MARCH (comparison of modified fractionation and standard RT)
patients
16 modalities of treatment
35 direct comparisons
deaths and events for PFS
CONFIDENTIAL
145 sur 157 HR (92,3%) 15

16. CONFIDENTIAL Network for OS ICTaxPF-LRT ICPF-LRT ICother-LRT
Label
Description
LRT
Standard RT +/- surgery
CLRTP
LRT + concomitant CT with platin
CLRTnoP
LRT + concomitant CT without platin
ICother - LRT
Induction CT other + LRT
ICPF - LRT
Induction CT PF + LRT
ICTaxPF - LRT
Induction CT TaxPF + LRT
LRT - AC
LRT + adjuvant CT
ICother - CLRT
Induction CT other + LRT+ concomitant CT
ICPF - CLRT
Induction CT PF + LRT + concomitant CT
ICTaxPF - CLRT
Induction CT Tax-PF + LRT + concomitant CT
HFRT
Hyperfractionnated RT
MART
Moderately accelerated RT
VART
Very accelerated RT
HFCRT
Hyperfractionnated RT + concomitant CT
ACRT
Accelerated (moderately and very) RT + concomitant CT
CLRTnoP - AC
CLRT without platin + adjuvant CT
CLRTP
ICTaxPF-CLRT
CLRTnoP
CLRTnoP-AC
ICother-LRT
ICPF-LRT
ICTaxPF-LRT
ICother-CLRT
ICPF-CLRT 16

17. Network meta-analysis in patients with non metastatic HNSCC: Results
Hyperfractionated radiotherapy with concomitant chemotherapy (HFCRT) was ranked as the best treatment in all analyses. The hazard ratios (HR) of HFCRT compared to platinum-based CRT was 0.80 [95% CI ] for OS (P-s 0.97) and 0.77 [95% CI: ] for progression-free survival (P-s 0.98). The superiority of HFCRT was robust to sensitivity analyses
Three other modalities of treatment had a better P-score than platinum-based CRT (P-s 0.78) but their HR for death were not significantly different: induction chemotherapy (TaxPF) followed by LRT (IC-LRT, (P-s 0.89)), accelerated radiotherapy with concomitant chemotherapy (ACRT, (P-s 0.82)) and induction chemotherapy (TaxPF) followed by CRT (IC-CRT, (P-s 0.79))
CONFIDENTIAL 17

18. Network meta-analysis in patients with non metastatic HNSCC: Results
Treatment comparison
Rank
Network meta-analysis
Number of trials per comparison HR
95% CI
Compared to platinum-based CRT
Hyp Frac (HF) Conc CT-RT (CRT) 1
0.80
[ ] 2
Ind CT (IC) (TaxPF) followed by LRT
0.90
[ ]
Acc CRT 3
0.97
[ ] 4
IC (TaxPF) followed by CRT
0.98
[ ]
Compared to LRT
HFCRT
0.62
[ ]
IC (TaxPF) followed by LRT
0.70
[ ]
ACRT
0.75
[ ]
0.76
[ ]
Platinum-based CRT 5
0.77
[ ] 23
CONFIDENTIAL 18

19. Network meta-analysis in patients with non metastatic HNSCC : conclusion
The results suggest the superiority of HFCRT for the treatment of LAHNC
Although toxicity is not addressed, these results, which ideally need to be confirmed by RCTs, could be clinically useful in advanced diseases with a high risk of locoregional failure, as represented by the patients in these meta-analyses
Network meta-analysis is a new method and results based on this method should be interpreted with caution (potential bias, robustness of the results) and the uncertainty of ranking be taken into account
CONFIDENTIAL 19

20. Remerciements
Les membres des groupes coopérateurs MACH-NC, MARCH et MAC-NPC et leur patients sans qui ce travail n’aurait pas été possible
Les membres du service de Biostatistique et d’Epidémiologie de Gustave Roussy, en particulier Françoise Delassus et le reste de l’équipe méta-analyse
L’INCa, la Ligue, l’ARC et Sanofi pour leur soutient financier sur le long terme 20

21. Gustave Roussy: Meta-analysis team