Articles

First-line nivolumab plus ipilimumab in unresectable

malignant pleural mesothelioma (CheckMate 743):
a multicentre, randomised, open-label, phase 3 trial
Paul Baas, Arnaud Scherpereel, Anna K Nowak, Nobukazu Fujimoto, Solange Peters, Anne S Tsao, Aaron S Mansfield, Sanjay Popat, Thierry Jahan,
Scott Antonia, Youssef Oulkhouir, Yolanda Bautista, Robin Cornelissen, Laurent Greillier, Francesco Grossi, Dariusz Kowalski,
Jerónimo Rodríguez-Cid, Praveen Aanur, Abderrahim Oukessou, Christine Baudelet, Gérard Zalcman

Summary
Background Approved systemic treatments for malignant pleural mesothelioma (MPM) have been limited to Lancet 2021; 397: 375–86
chemotherapy regimens that have moderate survival benefit with poor outcomes. Nivolumab plus ipilimumab has Published Online
shown clinical benefit in other tumour types, including first-line non-small-cell lung cancer. We hypothesised that January 21, 2021
https://doi.org/10.1016/
this regimen would improve overall survival in MPM.
S0140-6736(20)32714-8
See Comment page 348
Methods This open-label, randomised, phase 3 study (CheckMate 743) was run at 103 hospitals across 21 countries.
The Netherlands Cancer
Eligible individuals were aged 18 years and older, with previously untreated, histologically confirmed unresectable Institute and Leiden
MPM, and an Eastern Cooperative Oncology Group performance status of 0 or 1. Eligible participants were University Medical Center,
randomly assigned (1:1) to nivolumab (3 mg/kg intravenously once every 2 weeks) plus ipilimumab (1 mg/kg Amsterdam, Netherlands
intravenously once every 6 weeks) for up to 2 years, or platinum plus pemetrexed chemotherapy (pemetrexed (Prof P Baas MD); Pulmonary
and Thoracic Oncology,
[500 mg/m² intravenously] plus cisplatin [75 mg/m² intravenously] or carboplatin [area under the concentration- University of Lille, CHU Lille,
time curve 5 mg/mL per min intravenously]) once every 3 weeks for up to six cycles. The primary endpoint was INSERM U1189, OncoThAI,
overall survival among all participants randomly assigned to treatment, and safety was assessed in all participants Lille, France
(Prof A Scherpereel MD);
who received at least one dose of study treatment. This study is registered with ClinicalTrials.gov, NCT02899299,
Medical School, University of
and is closed to accrual. Western Australia Perth, WA,
Australia
Findings Between Nov 29, 2016, and April 28, 2018, 713 patients were enrolled, of whom 605 were randomly (Prof A K Nowak PhD);
Department of Medical
assigned to either nivolumab plus ipilimumab (n=303) or chemotherapy (n=302). 467 (77%) of 605 participants
Oncology, Sir Charles
were male and median age was 69 years (IQR 64–75). At the prespecified interim analysis (database lock Gairdner Hospital, Perth, WA,
April 3, 2020; median follow-up of 29·7 months [IQR 26·7–32·9]), nivolumab plus ipilimumab significantly Australia (Prof A K Nowak);
extended overall survival versus chemotherapy (median overall survival 18·1 months [95% CI 16·8–21·4] vs Okayama Rosai Hospital,
Okayama, Japan
14·1 months [12·4–16·2]; hazard ratio 0·74 [96·6% CI 0·60–0·91]; p=0·0020). 2-year overall survival rates were
(Prof N Fujimoto MD);
41% (95% CI 35·1–46·5) in the nivolumab plus ipilimumab group and 27% (21·9–32·4) in the chemotherapy Lausanne University
group. Grade 3–4 treatment-related adverse events were reported in 91 (30%) of 300 patients treated with nivolumab Hospital, Lausanne,
plus ipilimumab and 91 (32%) of 284 treated with chemotherapy. Three (1%) treatment-related deaths occurred in Switzerland
(Prof S Peters MD); MD
the nivolumab plus ipilimumab group (pneumonitis, encephalitis, and heart failure) and one (<1%) in the
Anderson Cancer Center,
chemotherapy group (myelosuppression). Houston, TX, USA
(Prof A S Tsao MD); Mayo
Interpretation Nivolumab plus ipilimumab provided significant and clinically meaningful improvements in overall Clinic, Rochester, MN, USA
(A S Mansfield MD); Royal
survival versus standard-of-care chemotherapy, supporting the use of this first-in-class regimen that has been Marsden Hospital, London,
approved in the USA as of October, 2020, for previously untreated unresectable MPM. UK (Prof S Popat FRCP);
Institute of Cancer Research,
Funding Bristol Myers Squibb. London, UK (Prof S Popat);
UCSF Helen Diller Family
Comprehensive Cancer
Copyright © 2021 Elsevier Ltd. All rights reserved. Center, San Francisco, CA,
USA (Prof T Jahan MD);
Introduction have been the only approved first-line treatment H Lee Moffitt Cancer Center
& Research Institute, Tampa,
Malignant pleural mesothelioma (MPM) is a highly regimens for MPM since 2004.4,5 However, long-term FL, USA (Prof S Antonia MD);
aggressive cancer and typically unresectable at diagnosis, survival outcomes remain poor with chemo­therapy;6–9 Hôpital Côte de Nacre CHU
with less than 10% of patients surviving 5 years or bevacizumab has been added to these regimens10 but its de Caen, Caen, France
beyond.1,2 Historically, age, sex, tumour grade and stage, use varies across regions. As such, there is an urgent (Y Oulkhouir MD); Centro
Médico Nacional Siglo XXI,
and histology have been shown to be independent need for new and effective therapeutic options. Mexico City, Mexico
prognostic factors. Notably, worse prognosis has been Nivolumab, a fully human anti-programmed cell (Prof Y Bautista MD); Erasmus
reported for non-epithelioid histology versus the death 1 (PD-1) antibody, and ipilimumab, a fully human MC Cancer Institute,
epithelioid subtype.1–3 Until October, 2020, platinum anti-cytotoxic T-lymphocyte 4 (CTLA-4) antibody are Rotterdam, Netherlands
(R Cornelissen MD);
agents plus folate antimetabolites, such as pemetrexed, immune checkpoint inhibitors with distinct but

www.thelancet.com Vol 397 January 30, 2021 375

 Articles

Aix Marseille University,

APHM, INSERM, CNRS, Research in context
CRCM, Hôpital Nord,
Multidisciplinary Oncology Evidence before this study phase 2 study (INITIATE) assessing nivolumab plus ipilimumab
and Therapeutic Innovations We searched PubMed and abstracts from major oncology in MPM showed that this regimen was tolerable and exhibited
Department, Marseille, congresses for studies published from database inception until encouraging clinical activity.
France (Prof L Greillier MD);
Fondazione IRCCS Ca’ Granda Oct 2, 2020, relevant to unresectable malignant pleural
Added value of this study
Ospedale Maggiore mesothelioma (MPM) and cancer immunotherapy regimens,
Here we provide results from the randomised CheckMate 743
Policlinico, Milan, Italy with a focus primarily on first-line phase 3 trials, using search
(F Grossi MD); Department of study, which is the first phase 3 study to show significant and
terms that included, but were not limited to (“mesothelioma”
Lung Cancer and Chest clinically meaningful improvements in overall survival with
Tumours, AND “nivolumab”) OR “chemotherapy” OR “pembrolizumab”
immunotherapy versus standard-of-care platinum plus
Maria Sklodowska-Curie OR “atezolizumab” OR “avelumab” OR “durvalumab” OR
pemetrexed chemotherapy for first-line treatment of
National Research Institute “ipilimumab” OR “tremelimumab” OR “PD-1” OR “PD-L1” OR
of Oncology, Warsaw, Poland unresectable MPM. This regimen was found to show clinical
“CTLA-4” (full names and abbreviations). Although we
(Prof D Kowalski MD); Centro benefit and tolerability, thus providing patients with a new
Oncológico, Médica identified several studies assessing immunotherapy in MPM,
first-line chemotherapy-free treatment option. Notably,
Sur-Instituto Nacional de we found no published randomised phase 3 studies
Enfermedades Respiratorias,
survival with nivolumab plus ipilimumab was similar in patients
investigating the efficacy or safety of immunotherapy
Mexico City, Mexico with both non-epithelioid and epithelioid histologies,
regimens in the first-line setting. Various phase 1 and 2 studies
(Prof J Rodríguez-Cid MD); suggesting that the regimen could be considered for all patients
Bristol Myers Squibb, in previously treated patients with MPM have suggested that
with unresectable MPM. Responses were durable, with a 2-year
Princeton, NJ, USA immunotherapy regimens might provide clinical benefit.
(P Aanur MD, A Oukessou MD,
duration of response rate of 32% of immunotherapy-treated
Notably, the multicentre, open-label, single-arm, phase 2
C Baudelet PhD); Bichat- patients. The safety profile of nivolumab plus ipilimumab was
MERIT study led to the approval of nivolumab monotherapy
Claude Bernard University consistent with that observed in first-line non-small-cell lung
Hospital, AP-HP, Université for unresectable recurrent MPM in Japan. However, with
cancer at this dose and schedule and no new safety signals were
de Paris, Paris, France recommended first-line systemic treatments limited to
(Prof G Zalcman MD) reported.
chemotherapy since 2004, with or without bevacizumab,
Correspondence to: there remains a need for new and effective therapeutic Implications of all the available evidence
Prof Paul Baas,
options. In the single-arm phase 2 DREAM study, first-line Nivolumab plus ipilimumab can provide notable and clinically
The Netherlands Cancer
Institute, Amsterdam, durvalumab plus chemotherapy exhibited promising activity meaningful improvements in overall survival versus the current
1066CX, Netherlands in 54 patients with MPM, but the combination requires standard of care. Data from CheckMate 743 support a
p.baas@nki.nl evaluation in a larger, randomised, phase 3 study. CheckMate favourable clinical benefit–risk profile for nivolumab plus
743 was designed to investigate the efficacy and safety of ipilimumab. Nivolumab plus ipilimumab is now indicated in the
nivolumab plus ipilimumab versus chemotherapy. A previous USA and Brazil as a first-line treatment for unresectable MPM.
non-comparative phase 2 trial (MAPS2) and single-arm

complementary mechanisms of action. Ipilimumab Additionally, NCCN guidelines recommend nivolumab

induces T-cell proliferation and de-novo anti-tumour plus ipilimumab as a preferred first-line option (category
T-cell responses, including in memory T cells, whereas 2A) for patients with biphasic or sarcomatoid histology
nivolumab restores the function of existing anti-tumour and is also an option for those with epithelioid histology.
T cells.11 Nivolumab plus ipilimumab is approved in
various tumours12 and has shown durable overall survival Methods
benefit in melanoma,13 renal cell carcinoma,14 and in Study design and participants
non-small-cell lung cancer (NSCLC).15 Furthermore, CheckMate 743 is a global, open-label, randomised,
National Comprehensive Cancer Network (NCCN) controlled, phase 3 study run at 103 hospitals across
For the most recent and Clinical Practice Guidelines in Oncology (NCCN 21 countries (appendix pp 2–4, 22). Eligible patients were
complete version of the NCCN guidelines) recommend nivolumab with or without aged 18 years or older with histologically confirmed
guidelines see NCCN.org
ipilimumab as a preferred treatment option (category 2A) unresectable MPM that was not amenable to curative
See Online for appendix
in second-line or later MPM settings based on results therapy (surgery with or without chemotherapy), and an
from three phase 2 trials,16–18 including the multicentre, Eastern Cooperative Oncology Group performance status
open-label, randomised, non-comparative IFCT-1501 of 0 or 1.19 Unresectability of the disease was determined
MAPS2 trial that showed encouraging clinical activity of by the investigator at individual sites using local standards.
the com­bination therapy.16 Patients must have completed any previous palliative
CheckMate 743 is a phase 3 study designed to assess radiotherapy 2 weeks or longer before initiating study
efficacy and safety of first-line nivolumab plus ipilimumab treatment, with no residual signs of toxicity, and have
versus platinum plus pemetrexed chemotherapy in measurable disease according to the modified Response
unresectable MPM. Here we present results from the Evaluation Criteria in Solid Tumors (mRECIST)20 for
prespecified interim analysis, which has led to nivolumab pleural mesothelioma. Patients without measurable
plus ipilimumab gaining approval in the USA.12 pleural lesions but with metastatic non-pleural lesions

376 www.thelancet.com Vol 397 January 30, 2021

 Articles

measurable per RECIST version 1.1 could be considered ipilimumab (1 mg/kg intravenous infusion once every
for inclusion after consultation with the study’s medical 6 weeks). Nivolumab was administered first, followed by
monitor. Patients were required to have tumour samples ipilimumab. Participants in the chemotherapy group
available for programmed cell death ligand 1 (PD-L1) were given an intravenous infusion of cisplatin
testing. Baseline laboratory tests required to assess (75 mg/m²) or carboplatin (area under the concentration-
eligibility included white blood cell counts, neutrophils, time curve 5 mg/mL per min) plus pemetrexed
platelets, haemoglobin, serum creatinine, alanine amino­ (500 mg/m²) every 3 weeks for a maximum of six cycles.
trans­ferase, aspartate amino­transferase, and total bilirubin Treatment was continued until disease progression,
(appendix p 6). unacceptable tox­icity, or for 2 years for immunotherapy.
Exclusion criteria included brain metastases (unless Treatment with nivolumab plus ipilimumab was
resected or treated with stereotactic radiotherapy and permitted beyond disease progression if prespecified
asymptomatic with no evolution within 3 months before requirements were met (appendix p 7). Dose reductions
study inclusion), autoimmune disease, and previous were permitted for chemotherapy, but not for nivolumab
treatment with drugs targeting T-cell costimulation or or ipilimumab; concomitant use of corticosteroids was
checkpoint pathways. Patients were excluded if they permitted. Patients could receive subsequent therapy
presented with primitive peritoneal, pericardial, tunica upon the discontinuation of study treatment in either
vaginalis, or testis mesotheliomas. Other exclusion group at the discretion of the investigator.
criteria included inadequate haematological, renal, or Tumour assessments were done 6 weeks after the date
hepatic function; known HIV infection; or interstitial of the first dose of study drug and then every 6 weeks
lung disease that was either symptomatic or might affect for the first 12 months. After 12 months, tumours were
the detection or management of suspected drug-related assessed every 12 weeks until blinded independent
pulmonary toxicity. Patients with current or previous central review (BICR) confirmed disease progression per
malignancy with less than 3 years of complete remission mRECIST or RECIST version 1.1 criteria, or both. At
(except for non-melanoma skin cancers and in-situ the time of investigator-assessed initial radiographic
cancers) requiring or likely to require concurrent progression, the site had to request the blinded indepen­
intervention during the study period were ineligible, as dent central review of progression from a third-party
were patients requiring systemic corticosteroids (>10 mg radiology vendor (E-research Technologies in St Louis,
daily prednisone or equivalent) or immunosuppressive MO, USA); if progression was not confirmed, treatment
medication within 14 days of the first dose of study drug. could continue.
More detail on eligibility criteria are in the appendix (p 5) Adverse events were assessed at baseline and
and study protocol (appendix pp 27–410). continuously throughout the study and during follow-up.
An institutional review board or independent ethics Adverse events were graded according to the National
committee at each study centre approved all versions Cancer Institute Common Terminology Criteria for
of the protocol. An independent Data Monitoring Adverse Events (version 4.0). Select adverse events
Committee provided general oversight of efficacy and consisted of a list of preferred terms with potential
safety for the trial. The trial was done in accordance immune aetiology grouped by specific category (gastro­
with the Declaration of Helsinki and the International intestinal adverse events, pulmonary adverse events,
Conference on Harmonisation Good Clinical Practice renal adverse events, hepatic adverse events, skin adverse
guidelines. All patients provided written informed events, infusion reactions, and endocrinopathies). The
consent. definition for serious adverse events is in the appendix
(p 6). Treatment-related adverse events were defined as
Randomisation and masking those reported between the first dose of study drug and
Patients were enrolled and randomly assigned (1:1) using 30 days after the last dose of study drug. According to
an interactive web response system, stratified by sex study sponsor practice, only events that led to death
and histology (epithelioid vs non-epithelioid [including within 24 h were documented as grade 5 events and
sarcomatoid and mixed subtypes]) to nivolumab plus reported as deaths here. Events leading to death more
ipilimumab or platinum plus pemetrexed chemotherapy. than 24 h after onset are reported with the worst grade
The trial was open label and so patients and investigators before death.
were not masked to treatment assignment. Tumour histology was determined by individual
sites using local protocols. Archival or fresh formalin-
Procedures fixed paraffin-embedded tumour samples were collected
Participants in both treatment groups were pretreated before randomisation. Optional on-treatment fresh
with folic acid (350–1000 µg orally daily) and vitamin B12 tumour samples were collected at weeks 6–8 and at
(1000 µg intramuscularly) 1 week before administration disease progression, at the discretion of the investigator.
of the first dose of study drug (appendix p 5). Partici­ Samples were sent to a central laboratory (Cancer
pants in the experimental group were given nivolumab Genetics, Rutherford, NJ, USA, and for patients in
(3 mg/kg intravenous infusion once every 2 weeks) plus China, PD-L1 testing was done at Covance, Shanghai) to

www.thelancet.com Vol 397 January 30, 2021 377

 Articles

determine the proportion of tumour cells showing appendix p 7).21 Overall survival was defined as the time
plasma membrane PD-L1 staining of any intensity using from randomisation to the date of death due to any cause.
the validated immunohistochemical 28-8 pharmDx assay Secondary endpoints were progression-free survival,
(Dako, Carpinteria, CA, USA). objective response rate, time to response, duration of
Laboratory tests were done within 14 days before response, and disease control rate (radio­graphic tumour
randomisation and within 3 days before each dose. Full assess­ments per adapted mRECIST for pleural lesisons
details of all assessments done are in the appendix (p 6). and RECIST [version 1.1] for the other lesions by BICR)
Hepatitis C RNA and HIV (where locally mandated) tests in all patients randomly assigned to treatment, as well as
were done at screening only. All tests had to be done at overall survival, progression-free survival, and objective
follow-up visits 1 and 2. response rate by PD-L1 expression.
Progression-free survival was defined as the time from
Outcomes randomisation to the date of the first documented tumour
The primary endpoint was overall survival in all patients progression or death due to any cause. Participants who
randomly assigned to treatment after the US Food died were considered to have progressed on the date of
and Drug Administration provided guidance to change death. Participants who received subsequent therapy
progression-free survival from a coprimary endpoint to a without previous reported progression were considered
secondary endpoint (protocol amendment April 25, 2019; to have progressed on the date of death or were censored
at the date of last evaluable tumour assessment before or
on initiation of subsequent therapy. Objective response
713 patients enrolled rate was defined as the proportion of patients with a best
overall response of partial response or complete response
and disease control rate was defined as the proportion
108 excluded
84 no longer met inclusion criteria of patients with a best overall response of complete
11 withdrew consent response, partial response, or stable disease. Duration
4 adverse event
2 poor compliance or non-compliance
of response was defined as the time between the date of
6 died first response to the date of the first documented tumour
1 other reason progression, or death due to any cause, whichever
occurred first.
605 randomly assigned to treatment Exploratory endpoints included safety and tolerability
in all treated patients. Analysis of other exploratory
endpoints that are ongoing but not reported here
include pharmacokinetics, biomarkers, patient-reported
303 assigned to nivolumab plus ipilimumab 302 assigned to chemotherapy out­comes, and immunogenicity; a full list is in the
appendix (pp 119–122).
3 did not receive allocated treatment 18 did not receive allocated treatment
1 withdrew consent 3 patient request Statistical analysis
2 no longer met inclusion criteria 11 withdrew consent For the primary endpoint of overall survival, a sample of
3 no longer met inclusion criteria
1 not reported approximately 600 patients randomly assigned to
treatment with 473 deaths would provide 90% power to
detect a target hazard ratio (HR) of 0·72 with a two-sided
300 received allocated intervention 284 received allocated intervention
type 1 error of 0·05, by means of a log-rank test. One
prespecified interim analysis of overall survival was
5 still on treatment 0 still on treatment planned for superiority at approximately 403 deaths
295 discontinued treatment 284 discontinued treatment (85% of total anticipated events). At the time of database
182 disease progression 44 disease progression
59 study drug toxicity 24 study drug toxicity
lock for the interim analysis, 419 patients had died
12 adverse event unrelated to study 9 adverse event unrelated to study (89% of total anticipated events); the boundary for
drug drug declaring superiority for overall survival was a p value of
4 patient request 10 patient request
6 patient withdrawal of consent 3 patient withdrawal of consent less than 0·0345, based on the Lan-DeMets alpha
10 maximum clinical benefit 1 lost to follow-up spending function with O’Brien-Fleming boundaries.
1 poor compliance or non-compliance 2 maximum clinical benefit
4 no longer met inclusion criteria 2 other None of the secondary endpoints were included in the
13 other reason 189 not reported testing procedure; therefore, we did no formal statistical
4 not reported 176 completed six cycles
testing or allocation of alpha values for progression-free
survival, objective response rate, and disease control rate.
303 analysed in efficacy analysis 302 included in efficacy analysis
We included all patients randomly assigned to
300 analysed in safety analysis 284 included in safety analysis treatment in demographic and efficacy analyses. We
stratified analyses for overall survival and progression-
Figure 1: Trial profile free survival by sex and histology. We estimated HRs

378 www.thelancet.com Vol 397 January 30, 2021

 Articles

and CIs (96·6% CIs for the overall survival primary

Nivolumab plus Chemotherapy
analysis [alpha adjusted for interim analysis], and ipilimumab group group (n=302)
95% CI elsewhere) using a stratified Cox propor­tional (n=303)
hazards model with treatment group as a single Age, years 69 (65–75) 69 (62–75)
covariate. We checked the proportional hazards <65 71 (23%) 96 (32%)
assumption only for the primary endpoint of overall ≥65 to <75 154 (51%) 127 (42%)
survival by adding a time-dependent covariate, defined ≥75 78 (26%) 79 (26%)
by treatment-by-time interaction, into the stratified Cox Sex
regression model of overall survival. We estimated
Male 234 (77%) 233 (77%)
survival curves and rates using the Kaplan-Meier
Female 69 (23%) 69 (23%)
method. We calculated exact two-sided 95% CIs for
Region
objective response and disease control rates using the
North America 32 (11%) 27 (9%)
Clopper-Pearson method. We did prespecified descrip­
Europe 177 (58%) 175 (58%)
tive subgroup analyses for overall survival, summarised
Asia 26 (9%) 39 (13%)
using HRs (with 95% CIs) calculated using an unstra­
Rest of the world* 68 (22%) 61 (20%)
tified Cox proportional hazards model. Safety analyses
Eastern Cooperative Oncology Group performance status†
included all patients who received at least one dose of
0 114 (38%) 128 (42%)
study drug. We also did exposure adjusted safety
1 189 (62%) 173 (57%)
analyses, taking into account all on-treatment events on
Smoking status
the basis of the total exposure time. We calculated the
person-year exposure as the sum over the participants’ Current or former 173 (57%) 171 (57%)

exposure expressed in years. More details on all Never 127 (42%) 122 (40%)

analyses are in the appendix (pp 7–8). Unknown 3 (1%) 9 (3%)

We did all statistical analyses using SAS software Histology
(version 9.2). An independent Data Monitoring Com­ Epithelioid 229 (76%) 227 (75%)
mittee reviewed efficacy and safety data on a periodic Non-epithelioid 74 (24%) 75 (25%)
basis and at the time of the preplanned interim ana­lysis. Sarcomatoid 35 (12%) 36 (12%)
This trial is registered with ClinicalTrials.gov, Mixed or other 39 (13%) 39 (13%)
NCT02899299. Stage
1 12 (4%) 20 (7%)
Role of the funding source 2 23 (8%) 22 (7%)
The study was designed by the funder (Bristol Myers 3 103 (34%) 106 (35%)
Squibb) and study steering committee. The funder had a 4 160 (53%) 149 (49%)
role in data collection with the investigators, data analysis Not reported 5 (2%) 5 (2%)
and interpretation in collaboration with the authors, and Previous cancer therapy
the writing of the report by funding professional medical Radiotherapy‡ 29 (10%) 28 (9%)
writing assistance. All authors had full access to all the Systemic therapy§ 1 (<1%) 0
data in the study and had final responsibility for the PD-L1 status
decision to submit for publication. Quantifiable 289 (95%) 297 (98%)
<1%¶ 57/289 (20%) 78/297 (26%)
Results ≥1%¶ 232/289 (80%) 219/297 (74%)
Between Nov, 29, 2016, and April 28, 2018, we enrolled
Data are median (IQR) or n (%). PD-L1=programmed cell death ligand 1. *Includes
713 patients, of whom 605 were eligible and randomly
Australia, Brazil, Chile, and South Africa. †On a score of 0 to 5, with higher scores
assigned to nivolumab plus ipilimumab (n=303) or indicating greater disability. One patient in the chemotherapy group had a
chemotherapy (n=302). 300 participants in the nivolumab baseline Eastern Cooperative Oncology Group performance status of 2 (protocol
plus ipilimumab group and 284 in the chemotherapy deviation). ‡Previous radiotherapy was provided for palliative support, pain
management, or prophylactic track irradiation for tumour biopsy. §Due to
group received at least one dose of study drug (figure 1). incorrect data entry, one patient was reported as having previous systemic cancer
At the prespecified interim analysis (database lock therapy in the nivolumab plus ipilimumab group. ¶Calculated as a proportion of
April 3, 2020), the median follow-up for overall survival quantifiable patients.
was 29·7 months (IQR 26·7–32·9), with a minimum Table 1: Baseline characteristics
of 22·1 months. Baseline characteristics were well
balanced between treatment groups (table 1). 467 (77%) of
605 participants were male and median age was 69 years treat­ment remained on treatment and no patients in the
(IQR 64–75). Overall, 456 (75%) of 605 patients had chemotherapy group remained on treatment (figure 1).
epithelioid tumour histology. The main reasons for treatment discon­ tinuation in
As of database lock, five (2%) of 300 patients in the nivolumab plus ipilimumab group were disease
the nivolumab plus ipilimumab group who received pro­gression (182 [61%] of 300) and study drug toxicity

www.thelancet.com Vol 397 January 30, 2021 379

 Articles

(59 [20%]); 25 (8%) of 300 patients completed 2 years of

immunotherapy. During the study, one patient in the
A Nivolumab plus Chemotherapy nivolumab plus ipilimumab group discontinued study
ipilimumab group group
100 drug but received subsequent therapy from the investi­
Median overall survival (95% CI), months 18·1 (16·8–21·4) 14·1 (12·4–16·2)
90 gator before BICR confirmation of disease progression.
Hazard ratio 0·74 (96·6% CI 0·60–0·91); p=0·0020
80 In the chemotherapy group, 176 (62%) of 284 patients
68% (95% CI 62·3–72·8)
70 completed all six cycles; 44 (16%) discon­tinued due to
Overall survival (%)

60 disease progression and 24 (8%) due to study drug

58% (95% CI 51·7–63·2) toxicity. Median duration of treatment was 5·6 months
50 41% (95% CI 35·1–46·5)
40 (IQR 2·0–11·4) in the nivolumab plus ipilimumab group
30 and 3·5 months (IQR 2·7–3·7) in the chemotherapy
20
27% (95% CI 21·9–32·4) group (appendix p 9). The median number of nivolumab
10
Nivolumab plus ipilimumab group doses received was 12·0 (IQR 5·0–23·5) and of
Chemotherapy group
0
ipilimumab was 4·0 (2·0–7·0). After randomisation,
0 3 6 9 12 15 18 21 24 27 30 33 36 39 104 (34%) of 302 patients in the chemotherapy group
Number at risk were given cisplatin and 180 (60%) were given
(number censored) carboplatin; 29 (28%) of 104 patients given cisplatin
Nivolumab plus 303 273 251 226 200 173 143 124 101 65 30 11 2 0
ipilimumab group (0) (2) (4) (5) (7) (11) (14) (16) (29) (49) (76) (93) (101) (103) switched to carboplatin after the first dose due to
Chemotherapy group 302 268 233 190 162 136 113 95 62 38 20 11 1 0 investigator decision. The median number of doses of
(0) (15) (18) (20) (20) (20) (21) (23) (36) (55) (66) (73) (82) (83)
cisplatin was 5·0 (IQR 3·0–6·0), of carboplatin was
B Nivolumab plus Chemotherapy 6·0 (4·0–6·0), and of pemetrexed was 6·0 (4·0–6·0).
100
ipilimumab group group Further information on treatment exposure is in the
Median overall survival (95% CI), months 18·7 (16·9–22·0) 16·5 (14·9–20·5) appendix (pp 9–10).
90
Hazard ratio 0·86 (95% CI 0·69–1·08) In the nivolumab plus ipilimumab group, 134 (44%)
80
69% (95% CI 63·0–75·0) of 303 patients were given subsequent systemic therapy,
70
Overall survival (%)

66% (95% CI 59·1–71·8) ten (3%) were given subsequent immunotherapy, and
60
131 (43%) were given subsequent chemotherapy. In the
50 42% (95% CI 35·0–48·1) chemotherapy group, 123 (41%) of 302 patients were
40
given subsequent systemic therapy, 61 (20%) were given
30 33% (95% CI 26·8–39·5)
subsequent immunotherapy, and 95 (31%) were given
20
subsequent chemotherapy (appendix p 11).
10
The study met its primary endpoint at the prespecified
0
0 3 6 9 12 15 18 21 24 27 30 33 36 39
interim analysis according to the recommendation of
Number at risk the independent Data Monitoring Committee. Given
(number censored) that the study was able to reject the null hypothesis at
Nivolumab plus 229 207 192 172 154 135 109 96 77 47 22 6 2 0
ipilimumab group (0) (1) (3) (4) (6) (7) (10) (12) (23) (40) (60) (74) (77) (79)
the interim analysis, this analysis is considered final.
Chemotherapy group 227 204 182 159 140 118 101 85 57 36 18 9 1 0 Median overall survival was 18·1 months (95% CI
(0) (11) (13) (14) (14) (14) (15) (17) (29) (45) (56) (63) (70) (71) 16·8–21·4) with nivolumab plus ipilimumab versus
C
14·1 months (12·4–16·2) with chemotherapy, with a
Nivolumab plus Chemotherapy
ipilimumab group group stratified HR of 0·74 (96·6% CI 0·60–0·91; p=0·0020;
100
Median overall survival (95% CI), months 18·1 (12·2–22·8) 8·8 (7·4–10·2) figure 2). The p value for the time-dependent covariate
90
Hazard ratio 0·46 (95% CI 0·31–0·68) was 0·9646, indicating that there was no evidence of a
80
non-constant treatment effect over time. Overall survival
70 63% (95% CI 50·9–72·9)
rates at 1 year were 68% (95% CI 62·3–72·8) versus 58%
Overall survival (%)

60
(51·7–63·2) and at 2 years were 41% (35·1–46·5) versus
50
38% (95% CI 27·0–49·5)
27% (21·9–32·4). Overall survival was similar between
40 chemotherapy regimens: median overall survival was
30 32% (95% CI 21·7–43·5) 13·7 months (95% CI 11·8–17·9) with pemetrexed plus
20 cisplatin, and 15·0 months (12·2–17·9) with pemetrexed
10
8% (95% CI 3·3–16·7) plus carboplatin (appendix p 25). Overall survival
0 favoured nivolumab plus ipilimumab across most
0 3 6 9 12 15 18 21 24 27 30 33 36 39
Time since randomisation (months)
Number at risk
(number censored)
Nivolumab plus 74 66 59 54 46 38 34 28 24 18 8 5 0 0
ipilimumab group (0) (1) (1) (1) (1) (4) (4) (4) (6) (9) (16) (19) (24) (24) Figure 2: Overall survival in all randomised patients (A) and in patients with
Chemotherapy group 75 64 51 31 22 18 12 10 5 2 2 2 0 0 epithelioid tumour histology (B) and non-epithelioid tumour histology (C)
(0) (4) (5) (6) (6) (6) (6) (6) (7) (10) (10) (10) (12) (12) The hazard ratio in part A is stratified by sex and histology. The hazard ratios in
parts B and C are from unstratified Cox proportional hazard models.

380 www.thelancet.com Vol 397 January 30, 2021

 Articles

subgroups, although survival in patients aged 75 years figure 3; appendix pp 23–24). Nonetheless, median
and older (n=157) was similar between treatment groups overall survival with nivolumab plus ipilimumab was
(figure 3). Notably, overall survival was improved similar in patients with tumours with PD-L1 expression
with nivolumab plus ipilimumab versus chemotherapy of 1% or higher (18·0 months [95% CI 16·8–21·5]) and
regardless of histology (study stratifi­ cation factor; of less than 1% (17·3 months [95% CI 10·1–24·3]);
figure 2). We found some evidence of higher treatment 1-year survival rates were 70% (95% CI 63·4–75·3) and
effect in patients with non-epithelioid histology 59% (45·5–70·9); and 2-year survival rates were 41%
(HR 0·46 [95% CI 0·31–0·68]) than in those with the (34·3–47·2) and 39% (25·9–51·3; appendix pp 23–24).
epithelioid subtype (0·86 [0·69–1·08]). Median overall Conversely, median overall survival with chemo­therapy
survival with nivolumab plus ipilimumab was similar differed between patients with PD-L1 expression of 1%
between non-epithelioid and epithelioid sub­­ types or higher (13·3 months [95% CI 11·6–15·4]) and less
(18·1 months [95% CI 12·2–22·8] vs 18·7 months than 1% (16·5 months [13·4–20·5]); 1-year survival rates
[16·9–22·0]), as were 2-year overall survival rates (38% were 55% (95% CI 48·2–61·8) and 64% (52·3–73·9);
[95% CI 27·0–49·5] vs 42% [35·0–48·1]). By contrast, and 2-year survival rates were 28% (22·1–34·7) and 25%
median overall survival with chemotherapy differed (15·5–35·0; appendix pp 23–24).
substantially between non-epithelioid and epithelioid The minimum follow-up for progression-free survival
subtypes (8·8 months [95% CI 7·4–10·2] vs 16·5 months was 19·8 months. Median progression-free survival was
[14·9–20·5]), as did 2-year overall survival rates (8% similar between treatment groups: 6·8 months (95% CI
[95% CI 3·3–16·7] vs 33% [26·8–39·5]). Overall survival 5·6–7·4) with nivolumab plus ipilimumab and 7·2 months
benefit by tumour PD-L1 expression level for nivolumab (95% CI 6·9–8·0) with chemotherapy (HR 1·00 [95% CI
plus ipilimumab versus chemotherapy was greater in 0·82–1·21]). However, progression-free survival rates at
patients with tumour expression of PD-L1 of 1% or 2 years were numerically greater with nivolumab plus
higher (HR 0·69 [95% CI 0·55–0·87]) than in patients ipilimumab (16% [95% CI 11·7–21·5]) versus chemo­
with expression of less than 1% (0·94 [0·62–1·40]; therapy (7% [4·0–11·7]; figure 4).

Number of Median overall survival (months) Unstratified hazard ratio

patients for death (95% CI)
Nivolumab plus ipilimumab Chemotherapy
group (n=303) group (n=302)

All randomly assigned 605 18·1 14·1 0·75 (0·62–0·91)

Age, years
<65 167 17·2 13·3 0·76 (0·52–1·11)
65 to <75 281 20·3 14·9 0·63 (0·48–0·83)
≥75 157 16·9 15·4 1·02 (0·70–1·48)
Sex
Male 467 17·5 13·7 0·74 (0·60–0·92)
Female 138 21·4 18·0 0·76 (0·50–1·16)
ECOG performance status*
0 242 20·7 19·5 0·87 (0·64–1·19)
1 363 17·0 11·6 0·66 (0·52–0·85)
Tumour histology
Epithelioid 456 18·7 16·5 0·86 (0·69–1·08)
Non-epithelioid 149 18·1 8·8 0·46 (0·31–0·68)
Stage of cancer
3 209 23·9 16·3 0·61 (0·44–0·86)
4 309 16·7 10·8 0·67 (0·52–0·87)
PD-L1 expression subgroups
PD-L1 <1% 135 17·3 16·5 0·94 (0·62–1·40)
PD-L1 ≥1% 451 18·0 13·3 0·69 (0·55–0·87)

0·25 0·5 1·0 2·0 4·0

Favours nivolumab plus ipilimumab Favours chemotherapy

Figure 3: Overall survival in predefined patient subgroups

ECOG=Eastern Cooperative Oncology Group. PD-L1=programmed cell death ligand 1. *One patient in the chemotherapy group had a baseline performance status of 2
(protocol deviation).

www.thelancet.com Vol 397 January 30, 2021 381

 Articles

A Nivolumab plus Chemotherapy

100 Nivolumab plus Chemotherapy ipilimumab group group
ipilimumab group group (n=303) (n=302)
90
Median progression-free survival 6·8 (5·6–7·4) 7·2 (6·9–8·0) Objective response rate
80 (95% CI), months
Progression-free survival (%)

70 n (%) 120 (40%) 129 (43%)

Hazard ratio 1·00 (95% CI 0·82–1·21)
60 95% CI 34·1–45·4 37·1–48·5
Nivolumab plus ipilimumab group
50 Chemotherapy group Best overall response
40 Complete response 5 (2%) 0
30%
30 Partial response 115 (38%) 129 (43%)

20 24% 16% Stable disease 112 (37%) 125 (41%)

10 Non-complete response and 0 3 (1%)

7% non-progressive disease
0
0 3 6 9 12 15 18 21 24 27 30 33 36 Progressive disease 55 (18%) 14 (5%)
Number at risk Unable to determine 4 (1%) 5 (2%)
(number censored)
Not reported 12 (4%) 26 (9%)
Nivolumab plus 303 198 135 89 64 52 45 36 22 15 7 2 0
ipilimumab group (0) (21) (34) (41) (51) (54) (54) (57) (66) (72) (79) (83) (85) Disease control rate
Chemotherapy group 302 222 144 71 44 33 27 21 10 6 3 1 0 n (%) 232 (77%) 257 (85%)
(0) (49) (63) (75) (81) (82) (82) (83) (86) (90) (91) (93) (93)
95% CI 71·4–81·2 80·6–88·9
B Time to response, months
100 Nivolumab plus Chemotherapy Median 2·7 2·5
ipilimumab group group
90 IQR 1·45–3·27 1·41–3·02
Median duration of response 11·0 (8·1–16·5) 6·7 (5·3–7·1)
80 Duration of response, months
Proportion with response (%)

(95% CI), months

70 Median 11·0 6·7
60 95% CI 8·1–16·5 5·3–7·1
47%
50 Proportion of patients with a response of at least 1 year or 2 years*
40 At 1 year 47% 26%
32%
30 95% CI 37–56 18–34
20 26% At 2 years 32% 8%
10 95% CI 23–41 3–15
8%
0
0 3 6 9 12 15 18 21 24 27 30 33 36 Data are n (%), unless indicated otherwise. Minimum follow-up for objective
response rate was 19·8 months. *Estimates are based on Kaplan-Meier estimates
Time since randomisation (months)
Number at risk of duration of response.
(number censored)
Nivolumab plus 120 98 74 54 45 41 37 21 12 8 2 2 0 Table 2: Tumour response, as per blinded independent central review, in
ipilimumab group (0) (5) (10) (12) (16) (16) (16) (28) (36) (39) (44) (44) (46) all patients randomly assigned to treatment
Chemotherapy group 129 99 57 33 23 19 16 8 3 1 1 0 0
(0) (8) (16) (18) (22) (23) (23) (25) (28) (29) (29) (29) (29)

Figure 4: Progression-free survival in all patients randomly assigned to treatment (A) and duration of (95% CI 23–41) in the nivolumab plus ipilimumab group
response in confirmed responders (B) versus 8% (95% CI 3–15) in the chemotherapy group.
Progression-free survival and duration of response are both per blinded independent central review. The hazard
ratio in part A is stratified by sex and histology. Safety is summarised in table 3, and all reported
grade 3 and 4 treatment-related adverse events are listed
An objective response was reported in 120 of 303 patients in the appendix (pp 13–16). Of 300 patients treated with
(40%; 95% CI 34·1–45·4) in the nivolumab plus nivolumab plus ipilimumab, 28 (9%) discontinued
ipilimumab group versus 129 of 302 patients (43%; 95% ipilimumab early. In the chemotherapy group, dose
CI 37·1–48·5) in the chemotherapy group (table 2). Com­ reduc­tions occurred in 89 (31%) of 284 participants who
plete responses were only observed in the nivolumab plus were given pemetrexed, 18 (17%) of 104 patients who
ipilimumab group (five [2%] of 303 patients). Disease were given cisplatin, and 85 (41%) of 209 participants
control was seen in 232 of 303 patients (77%; 95% CI who were given carboplatin, whereas dose reductions
71·4–81·2) with a median time to response of 2·7 months were not permitted for the nivolumab plus ipilimumab
(IQR 1·45–3·27) for the nivolumab plus ipilimumab group. Grade 3–4 treatment-related adverse events were
group versus 257 of 302 (85%; 95% CI 80·6–88·9) with a reported in 91 (30%) of 300 participants treated
median time to response of 2·5 months (IQR 1·41–3·02) with nivolumab plus ipilimumab and 91 (32%) of
for the chemotherapy group. Median duration of response 284 participants treated with chemotherapy. Any-grade
in all confirmed responders was 11·0 months (95% CI serious treatment-related adverse events were reported
8·1–16·5) in the nivolumab plus ipilimumab group versus in 64 (21%) patients treated with nivolumab plus
6·7 months (95% CI 5·3–7·1) in the chemotherapy group ipilimumab versus 22 (8%) patients treated with
(figure 4). The 2-year duration of response rate was 32% chemotherapy; grade 3–4 treatment-related serious

382 www.thelancet.com Vol 397 January 30, 2021

 Articles

Nivolumab plus ipilimumab group (n=300) Chemotherapy group (n=284)

Grade 1–2 Grade 3 Grade 4 Grade 1–2 Grade 3 Grade 4
Any 148 (49%) 79 (26%) 12 (4%) 141 (50%) 73 (26%) 18 (6%)
Diarrhoea 52 (17%) 10 (3%) 0 19 (7%) 2 (1%) 0
Pruritus 46 (15%) 3 (1%) 0 1 (<1%) 0 0
Rash 40 (13%) 3 (1%) 0 15 (5%) 0 0
Fatigue 38 (13%) 3 (1%) 0 50 (18%) 5 (2%) 0
Hypothyroidism 32 (11%) 0 0 0 0 0
Nausea 29 (10%) 1 (<1%) 0 97 (34%) 7 (2%) 0
Anaemia 5 (2%) 1 (<1%) 0 70 (25%) 32 (11%) 0
Decreased appetite 27 (9%) 2 (1%) 0 48 (17%) 2 (1%) 0
Constipation 12 (4%) 0 0 41 (14%) 1 (<1%) 0
Vomiting 8 (3%) 0 0 35 (12%) 6 (2%) 0
Asthenia 25 (8%) 0 0 32 (11%) 12 (4%) 0
Increased lipase 7 (2%) 11 (4%) 2 (1%) 0 1 (<1%) 0
Colitis 3 (1%) 7 (2%) 0 1 (<1%) 1 (<1%) 0
Increased amylase 10 (3%) 6 (2%) 1 (<1%) 1 (<1%) 0 0
Thrombocytopenia 0 2 (1%) 0 16 (6%) 4 (1%) 6 (2%)
Neutropenia 0 1 (<1%) 1 (<1%) 28 (10%) 31 (11%) 12 (4%)

Data are n (%). Safety was assessed in all patients who received at least one dose of study drug. Treatment-related adverse events with an incidence of ≥10% in any group or
grade 3 or 4 severity with an incidence of ≥2% in any group are shown. All grade 3 and 4 events are listed in the appendix (pp 13–16). Treatment-related adverse events
included those reported between the first dose of study drug and 30 days after the last dose of study drug. *Only events that led to death within 24 h were documented as
grade 5 and reported as deaths. Events leading to death >24 h after onset are reported with the worst grade before death.

Table 3: Summary of treatment-related adverse events in all treated patients*

events were reported in 46 (15%) patients treated with onset and resolution of treatment-related select adverse
nivolumab plus ipilimumab versus 17 (6%) treated with events, the proportion of patients requiring immune-
chemo­therapy (appendix pp 17–19). Any-grade treatment- modulating concomitant medication (mostly cortico­
related adverse events that led to discontinuation (due to steroids), and the duration of use of immune-modulating
either component of the regimen) were reported in concomitant medication are shown in the appendix
69 (23%) of 300 patients treated with nivolumab plus (p 21). The most commonly reported any-grade treatment-
ipilimumab and 45 (16%) of 284 patients treated with related select adverse events with nivolumab plus
chemotherapy, and 45 (15%) pa­tients treated with ipilimumab were skin (108 [36%] of 300 patients) and
nivolumab plus ipilimumab and 21 (7%) patients treated gastrointestinal (66 [22%]) events. Overall, 198 (66%) of
with chemotherapy had grade 3–4 events that led to 300 patients who were given nivolumab plus ipilimumab
discontinuation (appendix p 20). died, with 183 (61%) deaths due to disease progres­sion.
The most frequent any-grade treatment-related adverse 212 (75%) of 284 patients given chemotherapy died, with
events were diarrhoea in the nivolumab plus ipilimumab 199 (70%) deaths due to disease progression. Three (1%)
group (62 [21%] of 300 patients) and nausea in the treatment-related deaths occurred in the nivolumab plus
chemotherapy group (104 [37%] of 284 patients). The ipilimumab group, due to pneu­monitis, encephalitis, and
most frequently reported any-grade serious treatment- heart failure. One (<1%) treat­ment-related death occurred
related adverse events were colitis in the nivolumab plus in the chemotherapy group due to myelosuppression.
ipilimumab group (nine [3%]) and anaemia in the
chemotherapy group (six [2%]; appendix pp 17–19). The Discussion
median exposure time was 6·5 months (IQR 2·99–12·22) To our knowledge, CheckMate 743 is the first large,
for nivolumab plus ipilimumab and 4·5 months randomised, phase 3 study to show significant and
(3·65–4·68) for chemotherapy. Treatment exposure was clinically meaningful improvement in overall survival
220·3 person-years with nivolumab plus ipilimumab with immunotherapy versus standard-of-care platinum
and 94·5 person-years with chemotherapy. The overall plus pemetrexed chemotherapy for first-line treatment
exposure-adjusted incidence of treatment-related adverse of unresectable MPM. Based on these results, in
events was 502·1 per 100 person-years with nivolumab October, 2020, the US Food and Drug Administration
plus ipilimumab versus 1355·3 per 100 person-years with approved nivolumab plus ipilimumab for this patient
chemotherapy. population.12 With a median follow-up of 29·7 months,
A summary of treatment-related select adverse events nivolumab plus ipilimumab provided durable survival
(those with a potential immunological cause), time to benefit versus chemotherapy, with a 50% improvement

www.thelancet.com Vol 397 January 30, 2021 383

 Articles

in the 2-year overall survival rate (41% vs 27%). disease control with chemotherapy. However, radiographic
Furthermore, estimated rates of patients who still had assessments in MPM can be challenging because of the
a response at 2 years was 8% with chemotherapy absence of distinguishable tumour margins over time and
versus 32% with nivolumab plus ipilimumab. The safety successive CT evaluations.22 Thus, overall survival is
profile of nivolumab plus ipilimumab in this study was considered to be a more objective and reliable endpoint in
consistent with that seen previously in NSCLC at this this tumour type. Notably, nivolumab plus ipilimumab
dose and schedule15 and no new safety signals were provided long-term overall survival benefit, although the
reported. slight early survival benefit observed with chemotherapy
The frequencies of grade 3 or 4 serious treatment- was not durable.
related adverse events and those leading to discon­ The duration of response and durable survival benefit
tinuation were higher with nivolumab plus ipilimumab observed with nivolumab plus ipilimumab in patients
than with chemotherapy; however, most were manageable with MPM in CheckMate 743 builds on the existing body
and resolved with steroids or supportive treatment. of evidence that shows extended survival benefit with
Moreover, when treatment-related adverse events were this dual immunotherapy regimen across a number of
adjusted for exposure, the overall incidence of treatment- other tumour types, including NSCLC.13–15,23 Ipilimumab
related adverse events was lower with nivolumab plus is hypothesised to drive memory T-cell production leading
ipilimumab than with chemotherapy. to durable responses when combined with nivolumab.11
Benefit with nivolumab plus ipilimumab was observed Results of the current study also corroborate the pro­
in most subgroups assessed, with the exception of patients mising activity seen with anti-PD-1 or anti-PD-L1, and
aged 75 years or older. However, these subgroups were anti-CTLA-4 combination therapies in phase 2 studies in
small and did not have statistical power. As such, results second-line or later settings of MPM,16,18,24 and support the
from these subgroup analyses should be interpreted use of dual immunotherapy over single-agent anti-PD-1
with caution. Importantly, benefits were observed across or anti-CTLA-4 inhibitors, which have shown little benefit
histological groups, albeit at different magnitudes. For over chemotherapy.25,26
example, median overall survival with nivolumab plus Some treatment guidelines (eg, NCCN guidelines)
ipilimumab was consistent between patients with include the optional addition of the anti-angiogenic agent
epithelioid histology (18·7 months; HR 0·86 [95% CI bevacizumab to platinum plus pemetrexed chemotherapy
0·69–1·08]) and non-epithelioid histology (18·1 months; for first-line treatment of MPM in select patients, based
HR 0·46 [0·31–0·68]), showing clinically meaningful on the survival benefit seen in a phase 3 trial;5,10 however,
survival improvements across both groups; 1-year and this regimen is not approved by regulators. Nonetheless,
2-year overall survival rates were also similar between the given the durable survival benefit seen in CheckMate 743,
two histological subgroups. Notably, in the epithelioid combining nivolumab plus ipilimumab with other
subgroup, nivolumab plus ipilimumab showed an therapies, including anti-angiogenic agents or, as approved
improvement of 2 months in median overall survival for NSCLC in May, 2020, a short course of chemotherapy,12
compared with chemotherapy, with an HR favouring merits investigation to determine whether survival
nivolumab plus ipilimumab despite the 95% CI outcomes can be further enhanced. Similarly, future trials
overlapping 1. Furthermore, the 2-year overall survival rate assessing the benefit of second-line targeted therapies (eg,
in the epithelioid subgroup showed a long-term benefit of bevacizumab and ramucirumab) after nivolumab plus
nivolumab plus ipilimumab with a 9% absolute difference ipilimumab treatment are warranted.
versus chemotherapy. The larger magnitude of benefit Reliable biomarkers to predict the benefit of dual-
observed in the non-epithelioid subgroup was primarily agent immunotherapy in the treatment of MPM have not
driven by the inferior effect of chemotherapy in the non- yet been identified. Although PD-L1 expression is an
epithelioid subtype, as previously reported.3 This established biomarker for single-agent immunotherapy in
difference in outcomes between the subgroups treated NSCLC,27 its role in predicting treatment outcomes with
with chemotherapy could not be attributed to the type of dual immunotherapy regimens has not been estab­lished.
chemotherapy received because exploratory data from More specifically, in MPM trials investigating immuno­
CheckMate 743 suggest that patients derive a similar therapies, the association between PD-L1 expression and
overall survival benefit regardless of platinum backbone; efficacy is inconsistent.17,18,24 In CheckMate 743, overall
median overall survival was similar between pemetrexed survival outcomes with nivolumab plus ipilimumab were
plus cisplatin and pemetrexed plus carboplatin. similar in the subgroups with less than 1% and with 1% or
Median progression-free survival and objective response higher PD-L1 expressions and better outcomes were seen
rates were each numerically similar for nivolumab plus with nivolumab plus ipilimumab than with chemotherapy
ipilimumab and chemotherapy. Median progression-free at 2 years in both subgroups. However, survival with
survival was similar to results from previously reported chemotherapy was better in patients with tumour PD-L1
clinical trials in recurrent MPM.16,18 The progression-free expression of less than 1% than those with expression of
survival Kaplan-Meier curves crossed at approximately 1% or higher. These observations suggest that absence of
8 months, reflecting more rapid, although not durable, PD-L1 expression might be indicative of better prognosis

384 www.thelancet.com Vol 397 January 30, 2021

 Articles

with chemotherapy. However, these descriptive and Boehringer Ingelheim, Bristol Myers Squibb, F Hoffmann-La Roche,
exploratory data should be interpreted with caution given MSD, Novartis, and Pfizer (served on the advisory board, gave talks,
honoraria, and investigation in trials); received personal fees from
their potential limitations—ie, PD-L1 expression was not a Amgen, Clovis, Illumina, and Merck Serono (served on the advisory
stratification factor in the study and the sample size of the board, honoraria, and investigation in trials); received personal fees from
PD-L1 expression less than 1% group was small. As such, Takeda (gave talks, and honorarium); received personal fees from
the potential for imbalances in known or unknown Eli Lilly and Sanofi (served on the advisory board, honoraria, and gave
talks); received personal fees from AbbVie, Bayer, Biocartis, Biovent,
prognostic factors does not allow us to draw definitive Daiichi Sankyo, Debiopharm, Foundation Medicine, Janssen,
conclusions. Better characterisation of this heterogeneous Merrimack, Pharma Mar, Regeneron, Seattle Genetics, and Takeda
disease using transcriptomic and epigenetic profiling (served on the advisory board and honoraria). AST has received personal
should guide future patient selection and therapeutic fees from Bristol Myers Squibb, Eli Lilly, Genentech, Roche, Novartis,
Ariad, EMD Serono, Merck, Seattle Genetics, AstraZeneca, Boehringer
strategies, and aid in the identification of novel Ingelheim, Sellas Life Science, and Takeda (for advisory boards); and has
biomarkers.28,29 received grant support from Millenium, Polaris, Epizyme, EMD Serono,
In summary, first-line nivolumab plus ipilimumab and Seattle Genetics (for research grants). ASM’s institution received
provided a significant and clinically meaningful improve­ support from AbbVie, AstraZeneca, Bristol Myers Squibb, and
Genentech/Roche (paid honoraria to the institution); received grant
ment in overall survival versus platinum plus pemetrexed funding from Novartis and Verily (paid to institution); and ASM has
chemotherapy. Nivolumab plus ipilimumab has a received travel expenses from Roche; and ASM has acted as a
favourable clinical benefit–risk profile that has led to non-remunerated director for the Mesothelioma Applied Research
approval in the USA and should be considered as a new Foundation. SPo received personal fees from AbbVie, AstraZeneca,
Bayer, Beigene, Blueprint, Boehringer Ingelheim, Bristol Myers Squibb,
standard of care for previously untreated patients with Chugai, Daiichi Sankyo, EMD Serono, Eli Lilly, GlaxoSmithKline,
unresectable MPM, regardless of histological subtype. Guardant Health, Incyte, Janssen, Merck Sharp & Dohme, Novartis,
Contributors Pfizer, Roche, Seattle Genetics, Takeda, and Tesaro (served on the
PB, AS, AKN, NF, SPe, AST, ASM, SPo, TJ, PA, AO, CB, and GZ provided advisory board and provided consulting); received personal fees from
substantial contributions to the conception and design of the study. Elsevier (employment); and received personal fees from Paradox
PB, AS, AKN, NF, SPe, AST, ASM, SA, YO, YB, RC, LG, FG, DK, JR-C, (provided consulting). TJ has received personal fees from Atara
and GZ enrolled and treated patients. CB wrote the study statistical Pharmaceuticals; grant funding from AstraZeneca, Eli Lilly, Epizyme,
analysis plan, did all statistical analyses, and generated data. PB, AKN, Polaris, Springworks, and Trizell; and retired during manuscript
NF, SPe, AST, ASM, SPo, TJ, PA, AO, CB, and GZ analysed and development. SA has received personal fees from Bristol Myers Squibb
interpreted the data. PA and CB verified the underlying data from the (consulting or advisory role); personal fees from Achilles Biotech,
study. All authors reviewed the data, contributed to the development of Celcius Therapeutics, Memgen, Rapt Therapeutics, Venn Therapeutics,
the manuscript, and approved the final version for publication. Glympse, and Samyang (for advisory boards); personal fees from
AstraZeneca, Caris Life Science, G1 Therapeutics, GlaxoSmithKline,
Declaration of interests Merck, and Nektar (as an advisor); personal fees and non-financial
PB has received institutional grant funding from Bristol Myers Squibb support from Amgen (as an advisor and for travel fees); grant support
and MSD and has a consultancy or advisory role for Bristol Myers from Cellular Biomedicine Group (for clinical trial support); and
Squibb, MSD, Roche, Beigene, Epizyme, Takeda, Trizell, and Daichii- personal fees from EMD Serono (for a data review committee). YO has
Sankyo (all honoraria are paid to his institute). AS has received grant received personal fees from AstraZeneca, Bristol Myers Squibb, and
funding and personal fees from Bristol Myers Squibb (for provided work MSD (served on the advisory board). RC has received personal fees from
on advisory boards, consultancy, service on the speaker’s bureau, MSD and Roche (served on the advisory board), and personal fees from
provision of expert testimony, and for travel or accommodation Bristol Myers Squibb, Pfizer, and Roche (served on the speaker’s
expenses) and their institution has also received support from Bristol bureau). LG has received personal fees from AbbVie, AstraZeneca,
Myers Squibb (payment for work as a principal investigator or Boehringer Ingelheim, Bristol Myers Squibb, MSD, Novartis, Pfizer,
coprincipal investigator in clinical trials); has received personal fees Roche, and Takeda (advisory board). FG has received grant funding from
from AstraZeneca and MSD (for provided work on advisory boards, Bristol Myers Squibb; personal fees from AstraZeneca, Bristol Myers
consultancy, service on the speaker’s bureau, provision of expert Squibb, Eli Lilly, MSD, and Roche (served on the advisory board and
testimony, and for travel or accommodation expenses) and their served on the speaker’s bureau); personal fees from Amgen, Boehringer
institution also received support from AstraZeneca and MSD (payment Ingelheim, Pierre Fabre, and Pfizer (served on the speaker’s bureau);
for work as a principal investigator or coprincipal investigator in clinical and personal fees from Takeda and Bayer (served on the advisory board).
trials); and has received personal fees from Roche (for provided work on DK has received personal fees from Amgen, AstraZeneca, Boehringer
advisory boards, consultancy, service on the speaker’s bureau, provision Ingelheim, Bristol Myers Squibb, Merck, Merck Sharp & Dome, Pfizer,
of expert testimony, and for travel or accommodation expenses) and Roche, and Takeda (served on the advisory board and provided
their institution also received support from Roche (payment for work as consulting). JR-C has received other funding (sponsored research) from
a principal investigator or coprincipal investigator in clinical trials). Bristol Myers Squibb; grant support, personal fees, and non-financial
AKN has received grant funding from Atara Biotherapeutics and support (for advisory boards, as a speaker and in a research role) from
Douglas Pharmaceuticals; received non-financial, travel support, and Bristol Myers Squibb, MSD, and Roche; grant support and personal fees
grant funding from AstraZeneca; received personal fees from Bayer (for advisory boards, as a speaker and in a research role) from Takeda,
Pharmaceuticals, Pharmabcine, and Trizell (honoraria and provided Novartis, Pfizer, and AstraZeneca; personal fees (for advisory boards, as
consulting); received personal fees, non-financial, and travel support a speaker and in a research role) for Beigene; personal fees (for a
from Boehringer Ingelheim (honoraria, served on the advisory board research role) from Celltrion and Janssen; grant support and personal
and travel funding); received personal fees from Douglas fees (for advisory boards and as a speaker) from Merck and Bayer; and
Pharmaceuticals, Merck Sharp Dohme, and Roche Pharmaceuticals grant support, personal fees, and non-financial support (for advisory
(served on the advisory board and honoraria); and received personal fees boards and as a speaker) from Boehringer Ingelheim. PA was an
from Atara Biotherapeutics (served on the advisory board). NF has employee of Bristol Myers Squibb. AO and CB are employees of and
received personal fees from Bristol Myers Squibb and Daiichi Sankyo hold stocks in Bristol Myers Squibb. GZ has received grant funding
(honoraria) and received grant funding and personal fees from ONO from Inventiva and Roche; personal fees and reimbursement for
pharmaceutical (honoraria, and provided advice and consulting). SPe has attendance of international meetings from AbbVie, AstraZeneca, Bristol
received personal fees and non-financial support from AstraZeneca, Myers Squibb, Boehringer Ingelheim, Pfizer, and Roche (travel or

www.thelancet.com Vol 397 January 30, 2021 385

 Articles

accommodation expenses); personal fees from AstraZeneca, Bristol 14 Motzer RJ, Rini BI, McDermott DF, et al. Nivolumab plus
Myers Squibb, and Roche (served on the advisory board and honoraria); ipilimumab versus sunitinib in first-line treatment for advanced
personal fees from Bristol Myers Squibb and Inventiva (provided renal cell carcinoma: extended follow-up of efficacy and safety
consulting); and personal fees from MSD and Da Volterra (served on the results from a randomised, controlled, phase 3 trial. Lancet Oncol
advisory board). YB declares no competing interests. 2019; 20: 1370–85.
15 Hellmann MD, Paz-Ares L, Bernabe Caro R, et al. Nivolumab plus
Data sharing ipilimumab in advanced non-small-cell lung cancer. N Engl J Med
For the Bristol Myers Squibb The Bristol Myers Squibb policy on data sharing is available online. 2019; 381: 2020–31.
policy on data sharing see 16 Scherpereel A, Mazieres J, Greillier L, et al. Nivolumab or
Acknowledgments
https://www.bms.com/ nivolumab plus ipilimumab in patients with relapsed malignant
This study was funded by Bristol Myers Squibb. We thank the patients
researchers-and-partners/ pleural mesothelioma (IFCT-1501 MAPS2): a multicentre,
and families who participated for making this trial possible, and the
clinical-trials-and-research/ open-label, randomised, non-comparative, phase 2 trial.
investigators (appendix p 2) and clinical study teams who participated in Lancet Oncol 2019; 20: 239–53.
disclosure-commitment.html
the trial. We also thank Ama Day for contributions as protocol manager
17 Quispel-Janssen J, van der Noort V, de Vries JF, et al. Programmed
of this trial; Dako for collaborative development of the PD-L1 IHC 28-8 death 1 blockade with nivolumab in patients with recurrent
pharmDx assay; and Mhairi Laird, of Caudex (Oxford, UK), for her malignant pleural mesothelioma. J Thorac Oncol 2018; 13: 1569–76.
assistance in the preparation of the manuscript. The NCCN guidelines 18 Disselhorst MJ, Quispel-Janssen J, Lalezari F, et al. Ipilimumab and
were cited with the permission of NCCN. NCCN makes no warranties of nivolumab in the treatment of recurrent malignant pleural
any kind whatsoever regarding their content, use, or application, and mesothelioma (INITIATE): results of a prospective, single-arm,
disclaims any responsibility for their application or use in any way. phase 2 trial. Lancet Respir Med 2019; 7: 260–70.
References 19 Oken MM, Creech RH, Tormey DC, et al. Toxicity and response
1 Van Gerwen M, Alpert N, Wolf A, et al. Prognostic factors of survival criteria of the Eastern Cooperative Oncology Group.
in patients with malignant pleural mesothelioma: an analysis of the Am J Clin Oncol 1982; 5: 649–55.
National Cancer Database. Carcinogenesis 2019; 40: 529–36. 20 Byrne MJ, Nowak AK. Modified RECIST criteria for assessment of
2 Milano MT, Zhang H. Malignant pleural mesothelioma: response in malignant pleural mesothelioma. Ann Oncol 2004;
a population-based study of survival. J Thorac Oncol 2010; 5: 1841–48. 15: 257–60.
3 Billé A, Krug LM, Woo KM, Rusch VW, Zauderer MG. Contemporary 21 US Department of Health and Human Services, Food and Drugs
analysis of prognostic factors in patients with unresectable malignant Administration, Oncology Center of Excellence, Center for Drug
pleural mesothelioma. J Thorac Oncol 2016; 11: 249–55. Evaluation and Research (CDER), Center for Biologics Evaluation
and Research (CBER). Clinical trial endpoints for the approval of
4 Baas P, Fennell D, Kerr KM, Van Schil PE, Haas RL, Peters S. cancer drugs and biologics. December, 2018. https://www.fda.gov/
Malignant pleural mesothelioma: ESMO Clinical Practice media/71195/download (accessed Jan 14, 2021).
Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2015;
26 (suppl 5): v31–39. 22 US Department of Health and Human Services, Food and Drug
Administration, Oncology Center of Excellence, Center for Drug
5 Scherpereel A, Opitz I, Berghmans T, et al. ERS/ESTS/EACTS/ Evaluation and Research (CDER), Center for Biologics Evaluation
ESTRO guidelines for the management of malignant pleural and Research (CBER). Clinical trial endpoints for the approval of
mesothelioma. Eur Respir J 2020; 55: 1900953 cancer drugs and biologics: guidance for industry. December, 2018.
6 Santoro A, O’Brien ME, Stahel RA, et al. Pemetrexed plus cisplatin https://www.fda.gov/media/71195/download (accessed
or pemetrexed plus carboplatin for chemonaïve patients with July 24, 2020).
malignant pleural mesothelioma: results of the International 23 Ramalingam SS, Ciuleanu TE, Pluzanski A, et al. Nivolumab +
Expanded Access Program. J Thorac Oncol 2008; 3: 756–63. ipilimumab versus platinum-doublet chemotherapy as first-line
7 Taylor P, Castagneto B, Dark G, et al. Single-agent pemetrexed for treatment for advanced non-small cell lung cancer: three-year
chemonaïve and pretreated patients with malignant pleural update from CheckMate 227 part 1. J Clin Oncol 2020;
mesothelioma: results of an International Expanded Access 38 (suppl 15): 9500 (abstr).
Program. J Thorac Oncol 2008; 3: 764–71. 24 Calabrò L, Morra A, Giannarelli D, et al. Tremelimumab combined
8 Vogelzang NJ, Rusthoven JJ, Symanowski J, et al. Phase III study of with durvalumab in patients with mesothelioma (NIBIT-MESO-1):
pemetrexed in combination with cisplatin versus cisplatin alone in an open-label, non-randomised, phase 2 study. Lancet Respir Med
patients with malignant pleural mesothelioma. J Clin Oncol 2003; 2018; 6: 451–60.
21: 2636–44. 25 Popat S, Curioni-Fontecedro A, Dafni U, et al. A multicentre
9 van Meerbeeck JP, Gaafar R, Manegold C, et al. Randomized phase randomised phase III trial comparing pembrolizumab versus
III study of cisplatin with or without raltitrexed in patients with single-agent chemotherapy for advanced pre-treated malignant
malignant pleural mesothelioma: an intergroup study of the pleural mesothelioma: the European Thoracic Oncology Platform
European Organisation for Research and Treatment of Cancer Lung (ETOP 9-15) PROMISE-meso trial. Ann Oncol 2020; 31: 1734–45.
Cancer Group and the National Cancer Institute of Canada. 26 Maio M, Scherpereel A, Calabrò L, et al. Tremelimumab as second-
J Clin Oncol 2005; 23: 6881–89. line or third-line treatment in relapsed malignant mesothelioma
10 Zalcman G, Mazieres J, Margery J, et al. Bevacizumab for newly (DETERMINE): a multicentre, international, randomised, double-
diagnosed pleural mesothelioma in the Mesothelioma Avastin blind, placebo-controlled phase 2b trial. Lancet Oncol 2017;
Cisplatin Pemetrexed Study (MAPS): a randomised, controlled, 18: 1261–73.
open-label, phase 3 trial. Lancet 2016; 387: 1405–14. 27 Reck M, Rodríguez-Abreu D, Robinson AG, et al. Pembrolizumab
11 Wei SC, Duffy CR, Allison JP. Fundamental mechanisms of immune versus chemotherapy for PD-L1-positive non-small-cell lung cancer.
checkpoint blockade therapy. Cancer Discov 2018; 8: 1069–86. N Engl J Med 2016; 375: 1823–33.
12 Bristol Myers Squibb. OPDIVO (nivolumab) prescribing 28 Blum Y, Meiller C, Quetel L, et al. Dissecting heterogeneity in
information. US Food and Drug Administration, December, 2020. malignant pleural mesothelioma through histo-molecular gradients
https://packageinserts.bms.com/pi/pi_opdivo.pdf (accessed for clinical applications. Nat Commun 2019; 10: 1333.
Jan 14, 2021).
 29 Mansfield AS, Peikert T, Smadbeck JB, et al. Neoantigenic potential
13 Larkin J, Chiarion-Sileni V, Gonzalez R, et al. Five-year survival with of complex chromosomal rearrangements in mesothelioma.
combined nivolumab and ipilimumab in advanced melanoma. J Thorac Oncol 2019; 14: 276–87.
N Engl J Med 2019; 381: 1535–46.

386 www.thelancet.com Vol 397 January 30, 2021