See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327855792

Avelumab versus docetaxel in patients with platinum-treated advanced non-

small-cell lung cancer (JAVELIN Lung 200): an open-label, randomised, phase
3 study

Article  in  The Lancet Oncology · September 2018

DOI: 10.1016/S1470-2045(18)30673-9

CITATIONS READS

159 557

18 authors, including:

Hidenobu Ishii Marina Garassino

Kurume University Fondazione IRCCS Istituto Nazionale dei Tumori di Milano
48 PUBLICATIONS   645 CITATIONS    338 PUBLICATIONS   7,742 CITATIONS   

SEE PROFILE SEE PROFILE

Filippo de Marinis
IEO - Istituto Europeo di Oncologia
426 PUBLICATIONS   23,256 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NEPTUNE View project

Chemotherapy View project

All content following this page was uploaded by Osvaldo Aren on 16 October 2018.

The user has requested enhancement of the downloaded file.

 Articles

Avelumab versus docetaxel in patients with platinum-treated

advanced non-small-cell lung cancer (JAVELIN Lung 200):
an open-label, randomised, phase 3 study
Fabrice Barlesi, Johan Vansteenkiste, David Spigel, Hidenobu Ishii, Marina Garassino, Filippo de Marinis, Mustafa Özgüroğlu, Aleksandra Szczesna,
Andreas Polychronis, Ruchan Uslu, Maciej Krzakowski, Jong-Seok Lee, Luana Calabrò, Osvaldo Arén Frontera, Barbara Ellers-Lenz, Marcis Bajars,
Mary Ruisi, Keunchil Park

Summary
Background Antibodies targeting the immune checkpoint molecules PD-1 or PD-L1 have demonstrated clinical Lancet Oncol 2018
efficacy in patients with metastatic non-small-cell lung cancer (NSCLC). In this trial we investigated the efficacy and Published Online
safety of avelumab, an anti-PD-L1 antibody, in patients with NSCLC who had already received platinum-based therapy. September 24, 2018
http://dx.doi.org/10.1016/
S1470-2045(18)30673-9
Methods JAVELIN Lung 200 was a multicentre, open-label, randomised, phase 3 trial at 173 hospitals and cancer
See Online/Comment
treatment centres in 31 countries. Eligible patients were aged 18 years or older and had stage IIIB or IV or recurrent http://dx.doi.org/10.1016/
NSCLC and disease progression after treatment with a platinum-containing doublet, an Eastern Cooperative Oncology S1470-2045(18)30683-1
Group performance status score of 0 or 1, an estimated life expectancy of more than 12 weeks, and adequate Aix Marseille University,
haematological, renal, and hepatic function. Participants were randomly assigned (1:1), via an interactive voice-response Assistance Publique Hôpitaux
system with a stratified permuted block method with variable block length, to receive either avelumab 10 mg/kg every de Marseille, Marseille, France
(Prof F Barlesi MD); Department
2 weeks or docetaxel 75 mg/m² every 3 weeks. Randomisation was stratified by PD-L1 expression (≥1% vs <1% of of Respiratory Oncology,
tumour cells), which was measured with the 73–10 assay, and histology (squamous vs non-squamous). The primary University Hospital KU Leuven,
endpoint was overall survival, analysed when roughly 337 events (deaths) had occurred in the PD-L1-positive population. Leuven, Belgium
(Prof J Vansteenkiste MD);
Efficacy was analysed in all PD-L1-positive patients (ie, PD-L1 expression in ≥1% of tumour cells) randomly assigned to
Sarah Cannon Research
study treatment (the primary analysis population) and then in all randomly assigned patients through a hierarchical Institute, Nashville, TN, USA
testing procedure. Safety was analysed in all patients who received at least one dose of study treatment. This trial is (D Spigel MD); Division of
registered with ClinicalTrials.gov, number NCT02395172. Enrolment is complete, but the trial is ongoing. Respirology, Neurology, and
Rheumatology, Department of
Internal Medicine, Kurume
Findings Between March 24, 2015, and Jan 23, 2017, 792 patients were enrolled and randomly assigned to receive University School of Medicine,
avelumab (n=396) or docetaxel (n=396). 264 participants in the avelumab group and 265 in the docetaxel group had Kurume, Fukuoka, Japan
PD-L1-positive tumours. In patients with PD-L1-positive tumours, median overall survival did not differ significantly (H Ishii MD); Thoracic Oncology
Unit, Department of Medical
between the avelumab and docetaxel groups (11·4 months [95% CI 9·4–13·9] vs 10·3 months [8·5–13·0]; hazard ratio
Oncology, Fondazione IRCCS
0·90 [96% CI 0·72–1·12]; one-sided p=0·16). Treatment-related adverse events occurred in 251 (64%) of 393 avelumab- Istituto Nazionale dei Tumori,
treated patients and 313 (86%) of 365 docetaxel-treated patients, including grade 3–5 events in 39 (10%) and 180 (49%) Milan, Italy (M Garassino MD);
patients, respectively. The most common grade 3–5 treatment-related adverse events were infusion-related reaction Thoracic Oncology Division,
European Institute of
(six patients [2%]) and increased lipase (four [1%]) in the avelumab group and neutropenia (51 [14%]), febrile
Oncology, Milan, Italy
neutropenia (37 [10%]), and decreased neutrophil counts (36 [10%]) in the docetaxel group. Serious treatment-related (F de Marinis MD); Department
adverse events occurred in 34 (9%) patients in the avelumab group and 75 (21%) in the docetaxel group. Treatment- of Internal Medicine, Division
related deaths occurred in four (1%) participants in the avelumab group, two due to interstitial lung disease, one due of Medical Oncology,
Cerrahpaşa Medical Faculty,
to acute kidney injury, and one due to a combination of autoimmune myocarditis, acute cardiac failure, and respiratory Istanbul University, Istanbul,
failure. Treatment-related deaths occurred in 14 (4%) patients in the docetaxel group, three due to pneumonia, and Turkey (Prof M Özgüroğlu MD);
one each due to febrile neutropenia, septic shock, febrile neutropenia with septic shock, acute respiratory failure, Department of Lung Diseases,
cardiovascular insufficiency, renal impairment, leucopenia with mucosal inflammation and pyrexia, infection, Regional Lung Disease
Hospital, Otwock, Poland
neutropenic infection, dehydration, and unknown causes. (A Szczesna MD); Department
of Medical Oncology, Mount
Interpretation Compared with docetaxel, avelumab did not improve overall survival in patients with platinum-treated Vernon Cancer Centre,
PD-L1-positive NSCLC, but had a favourable safety profile. Northwood, Middlesex, UK
(A Polychronis PhD); Ege
University Hospital,
Funding Merck and Pfizer. Department of Medical
Oncology, Izmir, Turkey
Copyright © 2018 Elsevier Ltd. All rights reserved. (R Uslu MD); Centrum
Onkologii-Instytut Im M
Skłodowskiej-Curie w
Introduction non-small-cell lung cancer (NSCLC), and several of these Warszawie, Warsaw, Poland
Antibodies targeting the immune checkpoint molecules drugs have received regulatory approvals.1–3 Nivolumab (Prof M Krzakowski MD);
PD-1 or PD-L1 improve overall survival compared with and pembrolizumab (anti-PD-1 antibodies) were first Department of Internal
Medicine, Seoul National
standard-of-care chemotherapy in patients with metastatic approved in 2015 on the basis of data from randomised

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 1

 Articles

University Bundang Hospital,

Seoul National University Research in context
College of Medicine, Seoul,
South Korea (Prof J-S Lee MD); Evidence before this study Added value of this study
Center for Immuno-Oncology, Before the approval of anti-PD-1 and anti-PD-L1 immune To our knowledge, JAVELIN Lung 200 is the second phase 3,
Medical Oncology and checkpoint inhibitor antibodies, docetaxel was the mainstay randomised clinical trial of an anti-PD-L1 antibody in patients
Immunotherapy, University
Hospital of Siena, Siena, Italy
of treatment for patients with non-small-cell lung cancer with NSCLC and disease progression after platinum-based
(L Calabrò MD); (NSCLC) and disease progression after platinum-based chemotherapy. It differs from earlier phase 3 trials of anti-PD-1
Centro Investigación Clínica chemotherapy. We searched PubMed with the terms or anti-PD-L1 monotherapy because patients were enrolled at a
Bradford Hill, Santiago, Chile (“non-small cell lung cancer” OR “NSCLC”) AND (“PD-1” time when an increasing proportion of patients had access to
(O Arén Frontera MD); Merck,
Darmstadt, Germany
OR “PD-L1” OR “checkpoint inhibitor”) for clinical trials of an approved agent of these classes. In our study, avelumab was
(B Ellers-Lenz MS); EMD Serono, checkpoint inhibitors in NSCLC published in English from not associated with increased overall survival compared with
Billerica, MA, USA (M Bajars MD, database inception up to May 11, 2018. We identified 29 trials docetaxel in patients with PD-L1-positive tumours (defined as
M Ruisi MD); and Division of in patients with NSCLC, 11 of which exclusively enrolled or PD-L1 expression in ≥1% of tumour cells based on the
Hematology-Oncology,
Samsung Medical Center,
enrolled a specified cohort of patients who had received 73-10 assay). However, exploratory analyses showed that
Sungkyunkwan University previous platinum treatment—including trials of atezolizumab, avelumab was associated with improved overall survival relative
School of Medicine, Seoul, avelumab, durvalumab, nivolumab, and pembrolizumab. Of to docetaxel in patients whose tumours had higher levels of
South Korea (Prof K Park MD) these trials, five were randomised trials of anti-PD-1 or PD-L1 expression (ie, at ≥50% and ≥80% cutoffs). The overall
Correspondence to: anti-PD-L1 monotherapy compared with docetaxel (one safety profile of avelumab was better than that of docetaxel.
Prof Keunchil Park, Division of
phase 2 trial, one phase 2–3 trial, and three phase 3 trials); all
Hematology-Oncology, Samsung Implications of all the available evidence
Medical Center, Sungkyunkwan were international and open label. In all randomised phase 3
Although avelumab had clinical activity in patients with NSCLC
University School of Medicine, trials, overall survival with anti-PD-1 antibodies (nivolumab
Seoul 135-710, South Korea who had disease progression after platinum-based
and pembrolizumab) and an anti-PD-L1 antibody
kpark@skku.edu chemotherapy, it did not significantly improve overall survival
(atezolizumab) was significantly longer than that with
compared with docetaxel. Furthermore, the improved efficacy of
docetaxel, including in trials specifically done in patients with
avelumab relative to docetaxel in patients whose tumours had
squamous tumours, non-squamous tumours, or
higher levels of PD-L1 expression could support the use of
PD-L1-positive tumours, or all subpopulations. All reports of
tumour-cell PD-L1 expression as a biomarker to identify patients
randomised trials identified were published after enrolment
more likely to obtain a clinical benefit with agents of this class.
had begun for our trial of avelumab.

trials—two trials4,5 of nivolumab in patients unselected for Avelumab is a human anti-PD-L1 IgG1 antibody that
PD-L1 expression, and a trial6 of pembrolizumab in has durable antitumour activity and a tolerable safety
patients with PD-L1-positive disease (defined as PD-L1 profile in a range of solid tumours.10–13 In a large
expression in ≥1% of tumour cells based on the PD-L1 phase 1 study of avelumab in patients with various solid
IHC 22C3 pharmDx assay; Dako, Carpinteria,CA, USA)— tumours (n=1758), a cohort of patients with NSCLC who
showing significantly longer overall survival with each had already received platinum-based therapy (n=184)
drug than with docetaxel in patients with advanced received avelumab as second-line or later treatment.14 In
squamous or non-squamous NSCLC and disease pro­ this cohort, the proportion of patients who achieved an
gression after platinum treatment. In 2016, atezolizumab objective response was 14%, which increased with
became the first anti-PD-L1 antibody approved for increasing levels of PD-L1 expression. Here, we report the
treatment of metastatic NSCLC after platinum therapy on results of a phase 3 trial of avelumab versus docetaxel in
the basis of the results of the phase 3 OAK trial,7 in which patients with advanced NSCLC and disease progression
overall survival was significantly longer with atezolizumab after platinum-based chemotherapy. To our knowledge,
than with docetaxel in patients with squamous or non- ours is the second randomised trial of an anti-PD-L1
squamous tumours, irrespective of PD-L1 status. In the antibody in this setting.
trials4–6 of nivolumab and pembrolizumab, the relative
treatment benefit of each drug compared with docetaxel Methods
was greater in patients with squamous tumours than in Study design and participants
those with non-squamous disease. However, across all JAVELIN Lung 200 (EMR 100070-004) is an open-label,
trials,4–7 median overall survival in patients with non- multicentre, randomised phase 3 trial at 173 hosptials and
squamous tumours was still significantly longer in the cancer treatment centres in 31 countries (Argentina,
experimental groups than in the docetaxel groups. Since Australia, Belgium, Brazil, Bulgaria, Chile, Colombia,
these trials, positive outcomes from phase 3 trials of Croatia, Czech Republic, Denmark, Estonia, France,
anti-PD-1 and anti-PD-L1 antibodies in the context of Hungary, Israel, Italy, Japan, Latvia, Mexico, Peru, Poland,
earlier NSCLC have led to additional regulatory approvals, Romania, Russia, Slovakia, South Africa, South Korea,
further changing the treatment landscape in NSCLC and Spain, Switzerland, Taiwan, Turkey, UK, and USA).
improving patient outcomes.8,9 Eligible patients were aged 18 years or older and had

2 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

 Articles

histologically confirmed stage IIIB, IV, or recurrent primary analysis population (ie, the PD-L1-positive
NSCLC with disease progression after previous platinum population) was defined as patients with PD-L1 expression
doublet treatment, available fresh biopsy or tumour in 1% or more of tumour cells. Tumour samples stained
archival material for biomarker assessment, an Eastern with the 73-10 assay were subsequently rescored to
Cooperative Oncology Group performance status score identify subgroups with PD-L1 expression in 50% or
of 0 or 1, an estimated life expectancy of more than more and 80% or more of tumour cells.
12 weeks, and adequate haematological (white blood Participants in the avelumab group received 10 mg/kg
cell count ≥2∙5  × 10⁹ per L, absolute neutrophil avelumab intravenously over 1 h once every 2 weeks.
count ≥1∙5 × 10⁹ per L, lymphocyte count ≥0∙5 × 10⁹/L, An antihistamine and paracetamol (eg, oral diphen­
platelet count >100 × 10⁹ per L, and haemoglobin ≥90 g/L hydramine 25–50 mg and oral or intravenous paracetamol
[transfused blood permitted]), renal (estimated creatine 500–600 mg, or equivalent) were given 30–60 min before
clearance >30 mL/min according to the Cockcroft-Gault each infusion of avelumab. Participants in the docetaxel
formula or local institutional standard method), and group received 75 mg/m² docetaxel intravenously over 1 h
hepatic (total bilirubin concentration ≤1∙5 × upper limit of every 3 weeks according to label instructions and local
normal, and aspartate aminotransferase and alanine guidelines. Dexamethasone was given before each
aminotransferase concentrations ≤2∙5 × upper limit of infusion of docetaxel according to local standard of
normal) function. The presence of measurable disease care. Treatment was continued in both groups until
was not an inclusion criterion. Patients were not eligible unacceptable toxicity, progressive disease, substantial
if they had non-squamous cell NSCLC harbouring an clinical deterioration, or any other protocol-specified
EGFR or ALK mutation (before the second protocol criterion for withdrawal occurred, including occurrence of
amendment on July 10, 2015, patients with tumours with an exclusion criterion, withdrawal of consent, non-
EGFR mutations were eligible if they had also received compliance, or use of an unpermitted concomitant drug
previous treatment with a tyrosine kinase inhibitor) or if (appendix p 7). For patients in the avelumab group,
they had previously received a drug targeting a T-cell treatment could continue beyond initial disease
regulatory protein or systemic anticancer treatment after progression according to the Response Evaluation Criteria
disease progression with platinum-based combination in Solid Tumours (RECIST; version 1.1)17 if their
therapy. Other exclusion criteria included brain performance status remained stable, they were tolerating
metastases (unless treated locally and not associated with treatment, and if investigators decided that treatment
ongoing neurological symptoms related to brain beyond progression would not delay interventions to
localisation of disease), persisting toxicity after previous prevent serious complications of disease progression.
treatment, or other clinically significant diseases. Full Dose modification of avelumab was not permitted.
eligibility criteria are in the appendix (p 6). However, in patients in whom infusion was stopped early See Online for appendix
The study protocol (appendix p 23) was approved by because of an adverse event during the infusion, a dose
institutional review boards and ethics committees at reduction was defined as a patient receiving a non-zero
each institution. The study was done in accordance with dose of <90% of the planned dose. Dose modifications
the trial protocol, Good Clinical Practice guidelines, and and discontinuations of docetaxel were made in
the Declaration of Helsinki. All patients provided written accordance with label instructions and local guidelines.
informed consent. No crossover to avelumab was permitted. Treatment was
discontinued permanently in both groups in the event of
Randomisation and masking any grade 3 or worse treatment-related adverse events
Patients were enrolled by study investigators and were (with the exception of protocol-specified transient events).
randomly assigned (1:1) via an interactive voice-response Treatment was delayed in patients with grade 2 treatment-
system to either avelumab or docetaxel. The system was related adverse events that had not resolved to grade 1 or
used by study investigators, who were potentially assisted less by the time of the next scheduled infusion. Guidance
by colleagues and study coordinators. We used stratified for delaying or discontinuing treatment after adverse
permuted block random­ isation with variable block events is in the appendix (p 8).
length. Allocation was stratified by PD-L1 status Tumours were assessed by radiographic imaging at
(expression in ≥1% vs <1% of tumour cells) and NSCLC baseline, every 6 weeks for the first 12 months, then
histology (squamous vs non-squamous). The trial was every 12 weeks thereafter. Objective tumour response
open label, so neither investigators nor patients were was assessed according to RECIST (version 1.1) by
masked to assigned study treatments. blinded independent endpoint review committee (IERC).
Treatment decisions were based on investigator assess­
Procedures ments. Safety assessments were done at each treatment
PD-L1 expression in tumour tissue was assessed visit. Adverse events were coded in accordance with
centrally at baseline with the PD-L1 IHC 73-10 pharmDx MedDRA and graded according to the US National
assay, which identifies a higher proportion of PD-L1- Cancer Institute’s Common Terminology Criteria for
positive tumour cells than do other PD-L1 assays.15,16 The Adverse Events (version 4.03). Infusion-related reactions

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 3

 Articles

Outcomes
1399 patients assessed for eligibility The primary endpoint was overall survival, which was
defined as time from randomisation to death (irrespective
607 ineligible at screening
of cause). Secondary endpoints were progression-free
497 did not meet eligibility criteria survival (defined as time from randomisation until the
38 withdrew consent first documentation of objective progressive disease
13 adverse events
16 deaths according to RECIST and adjudicated by the IERC or
43 other reasons death from any cause, whichever occurred first), best
overall response (defined as the best response obtained
792 enrolled and randomly assigned among all tumour assessments after the date of
randomisation until documented disease progression,
according to RECIST (version 1.1) and adjudicated by the
IERC), changes in patient-reported outcomes or quality
396 assigned to avelumab (full analysis set) 396 assigned to docetaxel (full analysis set) of life (assessed by the EQ-5D, the European Organisation
393 received at least one dose 365 received at least one dose
(safety analysis set) (safety analysis set) for Research and Treatment of Cancer’s QLQ-C30, and
3 did not receive avelumab 31 did not receive docetaxel module QLQ-LC13 instruments), and safety. Prespecified
exploratory endpoints were duration of response (time
from complete or partial response until disease pro­
132 PD-L1 negative 131 PD-L1 negative
gression or death), tumour shrinkage in target lesions,
serum titres of anti-avelumab antibodies, the pharma­
264 included in primary analysis population 265 included in primary analysis population cokinetic profile of avelumab, the relationship between
(PD-L1-positive) (PD-L1-positive)
PD-L1 expression and clinical response parameters, and
assessment of potential biomarkers. Quality of life
221 discontinued treatment 241 discontinued treatment assessments, tumour shrinkage, pharmacokinetic data,
161 progressive disease 151 progressive disease and biomarker assessments are not reported in this
35 adverse events 40 adverse events
15 deaths 11 deaths paper.
7 withdrew consent 17 withdrew consent
3 other reasons 20 other reasons
2 did not receive treatment 2 lost to follow-up Statistical analysis
20 did not receive treatment We planned to enrol around 750 patients, which we
estimated would include 522 patients with PD-L1-positive
41 still on treatment 4 still on treatment
tumours and 337 events (deaths) in the primary analysis
population at the primary analysis. An interim analysis
was planned after approximately 253 events had occurred
Figure 1: Trial profile
in the PD-L1-positive population. Efficacy was analysed in
with avelumab were also assessed with a composite all patients randomly assigned to study treatment
definition that comprised infusion-related reaction, drug (intention-to-treat analysis) and in the PD-L1-positive
hyper­sensitivity, or hypersensitivity reaction occurring popu­lation (primary analysis set). Safety was analysed in
on the day of or the day after the infusion, or various all patients who received at least one dose of study
signs and symptoms of infusion-related reaction treatment. Overall survival was analysed with a one-sided
(including chills, pyrexia, back pain, hypersensitivity, log-rank test (overall α 2·5%) stratified for PD-L1 status
drug hypersensitivity, type I hypersensitivity, dyspnoea, and tumour histology. A hierarchical test procedure was
hypotension, and flushing) occurring on the day of the applied in the following order to control the overall
infusion and resolving within 2 days. Immune-related significance level at 0·025 (one sided): overall survival in
adverse events were identified with a prespecified list of the PD-L1-positive population, overall survival in the
adverse events followed by a comprehensive medical full analysis set, best overall response in the PD-L1-positive
review. Blood and urine samples were taken from non- population, progression-free survival in the PD-L1-positive
fasted patients at each trial visit for routine laboratory population, best overall response in the full analysis
analyses, including core serum chemistry, haematology, set, and progression-free survival in the full analysis set.
haemostaseology, and urinalysis. Full serum chemistry The α value in the primary analysis was 0·02041.
was assessed at weeks 3, 5, and 13 in the avelumab group Adjustment for multiple testing because of the interim
and at weeks 4, 7, and 13 in the docetaxel group. analysis was done according to the O’Brien-Fleming
Thereafter, serum chemistry was assessed every 6 weeks approach and based on the observed number of overall
in both groups. Free thyroxine and thyroid-stimulating survival events according to the Lan-DeMets method.
hormone concentrations were tested every 6 weeks in the Accordingly, CIs were adjusted to 96%. Treatment effects
avelumab group, and at week 13 and week 25 in the were estimated with a Cox proportional hazards model
docetaxel group; concentrations of all other hormones stratified by the randomisation strata (PD-L1 status and
were tested in both groups when clinically indicated. tumour histology) to calculate hazard ratios (HRs) and

4 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

 Articles

adjusted 96% CIs. The proportional hazards assumption PD-L1-positive population Full analysis set
was checked visually for the primary analysis by plotting Avelumab group Docetaxel group Avelumab group Docetaxel
log(−log(overall survival)) versus log(time) within each (n=264) (n=265) (n=396) group (n=396)
random­isation stratum. Additionally, Schoenfeld residuals, Age, years 64 (59–70) 63 (56–69) 64 (58–69) 63 (57–69)
including a locally weighted smoothing (LOESS) curve, Sex
were plotted to investigate graphically violations of Male 182 (69%) 185 (70%) 269 (68%) 273 (69%)
the proportional hazards assumption. We analysed Female 82 (31%) 80 (30%) 127 (32%) 123 (31%)
progression-free survival with the same log-rank test used Country or region
for overall survival. The proportion of patients with an
USA 9 (3%) 8 (3%) 10 (3%) 8 (2%)
objective response (ie, a confirmed best overall response of
Latin America 35 (13%) 28 (11%) 56 (14%) 44 (11%)
complete or partial response) was calculated with
Western Europe 62 (23%) 62 (23%) 96 (24%) 102 (26%)
corresponding two-sided exact Clopper-Pearson 95% CIs
Eastern Europe 55 (21%) 52 (20%) 79 (20%) 75 (19%)
and compared with the Cochran-Mantel-Haenszel test,
Australia 2 (1%) 3 (1%) 2 (1%) 3 (1%)
accounting for stratifi­cation factors. Exploratory analysis of
Asia* 69 (26%) 80 (30%) 100 (25%) 113 (29%)
PD-L1-positive subgroups (ie, ≥50% and ≥80% cutoffs) was
South Africa 3 (1%) 0 3 (1%) 0
prespecified in the statistical plan (statistical tests within
Middle East 29 (11%) 32 (12%) 50 (13%) 51 (13%)
subgroups were two sided and unstratified). To explore
Race
the interaction between treatment and subgroup factors,
White 182 (69%) 170 (64%) 273 (69%) 262 (66%)
we did likelihood ratio tests. Time-to-event end­ points
were estimated via the Kaplan-Meier method, and Black 3 (1%) 1 (<1%) 5 (1%) 1 (<1%)

two-sided CIs for the median were calculated with the Asian 71 (27%) 81 (31%) 102 (26%) 114 (29%)

Brookmeyer-Crowley method. Follow-up times for overall Native American or Alaskan 0 1 (<1%) 0 1 (<1%)
and progression-free survival were estimated with the Native Hawaiian or other 1 (<1%) 0 1 (<1%) 0
Pacific Islander
inverse Kaplan-Meier method. A Cox model based on an
Other 7 (3%) 12 (5%) 15 (4%) 18 (5%)
inverse probability of censoring weighting technique was
Eastern Cooperative Oncology Group performance status
used for a post-hoc analysis of the effect of post-study treat­
0 96 (36%) 91 (34%) 144 (36%) 134 (34%)
ment.18 Statistical analyses were done in SAS (version 9.4).
1 168 (64%) 174 (66%) 252 (64%) 262 (66%)
This study is registered with ClinicalTrials.gov, number
Months since diagnosis of 11 (2–124) 10 (3–157) 12 (2–126) 10 (3–157)
NCT02395172.
metastatic disease
Histology
Role of the funding source
Squamous 88 (33%) 92 (35%) 120 (30%) 122 (31%)
Merck provided the study drug and worked with a study
Non-squamous 176 (67%) 173 (65%) 276 (70%) 274 (69%)
steering committee and investigators on the trial design
M category at study entry
and plan, collection and analysis of data, and interpretation
M0 15 (6%) 17 (6%) 22 (6%) 25 (6%)
of results. Funding for a professional medical writer with
M1, M1a, or M1b 248 (94%) 247 (93%) 373 (94%) 370 (93%)
access to the data was provided by Merck and Pfizer.
Missing 1 (<1%) 1 (<1%) 1 (<1%) 1 (<1%)
FB and the corresponding author had full access to all the
CNS metastasis at baseline 27 (10%) 22 (8%) 46 (12%) 33 (8%)
study data and had final responsibility for the decision to
Activating EGFR mutation 5 (2%) 7 (3%) 11 (3%) 13 (3%)
submit for publication.
ALK rearrangement 1 (<1%) 0 1 (<1%) 1 (<1%)

Results Smoking status

Between March 24, 2015, and Jan 23, 2017, 792 patients Current or former smoker 220 (83%) 224 (85%) 324 (82%) 333 (84%)
were enrolled and randomly assigned: 396 to the avelumab Never smoker 43 (16%) 41 (15%) 70 (18%) 63 (16%)
group and 396 to the docetaxel group. 264 (67%) participants PD-L1 status (≥1% cutoff)†
in the avelumab group and 265 (67%) in the docetaxel Positive ·· ·· 264 (67%) 263 (66%)
group had PD-L1-positive tumours (figure 1). The data Negative ·· ·· 129 (33%) 132 (33%)
cutoff date was Nov 22, 2017. Baseline characteristics were Not assessable ·· ·· 3 (1%) 1 (<1%)
similar between the PD-L1-positive population and the full PD-L1 status (≥50% cutoff)
analysis set (table 1). In the PD-L1-positive popu­ lation, Positive ·· ·· 168 (42%) 147 (37%)
262 (99%) patients in the avelumab group and 245 (92%) in Negative ·· ·· 218 (55%) 243 (61%)
the docetaxel group received at least one dose of study Not assessable ·· ·· 10 (3%) 6 (2%)
treatment. 18 (90%) of the 20 PD-L1-positive participants in PD-L1 status (≥80% cutoff)
the docetaxel group who did not receive at least one dose Positive ·· ·· 120 (30%) 106 (27%)
withdrew consent, and thus information about subsequent Negative ·· ·· 266 (67%) 284 (72%)
treatment was scarce. Median duration of treatment in the Not assessable ·· ·· 10 (3%) 6 (2%)
PD-L1-positive population was 3·4 months (IQR 1·6–9·7) (Table 1 continues on next page)
in the avelumab group and 2·8 months (1·4–4·4) in the

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 5

 Articles

Overall survival in the PD-L1-positive population did not

PD-L1-positive population Full analysis set
differ between the two treatment groups. Median overall
Avelumab group Docetaxel group Avelumab group Docetaxel survival was 11·4 months (95% CI 9·4–13·9) in the
(n=264) (n=265) (n=396) group (n=396)
avelumab group versus 10·3 months (8·5–13·0) in
(Continued from previous page)
the docetaxel group (stratified HR 0·90 [96% CI
Previous lines of therapy
0·72–1·12], p=0·16; figure 2A). No major violations of
One 233 (88%) 238 (90%) 351 (89%) 353 (89%)
the proportional hazards assumption for overall survival
Two 30 (11%) 26 (10%) 43 (11%) 40 (10%) were identified.
Three 1 (<1%) 1 (<1%) 2 (1%) 3 (1%) In our subsequent efficacy analyses, which were now
Previous lines of therapy for locally advanced or metastatic disease rendered exploratory in nature, median overall survival
One 255 (97%) 257 (97%) 382 (96%) 381 (96%) in the full analysis set was 10·5 months (95% CI
Two 9 (3%) 7 (3%) 14 (4%) 14 (4%) 9·2–12·9) in the avelumab group and 9·9 months
Three 0 1 (<1%) 0 1 (<1%) (8·1–11·8) in the docetaxel group (stratified HR 0·90
Data are n (%) or median (IQR). *Includes Japan (36 [14%] vs 38 [14%] in the PD-L1-positive population and
[96% CI 0·75–1·08]; nominal p=0·12; appendix p 15).
48 [12%] vs 53 [13%] in the full analysis set), South Korea (31 [12%] vs 39 [15%] in the PD-L1-positive population and The overall survival results in the PD-L1-positive
47 [12%] vs 53 [13%] in the full analysis set), and Taiwan (two [<1%] vs three [1%] in the PD-L1-positive population population were consistent across most subgroups
and five [1%] vs seven [2%] in the full analysis set). †The number of patients with PD-L1-positive tumours based on a
examined (figure 3). However, prespecified exploratory
≥1% cutoff in the full analysis set is slightly different from that in the primary analysis population because of minor
differences in PD-L1 status reporting in the randomisation system. analysis of overall survival at higher PD-L1 cutoffs
showed improved overall survival with avelumab
Table 1: Baseline characteristics
compared with docetaxel in patients whose tumours had
PD-L1 expression in 50% or more of tumour cells
docetaxel group. The median number of infusions (figure 2B) or 80% or more of tumour cells (figure 2C).
received was seven (IQR 3–21) in the avelumab group and Prespecified exploratory analyses of overall survival by
four (2–6) in the docetaxel group. In the PD-L1-positive tumour histology in the PD-L1-positive population and
population, 41 (16%) participants in the avelumab group post-hoc analyses of overall survival by tumour histology
and four (2%) in the docetaxel group remained on in patients with PD-L1 expression in 80% or more of
treatment at data cutoff (figure 1). Figure 1 shows the tumour cells are shown in the appendix (pp 16–17).
reasons for treatment discontinuation in each treatment In the PD-L1-positive population, 323 of 529 patients had
group. a progression event (disease progression or death) at data
In the PD-L1-positive population, anticancer treatment cutoff, including 180 (68%) of 264 participants in the
after the study was received by 105 (40%) of 264 patients in avelumab group and 143 (54%) of 265 in the docetaxel
the avelumab group and 126 (48%) of 265 in the docetaxel group. In the full analysis set, 497 of 792 patients had a
group, and proportions were similar in the full analysis progression event, including 286 (72%) of 396 participants
set (157 [40%] of 396 vs 185 [47%] of 396). Post-study in the avelumab group and 211 (53%) of 396 in the
therapy in the PD-L1-positive population included an docetaxel group. In the PD-L1-positive population, median
immune checkpoint inhibitor (anti-PD-1, anti-PD-L1, or progression-free survival according to IERC judgment was
anti-CTLA-4 inhibitor) in 15 (6%) patients in the avelumab 3·4 months (95% CI 2·7–4·9) in the avelumab group
group and 70 (26%) in the docetaxel group (16 [4%] vs and 4·1 months (3·0–5·3) in the docetaxel group (strati­
104 [26%] in the full analysis set). In the docetaxel group, fied HR 1·01 [96% CI 0·80–1·28]; nominal p=0·53;
post-study immune checkpoint inhibitor treatment was appendix p 18). In the full analysis set, median progression-
more frequently received by patients with non-squamous free survival was 2·8 months (95% CI 2·7–3·5) in the
(58 [34%] of 173) than with squamous (12 [13%] of 92) avelumab group and 4·2 months (3·3–5·2) in the docetaxel
tumours. At data cutoff, 342 of 529 participants in the group (unstratified HR 1·16 [95% CI 0·97–1·40]; nominal
PD-L1-positive population had died: 169 (64%) patients in p=0·95). In the subgroup of patients with PD-L1 expression
the avelumab group and 173 (65%) in the docetaxel group. in 50% or more of tumour cells, median progression-free
At data cutoff, in the full analysis set, 520 of 792 patients survival in the avelumab group was significantly longer
had died: 257 (65%) of 396 participants in the avelumab than that in the docetaxel group (appendix p 19). Median
group and 263 (66%) of 396 in the docetaxel group. progression-free survival was also longer in the avelumab
Median follow-up for overall survival in the PD-L1- group than in the docetaxel group in the subgroup of
positive population was 18·9 months (IQR 13·5–23·1) patients with PD-L1 expression in 80% or more of tumour
in  the avelumab group, 17·8 months (12·6–22·4) in cells (appendix p 19). Prespecified exploratory analyses of
the docetaxel group, and 18·3 months (13·2–22·7) in progression-free survival based on different PD-L1 cutoffs
the overall PD-L1 population. Median follow-up for and post-hoc analyses of progression-free survival by
overall survival was similar in the full analysis set histology based on different PD-L1 cutoffs are in the
(18·9 months [IQR 13·2–23·0] in the avelumab group, appendix (pp 18–19 and p 20, respectively).
17·8 months [12·5–22·4] in the docetaxel group, and There was some discordance between investigator and
18·3 months [12·9–22·9] in the whole population). IERC assessments, whereby progressive disease was

6 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

 Articles

diagnosed by the investigator but not the IERC. This

discordance was noted more frequently in the docetaxel A
group than in the avelumab group (appendix p 9). 100 Avelumab group Docetaxel group
(n=264) (n=265)
In the PD-L1-positive population, a higher proportion 90
Median overall survival 11·4 (9·4–13·9) 10·3 (8·5–13·0)
of participants had objective responses in the avelumab 80 (95% CI), months
group than in the docetaxel group (50 [19%] of 264 vs 70
Stratified HR (96% CI) 0·90 (0·72–1·12)
Stratified p value (one-sided) 0·16
31 [12%] of 265; odds ratio [OR] 1·76 [95% CI 1·08–2·86];

Overall survival (%)

60
nominal p=0·011; table 2). In prespecified analyses,
50
objective responses were recorded significantly more
40
frequently with avelumab than with docetaxel in the
subgroups of patients with PD-L1 expression in 50% or 30
more cells and in 80% or more cells (table 2). Median 20
duration of response was not reached in the avelumab 10 Avelumab
Docetaxel
group (95% CI 9·9 to not estimable) and was 6·9 months 0
(3·5 to not estimable) in the docetaxel group. Reasons Number at risk
0 3 6 9 12 15 18 21 24 27 30
for patients in the PD-L1-positive population not being (number censored)
Avelumab group 264 (0) 218 (2) 172 (2) 150 (2) 109 (21) 73 (43) 52 (55) 29 (72) 16 (83) 5 (91) 0 (95)
assessable for best overall response are in the appendix Docetaxel group 265 (0) 210 (6) 160 (7) 132 (14) 94 (33) 65 (48) 38 (62) 18 (76) 10 (83) 2 (90) 0 (92)
(p 10). In the full analysis set, 59 (15%) of 396 patients in
the avelumab group and 44 (11%) of 396 in the docetaxel B Avelumab group Docetaxel group
group had an objective response (OR 1·40 [95% CI (n=168) (n=147)
100
0·92–2·13]; nominal p=0·055): five (1%) versus two (1%) Median overall survival 13·6 (10·1–18·5) 9·2 (5·8–12·4)
90 (95% CI), months
had complete responses, 54 (14%) versus 42 (11%) had Unstratified HR (95% CI) 0·67 (0·51–0·89)
partial responses, and 129 (33%) versus 158 (40%) had 80 Unstratified p value (two-sided) 0·0052
pinteraction 0·0052
stable disease. 70
Overall survival (%)

Exploratory post-hoc analyses of the effect of treatment 60

with an immune checkpoint inhibitor after discon­ 50
tinuation of study treatment on overall survival resulted in
40
an adjusted HR for avelumab versus docetaxel in an
30
inverse probability of censoring weighting model of the
20
PD-L1-positive population of 0·80 (95% CI 0·62–1·04). In
a naive sensitivity analysis in which patients were censored 10
at the date of subsequent anticancer therapy, the adjusted 0
0 3 6 9 12 15 18 21 24 27 30
HR was 0·86 (95% CI 0·68–1·09; appendix pp 11, 22). Number at risk
Avelumab was associated with fewer treatment-related (number censored)
Avelumab group 168 (0) 145 (2) 117 (2) 102 (2) 76 (15) 52 (31) 38 (40) 23 (52) 12 (62) 4 (69) 0 (72)
adverse events than was docetaxel, and most adverse Docetaxel group 147 (0) 110 (3) 80 (3) 71 (4) 50 (15) 37 (19) 20 (27) 11 (34) 6 (38) 1 (43) 0 (44)
events associated with avelumab were low grade (table 3).
Treatment-related adverse events of any grade occurred C
Avelumab group Docetaxel group
in  251 (64%) of 393 avelumab-treated patients and (n=120) (n=106)
313 (86%) of 365 docetaxel-treated patients. The pro­ 100 Median overall survival 17·1 (10·6–25·0) 9·3 (5·8–13·1)
(95% CI), months
portion of patients with a grade 3 or worse (39 [10%] of 90
Unstratified HR (95% CI) 0·59 (0·42–0·83)
393 vs 180 [49%] of 365) or grade 4 or worse (eight [2%] vs 80 Unstratified p value (two-sided) 0·0022
79 [22%]) treatment-related adverse event was lower in pinteraction 0·0023
70
the avelumab group than in the docetaxel group (appendix
Overall survival (%)

60
pp 12–13). In the avelumab group, the most common
50
treatment-related adverse events of any grade were
40
infusion-related reaction (65 [17%] of 393) and decreased
appetite (34 [9%]; table 3, appendix pp 12–13), and the 30

most common grade 3 or worse treatment-related adverse 20

events were infusion-related reaction (six [2%]) and 10
increased lipase (four [1%]; table 3, appendix pp 12–13). In 0
the docetaxel group, the most common treatment-related 0 3 6 9 12 15 18 21 24 27 30
Number at risk Time since treatment initiation (months)
adverse events of any grade were alopecia (97 [27%] of (number censored)
365), anaemia (69 [19%]), and decreased appetite Avelumab group 120 (0) 105 (2) 85 (2) 77 (2) 59 (11) 43 (23) 31 (31) 21 (38) 12 (46) 4 (53) 0 (56)
(66 [18%]), and the most common grade 3 or worse Docetaxel group 106 (0) 77 (3) 58 (3) 52 (4) 35 (13) 26 (15) 14 (21) 8 (26) 4 (29) 0 (33) 0 (33)

treatment-related adverse events were neutropenia Figure 2: Overall survival in the PD-L1-positive population at the ≥1% (A), ≥50% (B), and ≥80% (C) cutoffs
(51 [14%]), febrile neutropenia (37 [10%]), and decreased HR=hazard ratio.
neutrophil count (36 [10%]; table 3; appendix pp 12–13). Figure 2A was adjusted for multiple comparisons.

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 7

 Articles

Events/patients (n/N) Median overall survival (95% CI), months Hazard ratio (95% CI)
Avelumab group Docetaxel group Avelumab group Docetaxel group
Age (years)
<65 87/134 98/145 10·5 (8·8–13·6) 9·2 (6·4–11·9) 0·84 (0·63–1·13)
≥65 82/130 75/120 12·9 (6·6–15·6) 12·9 (8·5–17·0) 0·98 (0·71–1·34)
<75 152/235 162/244 10·6 (9·0–13·8) 10·2 (7·5–13·0) 0·89 (0·71–1·11)
≥75 17/29 11/21 14·5 (5·1–NE) 18·9 (8·8–NE) 1·16 (0·54–2·47)
Sex
Male 112/182 122/185 11·4 (9·0–14·9) 8·8 (6·2–11·9) 0·83 (0·64–1·08)
Female 57/82 51/80 10·8 (7·0–15·0) 13·1 (10·2–15·7) 1·08 (0·74–1·59)
Pooled region
USA and western Europe 44/71 47/70 9·1 (6·5–17·3) 10·3 (5·8–15·0) 0·87 (0·58–1·32)
Eastern Europe 36/55 33/52 10·5 (5·5–12·9) 8·8 (5·9–13·0) 0·92 (0·57–1·49)
Asia 48/69 56/80 14·5 (9·4–19·8) 12·9 (6·2–15·7) 0·84 (0·57–1·25)
Rest of the world 41/69 37/63 10·6 (7·0–15·3) 11·1 (5·6–16·8) 0·99 (0·63–1·54)
ECOG performance status
0 49/96 53/91 15·3 (12·8–24·0) 11·6 (9·1–15·3) 0·73 (0·50–1·08)
1 120/168 120/174 8·7 (6·1–11.4) 9·0 (6·1–13·1) 0·99 (0·77–1·28)
Histology
Squamous 52/88 64/92 10·6 (8·8–18·5) 7·5 (5·9–10·6) 0·70 (0·48–1·01)
Non-squamous 117/176 109/173 11·5 (8·0–14·5) 12·9 (9·3–15·0) 1·02 (0·79–1·33)
Smoking status
Never smoker 30/43 22/41 13·9 (8·6–15·0) 17·7 (12·0–19·8) 1·69 (0·97–2·95)
Ever smoker 139/220 151/224 10·6 (8·7–13·6) 8·6 (6·2–11·6) 0·83 (0·66–1·04)
Ethnicity
Non-Hispanic or Latino 125/197 130/201 10·6 (8·8–14.5) 9·9 (7·0–12·4) 0·87 (0·68–1·11)
Hispanic or Latino 24/38 18/29 8·6 (5·1–15·3) 8·6 (3·1–NE) 0·96 (0·52–1·77)
All other patients 141/228 144/227 10·5 (8·6–13·8) 9·3 (7·0–11·9) 0·89 (0·71–1·13)
Japanese living in Japan 28/36 29/38 14·7 (10·1–23·9) 15·4 (9·9–19·7) 0·95 (0·56–1·61)
PD–L1 status (≥50% cutoff)
Positive 96/168 103/147 13·6 (10·1–18·5) 9·2 (5·8–12·4) 0·67 (0·51–0·89)
Negative 153/218 158/243 9·4 (7·5–10·8) 9·9 (8·4–12·5) 1·11 (0·89–1·39)
PD–L1 status (≥80% cutoff)
Positive 64/120 73/106 17·1 (10·6–25·0) 9·3 (5·8–13·1) 0·59 (0·42–0·83)
Negative 185/266 188/284 9·4 (7·9–10·7) 9·8 (8·1–11·9) 1·08 (0·88–1·32)
All participants 169/264 173/265 11·4 (9·4–13·9) 10·3 (8·5–13·0) 0·90 (0·73–1·12)

0·5 1·0 10·0

Favours avelumab Favours docetaxel

Figure 3: Subgroup analysis of overall survival within the PD-L1-positive population

ECOG=Eastern Cooperative Oncology Group. NE=not estimable.

In an expanded analysis of infusion-related reaction as 28 (7%) of 393 patients in the avelumab group and
an adverse event of special interest in the avelumab 51 (14%) of 365 in the docetaxel group discontinued
group, events of any grade occurred in 107 (27%) patients— treatment because of a treatment-related adverse event.
including adverse events of grade 3 or worse in Dose reductions occurred in 16 (4%) participants in the
six (2%) participants—and led to avelumab discon­ avelumab group and 72 (20%) in the docetaxel group.
tinuation in six (2%) participants. First onset was within Serious treatment-related adverse events occurred in
the first three infusions in 104 (26%) participants, and at 34 (9%) patients in the avelumab group and 75 (21%) in
infusion four or later in three (1%). Immune-related the docetaxel group. The most common serious
adverse events of any grade occurred in 65 (17%) of treatment-related adverse events in the avelumab group
393 avelumab-treated patients, and grade 3 or higher were pneumonitis (five [1%]) and interstitial lung disease
immune-related events occurred in 11 (3%) participants. (three [1%]), whereas in the docetaxel group they were
The most common immune-related adverse events of febrile neutropenia (21 [6%]), neutropenia (ten [3%]),
any grade in the avelumab group that occurred in at least pneumonia (nine [2%]), diarrhoea (five [1%]), decreased
2% of patients were rash and related terms (25 [6%]), neutrophil count (four [1%]), asthenia (three [1%]),
hypothyroidism (19 [5%]), and pneumonitis (nine [2%]; interstitial lung disease (three [1%]), leucopenia
appendix p 14). (three [1%]), and septic shock (three [1%]). Deaths

8 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

 Articles

PD-L1 positive (≥1% cutoff)* PD-L1 positive (≥50% cutoff) PD-L1 positive (≥80% cutoff)
Avelumab group Docetaxel group Avelumab group Docetaxel group Avelumab group Docetaxel group
(n=264) (n=265) (n=168) (n=147) (n=120) (n=106)
Complete response 4 (2%) 1 (<1%) 4 (2%) 0 4 (3%) 0
Partial response 46 (17%) 30 (11%) 38 (23%) 15 (10%) 33 (28%) 11 (10%)
Stable disease 86 (33%) 104 (39%) 56 (33%) 48 (33%) 33 (28%) 32 (30%)
Non-complete response or 4 (2%) 12 (5%) 4 (2%) 7 (5%) 4 (3%) 6 (6%)
non-progressive disease
Progressive disease 93 (35%) 57 (22%) 47 (28%) 39 (27%) 33 (28%) 29 (27%)
Non-assessable† 31 (12%) 61 (23%) 19 (11%) 38 (26%) 13 (11%) 28 (26%)
Proportion of patients with 19% (14–24) 12% (8–16) 25% (19–32) 10% (6–16) 31% (23–40) 10% (5–18)
objective responses (95% CI)
Odds ratio (95% CI) 1·76 (1·08–2·86) ·· 2·93 (1·55–5·55) ·· 3·85 (1·85–8·03) ··
p value‡ 0·011 ·· 0·0007 ·· 0·0002 ··
Disease control %§ 140 (53%) 147 (55%) 102 (61%) 70 (48%) 74 (62%) 49 (46%)

Objective responses were assessed by the independent efficacy review committee. *PD-L1-positive population based on randomisation. †Includes missing and not assessable
participants (reasons in appendix p 10). ‡One-sided p values for the primary analysis population (≥1% cutoff) and two-sided p values for the ≥50% and ≥80% subgroups
(Cochran-Mantel-Haenszel test). §Complete response, partial response, and stable disease (including non-complete response or non-progressive disease).

Table 2: Objective responses in the PD-L1-positive population and subgroups

because of treatment-related adverse events occurred in The high frequency of post-study use of checkpoint
four (1%) patients in the avelumab group (three on study inhibitors in our study is indicative of the increasing use
and one 4 months after the last avelumab dose): two due of checkpoint inhibitors in the treatment of NSCLC, and
to interstitial lung disease, one due to acute kidney injury, might have affected overall survival findings in our study.
and one due to autoimmune myocarditis, acute cardiac The frequency of use of checkpoint inhibitors after
failure, and respiratory failure. Deaths due to treatment- docetaxel treatment in this study (26% in the docetaxel
related adverse events occurred in 14 (4%) patients in group vs 6% in the avelumab group) is higher than that
the docetaxel group: three due to pneumonia, and in previous trials in a similar setting, including a trial7
one each due to febrile neutropenia, septic shock, of docetaxel versus atezolizumab (17% vs 4%) and a trial
febrile neutropenia with septic shock, acute respiratory of docetaxel versus pembrolizumab (13% vs 1%).6
failure, cardiovascular insufficiency, renal impairment, Furthermore, the proportion of randomly assigned
leucopenia with mucosal inflammation and pyrexia, patients who did not receive any study treatment was
infection, neutropenic infection, dehydration, and un­ higher in the docetaxel group than in the avelumab
known causes. The total number of deaths in treated group (8% vs 1%), which is one of the limitations of a
patients in the safety analysis set irrespective of relation­ non-blinded randomised study and is suggestive of
ship to study treatment was 255 (65%) of 393 patients in selective discontinuation after treatment assignment.
the avelumab group and 241 (66%) of 365 in the docetaxel Other potential factors that might have affected trial
group. outcomes include methods of biomarker assessment,
In the avelumab group, serum anti-drug antibodies to patient characteristics, or drug characteristics.
avelumab were detected in 51 (14%) of 368 assessable Treatment outcomes after chemotherapy in patients
patients. Antibodies were treatment emergent in with NSCLC tend to be better in Asian patients than in
46 patients (13%) and pre-existing in five (1%). No patients from other regions.19,20 Our trial included a high
associations between efficacy or safety outcomes and proportion of Asian patients in the PD-L1-positive
serum anti-drug antibodies were identified (data not population (71 [27%] in the docetaxel group vs 81 [31%] in
shown). the docetaxel group), including a high proportion of
Japanese patients living in Japan (14% in both groups).
Discussion By comparison, in both the KEYNOTE-010 and OAK
In this randomised, phase 3 trial in patients with trials, 21% of patients were Asian (217 of 1033 patients
advanced or metastatic NSCLC and disease progression in KEYNOTE-010, and 180 of 850 in OAK).6,7 Cross-study
after platinum doublet therapy, avelumab was not comparisons should be interpreted with caution because
associated with improved overall survival compared with of differences in study designs, patient populations, and
docetaxel in the PD-L1-positive population (≥1% cutoff). dates of enrolment. However, the median overall
Thus, the JAVELIN Lung 200 trial did not meet its survival in the avelumab group in patients with
primary endpoint. However, avelumab did show clinical PD-L1-positive tumours in our trial was similar to that
activity and acceptable safety in these patients. in the pembrolizumab group of a phase 2–3 trial

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 9

 Articles

Avelumab group (n=393) Docetaxel group (n=365)

Grade 1–2 Grade 3 Grade 4 Grade 5 Grade 1–2 Grade 3 Grade 4 Grade 5
Alopecia 0 0 0 0 97 (27%) 0 0 0
Decreased appetite 34 (9%) 0 0 0 59 (16%) 7 (2%) 0 0
Diarrhoea 24 (6%) 0 0 0 51 (14%) 4 (1%) 1 (<1%) 0
Anaemia 5 (1%) 0 0 0 51 (14%) 17 (5%) 1 (<1%) 0
Nausea 20 (5%) 0 0 0 49 (13%) 3 (1%) 0 0
Fatigue 29 (7%) 1 (<1%) 0 0 45 (12%) 10 (3%) 0 0
Stomatitis 3 (1%) 0 0 0 38 (10%) 3 (1%) 0 0
Asthenia 29 (7%) 1 (<1%) 0 0 37 (10%) 14 (4%) 0 0
Myalgia 6 (2%) 0 0 0 37 (10%) 3 (1%) 0 0
Mucosal inflammation 2 (1%) 0 0 0 19 (5%) 4 (1%) 0 1 (<1%)
Leucopenia 1 (<1%) 0 0 0 7 (2%) 8 (2%) 2 (1%) 1 (<1%)
Pneumonia 2 (1%) 0 0 0 6 (2%) 5 (1%) 0 3 (1%)
Infusion-related reaction 59 (15%) 5 (1%) 1 (<1%) 0 6 (2%) 1 (<1%) 0 0
Neutropenia 1 (<1%) 0 0 0 5 (1%) 21 (6%) 30 (8%) 0
Decreased white blood 1 (<1%) 0 0 0 4 (1%) 13 (4%) 4 (1%) 0
cell count
Decreased neutrophil 0 0 0 0 2 (1%) 8 (2%) 28 (8%) 0
count
Febrile neutropenia 0 0 0 0 0 34 (9%) 1 (<1%) 2 (1%)
Lipase increased 1 (<1%) 3 (1%) 1 (<1%) 0 0 1 (<1%) 0 0

Data are n (%). The table include grade 1–2 adverse events that occurred in ≥10% of patients, and grade 3–5 events that occurred in ≥1% of patients. All grade 3–5 adverse
events are provided in the appendix (pp 12–13).

Table 3: Treatment-related adverse events

(KEYNOTE-010) in patients with PD-L1-positive the 22C3 assay in the KEYNOTE-010 trial (28%).6 By
tumours (based on tumour cell PD-L1 expression).6 contrast, in the OAK trial of atezolizumab, in which
Across trials of other drugs, median overall survival PD-L1 expression by tumour or immune cells was
with docetaxel (pooled for patients with squamous and measured with the SP142 assay, the proportion of patients
non-squamous tumours) has ranged from 8·1 months with PD-L1-positive tumours above a minimum thres­
to 9·6 months.6,7,21 hold (≥1% expression in tumour or immune cells) was
Our prespecified exploratory analyses suggested that in only 54%, which emphasises the differences between
subgroups of patients with higher PD-L1 expression, PD-L1 assays.7
overall and progression-free survival were significantly In this trial, the frequency of treatment with immune
longer, and objective responses were more common, checkpoint inhibitors after docetaxel discontinuation was
with avelumab than with docetaxel—a finding consistent higher in patients with non-squamous NSCLC than in
with those from other trials.6,21,22 Studies of the analytic those with squamous NSCLC (34% vs 13%). In previous
performance of different PD-L1 assays have shown that randomised trials4–6,21 of similar drugs, greater treatment
the 73-10 assay used in this trial has higher sensitivity to effects relative to docetaxel have been reported in patients
detect PD-L1-positive tumour cells than other assays with squamous tumours compared with those with non-
used in trials of other drugs, including the 22C3 squamous tumours, including a progression-free survival
(pembrolizumab; Dako, Carpinteria, CA, USA) and benefit in the trial4 of nivolumab in squamous NSCLC,
SP142 (atezolizumab; Ventana, Tucson, AZ, USA) but not in the trial5 of nivolumab in non-squamous
assays.15,16 In a direct comparison16 of PD-L1 staining with NSCLC.21 Additionally, the improved overall survival
the 73-10 and 22C3 assays in NSCLC samples, noted in patients with tumours with higher PD-L1
73-10 detected a higher proportion of PD-L1-positive expression during nivolumab treatment compared with
tumours with a 1% or greater cutoff (80% vs 59%), and docetaxel was more pronounced in patients with non-
the 80% or greater cutoff for 73-10 was highly concordant squamous tumours than in those with squamous
with the 50% or greater cutoff for 22C3 (identifying tumours, but signifigance testing was not reported.21
24% vs 20% of samples, respectively; overall percentage These findings might be because of biological differences
agreement 94%). Additionally, according to prevalence between histological subtypes that could affect responses
data, the 80% or greater cutoff with the 73-10 assay to immunotherapy, such as immunological characteristics
identified a similarly sized proportion of patients in this of the tumour microenvironment or viral integration,23–25
trial (226 [29%] of 792) to the 50% or greater cutoff with although further studies are needed to clarify the effects

10 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

 Articles

of these factors during checkpoint inhibitor treatment of Novartis, ONO Pharmaceutical, and Roche. All other authors declare no
NSCLC. Notably, in the high PD-L1 expression subgroup competing interests.

(≥80% cutoff) in our study, avelumab showed clinical Data sharing
activity (in terms of effects on overall survival, progression- For all new products or new indications approved in both the
European Union and the USA after Jan 1, 2014, Merck shares
free survival, and objective response) irrespective of patient-level and study-level data after deidentification, and redacted
NSCLC histology, suggesting that checkpoint inhibition study protocols and clinical study reports from clinical trials in patients.
has an increased likelihood of clinical benefit in both These data will be shared with qualified scientific and medical
squamous and non-squamous tumours that are researchers, upon researcher’s request, as necessary for legitimate
research. Such requests should be submitted in writing to Merck’s data For more about Merck’s
immunologically active. Analyses are underway to sharing portal. When Merck has a co-research, co-development, approach to data sharing see
establish whether tumour mutational burden or other or co-marketing or co-promotion agreement, or when the product has https://www.merckgroup.com/
biomarkers could have contributed to the overall efficacy been out-licensed, the responsibility for disclosure might be dependent en/research/our-approach-to-
results in this trial. on the agreement between parties. Under these circumstances, Merck research-and-development/
will endeavour to gain agreement to share data in response to requests. healthcare/clinical-trials/
In safety analyses, avelumab was associated with a lower commitment-responsible-data-
frequency of treatment-related adverse events of any grade Acknowledgments
sharing.html
This study was funded by Merck and is part of an alliance between Merck
and of grade 3 or worse adverse events than docetaxel. and Pfizer. Medical writing support was provided by
However, consistent with previous trials of avelumab,10–13,26 Abhijith Thippeswamy (ClinicalThinking, Manchester, UK), which was
the frequency of infusion-related reactions was higher in also funded by Merck and Pfizer. We thank the patients and their families;
the avelumab group than in the docetaxel group. Infusion- the investigators, co-investigators, and study teams at each participating
centre and at Merck (Darmstadt, Germany), EMD Serono Research &
related reactions were mostly mild, rarely led to treatment Development Institute (Billerica, MA, USA), and Quintiles (Durham, NC,
discontinuation, and typically occurred within the first USA); Joe Zhuo (EMD Serono), Vivek Pawar (EMD Serono), Isabell Speit
three infusions (99% of cases). Other treatment-related (Merck), Ingrid Jörg (Merck), and Matthew Silver (EMD Serono) for
assistance with data analysis and interpretation; and Vikram Chand
adverse events occurred at a similar frequency in our trial
(formerly of EMD Serono) for medical guidance during the trial.
to those in previous studies of anti-PD-1 or anti-PD-L1
References
antibodies compared with docetaxel.4,5,7,27 Avelumab was 1 Kazandjian D, Suzman DL, Blumenthal G, et al. FDA approval
associated with a low frequency of grade 3 or worse summary: nivolumab for the treatment of metastatic non-small cell
immune-related treatment-related adverse events. lung cancer with progression on or after platinum-based
chemotherapy. Oncologist 2016; 21: 634–42.
Overall, although this trial did not meet its primary 2 Sul J, Blumenthal GM, Jiang X, He K, Keegan P, Pazdur R.
endpoint, the clinical activity and safety noted in this FDA approval summary: pembrolizumab for the treatment of
study support further studies of avelumab in patients patients with metastatic non-small cell lung cancer whose tumors
express programmed death-ligand 1. Oncologist 2016; 21: 643–50.
with NSCLC, and several trials are underway in different
3 Weinstock C, Khozin S, Suzman D, et al. US Food and Drug
treatment settings. Findings from JAVELIN Lung 200 Administration approval summary: atezolizumab for metastatic
will help to inform the analyses of these trials, which non-small cell lung cancer. Clin Cancer Res 2017; 23: 4534–39.
could help to define a future role for avelumab in the 4 Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel
in advanced squamous-cell non-small-cell lung cancer.
treatment of NSCLC. N Engl J Med 2015; 373: 123–35.
Contributors 5 Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel
FB, MB, and KP designed the study. All authors contributed to data in advanced nonsquamous non-small-cell lung cancer.
collection, analysis, and interpretation, and writing of the Article, and N Engl J Med 2015; 373: 1627–39.
approved the final version. 6 Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel
for previously treated, PD-L1-positive, advanced non-small-cell lung
Declaration of interests cancer (KEYNOTE-010): a randomised controlled trial. Lancet 2016;
FB has provided advisory or consultancy services and received personal 387: 1540–50.
fees from AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, 7 Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus
Clovis, Eli Lilly, Merck, Merck Sharp & Dohme, Novartis, Pierre Fabre, docetaxel in patients with previously treated non-small-cell lung
Pfizer, Roche, and Takeda. JV has received institutional research funding cancer (OAK): a phase 3, open-label, multicentre randomised
from Merck Sharp & Dohme, provided advisory or consultancy services controlled trial. Lancet 2017; 389: 255–65.
for Apotex, AstraZeneca, Boehringer Ingelheim, Merck Sharp & Dohme, 8 Pai-Scherf L, Blumenthal GM, Li H, et al. FDA approval summary:
Novartis, and Roche, and provided speaker services for AstraZeneca, pembrolizumab for treatment of metastatic non-small cell lung
Bristol-Myers Squibb, Merck Sharp & Dohme, and Roche. MG has cancer: first-line therapy and beyond. Oncologist 2017; 22: 1392–99.
received personal fees from AstraZeneca, Bristol-Myers Squibb, 9 Antonia SJ, Villegas A, Daniel D, et al. Durvalumab after
Merck Sharp & Dohme, and Roche. FdM has received personal fees chemoradiotherapy in stage III non-small-cell lung cancer.
from AstraZeneca, Bristol-Myers Squibb, Merck Sharp & Dohme, N Engl J Med 2017; 377: 1919–29.
and Roche. MÖ has provided advisory or consultancy services to Astellas 10 Gulley JL, Rajan A, Spigel DR, et al. Avelumab for patients with
and received personal fees from Janssen. AP has received personal fees previously treated metastatic or recurrent non-small-cell lung cancer
(JAVELIN Solid Tumor): dose-expansion cohort of a multicentre,
from Amgen, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly,
open-label, phase 1b trial. Lancet Oncol 2017; 18: 599–610.
Merck Sharp & Dohme, and Roche. OAF has provided advisory or
11 Apolo AB, Infante JR, Balmanoukian A, et al.
consultancy services for AstraZeneca, Bristol-Myers Squibb, Novartis,
Avelumab, an anti-programmed death-ligand 1 antibody, in patients
Regeneron, and Roche, and received honoraria from Bristol-Myers with refractory metastatic urothelial carcinoma: results from a
Squibb, Novartis, and Roche. BE-L is an employee of Merck. MB and MR multicenter, phase Ib study. J Clin Oncol 2017; 35: 2117–24.
are employees of EMD Serono Research & Development Institute, 12 Chung HC, Arkenau H, Wyrwicz L, et al. Avelumab (MSB0010718C;
a biopharmaceutical business of Merck. KP has provided advisory or anti-PD-L1) in patients with advanced gastric or gastroesophageal
consultancy services and received personal fees from Astellas, junction cancer from JAVELIN solid tumor phase Ib trial: analysis
AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Clovis, of safety and clinical activity. J Clin Oncol 2016; 34 (suppl): abstr 4009.
Eli Lilly, GlaxoSmithKline, Hanmi, KHK, Merck Sharp & Dohme,

www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9 11

 Articles

13 Disis ML, Patel M, Pant S, et al. Avelumab (MSB0010718C; 21 Horn L, Spigel DR, Vokes EE, et al. Nivolumab versus docetaxel in
anti-PD-L1) in patients with recurrent/refractory ovarian cancer previously treated patients with advanced non-small-cell lung cancer:
from the JAVELIN Solid Tumor phase 1b trial: safety and clinical two-year outcomes from two randomized, open-label, phase III trials
activity. J Clin Oncol 2016; 34 (suppl): abstr 5533. (CheckMate 017 and CheckMate 057). J Clin Oncol 2017; 35: 3924–33.
14 Gulley J. Exposure–response and PD-L1 expression analysis of 22 Fehrenbacher L, von Pawel J, Park K, et al. Updated efficacy analysis
second-line avelumab in patients with advanced NSCLC: data from including secondary population results for OAK: a randomized
the JAVELIN Solid Tumor trial. J Clin Oncol 2017; phase III study of atezolizumab vs docetaxel in patients with
35 (suppl): abstract 9086. previously treated advanced non-small cell lung cancer.
15 Tsao MS, Kerr KM, Kockx M, et al. PD-L1 immunohistochemistry J Thorac Oncol 2018; 13: 1156–70.
comparability study in real-life clinical samples: results of Blueprint 23 Parra ER, Behrens C, Rodriguez-Canales J, et al. Image analysis-based
phase 2 project. J Thorac Oncol 2018; 13: 1302–11. assessment of PD-L1 and tumor-associated immune cells density
16 Feng Z, Schlichting M, Helwig C, et al. Comparative study of supports distinct intratumoral microenvironment groups in
two PD-L1 expression assays in patients with non-small cell lung non-small cell lung carcinoma patients. Clin Cancer Res 2016;
cancer (NSCLC). J Clin Oncol 2017; 35 (suppl): abstr e20581. 22: 6278–89.
17 Eisenhauer EA, Therasse P, Bogaerts J, et al. New response 24 Mazzaschi G, Madeddu D, Falco A, et al. Low PD-1 expression in
evaluation criteria in solid tumours: revised RECIST guideline cytotoxic CD8+ tumor-infiltrating lymphocytes confers an
(version 1.1). Eur J Cancer 2009; 45: 228–47. immune-privileged tissue microenvironment in NSCLC with a
18 Robins JM, Finkelstein DM. Correcting for noncompliance and prognostic and predictive value. Clin Cancer Res 2018; 24: 407–19.
dependent censoring in an AIDS clinical trial with inverse probability 25 Robinson LA, Jaing CJ, Pierce Campbell C, et al. Molecular evidence
of censoring weighted (IPCW) log-rank tests. Biometrics 2000; of viral DNA in non-small cell lung cancer and non-neoplastic lung.
56: 779–88. Br J Cancer 2016; 115: 497–504.
19 Kawaguchi T, Matsumura A, Fukai S, et al. Japanese ethnicity 26 Karen K, Infante JR, Taylor MH, et al. Safety profile of avelumab in
compared with Caucasian ethnicity and never-smoking status are patients with advanced solid tumors: a pooled analysis of data from
independent favorable prognostic factors for overall survival in the phase 1 JAVELIN solid tumor and phase 2 JAVELIN Merkel 200
non-small cell lung cancer: a collaborative epidemiologic study of clinical trials. Cancer 2018; 124: 2010–17.
the National Hospital Organization Study Group for Lung Cancer 27 Fehrenbacher L, Spira A, Ballinger M, et al. Atezolizumab versus
(NHSGLC) in Japan and a Southern California Regional Cancer docetaxel for patients with previously treated non-small-cell lung
Registry databases. J Thorac Oncol 2010; 5: 1001–10. cancer (POPLAR): a multicentre, open-label, phase 2 randomised
20 Ou SH, Ziogas A, Zell JA. Asian ethnicity is a favorable prognostic controlled trial. Lancet 2016; 387: 1837–46.
factor for overall survival in non-small cell lung cancer (NSCLC) and
is independent of smoking status. J Thorac Oncol 2009; 4: 1083–93.

12 www.thelancet.com/oncology Published online September 24, 2018 http://dx.doi.org/10.1016/S1470-2045(18)30673-9

View publication stats