Professional Documents
Culture Documents
IO in Review
IO in Review
doi:10.1093/annonc/mdx755
Published online 7 December 2017
REVIEW
*Correspondence to: Dr Vanessa M. Hubbard-Lucey, Anna-Maria Kellen Clinical Accelerator, Cancer Research Institute, 29 Broadway, 4th Floor, New York, NY 10006, USA.
Tel: þ1-212-6887515; E-mail: vlucey@cancerresearch.org
Advances from immuno-oncology (IO) are changing the standard of care of many types of cancer, and the paradigm of cancer
treatments and drug development is being rewritten on a regular basis. Moreover, an unprecedented number of new
investigational agents and companies are entering the field of IO. As such, it has become challenging for oncology physicians
conducting clinical trials, industry veterans developing IO drugs, and even regulators reviewing novel IO agents to keep track of
the rapidly evolving landscape. To help the key stake holders in the field understand the latest IO landscape, we sought to
present an unbiased, neutral, scientifically curated, and timely updated analysis of all the current IO agents in clinical
development and the clinical trials testing these agents. We based our analyses on information collected from numerous
trusted and publicly available sources. We have developed two databases. One database tracks 2004 IO agents (940 in clinical
stage and 1064 in preclinical stage) against 303 targets, from 864 companies; the other tracks 3042 active clinical trials of these
agents with a target enrollment of 577 076 patients. This report provides key analyses of these data. Furthermore, we will discuss
a number of important and actionable trends in the current IO landscape: a large number of companies developing agents
against the same IO targets; a rapid increase in the number of anti-PD-1/L1 combination studies, many of which are testing the
same combinations and following inefficient patterns; and a significant increase in the number of small, investigator-initiated
studies. For each of the findings, we speculate the causes and discuss a few initiatives that aim to address some of these
challenges. Finally, by making these landscape analyses available, we aspire to inform the cancer community as they seek to
strive for efficiencies and innovation while avoiding duplication.
Key words: immuno-oncology, cancer immunotherapy, landscape analysis, clinical trials, tumor immunology
C The Author 2017. Published by Oxford University Press on behalf of the European Society for Medical Oncology.
V
All rights reserved. For Permissions, please email: journals.permissions@oup.com.
Annals of Oncology Review
Table 1. The list of 26 approved immuno-oncology agents, company origins, and targets
journals. The large numbers of promising experimental therapies, PD-1, CD40, and GITR); (ii) other immunomodulators that act
active clinical trials, and planned patient enrollment suggest that on other immune cells or the tumor immune microenvironment
we are entering a new era that could change the standard-of-care to unleash antitumor immunity (e.g. agents modulating IFNAR,
in even more cancer types. CSF1R, IDO1, A2AR, and KIR); (iii) cancer vaccines that induce
However, the fast-expanding and ever-changing IO field could antigen-specific antitumor immunity (e.g. sipuleucel-T); (iv) cell
overwhelm even the most experienced clinical investigators as therapies that engineer immune cells such as T cells to directly
they seek treatment options for their patients. To help the cancer attack cancer cells (e.g. anti-CD19 CAR-T); (v) oncolytic viruses
community stay informed of the current IO landscape, the Anna- that rely on both direct tumor killing and activation of antitumor
Maria Kellen Clinical Accelerator of the Cancer Research immunity (e.g. T-VEC); and (vi) CD3-targeted bispecific anti-
Institute (CRI), the clinical research program of CRI, provides bodies that bring T cell to the targeted tumor cells for direct kill-
herein an unbiased, neutral, comprehensive, and scientifically ing (e.g. blinatumomab) (Figure 1).
curated landscape analysis that aims to inform the physician-
scientists, drug developers, regulators, and other stakeholders of
the latest trends in IO clinical development. Concentration of agents on a few targets. The 940 clinical-stage
IO agents are owned by 462 different companies or academic in-
stitutes (supplementary Table S2, available at Annals of Oncology
online), and these agents modulate 271 different targets.
The landscape of clinical immuno-oncology Interestingly, a closer investigation of the 940 agents revealed that
almost half of them modulate only 40 targets (Figure 2). For ex-
A large number of IO drugs in clinical development ample, we found 164 agents targeting PD-1 or PD-L1 (PD-1/L1),
with significant duplication with 50 clinical-stage agents targeting PD-1/L1, 34 of which are
monoclonal antibodies (Figure 3). This confirms our assumption
Overview of the Clinical Accelerator IO database. Our database of significant duplication, despite the fact that five anti-PD-1/L1
included 2004 IO agents as of September 2017, 940 of which are monoclonal antibodies have already been approved [9].
in clinical development. On the basis of different mechanisms of Although it is still unclear if targeting the PD-1 receptor or its lig-
actions, we classified these agents into six categories: (i) T-cell- and PD-L1 will result in any differences in efficacy and safety,
targeted immunomodulators that act on inhibitory or activating none of the agents or studies in our database show any head-to-
molecules expressed by T cells (e.g. agents targeting CTLA-4, head evaluation in patients. In another example, agents targeting
Figure 1. The overview of 2004 immuno-oncology (IO) agents. Six classes of IO agents are identified on the basis of different mechanisms of
actions.
Target
Unspecified TAA
CD19
PD-1
HER2
NY-ESO-1
PD-L1
CD20
CD40
EGFR
CSF1R
GD2
WT1
IFNAR1
MUC1
PSMA
TERT
Survivin
HPV E6
Personalization
BCMA
CEA
LAG-3 Therapy type
OX40
TLR T-cell targeted immunomodulator
TLR7
CD123 Other immunomodulator
GITR
GPC3 Cancer vaccine
HPV E7
IDO1 Cell therapy
IL2
mesothelin Oncolytic virus
PSA CD3-targeted bispecific mAb
TLR9
CD22
CD30
CD47
MAGE-A3
ADORA2A
CD25
0 10 20 30 40 50 60 70 80 90 100 110 120
Number of clinical agents
Figure 2. The top 40 targets with the most clinical-stage immuno-oncology agents. Unspecified tumor-associated antigen is defined as the
target of unspecified antigens from either individual patients and cell lines. In this sense, therapies against this target mainly consist of cell
line-based vaccines, autologous tumor vaccines, autologous DC vaccines, and tumor-infiltrating T-cell therapies. Personalization is defined as
the targets discovered by algorithm-based approaches from, e.g. the individual patients’ tumor samples or blood circulating DNA.
unspecified tumor-associated antigens (unspecified TAA) rep- Rapid proliferation of CAR-T-cell therapy, especially in China.
resent the largest number of clinical IO agents. There are now CAR-T-cell therapy has delivered unprecedented results, with
114 agents in the clinic, 98 of which are cancer vaccines. Of the two recent approvals of CD19-targeted CAR-T-cell therapies
114 unspecific TAA-targeted IO agents, only sipuleucel-T (an au- [11, 12]. CD19 is also the most common CAR-T target, with
tologous dendritic cell vaccine) has been approved for prostate 56 clinical-stage therapies against it (supplementary Table S3,
cancer [10]. This number of agents raises questions on efficien- available at Annals of Oncology online). Surprisingly, by interrog-
cies of resources and, importantly, the patients who enroll into ating our IO drug database, we found 291 different CAR-T thera-
clinical trials. The concentration of IO development and patient pies, with 162 of them being tested clinically. These agents are
resources on a few targets, some with already approved drugs, being developed by 136 different companies and academic cen-
could potentially be stalling future innovation. Rather, investing ters. Interestingly, companies and academic centers from China
these precious resources more efficiently could help accelerate seem to lead the total number of clinical-stage CAR-T therapies,
needed progress in finding cures to this deadly disease. with 98 clinical-stage agents owned by 46 companies from China
80
Bispecific antibody pies, chemotherapies, radiotherapies, or chemoradiotherapies. It is
Small molecule worth noting that 49 of the combination trials are testing agents
60 Other biologic that have not been approved, whereas the majority of combination
trials focus on the five approved agents (Figure 5).
40
30
20 Fragmentation and lack of coordination in PD-1/L1 combination
9 5 trial landscape. A deeper analysis of the 1105 trials reveals some
3 3
0 interesting observations. First, there has been a rapid increase in
Preclinical & Phase I Phase I/II Phase II Phase III Approved
Discovery new combination studies in the past 5 years. Analysis of our data-
base shows that in 2017 alone, 469 new studies were started, with a
Figure 3. The overview of 164 anti-PD-1/L1 agents. Of note, 164 target enrollment of 52 539 patients (Figure 6). Second, agents
agents are currently in development, 50 of whom are in clinical against 165 different targets are being combined with anti-PD-1/L1
stages. agents (supplementary Table S5, available at Annals of Oncology
98
US
Clinical stage
51 Approved
Phase III
Other
Phase II
13 Phase I/II
Phase I
80 60 40 20 0 20 40 60 80 100 Preclinical & Discovery
Number of agents
Figure 4. The landscape of CAR-T-cell therapy. China leads the total number of clinical-stage CAR-T-cell therapies, whereas United States has
the largest number of preclinical agents.
250
200
161
150
124
100
49
50 32
0
Pembrolizumab Nivolumab Durvalumab Atezolizumab Avelumab Others
Figure 5. Analysis of PD-1/L1 combination trial types. The majority of combination trials focus on the five approved agents. Combination
with other immuno-oncology agents, targeted therapies, and chemotherapies are the top common strategies.
Number
1 5 2 13 20 58 190 329 469
of new trials
Planned new
136 2473 582 4867 5031 11 276 39 821 46 153 52 539
enrollment
Planned
enrollment 136 495 291 374 252 194 210 140 112
per new trial
Figure 6. The rapid increase of new anti-PD-1/L1combination trials in the past 5 years. The size of the bubble correlates to the target enroll-
ment of patients. Multiple bubbles of the same color represent multiple cancer types that are being tested in these trials.
2. Chemotherapies: 170
3. Radiotherapies: 64
4. Anti-VEGFA agents: 43
5. Chemoradiotherapy combos: 42
Figure 7. The landscape analysis of targets of anti-PD-1/L1 combination trials. The size of the bubble correlates to the number of trials.
online). Many of these studies we observed have significant over- New trial designs should be considered for IO
lap and duplication of the combination partner targets. For in- development
stance, anti-CTLA-4 agents and chemotherapies are the two
most prevalent classes of therapies being combined with anti- Breakthrough treatments sooner. The early activity observed with
PD-1/L1 agents, with 251 and 170 studies, respectively (Figure anti-PD-1/L1 agents across multiple tumor types has accelerated
7). This could possibly be driven by the evidence of preclinical the use of novel trial designs, several of which have led to regula-
synergy of the combination [13], or the recent approvals of anti- tory approvals. There are several examples of novel study designs
PD-1/anti-CTLA-4 combination for advanced melanoma and where drugs have leaped through the drug development phases:
anti-PD-1/chemotherapy combination as the first-line treat- from first-in-human, dose selection, safety, efficacy, biomarker
ment of advanced non-small-cell lung cancer (NSCLC) [14, 15]. selection, all the way to regulatory approval, in one single study.
However, it should be questioned if 251 and 170 studies are Particularly, the proliferation of agents has resulted in a competi-
needed to answer clinically relevant questions. Other prevalent tive drug development landscape with a need to answer clinical
combination partners are anti-VEGF agents (predominantly questions faster within a shorter timeframe by using these novel
with bevacizumab) and radiochemotherapy, with some having trial designs. One example is KEYNOTE-001, a phase I trial that
already shown clinical benefits [16]. This wide coverage of dis- led to two FDA approvals, one in melanoma and the other in
tinct targets, in our opinion, reflects a fragmented and uncoor- advanced NSCLC [17]. Moreover, the PD-L1 biomarker was
dinated approach to drug development across the anti-PD-1/L1 evaluated and confirmed all within the same study. KEYNOTE-
combination trial landscape. 001 is arguably the largest phase I trial, which finally recruited
655
1 15 522 Summary and discussion
Sponsorship
In summary, the current clinical landscape of IO presents enor-
mous enthusiasm from academia and industry, which is exempli-
Non-industry-sponsored trials
fied by 940 clinical-stage IO agents, 3042 immunotherapy trials,
Industry-sponsored trials
and 1 105 anti-PD-1/L1 combination trials. The field is very
Figure 8. Comparison of total trial numbers and target enrollment be- promising and has the potential to deliver many breakthroughs
tween industry-sponsored and non-industry-sponsored anti-PD-1/L1 to change the standard of care of many cancer types. However, it
combination trials. is also very crowded, fragmented, and uncoordinated with
300 500
400
200
300
200
100
100
0 0
2009 2010 2011 2012 2013 2014 2015 2016 2017
Figure 9. Comparison of average target enrollment between trials with different sponsorship.