Annals of Oncology 28: 34–43, 2017

doi:10.1093/annonc/mdw413
Published online 11 October 2016

SPECIAL ARTICLE

Statistical controversies in clinical research: basket

trials, umbrella trials, and other master protocols: a
review and examples

L. A. Renfro* and D. J. Sargent

Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, USA

*Correspondence to: Dr Lindsay A. Renfro, Division of Biomedical Statistics and Informatics, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA. Tel: þ1-507-284-
3202; E-mail: [email protected]

In recent years, cancers once viewed as relatively homogeneous in terms of organ location and treatment strategy are now better
understood to be increasingly heterogeneous across biomarker and genetically defined patient subgroups. This has produced a
shift toward development of biomarker-targeted agents during a time when funding for cancer research has been limited; as a
result, the need for improved operational efficiency in studying many agent-and-target combinations in parallel has emerged.
Platform trials, basket trials, and umbrella trials are new approaches to clinical research driven by this need for enhanced efficiency
in the modern era of increasingly specific cancer subpopulations and decreased resources to study treatments for individual
cancer subtypes in a traditional way. In this review, we provide an overview of these new types of clinical trial designs, including
discussions of motivation for their use, recommended terminology, examples, and challenges encountered in their application.
Key words: basket trialbiomarker-based trialmaster protocolplatform trialprecision medicineumbrella trial

Introduction discoveries hinged upon identification of a fairly prevalent bio-

marker which happened to be an actionable mutation responsible
A new cancer treatment paradigm for driving the clinical behavior of a fairly common tumor; where
low prevalence mutations exist, particularly in rare diseases, less
In recent years, substantial progress in the areas of genomics tech- progress has been made to date. Additionally, immunotherapies
nology, tumor biology, computational analysis, and drug discovery including nivolumab [9, 10] and pembrolizumab [11] have been
have motivated exciting advances in clinical and translational can- approved for treatment of advanced squamous cell and non-
cer research. In particular, rapid development, decreased cost, and squamous cell non-small-cell lung cancer (NSCLC) and
increased availability of next-generation genomic sequencing and advanced melanoma, respectively.
other methods for molecular classification of tumors has produced Development of novel therapeutics is challenged by the hetero-
a new paradigm for understanding and treating cancer. In some geneity that exists not only among patients within any given
cases, this new molecular viewpoint of ‘precision oncology’ has led tumor type (inter-patient heterogeneity), but also the heterogen-
to therapeutic discoveries where the targeted treatment paradigm eity within an individual (intra-patient heterogeneity) as demon-
actually worked; for most other hypotheses, however, the signifi- strated by molecular evolution of a tumor through time (through
cance of molecular classification remains unclear. sequences of therapy) and space (primary tumor to metastasis)
The advantages of targeting oncogenic pathways believed to be [12]. The respective concepts of ‘oncogenic driver’ and ‘oncogene
responsible for the growth and metastasis of tumors is now well addiction’ have thus shifted the course of drug development in
established in a variety of cancer settings, including BRAF- oncology [13–15]. In parallel, increased understanding of the
mutant melanoma [1], HER2-positive breast cancer [2, 3], KRAS complex structural paradigm of genetic alterations activating
wild-type colorectal cancer [4, 5], and EGFR or ALK-mutated intracellular proteins along multiple pathways, as well as several
lung cancer [6–8], among others. The success of most of these years of early clinical experience with success of (and subsequent

C The Author 2016. Published by Oxford University Press on behalf of the European Society for Medical Oncology.
V
All rights reserved. For permissions, please email: [email protected]
Annals of Oncology Special article
resistance to) single-target therapies, has led to the widely ac- disease to enroll, whether to restrict enrollment to those patients
cepted hypothesis that multiple targeted therapeutics will likely hypothesized to experience the greatest benefit from targeted
be required to overcome tumor resistance and yield sustained therapy, or whether to do some combination of the two through
clinical benefit for patients [16–20]. mid-trial adaptive measures [32, 33]. In the case of a standard
Clearly, we have entered an era where a patient’s tumor and its ‘all-comers’ design, the targeted agent may only benefit a selected
treatment will no longer be viewed primarily in terms of fixed group of patients and thus a strong subgroup effect may instead
organ location and pathology, but also (or instead) will be viewed appear as a weak overall effect, though randomization of marker-
in terms of potentially dynamic genomic, proteomic, transcrip- positive and marker-negative patients to targeted versus non-
tomic, and immunologic abnormalities and features that may be targeted treatment can play an important role of validating the
specifically targeted with novel agents. treatment-by-marker interaction hypothesis. On the other hand,
so-called ‘enriched’ clinical trials enrolling only those patients
hypothesized to benefit (such as those whose tumor DNA harbors
Challenges of traditional clinical trial designs in a particular mutation) may demonstrate a large effect in theory,
the era of new treatment strategies but such a trial may be challenging to conduct in practice, par-
Despite remarkable advances in the development of molecularly ticularly within low-prevalence populations where a high screen
targeted agents, progress on the clinical front has outpaced mod- failure rate dampens enthusiasm for the trial. Choice of an un-
ernization of the clinical trial designs used to study novel therapies selected versus enriched design for a single marker-therapy com-
within possibly heterogeneous—and potentially rare—patient sub- bination should be made based on the existing level of evidence
groups [21]. Cancer statistician Dr Don Berry has called clinical tri- for the putative predictive biomarker [34].
als the ‘weakest links in the chain of knowledge for determining As many have noted, performing clinical development of tar-
therapeutic advances’, further stating that ‘it is ironic that we take geted therapies according to a traditionally isolated one-agent-at-
the same clinical trial approach to evaluate all manner of potentially a-time fashion lacks efficiency and may be prone to high overhead
amazing transformative experimental therapies and yet we don’t and low feasibility, particularly in low-prevalence subgroups of a
experiment with the design of the clinical trial itself’ [22]. given disease [16, 17, 22, 35]. While nearly 1.7 million adults are
Limitations inherent to both the existing phase I–II–III clinical de- diagnosed with cancer in the United States each year [36], only
velopment paradigm and the types of designs typically employed 3%–5% of these patients enroll in clinical trials, further contribu-
within single-phase cancer trials are evident when clinical cancer ting to underpowered studies and early trial discontinuation [37].
‘success’ is viewed on a broader scale. Currently, the time from ini- This changing cancer paradigm and ongoing segmentation of
tial drug discovery to clinical testing and regulatory review can take broadly recognized cancers into comparatively rare subcancers
up to 15 years [23]. While preclinical evaluations are currently have produced an urgent need to streamline the development
underway for thousands of compounds intended for the treatment and approval of new compounds [16, 19].
of cancer [24] and 771 new cancer medicines and vaccines are cur-
rently in development or awaiting regulatory review [25], it remains A movement toward increased efficiency in clinical
the case that only 13% of cancer drugs initiating phase I studies research
ultimately achieve final market approval [26]. Of those compounds
that make it to confirmatory phase III trials, only 34% may be ex- In response to the aforementioned challenges and pressures of con-
pected to achieve a ‘statistically significant’ result [27, 28]. ducting single biomarker-based trials in this new era of clinical
A 2010 report by the Institute of Medicine echoed these con- cancer research, a trend has emerged toward investigating multiple
cerns, highlighting the challenges faced by cooperative oncology target–treatment pairs in parallel, either within or across recog-
groups and calling for restructuring of the entire clinical trials nized tumor types. These so-called ‘master protocols’ may utilize a
system to increase efficiency and avoid redundancy at a time centralized screening platform and common protocol format for
where novel compounds waiting for evaluation are many and re- each of several biomarker-driven substudies, with benefits includ-
sources are limited [29]. In the same year, the National Cancer ing enhanced patient participation due to increased likelihood of
Institute’s Investigational Drug Steering Committee convened eligibility for at least one of the accruing cohorts. In the sections
The Clinical Trial Design Task Force, the members of which that follow, we review different types of master protocols including
endorsed several types of clinical trial design modifications to ‘basket’ and ‘umbrella’ trials that operationalize these advantages.
meet these objectives, including sequential learning for early ter- We further clarify terminology related to these types of trials, de-
mination (for efficacy or futility), possible trial extensions to es- scribe features they have in common, provide examples of recently
tablish or identify predictive subgroups, multi-stage (e.g. phase completed or ongoing master protocols, and discuss challenges
II/III) designs to enable seamless transitions to confirmatory and important considerations in their implementation.
studies, and ‘platform’ designs that allow for mid-trial adding or
dropping of new experimental treatment arms [30, 31].
Concerns regarding simple application of ‘off-the-shelf’ de- Overview of master protocols
signs within a traditional sequence of phase I, II, and III trials
(often utilizing different end points) are especially well founded Master protocols: motivation and common
in the context of targeted therapies studied within ever-shrinking features
‘targeted’ patient populations [22]. Specifically, when molecu-
larly targeted treatments are studied in early phase trials, it must The term ‘master protocol’ refers generally to a framework in
be decided whether to allow all eligible patients with a given which multiple parallel drug studies are operated under one

Volume 28 | Issue 1 | 2017 doi:10.1093/annonc/mdw413 | 35

Special article Annals of Oncology

Tissue submission

Biomarker profiling

Negative on all
Marker 1 positive Marker 2 positive Marker 3 positive … Marker N positive
markers

Sub-protocol 1 Sub-protocol 2 Sub-Protocol 3 Sub-Protocol N Non-match

(experimental
… (experimental (experimental … (experimental protocol (drugs
drug 1) drug 2) drug 3) drug N) vary)

Non-match
…Design 1 Design 2 Design 3 … Design N
design

Figure 1. General schema of a master protocol.

overarching (‘master’) protocol, wherein the parallel studies are treatments or cohorts are made later in time [16, 38–40]. In this
differentiated by the marker–treatment combinations under inves- case, the protocol serves as an operational structure allowing for
tigation (Figure 1) [16]. Master protocols require endorsement by new therapies to enter and subsequently exit a standing trial (for
a broad consortium of academic and industry partners, pharma- efficacy or futility) while the trial is underway, without the need
ceutical companies, and government agencies. The main goals of for overall protocol modifications and associated administrative
constructing a master protocol in place of several truly independ- delays. The use of a master protocol with a standardized screening
ent trials in biomarker-defined cohorts include increased genomic platform may also facilitate US Food and Drug Administration
screening efficiency, accelerated and streamlined clinical develop- (FDA) approval of new therapies, making them available to
ment timeline, and enhanced enthusiasm for patient accrual due to marker-defined subgroups of patients more quickly than may
inclusion of a broad range of molecular subtypes. have been possible otherwise.

increased efficiency. Master protocols increase genomic screening stakeholder enthusiasm. Inclusion of a potentially large number
efficiency in several ways, including: use of a common platform of marker–treatment combinations within the same study has the
or set of assays capable of detecting abnormalities in multiple po- potential to bolster the enthusiasm of study stakeholders, includ-
tential targets; immediate assignment of patients to an appropri- ing treating physicians, participating institutions, sponsors, and
ate substudy on the basis of screening results; strict guidelines patients themselves. In particular, where an additional cohort is
ensuring that sufficient tumor is evaluated for simultaneous as- included to capture patients negative on all markers (and where
sessment of multiple markers or abnormalities, and a structural such patients are assigned to an experimental therapy or standard
framework for allowing substudies of new agents and biomarkers of care), an initially eligible patient may enter the trial with the
to be added or dropped in a preplanned, expeditious fashion. knowledge that he or she will almost certainly be able to receive a
Although not always the case, substudies of master protocols are potentially promising experimental treatment regardless of the
often designed to detect only large efficacy signals (in single-arm outcome of the screening process [16, 35].
cohorts) or large treatment effects (in randomized cohorts), with
efficiency resulting from lower sample size requirements [16, 19, flexible objectives. Substudies based on biomarkers or genomic
22]. Additionally, master protocols often utilize a master budget groups of interest may share common design features (e.g. power,
with shared costs for a common infrastructure across substudies, sample size, type I error), or different study designs may be used
resulting in financial efficiency. across the master protocol, reflecting differences in study object-
ives among the protocol cohorts. A master protocol may contain
accelerated clinical development. Master protocols often include substudies with discovery-based or confirmatory objectives, or in
enhanced regulatory input to support accelerated clinical devel- some cases, sequential objectives are addressed by multi-phase
opment, as agencies usually review the high-level study design of designs within patient cohorts [16, 21, 38, 39]. An exploratory
a master protocol or platform trial up front, with lesser require- master protocol typically comprises single-arm studies within
ments for subsequent regulatory review when modifications to targeted cohorts where patients received a matched experimental

36 | Renfro and Sargent Volume 28 | Issue 1 | 2017

Annals of Oncology Special article
(targeted) drug; a confirmatory master protocol randomizes pa- subtrials where they receive (or perhaps are randomized to) a
tients to targeted versus non-targeted or standard of care therapy matched targeted treatment [16, 18–21, 39, 42]. In such trials, the
within targeted cohorts in order to confirm the predictive nature relevant markers are regarded as refinements of (rather than re-
of the biomarker for regulatory purposes [21]. placements of) tumor type. Common features of umbrella trials
include: an enrollment and screening process that occurs before
an actionable biomarker has been identified, and central profiling
Baskets, umbrellas, and platforms: definitions and
using a standardized platform or set of assays. While basket stud-
terminology
ies are generally limited to single-arm substudies with discovery
Rapid introduction of master protocols and the subsequent objectives, umbrella trials may include single-arm or randomized
frenzy of scientific interest have motivated widespread use (and subtrials, where the latter includes confirmatory objectives. An
misuse) of related terms in the literature such as ‘basket trial’, example schema for an umbrella trial is shown in Figure 3.
‘umbrella trial’, and ‘platform trial’, among others. A number of Hyman et al. [39] refer to umbrella trials as ‘molecular allocation
authors have noted confusion regarding the definitions of these studies’.
terms [16, 19, 22, 39, 41, 42]; we attempt to harmonize some def- Another type of master protocol described in the literature is
initions here. the platform trial (or ‘standing trial’), a generic term for a
A basket trial is a master protocol for which patient eligibility is randomized design with a common control arm and many differ-
defined by the presence of a particular biomarker or molecular al- ent experimental arms that enter and exit the trial as futility or ef-
teration rather than a particular cancer type. Basket trials are pre- ficacy are demonstrated, often according to Bayesian decision
dicted on the hypothesis that the molecular characterization of a rules [22, 38, 40]. The trial itself then comprises a platform or
particular tumor predicts response to a matched (targeted) treat- standing infrastructure to which novel therapies may be added or
ment to a greater extent or independent of tumor histology [16, from which they may be dropped. While biomarker cohorts in a
18–20, 21, 22, 39, 41, 42]. Basket trials are generally tumor- platform trial may not be explicitly separate, the treatment effects
agnostic to some degree; for example, enrollment to a basket trial of various experimental treatments are usually modeled as inde-
may be restricted to patients with solid tumors, while the molecu- pendent parameters across molecularly defined subtypes, often
larly defined subtrials (or ‘baskets’) may comprise patients with according to a Bayesian hierarchical model. Adaptive randomiza-
many different types of solid tumors. For this reason, randomiza- tion, i.e. mid-trial shifts in the randomization ratios for patients
tion to a common control arm within molecular cohorts is un- with a given biomarker signature to favor the treatment showing
common, due to differences in standard of care across tumor the most promise in that signature, may also be present. Berry
types. A general schema for a basket trial is shown in Figure 2.
Unlike basket trials, where drug–mutation pairs are tested on a
variety of tumor types, an umbrella trial generally restricts enroll-
Matched
ment to a single type or class of cancers. In an umbrella trial, pa-
therapy A
tients with tumors from the specified cancer type are centrally
screened and assigned to one of several molecularly defined Target A

Control

Tumor type
W

Molecular Matched
target A therapy B
Tumor type Specific Molecular
Target B
X tumor type portrait
Control

… …
Drug A/B
Matched
therapy N
Tumor type
Target N
Y

Molecular Control
target B
Tumor type
Z Non-match
Non-match
therapies
for all
(varies)
Figure 2. General schema of a basket trial (one ‘basket’ of potentially many baskets
shown). Figure 3. General schema of an umbrella trial.

Volume 28 | Issue 1 | 2017 doi:10.1093/annonc/mdw413 | 37

Special article Annals of Oncology
[22], Saville and Berry [38], and Hobbs et al. [40] have described 25 single-arm substudies based on disease classification by next-
platform trials in the context of randomization to a single control generation sequencing [20, 22, 49, 50]. In MATCH, it is hypothe-
arm versus many experimental arms (which may enter and exit sized that the overall response rate will be improved from 5% to
the trial seamlessly), in order to reduce overall sample size and 25% in one or more cohorts through the use of matched targeted
improve efficiency when compared with an equivalent set of two- therapies, and this hypothesis will be tested in independent ana-
arm trials using a common control arm. lyses of 30 patients with heterogeneous tumor types per substudy.
In the sections that follow, we describe basket trials, umbrella Each arm will enroll at least 25% of patients from rare tumor
trials, and other master protocol designs in more detail, including types and repeat biopsies over time may also occur, such that pa-
advantages, limitations, and examples. tients may be reclassified to other arms of the study if they remain
eligible. In the event that NCI-MATCH investigators detect a
promising efficacy signal within one of the ‘baskets’ that is mostly
Basket trials: marker specific, histology- or completely attributable to a particular tumor type, a location
or histology-dependent expansion or separate phase II or III
independent cohorts study could be launched to confirm the finding [50]. Patients are
not eligible for participation in NCI MATCH if they have a tumor
Advantages of basket trials type and mutation for which a targeted agent has been FDA
approved. Additional details of biopsy processing, molecular
Basket trials have several advantages. First, they can provide ac-
classification, and required levels of evidence for NCI MATCH
cess to molecularly targeted agents for patients across a broad
are described by Moore and Mannel [20].
range of tumor types, potentially including those not otherwise
While some discussions of NCI MATCH have been mostly
studied in clinical trials of targeted therapies [20]. Secondly, in
positive [20, 50], others have expressed strong concerns [22, 50].
many cases, molecular testing is carried out locally and confirm-
Importantly, all patients are regarded to be exchangeable across
ation by a central assay is not required before patient enrollment
tumor types in terms of expected response rate, even when pa-
[20, 39], though tumor and plasma are often banked for subse-
tients with different tumor types (e.g. breast versus colon) are
quent companion diagnostic testing and validation. This feature
understood to have very different prognoses. Currently, NCI
reduces the time between initial diagnosis and/or eligibility con-
MATCH has temporarily suspended accrual after experiencing a
firmation and later cohort assignment and initiation of treat-
lower-than-anticipated match rate among eligible patients [51,
ment. Thirdly, cohorts within basket trials are often small and
52]. At this time, feasibility is being assessed, along with the pos-
utilize single-stage or two-stage designs, which yield quick results,
sible addition of molecular cohorts to the trial to enhance the
given sufficient accrual.
overall match rate. Despite these challenges, NCI MATCH is one
of the first basket trials to be activated and will serve as a founda-
Limitations of basket trials tion for subsequent learning, both in terms of cancer biology and
One major limitation of basket trials is the assumption that mo- targeted trial design.
lecular profiling may be sufficient to replace histological tumor
typing, as, in some cases, histological tumor type has been found
to predict response to treatment more strongly than the bio-
markers or mutations comprising the studied cohorts (see, e.g. SIGNATURE. Another ongoing basket trial is SIGNATURE, a
[39]). Even outside the context of a basket trial, it was recognized basket trial sponsored by Novartis that is currently enrolling pa-
that V600E BRAF-mutant melanoma or hairy cell leukemia are tients harboring any solid tumor or lymphoma refractory to
responsive to BRAF inhibition, while colon tumors with the same standard therapies (excluding those where molecular agents have
BRAF mutation are not [18, 20, 43–46]. This issue may be antici- already proved effective). SIGNATURE comprises several inde-
pated, as it is well accepted that the environment and location in pendent biomarker-driven single-arm trials with matched tar-
which a tumor develops may impact its mutational profile as well geted treatments, and to date, three arms are open to enrollment
as differentially predict treatment response across similar profiles. [20, 53].
To this end, many have noted that current clinical evidence is in-
sufficient to conclude that molecular descriptors should replace
histological tumor typing [18, 22, 41, 47], and it has been sug-
gested that future studies integrate anatomic with mutational
and functional molecular profiling through the use of proteomic AcSé. AcSé is a large-scale multi-center phase II trial from the
technologies [47] and explore multi-gene signatures with com- French program UNICANCER assessing the efficacy and safety of
bination therapies [48]. crizotinib as monotherapy in 23 cohorts of patients harboring at
least one mutation among ALK, MET, RON, or ROS1 across a
Basket trial examples variety of solid tumors (gastrointestinal, breast, kidney, ovarian,
thyroid, and sarcomas, among others). While AcSé primarily re-
NCI MATCH. Activated in August 2015, NCI MATCH sembles a basket trial in that it enrolls patients from many tumor
(Molecular Analysis for Therapy CHoice) is an ongoing basket types, within AcSé, cohorts are defined both by alteration and
trial that was initially intended to screen 3000 patients with histopathology (e.g. gastric cancer with MET amplification), with
advanced solid tumors or lymphoma who have progressed on at each cohort following a single-arm, two-stage design. This study
least one therapy with the goal of assigning 800–1000 patients to is currently accruing patients [19, 54].

38 | Renfro and Sargent Volume 28 | Issue 1 | 2017

Annals of Oncology Special article
Umbrella trials: one tumor type, many LUNG-MAP. LUNG-MAP [16, 22, 59, 60] is a phase II/III study
of targeted therapies in patients with previously treated advanced
marker cohorts
squamous NSCLC led by the Southwest Oncology Group
(SWOG) in cooperation with NCI. In LUNG-MAP, four
Advantages of umbrella trials
randomized phase II trials of targeted therapy versus standard of
One immediate advantage of umbrella trials (relative to basket care are conducted in parallel within mutation-enriched cohorts,
trials) is the ability to draw meaningful conclusions that are spe- with graduation of a cohort to a phase III registration study if
cific to a tumor type and therefore less prone to chance tumor progression-free survival crosses an efficacy boundary during
heterogeneity present within a given trial cohort. Furthermore, phase II (phase II patients are included in the phase III analysis,
when randomization to targeted versus non-targeted treatments contributing critical additional length of follow-up). Once eligi-
within cohorts takes place (and particularly when a marker- bility is determined, patients in LUNG-MAP are screened for
negative cohort is included), the drug’s purported mechanism of mutations and amplifications of interest by the research version
action can be more thoroughly evaluated, and prognostic versus of the Foundation One panel of Foundation Medicine, Inc.,
predictive marker effects can be empirically distinguished. Such which improves screening efficiency in a disease where available
trial designs may lend stronger evidence to support the activity of tissue may be limited. LUNG-MAP was additionally designed as
a new drug with a readily describable population and indication. a platform trial allowing for seamless integration of new cohorts
via nomination by its Drug Selection committee. No cross-
cohort statistical analyses are planned, and patients without a
Limitations of umbrella trials mutation of interest were initially randomized to receive an anti-
A direct consequence of the greatest strength of umbrella trials PD-L1 therapy versus standard of care within a ‘non-match’
(conclusions applicable to a single tumor type) is one great disad- study. Patients with tumors harboring multiple markers are
vantage: feasibility. Particularly within rare diseases, further sub- randomized in a weighted fashion to qualifying cohorts with pri-
classification of an already-rare tumor type by molecular ority given to those cohorts with lower prevalence markers.
alterations may lead to poor accrual within cohorts, and slow Despite its sound design, regulatory efficiency, and ability to
progress of the trial overall. An already lengthy trial may be fur- assign each eligible patient to a promising experimental therapy,
ther exacerbated by inclusion of randomization (generally requir- LUNG-MAP has faced challenges since its initiation. In March
ing larger cohort-specific sample sizes) and changes in the 2015, shortly after the trial began, the FDA approved the im-
treatment landscape of the tumor type under study, e.g. introduc- munotherapy drug nivolumab for the same population based on
tion of new standard of care regimen to the market while the um- a trial where it had shown superiority over docetaxel, the stand-
brella trial is underway [38, 40]. ard of care drug used in the control arms of the LUNG-MAP co-
horts. Also, one cohort of the study (for c-MET-positive patients)
closed early for toxicity concerns when it became known that one
Umbrella trial examples of the study drugs had caused harm in patients with gastric can-
cer. Following temporary suspension to respond to these events
FOCUS4. One ongoing umbrella trial is FOCUS4, which enrolls [16], LUNG-MAP is currently recruiting patients to some of the
previously untreated metastatic colorectal cancer patients and as- study cohorts with treatment modifications [59].
signs them to one of four biomarker-enriched cohorts following
16 weeks of standard front-line chemotherapy. Within each co-
hort, patients are randomized to an experimental targeted agent
versus placebo [55, 56]. FOCUS4 further contains a cohort for Beyond baskets and umbrellas: platform
all-wild-type patients where patients may be randomized to treat- trials and other designs
ments showing promise in the marker-positive cohorts, thereby
Here, we provide several examples of master protocols that do
enhancing participation as potential access to experimental treat-
not fit neatly in the categories of basket or umbrella trials, includ-
ment is ensured for all eligible patients. FOCUS4 opened to en-
ing strategy trials and Bayesian adaptive platform designs.
rollment in 2014 and plans to follow patients for up to 5 years.

ALCHEMIST. An umbrella trial similar to FOCUS4 in its design

SHIVA
is ALCHEMIST (Adjuvant Lung Cancer Enrichment Marker In the multi-histology SHIVA trial, 195 patients with solid
Identification and Sequencing Trial), a study of targeted therapy tumors refractory to standard treatments were randomized to
in patients with resectable adenocarcinoma of the lung harboring targeted treatment according to molecular characteristics versus
EGFR or ALK mutations [16, 34, 57, 58]. In ALCHEMIST, fol- non-targeted treatment with standard therapy (physician’s
lowing initial treatment with standard non-targeted therapy, pa- choice), with crossover to targeted therapy upon progression
tients are randomized to experimental targeted agents versus allowed for patients randomized to the control arm [61]. SHIVA
placebo within respective target-enriched cohorts. Because the is considered a ‘strategy’ trial, as the strategy of each patient’s
prevalence of the mutations under study is quite low (10% com- treatment (biomarker-based versus standard) was the differenti-
bined), ALCHEMIST further enrolls all-wild-type patients whose ating feature at randomization, rather than comparison of spe-
tumors undergo whole exome sequencing at registration and at cific treatments. Unfortunately, the experimental targeted
relapse following standard care, so that the natural course of dis- strategy arm of SHIVA failed to show improved progression-free
ease may be studied in this group. survival compared with standard of care [62].

Volume 28 | Issue 1 | 2017 doi:10.1093/annonc/mdw413 | 39

Special article Annals of Oncology
NCI-MPACT [65–69], the utility of Bayesian adaptive randomization depends
on quick marker assessment, a relatively quick end point to in-
Similar to the SHIVA trial in its objectives, NCI-MPACT
form the randomization algorithm, and a slow-to-moderate ac-
(Molecular Profiling-Based Assignment of Cancer Therapy) is an
crual rate to ensure that early adaptations may benefit subsequent
ongoing pilot strategy trial for patients with advanced solid
patients.
tumors harboring a mutation in one of three specific genetic
pathways. In MPACT, patients are randomized in a 2:1 ratio to
receive targeted therapy for their identified pathway mutation CUSTOM
versus a treatment not known to be pathway-specific, with cross- The Molecular Profiling and Targeted Therapies in Advanced
over to targeted therapy allowed upon progression in the control Thoracic Malignancies (CUSTOM) study is a master protocol
arm [63, 64]. Currently, MPACT is studying treatments for the that simultaneously evaluated 5 targeted therapies in 15 different
DNA repair, PI3K, and RAS/RAF/MEK pathways, with a total of patient cohorts defined by 5 groupings of molecular features
four experimental treatment arms (two experimental arms are crossed with three different tumor histologies. One trial arm
included for the DNA repair pathway). It is anticipated that 700 studying erlotinib in NSCLC patients with EGFR mutations dem-
patients will be screened to enroll 180 patients to the initial 4 co- onstrated an overall response rate of 60%; in combination with
horts of the trial; additional pathway/treatment cohorts may be other published reports of erlotinib’s efficacy in this setting, this
added at a later date. Dual end points of this trial include overall arm of CUSTOM was terminated for early efficacy. Studies of tar-
response rate and 4-month progression-free survival. geted therapies in other arms are ongoing [18, 73].

BATTLE trials CREATE

BATTLE was a Bayesian adaptive trial in advanced NSCLC and The cross-tumoral phase II study with Crizotinib (CREATE)
one of the first clinical trials specifically designed to investigate study [19, 74] is evaluating crizotinib’s efficacy in patients with
differential biomarker-driven treatment effects [22, 65, 66]. advanced disease and ALK and/or MET mutations in one of six
BATTLE employed a master protocol and individual protocols heterogeneous tumor types. In CREATE, each tumor type consti-
for each of four treatment arms, among which response-adaptive tutes a subtrial, and each subtrial contains two subcohorts to en-
randomization was used to modify the randomization probabil- roll patients with ALK/METþ versus ALK/MET– tumors.
ities within each of five biomarker-based subgroups based on CREATE is hoping to recruit up to 420 patients from several
observed 8-week disease control rates within each marker– countries, and enrollment is ongoing.
treatment combination. The final results of the BATTLE trial and
its inherent challenge were detailed in several manuscripts [66–
70]. Kim et al. [66] stated that in a follow-up trial, BATTLE-2
[67], a prospectively defined learning period would occur, from
Discussion: considerations and challenges
which only biomarkers showing sufficiently strong predictive in master protocols
ability would be subsequently used to guide patient assignments. New challenges and considerations often affect the conduct and
BATTLE-2 is currently ongoing. feasibility of trials with a master protocol design. Here, we discuss
several in more detail. A detailed review of biomarker-based trial
I-SPY2 designs from a statistical design and analysis perspective was re-
cently given by Renfro et al. [33].
Another biomarker-based Bayesian adaptive design is I-SPY2, a
phase II trial of neoadjuvant treatment of women with locally
advanced breast cancer [22, 71]. In I-SPY2, patients are biopsied New collaborative paradigm
at baseline and directly assigned to one of many biomarker signa- To be successful, a master protocol requires a new collaborative
ture cohorts, wherein patients are subsequently randomized to paradigm defined by the close collaboration of multiple industry,
one of several experimental treatments. The primary end point academic, regulatory, and community oncology stakeholders,
for each cohort is pathologic complete response supported by often including participation by multiple pharmaceutical compa-
longitudinal MRI measurements. A drug performing well within nies providing drugs to the same trial [38]. Experts from many
a specific marker signature (in terms of Bayesian predictive prob- disparate areas are often involved, including cooperative group
ability) triggers adaptive randomization at higher probabilities and local hospital leadership, and specialists in oncology, path-
for subsequent patients enrolled within the same signature, and ology, molecular medicine, computational biology, clinical and
definitively successful drugs are ‘graduated’ to a phase III study translational research, pharmacology, biostatistics, and patient
within the signature. Meanwhile, treatments not showing prom- advocates. This is of clear long-term benefit to the research enter-
ise within a signature are randomized at lower probabilities and prise, but is a short-term challenge as new relationships and trust
are ultimately removed from consideration, and drugs reaching are established.
futility across all signatures are dropped from the trial. This plat-
form design framework allows novel targeted agents of interest to
continually enter and exit the trial protocol in an operationally
Inclusion of marker-negative patients
seamless manner, taking advantage of established infrastructure As described earlier, inclusion of a treatment cohort for all-
and site participation. To date, at least four cohorts have grad- negative or all-wild-type patients within a master protocol design
uated to the phase III setting [22, 72]. As noted by several authors may enhance enthusiasm for the trial, as all otherwise eligible

40 | Renfro and Sargent Volume 28 | Issue 1 | 2017

Annals of Oncology Special article
patients will be offered access to a potentially beneficial experi- Future directions of master protocols in oncology
mental treatment regardless of screening results. In some settings,
As next-generation sequencing continues to develop, master
however, inclusion of a marker-negative cohort may not be feas-
protocols including basket and umbrella trials are likely to see
ible or reasonable, or its patient population may shift due to mid-
increased use with more nuanced assignment of patients to
trial addition or cancellation of marker-positive arms. In the lat-
matched treatments, e.g. to combination treatments according to
ter case, potential regulatory approval of new therapies on the
multiple driving mutations, biomarkers, or pathways [18]. In the
basis of a master protocol’s marker-negative substudy would be
future, molecular evaluation of patients who show remarkable
challenged by a poorly defined patient population for labeling
improvement in their disease following treatment with cancer
purposes.
drugs that have otherwise shown low activity in other patients
(so-called ‘exceptional responders’ as defined by the NCI) may
Classification of patients with multiple genetic motivate future master protocol designs within specific disease
mutations types [78]. The FDA has also demonstrated a willingness to ap-
prove agents that have been evaluated in only a small number of
In many of the master protocols described, patients may poten- patients, even based on single-arm trials, as traditionally large
tially qualify for more than one targeted cohort on the basis of registration trials may never be feasible or ethically appropriate
multiple positive biomarkers. Assignment of such patients to just within some cancer subtypes [79]. How these large shifts in the
one qualifying matched cohort must be decided. If prevalence of standard paradigm of cancer clinical research will ultimately im-
a mutation or biomarker group is not too low and accrual is not a prove success rates for new therapies in oncology remains to be
concern, patients may be assigned to one of the cohorts at ran- seen, but we are confident that if we continue to learn and refine,
dom. In other cases, particularly where one of the mutations is the net impact will be better treatment of cancer patients.
rare and feasibility needs to be optimized, the patient may be dir-
ectly assigned to the rarest mutation or the cohort with the lowest
patient accrual to date. In all of these cases, it is also possible to re-
assess patients for mutations or markers upon initial disease pro- Funding
gression and, on the basis of those results, assign such patients to This work was supported by the National Institutes of Health
second matched cohorts within in the same trial (e.g. NCI [CTSA Grant Number KL2 TR000136 from the National Center
MATCH). for Advancing Translational Science (NCATS) to LAR]; and the
National Cancer Institute at the National Institutes of Health
(R01 CA174779 to DJS). Its contents are solely the responsibility
Effect size versus sample size
of the authors and do not necessarily represent the official views
The sample size versus effect size trade-off is often an issue when of the NIH.
defining targeted cohort-specific objectives within basket, um-
brella, or platform designs. To keep the sample size small within
cohorts and maintain overall trial feasibility, it is often necessary
to target a large effect size (versus a randomized comparator arm Disclosure
or historical control), often with lower power and higher type I The authors have declared no conflicts of interest.
error than traditional phase II or III trials. Several authors such as
Menis et al. [19] and Burock et al. [75] have stated that the goal of
cancer clinical trials in this era of precision medicine should be to References
conduct ‘trials designed to learn’ which lead to ‘trials designed to 1. Chapman PB, Hauschild A, Robert C et al. Improved survival with
conclude’, which begins with identifying large and meaningful vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med
differences within small, molecularly enriched groups of patients, 2011; 364: 2507–2516.
often referred to as ‘home runs’ [76]. 2. Slamon DJ, Leyland-Jones B, Shak S et al. Use of chemotherapy plus a
monoclonal antibody against HER2 for metastatic breast cancer that
overexpresses HER2. N Engl J Med 2001; 344: 783–792.
3. Swain SM. Pertuzumab, trastuzumab, and docetaxel in HER2-positive
Challenges in assay evaluation, validation, and metastatic breast cancer. N Engl J Med 2015; 372: 724–734.
implementation 4. Armado RG, Wolf M, Peeters M et al. Wild-type KRAS is required for
panitumumab efficacy in patients with metastatic colorectal cancer. J
Validation of biomarkers for use in patient screening and clinical
Clin Oncol 2008; 26: 1626–1634.
trials remains fraught with challenges, including a multitude of 5. Jonker DJ, O’Callaghan CJ, Karapetis CS et al. Cetuximab for the treat-
available assessment methods (e.g. immunohistochemistry, ment of colorectal cancer. N Engl J Med 2007; 357: 2040–2048.
fluorescence in situ hybridization, and next-generation sequenc- 6. Zhou C, Wu YL, Chen G et al. Erlotinib versus chemotherapy as first-line
ing), variable reliability (sensitivity and specificity), reproducibil- treatment for patients with advanced EGFR mutation-positive non-
ity, feasibility in terms of tissue requirements and turn-around small-cell lung cancer (OPTIMAL, CTONG-0802): a multicentre, open-
label, randomised, phase 3 study. Lancet Oncol 2011; 12: 735–742.
time, and related costs [19]. Several authors have noted that an
7. Shaw AT, Kim DW, Nakagawa K et al. Crizotinib versus chemotherapy
optimal drug/device co-development program should include in advanced ALK-positive lung cancer. N Engl J Med 2013; 368:
simultaneous evaluation of assay methodology and companion 2385–2394.
drugs from prospective or retrospective analysis of clinical trials 8. Shaw AT, Ou SH, Bang YJ et al. Crizotinib in ROS1-rearranged non-
[35, 77]. small-cell lung cancer. N Engl J Med 2014; 371: 1963–1971.

Volume 28 | Issue 1 | 2017 doi:10.1093/annonc/mdw413 | 41

Special article Annals of Oncology
9. Brahmer J, Reckamp KL, Baas P et al. Nivolumab versus docetaxel in 36. SEER stat facts sheet, cancer of any site. National Cancer Institute
advanced squamous-cell non-small-cell lung cancer. N Engl J Med 2015; Survey, Epidemiology, and End Results Program. http://seer.cancer.gov/
373: 123–135. statfacts/html/all.html
10. FDA Food and Drug Administration. FDA expands approved use of 37. A simple act: increasing clinical trial participation. American Association
Opdivo in advanced lung cancer. http://www.fda.gov/NewsEvents/ for Cancer Research. http://blog.aacr.org/a-simple-act-increasing-
Newsroom/PressAnnouncements/ucm466413.htm clinical-trial-participation/ (25 March 2015, date last accessed)
11. Robert C, Schachter J, Long GV et al. Pembrolizumab versus 38. Saville BR, Berry SM. Efficiencies of platform clinical trials: a vision of
ipilimumab in advanced melanoma. N Engl J Med 2015; 372: 2521–2532. the future. Clin Trials 2016; 13: 358–66.
12. Catennaci DVT. Next generation clinical trials: novel strategies to address 39. Hyman DM, Solit DB, Arcila ME et al. Precision medicine at Memorial
the challenge of tumor molecular heterogeneity. Mol Oncol 2015; 9: Sloan Kettering Cancer Center: clinical next-generation sequencing ena-
967–996. bling next-generation targeted therapy trials. Drug Discov Today 2015;
13. Weinstein IB, Joe A. Oncogene addiction. Cancer Res 2008; 68: 3077–3080. 20: 1422–1428.
14. Weinstein IB. Cancer. Addiction to oncogenes: the Achilles heal of can- 40. Hobbs BP, Chen N, Lee JJ. Controlled multi-arm platform design using
cer. Science 2002; 297: 63–64. predictive probability. Stat Methods Med Res 2016; epub ahead of print.
15. Vogelstein B, Papadopoulos N, Velculescu VE et al. Cancer genome land- 41. Simon R, Geyer S, Subramanian J, Roychowdhury S. The Bayesian basket
scapes. Science 2013; 339: 1546–1558. design for genomic variant-driven phase II trials. Semin Oncol 2016; 43:
16. Redman MW, Allegra CJ. The master protocol concept. Semin Oncol 13–18.
2015; 42: 723–730. 42. Mandrekar SJ, Dahlberg SE, Simon R. Improving clinical trial efficiency:
17. Galbraith S. The changing world of oncology drug development—a glo- thinking outside the box. Am Soc Clin Oncol Educ Book 2015;
bal pharmaceutical company’s perspective. Chin Clin Oncol 2014; 3: 20. e141–e147.
18. Redig AJ, J€anne PA. Basket trials and the evolution of clinical trial design 43. Hyman DM, Puzanov I, Subbiah V et al. Vemurafenib in multiple non-
in an era of genomic medicine. J Clin Oncol 2015; 33: 975–977. melanoma cancers with BRAF V600 mutations. N Engl J Med 2015; 373:
19. Menis J, Hasan B, Besse B. New clinical research strategies in thoracic on- 726–736.
cology: clinical trial design, adaptive, basket and umbrella trials, new end- 44. Flaherty KT, Puzanov I, Kim KB et al. Inhibition of mutated, activated
points and new evaluations of response. Eur Respir Rev 2014; 23: 367–378. BRAF in metastatic melanoma. N Engl J Med 2010; 363: 809–819.
20. Moore KN, Mannel RS. Is the NCI MATCH trial a match for gynecologic 45. Tiacci E, Trifonov V, Schiavoni G et al. BRAF mutations in hairy-cell leu-
oncology? Gynecol Oncol 2016; 140: 161–166. kemia. N Engl J Med 2011; 364: 2305–2315.
21. Billingham L, Malottki K, Steven N. Research methods to change clinical 46. Prahallad A, Sun C, Huang S et al. Unresponsiveness of colon cancer to
practice for patients with rare cancers. Lancet Oncol 2016; 17: e70–e80. BRAF(V600E) inhibition through feedback activation of EGFR. Nature
22. Berry DA. The Brave New World of clinical cancer research: adaptive 2012; 483: 100–103.
biomarker-driven trials integrating clinical practice with clinical re- 47. Klauschen F, Andreeff M, Keilholz U et al. The combinatorial complexity
search. Mol Oncol 2015; 9: 951–959. of cancer precision medicine. Oncoscience 2014; 1: 504–509.
23. Dilts DM, Cheng SK, Crites JS et al. Phase III clinical trial development: a 48. Stenzinger A, Weichert W, Lennerz JK, Klauschen F. Basket trials: just
process of chutes and ladders. Clin Cancer Res 2010; 16: 5381–5389. the end of the first quarter. J Clin Oncol 2015; 33: 2823–2824.
24. Pharmaceutical Research and Manufacturers of America. Drug discovery 49. National Cancer Institute. NCI-Molecular Analysis for Therapy Choice
and development. http://www.phrma.org/sites/default/files/pdf/ (NCI-MATCH) Trial. http://www.cancer.gov/about-cancer/treatment/
rdbrochure022307.pdf clinical-trials/nci-supported/nci-match
25. Pharmaceutical Research and Manufacturers of America. 771 medicines 50. Mullard A. NCI-MATCH trial pushes cancer umbrella trial paradigm.
in development for cancer. http://www.phrma.org/research/cancer Nat Rev Drug Discov 2015; 14: 513–515.
26. DiMasi JA, Reichert JM, Feldman L et al. Clinical approval success rates 51. National Cancer Institute. NCI-Match: status report and future direc-
for investigational cancer drugs. Clin Pharmacol Ther 2013; 94: 329–335. tions. http://www.cancer.gov/news-events/cancer-currents-blog/2016/
27. Sutter S, Lamotta L. Cancer drugs have the worst phase III track record. nci-match-update
Internal Medicine News Digital Network 2011. 52. Genomeweb. NCI-Match sees lots of enthusiasm in initial months; not
28. Amiri-Kordestani L, Fojo T. Why do phase III trials in oncology fail so many matches. https://www.genomeweb.com/molecular-diagnostics/
often? J Natl Cancer Inst 2012; 104: 568–569. nci-match-sees-lots-enthusiasm-initial-months-not-many-matches
29. Nass SJ, Moses HL, Mendelsohn J. Institute of Medicine: A National 53. Signature: bring the protocol to the patient. https://www.signaturetrial.
Cancer Clinical Trials System for the 21st Century: Reinvigorating the com/en/hcp?remember_choice=true#all
NCI Cooperative Group Program. Washington, DC: National 54. Clinicaltrials.gov. Phase 2 study assessing efficacy and safety of crizotinib
Academies Press, 2010. in patients harboring an alteration on ALK, MET or ROS1 (AcSe).
30. Seymour L, Ivy SP, Sargent DJ et al. The design of phase II clinical trials https://clinicaltrials.gov/ct2/show/NCT02304809
testing cancer therapeutics: consensus recommendations from the clin- 55. Kaplan R, Maughan T, Crook A et al. Evaluating many treatments and
ical trial design task force of the National Cancer Institute investigational biomarkers in oncology: a new design. J Clin Oncol 2013; 36: 4562–4570.
drug steering committee. Clin Cancer Res 2010; 16: 1764–1769. 56. Medical Research Council (MRC) Clinical Trials Unit. FOCUS4: mo-
31. Collins FS, Varmus H. A new initiative on precision medicine. N Engl J lecular selection of therapy in metastatic colorectal cancer: a molecularly
Med 2015; 372: 793–795. stratified randomized controlled trial programme. http://www.
32. Renfro LA, Mallick H, An MW, Sargent DJ, Mandrekar SJ. Clinical trial focus4trial.org
designs incorporating predictive biomarkers. Cancer Treat Rev 2016; 43: 57. Gerber DE, Oxnard GR, Govindan R. ALCHEMIST: bringing genomic
74–82. discovery and targeted therapies to early-stage lung cancer. Clin
33. Renfro LA, An MW, Mandrekar SJ. Precision oncology: a new era of can- Pharmacol Ther 2015; 97: 447–450.
cer clinical trials. Cancer Lett 2016; epub ahead of print. 58. Colwell J. Trial offers new model for drug development. Cancer Discov
34. Freidlin B, Korn EL. Biomarker enrichment strategies: matching trial de- 2014; 4: 266–267.
sign to biomarker credentials. Nat Rev Clin Oncol 2014; 11: 81–90. 59. LUNG-MAP. A groundbreaking collaborative approach to clinical trials.
35. Malik SM, Pazdur R, Abrams JS et al. Consensus report of a joint NCI http://www.lung-map.org
thoracic malignancy steering committee: FDA workshop on strategies for 60. Ferrarotto R, Redman MW, Gandara DR et al. Lung-MAP—framework,
integrating biomarkers into clinical development of new therapies for overview, and design principles. Chin Clin Oncol 2015; 4: 36–41.
lung cancer leading to the inception of ‘master protocols’ in lung cancer. 61. Le Tourneau C, Paoletti X, Servant N et al. Randomised proof-of-
J Thorac Oncol 2014; 9: 1443–1448. concept phase II trial comparing targeted therapy based on tumour

42 | Renfro and Sargent Volume 28 | Issue 1 | 2017

Annals of Oncology Special article
molecular profiling vs. conventional therapy in patients with refractory from Southwest Oncology Group randomized trials. J Clin Oncol 2008;
cancer: results of the feasibility part of the SHIVA trial. Br J Cancer 2014; 26: 463–467.
111: 17–24. 71. Barker AD, Sigman CC, Kelloff GJ et al. I-SPY 2: an adaptive breast can-
62. Le Tourneau C, Delord J-P, Gonçalves A et al. Molecularly targeted ther- cer trial design in the setting of neoadjuvant chemotherapy. Clin
apy based on tumour molecular profiling versus conventional therapy Pharmacol Ther 2009; 86: 97–100.
for advanced cancer (SHIVA): a multicentre, open-label, proof-of-con- 72. Park JW, Liu MC, Yee D et al. Adaptive randomization of neratinib in
cept, randomized, controlled phase 2 trial. Lancet Oncol 2015; 16: early breast cancer. N Engl J Med 2016; 375: 11–22.
1324–1334. 73. Lopez-Chavez A, Thomas A, Rajan A et al. Molecular profiling and tar-
63. NCI-MPACT: molecular profiling-based assignment of cancer therapy geted therapy for advanced thoracic malignancies: a biomarker-derived,
for patients with advanced solid tumors. https://clinicaltrials.gov/ct2/ multiarm, multihistology phase II basket trial. J Clin Oncol 2015; 33:
show/NCT01827384 1000–1007.
64. Do K, O’Sullivan Coyne G, Chen AP. An overview of the NCI precision 74. Clinicaltrials.gov. CREATE: cross-tumoral phase 2 with crizotinib.
medicine trials NCI MATCH and MPACT. Chin Clin Oncol 2015; 4: 31. NCT01524926. http://clinicaltrials.gov/ct2/show/NCT01524926?
65. Zhou X, Liu S, Kim ES et al. Bayesian adaptive design for targeted ther- term5NCT01524926&rank51
apy development in cancer—a step toward personalized medicine. Clin 75. Burock S, Mounier F, Lacombe D. How can innovative forms of clinical
Trials 2008; 5: 181–193. research contribute to deliver affordable cancer care in an evolving health
66. Kim ES, Herbst RS, Wistuba II et al. The BATTLE trial: personalizing care environment? Eur J Cancer 2013; 49: 2777–2783.
therapy for lung cancer. Cancer Discov 2011; 1: 44–53. 76. Sleiffer S, Bogaerts J, Siu LL et al. Designing transformative clinical trials
67. Liu S, Lee JJ. An overview of the design and conduct of the BATTLE tri- in the cancer genome era. J Clin Oncol 2013; 31: 1834–1841.
als. Chin Clin Oncol 2015; 4: 33. 77. Polley MY, Freidlin B, Korn EL et al. Statistical and practical consider-
68. Berry DA, Herbst RS, Rubin EH. Reports from 2010 Clinical and ations for clinical evaluation of predictive biomarkers. J Natl Cancer Inst
Translational Cancer Research Think Tank Meeting: design strategies for 2013; 105: 1677–1683.
personalized therapy trials. Clin Cancer Res 2012; 18: 638–644. 78. National Cancer Institute. Exceptional responders initiative: questions
69. Marchenko O, Fedorov V, Lee JJ et al. Adaptive clinical trials: overview of and answers. http://www.cancer.gov/news-events/press-releases/2014/
early-phase designs and challenges. Ther Innov Regul Sci 2014; 48: 20–30. ExceptionalRespondersQandA
70. Lara PNJr, Redman MW, Kelly K et al. Disease control rate at 8 weeks 79. Willyard C. Basket studies will hold intricate data for cancer drug ap-
predicts clinical benefit in advanced non-small cell lung cancer: results provals. Nat Med 2013; 19: 655.

Volume 28 | Issue 1 | 2017 doi:10.1093/annonc/mdw413 | 43