February 2021

Oncology Clinical Trials: An Emerging Paradigm Shift in Study Design

This white paper explains the different study designs for clinical trials and the emergence of master protocols as a way to simultaneously evaluate multiple drugs and/or disease populations in multiple studies. Find out the key features of master protocol trial designs, including the practical and statistical challenges associated with designs of this type.
Female scientist holding up a beaker  while a another scientist watches

In oncology, patients are generally classified by their primary cancer and the stage of the disease, and randomized controlled trials are conducted for each patient population to create standard therapies. Historically, cytotoxic agents have been developed based on this perspective, but in recent years, researchers have developed an improved understanding of cancer cell growth and progression at both the cellular and molecular levels. The identification of molecular markers or genetic mutations enabled classification of particular tumor types into several subtypes. Oncology drug development is shifting from treatments centered on cytotoxic agents to those using molecularly targeted agents, which act selectively on cancer cells. As the science behind therapeutic interventions has grown, clinical trial design methodologies in oncology have also evolved, leading to a substantial paradigm shift in the development of new cancer treatments.

Phase I Study Design: Rule-Based vs. Model-Based

Phase I trials are classically designed to assess the safety and maximum tolerated dose (MTD) of a novel drug or treatment. Phase I clinical trials in oncology are typically small, single-arm, open-label, sequential dose-escalation studies that include patients with a good performance status whose cancers have progressed despite standard treatments. The guiding principle for dose escalation in Phase I trials is to avoid exposing too many patients to subtherapeutic doses while preserving safety and maintaining rapid accrual.

There is currently a variety of methodologies for the execution of Phase I trials, ranging from classic simple rule-based design, such as 3 + 3 design, to sophisticated computational models involving Bayesian algorithms.

Although statisticians recommend model-based designs, most Phase I trials use rule-based designs such as the 3 + 3 design, primarily because of its simplicity.

With the emergence of new targeted therapies where the fundamental assumption of a linear relationship between dosage and toxicity may not be true, the need for a tailored design approach has become more apparent. There is currently no consensus on the most appropriate design for Phase I studies, requiring the review of the advantages and limitations of each design. Commonly used designs for Phase I studies are categorized into three groups: traditional rule-based design, pharmacokinetically (PK) guided dose-escalation design and model-based design.

Table 1: Understanding Phase I Study Designs1-7

Design Features Advantages Limitations
Traditional rule-based design
Standard 3 + 3 design

Three patients treated per dose level:

  • If no DLT, dose is escalated for the next cohort of three patients
  • If one DLT, three additional patients are treated at this level with dose escalation only if no additional DLTs
  • If two or more DLTs, prior dose level is defined as MTD

MTD is defined as the highest dose where no more than one DLT is observed among six patients.

Easy to implement and safe

Provides some data on PK interpatient variability

Many patients are treated at subtherapeutic doses

Slow dose escalation

Statistical simulations suggest RP2D often at lower doses than other designs

Only the result from the current dose is used for determining the dose of next cohort of patients;. iInformation on other doses is ignored.

Accelerated titration design

Acceleration and escalation phase in one design

More rapid dose escalation

May expose a greater proportion of patients at higher doses

Data from all patients, cumulative toxicity, and interpatient variability can be fit to a model to establish the RP2D

If model fitting is not performed (as is often the case in clinical practice):

  • Intrapatient dose escalation may mask cumulative or delayed toxicities.
  • Difficult interpretation of the results when intrapatient dose escalation is allowed
  • Uncertainty about the RP2D
Pharmacokinetically guided dose escalation design

Requires real-time PK measurement and assessment for dose modification

Assumes DLT can be predicted by plasma drug concentration

More rapid dose escalation

Provides some data on PK interpatient variability

Need to obtain real-time PK results

Interpatient variability may hamper dose escalation

Model-based design
Parametric - continual reassessment method (CRM)

A target level of toxicity defined at baseline

Increased dose levels defined with initial expectations of the probability of DLT at these doses to construct a statistical dose-toxicity model

May improve the accuracy of MTD level estimation based on statistical evidence

May overestimate dose for MTD

Uncertainty about toxicity of investigational agent may be reflected in initial dose-toxicity model

Escalation with overdose control (EWOC) method

Essentially a modified continual reassessment method with additional safety measures

Allows flexible patient enrollment, and conforms to the ethical goal of maximizing the number of patients receiving optimal doses

Dose-toxicity curve constantly remodeled requiring additional statistical support

Modified toxicity probability interval (mTPI) method

Bayesian framework used to calculate posterior probabilities of intervals, reflecting relative distance between toxicity rate of each dose level to the target probability with a fixed sample size

Fewer patients treated at doses above MTD

Software provided online

Computationally intensive

Non-parametric overdose control method

Each dose assignment is guided via a feasibility bound, which thereby can control the number of patients allocated to excessively toxic dose levels.

Have both the robustness and safety properties

Well suited to use for drug combinations

Semi-parametric dose finding method

May be applied to 3 + 3 design or CRM

Time to toxicity modeled using Kaplan-Meier estimator

May shorten duration of trial while maintaining accurate determination of RP2D

DLT = dose-limiting toxicity, MTD = maximum tolerated dose, RP2D = recommended Phase II dose

Phase II and III Study Design8-13

Historically, Phase II trials in oncology settings use a single-arm study design, to determine tumor response rate of a novel agent or a combination regimen to warrant further testing in a Phase III study, which establishes clinical efficacy.

In single-arm Phase II oncology studies, one-stage or two-stage designs are used. Single-stage designs are the simplest, examining a single group (single-arm) of patients (e.g. A’Hern design). In two-stage designs, patients are divided into two groups (or stages). At completion of stage one, an interim analysis is done to determine if stage two should be conducted. Simon’s design is the most frequently used single-arm Phase II design in oncology drug development. This design minimizes the maximum (minimax design) or expected sample size under the null hypothesis (optimal design) among all designs with the same significance level or power, and it includes the possibility to stop for futility after the first stage. Another design, known as Bryant and Day design, incorporates toxicity considerations into the design of two-stage Phase II clinical trials.

An uncontrolled single-arm Phase II, with tumor response as an endpoint, or a controlled multiple-arm randomized Phase II, with survival (or similar efficacy parameter) as an endpoint, is widely discussed and debated.

While transitioning from Phase II to Phase III trials with new molecularly targeted agents, new considerations and advanced features such as adaptive and biomarker-based trial designs should be taken into consideration.

Moving Away From a One-Size-Fits-All Approach

With expanding knowledge of human biology and biomarkers, oncology therapies are increasingly moving away from a one-size-fits-all approach. Advances in genomics, particularly in human genome sequencing, have improved our ability to differentiate cancers by their genetic mutation. These discoveries have energized developments in precision oncology (an innovative approach to cancer treatment where therapies are designed and selected to specifically target cancer based on their genetic mutations).

Traditionally, drugs are approved based on the tumor type; however, scientific advances in the last decade have led to the development of drugs tailored to the molecular profile of patients (e.g. BRAF mutant melanoma, HER2-positive breast cancer, KRAS wild-type colorectal cancer, EGFR or ALK-mutated lung cancer, etc.).

Some of the available biomarker-based clinical trial designs are listed in the table below.

Table 2: Understanding Biomarker-Based Study Designs14

Design Features Advantages Limitations
Enrichment design

Only marker positive patients are enrolled

Good for biomarker with clear evidence and/or low prevalence

Cannot gather treatment information for all populations

Cannot test for the companion diagnostic tool validity

Interaction design

Both biomarker-positive and -negative patients are enrolled

Marker status is used as a stratification factor and patients are randomized to treatment groups within each “marker-based subgroup”

Complete treatment information for overall population

Large sample size with high cost

Biomarker effect might be diluted

Biomarker strategy design

Design focused on the role of a biomarker in the treatment decision-making process

Patients are randomized to treatment strategy: based on biomarker vs. not based on biomarker (e.g., physician’s choice)

Complete information gathered

Able to test companion diagnostic tools

Scale and cost might be large

Adaptive enrichment design

Two-stage design with subpopulation selection at interim analysis

Flexible design with smaller expected sample size than all-comers but still contains information for biomarker-negative group

Logistically complicated

Need complex simulation studies to determine sample size

Multiple prerequisites needed to proceed

Group sequential design

Designs with several interim analysis to make go/no-go decisions

Smaller expected sample size

Decisions can be made during interims to save cost

Proven effective treatment can be accessible to all patients early

Logistically challenging

Extra complexity involved in the biomarker-related trials

Seamless design

A design in combination of two phases

Smaller expected sample size

Shortened duration of the drug development process

Logistically challenging

Complexity in sample size calculation

Master Protocols

In traditional trial designs, a single drug is tested in a single disease population in one clinical trial. A new trend has emerged toward investigating multiple target–treatment pairs in parallel, either within or across recognized tumor types.

Master protocol is a term describing trial designs using a single protocol to simultaneously evaluate multiple drugs and/or disease populations in multiple studies, allowing for efficient and accelerated drug development.

Master protocols are often classified into basket trials, umbrella trial, and platform trials. These trial designs and their corresponding pros and cons are listed below.

Table 3: Master Protocol Trial Designs15-18

Design Features Advantages Limitations
Basket trial

Evaluates one targeted therapy on multiple diseases or multiple disease subtypes

Patients with cancers of different histologies are enrolled based on presence of one specific molecular aberration

Generally used in early development phase, single-arm sub-studies

Relatively small sample size

Increased hit rate by enrolling patients with rare molecular features across tumor types

Offers an array of novel therapeutic agents to a broad group of patients who may benefit

Greater feasibility in evaluating rare diseases

Risk of overlooking impact of tumor histology type because tumor responses can differ for different tumor types despite targeting the same mutation (e.g., Vemurafenib, an oral BRAF inhibitor, in metastatic melanoma vs. metastatic colon cancer)

Prognostic heterogenicity across tumor type

Single-arm sub-studies generally require a tumor response rate endpoint (with a high bar)

Umbrella trial

Evaluates multiple targeted therapies for at least one disease

Commonly used in mid-to-late phase trials

Allow for better understood target treatment hypotheses

Often randomized with futility stopping to Phase III

Easier for more treatments to be tested efficiently

Ability to draw meaningful conclusions specific to tumor type minimizes heterogeneity within a given trial cohort

Overall trial feasibility, particularly within rare diseases, is compromised

Larger size, particularly when sub-trials are randomized

Longer duration

Difficulty enrolling rare-molecular subtypes of a single tumor type

Platform trial

Multiple treatments are evaluated simultaneously within a single master protocol, and allows adaptive features

With inclusion of an adaptive design, a platform trial offers flexible features such as dropping a treatment arm for futility, declaring one or more treatments superior or adding new treatment arms to an ongoing trial

Operational and statistical challenges (e.g. regulators contending with new statistical challenges)

Challenges and Considerations of Master Protocol Design

Trial success requires considering the practical and statistical challenges that often affect conduct and feasibility of clinical studies with master protocol designs.

  • Close collaboration between industry, academia, regulatory bodies and the medical community is necessary.
  • The sample size versus effect size trade-off is often an issue when defining the sample size. Keeping the sample size small within cohorts and maintaining the overall trial feasibility often require targeting a large effect size.
  • Inclusion of a marker negative cohort is encouraged but may not be possible in certain settings.
  • Patients may potentially qualify for more than one targeted cohort based on multiple positive biomarkers. Treatment assignment in such patients should be discussed at the outset of study design.

Next Steps

Substantial advances in the molecular understanding of cancer and clinical trial design methodology have occurred over the past few decades. Recently, use of master protocols has rapidly increased and agnostic-histology approvals are anticipated to continue to grow. To keep up with these changes, researchers should continue working on improving operational efficiency through development of novel trial designs, strategies for early-stage decision-making, and early selection of candidate drugs with a high likelihood of success.

To successfully plan and execute your oncology clinical trial with master protocol design, Amarex’s experts are here to guide you through this complex process. Our clinical team has in-depth knowledge and expertise in oncology clinical research applying master protocols that advance developments in precision medicine. Since 1998, Amarex has worked on nearly 600 projects, including 100 projects in oncology, which has led to the development of efficient, cost-effective product development solutions, tailored to our clients’ needs.


  1. Le Tourneau C, Lee JJ, Siu LL. Dose escalation methods in phase I cancer clinical trials. JNCI: Journal of the National Cancer Institute. 2009 May 20;101(10):708-20.
  2. Hansen AR, Graham DM, Pond GR, Siu LL. Phase 1 trial design: is 3+ 3 the best? Cancer control. 2014 Jul;21(3):200-8.
  3. Love SB, Brown S, Weir CJ, Harbron C, Yap C, Gaschler-Markefski B, Matcham J, Caffrey L, McKevitt C, Clive S, Craddock C. Embracing model-based designs for dose-finding trials. British journal of cancer. 2017 Jul;117(3):332-9.
  4. Rogatko A, Cook-Wiens G, Tighiouart M, Piantadosi S. Escalation with overdose control is more efficient and safer than accelerated titration for dose finding. Entropy. 2015 Aug;17(8):5288-303.
  5. de Araujo Toloi D, Jardim DL, Hoff PM, Riechelmann RS. Phase I trials of antitumour agents: fundamental concepts. ecancermedicalscience. 2015;9.
  6. Ananthakrishnan R, Green S, Chang M, Doros G, Massaro J, LaValley M. Systematic comparison of the statistical operating characteristics of various Phase I oncology designs. Contemporary clinical trials communications. 2017 Mar 1;5:34-48.
  7. Jaki T, Clive S, Weir CJ. Principles of dose finding studies in cancer: a comparison of trial designs. Cancer Chemotherapy and Pharmacology. 2013 May 1;71(5):1107-14.
  8. Brown SR, Gregory WM, Twelves CJ, Buyse M, Collinson F, Parmar M, Seymour MT, Brown JM. Designing phase II trials in cancer: a systematic review and guidance. British journal of cancer. 2011 Jul;105(2):194-9.
  9. Seymour L, Ivy SP, Sargent D, Spriggs D, Baker L, Rubinstein L, Ratain MJ, Le Blanc M, Stewart D, Crowley J, Groshen S. The design of phase II clinical trials testing cancer therapeutics: consensus recommendations from the clinical trial design task force of the national cancer institute investigational drug steering committee. Clinical Cancer Research. 2010 Mar 15;16(6):1764-9.
  10. Simon R. Optimal two-stage designs for phase II clinical trials. Controlled clinical trials. 1989 Mar 1;10(1):1-0.
  11. Bria E, Di Maio M, Carlini P, Cuppone F, Giannarelli D, Cognetti F, Milella M. Targeting targeted agents: open issues for clinical trial design. Journal of Experimental & Clinical Cancer Research. 2009 Dec;28(1):1-7.
  12. Roberts Jr TG, Lynch Jr TJ, Chabner BA. The phase III trial in the era of targeted therapy: unraveling the “go or no go” decision. Journal of Clinical Oncology. 2003 Oct 1;21(19):3683-95.
  13. Ocana A, Amir E, Vera-Badillo F, Seruga B, Tannock IF. Phase III trials of targeted anticancer therapies: redesigning the concept. Clinical Cancer Research. 2013 Sep 15;19(18):4931-40.
  14. He P. Personalized medicine: challenges in biomarker-related clinical trial design. Clinical Investigation. 2015 Feb;5(2):175-88.
  15. Hirakawa A, Asano J, Sato H, Teramukai S. Master protocol trials in oncology: review and new trial designs. Contemporary clinical trials communications. 2018 Dec 1;12:1-8.
  16. Renfro LA, Sargent DJ. Statistical controversies in clinical research: basket trials, umbrella trials, and other master protocols: a review and examples. Annals of Oncology. 2017 Jan 1;28(1):34-43.
  17. Park JJ, Siden E, Zoratti MJ, Dron L, Harari O, Singer J, Lester RT, Thorlund K, Mills EJ. Systematic review of basket trials, umbrella trials, and platform trials: a landscape analysis of master protocols. Trials. 2019 Dec;20(1):1-0.
  18. Food and Drug Administration. Master Protocols: Efficient Clinical Trial Design Strategies to Expedite Development of Oncology Drugs and Biologics. Guidance for Industry. Federal Register. https://www. federalregister.gov/documents/2018/10/01/2018-21313/master-protocols-efficient-clinical-trial-design-strategies-to-expedite-development-of-oncology. 2018.

About the Authors