External Controls in Cancer Clinical Trials –Challenges and OpportunitiesRajeshwari Sridhara, Ph.D.Director, Division of Biometrics V, CDER, FDAAcknowledgements:Jacqueline Corrigan-Curay, J.D., M.D. (Slides on Framework)Director, Office of Medical Policy, CDER, FDAYuan-Li Shen, Ph.D. (Slides on Examples of historical controls)Acting Associate Director, Division of Biometrics V, CDER, FDASridhara CTAC 2019

Potential Use of External Controls External controls could be considered to:– Understand natural history of disease– Designing future studies by establishing SOC effect– Used in place of randomized control arm (historical control)– Compare efficacy across treatment arms by supplementing concurrentcontrols in a prospective trial Source of data for the external control determines potential useSridhara CTAC 20192

Regulation and Guidances 21CFR 314.126“ historical control designs are usually reserved for special circumstances.Examples include studies of diseases with high and predictable mortality (forexample, certain malignancies) and studies in which the effect of the drug isself-evident (general anesthetics, drug metabolism)” ICH E10 (2001)Describes strategies for choosing a control group for clinical trialsintended to demonstrate efficacy. Considerations for using externalcontrols are described in Section E:“The inability to control bias restricts use of the external control design tosituations in which the effect of treatment is dramatic and the usual courseof the disease highly predictable”Sridhara CTAC 20193

21st Century Cures Deliverables FDA shall establish a program to evaluate the potential use of real worldevidence (RWE) to:– Help support approval of new indication for a drug approved under section505(c)– Help satisfy post-approval study requirements Program will be based on a framework to be issued by - 2018 Real-World Data (RWD) are data relating to patient health status and/or the delivery of healthcare routinely collected from a variety of sources. Real-World Evidence (RWE) is the clinical evidence regarding the usage and potential benefits orrisks of a medical product derived from analysis of RWD.4Real world evidence means data regarding the usage, or the potential benefitsor risks, of a drug derived from sources other than traditional clinical trials4

Published Dec. 2018 Intended for drug andbiological products Outlines FDA’s plan toimplement the RWE program Multifaceted program– Internal process– Guidance development– Stakeholder engagement– Demonstration projectsSridhara CTAC 20195

Incorporating RWE Into EvidenceGenerationMany factors must be considered at the same timeRegulatoryQuestionRelevancyEfficacy or safetyRelationship to availableevidenceValidationClinical context: rare,severe, or life-threatening,unmet needQualityassurance/controlNature of endpoint/concerns about biasRWDSridhara CTAC 2019FDA Guidance- Best Practices for Conducting and ReportingPharmacoepidemiologic Safety Studies Using Electronic Healthcare Data, 2013Methods/Design6

Framework for Evaluating RWD/RWE forUse in Regulatory Decisions The study conduct meets FDA regulatoryrequirement– Informed consent, appropriate oversightand monitoring,– Appropriate data standards for integrationfrom various sourcesSridhara CTAC 20197

Framework for Evaluating RWD/RWE forUse in Regulatory Decisions The trial or study design used to generate RWE canprovide adequate scientific evidence to answer or helpanswer the regulatory question– Randomized designs using RWD (explore pragmaticRCT),– Non-randomized, single arm trials with external control(guidance in development),– Observational studies (retrospective, prospective, roleof existing evidence (e.g.: natural history of disease –guidance under development))Sridhara CTAC 20198

Addresses importance of several aspectsincluding adequate description andunderstanding of natural history of thedisease, adequate understanding ofpathophysiology of the disease and drug’smechanism of action, nonclinicalpharmacotoxicology and human toxicologyconsiderations and selection of outcomeassessments and endpoints. Natural history studies can be retrospectiveor prospective and cross-sectional orlongitudinal studies Historical/external controls can beconsidered in serious rare diseases withunmet medical need provided disease ispredictable, such as high mortality, and thedrug effect is large and self-evident.Sridhara CTAC 2019

Natural history studies that systematically andcomprehensively capture data can help identify ordevelop biomarkers as a diagnostic biomarker, prognosticbiomarker and useful in guiding patient selection anddose selection in drug development programsUse of natural history study data: Adequate control to discriminate outcomes caused bynew drug from outcomes caused by other factors.Historical controls may be used as controls; however, maynot control certain biases. Use of external control assumes similarity betweentreated and control group with respect to diseaseseverity, duration of disease, prior treatments, and otheraspects that could affect outcomes and the timing ofoutcomes. Epidemiological approaches can be used to reduce bias.However, critical disease characteristics may not havebeen assessed in the historical/external control andstandard of care may have changed over a period of time.Sridhara CTAC 2019

Lessons From Safety Appropriateness of data source Pre-specified study protocoland statistical analysis plan Selection of study population –explicit inclusion and exclusioncriteria Exposure ascertainment Outcome ascertainment –validation, linkage Confounding adjustment –propensity score method Sensitivity analysis - ances/ucm243537.pdf.11

Consideration of using historical control arm Selection of patient population: blinded review; pre-defined eligibility criteria;contemporaneous patient cohort Endpoint ascertainment: prefer more objective endpoints; pre-specifiedcriteria; blinded review Comparable assessment timing and methods Adequate size of the historical control data Pre-specify/include important prognostic & confounding variables inthe data/analyses An adequate statistical analysis plan should be in placeSridhara CTAC 201912

Challenges for Drug Development in Rare Diseases Small number of eligible patients to participate in a given study Geographic distribution of patients Lack of knowledge about the clinical course/natural history ofdisease Dissimilar diseases Lack of appropriate comparator treatmentwww.fda.govSridhara AAADV 201913

Sources and Challenges of External ControlSources: Past clinical trials Registry Data Case Studies/Literature Real world dataChallenges: Collection of all confounders/factors that influence treatment assignment Unmeasured disease and patient characteristics Frequency of assessments (clinical trial vs. external control) and assessment method Time Lag Index date Survivor bias Follow-up timeSridhara CTAC 201914

Example 1: Blincyto supplemental approval Approved on 3/29/2018 for patients with precursor B-cell acute lymphoblasticleukemia (B-cell ALL), CR1 & CR2 with MRD Supported by a single arm study MT103-203-- Primary efficacy endpoint: complete MRD response within the firstcycle Supporting results (exploratory; not included in the label): Compare the single arm trial(Study MT103-203: reduce from n 113 to n 73) vs historical control arm (Study20120148;n 182)– Efficacy Endpoints: RFS and OS– Propensity score adjustment Selected baseline factors are balanced by using a weight function stabilizedinverse probability of treatment weight (sIPTW)Sridhara CTAC 201915

Blincyto : Results in the label and Supportive Resultscomparing with a HC arm Label: Based on n 86-- % MRD - : 81.4% (95%CI: 72%, 89%)-- Median hematological RFS : 22.3 months Sponsor’s supportive results (FDA presented in 3/7/18 ODAC):Sridhara CTAC 201916

Limitation in Blincyto supportive results using the HC arm3/7/2018 ODAC:While RFS appears to be in favor of Blincyto treated arm, there arelimitations of the propensity score adjusted analyses based on HC arm: 35% of Blincyto patient data excluded to match with external control group Confounding due to subsequent treatment, e.g. differential rates of HSCTand data are not contemporaneous Differential follow-up time between two armsSridhara CTAC 201917

Example 2: Vistogard Approved on 12/11/2015 for patients who experience 5-FU or capecitabine overdose or early onset of severe or life-threatening toxicities within 96 hours after5-FU or capecitabine treatment Studies report 0.5% mortality among 300,000 patients in U.S. receivingflurouracil due to toxicity This approval was supported by: Two expanded access single-arm studies :401.10.001 (US; n 60) and WELL401 (US&EU; n 75) Retrospective historical case report : n 25 Endpoint for the pivotal study: Survival at day 30 or resumption of chemotherapyif the resumption occurred first prior to 30 days 96% vs. 16% (historical control) survival rateSridhara CTAC 201918

Opportunities: Clinical Trial Design OptionsKey consideration: reduce sample size and maximize allocation to investigationdrugKey feature: adaptive design RCTs 2:1 or 3:1 randomization allocation with decision criteria to stop early Supplement concurrent control in a RCT with external control Crossover design with each patient as his/her own control Single arm trial with external control Basket/Umbrella/Platform trials with a Master ProtocolInternational Collaborations? Others? – Possibility to conduct RCTRegulatory considerations/flexibilities? – Depends on the disease and availableoptionsSridhara CTAC 201919

Conclusion When feasible RCT is the best way to understand, evaluate a treatment effect –takes care of known and unknown confounding factors Single arm trials supported with historical controls in general reserved for specialcircumstances If a historical/external control arm is used to support a submission, adequatedata based on pre-determined patient selection criteria and pre-specifiedstatistical analysis plan are required. FDA’s Framework serves as a roadmap for more fully incorporating RWD andRWE into the regulatory paradigm RWE remains a top FDA priority FDA is committed to understand its full potential; Multi-stakeholder effortSridhara CTAC 201920

If a historical/external control arm is used to support a submission, adequate data based on pre-determined patient selection criteria and pre-specified statistical analysis plan are required. FDA's Framework serves as a roadmap for more fully incorporating RWD and RWE into the regulatory paradigm, RWE remains a top FDA priority,