The use of external control arms in oncology is an evolving area that is generating increasing interest among regulators and sponsors alike. While traditional control arms in randomized trials are the mainstay of clinical research, long-running oncology trials highlight why the inclusion of a control group that receives a placebo isn’t always feasible or ethical. Rare disease research faces notoriously challenging recruitment, and no cancer patient wants to land on the placebo side of a trial when they could receive potentially life-saving treatment.

Fortunately, real world evidence (RWE) is increasingly being accepted as a mechanism to answer these challenges. Rather than populate a control arm with randomized patients from inside a trial’s recruited population, researchers can generate insights from an external comparator—a real-world cohort of patients with characteristics similar to those who are being studied in a clinical trial. This approach allows regulators to compare the safety and efficacy of new treatments to the existing standard of care, while ensuring all trial participants receive potentially beneficial options.

External control studies could be a suitable alternative to a traditional randomized trial, but there are myriad considerations to address validity issues and minimize bias. In February 2023, the U.S. Food and Drug Administration (FDA) released guidance for the use of external control arms, stating that, “the suitability of an externally controlled trial design warrants a case-by-case assessment.”¹ So, how to present the strongest case? This paper will focus on key regulatory and operational considerations to give trials the best chance of success with external comparators.

Takeaways from the FDA guidance on external comparators

The new guidance provides recommendations to sponsors and investigators, addresses considerations for the design and analysis of externally controlled trials, and describes considerations related to communicating with the FDA. IQVIA has long been at the forefront of exploring novel uses of RWE and was intimately involved with the development of this guidance through collaboration with Real World Evidence Alliance, an independent coalition of data and analytics organizations that provides input to regulatory bodies from the industry's perspective.

Key points from the guidance include:

• The evidentiary standard has not changed. This most recent guidance builds on the FDA’s 2018 RWE framework for using real world data (RWD) for regulatory decision making.² That means that, just like in the past, the FDA is looking to ensure data are fit-for-purpose, provide adequate scientific evidence to answer the question at hand, and meet the agency’s regulatory requirements.
• The level of regulatory rigor will depend on how sponsors are using the external comparator and for what purpose. The level of rigor the FDA deploys to assess a study design will depend on the question the study seeks to answer using RWD. For example, studies looking at contextualization around disease progression from a registry or previously published study will require less regulatory input. External control, on the other hand, is the most stringent application of FDA’s regulatory requirements as the RWD could potentially be used as primary evidence to support efficacy and safety. Remember, the FDA has not altered the evidentiary bar. These situations, which are the focus of this blog, involve conducting patient-level data-matching and then utilizing that data in the regulatory submissions.
• The FDA expects sponsors to have a finalized protocol before initiating an external control trial. The FDA is looking to dissuade sponsors from adding the external controls after the completion of a single-arm trial, and instead, proposes upfront planning to allow inclusion of the external control arm design and analytic approach along with the trial protocol. It also asks that sponsors pre-specify their plan, how they want to measure the different aspects of the design and the data, and how they would analyze any confounders and reduce sources of bias. Design elements for consideration include the study population (Is it exchangeable with the experimental arm? Does it have the same eligibility criteria?); treatment (Are the treatments comparable?); immortal time bias (Are the timings from exposure to outcomes the same?); and, outcome assessment.
• Be ready to show assessment of the pros and cons of using external control data. External controls could be created from RWD or from a historical trial. Regulators want sponsors to consider the pros and cons of using one type of external control over the other, as both may have advantages and disadvantages. For example, control arms developed from historical control data is protocolized and captured in a clinical setting, whereas RWD may be more contemporaneous and aligned with recent standard clinical practices. To help with this assessment of RWD, the FDA guidance lays out what to look for across 10 comparability considerations: time period, geography, diagnosis, prognosis, treatment, outcomes, follow-up period, intercurrent trials, missing data, and other factors.
• A statistical analysis plan should evaluate comparability and effect size. External controls are typically considered a good mechanism when there is a huge effect size, so the plan should show how to manage that effect size while reducing the bias. Sponsors should also have strategies prepared to address missing or misclassified data.

Given these factors, there is no doubt that early planning is key to successfully using external comparators. The remainder of this paper will discuss considerations for using that planning time in the most efficient and effective way possible.

External comparators in oncology bring unique strengths and challenges

Our knowledge of cancer has dramatically improved in recent years. We now know that, despite some common hallmarks, the concept of cancer comprises multiple different diseases. Even within one specific cancer type there can be numerous subtypes differentiated by features, e.g., biomarker or gene expression, histopathological features, and clinical features. To add to the complexity, there are also geographic and race/ethnicity variations not only in the incidence and mortality of these diseases, but also in their treatment and clinical presentation.

While all this knowledge is incredibly helpful in understanding the complexity of various cancers, it also poses a challenge to the traditional clinical trial construct. Namely, a randomized controlled clinical trial will sometimes enroll only a relatively homogenous patient population, challenging its representativity and generalizability to the broader population typically seen in the clinic.

This is where external comparators offer a huge benefit; they help us reach an extended clinical population with great specificity and expand our trial to be more inclusive and reflective of the clinical situation at hand. External comparators also allow us to address treatment effects in rare tumor types and subpopulations that are not sufficiently prevalent to be represented in a statistically powerful way in a randomized controlled trial. We can even address differences in geographical regions by adapting to specific regulatory requirements particular to a given region or country, allowing our studies to reflect traditionally underrepresented populations, like minority groups.

On the other side, variations in how certain cancer markers are tested and classified, as well as rapid change in the treatment landscape, have the potential to pose challenges to external comparator validation and should be well considered upfront. For example, the use of assays and classification systems of predictive and prognostic markers may vary across geographies and institutions. The same is true for clinical drivers of therapeutic decision and safety management. With the rapidly changing oncology landscape, an external comparator started later than the randomized trial needs to acknowledge the potential for changes in clinical practice, such as the introduction of novel therapies or potential safety issues that were not established when designing the randomized trial.

When we are assessing biomarkers, we also must consider whether retrospective tissue information will suffice, or if we need to have a prospective collection of tissue, e.g., in tumors where we know that expression can change due to prior treatment. Considerations should also be made for genetic biomarker testing, where validated assays are not always available. Finally, both efficacy and safety endpoints are dependent on the quality of the source database. The choice of which endpoints to collect and how to collect them needs to be well thought through and should aim to align with both the reflected randomized clinical trial and clinically relevant outcome measures for the study treatment.

Proposed approaches for building successful external comparators

These are complex considerations, but IQVIA’s extensive experience building and executing external comparator studies has surfaced best practices to ease the process. Below are several approaches to support their successful execution:

• Take a modular approach to building your external comparator. A modular approach will allow sponsors to address key risks early and allows for iterative study design. At IQVIA, we recommend a modular approach play out as follows:
  • Define the study’s needs, such as the target stakeholders and their evidence expectations (1-2 months)
  • Design the study, including discovery of what secondary data is available to support an external comparator, feasible designs, and how and when they should be executed (3-5 months)
  • Test and engage stakeholder feedback on the proposed design (3-6 months)
  • Generate fit-for-purpose evidence (12-24 months)
• Maximize the quality of evidence through careful study design. The internal validity of external comparator patients lies in the level of “exchangeability” with the trial that can be achieved. In this regard, the target trial emulation framework created by Miguel Hernán and James Robins offers a two-step structured process for early alignment between the trial and the external comparator.³ Remember to find an acceptable level of covariate balance between treatment groups, address missing data, and conduct bias analyses, as needed.
• Consider several statistical approaches that work well with external comparator studies. At IQVIA, we consider three approaches suitable for external comparators, each with varying strengths, granularity, and completeness required:
  • Real-world benchmark, which describes the demographics, clinical characteristics, and treatment outcomes for the treatment group and benchmark patients
  • Indirect comparison, which assigns larger weights to the outcome of patients from the trial whose baseline characteristics are closer to the baseline of the comparator population
  • Direct comparison, the most valid of the three which allows direct inputs to modeling support health technology assessment submissions
• Address common challenges occurring in the study design, data capture, and analysis phases with meticulous study planning and a strategic analytical approach. A recurring inquiry is whether precise measurement is required for valid inference, or if a proxy measurement or valid imputation is sufficient. Analyze and make an early action plan to address these common challenges at the outset of the study so they do not create delays downstream.
• Mitigate bias through careful design. For example, input from key opinion leaders can help with understanding of the local context and correct for potential confounding by indication. To mitigate bias arising at the point of data capture, consider conducting a probabilistic bias analysis informed by existing literature or expert knowledge.
• Consider external validation of study outcomes on a subset of patients. This is an option to avoid possible assessment bias and would include carrying out an independent central review (ICR) of outcomes where two independent reviewers are blinded to treatment and physician evaluation, with adjudication by a third reviewer. Concordance should then be assessed between ICR and treating physician evaluation.

Best practices for external comparators are still being established and debated within the scientific community, and with the FDA’s recent guidance, these will continue to take shape. With the current oncology R&D landscape focused in areas of high unmet need, external comparator studies are poised to play a key role in bringing life-changing—and potentially life-saving—treatments to patients.

References:

1. FDA Guidance for Industry. Considerations for the design and conduct of externally controlled trials for drugs and biological products. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-design-and-conduct-externally-controlled-trials-drug-and-biological-products.
2. FDA Framework for Industry. Framework for FDA’s real-world evidence program. https://www.fda.gov/media/120060/download.
3. Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. Am J Epidemiol. 2016 Apr 15;183(8):758-64. doi: 10.1093/aje/kwv254. Epub 2016 Mar 18. PMID: 26994063; PMCID: PMC4832051.