On March 25, 2019, FDA issued a draft guidance, “Rare Diseases: Natural History Studies for Drug Development,” to help inform the design and implementation of natural history studies that can be used to support the development of drugs and biological products for rare diseases (hereinafter “Rare Disease Natural History guidance”). This is the latest in a number of draft rare disease-focused guidance documents released by FDA (e.g., on common issues in drug development in February 2019 here and human gene therapies in July 2018 here).
The existence of a stand-alone guidance document on this topic demonstrates FDA’s recognition of the important role natural history studies play in rare disease drug development, which is understandable given that natural history has been underutilized as well as underappreciated when used. In addition, when natural history has played a key role its use has often not been thoroughly described or explained so that its use as a precedent was muted.
Informing Drug Development
The Rare Disease Natural History guidance endorses use of information from these studies as primarily helpful informing the design and conduct of adequate and well-controlled clinical trials of investigational drugs that can support review and eventual approval decisions. This includes identifying the appropriate patient population to study. The guidance notes that a natural history study may uncover important, detectable physiologic changes that are important predictors of disease progression or are clinically important in their own right. In addition, a natural history study can be useful in understanding patient subgroups and identifying which may benefit from a particular clinical trial. These things together can inform decisions on inclusions/exclusion criteria, the stage of the disease to treat, the duration of the trial, the frequency of data collection, and endpoints.
Specifically related to endpoints, FDA endorses natural history studies as a way to identify and develop two types: (1) clinical outcome assessments (COAs) and (2) biomarkers. COAs are measures of how a patient feels, functions or survives. The guidance states that a natural history study can help evaluate the ability of a new or existing COA to detect change in a particular disease, including in the pattern of the progression of the disease or its symptoms. FDA notes that natural history studies can be important vehicles for testing COAs to establish their performance and reproducibility for use in clinical trials.
Meanwhile, biomarkers are objectively measured indicators of biological processes, pathologic processes, or biological responses to therapeutic intervention, such as physiologic measurements, blood tests, and imaging. The guidance states that a natural history study can help identify or develop biomarkers that can be useful in guiding patient selection and dose selection in drug development programs and can also be predictive of treatment response. Evidence of a biomarker being predictive of treatment response is important to establishing the potential surrogacy of that biomarker, such as for use as an accelerated approval endpoint.
Use of Natural History Data as an External Control
The Rare Disease Natural History guidance also explores the use of natural history study data to serve as an adequate control group for a clinical trial to support marketing approval. While FDA previously explicitly recognized the use of historical controls (as described in its guidance on control groups here), and has approved drugs based on studies using historical controls, this guidance provides additional insights into when historical controls are most appropriate in the rare disease context.
Considerations in Selecting or Designing Historical Controls
The Rare Disease Natural History Guidance provides considerations for deciding whether to utilize such a control, given its inherent limitations (e.g., inability to control for certain biases), as well as ways to maximize the utility of these controls when planning for their use. FDA provides the following considerations:
- The historical control needs to be very similar to the treated group in all aspects, including disease severity, duration of illness, prior treatments, and other key prognostic factors that affect disease outcomes. FDA notes that patient level data can help support this comparison between treatment groups. In our experience, patient level data also allows you to appropriately match the historical control to the trial population on these important prognostic variables.
- Concerns of selection bias can be reduced if natural history studies are similar to the clinical trial in the following ways:
- Assessment/measurement of critical patient disease characteristics;
- Aspects of standard of care of the patient population;
- Data collection intervals and quality consistency; and
- Well-defined and reliable COAs.
Interpretability of External Controls
The Rare Disease Natural History guidance also sets forth scenarios when natural history controls are most interpretable. These include when the treatment effect:
- Is large in comparison to potential biases and the known variability in progression;
- Is not affected by patient or investigator motivation or choice of subjects for treatment;
- Can be objectively measured;
- Is measured in a manager that reasonably manages and minimizes bias;
- Has a strong temporal association with administration of the investigational drug; and
- Is consistent with expected pharmacological activity based on the target and perhaps shown in animal models.
While these factors are not unique to the rare disease setting, it underlines the importance of well-defined, carefully documented natural history study protocols that delineate who should be included, the information to be collected, how it is to be collected, the schedule for the data collections (if prospective), and the plan for analysis.
Using Historical Controls to Supplement Concurrent Control Arms
In one noteworthy expansion from previous FDA guidance, the Rare Disease Natural History guidance takes a more wholistic approach (or as the guidance states “a hybrid approach), that endorses the use of external control data to add to a concurrent randomized control arm in a clinical trial. While not explicitly stated in the guidance, the co-authors of this post view this as an opening to use the historical control to expand an existing placebo control arm to increase its size and, therefore, increase its ability for the trial’s ability to detect a between-group difference when testing its hypotheses.
We propose that sponsors of clinical trials plan their studies to include a small placebo-control arm – one that is minimally sized in order to power the maximum possible treatment effect expected, therefore so that it is still ethical to include such a placebo control. The sponsor would also collect natural history data in a way that minimizes bias and other weaknesses of historical controls as discussed as FDA lays out in its guidance (and as we outlined above). Upon completion of the trial, the sponsor would then compare the results of the placebo cohort with that of the natural history cohort and, if sufficiently similar, would use the natural history cohort to “add” to the placebo cohort.
Combining the two cohorts would be permissible if predefined correlations or other assessments of similarity are achieved, providing comfort that that there is less bias in the natural history cohort from known or unknown factors that is not present in the randomized placebo arm. The average numeric baseline and post-treatment values in the natural history cohort need not be the same as the placebo control cohort, since different patient populations will be expected to be different at baseline, but as long as the magnitude of the change from baseline is similar between the two cohorts, then the two can be deemed similar. Essentially anchoring the natural history data with the placebo arm data will give regulators an opportunity to assess and, in turn, feel more comfortable with the historical control, which may facilitate greater acceptance of historically-controlled trials.
Practical Considerations for Natural History Study Design
The Rare Disease Natural History guidance also includes discussions of different types of natural history studies (e.g., retrospective vs. prospective, cross-sectional vs. longitudinal), as well as an overview of natural history study design elements, which will help orient those not familiar with study design more generally. Of note, FDA recommends:
- Engage with patients and patient advocates from the early planning stage (to consider data to be collected, need for potential adjustments to an ongoing study, and potential uses of information), and maintain their involvement (for patient identification and recruitment, ongoing community engagement, identifying burdens that may be resulting in dropouts).
- Initiate natural history studies even before an investigational product has been identified to allow for collection of data over a longer duration and with a larger patient population.
- Consider tradeoffs between less convenient, centralized locations that might allow for better standardization of data collection and more convenient, local locations that may be less standardized. Consider also utilizing remote data collection to improve patient convenience.
Ultimately, the Rare Disease Natural History guidance builds incrementally on previous FDA guidance on selection of control groups and on rare disease drug development. Now patient groups, academic researchers, and drug developers have a single document that can help identify opportunities and considerations when planning a natural history study. While the majority of the principles in the guidance are not unique to rare diseases, FDA’s issuance of this document within this therapeutic context emphasizes the increased importance these studies play in orphan product development. This is also accentuated by the guidance’s parting words, calling for interactions with FDA on design and use of natural history studies, whether or not in the context of a particular development program.