Current Regulatory Landscape

Despite the growing attention given to the potential use of data from observational studies in evidence-based medicine, the regulatory requirement is still an evolving concept. Some of the important guidelines relating to the design, analysis, and reporting of observational studies have been issued by professional societies and other stakeholders. Notable examples include, the recommendations of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) (Berger et al. 2009); the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement (von Elm et al. 2008); and other resources for evaluating nonrandomized studies of comparative effectiveness (Deeks et al. 2003).

In the US, following the promulgation of the 21st Century Cures Act of 2016, the FDA has issued a framework for the use of RWE in regulatory decision-making (FDA 2018b). Key aspects of the framework include developing guidelines relating to: a) Whether the RWD are fit for use; b) Whether the trial or study design used to generate RWE can provide adequate scientific evidence to answer or help answer the regulatory question; and c) Whether the study conduct meets FDA regulatory requirements (e.g., for study monitoring and data collection).

With respect to effectiveness objectives, regulators are reluctant to draw a causal inference when treatment assignment is due to physician judgment, rather than randomly. FDA has stated that this must be addressed to support the acceptability of observational studies for effectiveness decisions (FDA 2018b, 2018c). There are examples of concordance between randomized trials and observational studies reaching similar conclusions about treatment effect (Anglemyer et al. 2014; Benson and Hartz 2000); however, there are also examples of discordant results (Cooper et al. 2014; Guadino et al. 2018; Hemkens et al. 2016). There have been recent efforts to use rigorous design and statistical methods to replicate randomized trial results with observational studies and to develop general rules to strengthen the validity of results in observational study designs (Franklin and Schneeweiss 2017). However, because of the discordant results given above and the uncertainty of the validity of observational results, it is unlikely that regulators will rely on results from observational studies for the purpose of effectiveness, except in certain special situations.

One important area of interest relates to enhancing non-randomized, single-arm trials through the use of external controls. Although the external control arm could use data from past RCTs, suitably chosen RWD might also be used to construct external controls. However, as highlighted elsewhere in this monograph, external controls have their own limitations, including lack of comparability of patient populations, lack of standardized diagnostic criteria, dissimilarity of outcome measures, and variability in follow-up procedures. The FDA Framework is anticipated to provide further guidance on the use of RWD to generate external control arms, complementing the ICH E10 guideline (ICH 2000).

Earlier, the US FDA issued a guidance document pertaining to the use of RWE to support regulatory decision-making for medical devices (FDA 2001). The document addresses important issues that arise in the evaluation of real-world data, including methodological rigor and data quality. More recently, a related guidance was issued, including recommendations for evaluating data sources used in pharmacoepidemiologic safety studies (FDA 2013).

There is a growing interest in pragmatic clinical trials, which involve randomization, and are typically integrated into routine clinical care. The study protocols for such trials specify minimal inclusion and exclusion criteria, and no treatment requirement other than the randomized assignment to one of the groups. Such trials may use EHRs or claims data to capture primary and secondary endpoints (see, e.g., Hernandez et al. 2015) or may be registry-based (e.g., Frobert et al. 2013).

Incidentally, there are several examples of RWE use in regulatory settings, including label expansion for new indications, fulfilling postapproval commitments, and even initial approval based upon external controls, especially in areas of high unmet medical need (Baumfeld Andre et al. 2019). Some use cases relate to label expansion based on EHRs, or postmarketing reports of claims databases. A recent example is the approval of palbociclib

(Ibrance) for the treatment of male breast cancer, expanding the earlier indication for the treatment of HR+, HER2- advanced/metastatic breast cancer in females (FDA 2019a). The approval was based on post-marketing reports and EHRs as part of the totality of evidence. Real-world data from EHRs showed encouraging signals of response rates with Ibrance, a CDK4/ 6 inhibitor in combination with an aromatase inhibitor or fulvestrant in the male patient population. The data also suggest that the safety of Ibrance in male patients was consistent with the tolerability observed in female patients who received palbociclib.

An example of label expansion using a pragmatic study concerns paliperidone palmitate, originally indicated for the treatment of schizophrenia in adults and treatment of schizoaffective disorder in adults as monotherapy and as an adjunct to mood stabilizers or antidepressants (Alphs et al. 2016).

Approvals have also been granted based on historical controls. An example is the accelerated approval of eteplirsen for Duchenne muscular dystrophy. The approval of eteplirsen used data on a historic control arm from a registry database (Mendell et al. 2016). In another case, label expansion was granted for blinatumomab based on the results of a single-arm trial supported by RWE to include indication for patients with minimal residual disease in which cancer cells are present at a low level that cannot be detected microscopically (Gokbuget et al. 2018).

Concluding Remarks

In this chapter, we considered some of the issues associated with the use of RWE in regulatory settings, and highlighted steps that need to be taken to maximize the evidentiary value of such data in drug development. Although the traditional RCT paradigm is the default approach for regulatory decision-making, there are considerable opportunities for RWE in label expansion and even initial approvals. With the growing cost of conducting RCTs, and the infeasibility of generating the requisite evidence for rare diseases, observational studies and RWD have garnered increased attention to support regulatory decision-making.

Historically, RWE from observational studies has been routinely accepted for satisfying postapproval safety commitments. With the development of guidelines and best practices, and gradual evolution of regulatory opinions, examples now abound illustrating regulatory decision-making based on RWE. Hybrid designs that incorporate randomization and minimal protocol requirements can be used to generate regulatory-grade evidence, especially in areas of unmet medical needs. In other cases, external controls involving RWD can be used to buttress single-arm trials.

Despite the various examples of approvals using RWE, the regulatory landscape pertaining to such data is by and large an evolving process. The onus is first on the sponsor to assess the appropriateness of the use of RWD from an observational database for the sponsors specific research purpose and that the results will have scientific rigor. There are myriad medical, statistical, regulatory issues to consider, as well as the nature of the observational database itself. An observational study should never be conducted simply because it may be more expedient than a clinical trial. Therefore, it is essential for sponsors to engage regulatory authorities early and obtain alignment of expectations. To facilitate the discussion with regulators and maximize the possibility of positive outcome, sponsors should apply best practices for study design, methodological rigor, and data quality. Special attention should also be paid to regulatory requirements for record retention, auditing, patient privacy, clinical endpoint validity, and reporting of study results. Sponsors should also put in place standard operating procedures to ensure transparency, including prespecification of protocols and analysis methods, data-quality assurance, and registration of studies and study results.