Newsletter Fall 2018 - Grant Williams, MD on Evidentiary Standards in the Accelerated Approval of Cancer Drugs

Let’s pretend I am an cancer drug sponsor with an immune-stimulating drug (Drug A) for the treatment of a refractory cancer. I plan to study Drug A in combination with an approved immunotherapy drug (Drug B), such as a PD1 inhibitor. I have recently seen many oncology drugs approved based on overall response rate (ORR) in a single-arm trial (SAT). I want the FDA to grant accelerated approval (AA) for my drug given in combination (Drug A + Drug B) based on ORR. What are the regulatory issues involved with this proposal? In the following discussion, I will shed light on these issues.


Accelerated approval in oncology

In the 1980s, the FDA usually required a survival benefit to support the approval of cancer drugs. In the early 1990s, the FDA stated that large effects on tumor endpoints may be viewed as a clinical benefit and could support drug approval. In 1992, the AA regulations were enacted, providing a path for drug approval based on a surrogate endpoint “reasonably likely to predict benefit.” The drug must also show a meaningful advantage over available therapy. Post-approval confirmatory studies were required to assess the clinical benefit of the drug. 

Advantage over available therapy

For the first AA requirement of a surrogate endpoint “reasonably likely to predict benefit” for oncology drugs, the FDA determined that a meaningful ORR with an acceptable response duration sufficed for many tumors. The evidentiary standard for demonstrating this effect will be discussed in a later section.

The second AA requirement was to show a meaningful advantage over available therapy. The usual regulatory approach of showing an advantage of one drug over another is to perform a randomized controlled trial (RCT). The FDA facilitated this process for oncology drugs by allowing the use of SATs for AA. The regulatory rationale for this practice was that for patients with refractory disease, in which no available therapy was effective for long, any meaningful benefit represented an advantage over available therapy. The paradigm for AA based on ORR in SAT in the refractory cancer setting was used many times in the ensuing years. The FDA reported that in the 25 years following the enactment of the AA regulations, 67 oncology drugs were approved based on ORR demonstrated in SATs (Beaver et al. 2018; Johnson et al. 2003).

With the advent of new cancer drugs yielding high ORRs, the FDA has also granted AA based on ORR in SATs in non-refractory disease settings. In such cases, advantages over available therapy was established based on cross-study comparisons of the ORRs of investigational drugs to the ORRs of available therapies. These studies were typically designed to demonstrate that the lower range of the 95% confidence interval of the investigational drug ORR excluded the point estimate of ORR of available therapy, which was often determined by a meta-analysis.

An example of this use of historical data as a basis for comparison is described in the FDA review of blinatumomab, which received AA in 2014 for the treatment of relapsed and refractory acute lymphoblastic leukemia (ALL). Historical ORR data were evaluated from a meta-analysis of clinical study results and by a model-based comparison of historical individual patient data. The FDA statistical review of this BLA provides an explanation of FDA’s position that the observed 30% complete response rate was better than what would be expected with available therapy.

There are other ways to show an advantage over available therapy besides demonstrating better efficacy. According to the FDA Guidance on Expedited Programs for Serious Conditions, examples include providing comparable efficacy with less toxicity and providing an important increase in compliance. It is important to note that neither regulations nor guidance describes a clear evidentiary standard for meeting the requirement to establish an advantage over available therapy; regulatory determination seems to involve regulatory and clinical judgement instead.

Substantial evidence of an effect

Did this new tradition of granting AA based on ORR replace the longstanding FDA efficacy standards for new drug approval? Though it is tempting to consider a SAT an uncontrolled study, the FDA believes otherwise. The FDA approves drugs based on the mandate of substantial evidence of efficacy from adequate and well-controlled clinical studies.

SATs are considered to be historically well-controlled studies with respect to the endpoint of ORR. The FDA and the oncology community have widely accepted that the historical ORR in patients with refractory cancer is less than 5%, if not 0%. Thus, if an investigational drug demonstrated a clinically meaningful ORR in a reasonably sized study (usually about 100 patients), this would represent substantial evidence of efficacy on ORR. The FDA considered duration of response a key supporting measure in demonstrating clinical meaning. I dwell on the term “substantial evidence of efficacy” because I believe many sponsors may be unaware of the special nature of ORR in a SAT that has allowed the FDA to rely on this evidence to support the approval of cancer monotherapy regimens.  

Several sources confirm that the FDA still believes that “substantial evidence of efficacy” is a relevant standard, even in the AA setting. One example took place during the Oncologic Drugs Advisory Committee (ODAC) discussion of apaziquone, a treatment for superficial bladder cancer. In this discussion, the FDA clearly noted that substantial evidence of efficacy is a requirement for AA. The apaziquone application proposed to pool the results of two large controlled studies and sought AA based on the combined analysis. The pivotal question for ODAC to support AA for apaziquone was, “Has substantial evidence of a treatment effect for apaziquone over placebo been demonstrated?” Another example can found in the current FDA template for NDA/BLA reviews. The executive summary review template has a section header entitled “1.2 Conclusions on the Substantial Evidence of Effectiveness.” This paragraph is used both for regular approvals and AAs (e.g., acalabrutinib approval in October 2017). Lastly, the FDA Guidance on Expedited Programs for Serious Conditions categorically states that the standard for efficacy under accelerated approval is “substantial evidence based on adequate and well-controlled clinical investigations.”

Substantial evidence in the drug combination setting

Let’s return to our example of the proposed trial design of Drug A + Drug B to support AA. I have frequently encountered sponsors who propose to use this SAT design to support AA of their investigational drug in combination with a standard therapy. In most cases, however, I believe this design is not likely to fulfill regulatory requirements for efficacy. When developing a drug given in combination with another approved active drug, we can no longer assume that the ORR observed in a SAT represents substantial evidence of an effect of either drug. We can no longer refer to a widely accepted historical rate of nearly zero as the comparator. If we seek approval of Drug A given in combination with Drug B, we need to show substantial evidence of the efficacy of Drug A. This would require comparing the ORR of the combination regimen to historical data on the ORR of Drug B. We would need to claim that these results represent substantial evidence of efficacy of Drug A demonstrated in an adequate and well-controlled study. I am not aware of any AA to date in which such ORR evidence has been accepted as the evidentiary basis for efficacy supporting AA. 

Two different evidentiary standards

The point of this short article is to convince you that a different standard of evidence is required for two of the key requirements for AA. What constitutes an advantage over available therapy is summarized in the FDA Guidance on Expedited Programs for Serious Conditions. Although there are many potential bases for claiming an advantage, neither regulation nor guidance provides clear standards for how such an improvement over available therapy must be demonstrated. On the other hand, the expedited therapy guidance clearly states that the standard for efficacy is the same for both accelerated and traditional approval: “substantial evidence of efficacy based on adequate and well-controlled studies.”


The FDA has used AA as a means to make promising cancer drugs available to patients. The FDA practice of granting AA based on ORR in SAT to date has been limited to drugs approved as monotherapy treatments. Currently, many new therapies are being studied as combinations of an investigational drug (Drug A) in combination with an approved drug (Drug B). This is particularly true for immunotherapies, in which the investigational drug may be an immunostimulant intended for use in combination with a PD1 inhibitor. Such immune-stimulating investigational drugs are not expected to be as effective as monotherapies. In these settings, sponsors may propose that promising ORRs observed from the drug combinations (Drug A + Drug B) should support AA.

The ORRs demonstrated in these SATs may be encouraging compared to ORR results reported for the standard drug partner alone, and the combination may appear to provide an advantage over available therapy based on cross-study comparisons of ORRs. However, it is important to consider and discuss with the FDA whether the trial design also meets FDA requirements in demonstrating substantial evidence of efficacy. If the efficacy of the new drug combination is extreme compared to the results from the monotherapy drug partner alone, then perhaps this approach is feasible. In any event, it may be prudent to understand the potentially precedent-setting nature of this approach when discussing this design with FDA.