ifty years of investment in biomedical research have revolutionized our understanding of human disease. No longer are many illnesses baffling tragedies that simply defy explanation. Particularly for rare diseases, we now can learn the molecular basis of the pathology and devise potential treatments and cures. Academic laboratories have identified the genetic and biochemical basis of hundreds of rare diseases and proposed treatments, bringing hope to millions of sufferers and their families.

But these hopes are often shattered. Imagine the anguish of a family when work is halted on a potential treatment that might save a loved one, often a child with a serious or fatal illness. Many times, that family will have contributed to the scientific development of the treatment. In the past few years, more than 1000 rare disease development programs have been put on the shelf, many before they even reached human testing. Why? Because even motivated investors are reluctant to put their money into a program with no clear path forward and a very low probability of success.

How can this be? In the US, drugs must undergo human testing and meet FDA standards for safety and efficacy before marketing. The standard for efficacy was established by Congress in 1962 and calls for “substantial evidence from adequate and well-controlled investigations, including clinical investigations” that would convince experts that the drug works. Since then, the FDA established, and the courts have supported, that these studies should include two randomized controlled clinical trials, each with a “p value” less than 0.05. For the first time, these requirements put in place a strong evidentiary foundation around the use of medicines and were rightly considered a scientific triumph, leading to most of the medicines widely used today.

Over time, as biomedical science advanced, variations on these requirements were developed. FDA utilized “accelerated approval” initially during the HIV epidemic and had been accepting “a single randomized trial plus confirmatory evidence” in cases of life-threatening illnesses. These variations were subsequently enacted into law. Accelerated approval was utilized for many of the new targeted cancer therapies that have been approved over the last two decades, generally believed to be responsible for the concurrent improvements in cancer survival in the US. FDA also published a regulation enabling approval of drugs for certain catastrophes (where human testing could not occur) based on animal model studies.

FDA originally developed these several variations to address new medical needs that could not be met by existing approaches. For example, once a large, randomized trial definitively shows a survival advantage with a new treatment, how can a second trial be done ethically when it would require people to be randomized to a placebo? New scientific knowledge about disease enabled the changes, since scientific gains since 1962 provided additional evidence to bolster the single trial results with a similar level of confidence.

Today we face another new and critical medical need, this time in rare disease drug development. Specifically, the construct of the randomized clinical trial (RCT) and a “p value” less than 0.05 is not fit for purpose in evaluating rare disease treatments where only small numbers of patients are available for study. Using such a study design when statistical power cannot be bolstered by enrolling additional patients creates a tremendous hurdle for such development programs. Usually only an extremely large effect can be detected. This means improvements routinely accepted as proof of effectiveness in common diseases will not show up in a small trial, and the trial will “fail.” Thus, in many cases, children with fatal disease will be denied treatments that could mitigate their suffering, since only “cures” will meet the hurdle. We must again enable alternative approaches, based on new scientific knowledge, that are feasible and will deliver adequate confidence that the therapies work.

In many cases, it is possible to devise non-RCT clinical studies that provide convincing evidence of efficacy, particularly when supported by additional lines of scientific evidence. FDA has on occasion accepted the results of such studies, based on the medical need and because the studies provided strong evidence. However, this acceptance is not predictable, and developers of rare disease programs are frequently told they must conduct placebo-controlled RCTs. As these are often infeasible or unlikely to succeed, this demand leads to prolonged delays or complete abandonment of promising programs. The lack of consistency in requirements has also caused “cognitive dissonance” among the FDA staff, since the RCT is generally considered the standard, but exceptions occur.

It is time to accept another variation in the efficacy standard. At the time of the 1962 amendments, little was known about rare diseases except for their description in case series. Today, their study is a central part of advancing our understanding of human biology and disease. We need to enable creativity and application of cutting-edge science, whether biological or statistical, to the performance evaluation of drugs for rare diseases, because the current approach is failing. Congress should again codify what has become intermittent practice and clarify that, in certain circumstances, evidence from non-RCT clinical studies, combined with additional scientific data, can provide substantial evidence that the drug works as intended.

What might these nontraditional RCTs be? One promising approach is using the self-controlled design, where each patient is followed for an appropriate period to identify their principal symptoms and disease trajectory. Patients could at this point be assigned to “open placebo” to help mitigate subjective placebo-driven effects. (Open placebo has been shown in multiple studies to be effective. Whether it is equally effective to “masked placebo” is not known for any given disorder, but certainly the impact of the placebo effect is partly mitigated by its use.) For each patient, the important symptom(s) and how it is measured is prospectively designated as the endpoint for that person, and the minimal important effect size identified. At that point, the individual is put on the active drug, and a before-and-after comparison is done, at an interval determined by the previously observed rate of progression. Documented improvement in one such patient would not be accepted by experts as clear evidence of benefit, but multiple such instances (how many would be contingent on the trajectory of the specific disease) would be very convincing to experts.

Another development scenario that would not have been feasible decades ago relies upon the progress of science. Many rare diseases result from enzyme deficiencies. Over the years, scientists have identified the genetic defects, the missing or malfunctioning enzyme, and the enzyme’s substrate, and have “connected the dots” on the resulting pathogenesis. In most cases, showing that the enzyme is active at the needed site(s), along with evidence of target organ improvement, should be considered adequate evidence of effectiveness for enzyme replacement. Clinical exposure is of course necessary for dose finding, determining immunogenicity and safety over time. Animal models could be used, when possible, to bolster the case, but the long history of successful enzyme replacement when active enzyme can get to needed sites should be given substantial weight.

Additional potential designs incorporate Bayesian statistics. FDA has been reluctant to utilize Bayesian methods in drug trials; however, their use can provide multiple advantages for small trials. Using an adaptive design in an RCT can minimize exposure to placebo. Other Bayesian approaches can improve the ability to discern an effect when one exists or can shorten the time between doses/trial phases, of key importance when dealing with critically ill children. No one design will fit all scenarios, but it is clear that creativity and good science can result in designs more fit for purpose for these devastating diseases.

Everyone in the biomedical research and patient communities wants the best outcomes for patients. No one wants ineffective therapies that will just add to the anguish of patients and families. What is needed is not a lowering of standards; it is applying the right tool for the job. Let’s get on with it.