Cerus Red Cell Results Show What a Non-Inferiority Trial Does and Does Not Prove

When Cerus Corporation reported on November 2 that its INTERCEPT red blood cells met the main goal of the Phase 3 RedeS trial, the key phrase was "non-inferiority." The term has a specific meaning in clinical research, and it is narrower than it may sound. A 2016 FDA guidance on non-inferiority trials lays out how regulators think about these studies.

What non-inferiority means

Many trials try to show that a new treatment beats a placebo or another treatment. A non-inferiority trial has a different aim. It compares the new product with an active comparator and asks whether the new product is worse by less than an agreed amount, called the non-inferiority margin. That margin is specified in advance.

FDA's guidance is explicit that this is not a test of sameness. It says the intent of such a trial "is not to show that the new drug is equivalent, but rather that it is not materially worse than the control."

The result is usually read from a confidence interval, the range of differences consistent with the data. If the whole interval stays on the acceptable side of the margin, non-inferiority is established. The guidance also describes a case in which a product can meet the standard while the data suggest it is somewhat less effective than the control: the product is "actually inferior" to the control "even while meeting the NI standard."

How that applies to RedeS

In its release, Cerus said the trial measured adjusted hemoglobin increment, which it called a commonly used measure of red blood cell transfusion effectiveness. The company-reported figures were 1.996 g/dL in the INTERCEPT arm and 2.050 g/dL in the control arm, an estimated difference of -0.054 g/dL with a 95% confidence interval of -0.210 to 0.103.

The prespecified margin was -0.32 g/dL. Because the lowest value in the interval, -0.210, sat above that margin, Cerus said non-inferiority was demonstrated. The interval also includes zero, so these figures do not show a difference in either direction.

The results are topline and have not been published in a journal. Cerus said the FDA has not yet reviewed the final clinical study report and that its assessment may differ from the company's.

Why the design is used, and its limits

The guidance says the usual reason for this design is ethical: it is chosen when it would not be ethical to use a placebo because an effective treatment already exists. The trade-off is that there is no placebo group inside the study. As the guidance puts it, "the effect of the active control is not measured in the study but must be assumed."

That is why FDA treats these trials as harder to interpret than superiority trials. A successful superiority trial, the guidance says, largely speaks for itself. A non-inferiority result depends on knowing that the comparator had its expected effect, and that knowledge relies heavily on evidence from outside the trial. Choosing the margin is therefore a central task, and the guidance says the determination is often difficult.

The document also warns against reading too much into the result. Its methods, it says, "are intended to show that a new treatment that demonstrates non-inferiority is effective, not that it is as effective as the active comparator."

One caveat applies to this example. The guidance, issued by FDA's drug and biologics centers, addresses drugs and biologics and does not cover the use of these designs to evaluate safety. Cerus has described a planned premarket approval application for its red cell system, so the guidance explains the concept rather than the specific standard that will apply to it.