Tony Hagen is senior managing editor for The Center for Biosimilars®.
Expecting real-world evidence to confirm safety and efficacy of biosimilars is asking too much, Uwe Gudat, a pharmacovigilance expert from Fresenius Kabi, said at the Terrapinn Festival of Biologics Basel 2020.
There is much effort these days to conduct real-world studies of biosimilars, but findings from these initiatives are not really suitable for confirming safety and efficacy trials used to gain biosimilar approval, said Uwe Gudat, head of clinical safety and pharmacovigilance for Fresenius Kabi, a Germany-based maker of oncology, diabetes, critical care, digestive, and medical products, including biosimilars.
Real-world evidence (RWE) is better suited to answer questions about practical use of products in the real world, Gudat explained at the Terrapinn Festival of Biologics Basel 2020 virtual conference.
RWE cannot do the job of detailed physicochemical characterization and functional assays, he said. “Do we really think that real-world evidence can reassure us more than well-controlled randomized clinical trials? I don't believe so.”
The value of a clinical trial is that it screens out random variables, or “noise,” such as patient comorbidities, that distort whatever “signal” the biosimilar is giving that it can successfully treat the target disease conditions, Gudat said. You can’t get the same results in the real world, because there the background noise is strong, he said. “For the most part, when we’re answering questions, the signal clarifies and the noise distracts.”
In the clinical trial setting, the efficacy of a drug candidate is being evaluated; and in the real world, the actual effectiveness of the drug under conditions of everyday medical practice is being judged.
The Car Analogy
The analytical comparison of a biosimilar candidate done in a controlled setting can be compared to development of a car and the systematic testing of its constituent parts, he said. In the real world, “we’re taking the car out for a spin.” But it’s then that many other factors come into play. Continuing the car analogy, he explained that road conditions, driver fitness, driving skills, and time of day are variables that have the potential to “confound your comparison,” and the variations from what you observed in a controlled setting could be extreme and not the fault of your product.
Even in a drug trial, it is unlikely that clinicians will get the same results every time. It is a fundamental of statistics that patient outcome samples will vary, and these must be pooled to form a more comprehensive picture of how the biosimilar will perform, Gudat said.
He noted that clinical studies of US- and EU-sourced samples of reference adalimumab (Humira) have demonstrated variations that could be explained by local conditions under which the drugs were administered—not by variations between the US and EU biosimilars. “We don’t believe that studies conducted by one sponsor vs another sponsor are fundamentally different and the products used are fundamentally different; much more, we think there are certain circumstances that the study introduces that drive the observation,” Gudat said. "We have to acknowledge that there’s randomness."
Genuine product characterization involves pooling multiple samples of patient outcomes to form a statistical curve. Most of the samples will cluster in the center of this curve, but inevitably there will be outliers. The confidence interval, or the probabilistic margins of drug performance, cuts away the outliers, but those outliers still reflect what could happen when patients use the drug being studied, he said. “This is important, because it helps us understand what kind of numbers we’ll be looking at and what they mean or don’t mean.”
What's Driving Real-World Differences?
The real-world differences observed between US- vs EU-sourced Humira could be explained by non-drug–related variables, such as improper drug administration or the geographic dispersal of study populations and their epidemiological history—ie, tuberculosis, Gudat said.
For those reasons, he said, RWE doesn’t really help evaluate for biosimilarity. “It compares the products embedded in a multitude of circumstances. You have to look at the whole package. I’m really illustrating why the signal-to-noise ratio in the real-world setting is actually not necessarily conducive to generating comparative evidence for a biosimilar. You’ve just got so much noise; and if you do see a difference, you’ve got no idea where it’s coming from.”
That doesn’t mean RWE is of no value. There are questions that RWE can answer and should be used to answer, he said.
RWE studies can provide important information along the lines of who has gained access to the biosimilar, when is it being used in real world practice, how it is administered, and in what settings the biosimilar has been embraced, Gudat said.
By following this line of inquiry, real-world studies can complement clinical efficacy studies, he said.
Gudat suggested a number of potential real-world studies that would accomplish this objective:
“These are other questions equally important or also important that [RWE] can help us to answer,” Gudat said.