On paper, biosimilars have great potential to cut development costs, expand patient access to biologics, and improve affordability. The practical reality is that these goals are difficult to achieve.
Technology isn’t the biggest problem, say experts. Intellectual property challenges, added costs to characterize reference products, and uncertainty around what constitutes acceptable similarity are the stumbling blocks. That said, the biopharmaceutical industry is pushing forward aggressively to overcome the difficulties, and it’s making substantive progress.
Quality by design (QbD) approaches can reduce biosimilar development risk. Single-use technologies can cut costs. Modern bioprocessing methods produce higher titers than when originator products first reached the market. The ability to characterize large complex biologics, such as monocolonal antibodies (mAb), has improved. And the recent draft guidance from the FDA is further clarifying the regulatory landscape.
“We are all happy to have guidance documents,” says Joerg Windisch, Ph.D., head of global technical development for Sandoz Biopharmaceuticals, a player in biosimilar and innovator biologics. “They are very science-based and give FDA flexibility. We like many things about them.”
For example, the draft guidance allows global development, sparing biosimilar developers the need to conduct two clinical studies, one against the European product and one against the U.S. reference product. “We were very worried about that,” Dr. Windisch admits.
One way to help reduce development uncertainty is to use QbD concepts. “In this context, you design the manufacturing process for the biosimilar to produce a molecule that is very much the same as the original molecule,” notes Dr. Windisch.
The first step, of course, is to define the target by characterizing the reference product. This is an expensive, time-consuming process that involves buying multiple batches of the reference product and profiling the critical quality attributes (CQAs) with an arsenal of modern analytical tools.
A nagging issue is the range of variability in CQAs common to originator products including changes in the variation range over time. The next step is designing a process to produce a biosimilar that will inevitably use a different cell line, more advanced technology, and more sensitive analytics.
Sandoz employs a systematic approach, using design-of-experiment (DOE) methodologies and hundreds of experiments to characterize products and processes. Various critical attributes—basic and acidic variance, structural variation, PK/PD activities, immunogenicity, functionality, etc.—are measured and used in building mathematical models to predict process performance and product quality.
Sandoz’ work with rituximab is a good example, notes Dr. Windisch. Rituximab works largely by inducing ADCC (antibody-dependent cell-mediated cytotoxicity), which in turn is controlled by two different glycosylations on rituximab. Sandoz developed a process model that accurately predicts how much of either glycosylation occurs in response to even small changes in the process.
“We have a mechanistic model in that we have an algorithm of how different process parameters (pH, temp, addition of sugar precursors, etc.) systematically influence both of these glycosylation parameters, and we can forecast if we run a process a certain way what levels of ADCC response we will get,” he says.
Processing technologies have advanced such that “you really can’t go back and recreate a cell line and expression system that would be the same as the innovator company,” says Morrey Atkinson, Ph.D., CSO and vp of R&D and drug manufacturing for Cook Pharmica.
On the upside, modern processes have higher yields. One downside is that the biosimilar’s purity profile can be quite different from the original’s purity profile. “That presents additional challenges in terms of understanding the impacts on safety and efficacy during your fairly abbreviated preclinical and clinical development. When innovator compounds came to market, there was no specification or no standard that they had to compare to,” says Dr. Atkinson.
“We know more now but we don’t yet have the ability to dial in the characteristics we want. It’s why the agencies talk about the totality of evidence when evaluating submissions.”
The challenge, explains Dr. Atkinson, is to screen clones, cell lines, and cell culture conditions that would match the reference product’s biological activity and then develop purification processes “where your acidic and basic variance and polymer levels, all of those things, are within some predefined criteria the innovator compound has established.”
Regulators struggle with the same issue. “It seems to me that EMA is trying to be more formulaic and FDA is trying to be more case-by-case. That doesn’t mean your outcome would be much different in either case. FDA is saying it wants to drive the discussion and be involved in design of the programs you develop. EMA is not doing that; it is saying these are our guidelines per class of molecule, meet these guidelines and then bring forward your application,” says Dr. Atkinson.
A surprising point in FDA’s guidance “is you don’t need to match the closure or delivery system of the current marketed product. One would think you should be close if you thought that had any effect on PK/PD. It was interesting in the FAQs they specifically say you do not need to even match the dosage form of the market product. For product people, that raises eyebrows.”