The Challenge of Defining the Higher Order Structure of Biopharmaceutical Products

Characterizing the higher order structure (HOS) of protein drugs increases manufacturers' understanding of stability and batch-to-batch variability, and may make it possible to link variants or aggregates to safety and efficacy. Yet at the January 24 CMC Strategy Forum in Washington, DC, regulators expressed concern that methods to characterize the three-dimensional structure of proteins are not routinely applied to biotechnology products.

"We receive a lot of information in filings about primary sequence but not much about higher-order structure," said Emily Shacter, PhD, chief of the Laboratory of Biochemistry in the Division of Therapeutic Proteins in the US FDA's Office of Biotechnology Products. As a result, she said, it can be difficult to know if important information is missing. "Biotech products have good track record for safety, but we should be seeing more characterization of higher-order structure," she said.

Brigitte Brake, PhD, head of the Pharmaceutical Biotechnology Unit at BfArM, Germany's regulatory agency, said the European regulatory experience has been similar. "The main difference is that we get fairly detailed data on glycosylation mainly because we have asked for it for a long time," she said. It was only when they began reviewing biosimilars applications that they began to see more information about higher order structure. "When companies compared their molecules to molecules of other companies, we got a lot more information that we did not have before," she said. "It was astonishing."

Method Limitations
Companies will file more information on higher order structure when better methods become available, said Rohin Mhatre, PhD, vice president of technical development at Biogen Idec. "We saw that with primary structure. Ten or 15 years ago, those techniques were not that sophisticated. Now they are and we use them routinely," he said. "We may be entering into a new phase in terms of HOS, but we are not quite there yet."

Speakers presented information on a variety of advanced analytical methods for evaluating HOS, including nuclear magnetic resonance (NMR), microcalorimetry, advanced fluorescence methods, hydrogen-deuterium exchange with mass spectrometry (H/DX–MS), and single-molecule analytical characterization. All have significant capabilities, yet important limitations remain, such as the time required to characterize a protein, decreasing day-to-day drift, sensitivity to slight changes in sample concentrations, and cumbersome data analysis. More importantly, none offer robust detection of very small quantities of change, such as 1–2% unfolded isoforms, which may affect drug safety.

Given the limitations of newer techniques for characterizing HOS, products should not fail to gain approval when they are not used, said Keith Webber, PhD, deputy director of the Office of Pharmaceutical Science, CDER, at the US FDA. Yet he reinforced the message that improvement is needed. "The purpose of this meeting and the demonstrations of the up-and-coming technologies is for us to gain a better understanding of the technologies, so products and the impact of changes can be better understood," he said. "We need a future that holds better understanding."

Webber noted that as Quality by Design is implemented, more companies will probably define the HOS of their products in their filings and seek to make clearer correlations between HOS and function. Correspondingly, regulatory expectations are likely to increase in this regard.

Using HOS in Comparability Studies
The International Conference on Harmonization's Q5E guideline on comparability encourages companies to assess HOS changes in proteins before and after manufacturing process changes. A number of challenges remain, however, in using HOS to assess comparability.

It is not always known whether differences seen in analytical test results are actual changes in the product or result from variability in samples, assays, or processes. Also, HOS techniques may indicate that a product's thermal stability has decreased or that there are differences in the position of amino acids on the surface, but cannot identify the underlying structural change. Furthermore, most techniques produce data from averages, so differences might not be detected if they are balanced out or diluted by multiple or larger signals.

In addition, some higher order structures may not be preserved in archived samples. "Some higher order structures are altered when you freeze and thaw, or they change with time," said Tom Patapoff, a senior scientist at Genentech. "That's the thing that scares me-you may be introducing artifacts and following the wrong path."

The biggest concern, however, is always the difficulty in assessing whether structural changes will affect patients.

"The only way we can know is if we track this during clinical development," said FDA's Shacter. For that reason, she said, the agency pushes companies to characterize their products as early as possible and evaluate the product for potential changes, so they can see if a change in a manufacturing process affected the molecule-and whether that changed outcomes in the clinic.

"We don't want to be overbearing, but we need some data to make a science-based decision," she said. "We don't want to be taken by surprise because we didn't look and didn't require manufacturers to look."