Amgen v. Sanofi: Critical Impact on the Value of Innovative Science in Antibody Discovery

The Supreme Courtis considering a patent dispute between pharmaceutical giants Amgen and Sanofi that will have broad implications on future discovery and development of monoclonal antibody drugs, with possible extension to other classes of biological treatments of disease (e.g., gene editing). The heart of the disagreement relates to whether disclosure of the discovery of functional monoclonal antibodies, binding to a specified region of a targeted substance (referred to below as the “target”), provide sufficient enablement to scientists wishing to develop and use the invention. The competing argument, which is in-line with Sanofi’s position, is that enablement requires definition of the amino acid sequence of the monoclonal antibody, and that antibody patents should not extend to members of a class or genus (e.g., based on functional ability to bind to a target epitope) beyond examples for which sequences have been specified within the patent. This argument would require listing of all antibody sequences capable of binding to the target, which could be potentially millions of sequences, to allow patent protection for the full scope of the genus. If this standard is implemented, it would be impractical or impossible to achieve, and it would mark the end of traditional, broad patents for antibody drugs. There are advantages and disadvantages to each side of this argument. This commentary provides some background on the discovery and development of “small molecule” drugs and antibody drugs, and highlights scientific (not legal!) justifications for genus coverage for antibodies based on consideration of the structural heterogeneity of antibody drugs and the absence of innovation in the discovery of “follow-on” antibody drugs. Advantages and disadvantages for the award of broad (genus) vs. narrow (sequence-specific) patent coverage are summarized, and a simple compromise solution is suggested.

Drug discovery and development is a costly, time-consuming process that is typically described as progressing through a series of steps that include the following: (a) target identification, (b) target validation, (c) the generation of “hits,” (d) hit-to-lead engineering, (e) lead optimization/candidate selection, and (f) preclinical and clinical evaluation of the drug candidate (i.e., development) (Fig. 1). Although the costs and risks are greatest during development, it may be argued that all innovation associated with the generation of new drug therapies occurs during the discovery phase, which is represented by steps a–e, above.

Fig. 1.

Overview of drug discovery and development. Key areas of innovation for discovery of small molecule drugs include the first development of hits (agents that are identified to bind to the target), hit-to-lead engineering (generation of functional small molecule drugs), and lead optimization (leading to the identification of a drug candidate with drug-like qualities (good oral bioavailability, etc.)). In the case of monoclonal antibody drug discovery, key areas of innovation are the identification and validation of targets, including the first demonstration that antibodies with binding to target epitopes deliver the desired function

The starting point for modern drug discovery and development is typically tracked to the work of Dr. Paul Ehrlich, who introduced “receptor theory” in the beginning of the twentieth century. Drugs act primarily by binding to and modulating the behavior of natural substances, such as signaling proteins, enzymes, transcription factors, and nucleic acids. Binding to specific sites is key to drug function, famously conveyed by Ehrlich as “corpora non agunt nisi fixate,” which has been translated to “agents only work when they are bound” (1). In the case of small molecule drugs, binding is only one element of many that determines their therapeutic utility. For example, for the vast majority of small molecule drugs, target-binding properties (e.g., affinity) do not appreciably influence their pharmacokinetics. Rather, the combined features of all elements of the chemical composition of a small molecule influence its solubility, lipophilicity, polarity, liability to biotransformation, membrane transport and, thus, pharmacokinetics. As such, much investment and innovation are often required to discover new small molecule chemicals with desirable binding to the pharmacological target and with desirable physicochemical properties for acceptable drug-like attributes (i.e., good oral absorption, sufficient distribution to enable engagement of the target, sufficient metabolic stability to allow a long biological half-life, etc.). In part due to the importance of all aspects of the chemical structure in determining pharmacokinetics, pharmacodynamics, and toxicodynamics, the steps of greatest importance in small molecule drug discovery are steps c–e: the identification of hits and the engineering steps leading to the selection of the drug candidate (hit-to-lead engineering and lead optimization).

In contrast to small molecules, it was well recognized that the earliest antibody therapies could not be defined as a single chemical. Early antibody preparations were typically developed by immunization of animals and by the collection and purification of immunoglobulin from blood. These preparations contain a wide distribution of active molecules, which may differ in terms of their primary amino acid sequences and their key structural features (referred to as the antibody isotype and subclass). Due to this molecular heterogeneity, patents have defined the antibody drug by citing the structural class of the antibody (e.g., immunoglobulin G) and by detailing functional binding to the therapeutic target.

Advances in the fields of immunology, biology, and molecular biology, including the work of Köhler and Milstein in innovating strategies for isolating immortalized, antibody-secreting cells (hybridomas) (2), enabled the production of monoclonal antibodies (mAb) from cloned cells in culture, where the drug product could be defined more precisely. Although antibody development is time consuming, often requiring several months to advance through the generation of hits to select a drug candidate, the process is considered to be trivial to those skilled in the art. As a consequence, the true innovation in the development of antibody drugs lies in target identification and target validation (steps a and b, above), including validation of therapeutic utility derived from antibodies capable of the desired binding functionality. The first demonstration of therapeutic utility of an antibody with a desired binding profile represents the key innovation in antibody discovery and development; the generation of additional, follow-on antibodies that bind to the same target epitope is neither difficult nor innovative.

Through the late 1970s until present, the favored structure of mAb drugs has been immunoglobulin G (IgG). The domains of an IgG antibody that bind to the target, the complementarity determining regions (CDR), relate to a very small fraction of the amino acid residues of the mAb (~ 5%), with the remaining residues highly conserved among all IgGs. As such, in some ways, it is possible to consider all antibody drugs of a given isotype (e.g., IgG1) to be the “same,”¹ with exception for the CDRs and the binding function of the antibody. In other words, the unique elements of the mAb may be considered to be limited to their binding domains, and the other elements of the mAb, which contribute largely to their solubility, stability, and pharmacokinetics are nearly constant from mAb to mAb. This is not to say that antibody binding is not an important determinant of antibody pharmacokinetics—it often is. But, due to the highly shared structure of IgG molecules, the binding functionality of a new antibody (e.g., binding affinity and the site of antibody binding on the target) is key in determining both cellular pharmacodynamics and systemic and cellular pharmacokinetics. Indeed, physiologically based pharmacokinetic models, developed by my group and by others, have demonstrated that the complex, non-linear pharmacokinetics of mAb may be predicted successfully based on antibody isotype (e.g., IgG) and by mAb-target binding (3, 4).

Simple consideration of the relationships between the global structure, target binding, and pharmacokinetics, for small molecule drugs and for mAb, helps to show why it is necessary to define the entire chemical structure of a small molecule drug (e.g., as pharmacokinetics are dependent on all elements of the chemical structure of the small molecule) and why it is possible to define appropriately an antibody drug based on its general structure (e.g., as an IgG mAb) and based on its target binding (due to the key role of target binding functionality as a determinant of pharmacokinetics and pharmacodynamics).

Much more importantly, mAb drugs cannot be described as a simple chemical composition. While monoclonal antibody technology enables the production of agents with a defined primary amino acid sequence, mAb drugs are not homogenous preparations (5). Monoclonal antibodies are not chemically synthesized, but are biogenically produced by cellular expression systems, and the mAb protein is subject to a multitude of mechanisms leading to post-translational modification. Modern monoclonal antibody drug products are a collection of thousands of unique antibody molecules, where the molecular distribution may be influenced by many factors, including the methods of production and purification. For this reason, regulatory agencies (e.g., the US Food and Drug Association) require complex testing of biosimilarity beyond the demonstration of primary amino acid sequence, with a range of functional assessments including safety and efficacy testing in human subjects, prior to the approval of “generic” monoclonal antibody drugs (6). If the standard for patenting antibody drugs shifts to sequence and away from binding and function, what will be next? Will future infringers seek to more narrowly limit patents for antibody drugs by arguing for specification of primary sequence and glycosylation distributions? Or perhaps specification of sequence, distribution of glycosylation, and post-translational modifications of the N-terminal and C-terminal amino acids? Where would this end?

In part due to the recognition that the key innovation for antibody drugs relates to the first demonstration of utility of a target-binding antibody preparation for the desired therapeutic application, and in part due to the historical complexity, and current complexity, of antibody therapeutics, antibody patents have been defined by function, and have enabled protection for all antibodies that exhibit the given function (e.g., binding to a particular epitope on a target). This approach recognizes that even monoclonal antibodies are not a single, homogenous agent, but rather a family of agents with desired target binding attributes, and places appropriate value on the area of true innovation (i.e., demonstration of utility of a target-binding antibody for a given purpose).

If the argument in favor of narrow patent coverage is accepted (i.e., limited to primary amino acid sequences), this will reward the development of follow-on products, in most cases devoid of any innovation, and it will disincentivize investment in the truly innovative aspects of the discovery and development of new antibody drugs. True innovation in antibody discovery ties together existing knowledge and experimental work to identify a new target, validates the new target as being “drugable,” and demonstrates and teaches the antibody drug invention by reduction to practice.

Limiting the scope of the invention of a new antibody drug, not based on binding function, but based on primary amino acid sequence, allows others to perform the non-innovative, trivial steps of antibody development and claim the market from the innovator. A decision in favor of narrow patent coverage would be, in my opinion, a travesty, as it would: (i) disincentivize much innovation, including pursuit of mAb that bind to unique epitopes on the target (i.e., innovation outside of the genus), which may demonstrate superior safety and efficacy; (ii) disincentivize publication of information about the development and use of new antibody therapies for newly identified therapeutic targets, particularly by academic scientists and by small biotech companies, as this may merely benefit large companies who could race to the market with follow-on mAb; and (iii) disincentivize large companies from investing in the discovery of new targets; they could simply wait for others to innovate, develop follow-on agents, and then out-resource innovators to claim the market.

The legal arguments against broad patent coverage for antibody drugs focus on issues of enablement (i.e., questioning whether the description of methods for the generation/identification of antibodies with binding to a target epitope requires undue experimentation and testing to produce agents within the protected genus). Although the definition of undue experimentation is debatable, the majority of scientists within the antibody discovery field would likely agree that the development of mAb with specificity for a given epitope is time-consuming, yet not technically challenging, such that definition of the desired epitope is indeed enabling. Scientific and practical arguments suggest that broad patents may limit opportunities for innovation (i.e., within the genus), while also limiting potential for competition and patient access (7). Additionally, concerns have been raised that the issuance of broad patents for mAb places too much value on early discovery, and too little value on the very expensive and risky process of clinical development.

Possible Compromise Solution?

Regardless of the decision that will be made by the Supreme Court, Congress will have the opportunity to act and define special terms for the patenting of new therapeutic monoclonal antibodies (and, perhaps, other classes of drugs). One simple solution would be to define acceptable conditions for the award of genus coverage, and to mandate legally that holders of genus patents accept any proposed licensing agreement from any competitor in exchange for a reasonable, legally defined royalty fee (e.g., 10% of net sales of the licensed product). This arrangement would allow true innovators to claim value for their discoveries, and thus promote innovation by all participants within the antibody discovery field (i.e., ranging from academic labs to small biotechs, to large pharma companies). There would be motivation to innovate within and outside of identified sweet spots on antibody targets. Additionally, this arrangement would allow and facilitate competition, providing the opportunity for several options within an antibody genus for patients, while also placing appropriate value on the costly and risky steps of preclinical and clinical development. Will Congress seize this opportunity to protect and promote the interests of antibody innovators and developers and patients?