Is an antibody defined solely by broad, functional terms in a patent, without specifying its amino acid sequence, eligible for SPC protection? Such patents are not unusual in the field of antibodies, where early patent filings often describe antibodies by their ability to bind a target (e.g., “An antibody that binds to X”). The EPO Guidelines recognise this approach, permitting antibodies to be defined functionally by the antigen they bind (G-II, 6.1.2; Guidelines for Examination in the EPO, April 2025 edition). Would an antibody defined in this manner be ‘protected by a basic patent’ under Article 3(a) of the SPC Regulation?
The CJEU’s decisions in Teva (C-121/17) and Royalty Pharma (C-650/17) offer valuable guidance in this context. These rulings establish a framework for assessing whether a product, defined functionally in the patent claims but not individually identified in the specification, is protected by the patent. According to this framework, a product is protected by a basic patent if it is ‘necessarily and specifically’ covered by the claims of the patent, for which the product must:
- necessarily come under the invention covered by that patent, based on a skilled person’s perspective and in the light of the description and drawings of the basic patent; and
- be specifically identifiable in light of all the information disclosed by that patent, on the basis of the prior art at the filing date or priority date of the patent concerned.
Even if these two cumulative conditions are met, a product is nevertheless not protected by a basic patent if it was developed after the filing date of the application for the basic patent following an independent inventive step.
The Cour de cassation (French Supreme Court) has been leading the charge in interpreting these criteria as applied to antibody SPC applications, with four decisions issued in the past three years (Case nos 21-13.664, 21-13.663, 22-18.374 and 23-20.000). In its latest ruling earlier this year, the Court upheld a decision stemming from the French patent office (INPI) to reject Dana-Farber Cancer Institute, Inc.’s (Dana‑Farber) SPC application for atezolizumab for alleged non-compliance with Article 3(a). The following sections will examine the facts underlying the decision, the Court’s reasoning, and the potential implications of the ruling.
Background
Atezolizumab is a therapeutic antibody that belongs to a class of anti-cancer therapies known as immune checkpoint inhibitors. Such immune checkpoint inhibitors target the programmed death-1 (PD-1) receptor or its ligand (PD-L1) and enhance immunity against several cancers. Atezolizumab binds to and inhibits PD-L1, and other examples of such inhibitors include avelumab and durvalumab. Atezolizumab is a ‘humanised’ antibody, in that it has sequence elements that are derived from non-human antibody sequences, while avelumab and durvalumab are ‘fully human’ antibodies i.e., their sequence is entirely derived from human antibody sequences.
Dana‑Farber applied for an SPC for atezolizumab based on its EP 1 210 424 (‘424) patent and a marketing authorisation that was issued for Roche’s Tecentriq®. Claims 17 and 27 of the ‘424 patent generally refer to an antibody that ‘selectively binds’ to a B7-4 (PD-L1) polypeptide, but the patent does not explicitly disclose the amino acid sequence of atezolizumab. The amino acid sequences of atezolizumab were disclosed in a later patent stemming from Genentech as applicant, filed around nine years after the ‘424 patent.
Atezolizumab decision
INPI rejected Dana-Farber’s application, stating that atezolizumab was not specifically identifiable based on the disclosure of the ‘424 patent. INPI’s decision was upheld by the Paris Court of Appeal. Dana‑Farber appealed to the French Supreme Court, which confirmed the Paris Court of Appeal’s judgment and dismissed the appeal.
The Paris Court of Appeal noted that the ‘424 patent relates to methods for identifying antibodies and provides examples of two such techniques that produce either murine or human antibodies (hybridoma or phage display techniques). However, atezolizumab is a humanised antibody, and the Supreme Court agreed with the lower court’s assessment that the ‘424 patent did not explain how to produce such humanised antibodies, which would require further research.
The Supreme Court also concurred with the Court of Appeal that the ‘424 patent did not describe all steps necessary for a skilled person to identify atezolizumab even based on the hybridoma or phage display techniques disclosed therein. On this front, the Court of Appeal had noted that Dana-Farber had not shown that such steps constitute simple, routine tasks for a skilled person and had dismissed Dana-Farber’s reliance on cases cited in the EPO Guidelines as evidence for establishing that such steps were part of the skilled person’s common general knowledge. The Court of Appeal found these Guidelines irrelevant as they were published after the priority date of the ‘424 patent and concerned different scenarios.
Finally, Dana-Farber’s arguments based on the disclosure of the later-filed Genentech patents, which describe the identification of atezolizumab, did not persuade the Court of Appeal or the Supreme Court. The Genentech patent describes the use of phage display and hybridoma techniques to develop atezolizumab and refers to publications predating the ‘424 patent’s filing date that describe methods for preparing and humanising monoclonal antibodies. Dana-Farber argued that this indicated that the skilled person knew how to implement equivalent techniques described in the ‘424 patent. However, the Supreme Court noted from the Court of Appeal’s decision that a reference to equivalent techniques in the Genentech patent does not necessarily show that the skilled person also knew how to implement these techniques at the priority date of the ‘424 patent.
Overall, the Supreme Court agreed with the assessment by the lower court and INPI that atezolizumab was not specifically identifiable in the ‘424 patent.
Conclusions
The Supreme Court’s decision offers some clarity on the extent to which a product could be considered specifically identifiable based on the disclosure e.g., of methods for screening for a functional activity. This decision appears to contrast with the French Supreme Court’s decision from last year concerning Dana-Farber’s SPC application for avelumab that appeared to signal a move in a patentee-friendly direction.
In last year’s decision, the Supreme Court agreed that avelumab was specifically identifiable based on the antibody screening methods disclosed in the basic patent and given that the hybridoma technology for generating antibodies was well known since 1975. Notably, the basic patent in that case was also the ‘424 patent. However, a key distinction between the two cases is that avelumab is a ‘fully human’ antibody, while atezolizumab is a ‘humanised’ antibody. Taken together, it appears that the Supreme Court seems to view methods for generating fully human antibodies to be adequately disclosed in the ‘424 patent such that they are specifically identifiable, but this does not extend further to humanised antibodies. Dana-Farber’s SPC application for durvelumab (a fully human anti-PD-L1 antibody), also based on the ‘424 patent, is awaiting the outcome of appeal proceedings at the Paris Court of Appeal, so it will be interesting to see if further guidance on assessing compliance with Article 3(a) might be forthcoming.
While the technical considerations underlying the atezolizumab decision are specific to the antibody field, the general principles ought to find broader applicability to other biologics, and possibly small molecule compounds. For instance, this decision underscores the importance of adequately describing methods for screening for and generating a product in early patent filings that may not disclose sequences/structures of such products. Moreover, this decision also appears to suggest that possible gaps in information from the patent disclosure may be bridged if it can be shown that this information forms part of the skilled person’s common general knowledge.
Overall, this decision sheds light on the level of disclosure that may be required to meet the requirement for a product to be “specifically identifiable” i.e. the second arm of the framework established by Teva (C-121/17) and Royalty Pharma (C-650/17) noted above. However, the Supreme Court did not need to consider whether the product also satisfied the first requirement, i.e. whether it necessarily comes under the invention covered by the patent, based on the underlying facts in this case. This requirement has recently come to the fore following the CJEU’s judgment in joined cases C-119/22 and C-149/22 Merck Sharp & Dohme, which held that it is necessary for the product to fall under the invention covered by the patent, even in cases where the product is expressly mentioned in the claims of a patent. While the atezolizumab decision in France was handed down subsequent to the Merck Sharp & Dohme judgment, the CJEU’s reasoning could not have been applied ex tunc in any event as the Supreme Court was required to place itself in conditions that existed at the time when the lower Court’s decision was taken.
Originally published in the CIPA Journal