BioNTech SE & Anor v CureVac SE & Anor

For a patent claim directed to a very broad class of biological constructs, the specification must disclose the asserted technical effect and provide a positive reason for thinking that the effect is plausible across substantially the claim’s scope. A possibility that some constructs may work is insufficient. Later evidence may establish that the effect is not achieved across the claim, including through a careful scientific assessment of experimental data. Statistically non-significant results do not prove equivalence, but the court may assess whether the data show no meaningful improvement by a careful visual and scientific evaluation. A real improvement at a particular time may suffice where the patent defines the benefit in that way, but trivial or isolated differences do not. The claims were also obvious over the cited prior art.

BioNTech SE and Pfizer Inc. sought revocation of CureVac’s EP668 and EP755 patents concerning artificial mRNA molecules containing split poly(A) sequences. Infringement by the Comirnaty vaccines was not disputed if the patents were valid. The trial concerned insufficiency, including disclosure and plausibility of the alleged improvement in protein expression across the claims, sufficiency in fact, obviousness over Thess, and added matter. The patents claimed mRNAs with two poly(A) sequences separated by a linker and intended for intramuscular administration. The central questions were whether the split poly(A) feature was disclosed as causing improved expression, whether that effect was plausible across the breadth of the claims, whether it was actually achieved, and whether the claimed subject matter was obvious.

1. Disclosure. The patents did not disclose that a split poly(A) tail itself improved protein expression. The general statement of improved expression as an object of the invention, read with the claims, did not identify the split poly(A) feature as the cause. The examples and Figure 3 were better explained by the known effects of additional adenines, masking and plateauing. The reference molecule in paragraph [0144] did not provide an appropriate comparator for a split poly(A) tail.
2. Plausibility. Applying the standard in Warner-Lambert, the skilled person needed a positive reason, beyond speculation, to expect the asserted effect across substantially the claim. Even assuming that Figure 3 disclosed the effect, the common general knowledge showed that mRNA degradation pathways were complex and incompletely understood. The evidence did not support a prediction that substantially all linkers and claimed constructs would work. The broad claims therefore lacked plausible technical contribution.
3. Sufficiency in fact. Improved expression at a particular time could constitute a relevant benefit, but it had to be real, meaningful and potentially useful. Trivial differences, isolated outliers and reliance on overlapping error bars were insufficient. A statistically significant difference could nevertheless be practically immaterial. Conversely, the absence of statistical significance did not establish equivalence; the court could assess the data scientifically, including by careful visual comparison. In vitro evidence was not automatically excluded merely because the claims concerned intramuscular administration.
4. The experiments as a whole showed that the technical effect was absent from a significant part of the claim. The evidence did not need to demonstrate failure for every possible construct. The claims were therefore insufficient because the effect was not achieved across substantially all of their scope.
5. Obviousness. Applying Actavis v ICOS, the skilled person would use Thess as a starting point, would be motivated to investigate its apparently synergistic histone stem-loop/poly(A) combination, and would consider testing repeated elements, including constructs with split poly(A) sequences. The claimed subject matter was obvious over Thess.
6. Added matter. The attack failed. The general definition of a poly(A) sequence, read with the passage concerning at least two poly(A) sequences and a lower limit of 60 adenines, provided an individualised disclosure of the relevant claim features.
7. The patents were invalid for insufficiency and obviousness. The added-matter attack failed. The court directed that the parties address the form of order.

First-instance revocation judgment. No appellate history was stated in the judgment.