D1 or D2: Selecting the Best Prior Art for Inventive Step Attacks at the UPC

Could-would arguments are not indispensable at the EPO

In inventive step analysis at the EPO, could-would arguments focus on “pointers” in the closest prior art document D1. Pointers are typically text passages where a topic relevant to the objective technical problem has been discussed.

When arguing against inventive step over D1, a could-would argument can state that D1 contains a hint (the pointer) which would have led the skilled person to the claimed solution. If we want to argue in favor of inventive step, we turn the other way: the pointer in D1 leads the skilled person to doubt that the claimed solution could work, therefore the claim must be inventive over D1. The could-would approach is applied in the same way at the UPC.

However, in practice, many prior art documents don’t contain any information pertinent to the objective technical problem. They don’t explicitly point the person skilled in the art in any direction. This is not particularly significant in EPO oppositions, since lack of inventive step can be argued by other means when no pointers are available. We look for prior art which meets a few general proximity criteria and (assuming that it’s not a novelty objection) leads to an easily solvable objective technical problem. A could-would argument is a welcome addition if it can be made, but it is not indispensable.

Recent case law illustrates the importance of could-would arguments at the UPC

The UPC Court of Appeal’s emphasis on could-would reasoning is an interesting aspect of its inventive step analysis, where the EPO’s problem-solution approach is not utilized.

In a recent order concerning a request for provisional measures, UPC-CoA_901/2025 (Abbott v Sinocare), the UPC Court of Appeal discussed a continuous glucose monitoring system. The order offers an interesting illustration of how central could-would reasoning is when inventive step is assessed at the Court.

This leads to the question: Does the absence of a pointer in a prior art document more or less automatically mean that the claim is inventive over that document?

If the answer would be yes, then the optimal prior art in UPC proceedings may differ from the prior art we would prefer to use before the EPO.

Abbott v Sinocare

The Court of Appeal begins its inventive step analysis by restating the now familiar framework: the claim as a whole is compared with the prior art in the context of the description and drawings to formulate an objective problem. Obviousness hinges on whether the skilled person, starting from a realistic starting point and wishing to solve that objective problem, would have arrived at the claimed solution (paragraphs 91-92).

The attacking party carries a positive burden to articulate a coherent line of reasoning: why the chosen starting point is realistic, what the differences are, what problem is addressed, and crucially why the skilled person would be motivated to modify the starting point in the claimed direction. The Court’s treatment of the main starting point, Saffer, demonstrates what is required.

The Court accepts that Saffer is a realistic starting point because it addresses a similar technical field and problem (paragraphs 96–99). The Court then carefully identifies the features that are missing from Saffer (paragraphs 105–110). From that point on, the analysis is driven by the absence of a pointer (paragraphs 115-118). The Respondents attempt to derive a pointer from Saffer’s statement that “other information regarding the graph” may be displayed on the glucose overview screen, and from references to lines indicating when insulin should be administered. The Court rejects this argument.

The absence of a pointer also becomes decisive when the Court considers the proposed combinations with secondary documents such as Stocker, Matian, and Schwartz. In each case (paragraphs 128, 138 and 145), the Court stresses that these documents lack pointers. Against that background, the Court concludes that combining Saffer with these documents would require the skilled person to isolate selected elements from systems with different aims and architectures and transplant them into Saffer’s glucose overview screen without any teaching or motivation to do so.

This reflects how the UPC’s inventive step framework functions in practice. Because the test turns on whether the skilled person would take a particular step, and because that assessment hinges on concrete motivations and technical context, an obviousness attack must articulate a credible narrative of why the skilled person would move from the starting point to the claimed invention.

Conclusions

For parties opposing patents at the UPC, there are three main takeaways:

(1) Due to the way the UPC assesses inventive step, prior art documents which don’t touch on the same problem as the target patent may have little value despite technical proximity. Even a good starting point (such as Saffer) may be insufficient if it lacks a clear pointer to how the objective problem might be solved. Potential prior art should therefore be carefully reviewed not just for how it matches the features claimed in the target patent, but also for its general teaching.

(2) Both at the EPO and the UPC, a prior art document needs to provide a realistic starting point for the person skilled in the art. Beyond that, EPO-optimal prior art may be in need of supplementation before the UPC. If D1 seems like excellent prior art from an EPO perspective but lacks a pointer, we might want to search for a pointer-containing D2 to be able to formulate a D1+D2 objection at the UPC.

(3) Breadth of prior art is no substitute for depth of reasoning. Without a properly articulated “would” chain, even a large collection of prior art may fail to undermine inventive step at the UPC.

Footnote

A noteworthy additional aspect of this Abbott v Sinocare decision concerns the treatment of allegedly non technical features. The Respondents argued that the measures of feature group 1.13 should be excluded from inventive step analysis as non technical user interface features (paragraph 111). The Court rejects this and explicitly draws on the EPO’s COMVIK style reasoning, stating that a feature should not be excluded merely because it is non technical as such (paragraphs 113–114).