ECJ’s Servier judgments clarify the analytical framework in “pay-for-delay” cases

In these judgments (see, in particular, here and here) the ECJ analyzed the so-called “pay-for-delay” agreements between Servier and several generic medicine manufacturers under EU antitrust law. In doing so, it refined the criteria set out in its previous rulings in Generics (UK) and Lundbeck (and related appeals)—see here for our article examining the relevant analytical framework in these judgments.

The ECJ followed the Advocate General (AG)’s opinions, finding several errors in the General Court (GC)’s judgments and confirming that all “pay-for-delay” agreements between Servier and the generic companies – including with Krka – constituted restrictions of competition by object.

Background

In 1981, the compound patent for perindopril, a medicinal product developed by the Servier group for use in cardiovascular medicine (intended primarily for the treatment of hypertension and heart failure), was filed with the European Patent Office for protection.

However, the perindopril compound patent gradually expired, over the course of the 2000s, in various EU Member States. This led Servier to seek supplementary patent protection for perindopril and its manufacturing processes. These new patents were granted in 2004.

Several manufacturers of generic medicines challenged the validity of the patents before the European Patent Office and national courts. To resolve these disputes, Servier entered into separate settlement agreements with the generic companies — Niche/Unichem (Niche’s parent company), Matrix (now Mylan Laboratories), Teva, Lupin and Krka.

Each of these companies undertook to refrain from entering the market with generic versions of perindopril (allegedly in breach of Servier’s patents) and from challenging those patents, in return for a value transfer by Servier. As such, these agreements were aimed at delaying the market entry of manufacturers of generic perindopril to the benefit of Servier. They were, therefore, designated as “pay-for-delay agreements”, i.e., agreements in settlement of patent disputes between an originator holding a pharmaceutical patent and a manufacturer of generic medicines, involving a value transfer (monetary or otherwise).

The European Commission’s decision (2014)

In July 2014, the European Commission (EC) adopted an infringement decision, finding that the agreements at issue:

• constituted restrictions of competition by object and by effect, in breach of the prohibition on anticompetitive agreements under Article 101 of the Treaty on the Functioning of the European Union (TFEU)
• formed part of an exclusionary strategy implemented by Servier, amounting to an abuse of dominance in breach of Article 102 TFEU

The EC imposed total fines of EUR427.7 million on Servier and the generic manufacturers.

The General Court’s judgment (2018)

In December 2018, the GC annulled in part the EC’s decision. In essence:

1. The GC confirmed that the “pay-for-delay” agreements with Niche/Unichem, Matrix, Teva and Lupin, by virtue of their object, were anticompetitive and upheld the EC’s decision in this respect. This part of the judgment was appealed by Servier to the ECJ.
2. By contrast, the GC considered that Servier’s agreements with Krka were not anticompetitive. It thus overturned the EC’s classification of the agreements as restrictions of competition by object or by effect.
3. The GC also held that the EC, when applying the abuse of dominance rules, had wrongly defined the relevant market as being limited solely to originator and generic versions of perindopril. In particular, the GC concluded that the EC was wrong to consider that perindopril differed (in terms of therapeutic use) from other ACE inhibitors, underestimated the propensity of patients treated with perindopril to change medicines and attached too much importance to the price factor in analyzing the competitive constraints. The GC therefore considered that the EC had failed to show that the finished product market was limited to the perindopril molecule alone, when the latter could be exposed to non-price competitive pressures from other medicines of the same therapeutic class. It ultimately disagreed with the EC’s conclusion that Servier held a dominant position in certain markets and annulled the abuse of dominance part of the decision. The EC appealed this aspect of the judgment.

The AG’s opinions in the appeals (2022)

In July 2022, AG Kokott delivered her opinions in both cases (see here and here). Broadly, she invited the ECJ to rule that all agreements between Servier and the generic companies— including with Krka—constituted restrictions of competition by object:

1. AG Kokott proposed that the ECJ should uphold the GC’s findings that the agreements with Niche/Unichem, Matrix, Teva and Lupin constituted, by virtue of their object, prohibited restrictions of competition. She considered that the GC was correct in assessing, consistent with the ECJ’s earlier case-law, whether the “net gain” derived by the generic manufacturers from Servier’s value transfers could be justified by the existence of any consideration—in other words, whether the value transfers could be explained by anything other than the parties’ interest or desire not to compete. Here, Servier had not demonstrated that the patents at issue had an economic value. In AG Kokott’s view, the GC was therefore correct in finding that these payments were not justified by any consideration other than the agreement not to compete with Servier, as the patent holder.
2. However, AG Kokott recommended that the ECJ should rule that the GC was wrong to hold that the agreements with Krka were not anticompetitive. In her opinion, the GC incorrectly took the view that the agreements—which included a settlement agreement and a licensing agreement—were based on the recognition of the validity of several patents by Krka. Instead, she believed that the EC was right in qualifying the agreements as restrictions by object because the license constituted a significant value transfer from Servier to Krka in exchange for the latter’s commitment not to compete in some EU markets (not covered by the licensing agreement). AG Kokott also considered that the EC had established to the requisite legal standard that the agreements prevented Krka’s entry onto the market and therefore removed an important source of potential competition for Servier.
3. Finally, AG Kokott considered that the GC provided insufficient and contradictory reasoning when it annulled the part of the EC’s decision on definition of the relevant market for the purposes of applying the abuse of dominance provisions. She also considered that the GC was wrong to base its assessment of the relevant market only on the objective characteristics of the products in question, in this case therapeutic substitutability. In AG Kokott’s view, it is also necessary to determine whether such substitutability gives rise to an effective competitive constraint, taking into account the conditions of competition and the structure of demand and supply on the market concerned.

The ECJ’s judgments (2024)

The ECJ has now delivered nine judgments in the case. It followed the AG’s opinions, dismissing the appeals of Niche/Unichem, Matrix, Teva and Lupin, and finding that Servier’s “pay-for-delay” agreements were all anticompetitive by their very nature, including the agreements with Krka.

The ECJ found several errors in the GC’s judgment with respect to: (a) the classification of the Krka agreements, (b) market definition for the purposes of the abuse of dominance assessment, and (c) the duration of Servier’s infringement in relation to its agreement with Lupin.

1. Regarding the agreements with Krka, the ECJ ruled that two out of three of the agreements constituted restrictions of competition by object. It concluded that the value transferred in relation to the settlement and license agreements was substantial and could only be explained by the parties’ interest not to compete. The ECJ found that the GC erred in concluding that Krka was not a potential competitor of Servier. It held that the license agreement between Servier and Krka ought to be examined jointly with their settlement agreement, as the former would not have been possible without the latter and, together, they implemented a market sharing strategy. According to the ECJ, the analysis should focus on the economic context rather than an agreement’s form and, confirming Superleague, it reiterated that the parties’ intentions cannot be decisive for the finding of an agreement’s object as restrictive. Additionally, diverging from the framework of assessment suggested in Generics, the ECJ held that the GC should not have considered procompetitive or anticompetitive effects in determining whether an agreement is anticompetitive by object. Finally, it referred a third agreement between Servier and Krka (the assignment and license agreement) back to the GC for reassessment.
2. On the definition of the relevant market for the purposes of the abuse of dominance assessment, the ECJ again aligned with the AG and overturned the GC’s findings. It found that the GC was wrong in holding that the relevant market was broader than perindopril. The ECJ considered that the assessment should not be limited to a functional analysis, emphasizing the importance of analyzing cross-price elasticity of demand for pharmaceutical products, by determining whether consumers of a product subject to a small but permanent price increase would switch to substitute products.
3. The ECJ reduced Servier’s fine for the agreements with Lupin from EUR37.1m to EUR34.7m, after concluding that the GC had incorrectly calculated the infringement duration.

Impact of the rulings

The ECJ’s judgments in the Servier saga serve to further clarify the analytical framework for “pay-for-delay” agreements, as well as for restrictions of competition by object more generally.

The court refines the Generics and Lundbeck criteria by reconciling them with the more recent EU case-law. In particular, the ECJ makes it clear that procompetitive effects are not relevant when assessing whether a restriction of competition by object arises. It thereby abandons the suggestion in Generics that procompetitive effects of an agreement “must, as elements of the context of that agreement, be duly taken into account for the purpose of its characterisation as a 'restriction by object'”.

The ECJ further clarifies that an infringement of the prohibition on anticompetitive agreements under Article 101(1) TFEU can only be established if the regulatory context allows for competition. In Servier, the court examined the short-term market entry possibilities of Krka, concluding that Krka was a potential competitor to Servier.

Drafted with the help of Maria Markou.