14 Sep 2016

(by Sven Gallasch) On 8 September, the General Court handed down its eagerly awaited decision in – the first ever European judgment concerning so-called pay for delay settlements. The Commission’s decision in this case was heavily criticised by practitioners as well as academics like myself for taking the view that agreements in question would constitute a ‘restriction by object’. In a previous I argued that the Commission might have pushed it too far by finding this kind of agreement an object restriction, especially in the light of the Court of Justice’s decision in Groupement des Cartes Bancaires, where it was held that such restrictions should be interpreted ‘restrictively’. It is therefore perhaps surprising that the General Court has rejected every one of the 10 arguments (by my count) put forward by Lundbeck, and has upheld the Commission’s decision in its entirety – even the level of the fine.

However, having read the decision and considering the facts of the specific case I can now see why the Commission has opted for a ‘restriction by object’ approach, despite the Court of Justice’s recent ruling in Cartes Bancaires. The Commission presents a very strong narrative, which proved very difficult to rebut.

In essence the agreements between Lundbeck and the potential generic competitors was regarded as “cartel by contract” aimed at the exclusion of competitors resulting in market division and output limitation. (para 435) In return for abstaining from competing in the market, the generic companies received an unusually large payment. Moreover the Commission was able to show that the size of the payment was linked to the estimated profit that the generic could have expected upon entry (supported by numerous pieces of evidence (para 388)). This ‘link’ was not disputed by the Lundbeck. (para 362) It tied in with the underlying of pay for delay settlements and led the GC to uphold the argument that the generic exclusion was based on the payment itself rather than the patent settlement. The Commission’s story was further supported by rather incriminating internal statements such as: ‘There is a ‘pile of $’ to be shared’. (para. 368)

Is this decision likely to be the blueprint for all EU pay for delay decisions to come?

This is the key question and the answer may be no, as the Commission did in fact pick an unusual case for its first pay for delay infringement decision. The underlying patent of the settlements in Lundbeck was ‘just’ a secondary process patent and not for example a main patent on the active ingredient – these had already expired. This consideration seems to have significant implications for the entire case:

  • The generics seem to be more likely to be potential competitors due to the process patent. According to the Court the threat of generic entry was sufficiently credible due to the doubt over the validity of the patent and estimated probability of success by the parties (para 125) This meant that the generics had a concrete possibility to enter the market – making them potential competitors (which is one of the key requirements for the finding of an anticompetitive agreement)
  • Highlighting that the patent is question is ‘only’ a secondary process patent, the Court also held that the Commission did not disregard the presumption of validity of Lundbeck’s patents. It rather concluded that the Commission had correctly shown that the relevant patent was simply not capable of blocking entry. (para 166)
  • Considering the facts of the case, the Commission also argued that the settlements were based on Lundbeck’s mere speculation and its subjective assessment of the strength of its process patent (para. 494)

Against this backdrop, one wonders whether the story would have been a different one, if the patent concerned had not been a secondary process patent but rather one of the key patents, and had the parties not been so explicit in their settlement negotiations.

So what can we take away from the decision?

Although the Court reiterates that the Commission expressly stated that it is not required to lay down a generally applicable legal test, but only has to determine whether the specific agreement infringes competition law (para 415), a number of general considerations and statements can be derived from the judgment that are very likely to guide future investigations:

  • First and foremost it is stated that not every patent settlement that includes a value transfer from the brand company to the generic is anticompetitive (para 334, 402)
  • However, the presumption of validity cannot be equated with a presumption of illegality (para 121)
  • An agreement that transforms the uncertainty with regard to the litigation outcome into the certainty that generic competitors would not enter the market, are a restriction by object, when such limits do not result from the assessment of the merits of the patent but rather from the size of the reverse payment which induces the generic company not to compete. (para 336) [This seems to be the key statement, as it is repeated throughout the judgment on numerous occasions]
  • The size of the payment does matter; if it can be linked to averting litigation costs it should be permissible; however, if it is linked to the estimated generic company’s profits it is likely to be a strong indicator of anticompetitive conduct
  • Based on these considerations, the brand company is not entitled to substitute their own subjective assessment of validity (and the infringing nature of the generic’s conduct) for that of an independent judge while paying off the generic party to the settlement for complying with this assessment. (para 390)

In conclusion, is seems that the Commission and the Court were careful to highlight that Lundbeck was a special case because the pay for delay settlements were based on a ‘weak’ process patent, potentially leaving the door an inch open for strong main patents on the active ingredient. We are also left with some guidance.

However one issue remains and that goes back to the notion that pay for delay settlements should be regarded as restrictions by object. Lundbeck seem to claim that if they had intended to prevent all sales of citalopram (the drug in question), they would have had to enter into agreements with all potential entrants (of which there were 300 within the EEA at the time). (para 536) The Commission swept this claim aside purely by relying on its key statement that is mentioned in the third bullet point above. It is a shame that the Commission did not fully engage with this claim, as it generally could have a significant impact on the actual anticompetitive effects of the agreement. The situation is not too dissimilar to the case in Delimitis, where the Court opted for a truncated effects based approach.

With this in mind, I believe that the object approach might be workable in this very case, but that more generally, it should be substituted for the structured effects-based approach that I have developed and has recently been published .