On 25 July 2023, the Court of Appeal handed down its decision in Teva & Sandoz v Astellas concerning the validity of Astellas’ patent to mirabegron for use in the treatment of overactive bladder (“OAB”). At first instance, Meade J had held the patent valid and infringement by the generics’ proposed acts was not separately contested. On appeal, Teva and Sandoz contended that the judge had erred in the application of the law as to obviousness. The Court of Appeal dismissed the appeal, with Lord Justice Arnold giving the leading judgment and Lord Justice Stuart-Smith and Lady Justice Falk in agreement.
First Instance decision
In 2020, Teva and Sandoz issued proceedings seeking to revoke Astellas’ European Patent (UK) 1 559 427 B1 (the “Patent“) and associated SPC. The Patent claims priority to November 2002 and, as noted above, protects mirabegron for use in the treatment of overactive bladder. Mirabegron is a β3-adrenoceptor (“β3-AR“) agonist which results in relaxation of the bladder smooth muscle, expanding the bladder’s capacity and relieving patient symptoms. Astellas market mirabegron under the name Betmiga® and Teva and Sandoz wished to clear the way (having accepted that there would be infringement if the Patent was held to be valid).
By trial, the main issue in dispute was whether the Patent was obvious over one piece of prior art: Australian patent application 199889288 (“AU 228“). AU 228 discloses a series of compounds according to a Markush formula which are described as being therapeutic agents for diabetes mellitus and having anti-obesity and anti-hyperlipemia actions due to selective stimulation of β3-AR (i.e. agonism). While the compounds are defined by Markush formula, six example compounds are disclosed: compound 5 is mirabegron. The only numerical data provided are in relation to compound 6 which is tested in a hyperglycaemic mouse model. There is no discussion of OAB and no numerical data demonstrating the selectivity of the compounds for β3-AR.
Teva and Sandoz argued that it was common general knowledge that selective β3-AR agonists had potential to treat OAB. Further, there was a shortage of human, selective β3-AR agonists. Accordingly, the skilled team would be interested in the β3-AR agonists in AU 228 and test them in the appropriate OAB model. Upon obtaining positive results, the skilled team would then take mirabegron into clinical trials with an expectation of success. Astellas’ case was that β3-AR agonism was one of a number of potential ways to treat OAB. There was no clinical evidence that β3-AR agonism would work to treat OAB; in fact, a β3-AR agonist had failed in human clinical trials to treat obesity. AU 228 is silent as to OAB and mirabegron’s activity (i.e. selectivity and potency). Even if β3-AR agonism were to be pursued, there were many other compounds to choose other than mirabegron.
As regards the CGK, Meade J found that whilst there was “momentum” in the idea of using β3-AR agonists in the treatment of OAB, it was known that certain β3-AR agonists had failed in obesity clinical trials. Further, it was established that not all β3-AR agonists behaved the same way, only a handful were human-selective and their potency varied. The evidence also showed that, at the priority date, there were a number of other approaches being considered to treat OAB. Meade J found that the generic companies had overstated the confidence that the skilled team would have in treating OAB with β3-AR and had oversimplified the situation, in that they argued that the skilled team would consider any β3-AR agonist would be likely to succeed as a treatment. In Meade J’s view, the skilled team could not draw any conclusions about the efficacy of the compounds in AU 228 without further testing. Given the failed obesity clinical trials involving β3-AR agonists, their known unpredictability and the absence of data in AU 228, there was no reasonable expectation of success of mirabegron as a treatment for OAB. Accordingly, the obviousness attack based on AU 228 failed.
On appeal, Teva and Sandoz contended that the judge had erred in the application of the law as stated in Pozzoli and Philips, in particular the premise that in order to be inventive the patent must solve the problem to which is it addressed. On the basis that the Patent only provides data from rat models for OAB (i.e. no human clinical trial data or data from human tissue assays), the generics argued that Meade J’s reasoning depended on two uncertainties that the Patent had failed to solve: (1) the uncertainty regarding β3-AR agonism to treat OAB without human clinical trial data; and (2) the uncertainty as to whether mirabegron was a human β3-AR selective agonist or a sufficiently potent agonist.
Arnold LJ did not accept these arguments for two reasons. First, it was not the case that Meade J had held the Patent to be inventive by overcoming the technical problem of side effects associated with cross-reactive β3-AR agonists; indeed, at first instance, Astellas had accepted that the Patent does not disclose mirabegron’s selectivity for β3-AR. Instead, Meade J had found that, having read AU 288 in light of the common general knowledge, it was not obvious to try mirabegron as a treatment for OAB with a reasonable expectation of success. This was due, in part, to the skilled team’s lack of confidence in β3-AR agonism as a potential therapy for OAB and the lack of disclosure in AU 288.
Second, citing Conor Medsystems, the question of obviousness does not depend on the amount of evidence presented in the specification. Provided that the Patent makes it plausible that mirabegron is efficacious for the treatment of OAB (which the generics accepted), then obviousness should be assessed in the normal way, i.e. whether AU 288 read with the common general knowledge made it obvious to try mirabegron as a treatment for OAB with a reasonable expectation of success. Arnold LJ considered that Meade J had done so and thus saw no reason to interfere with his judgment, thereby dismissing the appeal.
As UK patent litigators will know, obviousness attacks at first instance before the English Patents Court hinge on a multi-factorial evaluation of all the available evidence – both the evidence given in writing by expert witnesses in their reports and also the oral testimony in cross-examination. There is no cross-examination of witnesses in the Court of Appeal and fresh evidence is hardly ever permitted. As a result, the Court of Appeal is generally reluctant to interfere with the findings of the first instance judge absent an error of law or principle. In this instance, permission to appeal was granted on the basis of an arguable tension between the law on obviousness in Conor Medsystems and in Pozzolli/Philips and in particular about the assessment of the technical contribution of the patent. However, by the time the appeal reached Arnold LJ, the generics accepted that there was no conflict between the cases and instead ran the argument that the judge erred in principle because he did not correctly apply the law as stated in Pozzoli and Philips. After having noted the generics faced an obstacle in that Meade J’s judgment “contains a very careful, detailed and nuanced consideration of the evidence and issues”, Arnold LJ’s decision succinctly deals with the generics’ appeal. This decision is therefore a reminder of the challenges facing litigants attempting to appeal a finding on obviousness. In short, unless a real error of principle can be identified, appellants will face an uphill battle to persuade the Court of Appeal that a first instance judge has erred in their assessment of inventive step.
  EWCA Civ 880
  EWCA Civ 588
  EWCA Civ 2230
  UKHL 49