Bayer Intellectual Property GmbH owns the compound patent ZL00818966.8 of the blockbuster anticoagulant drug rivaroxaban, which is marketed as "Xarelto". Rivaroxaban is a direct inhibitor of factor Xa (FXa), a coagulation factor at a critical juncture in the blood coagulation pathways leading to thrombin generation and clot formation.

In the space of one year, a cohort of pharmaceutical companies, including Nanjing Chia Tai Tianqing Pharmaceutical Co, Ltd (the "petitioner"), launched multiple invalidity challenges targeting Bayer's patent. In the petition for invalidation, the petitioner asserted that the rivaroxaban compound (Figure 1) was devoid of inventiveness, citing several pieces of evidence and proposing numerous combinations of evidence. Taking Compound A of Exhibit 3 (Figure 2) as the closest prior art, the petitioner alleged that both rivaroxaban and Compound A are inhibitors of FXa. The petitioner also argued that the two compounds shared two identical rings (phenyloxazolidinone) in terms of structure and differed in the morpholinone fragment on the left of rivaroxaban and the 5-chloro-thiophene group on the right, but that those fragments could be found in the relevant compounds cited by other evidence. The petitioner thus concluded that there was motivation for obtaining rivaroxaban.

Figure 1: rivaroxaban

Figure 2: Compound A of Exhibit 3


On 3 September 2020, the China National Intellectual Property Administration (CNIPA) rendered an invalidity decision, finding the arguments inadmissible. It opined that pursuant to the three-step method commonly used in the assessment of an inventive step, the technical problem actually solved by the patent was to provide a compound that had good inhibitory activity against FXa. Nevertheless, in order to solve this problem, those skilled in the art would not be motivated to modify Compound A with those distinct structures and obtain rivaroxaban. The CNIPA therefore concluded that the claimed compound, rivaroxaban, was inventive.

Exhibit 3 was a patent application entitled "Benzamidine derivatives", with all the example compounds containing benzamidine or amidine radical. Rivaroxaban and Compound A shared the same structure of the two rings (phenyloxazolidinone) in the middle, yet many example compounds cited in Exhibit 3 were devoid of such structure. Therefore, a conclusion could not be drawn as to whether compounds with the same structure had inhibitory activity against FXa. To improve the inhibitory activity of FXa, Exhibit 3 gave the technical teaching to keep the structure of benzamidine or amidine unchanged and modify Compound A at other positions, rather than to remove the structure of benzamidine or amidine and retain the moiety of the phenylene oxazolidinone amino structure.

In practice, the first step a skilled person tasked to structurally modify a compound would take would be to identify the site for modification. This would be followed by the selection of a radical in the field in view of the properties of radicals or teachings about structures and properties. Apparently, structural modification is not a process of cobbling together individual radicals from compounds with different structures in the prior art.

As regards the morpholinone structure, notwithstanding the presence of the morpholino radical in the compound cited in Exhibit 5, there was no technical teaching that piecing the morpholine radical to Compound A to replace the amidine radical would achieve better inhibitory activity against FXa. Moreover, according to the prior art, the amidine radical in the benzamidine compounds binds with the S1 pocket of FXa. However, the radicals that the petitioner found to replace the amidine radical of Compound A from other evidence were not the radicals bound with the S1 pocket; therefore, there was no substitution motivation.

As regards the moiety of 5-chlorothiophen-2-yl, those skilled in the art, by leveraging the overall teaching in Exhibits 3 and 5, would be incapable of determining how to select a site for modification. The structures of the compounds cited in Exhibit 5 were fundamentally different. There was no technical teaching in the prior art regarding identifying and substituting the radical in a compound with a completely different structure.

Based on the above reasoning, the CNIPA maintained the validity of the patent on the basis of the amended claims.


This case was selected as one of the CNIPA's top 10 patent re-examination and invalidation cases in 2020.

Pharmaceutical compound patents are the jewel in the crown of drug patents. Drugs with good market prospects often fall victim to patent invalidation actions initiated by rivals or generic drug companies, with the validity of compound patents frequently being challenged. The focus of such contests ultimately lies in the inventiveness of the compound.

Chapter 10 of the newly revised Guidelines for Patent Examination, which entered into force on 15 January 2021, affirms that pharmaceutical compound patents follow the three-step method in the determination of non-obviousness, with reference to auxiliary factors including unexpected technical effects. Where the determination of non-obviousness suffices to lead to a conclusive finding on the inventiveness of the patent, auxiliary factors are unlikely to take the centre stage in the determination of inventiveness. In this case, although the patentee raised unexpected technical effects in its validity defence and adduced corroborative evidence, the invalidity decision ascertained the inventiveness of the patent based on the non-obviousness finding without addressing this particular issue.

In practice, controversy may easily arise in each step of applying the three-step method in ascertaining the obviousness of a compound – namely:

  • how to determine the closest compound in the first step;
  • how to determine the technical contribution of the invention so as to reasonably determine the actual technical problem being solved; and
  • how to ascertain whether there is technical motivation for structural modification in the prior art.

The controversy in this case resided in the issue of technical motivation – that is, how those skilled in the art would modify the structure of Compound A and whether they would be motivated to adopt the structure of the invention to solve the corresponding technical problem.

A further analysis of the invalidity decision affirms that the specific drug structure-activity relationship teaching is pivotal in ascertaining whether there is a motivation in the prior art as to transforming Compound A into the rivaroxaban structure. This aligns perfectly with the orientation of the three-step method in solving the technical problem. In practice, the structure-activity relationship also navigates the drug discovery process.

First, a full understanding of the structure-activity relationship is conducive to reasonably identifying the core structure for modification. The structural modification of a compound is usually the result of a radical substitution in the core structure. In the context of the three-step method, the closest compound is the one with the most similar structure selected from the prior art with prior knowledge of the structure of the patented compound, which is a retrospective process. The identical structural part shared by the cited compound and the patented compound is naturally used as the basis for the next group substitution. In theory, however, structural modification could occur at any position of the closest compound. The question that needs to be answered in the first place is why those skilled in the art would retain this same structure and make it the core structure to modify other moieties, rather than directly modify the same structure. The structure-activity relationship can help to answer this question.

As the CNIPA's decision stated, "[t]o modify a compound, the first step is to identify the site for modification". In this case, the petitioner argued for the substitution of the benzamidine moiety on the phenyloxazolidinone core structure of Compound A, which contravened the structure-activity relationship teaching revealed in Exhibit 3 – that is, that benzamidine other than phenyloxazolidinone was a necessary structure for the compounds throughout Exhibit 3.

Second, the structure-activity relationship guiding the structural modification should be explicit rather than broad. This particular structure-activity relationship embodied a specific mode of interaction between the compound and the specific target. The process of modern drug discovery usually starts from the discovery and identification of a target. Based on the interaction of the compound structure and target, lead compounds are singled out and optimised, a process through which candidate compounds are generated. The structure-activity relationship in this case was embodied in the interaction between the compound and FXa. In this sense, anticoagulant compounds with different targets of action do not provide structural guidance for FXa inhibitors. Furthermore, a structure that binds with one site of FXa does not provide guidance for a structure that binds with another site. As the CNIPA put it:

as for the radicals in Exhibits 6, 8 and 9, based on the studies on S1 and S4 pockets in Exhibits 4, those cited in Exhibits 6, 8 and 9 do not bind with S1 pocket, not to mention technical motivation for using the aforesaid radicals to make the modification.

Finally, the exploration of the structure-activity relationship should take root in the prior art, especially the holistic teaching of the closest prior art documents. The modification of the closest compound by those skilled in the art is first influenced by the closest prior art documents. In this case, the closest prior art document (ie, Exhibit 3) was an important basis for the structure-activity relationship. Exhibit 3 neither directly addressed the structure-activity relationship nor gave any data regarding the activity of the compounds. However, the title of the invention, the general formula of the definition, the structure of the specific compounds prepared and its other overall contents still conveyed the information in terms of the direction of structural modification to those skilled in the art – that is, that the benzamidine structure was a necessary structure for the realisation of FXa inhibitory activity. This information was corroborated by the teachings of the petitioner's other evidence; therefore, the understanding of the structure-activity relationship of the compounds cited in Exhibit 3 by those skilled in the art could be established.

The structural modification of drug compounds cannot digress from the structure-activity relationship. In the process of structural modification of a compound, it is always important to ask why those skilled in the art are making such modifications and how the overall activity of the final compound is expected to change after such modifications. Only in this way can the hindsight regarding the determination of the technical motivation be avoided to the greatest extent possible. Jurisprudence does not contradict common sense in the research and development practice of drugs.

For further information on this topic please contact Hu Honghui at Wanhuida Intellectual Property by telephone (+86 10 6892 1000) or email ([email protected]). The Wanhuida Intellectual Property website can be accessed at www.wanhuida.com.

An earlier version of this article was published by Managing IP.