E cdip/15/inf/2 original: english date: january 8, 2015 Committee on Development and Intellectual Property (cdip) Fifteenth Session Geneva, April 20 to 24, 2015

Linking products to trademarks

The ISP database provides the names under which drugs are marketed as well as their owners and potential licensees that might market products under their own name. We have all trademark filings with INAPI for the period 1991-2010, which contains filings by both residents and non-residents. To associate registered trademarks to pharmaceutical products, we search for product trademarks associated with the drugs’ names as well as the owners as reported by ISP in INAPI´s trademark database. Needless to say, this is a very complex process and the current data file by no means exhausts the trademarks that might be associated with our products.

To give an idea of the difficulty, recall that there are about 12,000 pharmaceutical products in the ISP database. The trademark database has about 780,000 registrations (averaging 2 registrations per each distinct trademark), of which there are about 150,000 registrations in the NICE classes 3 (soaps and cosmetics), 5 (pharmaceuticals, dietary, medical supplies), 10 (medical and surgical instruments), and 44 (medical services & beauty care). About 50,000 of these registrations are renewals, leaving 100,00 unique trademarks. Matching even 12,000 names with 100,000 names requires an automated approach. Our initial algorithm cleaned each name (product and trademark) for special characters and did some standardization by removing frequently repeated words from the product name (e.g. “acido” or “compuesto”). We then matched on the first word in each name. The result of this match was examined for obvious errors, and those were removed. A manual search of the trademark database using the remaining unamtched drugnames was then performed, which added a few more matches.

The resulting match contains 10,461 unique trademark registration numbers for 4,255 unique trademarks. 9,273 of the 12,116 product names (76%) have been matched to at least one trademark.⁴⁴ There are 1,323 unique names of trademark owners. About half of the registrations are renewals and the vast majority of the trademark names are from Nice class 5 (pharmaceuticals), as shown in Table A5 below.

Table A5

Table A6 shows the trademark status of the matched and unmatched patents. The majority (77%) of trademark applications are granted and about 21% are rejected or abandoned. As in the case of patents, pending and granted trademarks are much more likely to have been matched to a product in our ISP dataset, although the share that matched is still rather low.

Table A6

Therapeutic classes

The final step in our data construction was to standardize the therapeutic classes attached to each ISP registration. The raw data in the ISP register contained a total of 1,542 distinct therapeutic classes. 248 (1.7 per cent) of the ISP registrations were missing the therapeutic class and we filled in the missing information. The classes in the raw data do not follow a common structure and the same classes may be labelled in different ways. In addition, each entry potentially contains multiple therapeutic classes. We translated these classes and standardized them which included spelling corrections, name harmonizations, and the grouping of related classes (for example we group “antidepressant selective inhibitor of serotonin reuptake” and “antidepressant”), yielding 594 standardized therapeutic classes. In a final step we match the cleaned and standardized therapeutic classes to a hierarchical classification system maintained by www.drugs.com. This allows us to group therapeutic classes under broad headers and to collapse our data into 19 broad thereapeutic groups consisting of 183 classes; we use these classifications for the analysis.

Table A7 shows the number of ISP registrations by broad therapeutic group and Table A8 shows the number of registrations for the more detailed classes that have 100+ associated registrations. In both cases, the numbers are weighted by the inverse of the number of classes attached to that registration.⁴⁵ Table A8 shows that many of the most common registrations are for products that are “off-patent”, such as NSAIDs, vitamins, analgesics, penicillin, etc., as we expect.

Table A7



Table A8

[End of Annex and of document]

1^ The authors thank INAPI, in particular the pharmaceutical patents team, Aisen Etcheverry, Maria Lorena Chacon, and Alhena Fuentes for their generous support in constructing the data. We also thank representatives of ASILFA and Camara de la Innovacion Farmaceutica for insightful discussions. We also benefitted from the comments of Catalina Martinez, Keith Maskus, and participants of the 2014 Meide conference in Santiago and seminars at INAPI, UC Berkeley, and UC Davis. The opinions expressed in the present study are the sole responsibility of the authors and do not necessarily represent the points of view of the Member States or Secretariat of WIPO.

2^ Still, a number of provisions that allow signatories to put restrictions on the granting and use of patents on pharmaceuticals such as price controls and compulsory licensing are included in the TRIPS Agreement.

3^ TRIPS Article 27.3(a) allows countries to exclude therapeutic and diagnostic methods from patentability, which offers another legal justification for excluding new uses of existing drugs from patentability.

4^ Section 3(d) of India’s Amended Patents Act of 2005 excludes from patentability the “mere discovery of a new form of a known substance which does not result in the enhancement of the known efficacy of that substance or the mere discovery of any new property or new use for a known substance or of the mere use of a known process, machine or apparatus unless such known process results in a new product or employs at least one new reactant.”

5^ Note that there is no primary/secondary patent distinction in patent law. However, this distinction is commonly used to distinguish between different claim types in pharmaceuticals. For example, see Kapczynski, Park, and Sampat (2012). Patents on active ingredients are referred to as primary patents. In later phases of the drug development, patents are filed on other aspects of active ingredients such as different dosage forms, formulations, production methods etc. These patents are referred to as secondary patents. Secondary patents also emerge from changes to formulations and dosages or applications in new therapeutic classes, discovered during clinical trials.

6^ In Brazil, Article 3 of Bill No. H.R. 5402/2013 proposes to explicitly exclude new uses and new forms of existing medicines (including salts, esters, ethers, polymorphs, metabolites, isomers etc.) from what is considered an invention. South Africa’s Draft National Policy on Intellectual Property released in 2013 proposes similar provisions.

7^ Drugs are products that are marketed. Different products may contain the same active ingredient and a given product may contain multiple active ingredients.

8^ We use the combination of priorities and inventors to identify equivalents. Our set of equivalents was verified by INAPI’s patent examiners.

9^ In Chile, all pharmaceutical products that are to be sold on the domestic market have to be registered with the ISP.

10^ Pharmaceutical process inventions remained patentable, although their patent life was restricted to 7 years counting from filing date whereas the life of any other patent was 14 years. The 1970 Patent Act also enacted provisions that allowed compulsory licensing of pharmaceutical drug related process patents.

11^ However, according to TRIPS regulations, patentees were allowed to file pharmaceutical product patent applications already during this 10-year transition period through a so-called “pipeline” system (the mailbox system), adopted during the Uruguay round of negotiations.

12^ Despite the existence of secondary patents, there may be generic entry when the primary patent on an active ingredient expires. It is still true, however, that the existence of secondary patents may create uncertainty about potential patent infringement even when the patent on the active ingredient has expired (EU Commission, 2009).

13^ See Sanofi-Aventis Australia Pty Ltd v Apotex Pty Ltd (No.3) [2011] FCA 346.

14^ This finding is reinforced by the fact that the EU Commission has found an increase in effective patent terms in Europe (EU Commission, 2009), which could be partly explained by weaker incentives for generic producers to challenge existing patents in Europe compared to the U.S.

15^ Note that an existing FDA approval or the approval of any other foreign agency does not eliminate the ISP registration requirement.

16^ For a complete list of active ingredients see http://www.ispch.cl/medicamentos-bioequivalentes.

17^ If a generic drug contains the same quantity of the same active ingredient as the originator drug, drugs are said to be pharmaceutically equivalent. However, to prove bioequivalence, additional conditions have to be met (quality, efficacy and safety). In the U.S. for example, generics have to prove bioequivalence to obtain market approval.

18^ In the U.S., existing patents are taken into consideration in drug approval. Generics are not granted market approval if patents on the original drug are still in force, although the generics company may challenge the patent’s validity or claim not to infringe the patent. In Chile, a proposal was submitted in 2012 for an amendment of Law 19.039 with the aim to create this type of patent linkage (Boletín N° 8183-03: Proyecto de Ley “Linkage”).

19^ Before 1991, Law DL 958 Article 5 established that all pharmaceutical drugs are excluded from patentable subject matter. Note that Chile also joined the Paris Convention in 1991.

20^ A specific feature of Law 19.039 is that patents cannot be invalidated after they have been in force for 5 (formerly 10) years (Article 50).

21^ The provision limited the patent life in Chile for such patents to either the remaining lifetime abroad or 15 years, whichever is shorter.

22^ Law 19.039 Transitional Provisions, Article 1.

23^ Law 17.336 in 2010 did not affect the patents contained in our sample.

24^ Supplementary patent protection is granted if the grant decision has taken more than five years counting from the application date of the patent or three years counting from the date when examination was requested. Supplementary patent protection is also granted if the sanitary registration takes more than a year.

25^ Data exclusivity is only available to new active ingredients. Article 90 of Law 19.039 defines how new active ingredients are distinguished from therapeutic uses, dosages, formulations, combinations, and polymorphs. Article 91 provides exceptions to this kind of protection. For example, when the holder of the information has engaged in conduct or practices declared anti­competitive; on justified grounds of public health, national security, non­commercial public use, national emergency or other extremely urgent circumstances declared so by the competent authority; or when a drug had already been registered abroad for at least 12 months.

26^ Prior to the 2005 amendment, new uses of known substances were patentable as long as they solved a technical problem (Law 19.039 Article 37 e) and improved the existing patented invention (Law 19.039 Article 40). Law 19.996 eliminated the improvement requirement (Article 40).

27^ Article 49 effectively legalized parallel imports as long as the products were marketed abroad by the patent holder (or with the patent holder’s consent).

28^ Under the PCT system, applicants file a single application with the World Intellectual Property Organization (WIPO) that is, however, examined separately in each jurisdiction party to the PCT system in which patent protection is sought.

29^ The INN is the active ingredient’s generic name that cannot be trademarked.

30^ A Markush structure describes sets of specific molecules with a common chemical structure.

31^ In the case of patents, matching a larger share of granted patents to the ISP register is to be expected since we checked all granted patents whereas we matched patents that have not been granted only if (a) they protect a new active ingredient registered with the ISP for the first time between 2005-2010 or (b) if they match via the Orange Book or the Merck Index (see Section V).

32^ We focus on the 1991-2010 period because pharmaceutical patents were introduced in Chile in 1991 (see Section IV). Note that there are truncation issues at both ends of the distribution. We recomputed the chi-squared for observations with lags between -10 and 10 and obtained a value of

26.5.

33^ There is usually more than one class for a given patent, so the total number of entries in the table is 1,246 rather than 569.

34^ E.g., pentahydrate recorded as 5-hydrate on occasion. Calcium spelled out or recorded as the Chemical symbol Ca.

35^ 439 (3%) of the registrations were missing the ISP registration date and are not included in these figures.

36^ The raw file contains about 104,000 entries, with several for each ISP id, firms listed more than once for a single id if they performed multiple functions, and some duplication due to simple spelling errors.

37^ Note that almost all of the organization names are in fact firm names, with a few individuals and universities in addition.

38^ The INN is the official nonproprietary or generic name given to a pharmaceutical substance designated by the World Health Organization (WHO).

39^ This organization names new compounds according to the rules of organic chemistry.

40^ This is due the “Markush” formula. This formula represents a group of compounds related with an active ingredient. These related compounds are usually modifications of the original active ingredient.

41^ The OB does not provide historical patent data, that is, if a patent expires or lapses at the USPTO, the patent is deleted from the register. Bhaven Sampat provided historical records that allow us to correct this problem.

42^ We could have also searched directly for the products contained in the ISP register, but going via active ingredients seems to be the `cleaner’ way of proceeding.

43^ The search for additional matches to the ISP ingredients in these 3,000 granted patents yielded 70 patents (2%). Therefore the remaining unsearched patent applications are unlikely to contain many additional matches, especially since they include a majority of rejected or abandoned filings.

44^ Multiple registrations correspond to the same trademark text: there are many renewals, and text that is the same even if the owner and true trademark are different. So the statistics here may requre more work.

45^ There may be as many as 4 classes per registration, although most have only one or two.