Preventing Poor Pharmaceutical Patents: A Cross-National Comparative Analysis 1. Tahir Amin, Bhaven Sampat, Ken Shadlen

Transcription

1 Preventing Poor Pharmaceutical Patents: A Cross-National Comparative Analysis 1 Tahir Amin, Bhaven Sampat, Ken Shadlen Version 6.0, March 2012 rough and incomplete -- not for citation 1 Funding has been provided by a seed grant from the joint LSE-Columbia University Research Committee. The authors are grateful to Meriem Bekka for assistance with data collection.

2 1. Introduction The introduction of pharmaceutical patents in developing countries raises significant challenges. In many countries the eligibility of product patents to pharmaceutical innovations is new, only introduced in recent years as countries came into compliance with new international obligations. Prior to the finalization of the World Trade Organization s Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), countries could and many did treat pharmaceuticals as non-patentable types of inventions. With non-patentability no longer an option, patent offices in developing countries have been flooded by applications in this area. Because any given medication is likely to be the subject of multiple patents covering, for example, the base molecule, alternative molecular structures, formulations, and dosages, and because each of these patents will have different expiration dates, the liberal granting of pharmaceutical patents can extend periods of exclusivity and thus generate worrisome effects on national health budgets and public health. For example, if a patent on a drug s base molecule is set to expire in, say, 2012, then the granting of additional patents on modifications of the drug, with expiration dates up to 2018, would mean that generic competition (and the accompanying reduction in prices) would not begin for another six years. Given the new obligation to allow pharmaceutical patents, in general, and the concerns that doing so may deter competition and affect prices in these ways, countries may introduce measures to minimize the granting of patents on incremental pharmaceutical innovations (IPIs). The sorts of measures that countries may introduce, which we label pre-emptive, aim to prevent potentially term-expiring patents from being granted. As we shall explain later, the pre-emptive approach differs from relying on litigation between private actors at the end of patent terms over the validity of existing patents. Instead, motivated by a concern that once patents even weak patents are granted they may be difficult to overturn, 2 countries can pre-emptively seek to stop such patents from being granted in the first place. Although countries have legal rights, e.g. under TRIPS (CIPR 2002; Correa 2007; Basheer 2005), to introduce pre-emptive measures, we know little about what sorts of steps different countries have taken and even less about the effectiveness of such steps. What sorts of preemptive mechanisms have countries introduced? Do countries that have adopted such mechanisms end up granting fewer pharmaceutical patents than they otherwise would? This paper aims to fill both of these gaps, offering a preliminary mapping of different approaches, and then empirical analysis of patent application outcomes in two countries. By providing empirical analysis of pharmaceutical patent policy the paper addresses an enormous gap between what observers write about and what observers know about in this area. The introduction of pharmaceutical patents in the developing world has triggered exceptionally heated debates, with substantial attention being paid to developing countries flexibilities under TRIPS and the importance of preserving the formal and effective space for the exploitation of such flexibilities. Yet, notwithstanding all of the attention that the existence, preservation, and use of flexibilities has received, empirical analyses of which countries have taken what steps and to what effect are exceedingly rare. Nor has the flurry of analyses produced to celebrate the tenth anniversary of the 2001 Doha Declaration ( Doha 2 As we shall explain below, once patents are on the register they are difficult to remove -- even non-deserving, low-quality patents. 1

3 Plus 10 ) added empirical support, but rather continued to highlight the existence of flexibilities, underscore the importance of preserving such flexibilities, and encourage policymakers to exploit these opportunities and implement appropriate policies (references). 3 One of the main contentions of this paper is that analysts need to focus not just on what measures countries can take, but on what measures countries do take, why they take them, and to what effect. In this paper we take initial steps toward addressing these lacunae. The remainder of the paper consists of five sections. In section Two we examine the significance of incremental innovations to the pharmaceutical industry and the challenges that patents on IPIs present to developing (as well as developed) countries. In section Three we introduce a typology for comparing pre-emptive strategies, we provide a preliminary survey of pre-emptive strategies different countries have utilized, and we we discuss the strategies of two countries Brazil and India in detail. In Section Four we describe new data on patent applications filed in these two countries and the outcomes of these applications, for a sample of drugs approved in the U.S. between 1996 and The data allow us to compare the two countries, and to begin to assess how well their respective strategies are working. We conclude in Section Five with early thoughts on why these systems exhibit the differences we show in the empirical analysis. The differences in patent prosecution outcomes might, perhaps, be due to differences identified in our typology, though in addition to such institutional factors we also consider how the structure of interests that support and oppose the outputs that these systems generate (or aim to generate) may apply pressure on the way the systems actually operate. 2: Incremental Pharmaceutical Innovation, Evergreening, and Patent Examination In this section we provide a brief overview of IPI and, importantly, the challenges that IPI patents pose for developing countries. In doing so we provide the rationale for introducing pre-emptive approaches, and we emphasize the particular importance of pre-emptive (rather than ex post) approaches for developing countries. Historically, most developing countries did not grant patents on pharmaceuticals, 4 or if they did, only for manufacturing processes (i.e., not for chemical substances per se). Since the 1990s, however, with the introduction of TRIPS, all WTO members are required to make patents available for pharmaceutical and pharmo-chemical products and processes. As countries have changed their national patent laws to comply with TRIPS, their patent offices have subsequently received enormous quantities of applications. 5 Most pharmaceutical patent applications come from abroad. Table 1 presents data on worldwide patent applications, by technology class, in the period The data, collected by World Intellectual Property 3 To the extent that empirical analyses and case studies exist, most of the attention has been paid to one particular policy instrument, compulsory licenses (refs). In contrast to compulsory licensing, which occurs after patents have already been granted, examination practices may prevent particular patents from being granted in the first place. 4 Pharmaceuticals are the final product that we consume, i.e., drugs or medicines. Pharmochemicals, a step backward on the production chain, are the inputs, including the active ingredients (APIs) and intermediates used to make APIs. We use pharmaceuticals to refer to the whole sector, though we distinguish where it is important to do so. 5 Patents are national, so the same applications are examined in each country where the applicants file. 2

4 Organization through annual IP surveys of national patent offices and through the Patent Cooperation Treaty, regard published patent applications (WIPO 2011: 10). WIPO derives technological breakdowns by using a concordance table (Schmoch 2008) that links International Patent Classification (IPC) symbols assigned to each application to technological fields. 6 The degree of concentration is notable: six countries (the United States, Japan, Germany, France, United Kingdom, and Switzerland) account for 77 percent of pharmaceutical and 84 percent of pharmo-chemical patent applications in this period. -- Table 1 -- Incremental pharmaceutical innovation (IPI) consists of developing alternative forms of existing molecules, new formulations, dosing regimens, and new uses of known molecules. Incremental innovation plays an important role in pharmaceutical development, simply because base molecules do not become drugs (i.e. useful products) without additional followon, incremental innovations. Alternative forms of the same base molecules may perform differently in the human body, for example, in that they may be more stable and effective. 7 Devising alternative molecular forms, then, amounts to a type of innovation that contributes to pharmaceutical development. Likewise, once stable and effective molecules are settled upon, pharmaceutical innovation also consists of figuring out how to deliver them. The reader should simply consider that medications are typically consumed in tiny dosages, sometimes as little as 5 milligrams or as much as 500 milligrams of the active pharmaceutical ingredient (API). Delivering such miniscule amounts would be exceedingly difficult, if not impossible, were they not combined with a range of additional inert ingredients (i.e. excipients). Yet while excipients role may ultimately be simply to consume space and to facilitate handling, they must be selected and included in the manufacturing process in such a way as to make for a deliverable and consumable medication that retains the desired biological effects of the API. And, of course, separate formulation processes must be undertaken for each dosage and different systems of delivery (e.g. capsules, pills, ointments, syrups, and so on). Finally, often the same molecules affect multiple parts of the body differently; when molecules intended for one therapeutic use turn out to be more effective in ways other than intended, additional work is entailed to develop the drug for the new uses. Incremental innovation, thus, is a bedrock of the international pharmaceutical industry and a central part of firms intellectual property management strategies. As the International Chamber of Commerce testified to the World Health Organization, the vast majority [of pharmaceutical patents] cover innovations which build on inventions of others (ICC 2005: 4). And when we examine patent landscapes of many drugs, it is common to see one medication protected by tens of patents on diverse molecular forms, formulations, processes and uses (Howard 2007; Rathod 2010). 8 6 WIPO (2011: 10-11) describes the data collection used and provides figures on a range of indicators. The specific data displayed in Table 1, however, are not in the WIPO publications (or on WIPO s website), but were provided to the authors by WIPO. 7 Often molecules that successfully act on target receptors in vitro lack stability or produce undesirable side effects in the human body, while alternative forms of the molecules perform better. 8 Hemphill and Sampat (2012b[pharmathickets]) make this point as well, that pharmaceuticals are misrepresented as a sector marked by discrete patenting patterns. Also note that WIPO (2011: 181, Annex A) classifies pharmaceuticals and pharmo-chemicals as discrete (rather than complex ) technologies, which suggests that a binary approach (complex vs. discrete) is probably not appropriate. 3

5 4 Incremental innovation is not simply foundational for the pharmaceutical industry, but it has become increasingly so. Previous analyses based on U.S. drug approvals show that the share of drugs that are not new molecules has increased sharply over the past quarter century (NIHCM 2001). Recent work examining patent portfolios in the U.S. (Hemphill and Sampat 2011, 2012b) shows the share of drugs with secondary patents those covering aspects of the drugs other than its active ingredient has also increased over time. Among drugs approved between 1985 and 1987 less than 60 percent had such patents; among drugs approved between 2000 and 2002 over 85 percent did. These low quality patents also generate market exclusivity beyond the additional patent in the U.S. Thus nominal patent term has increased from about 14 years to 16 years over the and period. The WIPO data on patent applications presented in Table 1 provide a window to understand the pharmaceutical industry s increased reliance on incremental innovations not just in the U.S. but globally. Applications are classified as organic chemistry (IPC code: C07) when their central claims cover molecules and molecular forms, while applications are classified as pharmaceuticals (IPC code: A61) when their central claims cover formulations, dosages, and uses. In general, though not without exceptions, pharmaceutical patents are more likely to be incremental. 9 The WIPO categories almost certainly understate the extent of incremental innovations, since many pharmo-chemical patents are polymorphs, salts, isomers, and other alternative forms of known molecules. 10 Figure 1 tracks the ratio of pharmaceutical patent applications to pharmo-chemical (organic chemistry) patent applications from , showing a consistent and unmistakeable increase over this period: for every 100 pharmo-chemical applications filed in 1995 roughly 81 pharmaceutical patent applications were filed, while for every 100 pharmo-chemical applications in pharmaceutical applications were filed. The measure, of course, is crude and simplistic, but it s no more crude and simplistic when looking at applications for one year than for another year, so as a basic way of assessing change it is illuminating: the pharmaceutical industry is increasingly dependent on incremental innovation. -- Figure 1 -- What is driving these changes? One reason for the increased prominence of secondary patenting may be regulatory: measures designed to speed the entry of generic competition upon expiration of patents in the U.S., for example, have encouraged originator firms to patent as many incremental steps as possible to prolong the periods of exclusivity their products enjoy. 11 Another reason is science-based: it is increasingly easier to modify than create, as the knowledge and technologies that inspired the development of new molecular 9 To examine this, we collected all patents for drugs approved by the U.S. FDA between 1988 and 2005, and determined their IPC as well as whether the patent claims covered the active ingredient (Hemphill and Sampat 2011). Of the patents in class A61, only 28 percent were active ingredient patents, compared to 67 percent of patents in class C To mitigate against this bias the WIPO concordance does not include pharmo-chemical patents that have a secondary listing as pharmaceuticals. 11 Thus, managing patent life-cycles has become a critical element of originator firms business strategies. This phenomenon, in turn, inspires follow-on producers to apply for patents on their own modifications of existing drugs, with the result being exceptionally cluttered patent landscapes with individual drugs covered by multiple patents. Howard (2007); Hemphill and Sampat (2012b), that potential entrants often face a a web of overlapping claims.

6 entities exhibit diminishing returns (i.e., the low-hanging fruit of the revolution in molecular biology has been picked). Taking out multiple patents on different aspects of a drug in order to cordon off competitors and extend periods of exclusivity is standard practice in the pharmaceutical industry. Critics refer to this pejoratively as evergreening. Although the industry vehemently rejects use of this specific term (GSK 2007), devising patenting strategies to extend periods of protection and protect market shares remains fundamentally important to the pharmaceutical industry. Consider, for example, Carlos Correa s definition of evergreening as a patenting strategy consisting of acquiring patents on minor, often trivial, modifications of existing pharmaceutical products or processes in order to indirectly extend the period of patent protection over previously patented compounds. Suppose we remove the adjectives minor and trivial, referring simply to a patenting strategy consisting of acquiring patents on modifications of existing pharmaceutical products or processes in order to indirectly extend the period of patent protection over previously patented compounds. The new definition is hardly different from what the pharmaceutical industry (and patent lawyers) enthusiastically refer to as patent lifecycle management. Beyond the semantic debates, why might granting additional patents on incremental pharmaceutical innovations be worrisome from a public-health perspective? The principal reason is that patents on incremental innovations can extend periods of exclusivity beyond the dates in which they would otherwise expire, thus postponing the entry of competition and, subsequently, the effects that generic competition ordinarily has on prices. By way of illustration, consider an imaginary pharmaceutical firm Px that, in 1994, applied for a patent on a new molecule XYZ for treating diabetes. Through subsequent research Px came to believe that an alternative form of the molecule, XYZ*, would work better and Px applied for a patent on XYZ* in In 2003, after completing clinical trials, Px received authorization from health authorities to place a drug based on XYZ* on the market. The patent on the base molecule XYZ expires in 2014, while the patent for the alternative form XYZ*, if granted, would not expire until To add further complexity, what we are calling XYZ* is, in reality, more likely to consist not of a single patent application but rather a variety of patent applications on different molecular forms and drug compositions, each with different (post-2014) expiration dates. 12 How countries treat applications for patents on incremental innovations of this sort is of paramount importance. XYZ* may be an innovation (or maybe not), 13 but any country that 5 12 So, perhaps, XYZ* would expire in 2020, XYZ+ in 2021, XYZ# in 2023, and so on. 13 The material in this note could be expanded and included in text. There are plenty of examples of XYZ* patents that lack novelty or inventiveness and should never have been granted in the first place. In this example, we present a more challenging case where there does appear to be an innovation (XYZ* relative to XYZ) but granting the secondary patent on the incremental innovation nevertheless will extend Px s period of market exclusivity. In that light, it is also worth noting that not all incremental innovation patents have equal blocking effects: the extent of the blocking effect of a patent on XYZ* depends on the pharmacological usefulness of XYZ and market dynamics. The important issue here is that there are two different types of evaluation, one (legal) about patentability and one (health) about safety and efficacy, and they do not necessarily co-vary. A new molecule may satisfy the former but not the latter, a secondary innovation may satisfy the latter but not the former.

7 grants this secondary patent will be extending Px s period of market exclusivity for six years beyond 2014, which will affect prices, the government s health budget, and patients access to medicines. From a consumer welfare perspective, then, the status of these IPI patents is hugely significant. Any welfare calculation would also need to consider effects on innovation. Unfortunately, this is difficult. The theoretical and empirical literature on the effects of patent protection in developing countries on innovation is mixed (refs), and there is very little literature on whether patents on incremental innovation are needed to stimulate R&D. The extent to which incremental innovations offer health benefits is also the subject of debate (refs). And while there is no reliable data on the costs of R&D to generate incremental innovations, there is some belief that many incremental patents represent research that was at least obvious to try (refs). Perhaps the best we can say about the innovation side of the ledger is that, by revealed preference, some developing countries may believe that the benefits to access from eliminating these patents exceed the potential negative effects on innovation incentives. Challenges surrounding secondary patents are not unique to developing countries, but rather to all countries that grant pharmaceutical patents. In the US, for example, evaluation (or, more precisely, re-evaluation) of pharmaceutical patents tends to occur, de facto, by courts at the end of patent terms. Taking the example from above, the period after XYZ s patent expires in 2014 but before all of the incremental patents expire tends to feature extensive litigation as generic firms seeking to enter the market find themselves sued for alleged infringement of these secondary patents, and they, in turn, often challenge the validity of the relevant later-expiring patents. The validity of the incremental patents are essentially assessed ex post in the course of litigation rather than ex ante through rigorous patent examination. Given the complexity of patent examination and the fact that most patents are never worked anyhow as drug projects are abandoned along the way, simply granting patents and allowing interested parties to litigate, when they see fit, may be a rational way of addressing the challenge of IPIs. 14 This appears to be the strategy the U.S. and many developed countries have chosen. Yet there are well-known difficulties here, most importantly that seeking to invalidate weak patents through litigation is not just uncertain, expensive and risky, but has the characteristics of a public good: the challenger bears the costs of litigation (and the risk of being liable for damages if unsuccessful), but, if successful in invalidating a patent, the benefits are available to all since invalidation places the knowledge in the public domain. Precisely to surmount this challenge, and thus stimulate the supply of such public goods, the USA offers a bounty to the first successful entrant (Hemphill and Sampat, forthcoming). 15 While Hemphill and Sampat (2012a) have shown that, in the U.S., ex post litigation is an effective mechanism for reducing the extension of patent terms of drugs based on weak IPI patents, a number of factors mitigate against the likelihood of this effect being replicated in developing countries. Most simply, few (if any) other countries offer U.S.-style bounties, so 6 14 Lemley (2001) on rational ignorance. In contrast, Farrell and Merges (2004) question the appropriateness of ex post re-evaluation via litigation. 15 Our sense is that the US system is unique in this regard, though further research on the variety of ex post (Hatch-Waxman-like) mechanisms would serve as a useful complement to our paper.

8 the collective action problem remains paramount. But even if the Hatch-Waxman Act were to be replicated across the globe, there are many reasons to believe that reliance on ex post litigation would not be an adequate response or strategy in the case of developing countries. First, the smaller size of markets means the gains to successful litigation are smaller (thus reducing the incentive to litigate). Second, the greater resource asymmetries between owners and challenges means the likelihood of succeeding in litigation may be smaller. Third, in many developing countries the introduction of pharmaceutical patenting and the flood of pharmaceutical patents may simply overwhelm the capacities of local legal systems to properly handle litigation. 16 A fourth issue simply has to do with uncertainty and search costs: not knowing how many patents exist on a given drug creates uncertainty, conducting searches on patent landscapes is costly; the costs can serve as deterrents to pre-expiration challenges that might be successful, were they to occur. The alternative to ex post re-evaluation via is pre-emptive screening. That is, rather than grant secondary patents with sequential expiration dates at t 1 and rely on litigation among interested private actors at t 15 to determine validity, countries can introduce mechanisms to minimize the grant of weaker secondary patents in the first place. One way of doing so would entail the rigorous application of patentability criteria to assure that only innovations with genuine novelty and demonstrating high levels of inventiveness receive patents. To return to the example above, a strict application of patentability criteria may lead examiners to conclude XYZ* is not novel (perhaps this form was revealed in the original patent or prior art) or that it is not sufficiently inventive (perhaps on the basis of existing knowledge and scientific literature any person skilled in the art could have obtained the molecular form XYZ*). Another strategy of pre-emptive denial would consist of declaring certain sub-types of IPIs, e.g. salts or new uses of existing drugs, as non-patentable. Although simply declaring that incremental innovations are not patentable may be a violation of TRIPS, countries are permitted to introduce pre-emptive mechanisms to minimize the granting of such patents. The remainder of this paper focuses on pre-emptive approaches in the developing world. 3. Characterizing and Mapping Pre-Emptive Mechanisms in the Developing World (Brazil, India, and Beyond) While pre-emptive mechanisms are acceptable under TRIPS and are widely encouraged by legal scholars (e.g. Correa) and health activists (South Centre 2011, ICTSD, UNAIDS 2011), we know very little about what countries have actually done in terms of introducing such measures. It is commonly pointed out that Brazil and India have such measures (which we shall discuss below), but little cross-national research has been conducted to understand the spread (or non-spread) of pre-emptive mechanisms. Morevoer, what we do know tends to be non-systematic, making comparisons difficult. Musungu and Oh (2005) provide data on roughly 45 countries in Africa, Asia, and Latin America, focusing largely on whether or not countries have specific exclusions to patentability in their patent laws. 17 More recently, the 16 In Shadlen s fieldwork in Argentina, Brazil, and Mexico, a constant complaint expressed by lawyers from both the transnational and national sectors regards the poor quality of legal reasoning in pharmaceutical patent litigation on the part of lower-court judges. This is not terribly surprising: the field is complex and the need for technical legal knowledge in the area is new, as pharmaceutical products were until recently outside of the ambit of patent law. 17 Some of these data are replicated in Deere (2008: 78-80, Table 3.4). See also Amin (2012, forthcoming). 7

9 South Centre reports that the East African Community (EAC) has been discussing a regional strategy toward the use of TRIPS flexibilities that potentially includes rigid assessment of pharmaceutical patents (South Centre 2011: p. 6), though no evidence is provided that such a regional strategy has been adopted (let alone implemented at the national level); 18 and on the same page of the report, when providing illustrations, the South Centre points to India as the sole example. Nor does UNAIDS (2011) provide details. In 2008 the Philippines adopted virtually identical language in the Universally Accessible Quality and Cheaper Medicines Act (Correa and Matthews 2011: 20). In Thailand, evidently, there is also discussion about a legislative project to emulate India, but, again, little appears to have happened. 19 To return to a point made in the introduction, the gap between what we talk about and what we know about is frustratingly large. 20 To update the stock of knowledge with regard to who is doing what, and to enable comparisons, we introduce a typology of pre-emptive mechanisms and embark on a preliminary mapping of national practices. We characterize pre-emptive mechanisms along three axes of variation that correspond to (1) modes of operationalizing and applying patent law, (2) distribution of roles and responsibilities among government agencies, and (3) the nature of third-party participation. The first axis of variation regards approaches to patent law, whether the mechanisms are designed to apply standard patentability criteria of novelty and inventive step in particular ways for pharmaceutical applications (as suggested by Correa 2007), or whether they include outright statutory exclusions from patentability of some sorts of IPIs. The second axis of variation regards whether the responsibility for pharmaceutical patent examination belongs exclusively done to patent offices or is shared with other (e.g. health) agencies. The third axis of variation regards the availability and nature of opposition systems. We accessed relevant laws from WIPO s database, 21 downloading not only patent laws per se but reforms and other legislative and regulatory items of relevance. 22 In the patent laws we 8 18 See the EAC Regional Intellectual Property Policy, the Protocol on the Public Health Related WTO-TRIPS Flexibilities and the EAC Regional Pharmaceutical Manufacturing Plan of Action, available at According to the South Centre (2011: 11, note 19), The Protocol seeks to provide guidance to the EAC Partner States on how their IP legislation should be adjusted to enable them to fully use the public health related TRIPS flexibilities, to restrict patentability of pharmaceutical products and medical devices in order to keep them in the public domain, promote a local pharmaceutical industry and ensure access to affordable medicines (emphasis added) Moreover, few seem to care about not knowing. The authors asked numerous scholars and activists working in the field for updates, and we also posted an inquiry on the IP-Health listserv. We received few responses, and to extent that anyone referred us to papers with updated research, the papers typically repeated what everyone already knows: that countries can introduce pre-emptive mechanisms and that India has done so! Our approach has been to start with a set of fifteen larger developing and transition countries with growing pharmaceutical markets, i.e. pharmerging markets (IMS 2010), and

10 searched the standard sections that define inventions (or at least declare what is not considered an invention) and stipulate lists of non-patentable inventions. We also examined transitory sections where sector-specific provisions are often included. And in all documents we undertook content searches, using the following keywords: pharmaceutical, health, drug, medicine, patentability, inventive, novel, inspection, opposition, oppose, observation, third part(y/ies). 23 Our approach has multiple limitations, in that elements of pre-emptive legislation may not always be in the laws on WIPO s website but in amendments and implementing regulations. Furthermore some exclusions are in patent office guidelines or judicial rulings (e.g. Andean Court of Justice) that may not be recorded on the WIPO website. Notwithstanding these limitations we believe this is a useful first step, and in and of itself a contribution to the state of knowledge about pre-emptive mechanisms. 24 Table 2 present a preliminary mapping of pre-emptive measures in the developing world on the basis of this typology. Brazil and India, the two countries that we focus on in subsequent sections, are listed at the top. The table shows differences in the types of pre-emptive mechanisms used in these countries (as we discuss below). Moreover, the data collected so far suggests that pre-emptive mechanisms, while not limited widely diffused throughout the developing world, are not limited to Brazil and India. -- Table 2 -- As indicated, the two countries that have received the most attention in terms of pre-emptive screening mechanisms are Brazil and India. Both of these countries began offering product patents on pharmaceuticals in the wake of the TRIPS agreement, though they did so differently and subsequently introduced different sorts of pre-emptive measures. In Brazil, pharmaceutical patents (for products and processes) were made patentable in In addition to being marked by early implementation, in that Brazil could have waited until 2005 to being issuing pharmaceutical patents, an important characteristic of the Brazilian pharmaceutical patent regime was the inclusion of pipeline patents. Pipeline patents recognize, retroactively, patents that had already been granted in other countries. 25 In 2001 Brazil introduced a requirement that all pharmaceutical applications that are approved by the National Institute for Industrial Property (INPI) must be sent to the Ministry of Health for review. Pharmaceutical patents can only be granted after the Ministry s health surveillance agency (ANVISA) issues its prior consent (Shadlen 2009; 2011; Shadlen 2012a). 26 Thus, since 2001 Brazil has experimented with sharing the role of pharmaceutical patent examination among two state institutions, the patent office (INPI) and the health surveillance agency (ANVISA). 9 then build out with additional countries. Language and information-availability issues have altered the order by which we ve proceeded. 23 Of course we used truncated versions of these search terms, and translated versions in Arabic, French, and Spanish for materials in those languages. 24 In subsequent research we would undertake surveys of in-country patent attorneys to help us uncover aspects of national practices that are omitted from our text-based review. 25 Shadlen (2012b, forthcoming: Chapter 5) analyzes the politics of Brazil s early and retroactive implementation. 26 Formally the prior consent requirement was introduced in 1999, as part of a provisional measure (presidential decree), but in reality the change did not occur until after the 2001 reform of the patent law.

11 10 The terms of the Brazilian law are ambiguous. The 2001 reform simply inserted a single clause into the patent law (Article 229-C), which states The concession of patents for pharmaceutical products and processes depends on the prior consent of the National Agency for Sanitary Vigilance [ANVISA]. Thus, while it is clear that ANVISA s prior consent is required for pharmaceutical patents to be granted, the terms on which ANVISA is supposed to exercise this role and grant or deny its prior consent are left unstated. Moreover, the lack of legislative debate, with regard to either the initial 1999 provisional measure or the 2001 reform, provide little record of legislative intent. The little information we have to go on is the letter that the Executive submitted to Congress in making the original change, and this simply says, in the spirit of two heads are better than one, that the purpose of involving ANVISA was to improve the examination of pharmaceutical patents. 27 From ANVISA s perspective this lack of detail was not problematic, but to the contrary pregnant with meaning. After all, since TRIPS and Brazil s patent law both stipulate the criteria on which patents must be granted or denied (novelty, inventive step, utility), the logical interpretation was that ANVISA s role must be to verify that pharmaceutical patents satisfied these standard patentability criteria. 28 Thus, Brazil ended up with a dual examination system in the area of pharmaceutical patents that operates like this: INPI receives and examines a patent application. If INPI determines that the patent should not be granted, then it is rejected. If INPI judges that the patent should be granted (approves), then it is passed to ANVISA. ANVISA examines the application and INPI s technical report, often requesting additional material from the patent office and the applicants. ANVISA may deny or give its consent to the patent, but even when it consents it may do so only after requiring that the applicant narrow some of the patent s claims. ANVISA s report is returned to INPI, which can formally grant the patent only if ANVISA has given its Prior Consent. The Prior Consent rule has generated intense controversy, both internationally and nationally. On the international front, the transnational sector regularly complains of ANVISA s involvement. Within Brazil, prior consent sparked conflict between the two bodies involved in examining pharmaceutical patents in Brazil, on account of their substantive differences over how to examine pharmaceutical patent applications. On most patent applications INPI and ANVISA are in agreement, as we shall see below, yet in the area of secondary patents the two agencies approach examination with different perspectives. While INPI s examination guidelines facilitate patenting of incremental changes to molecules and new uses of existing drugs, for example, ANVISA s approach is significantly more restrictive. Silva (2008) provides a detailed analysis of the prior consent process in its early years, examining every patent application received by ANVISA from and, critically, the reports that ANVISA s examiners made after their preliminary examinations. The data show that in the cases where ANVISA had substantive concerns about a patent application previously approved by INPI (recall, ANVISA only receives applications that INPI has already approved and is prepared to grant), 45.6% of the time the concerns regarded degree of novelty, degree 27 Basso (2006) and Kunisawa (2009: ) also discuss the background of Prior Consent. 28 Personal communication with ANVISA officials.

12 of inventiveness, or whether the innovation amounts to a therapeutic use, while another 36.4% of the concerns regarded insufficient description of the invention. In short, the two agencies have divergent and irreconcilable perspectives on what sorts of incremental innovations merit patent protection. 11 In India pharmaceutical products were not introduced until 2005, as India took full advantage of the transition periods available under TRIPS. India s implementation differs from Brazil s not just in timing, but in the absence of retroactivity: in contrast to Brazil s pipeline patents provision, India adopted a mailbox. Accordingly, India began accepting patent applications as of 1995, but did not publish or examine the applications until 2005, when the mailbox was opened. In the final amendment to the patent act, introducing pharmaceutical patents, India included a provision (section 3d) that defines some IPIs as ineligible for patents. In contrast to Brazil, where the sources of Prior Consent are clouded in uncertainty, the motivations driving the inclusion of Section 3d are clear: this was in response to concerns about evergreening, that secondary patents on existing substances would be used to extend market life and delay generic competition. 29 The final law, passed after March 2005, included a provision that limits patentable subject matter as follows: 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 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. For the purposes of this clause, esters, ethers, polymorphs, metabolites, pure form, particle size, isomers, mixtures of isomers, complexes, combinations, and other derivatives of known substance shall be considered to be the same substance, unless they differ significantly in properties with regard to efficacy. As in Brazil, the pre-emptive measure has been the source of significant controversy. The U.S.-India Business Council, in a report on Section 3(d), exclaims (with some exaggeration), that despite the increasing harmonization of intellectual property norms, India stands alone as the only country in the world to exclude the full range of incremental pharmaceutical innovations from patent eligibility (US India Business Council 2009, 2). Concerns about 3(d) rose to prominence in 2006, when the IPO rejected an application by Swiss pharmaceutical firm Novartis for a patent on 3(d) grounds. Novartis, in turn, filed a lawsuit challenging the constitutionality of 3(d), and whether it was TRIPS compliant. Though the Madras High Court ruled that 3(d) was constitutional, and that it lacks the jurisdiction to decide on TRIPS-compliance, the conflict is far from resolved, as Novartis s appeal of the IPO s decision continued and will be heard by the Indian Supreme Court Until the very last days of the debate, there was uncertainty about what this aspect of India s TRIPS-compliant patent law would look like. Thus, even in January 2005, a New York Times editorial ( India s Choice, January 18, 2005) noted that the legislation was uncomfortably vague about whether companies could engage in `evergreening. 30 See Vikas Bajaj and Andrew Pollack, India s Supreme Court to Hear Dispute on Drug Patents, New York Times, 6 March 2012 (

13 12 Consider these two countries according to the typology introduced above (see Table 2). In Brazil, Prior Consent entails ultra-rigorous application of standard patentability criteria (pharmaceutical patent applications that are approved by INPI are passed to ANVISA, where this agency undertakes a second evaluation that denies, or reduces the claims of, applications for innovations that are deemed to lack novelty or sufficient levels of inventiveness), while the Indian system is based on the restriction of patentable subject matter, in that the patent law (Section 3d) stipulates that many incremental innovations are not eligible for patents unless they demonstrate increased efficacy. 31 In Brazil responsibility is shared between INPI and ANVISA through the Prior Consent mechanism, while in India the IPO retains solely discharges the role of examination. In Brazil pre-grant opposition is much less extensive than in India, which has a robust system of pre-grant opposition that invites actors from civil society and industry to challenge patent applications that the IPO is examining. Despite significant controversy surrounding the pre-emptive mechanisms in India and Brazil, little is known about their effects. More generally, there is a paucity of data on the outcomes of patent prosecution in India and Brazil in the post-trips era. The following section describes empirical analyses that aim to help fill that void, by examining the outcomes of patent applications filed in India and Brazil for a large sample of marketed drugs. 4. Empirical Analyses Data Drugs We began by collecting data on all 170 non-injectable new chemical entities approved by the US Food and Drug Administration (FDA) between 1996 and 2004 that have at least one U.S. patent listed on the FDA s Orange Book. See Hemphill and Sampat (2011) for a description of these data. For each drug, we collected information on sales in the U.S. four years after approval from IMS Health, the leading commercial provider of data on pharmaceuticals. We also collected information of all U.S. patents listed on the FDA's Orange book for each drug. As Hemphill and Sampat (2011) note, the Orange Book list most (though not all) of the pertinent patents for drugs marketed in the U.S. Having this set is useful as a basis for locating patents in other jurisdictions, as we describe below. Patent applications As noted above, locating drug patent applications in developing countries is difficult. This creates difficulties not only for potential generic entrants, but also for empirical analyses such as ours. India and Brazil, like most developing (and developed) countries, provide no official link between drugs and their patents similar to the U.S. Orange Book. We thus had to construct patent landscapes for each drug ourselves. To do so, we relied on two primary sources. First, we mapped each US patent to its family applications in other jurisdictions, based on data from a private vendor, the Derwent World Patent Index. Second, we collected information from another IMS database, Patent Focus, 31 Note, however, that according to Kapcyzinski (2009) India also has rigorous application of patentability criteria in the patent law too.

14 which provides global patent data for commercially important pharmaceutical products (including all of the drugs in our sample). 13 The Patent Focus database also includes patent types for each application it covers: Product, Composition, Method of Use, Process, and Drug Delivery Systems. Typically, Product patents are typically thought of those covering novel active ingredients. However, there are different ways to operationalize this: for example, to include alternative isomers, crystalline structures, salts, and metabolites as well as claims for new active ingredients (Hemphill and Sampat 2012a). IMS appears to define this category broadly, though we nevertheless use their characterization for our Product patent category as a proxy for active ingredient patent. 32 Composition Use and Delivery patents are typically viewed as secondary patents (Correa 2007; Howard 2007; Kapcynski et al. 2012). Accordingly, in the analyses we collapse these to a broad Secondary category. Process patents may or may not correspond to secondary patents and we treat these separately. Finally, we create a category for patents that resulted from the DWPI match but were not listed in IMS Patent Focus, Non-IMS patent applications. As a practical matter, almost all of the Brazilian applications located through the Orange Book Derwent approach were also in IMS (about 92 percent). We located a total of 470 Brazilian applications for the 170 drugs in our sample. About half of these (238) were only in IMS (i.e. not identified via Derwent), 213 were in both IMS and Derwent, and 19 in Derwent only. There were only about half as many patent applications for these drugs in India: 237 applications. This may reflect the later date of TRIPS-implementation in India, discussed above. Of the 237 applications about half (111) are in IMS only, 72 in Derwent only, and 54 in both sources. While decidedly inferior to painstaking drug-by-drug searches (Amin 2010), the landscaping approach described above allows us to assemble patents for the universe of drugs. We believe that, together, the two sources do appear to provide a reasonable characterization of the patent landscape for the drugs. More importantly, there is no reason to believe that any false positives or false negatives are systematically related to drug or patent characteristics. Patent outcomes Next, we collected information on the outcomes of the patent applications in each country, as of March For India, we searched the Indian Patent Office s Application Status database for each application s status, and aggregated to: granted (indicating the patent application was granted), pending (indicating the application is still being examined) and rejected/abandoned/withdrawn (indicating that application was rejected, abandoned by the applicant, or formally withdrawn). We collapse the last three categories since withdrawals and abandonments themselves could be responses to examiners requests for information that may suggest unpatentability (Lemley and Sampat 2008). 32 In future work, we intend to code each of the patent applications in India and Brazil by hand, to develop categories that correspond better with the distinctions between primary and secondary patents embodied in the pre-emptive mechanisms in the two countries patent laws.

15 In cases where patents are rejected, we consulted Controller s Decisions to determine the reasons for rejections. To examine the effects of 3(d) we coded each rejection based on whether 3(d) was a reason for the rejection, and also whether it was the reason for rejection, i.e. whether or not other arguments (lack inventive step or novelty, lack of enablement) were cited as well. For Brazil, we searched the INPI database for each of the 470 applications. The INPI website provides data on each transaction that occurs during patent prosecution, which we used to determine whether the application was granted, rejected, or pending. We also consulted information from an ANVISA list of all patent applications the agency has received (since 2001) together with its action on these applications. Using these data, we determined whether Brazilian patent applications were granted, pending, or rejected. For rejected applications, we determined whether the application was rejected by INPI alone, in which case it would never be forwarded to ANVISA, or whether ANVISA was involved. We coded two types of ANVISA involvement associated with rejections: whether it did not consent to a grant, or whether it raised questions of INPI s evaluation that resulted in the patent office reversing its original approval. Results India Of the 170 drugs, 76 had one or more Indian patent applications. As Figure 1 shows, the likelihood of an Indian application increases sharply with sales, with over two-thirds of drugs in the top-sales quintile having an Indian application. 14

16 While most of our analyses are at the drug level (recall that drugs can have multiple patent application), it is also interesting to examine the share of drugs overall that have grants and rejections. Figure 2 shows that 62 percent of Indian drugs have received at least one patent application granted, compared to 30 percent that have had at least one patent application rejected. That pharmaceutical firms are obtaining patents on over three-fifths of the drugs where they file patent applications already suggests that India s pre-emptive mechanisms do not prevent patents on the majority of drugs. This is not surprising since 3(d), like most other anti-evergreening provisions, is aimed to affect but a subset of patent types. 15 We next examined the outcomes for the patent applications on these drugs. Of the 237 applications, 224 are distinct, i.e. in some cases patent applications are associated with multiple drugs. For the application-level analyses we focus on the 224 distinct patent applications. Overall, 34 percent of the applications were granted, 28 percent rejected, and 38 percent remain pending as of March Among the 139 applications that received final disposition, i.e. the granted and rejected applications, the majority (55 percent) are granted. Figures 3 and 4 show the distribution of outcomes for Indian applications by patent type. Grant rates are highest for process patents: this is not surprising since process patents have long been patentable in India. Surprisingly, product patents have the lowest grant rates and highest rejection rates. Indeed, secondary patents have higher grant rates and (slightly) lower rejection rates than product patents. However, these results should be viewed as tentative, for several reasons. First, a substantial share (nearly one-third) of Indian patent applications were not in IMS (i.e. were located through Derwent) and we thus lack sufficient information about

17 patent type for a significant share of the applications in our dataset. Second, as emphasized above, the IMS Product patent category may be over-inclusive, and include a number of patents that are less basic than chemical compound patents. In future work, we aim to read and code these patents independently, to overcome these data constraints. 16

18 Next, we examine the roles that Section 3(d) had on the prosecution of these applications. Figures 5 and 6 show the overall outcomes, breaking out rejections into those that did not involve 3(d), those that involved 3(d) but also included other bases for rejection (e.g. lack of inventive step) and pure 3(d) rejections. 17

19 18

20 19 Despite the attention that scholars and stakeholders have given this pre-emptive provision, that vast majority of rejections in India do not involve section 3(d). Among applications with final decisions, the vast majority (86 percent) are either granted, or rejected on grounds other than 3(d). Only about 2 percent of these applications include pure 3(d) rejections. Brazil Overall, about 81 percent of the drugs in our sample had Brazilian applications, nearly twice as high as the share in India. A substantial share of these applications are pipeline applications and will be treated separately in the analyses below. Including these applications, the general trend in Brazil is similar to that in India, with the likelihood that a drug has a Brazilian application increasing sharply over the sales distribution, as Figure 7 shows. Figure 8 shows the share of drugs with a patent grant is about 72 percent, while the share with at least one patent rejected is 46 percent:

21 20 Next, we examined the outcomes of the 470 patent applications associated with these drugs. Overall, about 30 percent of the applications (143) are pipeline applications. Since these are not subject to the same examination procedures, 33 we drop them from subsequent analyses, leaving a sample of 327 applications. Of these, 14 percent are granted, 58 percent pending, and 27 percent rejected. By contrast to India, Figures 9 and 10 show the grant rate is lower (and the rejection rate higher) for product applications than secondary applications. Subject to the caveats noted earlier about the coarseness of IMS categories, the data suggest that Brazil appears to be more effective in restricting secondary patenting in pharmaceuticals. 33 To repeat, pipeline patents are not examined, but rather revalidated. If the patent was granted elsewhere, provided that no product was already being marketed, the patent was granted in Brazil too (with the same claims as granted abroad). Thus, to the extent that pipeline patents are assessed, they are not assessed with regard to novelty and inventiveness but simply to assure that (1) the patent had been issued abroad, and (2) the drug was not on the market (i.e. it was still in the pipeline ).

22 21

23 However, just as 3(d) has a relatively minor role in affecting overall Indian outcomes, ANVISA has had a limited role in influencing overall Brazilian outcomes. Figures 11 and 12 show that ANVISA was not directly involved in the vast majority of rejections. Moreover, in 6 of the 7 cases where ANVISA was involved, INPI rejected the application following ANVISA requests for information. Only one patent application in our sample (a method of use patent application for the anti-epilepsy drug pregabalin) was rejected due to ANVISA terminating its own evaluation and issuing non-consent. To be sure, the second set of eyes provided by ANVISA may shape INPI s examiners behavior indirectly anticipation of ANVISA scrutiny may lead examiners to be more rigorous---but the controversial nonconsent rule, which requires significant inter-agency cooperation, appears to have had only a limited direct effect on pharmaceutical patent prosecution in Brazil. Indeed, while we have but a single instance of a pure prior consent rejection in our dataset, we have three cases of ANVISA s non-consent being ignored by INPI, where the patent office simply refuses to accept the input of health authorities. 22

24 23 6. Conclusion Although the introduction of pharmaceutical patent protection in developing world has inspired endless conflict and debate, and an extensive literature, very little empirical work has been done about its effects. In this paper we have attempted to address this gap with analysis of pre-emptive mechanisms designed to minimize the granting of patents on incremental pharmaceutical innovations. In particular, we have focused on India s Section 3(d) and Brazil s Prior Consent. One conclusion from our analysis is that many of the debates regarding these two controversial measures suffer from exaggeration. Neither India nor Brazil are patent free zones, as critics of 3(d) and Prior Consent imply (and at times state); neither of these countries are accurately characterized as places that severely limit the granting of pharmaceutical patents. To the contrary, most pharmaceutical patent applications are granted. This observation also qualifies the importance of pre-emptive screening and suggests that, even in their most effective forms, these mechanisms do not have all so much effect. Quite simply, the ability to pre-emptively deny patents to secondary pharmaceutical patents is not a functional substitute for the pre-trips ability to deny patents to all pharmaceutical patents; there is no return to Eden. Going forward, where pre-emptive mechanisms are likely to have the greatest effect, then, is not with regard to the existence of pharmaceutical patents, but rather with regard to the duration of periods of exclusivity conferred by pharmaceutical patents. Our analysis also provides a suggestive evidence that India is less effective at screening out secondary patent applications than Brazil, at least based on comparison of grant rates. At first blush this finding is somewhat surprising, since India s pre-emptive mechanism appears