Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 Intellectual Property and Access to Research Tools and Data Scientific research often leads to the generation of intellectual prop- erty a term used to refer to a wide range of rights associated with inventions, discoveries, writings, product designs, and other cre- ative works (Eisenberg, 1997~. In recent years, the assignment and use of intellectual property rights have become more common in biomedical research, in some cases generating considerable controversy in the pro- cess. Patents in particular have been a recurrent cause for debate because of their potential effects on the dissemination and use of new knowledge, and thus on the progress of science. This issue may be especially salient for large-scale, collaborative projects that generate research tools1 and products that may be useful to a large number of scientists in the field. Intellectual property issues may also be especially contentious for large- scale projects because providing a few scientists or institutions with very large amounts of money to conduct a large-scale project and then also rewarding them with many revenue-generating patents on the products of the research could be viewed as unfair. The patent system is intended to promote innovation by rewarding inventors with the right to exclude others from using the invention in ex- 1 The term "research tool" is defined as any discovery or invention that can facilitate or be used in subsequent research, including such things as reagents, devices, and databases. However, this term is not found in patent law, and no legal consequences arise from desig- nating a particular discovery as a research tool. 162
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 163 change for an "enabling disclosure"2 of the invention. (For definitions of terms associated with intellectual property, see Box 7-1~. For the system to be successful, patents must restrict rights to products that are valuable and unlikely to be obtained by other means. This can be accomplished through the imposition of threshold requirements, such as novelty and utility. Patents are generally thought to promote technological progress in two ways: by providing an economic incentive to devise new inventions and develop them into commercial products, and by promoting disclo- sure of new inventions to the public (Eisenberg, 1997~. In the absence of patents, competitive forces encourage inventors to protect their ideas by keeping their inventions secret. Such an environment can lead to duplica- tion of effort and reinvention because scientists are not aware of the ad- vances of competitors. In contrast, if patent protection is available, scien- tists may learn more easily of advances in the field, allowing them to focus their energies on developing a subsequent invention (Thorner, 1997~. However, scientists may have incentives for both withholding and dis- closing research results (see Box 7-2), and disclosure need not necessarily take the form of a patent. The traditional method of scientific disclosure in the academic world has been simply through publication in peer-reviewed journals, rather than through patents. Numerous developments over the last two decades including in- creased commercial interest in the field and changes in federal policy to encourage patenting the results of government-sponsored research have contributed to the increasing significance of intellectual property in bio- medical research (National Research Council, 1997~. In 1980, in an effort to promote commercial development of new technologies initiated by small businesses and nonprofit organizations, Congress passed legisla- tion that uniformly3 encouraged patenting of discoveries arising from federally supported research and promoted their commercial utilization (see Box 7-3~. Although the Bayh-Dole Act targeted primarily small busi- nesses rather than universities, the number of university-based patents has since increased greatly. The largest portion of the increase in univer- sity patenting has been in the biomedical sciences (National Research Council, 1997~. 2 Until 1999, when Congress amended the Patent Law, U.S. patent applications were not published. Disclosure, through publication of the patent, occurred only after the patent had been granted. In 1999, Congress passed an amendment to the U.S. Patent Law requiring the publication of most patent applications 18 months after filing, similar to the long-standing requirement abroad. 3 Previously, universities could use Institutional Patent Agreements to take title of intel- lectual property generated with federal funding. However, agreements had to be negoti- ated on a case-by-case basis, and they varied across federal agencies. The new legislation established a uniform policy across federal agencies.
164 LARGE-SCALE BIOMEDICAL SCIENCE Prior to 1980, the federal government sponsored primarily basic or "upstream" research, and broad, unpatented dissemination of results in the public domain was the norm for universities (Heller and Eisenberg, 1998~. Fewer than 250 U.S. patents were issued to universities each year.4 As a result, patents belonged almost exclusively to industry, where scientists 4 See <htip://www.autm.net/index_ie.html>.
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 165 and engineers were doing product-oriented "applied" research. Most pat- ented innovations were incorporated into finished or near-market products and processes, because only then were they considered to be worth the costs associated with obtaining and protecting a patent. The new legisla- tion, combined with the advent of biotechnology, has been a factor in chang- ing this scenario and has contributed to a blurring of traditional distinctions between basic and applied research. Nonprofit institutions are now much more likely to patent their discoveries. In recent years, U.S, patents being issued to universities have exceeded 3,000 per year, with more than 3,7005 issued to Universities in the year 2000 (Pressman, 2002~. Nonprofit institu- tions now often pursue avenues of research that are similar to those in private industry, and research collaborations between industry and aca- demic institutions are now widespread as well. Furthermore, many bio- technology companies have their origins in university-based research (Ducor, 1997~. In fact, commercial biotechnology firms have attempted to fill a niche in research and development somewhere between the tradi- tional basic science of academic laboratories and the targeted product de- velopment of pharmaceutical firms (Heller and Eisenberg, 1998~. The privatization of upstream biomedical research has led to intellec- tual property claims on research results that, in an earlier era, would have been made freely available in the public domain (Heller and Eisenberg, 1998~. Indeed, many biotechnology patents are considered research tools rather than traditional end products, since they are useful primarily for 5 This total, representing 2.4 percent of all utility patents issued in the year 2000, is the aggregate figure for 190 institutions.
166 LARGE-SCALE BIOMEDICAL SCIENCE further scientific research. Most large-scale research projects, as defined in this report, also produce research tools. The term "research tool" is not found in patent law, and no legal consequences arise from designating a particular discovery as a research tool; nonetheless, some patents that fall into this category have been the most contentious with regard to their impact on the progress of science. But the key question relates to access to the research tools, rather than to whether the tools should be patented or not. There are many different ways to transfer patented technologies to other institutions, and the methods chosen can have a significant impact on the availability of research tools. A research project may require access to many research tools, and the costs and administrative burden can
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 167 mount quickly if it is necessary for researchers to negotiate separate li- censes for each of these tools. NONEXCLUSIVE AND EXCLUSIVE LICENSING There has been considerable debate over the strategies used by NIH and universities to disseminate patented technologies developed with federal funds. Ultimately, the value of a research tool is likely to be great- est when it is widely available to all researchers who can use it, but there is no universal agreement as to how broad dissemination is best achieved. Once a research tool has been patented, there is a wide spectrum of op- tions for exercising the intellectual property rights associated with that patent. At one extreme is no protection of the patent, in which no effort is made to prevent infringement. In essence, such a strategy makes the dis- covery freely available to anyone, but precludes others from restricting access through a patent of their own. More commonly, patented research tools are licensed to another party for use in research or product develop- ment, or for purposes of sublicensing to others. A license may be exclu- sive with a single company or nonexclusive, in which case anyone willing to negotiate a contract may have access to the technology (see Box 7-1~. At many major universities, it is now common for industry-sponsored re- search agreements to stipulate that while ownership of any resulting pat- ents will be retained by the university, the sponsoring company will have a first option to an exclusive license. And with the increase in university- industry partnerships, this approach applies to more research than in the past years (Ducor, 1997~. Although the Bayh-Dole Act does not specify a preference for either exclusive or nonexclusive licenses, it does mandate a preference for licensing to small firms. But because small companies, es- pecially start-up businesses, may depend on exclusive rights to establish a competitive advantage and ensure access to high-risk capital, the law may indirectly encourage universities to grant exclusive licenses (Henry et al., 2002~. However, many scholars have suggested that nonexclusive licenses are more effective in ensuring the development and broad use of . . new c discoveries. Examples exist to support both sides of the debate. The Cohen-Boyer patent on basic recombinant DNA technology is an early example of a nonexclusive licensing policy that led to modest pricing and wide distri- bution of the technology. The decision to negotiate a nonexclusive license was critical to the industry, as this technology has contributed enormously to subsequent development of commercial biotechnology (National Re- search Council, 1997; U.S. Patent and Trademark Office, 2000~. On the other hand, the technology for DNA sequencing instruments developed and patented at Caltech was licensed exclusively to Applied Biosystems
168 LARGE-SCALE BIOMEDICAL SCIENCE Incorporated (ABI) at the company's insistence. The technology has since been broadly disseminated and is widely available to researchers, usually through core facilities at universities. ABI is currently the leader in the world market for DNA sequencers, but other companies still have impor- tant market shares (National Research Council, 1997~. One case in which an exclusive license has been widely criticized for restricting the use of a common research tool is that of the patented "oncomouse" (National Research Council, 1994b; Institute of Medicine, 1996; Marshall, 2002a). In the 1980s, Philip Leder and his colleagues at Harvard developed a transgenic mouse that overexpressed the oncogene c-myc and was thus prone to developing cancer. Harvard has been granted three related patents on this oncomouse, and all three have been licensed exclusively to DuPont. The first patent, granted in 1988, claims rights to all transgenic animals predisposed to cancer. Thus, any scientist studying a transgenic animal that is prone to cancer must ob- tain a license from DuPont for permission to use it, regardless of who created the particular transgenic mouse line or what cancer-related gene was altered in its germline. In 2000, NIH brokered an agreement with DuPont in which the com- pany agreed to provide a "free research license" to any NIH scientist or NIH grantee doing noncommercial studies with an oncomouse. How- ever, scientists and their institutions must agree to the terms of the con- tract, which stipulate reach-through license agreements on the resultant downstream research (see the next section of this chapter). Anyone who wants to use an oncomouse in drug screening must obtain a commercial license and pay a considerable fee. Some academic institutions have refused to sign a contract with DuPont, and many have suggested that the broad claims of the patent would not survive a court challenge; however, universities are reluctant to pursue costly litigation (Marshall, 2002a). The patents have already withstood legal challenges in Europe and Japan, but the Canadian Su- preme Court recently ruled that the oncomouse is unpatentable in Canada. Restricting the use of trans~enic mice could Neatly impede cancer - ~ - - --- - - - -I- - -- --- - ~ -- - J r -- - -- -- research because such mice serve as basic research tools and models for human cancer and can also be used to screen for or test new therapeutics. Thus, DuPont's aggressive enforcement of the oncomouse patents could be an obstacle to achieving the goals of NCI's Mouse Models of Human Cancer Consortium (described in Chapter 3~. A recent survey of U.S. institutions holding gene sequence patents showed that companies and nonprofit organizations tend to favor differ- ent strategies for licensing their discoveries (Henry et al., 2002), perhaps reflecting different goals or stages of product development. For example, companies may have more end-stage products for which a nonexclusive
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 169 license would generate the most revenue, while nonprofits may have more upstream products. Whatever the reason, private firms reported that an average of 27 percent of all licenses granted were exclusive, while nonprofits reported an average of 68 percent. An earlier survey of aca- demic technology transfer executives showed that about 50 percent of licenses granted by universities were exclusive; in contrast only 22 per- cent of licenses granted by the NIH in 2001 were exclusive (Pressman, 2002). NIH appears to favor the nonexclusive approach to licensing, given the set of rules it adopted in 1999 with the intent of promoting greater sharing of tools and new materials. The guidelines6 for sharing research tools encompasses four principles: · Scientists who receive federal funds must avoid agreements that stifle academic communications. · Scientists should not seek or agree to exclusive licenses on research tools (defined as inventions whose "primary usefulness" is "discovery" and not a product to be approved by the Food and Drug Administration [FDA]~. · Academic scientists should "minimize administrative impedi- ments" on exchanges of materials by refusing "unacceptable conditions" (such as reach-through provisions). · Academic institutions should be as flexible in dealing with others (including companies) as they would have others be with them. Perhaps not surprisingly, some smaller biotechnology companies, whose survival may depend on selling such research tools, are not enthusiastic about these guidelines (Marshall, 1999~. REACH-THROUGH LICENSE AGREEMENTS One difficulty in licensing research tools is that the value of the li- cense is impossible to determine in advance, so it can be difficult to define mutually agreeable license terms. One of the most contentious issues re- garding the licensing of patented research tools is the "reach-through" clause. Reach-through license agreements (RTLAs) give the owner of a patented invention used in upstream stages of research rights in subse- quent downstream discoveries. Such rights may take the form of a royalty on sales that result from use of the upstream research tool, an exclusive or nonexclusive license on future discoveries, or an option to acquire such a license. In principle, such agreements could offer advantages to both patent holders and the scientists who use the patented tools in their re- 6 Guidelines available at <htip://ott.od.nih.gov/NewPages/RTguide_final.html>.
170 LARGE-SCALE BIOMEDICAL SCIENCE search because they impose an obligation to share profits of successful research without adding to the costs of unsuccessful research. For ex- ample, RTLAs could allow scientists with limited funds to use tools and defer payment until the research yielded valuable results. Patent holders may also prefer a chance at larger payoffs from sales of downstream products, rather than certain but smaller up-front fees. In practice, how- ever, companies fear that RTLAs may lead to stacked, overlapping, and inconsistent claims on potential downstream products. From their per- spective, each RTLA royalty obligation becomes a prospective tax on sales of a new product, and the more research tools are used in developing a product, the higher the tax burden will be (Heller and Eisenberg, 1998; Eisenberg, 1997~. RTLAs can also create challenges for universities, as scientists may find it more difficult to obtain research funding for projects in which they are entailed. RESEARCH EXEMPTIONS One potential way to prevent technology licenses from impeding the progress of scientific research is an experimental-use license exemption, or research exemption, for patented research tools. In principle, such an exemption allows scientists to use research tools without obtaining a li- cense if the research is purely experimental and is not aimed at develop- ing a patentable or marketable product. The goal of this exemption is to facilitate widespread use in subsequent research while preserving the financial interests of the patent holder. Most European countries and la- pan have included the principle of research exemption in their patent statutes, but the U.S. Patent Act has not.7 U.S. courts have recognized a research exemption in theory, but most court cases have arisen in in- stances in which the commercial stakes are high, and the exemption is unlikely to be sustained (Eisenberg, 1997~. Unfortunately, it is difficult to define experimental use in such a way as to maintain the commercial value of research tools for the patent holder. According to Eisenberg (1997: 13~: The problem is that researchers are ordinary consumers of patented re- search tools, and that if these consumers were exempt from infringe- ment liability, patent holders would have nowhere else to turn to collect 7 Congress has enacted laws for two specific research use exceptions. The first permits basic research on an invention during the life of a patent if the research is to develop and submit information to FDA (the Drug Price Competition and Patent Term Restoration Act, 35 U.S.C., Section 271 leg. The second permits the use and reproduction of a protected plant variety for plan breeding or other bona fide research (Plant Variety Protection Act, 7 U.S.C. 2321 et seq., Section 2544~.
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 171 patent royalties. Another way of looking at the problem is that one firm's research tool may be another firm's end product. This is particularly likely in contemporary molecular biology, in which research is big busi- ness and there is money to be made by developing and marketing re- search tools for use by other firms. An excessively broad research ex- emption could eliminate incentives for private firms to develop and disseminate new research tools, which could on balance do more harm than good to the research enterprise. Furthermore, some industry executives have noted that although in- dustry previously often gave university research a de facto research ex- emption, they are now often more reluctant to do so because in many cases, university researchers are seen as competing directly with their own research.8 As a result, companies may feel burdened by the require- ment to license the results of university research that have been patented (but formerly would have been freely available) when the universities continue to expect an exemption for use of the companies' patented re- search tools. A recent court rulings may also make it more difficult for universities to a claim research exemption if they are sued for patent infringement. In ordering a district court to reevaluate its decision in an infringement case brought against Duke University, a U.S. Court of Appeals for the Federal Circuit found that the district court had applied an overly broad concept of the very narrow and strictly limited experimental-use defense (Ergenzinger and Spruill, 2003~. In ruling against Duke, the appeals court stated that even if a university is pursuing research "with no commercial application whatsoever," the institution should not presume that its ac- tions automatically qualify as an exception to infringement. Duke had argued that its use of patented laboratory equipment did not constitute infringement because the equipment was used under the authority of a government research grant and was covered by an excep- tion for experimental uses. However, the scientist claiming infringement offered the counter argument that Duke is in the business of obtaining grants and developing possible commercial applications for the fruits of its academic research, and therefore the research exemption should not apply. Previous court rulings have established that the experimental-use ex- ception can be claimed only for actions performed for amusement, to satisfy idle curiosity, or for strictly philosophical inquiry. The defense ~ Personal communication with Richard Nelson, School of International and Public Af- fairs, Columbia University. 9 Madey v. Duke University, USCAFC, 01-1567. Decided October 3, 2002.
72 LARGE-SCALE BIOMEDICAL SCIENCE does not apply when the use is undertaken "in the guise of scientific inquiry" or in furtherance of a party's legitimate business. The new appel- late ruling states that in the case of a major research university such as Duke, such business includes research that educates students and faculty members, attracts additional grants, and helps "increase the status of the institution and lure lucrative research grants, students, and faculty." Sev- eral major research universities are now petitioning the Supreme Court to review the decision because they believe it will hinder research by forcing scientists to obtain permission before using patented technologies (Mala- koff, 2003~. PATENT POOLS Perhaps a more viable approach to reducing potential licensing ob- stacles associated with research tools is to establish patent pools. Over the last 150 years, patent pools have played an important role in shaping other fields, such as the automobile, aircraft, and telecommunications industries. These patent pools have emerged, sometimes with the help of government, when licenses under multiple patent rights have been neces- sary to develop important new products (Heller and Eisenberg, 1998; U.S. Patent and Trademark Office, 2000~. A patent pool is an agreement be- tween two or more patent holders to license their patents to each other or to third parties. In theory, this type of arrangement can simplify research and reduce transaction costs by allowing interested parties to gather all the tools needed to practice with a certain technology in one place ("one- stop shopping") rather than obtaining licenses from each patent owner individually (see Box 7-4~. Patent pools can also facilitate information exchange and the distribution of risks associated with research and devel- opment. But patent pools can also potentially have detrimental effects (see Box 7-4~. For example, a patent pool could theoretically shield invalid patents, eliminate competition, and inflate prices. Patent pools can also conflict with antitrust laws that were designed to prevent the creation of monopolies and restraints on interstate com- merce (U.S. Patent and Trademark Office, 2000~. As a result, the Depart- ment of Justice evaluated all patent pools prior to the 1960s and created a list of patent licensing practices that were per se antitrust violations. More recently, the Department of Justice and the Federal Trade Commission (FTC) have recognized that patent pools can have significant procom- petitive effects and may also improve a business' ability to survive in a time of rapid technological innovation in a global economy. In 1995, the Department of Justice and the FTC issued Antitrust Guidelines for the Licensing of Intellectual Property and set forth enforcement policies (see Box 7-5~. These guidelines can be summarized in two broad questions:
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 173 · Is the proposed licensing program likely to integrate complemen- tary patent rights? · If so, are the resulting competitive benefits likely to outweigh the competitive harm posed by other aspects of the program? The recent formation of patent pools in a number of fields suggests that the social and economic benefits of these arrangements can outweigh their costs (U.S. Patent and Trademark Office, 2000~. Patent pools could offer similar potential benefits to the field of biotechnology, serving the interests of both the public and private industry by eliminating patent stacks and licensing bottlenecks. In the case of biomedical research, how- ever, the obstacles to establishing patent pools may be greater than in other fields. Because patents are often particularly important to the phar- maceutical and biotechnology industries, firms may be less willing to participate in patent pools that undermine the gains of exclusivity (Levin et al., 1987~. For example, it can be more difficult to "invent around" gene or protein patents than the device or process patents that are more com- mon in other industries. Conflicting agendas of the various patent holders and their tendency to overvalue their contribution to the pool can also make it difficult to reach mutually satisfactory agreements or to develop standard license terms. Patents on research tools are likely to include a diverse set of techniques, reagents, sequences, and instruments, making it difficult to compare the value of the various patents in a potential pool. This heterogeneity is complicated by the fact that licensing agreements are likely to be negotiated early in the course of research and develop-
174 LARGE-SCALE BIOMEDICAL SCIENCE meet, when the outcome is uncertain and the value of downstream prod- ucts is unknown (Heller and Eisenberg, 1998~. UNIVERSITY POLICIES AND TECHNOLOGY TRANSFER OFFICES Regardless of what approach is taken to licensing patented innova- tions, perhaps one of the key issues in fostering the development of tech- nologies developed at universities with federal funding lies in the policies and strategies used by universities to disseminate their innovations and to capitalize on their patented intellectual property. A major intent of the laws that encourage universities to patent the results of federally funded
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 175 research is to facilitate technology transfer for further research and devel- opment. But patents also provide universities with opportunities for new revenue sources with none of the restrictions that may be associated with traditional funding sources. Optimizing technology transfer to promote the commercialization of discoveries and maximizing licensing revenues may entail different strategies; thus the policies of universities regarding intellectual property rights should be considered carefully, within the context of their academic values and mission to advance knowledge. Since 1980, NIH and many universities have created technology trans- fer offices to patent and license their discoveries for further development. Protection of intellectual property rights has helped researchers and insti- tutions generate research funding and has also helped new biotechnology firms raise investment capital and pursue product development (National Research Council, 1997~. A recent survey conducted by the Association of University Technology Managers (AUTM) found that the adjusted gross income received by universities from licenses and options was $1.26 bil- lion in 2000. The survey also found that at least 454 new companies were created and 347 products were made commercially available that year based on academic discoveries (Pressman, 2002~. However, these num- bers must be placed in proper perspective. Despite the perception of many that the role of technology transfer offices is to generate revenue for re- search, university-based technology is generally not very lucrative, ex- cept for a few of the most prominent research institutions. Data from the AUTM survey consistently document that gross revenues generated from a university's patent licensing activity average approximately 4 percent of the research dollars spent by that institution (net revenues are even less after the administrative and legal costs of the technology transfer activity have been paid). Furthermore, few campuses benefit from patents for "blockbuster" products. Of the reported 20,968 active licenses in fiscal year 2000, only 125 (0.6 percent) generated more than $1,000,000 in roy- alty income (Pressman, 2002~.~° Thus, it may be beneficial to the public for universities to consider a variety of patenting and licensing alternatives that may yield lower rev- enues but ensure wider public use. For example, for a given patent, an exclusive license might generate the most revenue or might be the fastest way to generate income, but such a license might not necessarily be the best strategy for fostering the most downstream research activity. In some instances, the objectives of the Bayh-Dole Act might also be achieved by making patented university research results available to all who want to use them at a very low transaction cost. A By comparison, revenues of $100,000/year are roughly equivalent to increasing a uni- versity's endowment by $3 million, assuming a 3 percent rate of interest.
176 LARGE-SCALE BIOMEDICAL SCIENCE Further research is needed to determine whether the frequent inclina- tion of universities to license out their patented inventions exclusively is warranted by the need to stimulate investment in downstream product development or whether exclusive licensing has simply become a default method for transferring technology to industry (Henry et al., 2002~. If the latter is true, NIH could facilitate access to research results on a case-by- case basis by specifying broad licensing expectations whenever feasible in issuing program announcement (PAs) or requests for applications (RFAs) for new initiatives. EXAMPLES OF INTELLECTUAL PROPERTY AND DATA SHARING ISSUES ASSOCIATED WITH LARGE-SCALE PROJECTS Genomics and DNA Patents Debates regarding patents on DNA sequences and access to sequence data have been frequent and often quite contentious in recent years. A1- though raw DNA sequences are viewed as "products of nature" and are therefore not patentable, purified and isolated DNA sequences are now routinely patented. Such DNA sequences are considered large chemical compounds that may be patented as "compositions of matter" under the same principles previously applied to smaller molecules. The number of DNA-based patents issued per year increased expo- nentially between 1980 and 1998. In 1999 the number leveled off, and in 2000 the number began to decrease for the first time (see Figure 7-1~. During the brief history of genomics research, the public and private strategies for publication and patenting of DNA sequences have varied and often overlapped, at times favoring patent applications, with or with- out pursuing patent protection, and at times using rapid publication to disclose findings (Eisenberg, 2000; see Box 7-6~. Indeed, although about half the patents are held by for-profit companies, the U.S. Government holds more patents on DNA sequences than any single private firm or institution (see Figures 7-2 and 7-3~. From an international perspective, the vast majority of DNA-based patents are assigned to owners based in the United States (see Figure 7-4~. Between 1980 and 1993, 80 percent of DNA patents were based in this country. The next-largest holder of pat- ents was Japan, with 7 percent of the total. All other countries held less than 3 percent. The most significant obstacles to obtaining patent protection for DNA sequences have been the utility requirement, which limits protection to inventions that have a demonstrated practical use, and the disclosure requirements, which limit the scope of allowable claims (Eisenberg,1997~. In fact, these issues have been at the heart of the debate surrounding
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 177 4,500 - 4,000 3,500- Q ARC A o E z 3,000 - 2,500 - 2000 1 ,500 500 it' ·1 1'''' ''''I I o l ~ 9~\0 9~\~ 9~\~ 9~\~ 9~\oo DOOR HOOF cool cook 9OoOO 990 99~ 99> 99~ to OF OF Year FIGURE 7-1 DNA-based U.S. patents, 197~2002. SOURCE: LeRoy Walters (<htlp://www.stanford.edu/class/siwl98q/websites/ genom~cs>) and DNA Patent Database (<www.genom~c.org>~. patent applications for gene fragments known as expressed sequence tags (ESTs) (see Chapter 3), which are often accompanied by few data regard- ing the biological function of the sequence beyond what can be hypoth- esized from comparisons with similar sequences with known function. This situation led the U.S. Patent and Trademark Office to adjust its utility standards in 2000, essentially making the requirements for DNA sequen- ces more stringent (Kyd, 2000~. Even in the case of full-length genes, however, patents can generate controversy. Most often, discovery of a disease-associated gene first leads to the development of a diagnostic test. If patent holders choose to enforce their patent rights aggressively, they can preclude scientists and physi- cians from testing patients in their own laboratories, which leads to in- creased costs. Such was the case for the breast cancer-related genes BRCA1 and BRCA2. In other instances, such as Canavan's~ disease, the tests could be performed locally but only after paying a royalty for each test that was performed, again leading to higher costs. In the latter example, the issue was especially contentious because patient advocacy groups had been very involved in collecting patient DNA samples that were used to identify the gene. They then found that access to the resultant diagnos- ~ Canavan's disease is a progressive, degenerative disorder of the central nervous sys- tem. It develops in infancy, usually between the ages of 3 and 9 months. Death usually occurs within 18 months after onset of symptoms.
178 LARGE-SCALE BIOMEDICAL SCIENCE tic test was hindered by the licensing policy of Miami Children's Hospi- tal, the gene patent holder (Marshall, 2000~. As a result of such cases, legislation has been proposed that would exempt researchers and clini- cians who use genetic-based diagnostic tests from patent infringement- similar to legislation enacted in 1996 exempting doctors from suits for using patented medical or surgical procedures (Boahene, 2002~. The Bio- technology Industry Organization opposes such legislation on the grounds that it would devastate the biotechnology industry and drive investors to other industries (Warner, 2002~. A related controversial issue regarding DNA patents hinges on the level of effort and inventiveness associated with such discoveries. The
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 179 Non-profit Other NIH 6% 6% Research Institutes Private University 14% Public University 9% Companies 52% FIGURE 7-2 U.S. DNA-based patent assignees. SOURCE: Stephen McCormack and Robert Cook-Deegan, (<http://www. stanford.edu/class/siwl98q/websites/genomics/entry.htm>) and DNA Patent Database (<www.genomic.org>), August 1999. 500 450 400 - `,, 350- Q o 5] A 250 - 150 - 100 - O- ...... <~ ~~°! ~~: an an l ;?02't494; W`~c Patent Holder FIGURE 7-3 Number of patents in the DNA Patent Database assigned to vari- ous entities in academia, industry, and government, 1980-1999. SOURCE: Robert Cook-Deegan (<http://www.stanford.edu/class/siwl98q/ websites/genomics/entry.htm>) and DNA Patent Database (<www.genomic. Org>), December 1999.
180 LARGE-SCALE BIOMEDICAL SCIENCE United Kingdom 2.1% France 2.4% Japan 7.1% Germany Other 1.9% 7.1% United States 80% FIGURE 74 Ownership (assignee country) of 1028 DNA-based patents 1980-1993. SOURCE: Stephen McCormack and Robert Cook-Deegan (<http://www. Stanford. edu/class/siwl98q/websites/genomics/entry.htm>) and DNA Patent Database, (<www.genomic.org>), August 1999. patentability of a novel DNA sequence depends on the absence of disclo- sure of structurally similar DNA molecules rather than on the level of inventive skill necessary to obtain the sequence. Research scientists may view this as incompatible with the traditional perceptions of scientific achievement. And by simply identifying a gene sequence, a patent holder can claim broad rights to future treatments that target the gene product. Indeed, if patent law systematically allocates stronger rights to those who identify novel DNA sequences while withholding effective patent protec- tion from those who undertake the more difficult tasks of elucidating gene function and developing new therapies, it stands to reason that the latter two activities will be less lucrative than the business of identifying novel DNA sequences (Eisenberg, 1997~. There is currently little hard evidence that this scenario is being played out, and in fact, firms that own the sequence patents will still undoubtedly be motivated to learn the functions of those genes. However, narrowing the scope of patents on biotechnology innovations that are still far removed from clinical applica- tion might stimulate a broader range of research activity by more institu- tions, and thus lead to better therapeutics in the long run. Many questions remain regarding the potential strength and value of patents that cover related subjects, such as proteins that are coded by patented DNA se- quences, as is discussed further in the next section.
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 181 Protein Patents Analysis of the human genome sequence has led to the conclusion that the genome may contain significantly fewer genes than the previous estimate of 100,000. Although current methods for identifying functional protein coding sequences may underestimate the actual number of genes, recent estimates suggest that there are fewer than 50,000 (Lander et al., 2001; Venter et al., 2001~. Nonetheless, researchers speculate that there may be as many as 2 million different proteins, suggesting that many variant proteins can be produced from a single gene. The difference be- tween the two numbers lies in alterations in DNA transcription and RNA splicing, as well as post-translational modifications of the protein prod- ucts, all of which can have a profound influence on the function and activity of the resultant proteins. According to the U.S. Patent and Trade- mark Office, researchers can make separate patent claims on these variant proteins even if the parent gene is already patented, as long as the identi- fied changes lead to new and unclaimed functions and uses. The result could potentially be a confusing landscape of competing gene and protein patent claims, perhaps setting the stage for legal battles for control over further research and development. This may be especially true if a protein variant shows a stronger correlation with disease than is the case for the earlier gene that was patented (Service, 2001b). To avoid the possibility of expensive litigation, companies may find themselves cross-licensing many related patented discoveries. On the other hand, some companies are hoping to develop ways of circumventing the claims of genomics compa- nies that have patent rights for making proteins from patented gene se- quences in bacteria. For example, GeneProt, a new proteomics firm, plans to synthesize proteins chemically (Service, 2001a). Many companies appear to be banking on the patentability and prof- itability of identifying and characterizing unique protein variants. Doz- ens of new biotechnology firms have emerged in the past few years either to conduct large-scale searches for proteins (proteomics) or to sell re- search tools to those doing the searching. Most pharmaceutical compa- nies have also launched their own proteomics efforts. All are now racing to find and patent as many proteins as possible. One of the leading com- panies intended to file 4,000 patents on proteins whose functions are known and linked to disease by the end of 2001. Investors also initially appeared to have confidence in the profitability of this approach. Pro- teomics companies attracted more than $530 million in venture capital funds in 2000 and 2001, and stock offerings have raised hundreds of mil- lions more (Service, 2001c). More recently, however, investment in pro- teomics companies has declined because of predictions of lower profit- ability (Warner, 2002~.
182 LARGE-SCALE BIOMEDICAL SCIENCE Because of all the competing efforts, many proteomics researchers are concerned that all the data will be locked up by various private compa- nies. As a result, there is great interest in undertaking publicly funded proteomics projects whose results would be deposited and organized in a freely accessible database. In October 2001, scientific leaders in the field met with representatives of NIH and other government funding agencies, as well as proteomics companies, to discuss launching a coordinated ini- tiative, perhaps modeled after the NIH-funded Alliance for Cell Signaling (see Chapter 3~. The group recommended pilot projects in three areas- profiling protein expression in selected tissues, detailing proteins' func- tions, and creating new bioinformatics tools (Service, 2001c). Databases With the recent increase in large-scale biomedical research projects that generate immense datasets have come concerns about the organiza- tion and accessibility of databases. To optimize the progress of science, scientists may have to combine data from a variety of academic and com- mercial sources, but this data aggregation can present a serious obstacle for both technical and proprietary reasons. A lack of uniformity or stan- dardization in quality control can be a serious impediment to combining data from different sources. Attempts to protect the intellectual property value of data add additional challenges. Over the past 5 years, both academic and commercial biologists have attempted to use "pass- through" rights that place restrictions on data even after they have been incorporated into other databases. This practice creates opportunities for gridlock. But biotechnology companies in particular worry about making a large financial commitment to a project if there is a risk that the under- lying data could belong to someone else. U.S. companies have resorted to a sophisticated assortment of strategies to prevent copying, including contracts, download restrictions, and frequent updates. At the other end of the spectrum, the "open source code" model that was used to develop free computer software was discussed as an option for the public data- bases containing human genome sequences, although that strategy was ultimately rejected (Sulston and Ferry, 2002~. In such a model, anyone could freely use the information in the database to conduct research, to develop products, or to redistribute the information in any form. How- ever, anyone who did so would not be allowed to place new restrictions on further development or redistribution of the data. Intellectual property law in the United States does not cover data- bases. Recently, however, the concept of database protection has been discussed frequently, in part as the result of a 1996 directive from the
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 183 European community to its members to create in Europe a new type of database that restricts use of the information. In fact, several unsuccessful attempts have been made to pass similar legislation in the United States since that directive was issued (Maurer et al., 2001~. Although the Euro- pean approach has been offered as a model to address the complex issues of database access and use, the legislation is actually unlikely to alter the strategies currently used to protect databases. In fact, critics have argued that the European Community directive has eroded the public domain, overprotected synthetic data of doubtful value (e.g., telephone numbers), and raised new barriers to data aggregation (Maurer et al., 2001~. The threshold requirements of the directive have proven to be quite low, and most lawsuits have been brought by a small number of companies that create synthetic data, in essence making such data more valuable than genuine information. A recent report of the National Research Council (1999) examines trends in access to scientific databases and makes recommendations for striking a balance between legitimate rights to protection and open access for the public good. A symposium hosted by the National Academies in 2002 further examined the potential negative effects of a diminishing pub- lic domain for scientific data, caused in part by pressures to commercial- ize and legislative efforts to protect intellectual property rights (Jenkins, 2002b). Patient Confidentiality and Consent Many of the theoretical large-scale projects described in chapter 2, as well as many of the ongoing projects described in chapter 3, entail the collection and analysis of human samples. As such, these endeavors re- quire additional considerations with regard to data access and research on human subjects. There is an inherent tension in biomedical research between the need to protect the confidentiality of individuals and the need for access to information in order to make progress in understand- ing and treating disease. Because the data collected in large-scale projects are often placed in publicly accessible databases, considerations of pri- vacy, confidentiality, and informed consent must be taken into account before, during, and after the study. A common approach to obtaining informed consent for the use of human samples in specimen banks is to develop a very general consent form that will allow future, unspecified research to be conducted without the need to reacquire consent for every subsequent study. However, such an approach may come into question if a future project entails such objec-
184 LARGE-SCALE BIOMEDICAL SCIENCE fives as genetic analysis with linkage to health information or the inclu- sion of data in public databases, or even if the samples are shared more widely within the scientific community than was originally anticipated. Because of the potential for a breach of privacy and subsequent discrimi- nation or other repercussions, great care must be taken to protect the identity of sample donors. The NCI has developed guidelines for protect- ing the identities of tissue donors while still maintaining links to data on clinical information, but researchers and their institution are respon- sible for protecting human subjects in studies carried out under their purview. The tremendous concern about patient confidentiality in the United States is due in part to both hypothetical and actual lapses in the routine practice of medicine for example, the management of the medical rec- ords in the ordinary setting of day-to-day hospital business, or the misuse of patient information by health care insurers. However, recent legisla- tion aimed at redressing such lapses could also affect researchers' access to patient samples. Known as HIPAA, or the Health Insurance Portability and Accountability Act, the legislation contains rules that protect indi- vidually identifiable health information (Box 7-7~. Because the enforce- ment of the rules will only begin in April of 2003, the impact of HIPAA on biomedical research is not yet known, but there is great concern within the biomedical scientific community. EFFECTS OF INTELLECTUAL PROPERTY CLAIMS ON THE SHARING OF DATA AND RESEARCH TOOLS Although concerns have been raised that intellectual property claims could inhibit access to data and research tools, little quantitative assess- ment has been undertaken to determine whether those concerns are valid. However, a survey of academic life scientists undertaken in 1997 found that 20 percent of respondents had delayed publication of their research results by more than 6 months at least once in the last 3 years to allow for patent application, to protect their scientific lead, to slow the dissemina- tion of undesired results, to allow time to negotiate a patent, or to resolve disputes over the ownership of intellectual property (Blumenthal et al., 1997~. Multivariate analysis indicated that participation in an academic- industry research relationship and engagement in the commercialization of university research were significantly associated with delays in publi- cation. Notably, scientists who reported conducting research on goals similar to that of the Human Genome Project were also more likely to ~2 <htip://www3.cancer.gov/confidentiality.htmI>.
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 185 deny requests for information, data, and materials than were other life scientists. More recently, a survey focusing on geneticistsl3 found that among those who had requested published data, materials, or techniques from another academician, 47 percent reported that at least one request was 13A stratified sample of 3,000 faculty members was selected. The sample included 219 grantees of the Human Genome Project and 1,547 faculty members in genetics or human genetics departments. The remainder of the sample (n = 1234) was randomly selected so that half came from nonclinical departments (n = 617) and half from clinical departments (n = 617~.
186 LARGE-SCALE BIOMEDICAL SCIENCE deniedl4 (Campbell et al., 2002~. The requests most likely to be denied were for biomaterials such as mice or viruses (35 percent), followed by sequence data (28 percent), findings (25 percent), phenotypes (22 per- cent), and laboratory techniques (16 percent). Figure 7-5 shows the rea- sons given by geneticists for intentionally withholding from other scien- tists information, data, or materials concerning their own published research. A number of respondents reported that such withholding of data had adverse effects on their ability to reproduce the work of other investigators, on the timeliness of their own publications, and on their ability to pursue chosen research directions. These adverse effects on re- search progress at the individual and field levels were more likely to be reported by geneticists than by investigators who had experienced such withholding in other life science fields. Many geneticists indicated that the situation was having a negative impact on communication within their field, the education of young scientists, and the rate of scientific progress. Even when data are placed into a publicly accessible database, con- flicts over rights to and use of the data can arise (Marshall, 2002c; Roberts, 2002~. For example, a group of collaborating scientists led by the Marine Biological Laboratory (MBL) in Woods Hole, Massachusetts, won a 5- year, $2.6 million award from NIH in 1997 to sequence the genome of a disease-causing protozoan. Because NIH required rapid availability of the data, raw sequence data were routinely posted on a public website. However, the group also posted guidelines that limited database users to reagent development or mutually agreed-upon projects. When two scien- tists published a paper that included analysis of data from a variety of public databases, including the MBL site, the MBL organizers protested and closed down the website. This is a relatively new issue associated with large-scale projects that generate databases for use by the scientific community. Traditionally, scientists are expected to share data and re- agents following publication, but prior to publishing, scientists have had discretion in choosing what to share and with whom. However, NIH has required grantees to release data as soon as they are generated, often well before publication in peer-reviewed journals. This policy is designed to speed the pace of research and to allow the field as a whole to benefit from the large investment made to generate the data, but it clearly raises new questions regarding the data's use, analysis, and publication. The advisory committee for the International Nucleotide Sequence Databases, which includes Genbank, recently endorsed a data sharing 14Respondents estimated that they had made an average of 8.8 requests for information, data, or materials regarding published research in the previous 3 years, with 10 percent of those requests being denied.
INTELLECTUAL PROPERTY ID ACCESS TO RESEARCH TOOLS AND DATA 187 Reasons for denying request for materials or data Protection of commercial value of results Protection of patient confidentiality Requirements of industrial sponsor Concerns of lack of reciprocation Cost of transfer Protection of unpublished work Protection of students or junior scientists Effort required for transfer 0 20 40 60 80 100 Percentage saying reason is important FIGURE 7-5 A survey of geneticists examined the reasons for which requests for data and materials were denied. SOURCE: Campbell et al., 2002. policy that specifically prohibits use or publication restrictions such as those described above, as well as licensing requirements (Brunak et al., 2002~. A recent report of the National Research Council (2003) urges sci- entists, funding agencies, and publishers to adhere to a uniform policy for sharing data and reagents. Using the acronym UPSIDE (universal prin- ciple of sharing integral data expeditiously), the report reinforces an author's obligation to release data and materials quickly to allow others to verify or replicate published findings. New guidelines for sharing data are also being developed by NIH. For instance, the agency will expect investigators to include information in their research applications about how they plan to share the resultant data or why they are unable to do so (Spieler, 2002~. But it is not clear how adherence to the guidelines will be monitored and enforced, or whether they will lead to real changes in behavior within the scientific community. In principle, NIH has numer- ous legal authorities available to assist in improving access to research tools (see Box 7-8~. In practice, however, the exercise of some options may
188 LARGE-SCALE BIOMEDICAL SCIENCE
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 189
190 LARGE-SCALE BIOMEDICAL SCIENCE be perceived as extreme, and may have far-reaching and perhaps unpre- dictable consequences. Providing funding to cover the costs of sharing materials could per- haps also facilitate a greater willingness to provide requested materials to fellow academic scientists (National Research Council, 2003~. Ultimately, assessing the impact of the increased assertion of intellec- tual property rights in academia due to the Bayh-Dole Act is difficult because few data that could be used for this purpose are reported back (Bar-Shalom and Cook-Deegan, 2002~. A report of the U.S. Department of Health and Human Services (DHHS) (1994) notes many deficiencies in NIH's capability for monitoring patents that result from NIH funding. Consequently, NIH established a database to monitor invention disclo- sures and patents. Nonetheless, the General Accounting Office and the DHHS inspector general have since documented that NIH and other fund- ing agencies still are not being notified of many patented inventions de- veloped with federal funding (U.S. General Accounting Office, 1999~. The General Accounting Office is currently conducting an investigation of the administration, use, and benefits to federal agencies of intellectual prop- erty derived from federally sponsored research (Aker, 2002~. The Presi- dent's Council of Advisors on Science and Technology has also been charged with assessing the benefits and difficulties of the Bayh-Dole Act with regard to the commercialization of products resulting from federally funded research, as well as with providing a review of technology trans- fer mechanisms. After collecting data and soliciting input from various stakeholders, such as small businesses and venture capital groups, the council plans to present the Bush Administration with its technology transfer recommendations in March 2003. Suggestions for major modifi- cations to the act are not expected, however (Jenkins, 2002a, 2003a). SUMMARY Concerns have been raised in recent years about the willingness and ability of scientists and their institutions to share data, reagents, and other tools derived from their research. Many factors contribute to these diffi- culties, including the time and expense of sharing data and materials, the desire to protect raw or unpublished data and intellectual property, the incentive to maintain a lead in a particular research area, and the need to protect patient confidentiality. Since a primary goal of many large-scale biomedical research projects is to produce data and research tools these issues are of great importance when planning and conducting such projects. NIH should facilitate the sharing of data and the distribution of reagents to the extent feasible, by providing funds for the maintenance and distribution of reagents produced through large-scale projects, and
INTELLECTUAL PROPERTY AND ACCESS TO RESEARCH TOOLS AND DATA 1 Al by promoting broad dissemination of data and research tools generated with federal funds. Over the course of the last two decades, the assignment and use of intellectual property rights have become increasingly common in the bio- medical sciences. This phenomenon is due largely to changes in federal policy and legislation aimed at promoting the commercial development of discoveries made with public funding. There is evidence to suggest that biomedical patents assigned to academic research institutions are indeed generating research funds for those institutions and spawning research activities in the private sector. However, many questions have been raised regarding the licensing practices used by institutions to trans- fer their technologies to other institutions, both public and private. A number of licensing strategies exists for technology transfer, and the strat- egy chosen could greatly impact the accessibility of new discoveries to the scientific community. Unfortunately, however, little effort has been de- voted to studying the impact of licensing practices on the use of patented biomedical innovations and on the progress of scientific research. Patents and licensing practices have perhaps been most contentious for innovations that can be used as research tools for further research. Even a basic research project may require the use of several patented research tools, so it can be difficult, expensive, and time-consuming to acquire licenses for conducting such a project. Because the goal of many large-scale projects is to produce data and reagents that can be used as research tools, this issue may be especially salient for large-scale endeav- ors. NIH has many tools at its disposal to encourage and facilitate easy access to tools and discoveries derived from federally funded research. However, a lack of information on and scholarly assessment of licensing and technology transfer practices may be hindering effective action on the part of NIH. Thus, a systematic examination of the ways in which licens- ing practices affect the availability of research tools produced by and used for large-scale research projects could be extremely useful in formulating future NIH policies and actions.
