Association For Molecular Pathology et al v. United States Patent and Trademark Office et al

Filing 225

DECLARATION of Fiona E. Murray, Ph.D. in Opposition re: 175 MOTION for Judgment on the Pleadings., 152 MOTION for Summary Judgment.. Document filed by American Society For Clinical Pathology, College of American Pathologists, Association For Molecular Pathology, Haig Kazazian, Arupa Ganguly, Wendy Chung, Harry Ostrer, David Ledbetter, Stephen Warren, Ellen Matloff, Elsa Reich, Breast Cancer Action, Boston Women's Health Book Collective, American College of Medical Genetics, Lisbeth Ceriani, Runi Limary, Genae Girard, Patrice Fortune, Vicky Thomason, Kathleen Raker. (Attachments: # 1 Exhibit 1, # 2 Exhibit 2, # 3 Exhibit 3)(Hansen, Christopher)

Association For Molecular Pathology et al v. United States Patent and Trademark Office et al

Doc. 225 Att. 2

EXHIBIT 2

Dockets.Justia.com

P O L I C Y F O RU M
I N T E L L E C T UA L P RO P E RT Y

Intellectual Property Landscape of the Human Genome
Kyle Jensen and Fiona Murray*

ene patents are the subject of cons i d e r a bl e debate and yet, like the term "gene" itself, the definition of what constitutes a gene patent is fuzzy (1). N o n e t h e l e s s , gene patents that seem to c a u s e the most Enhanced online at c o n t r ov e r s y are www.sciencemag.org/cgi/ t h o s e claiming content/full/310/5746/239 h u m a n proteinencoding nucleotide sequences. This categor y is the subject of our analysis of the patent landscape of the human genome (2). Critics describe the growth in gene sequence patents as an intellectual property ( I P ) "land grab" over a finite number of h u m a n genes (3, 4). They suggest that overly broad patents might block follow-on research (5). Alternatively, gene IP rights may become highly fragmented and cause a n anticommons effect, imposing high costs on future innovators and underuse of g e n o m i c resources (6). Both situations, critics argue, would increase the costs of genetic diagnostics, slow the development of new medicines, stifle academic research, and discourage investment in downstream R&D (7­11). In contrast, the classic argument in support of gene patenting is that strong IP protection provides incentives crucial to downstream investment (12, 13) and the disclosure of inventions. Patents are also regarded as the cornerstone of vibrant markets for ideas (14) and central to the biotech boom of the 1980s and 1990s (15). Policy-makers are hampered by the lack o f empirical data on the extent of gene p a t e n t i n g . Most analyses have relied on anecdotal evidence (11, 16­18) and empirical analyses have been hindered by (i) limited (and poorly defined) coverage of DNA s e q u e n c e patents (17, 19); (ii) difficulty s e p a r a t i n g patents that claim gene sequences per se from those merely disclosing DNA sequences (20­22); and (iii) dis-

G

K . J e n s e n is in the Department of Chemical E n g i n e e r i n g , F. M u r r a y is in the Sloan School of Management, Massachusetts Institute of Technology, Cambridge, MA 02139, USA. * A u t h o r for correspondence. E - m a i l : f m u r r a y @ mit.edu

tinguishing patents on the human genome from those on other species (23). Our detailed map was developed using b i o i n f o r m a t i c s methods to compare nucleotide sequences claimed in U.S. patents to the human genome. Specifically, this map is based on a BLAST (24) homology search linking nucleotide sequences disclosed and claimed in granted U.S. utility patents to the s e t of protein-encoding messenger RNA transcripts contained in the National Center f o r Biotechnology Information (NCBI) RefSeq (25) and Gene (26) databases. This method allows us to map gene-oriented IP rights to BMP2 specif ic physical loci on JAG1 the human genome (27) (see figure, right). Our approach is highly specif ic in its identification o f patents that actually BCL2L1 claim human nucleotide GDF5 sequences. However, by l i m i t i n g the search to patents using the canoniCD40 cal "SEQ ID NO" claim language we do not conBMP7 s i d e r claims on genes def ined through amino a c i d sequences. (See table S1 for a sensitivity Physical mapping of patent activity on chromosome 20, divided analysis.) i n to 300-kb segments. E a c h horizontal bar represents a unique Our results reveal that patent claiming a gene sequence located in that region. Orange repren e a r l y 20% of human sents the number of unique patent families in a region (28). Labels g e n e s are explicitly show the loci of highly patented genes (see table S1). claimed as U.S. IP. This represents 4382 of the 23,688 of genes in the B M P 7 , an osteogenic factor, and NCBI's gene database at the time of writing CDKN2A, a tumor suppressor gene, were (see figure, right). These genes are claimed in t h e most highly patented genes in the 4270 patents within 3050 patent families (28). g e n o m e [their sequences were each Although this number is low compared with c l a i m e d in 20 patents (table S2)]. The prior reports, a distinction should be made p a t e n t s on CDKN2A are distributed between sequences that are explicitly claimed between nine different assignees and, coland those that are merely disclosed, which lectively, claim all three splice variants of outnumber claimed sequences roughly 10:1. t h e gene. Nearly all of these patents are The 4270 patents are owned by 1156 different d i r e c t e d toward diagnostic applications. assignees (with no adjustments for mergers In contrast, the patents on BMP7 are for a n d acquisition activity, subsidiaries, or the use of BMP7 proteins in implants to s p e l l i n g variations). Roughly 63% are stimulate bone growth. However, a numassigned to private firms (see figure, above). ber are directed towards more speculative Of the top ten gene patent assignees, nine are u t i l i t i e s , such as drug-screening probes, U. S . - b a s e d, including the University of w h i c h suggests a strategy of "scienceSCIENCE VOL 310 14 OCTOBER 2005

California, Isis Pharmaceuticals, the former SmithKline Beecham, and Human Genome Sciences. The top patent assignee is Incyte Pharmaceuticals/Incyte Genomics, whose IP rights cover 2000 human genes, mainly for use as probes on DNA microarrays. Although large expanses of the genome are unpatented, some genes have up to 20 patents asserting rights to various gene uses a n d manifestations including diagnostic u s e s , single nucleotide polymorphisms (SNPs), cell lines, and constructs containing the gene. The distribution of gene patents was nonuniform (see figure, page 240, top right): Specific regions of the genome are "hot spots" of heavy patent activity, usually with a one-gene-many-patents scenario (see f i g u r e , below). Although less common, there were cases in which a single patent claims many genes, typically as complementary DNA probes used on a microarray (see figure, p. 240, bottom).

www.sciencemag.org

239

Published by AAAS

POLICY FORUM
based" rather than "disease-based" Patent ownership by assignee characteristics A C 1000 (N = 4,270 patents) Ownership fragmentation by gene patenting. Other 1% U n s u r p r i s i n g ly, other heavily Private 63% Japan 4% patented genes tended to have releEurope 6% 750 Patents per gene Canada 2% vance to human health and diseases: 1 Unclassified 9% Unclassified 9% e . g . , BRCA1 (breast cancer), 2­5 US 78% Public 28% PIK3R5 (diabetes), and LEPR (obe>5 500 s i t y ) . Of the 291 cancer genes Gene "ownership" by assignee characteristics B reviewed by Futreal et. al. (29), 131 (N = 23,688 genes) a r e patented--significantly more Other <1% Unclassified 2% 250 US 15% than expected for a random sample Public 3% Unclassified 2% ­32 based on binoof genes (P = 1.2 Private 14% Japan <1% mial distribution). Moreover, these Europe 1% 0 1000 2000 3000 4000 5000 6000 7000 Canada <1% g e n e s contain a higher number of Unpatented 82% N [Herfindahl index] Unpatented 82% patents per gene than expected by ­11 b a s e d on a chichance (P = 9.4 Patent and gene ownership characteristics. (A) and (B) Distribution of gene patent assignees "public" squared test) (30). includes governments, schools, universities, research institutions, and hospitals. (A) Ownership breakdown for Of the 4000+ patented genes, at the 4270 human gene patents. Fractional ownership is based on the number of assignees on a single patent or l e a s t 3000 have only a single IP the number of patents on a gene. (B) "Ownership" breakdown of the genes in the human genome. (C) The fragrights holder. For the remainder, we mentation of gene ownership by the Herfindahl index, rounded to the nearest 1000. (31). (The 3002 genes examined whether IP ownership was with an index of 10,000 are not shown; those for 8000 to 9000 would not be visible on the graph.) The assignee fragmented by constructing a meas- names were used as listed on the patents by the European Patent Office. As such, the Herfindahl indices are ure based on the Herfindahl index likely to overestimate the "true" fragmentation because they do not reflect assignee name changes, mergers, (31) (see figure, top right; part C). acquisitions, splits, partnerships, or other events that usually lead to a consolidation of IP rights. The two genes with the most fragmented ownership were PSEN2, the amy- rapidly growing IP surrounding non­pro- 12. M. Stott, J. Valentine, Nat. Rev. Drug Discov. 3, 364 ( l o i d precursor protein (8 assignees for 9 t e i n coding components of the human 13. J.2J004)l.l, Science 280, 689 (1998). . Do patents), and BRCA1, the early onset breast genome, such as microRNAs, ribozymes, 14. J. Gans, S. Stern, J. Econ. Manage. Strat. 9, 485 (2000). 15. R . L e v i n , A . K . K l e v o r i c k , R . R . N e l s o n , S . G . Wi n t e r, cancer gene (12 assignees for 14 patents). and cis-regulatory elements. Brookings Pap. Econ. Act. 3, 783 (1987). Such fragmentation raises the possibility 16. R. Eisenberg, C. R. Biol. 326, 1115 (2003). t h a t innovators may incur considerable References and Notes 17. M. Stott, J. Valentine, Nat. Biotechnol. 21, 729 (2003). 1. E. F. Keller, The Century of the Gene (Harvard Univ. 18. J. P. Walsh, A. Arora, W. M. Cohen, Science 299, 1021 costs securing access to genes via structur(2003). Press, Cambridge, MA, 2000). ing complex licensing agreements. 19. G. Xu, A. Webster, E. Doran, World Patent Inform. 24, 95 2. We use the term "gene" to mean a set of cotranscribed (2002). O u r analysis suggests a number of protein-encoding exons. Here, "gene patent" means a 20. G. Dufresne, M. Duval, Nat. Biotechnol. 22, 231 (2004). p a t e n t disclosing and claiming a human gene avenues for further research: It would be 21. Georgetown University Kennedy Institute of Ethics, sequence or some fraction thereof. valuable to examine whether current pracD N A patent database (http://dnapatents.george3. D. Crease, G. Sc hlic h, Nat. Rev. Drug Discov. 2, 407 town.edu/). tice in patent examination has allowed mul(2003). 22. National Genome Information Center, Patome data4. T. Hollon, Nat. Med. 6, 362 (2000). tiple conflicting patents on the same gene. base (http://bach.ngic.re.kr:8000/patent/en/). 5. S. Scotchmer, J. Econ. Perspect. 5, 29 (1991). I n addition, genes with multiple patents 23. G. Dufresne, L. Takacs, H. C . Heus, J.-J. Codani, M. Duval, 6. M. A. Heller, R. S. Eisenberg, Science 280, 698 (1998). Nat. Biotechnol. 20, 1269 (2002). and IP owners provide a valuable context in 7. U. S. Patent and Trademark Office, Fed. Regist. 66 (4), 24. S. Altschul, G. Warren, W. Miller, E. Myers, D. Lipman, J. 1092 (January 2001). wh i c h to explore the variety of arrangeMol. Biol. 215, 403 (1990). 8. Nuffield Council on Bioethics, "The ethics of patenting ments used to facilitate or block access to 25. K. D. Pruitt, T. Tatusova, D. R. Maglott, Nucleic Acids Res. D N A" ( Te c h . r e p . , N u f f i e l d Council on Bioethics, 33 (database issue), 501 (2005). g e n e - b a s e d research and the impact of London, UK, 2002). 26. D. Maglott, J. Ostell, K. D. Pruitt, T. Tatusova, Nucleic 9. T. Caulfield, E. Gold, M. Cho, Nat. Rev. Genet. 1, 227 these arrangements on future innovators. Acids Res. 33 (database issue), 54 (2005). (2000). 27. Materials and methods are available as supporting Fi n a l ly, whereas our study includes only 10. L. Andrews, Nat. Rev. Genet. 3, 803 (2002) m eria on Science O e. protein-coding genes, future studies should 11. S. M. Thomas, M. M. Hopkins, M. Brady, Nat. Biotechnol. 28. Watuse lthe definition nlinpatent family recommended e of c h a r a c t e r i z e the nature and extent of the 20, 1185 (2002). by the International Patent Documentation Center:
2800 Number of genes with Herfindahl index = N

Genes claimed by N patents

Genes by number of times patented Patents claiming N genes

3829 350 300

Patents by number of genes claimed

29. 30.

2400 2000 1600 1200 800 400 0 55 38 17 13 9 4 1 1 2 2 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 20

31. 250 200 150 32. 100 50 0 1 2 3 4 5 6 7 8 9 10+

Any two patents linked directly or indirectly by a priority document are in the same family. P. A. Futreal et al., Nat. Rev. Cancer 4, 177 (2004). We found similar patenting rates for the 1456 genes listed in the Online Mendelian Inheritance in Man (32) w i t h a well-characterized association to disease phenotypes (517 of 1456, P value = 1.6­67). The Herfindahl index is the sum of the squares of the patent shares (in percentage terms) of each patent assignee (range 0 to 10,000), where 10,000 represents a "monopolist" with 100% of patents owned by one assignee and low numbers representing more fragmentation. O n l i n e Mendelian Inheritance in Man, O M I M , M c Ku s i c k - N a t h a n s Institute for Genetic Medicine, J o h n s Hopkins University (Baltimore, M D ) and N a t i o n a l Center for Biotechnology Information, National Library of Medicine (Bethesda, MD), 2000.

N [patents]

N [genes]

Supporting Online Material www.sciencemag .org/cgi/content/full/310/5746/240 DC1 10.1126/science.1120014

Global characteristics of the patent map. (Left) Distribution of genes by the number of times they are patented. (Right) Distribution of patents by the number of unique genes they claim.

240

14 OCTOBER 2005

VOL 310

SCIENCE

www.sciencemag.org

Published by AAAS