# FromBilskiBack toBenson: Preemption, Inventing Around, and the Case of Genetic Diagnostics **Authors:** Rochelle C. Dreyfuss, James P. Evans **Citation:** "From Bilski Back to Benson: Preemption, Inventing Around, and the Case of Genetic Diagnostics," 63 *Stan. L. Rev.* 1349 (2011) (with James P. Evans) **Abstract:** The long-anticipated decision in Bilski v. Kappos was supposed to end uncertainty regarding the patentability of process claims (or, at the least, business method claims). Instead, the opinion featured a series of anomalies: The Court emphasized strict construction of the Patent Act, but acknowledged three judgemade exceptions to patentability. It disapproved State Street, the Federal Circuit case that had upheld business method patents, but could muster only four votes for the proposition that business methods are in fact unpatentable. But even though the Court upheld business method patents, it invalidated all of Bilski's hedging claims. And while the Justices agreed on one thing--a patent that "preempts" something (a mathematical formula, an approach, a commonly used idea, a wide swath of technological developments, the public's access) is bad--they failed to operationalize the concept. That problem had plagued the law prior to State Street; in the interest of preventing the same set of problems from recurring, this Article uses recent empirical studies on gene patents to tease out indicia ("clues") to supplement the machine-or-transformation test for determining when a claim is preemptive and therefore invalid. Chief among these clues is the inability to invent around claims that cover broad prospects. **Source:** http://www.stanfordlawreview.org/wp-content/uploads/sites/3/2011/06/Dreyfuss-Evans-63-Stan-L-Rev-1349.pdf ## INTRODUCTION *p. 2* This was supposed to be the end of an era. After prolonged uncertainty regarding the patentability of claims drawn to business methods, Bilski v. Kappos 1 was expected to provide guidance on when they constituted patentable subject matter. But while the Court explicitly laid to rest both the Federal Circuit's broad approach in State Street Bank & Trust Co. v. Signature Financial Group 2 and its narrow approach in In re Bilski, 3 the Justices otherwise provided little information on how to determine whether particular subject matter is statutory. In a fractured set of decisions, the Court appeared to do no more than state the obvious. The Patent Act should be read broadly, but "laws of nature, physical phenomena, and abstract ideas" are not within the ambit of protection. 4 The Justices' opinions featured a series of anomalies. The majority insisted on strict construction of the statute. 5 However, the three exceptions the majority recognized-laws of nature, physical phenomena, and abstract ideas-had all been imposed judicially. State Street, the case approving business method patents, was deemed bad law, but it was impossible to attract five votes for the proposition that business methods are not patentable. 6 And even though business methods are, apparently, patentable, the hedging claims at issue in Bilski were all held invalid-not just the broadest claim, but even the narrow ones. 7 The Court held that the Federal Circuit's "machine-or-transformation" testunder which inventions are unpatentable unless they are tied to a machine, or they transform an article into a different state or thing-is a mere "clue" to patentability; 8 but the Court never indicated how that clue should be used. It is clear that a claim that fails the test is not necessarily invalid, but it remains uncertain whether a claim that passes the test is necessarily valid. Nor June 2011] PREEMPTION AND GENETIC DIAGNOSTICS 1351 effect" test, 9 a "technological arts" doctrine, 10 greater attention to the usefulness of the art, 11 or a return to the "mental steps" doctrine. 12 But aside from references to the "technological arts" in Justice Stevens's concurrence, none of these approaches were discussed. As Justice Stevens put it, "The Court . . . never provides a satisfying account of what constitutes an unpatentable abstract idea." 13 In fact, the Court seemingly went out of its way to say nothing. Justice Kennedy's plurality opinion emphasized that "the Court today is not commenting on the patentability of any particular invention." 14 The Justices did, however, agree on one thing: a patent that "preempts" something (e.g., a mathematical formula, an approach, a commonly used idea, a wide swath of technological developments, the public's access) is very bad indeed. "Preempt" is used in each of the Bilski opinions. 15 Convergence on the term could provide an important hint to the Court's concerns-if, that is, the term had a meaning within scientific and technological discourse. Instead, its use is entirely within the legal domain, where it most often describes the displacement of one law (such as state law) by another (federal law); 16 in an earlier time, it was also used to describe various technical issues arising in claim drafting and prosecution. 17 Justice Douglas elevated the concept to center stage 9. See Brief of TELES AG as Amicus Curiae in Support of Neither Party at 17-21, Bilski, 130 S. Ct. 3218 (No. 08-964) (arguing for harmonization with European law, which requires that the claimed subject matter contain "technical elements"). *p. 3* 10. See Brief Amicus Curiae of International Business Machines Corp. in Support of Neither Party at 7-8, 10-11, Bilski, 130 S. Ct. 3218 (No. 08-964) (arguing that the test should examine whether the claimed process makes a "technological contribution"). The "technological arts" test was also favored by the original patent examiner in Bilski, see 130 S. Ct. at 3233 (Stevens, J., concurring in the judgment), and is considered by "[n]umerous scholars" to reflect the original meaning of the term "useful arts," see id. at 3244. 11. See Brief of Regulatory Datacorp, Inc. et al. as Amici Curiae in Support of Neither Party, Bilski, 130 S. Ct. 3218 (No. 08-964). *p. 3* 12. Brief of Amicus Curiae Law Professor Kevin Emerson Collins in Support of Neither Party, Bilski, 130 S. Ct. 3218 (No. 08-964). *p. 3* 13. Bilski, 130 S. Ct. at 3236 (Stevens, J., concurring in the judgment). 14. Id. at 3228 (plurality opinion). 15. See id. at 3230-31 (majority opinion) (quoting the "nutshell" summary from Gottschalk v. Benson, 409 U.S. 63, 72 (1972)); id. at 3253 (Stevens, J., concurring in the judgment); id. at 3258 (Breyer, J., concurring in the judgment). *p. 3* 16. See, e.g., Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 154 (1989). 17. See, e.g., Pangborn Corp. v. Am. Foundry Equip. Co., 159 F.2d 88, 99 (3d Cir. 1946) ("Peik could not preempt Rosenberger and Keefer's form of wheel as claimed in the counts of the interference."); In re Collins, 75 F.2d 1000, 1002 (C.C.P.A. 1935) ("Appellant, under the circumstances of this case, cannot pre-empt the entire field of useful inventions such as were claimed by Pieper by broadly teaching that useful results may be obtained by mixtures and combinations of a broad general group of materials without specifically naming such materials . . . ."). in Gottschalk v. Benson when the issue of protecting computer programs first reached the Supreme Court. 18 Since then, it has caused endless confusion. 19 Nonetheless, we are apparently now back to Benson; and with the return of preemption, it is time to operationalize the concept. Part I briefly recounts the Supreme Court's attempts to define patentable subject matter, with the aim of identifying the concerns that led the Bilski Court to invoke the language of preemption. It concludes that the real question is not whether an advance is in a field where patenting is appropriate, but how claims are drafted. Claims that "preempt" competitive development-that cover prospects that cannot be efficiently mined by individual right holders-are barred. Part II moves on to consider, as a case study, the field of genetic diagnostics. This is an area particularly ripe for attention. Justice Breyer's dissent from the dismissal of certiorari in Laboratory Corp. of America Holdings v. Metabolite Laboratories, Inc., 20 which ignited the debate over the patents issuing under State Street, was a medical diagnostics case. Considerable empirical work on the effect of patenting has been done in this area; 21 there are cases waiting in the wings, 22 and promising medical and scientific advances are on the horizon. 23 The case study suggests that, at its core, the preemption problem arises when an advance cannot be invented around. When such advances cover broad prospects, patenting would, as Justice Breyer suggested in Metabolite, "impede rather than 'promote . . . [p]rogress.'" 24 Part III concludes with thoughts about other indicia for determining when a claim is preempted. It is not especially surprising that this has been an era of uncertainty in patent law. As new technological opportunities emerge, it is inevitable that there will be questions about how the law applies. It happened when the power of steam was first exploited, 25 when the effects of oxygen were discovered, 26 when it became possible to manipulate electric current, 27 and when differential solubility was understood. 28 The recent rapid development of new sciencessuch as molecular biology, genomics, electrical engineering, 29 and information and communication technology 30 -creates many fresh challenges. In theory, each technology raises two categories of questions. The first is "whether"whether existing patent law is appropriate to the new field, or a different (or entirely novel) intellectual property system is necessary. The second is "how"how the requirements of the system should be applied to the new technology. ## A. The "Whether" Inquiry *p. 5* One might have thought that the "whether" inquiry would be labeled the "statutory subject matter question," and the "how" inquiry would be conceived of as addressing issues on the interpretation and application of other provisions of patent law. 31 And indeed, lawmakers have ostensibly followed that approach. Thus, advances in computer science initially raised the question whether software should be considered a literary work for copyright purposes. A national commission was appointed. After it answered in the affirmative, copyright law was amended to deal with foreseeable problems. 32 As it became clear that copyright protection for software would be highly limited, 33 the action moved to patenting-leading to a large number of cases on whether software fit within that realm. 34 Similarly, in State Street, the Federal Circuit was confronted with the question whether business methods are patentable. Taking its cue, perhaps, from Diamond v. Chakrabarty's hospitality to patenting in emerging technologies, 35 the court answered with a broad holding: anything that achieves "a useful, concrete and tangible result" is patentable. 36 Because State Street led to the patenting of highly diverse advances-from medical diagnostics to tax-minimization strategies 37 to methods for training janitors to dust 38 -the Supreme Court used the dismissal of certiorari in Metabolite to signal a need to reevaluate. In a series of cases, the Federal Circuit considered a variety of formulations. 39 Eventually, In re Bilski was taken en banc. 40 In that decision, the Federal Circuit narrowed the criteria for patent eligibility. Adopting a test that it thought derived from the Supreme Court's software cases, the court held that a process is statutory subject matter when it is tied to a machine or transforms materials to a different state or thing (the machine-or-transformation test). 41 The Supreme Court granted certiorari in Bilski, and at oral argument the Justices pursued the same analytical framework, searching for categorical limits to patentable subject matter. Thus, the Justices asked questions about whether specific enterprises-speed dating, training horses, and teaching students- It did not, however, indicate how the clue should be used. The main limit the Court identified was an old one-that "laws of nature, physical phenomena, and abstract ideas" are not protectable. 45 In addition, the Court resurrected the preemption trope developed in Benson, the first case on the patentability of a computer method. But while every Justice who wrote an opinion used the term "preempt," no one explained what it meant. 46 Instead, the Court appeared to rely on the "nutshell" with which Justice Douglas summed up Benson: *p. 7* The mathematical formula involved here has no substantial practical application except in connection with a digital computer, which means that if the judgment below is affirmed, the patent would wholly pre-empt the mathematical formula and in practical effect would be a patent on the algorithm itself. 47 After Bilski, the question is thus what the Court means by "preemption." The law subsequent to Benson had not been a model of clarity. In part, the problem was that Justice Douglas not only adopted the term "preemption," he also deployed another new concept, the "algorithm." Although he defined it as "[a] procedure for solving a given type of mathematical problem," 48 it was a word unfamiliar to patent law, and the definition itself left much to be desired. The reference to a mathematical problem was misleading: in common parlance, any set of steps to solve a problem is an algorithm. Justice Douglas may have been trying to get at the notion that scientific truths are unpatentable, 49 Unfortunately, it is equally difficult to decide whether an algorithm-once identified-is "preempted." After Benson, drafters had a field day. Some attempted to waive rights over particular uses of the algorithm, hoping that if some uses (such as for academic research) were left in the public domain, the claim would not be considered preemptive. 51 Others confined their claims to specific fields (much as Bilski did). Some drafters added data-gathering steps or postsolution activities. 52 Alternatively, they embedded the algorithm in a traditional industrial process where patenting was common 53 or claimed the machine that implements the steps of the algorithm. 54 Some of these attempts were successful; others were not. 55 Prior to State Street, the courts and the Patent and Trademark Office (PTO) formulated a series of tests, each designed to create a procedure (ironically, an algorithm) for identifying advances that did not qualify for protection. 56 In retrospect, it is evident why it was so difficult to get a handle on preemption. In fact, the concept does not answer the question of whether a particular field is suitable for patenting. 57 Or as Judge Rader suggested in his dissent to the Federal Circuit's decision in In re Bilski, nothing explains "why . . . some categories of invention deserve no protection." 58 Justice Kennedy's analysis is no better. In deciding that Bilski's hedging claims were too abstract to patent, the Court did not discuss hedging as a category. Rather, what it determined was that the claims had not been framed in a way that was acceptable. In other words, the question the court answered was not the "whether" question, but the "how" question-how should claims in this field be drafted? *p. 8* Significantly, a close look at prior subject matter decisions reveals a similar pattern. There was probably very little doubt about whether the technology in many of the early cases-blast furnaces, electrical and chemical innovations- was patentable subject matter; the only questions in those cases were "how" questions. For example, the issues in O'Reilly v. Morse and the Telephone Cases were in reality about how to disclose and claim advances in the inventors' respective fields: Morse lost because he claimed all the ways of "printing intelligible characters . . . at any distance[]," but had not identified all of the ways; 59 Bell won because his claims were (allegedly) limited to the methods for "transmitting vocal . . . sounds telegraphically" that he described. 60 It is improbable that successive generations of Supreme Court Justices have overlooked the "whether" question; something else appears to be going on. Perhaps attempts to define patentable subject matter are doomed to failure because there is nothing categorical that can be said about the fruits of innovation, hence the rejection of "technical effect," "technological arts," and "useful arts" tests, and the Court's refusal, despite clear misgivings, to exclude business methods. In a sense, then, the real problem with Bilski isn't that it rejected State Street without providing a substitute; the real problem is that it approved the broad holding in State Street without acknowledging that it was doing so. As long as the nation is committed to using the patent system to spur creative development, perhaps the best the Supreme Court can do is keep all fields open to patenting-exactly what Judge Rich was trying to accomplish in State Street. ## B. The "How" Inquiry *p. 9* To put this another way, Judge Rader's question why some categories of invention deserve no protection cannot be answered by examining specific endeavors. Rather, certain claims do not deserve protection-and the way to understand what the Court means is by formulating a reason why. Judge Rader tried, saying: *p. 9* Natural laws and phenomena can never qualify for patent protection because they cannot be invented at all. After all, God or Allah or Jahveh or Vishnu or the Great Spirit provided these laws and phenomena as humanity's common heritage. . . . An abstract idea must be applied to (transformed into) a practical use before it qualifies for protection. 61 Invocation of a higher authority will not appeal to those with a more secular bent. But the passage is critical for two reasons. First, it recognizes that issues like abstractness cannot be answered in the abstract-one needs to understand the goal that the exception is there to further in order to apply the rule correctly. Second, Judge Rader provides important hints as to what that goal is: preventing inventors from claiming "before" something has been "transformed." *p. 9* The issue, in short, is timing. Patent claims cannot be made too early in the development of a field because there is a danger of preemption: exclusive rights may preempt others from competing and thereby diminish the vibrancy of the marketplace or the vigor of the creative environment. To use the language of Morse, early claiming can pose an obstacle to "the onward march of science" 62 (or business) and it does so by limiting the number of approaches, experiences, bodies of knowledge, and interests that can be brought to bear in mining the initial insight. 63 To be sure, other requirements for patentability can also be construed as aimed at timing. For example, Ariad Pharmaceuticals, Inc. v. Eli Lilly & Co., a case on exploiting the properties of the protein complex NF-κB, was decided on the ground that the applicant filed before it was able to supply a written description of the substances that could be used to achieve the claimed result. 64 The case could have been equally well argued along preemption lines-that knowledge about NF-κB, which appears to play a role in a wide variety of conditions, including memory loss and susceptibility to diseases such as cancer, 65 is not patentable because the patent would preempt those who would follow on, elucidate the impact of NF-κB, and find ways to utilize the information to treat patients. Rebecca Eisenberg and Robert Merges have similarly argued that the utility issue is aimed at delaying the onset of patent protection to the point where more is known about the invention. 66 Claim construction also plays a role in trimming the impact of a patent. Thus, Robert Merges and economist Richard Nelson have questioned the traditional broad protection accorded to pioneer inventions. 67 While they recognized that incentives can be important at the inception of a new field, and that centralizing control over an opportunity can improve planning and reduce wasteful duplication, their examination of a wide variety of fields led them to conclude that competitive development is the superior approach to promoting progress: ## June 2011] PREEMPTION AND GENETIC DIAGNOSTICS 1359 *p. 11* When a broad patent is granted . . . , its scope diminishes incentives for others to stay in the invention game, compared again with a patent whose claims are trimmed more closely to the inventor's actual results. . . . This would not be undesirable if the evidence indicated that control of subsequent developments by one party made subsequent inventive effort more effective. But the evidence, we think, points the other way. 68 But even though there are other doctrines that can be used to protect competitive development, a preemption doctrine is nonetheless critical. All the other requirements permit patents-they will simply be narrower than might otherwise be claimed, or delayed until a use is identified. Yet because patentsonce issued-cover all uses, there will be situations where even very narrow patents block off too much, especially in areas (like computer science and genetic diagnostics) where applications flow easily from basic discoveries. 69 Armed with that understanding, the anomalies in Bilski largely disappear. The Court did not outright invalidate business methods, but it was skeptical about them because concentrating a broad business opportunity in a single entity can distort the market and harm the economy. 70 The Court rejected State Street's "useful, concrete, and tangible result" test because insights into broad technological opportunities are clearly "useful" and some will have "concrete and tangible results." 71 Thus, that approach could lead to too much control over important prospects. In contrast, the machine-or-transformation test was accepted as a clue because once an insight is instantiated in a product or in a physical transformation, the claims are unlikely to have such a broad reach that they cut off lines of inquiry or limit competition. But since they can have that effect-for example, tying a process to a general-purpose computer may not reduce the reach of the process significantly-it is unlikely that the Court meant to convert the machine-or-transformation test into a safe harbor. *p. 11* It is also possible to understand the Court's attitude toward other attempts to limit patenting. Thus, the impulse behind the mental steps doctrine, which would bar patents on processes that can be accomplished in the mind, appears based on the idea that thinking usually comes "before" (in the Judge Rader sense) applying fundamental insights to concrete problems. The flaw in the reasoning is that not all thought is of that character; some thoughts are highly complex and so focused on a particular problem that they can be privatized Claim limitations are similarly ambiguous. For example, attempts like Bilski's to limit claims to particular sectors will not work as a general rule because competitive development can be important even within a specified field. The genetics case study in Part II furnishes an example. However, if the true concern is with fostering multiple pathways to development, waiving rights over research uses is closer to the mark. When the cases on waiver were decided, there was a broad common law research exemption; waivers thus contributed little to innovation. Now that the Federal Circuit has largely eliminated the exemption, 73 a waiver is significant. But it is probably not significant enough to furnish a clue to patentability. Because patentees and followers are unlikely to agree on what was relinquished, a waiver could foster litigation, dampen innovation, and impair business. *p. 12* References to data gathering or postsolution activity may, in contrast, furnish somewhat better clues as there may be some activities that are specific enough to limit the reach of a broad claim. But their use must be handled carefully because, as Bilski implicitly recognized, 74 these activities may be too generic to release a prospect for general use. For example, because all diagnostics start by drawing blood and end with associating a variable to disease, the steps in Metabolite-"assaying a body fluid" and "correlating" (or, alternatively, "diagnosing")-are not limiting. 75 In contrast, the method for optimizing therapeutic efficiency at issue in Prometheus Laboratories, Inc. v. Mayo Collaborative Services, which requires the administration of a drug prior to assaying and diagnosing, is arguably a sufficiently specific limitation. 76 Significantly, cases like Metabolite and Prometheus can be distinguished in another way. In Metabolite, there is no way around the claim; the practice of medicine is all about the activities at issue there-examining patients and interpreting findings in light of "associated" symptoms. The "limiting steps" thus do nothing to reduce the power of the claim to prevent scientists or physicians from understanding biology or treating patients; in Prometheus, however, there are arguably other ways to achieve the goals of the patent. In short, a better way 72. Cf PREEMPTION AND GENETIC DIAGNOSTICS 1361 to grapple with preemption may be to ask whether the claim can be practiced in other ways-or as patent lawyers say, "invented around." 77 Judge Rich condemned the approach early on as "an essentially illogical distinction unwarranted by, and at odds with, the basic purposes of the patent system." 78 Nonetheless, there is much to recommend "inventing around" as a clue to patentability. The ability to work around the patented method limits the patent holder's grip-it sets a cap on the price that can be charged and makes it possible for others to mine a prospect even when the patentee refuses to do so. 79 Furthermore, since science must deal with the natural world, the inability to invent around is also a clue to Bilski's other exclusions: laws of nature and natural phenomena. Indeed, as Part II demonstrates, the inability to invent around may be the best evidence of what the Court means by "preemption." *p. 13* II. GENETIC DIAGNOSTICS Some have complained that Bilski provides so little guidance, a Bilski defense will be raised in every case. 80 The above suggests, however, that the Court's concerns are fairly specific: the opinion is aimed at fostering business by protecting the competitive environment and at promoting innovation by assuring public access to broad technological prospects. To be sure, identifying the claims that pose a danger to business or innovation will not always be easy, but once courts start looking at the problem in a more directed way, they will surely develop a better grasp of the issue. 81 Further, they may consider other strategies, such as new defenses to infringement or denial of injunctive relief, 82 to protect business and innovation without sacrificing incentives to invent. While it is likely that the way forward will be somewhat sector dependent, a study of issues arising in genetic diagnostics is illuminating. This is a field where the impact of patenting genes and diagnostics has been examined, so it is possible to see how exclusivity affects "the onward march." Furthermore, this area is of interest in its own right; there are pending cases and the scientific promise is considerable. ## A. The Science of Genetics *p. 14* The field of genetics is concerned with the storage, expression, and transmission of biological information. In the most general terms, genetics seeks to explain why children look like their parents but also why they are unique. Genetics illuminates both the diversity of life on earth and its unity. On the one hand, the differences between the genome of a chimp and that of a human are responsible for our unique attributes as distinct species; on the other hand, genetics demonstrates that we share a common genetic code (and indeed the molecular details of life) with earthworms, gazelles, and bark beetles. 84 At the root of genetics is deoxyribonucleic acid (DNA). DNA is a long chain of nucleotides-chemical subunits. The DNA chain that encodes the instructions for a human being consists of about three billion nucleotides and is about six feet long. A single copy of this chain, intricately folded upon itself, resides in each of the approximately 10 14 (one hundred million million) cells in the human body. DNA has only two jobs in the living organism. It serves as a store of information, and it instructs the cell how to synthesize proteins (which execute the work necessary for living cells) and RNA (which carries information and possesses regulatory functions). DNA performs both of these tasks by encoding information via a digital code, represented as the order of the individual nucleotides in a given stretch of the DNA chain. *p. 14* If one travels along the three-billion-link chain that is the human genome, one encounters particular linear stretches of DNA, usually extending several thousand nucleotides, that encode the instructions for making a particular protein. After traveling along the entire length of the chain one would have encountered about twenty-five thousand intervals in which a distinct protein or RNA molecule is encoded. That is, one would have encountered approximately twenty-five thousand human genes. Between the "coding" intervals of DNA, each specifying a unique RNA or protein molecule, reside "noncoding" regions. Some of this noncoding DNA is regulatory in nature, directing the cell as to which genes to activate and which to leave dormant. For example, a white blood cell does not need to activate or "transcribe" a gene that encodes the instructions for skin pigment, so it leaves that gene in a dormant state. Much of the noncoding DNA appears to simply be "junk" left over from evolution, un- Noncoding DNA does not exist only between genes; the vast majority of genes themselves are interrupted by stretches of noncoding DNA. Within a given gene (that is, a stretch of DNA that encodes a particular RNA or protein molecule), there thus exist "introns" (interrupting noncoding DNA) and "exons" (expressed segments of DNA). It is primarily in the exons that the meaningful information for directing the synthesis of a protein or RNA molecule resides. Indeed, the introns can typically be readily removed from a gene without materially altering its informational content. This is the difference between a socalled cDNA version of a gene and a "genomic" copy of a gene. The cDNA version has simply had the unnecessary interrupting segments (introns) snipped out, and therefore includes only the coding regions (exons). Thus, it is shorter; it is also easier to manipulate. It is worth pointing out that the cell machinery splices out the introns on a routine basis just as the process of snipping out introns occurs in nature every time a cell expresses a given gene. 85 The way in which a linear stretch of DNA specifies the synthesis of a protein or RNA molecule is though the unique order in which the nucleotides are arranged along that stretch of DNA. Within any given gene, each successive triplet of three nucleotides specifies a particular amino acid (of which there are twenty), which together form the building blocks of proteins. Thus a uniquely ordered stretch of nucleotides along the DNA molecule specifies (encodes) a unique order of amino acids and thus a unique protein. The only difference between a stretch of DNA that directs the synthesis of a skin pigment protein (i.e., a "pigment gene") and a stretch of DNA that encodes a globin chain (i.e., a "globin gene") that carries oxygen in the blood, is the particular order of the nucleotides that make up that stretch of DNA. *p. 15* Elucidation of the double-helical structure of DNA by Watson and Crick (based on data unwittingly provided by Rosalind Franklin) in 1953 immediately illuminated the way in which DNA serves as a store of information and is transmitted from generation to generation. 86 In essence, by consisting of two chains, the double helix contains a copy (or more precisely, a mirror image) of itself. Each of the two chains in the double helix is made up of three billion nucleotides arranged in a particular order. There are only four types of nucleotides, designated as A, T, G, and C. Thus, one short stretch of a nucleotide chain might read: AATGGCTCGGAT and so on. The two chains of the double helix are held together by the fact that A binds specifically to T and G binds specifically to C. So if one chain has an A at a particular site along its sequence, we know that the other chain will have a T in the corresponding posi- tion. Likewise, if a G is located in a particular site, the other chain will have a C at that corresponding site. When a cell divides, it sends one chain to one daughter cell and the "complementary chain" to the other daughter cell. The cells can then easily reconstruct the double helix by building the other, complementary, chain based on this base pairing. Modern molecular techniques have been developed that allow for the "isolation" of any given gene. This simply means purifying (or using enzymes to construct in a test tube) a particular stretch of DNA that represents a given gene. For example, the gene for insulin is 1430 nucleotides long and the part that actually encodes the insulin protein is 153 nucleotides. There are two introns which are removed ("snipped out" or "spliced") in the process of expressing the gene. When the human insulin gene is isolated, investigators may also snip out these introns to produce a cDNA version as above. *p. 16* A mutation is simply an error in the DNA sequence that disrupts the ability of the gene to encode a functioning protein. A mutation may consist of a single missing nucleotide, deletion of several (or several thousand) nucleotides, an insertion of a single or many nucleotides, or the substitution of a nucleotide (e.g., a T where there should be a C). A mutation in a gene disrupts the ability of that gene to encode a protein and may result in disease. For example, a mutation in the human CFTR gene causes cystic fibrosis. The field of DNA diagnostics hinges on assaying a gene for its sequence integrity. The most common and generally most effective means of assaying a gene is to "sequence" it-that is, to determine the precise order of the nucleotides that comprise that given gene in an individual. If a patient with a family history of early-onset breast cancer has a mutation in the BRCA1 gene (for example, an extra nucleotide, a missing nucleotide, or the wrong nucleotide at a given position), the gene cannot regulate cell growth, and the patient has increased susceptibility to breast cancer. Geneticists refer to an individual's underlying genetic sequence as her "genotype." The "phenotype" of an individual refers to the ultimate effect of that genotype on the individual organism. Thus, an individual's genotype may indicate that she carries a mutation in the BRCA1 gene. The resulting phenotype of that individual is her marked propensity towards breast cancer at a young age. *p. 16* Geneticists have discovered many disease-gene linkages in which mutations in a given gene are responsible for a given disease. This involves studying the relationship between phenotypes and genotypes. Typically, this process involves looking at families or a large number of individuals with a given disease and sequencing the patients' genes to detect mutations that track with (in technical jargon, are "linked to") the presence of the disease. For such purposes large numbers of individuals with the disease in question are necessary. Alternatively, large families with a predisposition to the disease may be assayed. While in classic genetic diseases the relationship between harboring a mutation and developing the corresponding disease is very strong (e.g., 100% of people with a mutation in the Huntington gene eventually develop Huntington dis-ease), geneticists are now learning about many weaker associations that predispose an individual to develop the corresponding disease. *p. 17* Finally, geneticists are learning that individuals differ in their nucleotide sequence at many sites throughout the genome and many (if not most) of these differences are unimportant to their health. Thus, any one individual will, on average, differ from his neighbor at over one million sites in his genome. For example, at a particular site one individual may have a G whereas others will have an A. These subtle differences among individuals are called "single nucleotide polymorphisms" (SNPs). Sometimes, they lead to small changes among individuals, like eye color, while at other times, such polymorphisms can be associated with disease-or may have no impact at all on one's phenotype. Sorting out innocuous polymorphisms with no health implications from those which have health effects is a major challenge for the future of genomic medicine and will be a difficult task, requiring the pooling of sequence information and health information from many individuals. *p. 17* Because genetic research requires broad access to both phenotype and genotype information, researchers have a strong commitment to putting sequencing data into publically available databases. 87 This commitment does not, however, mean that genetic information cannot be involved in patents. In fact, about twenty percent of the genes in the human genome are associated with patents. 88 Some patents claim products covering "purified DNA"-essentially, cDNA sequences comprising specific genes or mutations. Others claim processes, such as for detecting a specific sequence or for using the sequence to diagnose a predisposition to disease. ## B. The Effect of Patenting *p. 17* Because patenting behavior in this field has been highly variable, it is possible to conduct a natural experiment on the effects of patents both on the "business" (i.e., practice) of medicine and on innovation in medical science. In a series of eight case studies conducted at the behest of the Health and Human Services Secretary's Advisory Committee on Genetics, Health, and Society (SACGHS), Robert Cook-Deegan and his associates examined ten clinical conditions involving heritable disorders for which genetic tests were available. 89 Some of the conditions were associated with patents, some not; some patents 87. See (1) breast/ovarian cancer, (2) colon cancer, (3) hearing loss, (4) cystic fibrosis, (5) inherited susceptibility to Alzheimer disease, (6) hereditary hemochromatosis (HH), (7) spinocerebellar ataxias (SCA), (8) long QT syndrome (LQTS), (9) Canavan disease, and (10) Tay-Sachs disease. Id. were widely licensed, others not; some of the conditions studied had high prevalence in the population, others afflicted small groups. In each case, the associations were known for at least ten years-long enough for the use of the diagnostics to be well established within the medical community and for the effects of patenting to become evident. 90 By comparing the experiences under a variety of patenting and licensing strategies, the investigators isolated and quantified the effects of patents on the development of gene diagnostics and on their availability to patients. 91 The results demonstrate how upstream patents can impact downstream activities. In cases where there was broad access (either because there were no patents or the patents were widely licensed), there were many laboratories conducting diagnostic tests, spanning the spectrum from academic labs to industry. 92 In settings where numerous labs offered genetic diagnostic tests for the same condition, these laboratories competed on the basis of quality, price, innovation, and the specific nature of the test employed (e.g., sequencing versus looking only for specific mutations). In contrast, when there were patents held exclusively by a single entity, both clinical practice and scientific development were impaired. 93 On the practice end, exclusivity for a given genetic test was associated with a number of harms. Patent holders (who, significantly, were never the first to market in any of the case studies), sometimes cleared the field once their patents issued. 94 Doctors and patients could no longer obtain second opinions on tests that can carry considerable medical implications (such as a recommendation for major surgery or lifelong surveillance). In addition, laboratorians expressed concern about the quality of genetic diagnostic tests. When only a single lab offers a given test it is impossible to apply the "gold standard" of quality assurance-proficiency testing-which requires analysis of the same sample by more than one provider. 95 In some cases, tests deemed necessary for patient care were simply not available. For example, patent holders did not always develop tests needed by a segment of the population deemed insufficiently large, but nonetheless enforced the patent against academic labs that routinely cater to such small populations. 96 Some providers failed to offer prenatal screening. 97 providers (such as state Medicaid offices), poor patients were denied access to testing. Charity testing programs, which are difficult to use, were generally insufficient to make up for the insurance shortfall. 98 Finally, in at least one example, a test for a life-threatening cardiac condition (long QT syndrome) was practically unavailable for eighteen months when the exclusive rights holder failed to either offer the test clinically or license it so that another lab could perform it. 99 The SACGHS report and its underlying case studies focused more on health delivery questions than on innovation concerns. Still, the report pointed out several potential impacts on research. Because many clinically identical diseases can result from mutations in widely disparate genes, fears were articulated that patent thickets and holdouts could obstruct the development of new diagnostic methodologies, such as multiplex testing (testing multiple genes simultaneously) and new therapeutic techniques. For example, while sequencing an individual's whole genome will soon be a practical reality, it may be a legal impossibility given the number of patents that would be infringed in one fell swoop. 100 Other evidence on the effect of patents on basic research tends to support these fears. Proponents of patenting cite the work of Wesley Cohen and coauthors, who conducted surveys of scientists in a variety of fields. Their work suggests that research is unimpeded by patents, largely because scientists tend to ignore them. 101 However, the Cohen studies have limited application to diagnostics. Whereas researchers were rarely sued in most of the fields Cohen studied, the case studies (and Cohen) found that geneticists do receive threatening letters. 102 Furthermore, although Cohen's interviewees did not think patents were hindering research, they mentioned other impediments, such as withholding research results or key physical materials, 103 which may be linked to patenting in ways the interviewees did not appreciate. Significantly, a type of this "self-help exclusivity" is also occurring in the diagnostics realm, where it takes the form of failing to deposit new mutations and human variants in public data- bases. 104 That activity is particularly damaging in that any understanding of gene function and the role of a given gene in health and disease has an absolute dependency on broad sharing within the scientific community. Finally, there is other empirical work suggesting that patents do impede research. Using event studies, Fiona Murray and her coauthors showed that patenting is associated with a decline in research and a decrease in the number of lines of research pursued. 105 Similarly, Heidi Williams found that product development involving genes subject to exclusive rights lags behind the development of genes in the public domain. 106 In any event, it is hardly a ringing endorsement of the patent system that its functioning depends on its being ignored. *p. 20* It may seem surprising that the downstream impact of gene patents is so profound. As Judge Markey, the first Chief Judge of the Federal Circuit, took pains to stress, patents are rarely true monopolies; usually alternative ways exist to achieve a result similar to the one for which the patented invention is utilized. 107 Richard Epstein has applied that idea to the genetics landscape, arguing that it is possible to sidestep the use of a patented gene by relying on another gene involved in the same condition. 108 Genetics is, however, hostage to biology. Genes evolved over millions of years to serve a specific biological purpose; that is why disruptions by mutation result in disease. These evolved genes are unique and the key value in isolation and purification is to produce the identical sequences to the genes found in nature. *p. 20* Thus, there is no possibility of sidestepping. While many genetic conditions demonstrate the phenomenon of "genetic heterogeneity," in which a mutation in one of any number of different genes can result in a clinically identical disease, the mutations are not substitutes for each other. A prominent example is that of BRCA1, BRCA2, and p53. Each of these genes is associated with early-onset breast cancer. But each gene has a distinct function. When a patient's family exhibits characteristics of hereditary breast cancer, it is necessary to assay all these genes since a derangement of any one of them can cause the phenotype of breast cancer. It is not possible to bypass BRCA1 and BRCA2, which are patented, and assay only for mutations of p53, which is not. Doing so would be medical malpractice because the testing would fail to detect BRCA1 and BRCA2 mutations, which account for more than two-thirds of early-onset breast cancers. It is thus necessary for clinicians to deal with Myriad Genetics, one of the firms identified in the SACGHS report as raising barriers to patient access to breast cancer diagnoses, 109 and which is also failing to deposit new mutations in the public database. 110 Arguably, there are other ways to invent around patented genes in order to identify hereditary conditions. However, for most genetic conditions, none can supplant the direct analysis of genes. Thus, some have suggested exploiting the phenomenon of linkage disequilibrium (LD), an association between a specific mutation (e.g., in a disease gene) and DNA sequence variants that reside some distance from the mutation in question and can thus act as "markers" for the presence of the disease gene. 111 In theory, LD allows one to evade the constraints of a patent by assaying for the marker that is "linked" to the patented gene. But while it is indeed true that sometimes such associations exist, there are two fundamental-and biologically insurmountable-problems with using such a strategy in the real-world diagnostic arena. First, the linkage between a marker and a gene is always imperfect: as the distance between the linked marker and the mutation of interest grows, LD testing becomes increasingly imprecise. A testing strategy with a high (and known) error rate is wholly inadequate for diagnostic purposes. Second, even if one were (perversely) satisfied with a laboratory test guaranteed to give wrong answers for a subset of patients, the biology of the situation makes it impractical for the majority of genetic tests: the linkage between a disease-causing mutation and a marker depends on historical and genetic contingencies that are only sometimes met. It is a distinct minority of genetic conditions that even demonstrate consistent linkage between a common mutation and a marker. *p. 21* Finally, it has been argued that one could circumvent a gene patent by instead analyzing the protein the gene produces ("expresses"). 112 For a number of reasons, this is an entirely infeasible alternative to genetic testing in the vast majority of situations. Proteins are often only expressed in specific tissues at specific times. For example, many genes are expressed only in the brain or only for a short period of time during fetal development. In contrast, the DNA of any of the body's cells reflects the mutational status of all others. Thus, to detect a mutation in a gene which is expressed only in the brain is a simple matter of analyzing the DNA from blood cells or material from a cheek swab. If one were forced to examine the pertinent protein it would necessitate a brain biop- sy. Likewise, to query the genetic reasons why a child has multiple malformations would be impossible if one were reliant on protein analysis since many of the critical proteins are no longer expressed anywhere in the child's body, having done (or not done) their job during a specified period during the child's embryonic development. *p. 22* Of course, there may be situations where avoiding a patent will be possible. Sometimes, an LD or protein test will work; some day new forms of imaging may make sequencing unnecessary. From a legal perspective, some process or product claims are so narrowly drawn, they can be circumvented. 113 Work can also be done offshore and the results sent back to the United States. 114 But alternative strategies are not regularly available. 115 And not all patents are narrow-patentees claim as much as possible and learn from earlier cases how to draft new claims more broadly. From a clinical perspective, offshoring is impractical, and for research, it could undermine U.S. competitiveness in global markets. *p. 22* Given the difficulty in finding effective substitutes for genetic information, it is no wonder that courts have begun to question the validity of these patents. In Molecular Pathology, 116 product claims to BRCA1 and BRCA2 mutations were invalidated on subject matter grounds, as were process claims for methods of diagnosing a predisposition to breast cancer from BRCA sequences. Molecular Pathology was decided while Bilski was pending in the Supreme Court. However, one Federal Circuit judge has already suggested that Bilski's preemption test raises serious questions about patents on isolated DNA molecules. 117 And as noted earlier, Justice Breyer has been skeptical of claims on associations between patents and disease. When a process or a product patent cannot be invented around, both product markets and innovation markets are badly distorted. ## III. LESSONS *p. 22* The problems encountered in the application of patented genomic advances in both clinical and research settings illustrate why the Supreme Court is wary ## June 2011] *p. 23* PREEMPTION AND GENETIC DIAGNOSTICS 1371 of claims that preempt rivals from competitive development. Admittedly, these effects are most evident through the kind of study described above: after the patents have issued and the inventions are widely distributed. Thus, it can be argued that even if a preemption approach to patentability is desirable in theory, there is no way for the PTO to administer a system that requires such highly evidence-based decisions. *p. 23* There are two responses. First, many issues in patent law cannot be fully implemented by the PTO. For example, application of the novelty and nonobviousness doctrines requires knowledge of the prior art. Some of that art (e.g., prior use and sale) is of a form that examiners have difficulty finding. 118 Yet the requirements are nonetheless maintained; in cases where the art comes to light later, the requirements are implemented through postissuance challenges. 119 Second, as the Bilski Court intimated, there are clues to patentability that both the PTO and courts can use. The machine-or-transformation test, mental steps doctrine, and absence of claim limitations were discussed in Part I. The genetics case studies suggest others. ## A. The Ability to Invent Around *p. 23* A critical feature of patents in the context of diagnostics is that claims to gene sequences and associations between sequences and predisposition to disease cannot be easily invented around. Patent holders can raise prices, refuse to license laboratories, or fail to develop needed tests without fear that an alternative technology will usurp the market for their advance. If society's interest in the development of the field is not aligned with the patent holder's, then it is society that is the loser: it is "preempted" from finding alternatives or leapfrogging over the existing invention to achieve results that are substantially better. 120 In genetics, the problem is that geneticists must work with the physical phenomena of the genes, but the same problem-the fundamental impossibility of circumventing-arises when claims are drawn in the abstract or to principles of nature. *p. 23* The inability to invent around can, however, be no more than a clue to patentability. After all, patents are intended to produce exclusivity; at some level, no claim can be invented around. The issue, then, is one of degree. Furthermore, the determination is sensitive to context. In the genetic realm, for example, the case study demonstrates how patents on diagnostic processes can impede the delivery of healthcare. However, patents on therapeutic products could be circumvented. When genetic information is used for therapeutic purposes, new substances are introduced into the body. These could differ from the patient's own sequences to improve their efficacy or reduce side effects. Because inventing around is not only possible but desirable, product patents on isolated genes would be acceptable if the rights could be limited to therapeutic uses-for example, by creating exemptions for diagnostic and research uses, or by limiting patent scope. 121 Applying an inventing-around criterion to the subject matter issue will thus require both a grasp of the field and an understanding of the patented invention's epistemic significance within it. These are not easy tasks. The National Academies has suggested the PTO convene panels of experts to advise it on patent policy; 122 that idea could be extended to the development of guidelines on the possibilities for inventing around within particular sectors. As previously noted, however, many of these decisions will likely be made post-issuance, when the impact of the patent on the field is evident. Significantly, however, in that context, the question may not be terribly different from the inquiry made when remedies for patent infringement are calculated and the issue of noninfringing substitutes arises. 123 In both situations, the issue is whether there are other ways to reach the ends achieved by the claimed invention. A quick look at the remedies cases suggests that district courts are able to accurately follow the criteria laid down by the Federal Circuit. Of the noninfringing-substitute cases appealed since 1978 (when the current regime went into effect) 124 to the present, the trial court was reversed only 8.7% of the time. 125 In contrast, Jeffrey Lefstin has computed overall reversal rates in the neighborhood of 14%. 126 ## B. Interoperability *p. 24* A closely related concern is interoperability-the demand for equipment that can easily interact. The most familiar example is in the computer arena, where consumers want software that works with the hardware of their computers, computers that work with their printers, and backwardly compatible upgrades. In science, researchers need to compare their results and so require ## June 2011] *p. 25* PREEMPTION AND GENETIC DIAGNOSTICS 1373 wide access to the same (or compatible) research tools. 127 Similarly, the hope of synthetic biology is that a stable set of "parts" (synthesized DNA sequences) will become-like mechanical parts, such as sockets and plugsinterchangeable elements that can be utilized in a wide array of products. 128 In these situations, there may be a variety of ways to achieve a particular result. However, once a choice is made, those who come later are hostage to earlier decisions in much the way that geneticists are hostage to biology. *p. 25* In an important paper on reverse engineering, Pamela Samuelson and Suzanne Scotchmer analyzed this problem in the software sector. Although the authors conceded that intellectual protection can be necessary to encourage the development of platforms, they concluded that net welfare is enhanced when application developers are permitted to utilize and build upon the work of others. More applications are developed, there is less waste, and a competitive marketplace is preserved. 129 As with genetic diagnostics, it is socially preferable to put the first developer's advances into a legal domain where they can be utilized by all. While the authors restricted their policy prescriptions to copyright and contract law, 130 within the patent regime, interoperability concerns may, like the inability to invent around more generally, be taken as a clue to nonpatentability. ## C. Breadth of Prospects *p. 25* Of course, there is a sense in which every invention is unique. Accordingly, the inability to invent around cannot be taken as dispositive of preemption. As important is the patent's dominance. As we saw, information about genetic sequences and about relationships between phenotypes and genotypes open many important opportunities to both clinicians and researchers. In none of the cases studied did it appear that these opportunities were fully utilized when the patent was controlled by a single patent holder or licensee. Indeed, especially in the realm of relating genotype to phenotype, a lack of broad distribution has a profound quelling effect on future development. The number of opportunities a claim produces thus furnishes another, related, clue to the possibility that the claim is preemptive and should not be regarded as patentable. *p. 25* Admittedly, there are counterarguments. Thus, it has been suggested that research is never stymied: patentees are rational; if they are uninterested in de- veloping a prospect, they will license it out. 131 However, the case study suggests that broad licensing is not always the norm. It is easy to understand why. Rationality is bounded by intellectual and informational capacity. Patentees may, for example, have a difficult time understanding the potential for their advance in fields that are remote from their own area of expertise. There can also be significant barriers to licensing, especially between entities like universities and commercial firms that have very different goals. 132 Further, some decisions to hold out are highly rational: the right holder may be afraid that superseding inventions will destroy its market, especially with products that are not encompassed by the patents and therefore escape demands for royalties. Finally, potential licensees can be risk averse and fail to seek licenses when the likelihood is low that their ideas will pan out. 133 ## D. The Identity of the Inventor *p. 26* Another useful clue may be gleaned from the status of the inventors named in the patent. For example, the genetics case studies show that associations between genotype and specific diseases are most often identified by academics. 134 From a practical perspective, that finding is significant because these inventors are not primarily motivated by the promise of patents. 135 PREEMPTION AND GENETIC DIAGNOSTICS 1375 the purpose of determining preemption, academic rewards tend to depend on "abstract knowledge production." 136 Accordingly, work that comes out of academia is likely to be fundamental-and thus raise preemption concerns. Of course, this will not necessarily be the case-an academic who has discovered a broad prospect may also be the one to identify narrow applications. Nonetheless, academic involvement furnishes a clue to preemption concerns. 137 Academics can also be considered examples of a broader class of inventors whose work requires greater scrutiny: what Eric von Hippel calls "userinnovators" (or "lead user" innovators). 138 These are inventors who develop technology for their own use. Thus, they are not primarily working for the rewards associated with patents. 139 More important, the advances they make are often penultimate in the sense that they are made for the purpose of achieving other goals. In the case of diagnostics, for example, clinicians develop associations in order to treat their patients, find new cures for diseases, understand the biology of disease, and reap the reputational awards that will advance their careers. Similarly, research tools are primarily developed to facilitate further research. As Fiona Murray and her coauthors have shown in connection with the oncomouse, which is used in cancer research, patents on research tools can reduce lines of research and retard technological development. 140 Accordingly, they will often raise the same concerns that underlie the Bilski Court's focus on preemption. ## CONCLUSION *p. 27* At the end of the day, the question is whether to foster a culture of innovation or a culture of intellectual property. They are not the same because patenting far upstream can yield royalties while also delaying innovation. The Bilski Court's willingness, despite a commitment to statutory language, to read three exceptions into the subject matter requirement suggests that the Court understands the statute as promoting a culture of innovation. *p. 27* As new technological opportunities emerge, and as universities and other upstream innovators become increasingly aggressive in pursuing patent protection, promoting a culture of innovation becomes ever more difficult. There are many ways to preserve a robust creative environment, including through the disclosure and utility requirements, defenses to infringement, discretion over injunctive remedies, and antitrust law. Many of these approaches may be easier to apply than Bilski's preemption doctrine, but the courts have significantly narrowed two of them-defenses to infringement and antitrust law. The result is significant pressure on the subject matter doctrine. And there are core advances that should remain in the public domain. The hallmark of such an advance is an invention so close to nature that it creates broad prospects that cannot be exploited by inventing around the patent. Other clues include the absence of physicality (the machine-or-transformation test), claims that recite only steps performed in the mind, the absence of claim limitations, the demand for interoperability, and academic or user-innovator involvement. Presumably, as the Federal Circuit begins to apply Bilski, it will identify other ways for determining when a claim is too preemptive to patent. ## Footnotes > June 2011]PREEMPTION AND GENETIC DIAGNOSTICS > 20. 548 U.S. 124, 125-39 (2006) (Breyer, J., dissenting). 21. See SEC'Y'S ADVISORY COMM. ON GENETICS, HEALTH & SOC'Y, GENE PATENTS AND > 51. See Samuelson, supra note 19, at 1101-02. 52. See, e.g., In re Phillips, 608 F.2d 879, 882 (C.C.P.A. 1979). 53. See Parker v. Flook, 437 U.S. 584, 586, 589-90 (1978). 54. See, e.g., In re Bradley, 600 F.2d 807, 812 (C.C.P.A. 1979), aff'd by an equally divided Court sub nom. Diamond v. Bradley, 450 U.S. 381 (1981). 55. Compare Flook, 437 U.S. 584 (holding algorithm unpatentable), with Diamond v. Diehr, 450 U.S. 175 (1981) (holding algorithm patentable). 56. See In re Abele, 684 F.2d 902, 905-07 (C.C.P.A. 1982); 58. 545 F.3d 943, 1012 (Fed. Cir. 2008) (en banc) (Rader, J., dissenting).