Patents offer a powerful incentive for the development and dissemination of new technologies. But patents can be building blocks or bottlenecks—patent 'exclusivity' can drive rapid development; it can also be a tool to suppress innovation. Errors made by the grant of excessively broad or “reach through” claims can choke off a flood of valuable technology. If a patent is defective in ways that cannot be properly corrected by the process of re-examination before the US Patent and Trademark Office, the only other remedy available in the United States is formal litigation, with all its perils and unpredictability. The cautionary tale told here (SIBIA v CADUS) is one more example that fuels a growing pressure for a new round of US patent reform.
Judges and juries are not adequately equipped to make the right decisions in complex biotech cases. Legislative reform in 1982 created the Court of Appeals for the Federal Circuit (CAFC) to ensure even handed and more predictable outcomes in patent litigation. The record to date has not been reassuring. The case reviewed here, chosen from early patents on high throughput screening (HTS), illustrates a powerful lesson – even the victorious litigant may be the worse for it, gaining neither fortune nor wisdom from the proceedings.
A critical stage in a US patent lawsuit is the Markman hearing, named after a 1996 ruling by the Supreme Court. Most District Courts hold Markman hearings just prior to or just after the close of discovery; in the period of litigation when dispositive motions are filed before trial. During the Markman hearing, the presiding District Court Judge will interpret, and if necessary reconstruct, the plain meaning of the issued claims in the patent at issue. If the case goes to Appeal, there may be a second round of claim interpretation at the Federal Circuit level. The review is de novoat both levels, ie. the meaning of the claim language as construed by the litigants, the patent office, or a lower court, is not binding on the Federal Circuit.
More than a few recent patent lawsuits have been all but decided in advance of trial by a favorable or unfavorable interpretation of the meaning of the claims by a Judge or Panel at a Markman hearing. The benefit of hindsight allows us to better understand the workings of this relatively recent innovation in patent litigation, and to illustrate the difficulties and dangers in current law and procedure … when cutting edge biotechnology goes to trial.Some Principles of Patent Law:
A US patent application is a snapshot of the state of the invention at the time of filing, and cannot thereafter be expanded to add new matter. The application includes any claims, drawings, biological specimens, and “specification” necessary to describe the invention. Title 35 USC §112 ¶1 sets out a key requirement of patentability:
“The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art … to make and use the same”. (emphasis added).
What is a “full, clear, concise, and exact” description? The answer is that a patent is directed to a relatively sophisticated reader. Therefore it is customary not to include material drawn from common knowledge or laboratory practice readily familiar or available to “any person skilled in the art”, that hypothetical person who more and more has at his fingertips all the skills of a whole team of scientists working together in discovery phase research, the entire corpus cognitum of bioscience and chemistry, pharmaceuticals and medicine, or engineering and architecture. However, this multifarious, polytechnical entity lacks a key unique dimension—creativity, and it is the “new and non-obvious” that must be fully, clearly, concisely and exactly described in a patent application. To meet the standard of USC §112 ¶1, the description must describe the invention in such scope and detail that the inventive concepts or “teachings” become transparent and obvious by hindsight, in all their manifestations as claimed.
Such a description is said to be “enabling” if our unimaginative but polytechnical scientist, upon studying the application, can then grasp and copy the claimed invention, perhaps even expanding on it by straightforward extrapolation of the teachings.
It is also true that the standard for enablement is higher in complex and unpredictable arts such as biotechnology and organic synthesis. Whereas the inventor will often be given the benefit of the doubt in a simple mechanical case, and a few examples will serve to illustrate a whole panoply of embodiments, this is not so when living cells are an element in the invention, simply because their behavior in one case may be very different in another seemingly related one. Decisions more strictly limiting the scope of biotechnology claims to the scope of the disclosure run in a strong current of black letter law back to University of California v Eli Lilly, Amgen v Chugai, more recently elegantly affirmed in Plant Genetic Systems, and discussed further below.
The written description serves a second important function. While it is the claims that provide notice to the public and delimit the res of the patented invention, the claims must be read in light of the specification, ie. the specification must fully “support” the claims. The court has a longstanding concern that the “public notice” function of a patent has been impaired by claim language that is overly broad or ambiguous, and for that reason has reserved to itself the function of determining the exact meaning of a claim for purposes of establishing claim validity and infringement. A patent claim, in the court’s view, must draw a clear and finite circle around the claimed invention, so that the public is fully apprised of what constitutes infringement, and what does not. This circle must be ‘a bright line’. Thus the Markman hearing.
Some inventions consist of only a single embodiment or “species.” If a family or “genus” of compositions or methods is claimed, then our copyist must be able to make or use all the species of the genus. There is no need to supply a separate description of every possible species if the description given is sufficient to demonstrate to the public that each species in the genus has been individually considered by the inventor and is deemed likely to be useful in the invention. However, a mere “laundry list” of species does not suffice to support a genus claim.
We have already described the hypothetical ‘person skilled in the art’ who must go about copying the teachings of the patent. Following publication of the patent, this person’s knowledge and skill set is expanded to incorporate the teachings of the new invention. But at some point technical advance must slow again as the limits of those teachings are approached. The courts have frowned on "reach through" or downstream claiming of those fruits of an invention not obvious from the written description. In order to limit the claimed invention's reach into future discoveries, some very rational means tests have been devised to establish the radius at which the patent’s teachings no longer illuminate the way, the distance corresponding to the outermost boundary of what the patent enables and may claim. These means tests are called the “Wands Factors,” and should be familiar to every discovery phase scientist.
If infringement is to be shown and words describing the allegedly infringing article or method are not found in haec verbae in the patent claims, the teachings of the patent may be extrapolated only so far, in the sense that the path to the allegedly infringing item must not entail “undue experimentation” or further inventive step. Experimentation to extend the teachings of a patent is ‘undue’ if:
1. the quantity of experimentation required is unreasonable;
2. the amount of direction or guidance provided by the patent is insufficient;
3. relevant working examples are not presented in the patent;
4. and/or the claims asserted in infringement are excessively broad in proportion to the scope of the written description.
A determination that the level of experimentation is ‘undue’ must also take into account:5. the level of predictability or unpredictability of the art; 6. the degree of novelty and nature of the invention; 7. the state of the prior art; 8. and the level of skill typically possessed by those skilled in the art.
These eight Wands Factors set a high bar for enablement in notoriously unpredictable arts such as biotechnology or organic chemistry, where the figurative “bright circle” that surrounds any claim is necessarily drawn tightly around the boundaries of original written description. Very broad, enabling claims, such as the Cohen/Boyer claim to recombinant DNA, are the exception rather than the norm.
There are many cases that illustrate the court’s level of skepticism in extrapolating patent claims in biotechnology. Not all these decisions have been popular, but taken on the whole there is a relatively comforting consistency to them.
In Enzo 1999, an embodiment was described for an antisense nucleic acid sequence as used to block expression of three E coli cell wall genes important in antigenicity and induction of endotoxic shock. However, Enzo’s claims covered the use of antisense technology in both bacteria such as E coli, but also in eukaryotic, vertebrate and mammalian cells, such as those of man. The court ruled that, because antisense technology is complex and unpredictable, and because of the undue time and cost of experimentation that would be required to develop each particular embodiment or to adapt a method from a prokaryotic embodiment for use in higher animals, the Enzo specification did not provide sufficient written description to support a broad claim to both categories of the larger genus of methods. Absent a working example, the inventors had not provided any evidence that they were in possession of the skill to actually perform the invention in animals without further, “undue” experimentation, such experimentation requiring excessive time, cost and uncertainty on the part of our hypothetical copyist, and probably also some creativity. The decision made clear that to claim eukaryotic cells, a working or prophetic example describing use of the technology in at least one eukaryotic cell was a prerequisite.
Taking In re Fisher as another example, claims to compositions of ‘at least’ 1.0 IU adrenocorticotrophic hormone (ACTH) were ruled invalid because the open-ended “at least” limitation was not supported by the two preparations, potencies of only 1.11 and 2.30 IU, described in the specification. The two preparations were novel and non-obvious in the sense that they were much more potent than had previously been achieved, but the court reasoned that future experimenters who might yet achieve even greater potencies of ACTH by inventive means should not be required to surrender their advances simply because the open-ended Fisher claim read on, or “reached through” to cover their yet-higher concentration compositions. The court further opined, clearly drawing the issue in light of statute, that “[Fisher] must not be permitted to achieve this dominance by claims which are insufficiently supported and hence not in compliance with the first paragraph of 35 USC §112. That paragraph requires that the scope of the claims must bear a reasonable correlation to the scope of enablement taught in the specification to persons of ordinary skill in the art.” (166 USPQ at 24). Note that when a US court invalidates a patent claim, it is too late to go back and amend the claim. Overreaching in a patent claim is a fatal error at trial.
In re Fisher goes on to affirm, “In cases involving predictable factors, such as mechanical or electrical elements, a single embodiment provides broad enablement in the sense that, once imagined, other embodiments can be made without difficulty and their performance characteristics predicted by resort to known scientific laws. In cases involving unpredictable factors, such as most chemical reactions and physiological activity, the scope of enablement obviously varies inversely with the degree of unpredictability of the factors involved” (166 USPQ at 24).
The court recognizes a public interest in prohibiting claims that seek to stake out more than the inventor actually contributes. The British have called these the “free beer” claims, defined as any broad, “overreaching claim” that essentially invites others give some novel combination a try, without providing sufficient guidance or specific instructions to predictably make it work, and then claims the credit (and a royalty) if others are actually successful., Similar prohibitions are made by the European Patent Office (EPO); see Article 83: Insufficiency of the disclosure is grounds for invalidity of the claim, and Article 84: The description must enable the invention to be performed over the entire scope that is claimed. While there may be points of law where parties disagree with the methods of the court, there should be no argument that there is a public interest in preventing fuzzy, puffed up and prematurely filed patents from dominating future, more diligent, and oftimes more creative research in a new field. Hyperinflated patents are likely to have the “anticommons” effect described by Heller. These claims, also termed “reach through” claims, because they read on matter or methods not even known at the time the claim was written, are an invitation for a lawsuit.
In other decisions, the court went even further. In Fujikawa v Wattanasin, the court rejected genus claims to chemical compositions substituted with cyclopropyl, even though the cyclopropyl radical was actually listed among other possible substituents. The court reasoned that a mere list of possible substituents did not constitute enabling disclosure of a particular substituent -- in other words, “Were [a mere list to suffice], a ‘laundry list’ disclosure of every possible moiety for every possible position would constitute a written description of every species in the genus. This cannot be because such a disclosure would not “reasonably lead” those skilled in the art to any particular species.” (39 USPQ2d at 1905).
We frequently see in biotechnology and pharmaceutical patents, long lists of reagents, biologicals, and utilities offered in place of working examples spanning the scope of the claims. It is quite possible that these will not be found tenable under §112 ¶1 if litigated. Invention is the act of choosing from a list, not the act of listing. As far back as 1935, this principle was articulated by the respected Learned Hand, who wrote,
“All machines are made up of the same elements: rods, pawls, pitmans, journals, toggles, gears, cams, and the like, all acting their parts as they always do and always must. All compositions are made of the same substances, retaining their fixed chemical properties. But the elements are capable of an infinity of permutations and the selection of that group which proves serviceable to a given need may require a high degree of originality. It is that act of selection which is the [patentable] invention” (Emphasis here).
In Fujikawa 1996, the court held that a patent claim cannot be broadened by mere supplementing of the disclosure with long ‘laundry lists’ of elements. Such lists do not substitute for actually demonstrating at least a firm conceptual grasp and conviction as to how each of the elements of the list are operative in the invention. Mere listing without making the functional nexus is equivalent to a sort of “scorched earth” practice in that it essentially places the elements of the lists and their combinations in the public domain with no sure coverage by the claims, and most undisputedly blocks follow-on “selection” patents not only by others, but also by the original inventor.
The SIBIA 2000 case is instructive as to how some of these considerations play out in the real world of business and innovation. It involves an early patent claiming methods for high throughput screening (HTS) as a tool for identifying cell surface receptor agonists.
The case arose because SIBIA, a commercial offshoot of the Salk Institute, had filed a patent application in 1990 that claimed an elegant screening method for identifying receptor ligands of potential therapeutic use (Harpold US 5,401,629, hereinafter the "SIBIA" patent). Keep in mind that in 1990, HTS was a brand new, cutting edge technology. A strong patent was likely to be of substantial value.
The patent was granted in 1995 with a broad independent claim to, “a method that comprises:
1) genetically modifying a suitable host cell to express a functional mammalian cell surface receptor or ion channel, wherein said host cell is suitable for modification because it is shown not to contain a functional endogenous form of said receptor or ion channel;
2) further genetically modifying said host cell with a reporter gene construct so that transcriptional activity of said reporter gene construct is functionally linked to said receptor or ion channel;
3) treating the genetically modified host cell with a compound or compounds [for example, Insulin];
4) detecting the transcription of said reporter gene construct by assaying the cell or cellular milieu for a signal indicative of transcriptional activity, and;
5) selecting such compound or compounds that modulate transcriptional activity of said reporter gene construct.” [Note: claim reconstructed for clarity]
This claim covers a method for screening for drugs active on any mammalian receptor or ion channel, an enormous piece of the pie. The claims also “reach through” to read on any use of any derivative method for yet-to-be-discovered receptors, transcriptional control elements, reporter genes and ion channels until the year 2012 (ie. 17 years from issue).
A problem arose at trial because no definition was given for the word “cell” in the specification. In order to draw that “bright line” around the claimed invention, one must immediately ask whether the language reads on methods developed for use in either bacteria, frogs, seaweed, or mammals, and whether it reads on any cell whatsoever. The claims and the specification were ambiguous on this point.
CADIS Pharmaceuticals had developed a line of yeast cells expressing heterologous mammalian cellular receptors and containing reporter genes. Depending on the interpretation of the word “cell” in the claims, CADIS was potentially infringing and had not licensed the SIBIA technology. Discussions of terms, exclusivity, TBDF.
SIBIA pressed forward with a lawsuit in 1997, arguing that the claim was constructed to read on any cell whatsoever, including the CADUS yeast cell libraries. In truth however, the actual SIBIA disclosure was much more limited. No xenobiotics were actually screened and only one assay was actually described, an assay testing ligands of the human muscarinic receptor HM1 in a transfected Chinese Hamster Ovary (CHO) cell line. Known agonists and antagonists were shown to behave predictably in the assay. Because this mammalian host cell line was the only embodiment of the assay disclosed in an enabling example, the defendant properly sought to limit the scope of the meaning of the word “cell” to only “mammalian cells”.
In the written description, the host cell is variously referred to as “a cell line” or “an eukaryotic cell”. The District Court held a Markman hearing, and ruled that “cell” must refer to all eukaryotic cells, the full scope of the inventor’s apparent conception, and was silent about the issue of enablement in non-mammalian cells.
When this fell into place, all remaining issues were also resolved in favor of the plaintiff. The defendant, CADUS was obligated to deposit an $18M bond to carry an appeal. The $18M award was actually quite high. In the highly publicised CellPro case, treble damages reflecting intentional infringment amounted to only $7.6M, and the higher figure here, in the absence of intentional misconduct, was consistent with the potential commercial value of the technology. On contingency, CADUS obtained representation by Pennie & Edmonds LLP for an appeal to CAFC; the expected interest on the $18M bond was sufficient to guarantee a $1M flat fee to the lawfirm in return for a successful reversal of the decision against CADUS.
CADUS would base its appeal to CAFC on two points:
A subordinate, affirmative defense argued that the claims were not supported by the written description (USC 35 §112 ¶1), in short that if the definition of the word “cell” were to be limited to a scope supported by the specification, then it should be limited to “mammalian cells.” This same argument was made unsuccessfully at the first trial.
However, there were actually very strong technical grounds in support of this position. CADUS, in reducing the yeast screening assay to practice, had encountered a fatal problem with the assay as envisaged by the SIBIA inventors. Simply put, it didn’t work, the description in the patent was not enabling when applied to yeast. When CADUS transfected a mammalian receptor into compliant yeast cells, the cognate G-protein failed to operatively couple with the heterologous receptor, no second messengers were released, and transcriptional control of a reporter gene construct could not be demonstrated. While it may have seemed obvious to simultaneously transfect the yeast with a complementary set of mammalian G-proteins, that had not worked either.
These negative findings were also confirmed by independent academic researchers,,. Investigators had transfected yeast cells with the HM1 receptor and showed that the mammalian receptor was not coupling to the endogenous yeast G-proteins. Controlling copy number also turned out to be critical, and that was not disclosed in the SIBIA patent either.
In fact, the teachings of the SIBIA patent offered no guidance for those working with yeast. CADUS, rather ingeniously, had then gone about constructing a chimeric G-protein subunit Ga1 that coupled with both the mammalian receptor and also with the remaining subunits of the G-protein supramolecular complex, a functional coupling. This device enabled transcriptional control of suitable yeast reporter gene constructs, including fluorescent endpoints not described in the SIBIA filing. Published descriptions show that a very large amount of experimentation by trial and error had gone into this solution.
The design and execution of this research strategy had not been easy. Ga1 of yeast contains 472 amino acids; known mammalian Ga1 proteins have between 340 – 350 amino acids. Homologies permit sequence alignment, and the function of certain highly conserved regions such as the GTPase domain could be recognized. But other sequences formed loops extending out of the core GTPase domain, and the function of these loops or “inserts” had not been established. Inserts 3 and 4 were believed to participate in binding to Gg . In addition, it was known that the n-terminus was myristylated at Gly, position 2, indicating a requirement for post-translational modification, and that the flanking n-terminal sequence participated in associations with Gb..g of the G-protein complex. These were reversible species-specific associations. Finally, it was the carboxy terminus and flanking sequence of Ga1 that was implicated in the association of the G-protein transductional complex with the ligand-binding transmembrane receptor and its 7 helices.
The CADUS team designed a tic-tac-toe strategy involving substitution of various lengths from the mammalian Ga1 protein with yeast n-terminal sequences, or various lengths of the yeast Ga1 protein with mammalian n-terminal sequences. After study, it was determined that about 10 to 20 amino acids at the n-terminus were sufficient to induce association of the mammalian Ga1 with the remaining subunits of the yeast G-protein complex. Other hybrids that involved “double-crossover” were also tested, as were hybrids involving gene fusion between the various G-protein subunits. In the course of this work, CADUS discovered a possible mechanism for G-protein subunit association between Ga1 and Gb..g involving coupled reversible winding and unwinding of mated amino-terminal a -helices of the G a1 and Gb..g subunits, analogous to a leucine zipper. Ga1 activates adenylate cyclase, Gb..g reverses it in mammalian cells. Therefore, a sequence swapping strategy of taking 7 amino acids at a time (each 7 amino acids corresponding to one turn of the a-helix) was tested. Simultaneously, CADUS also built a library of mutants, and the mutants were used also in generating a large library of Ga1 hybrids. Mutatis mutandis, point mutations of Gb..g sequences likely to be involved in subunit association were also generated. The constructs were then transfected into yeast cells so that subunits were coordinately expressed, or at least a rough proportionality of subunits was maintained. Expression of the resulting construct in a low copy plasmid, or by genomic integration under the transcriptional control of the GPA1 promoter, provided suitable control.
The reporter construct CADUS relied on was the well-characterized yeast pheromone-responsive mating system—specific proteins are expressed in response to pheromone binding at the STE2 ligand receptor on the cell surface and the cells toggle between G1 and S phases of growth. The transcriptional control elements for this system have frequently been used to generate fusion hybrids expressing foreign proteins such as the fluorescent Green Protein and were readily adapted as reporter gene constructs in the CADIS assay.,
Some years later, NIH researchers also achieved functional expression of human muscarinic receptors in yeast, but only when a Ga1/Gaq fusion (a general approach CADUS had failed at) engineered to contain the C-terminal mammalian Gaq sequence, was co-transfected. This work was also outside the scope of that envisaged in the SIBIA patent.
Earlier, King demonstrated functional coupling of a mammalian receptor protein with the yeast mating system via a G-protein. But the key involved re-engineering the mammalian b2-adrenergic receptor to contain 16 amino acids from the STE2 protein of yeast (STE2 is the yeast sex hormone receptor that couples to the native yeast G-protein). This solved the problem in different way, and American Cyanamid obtained a patent for those efforts. Note that the American Cyanamid patent was not found to be in "interference" (ie. claiming the same invention) with the SIBIA patent.
But for its troubles, CADUS was handed a patent infringment lawsuit. The company, a spin-off from ImClone, had been commercially successful in lining up partners for its assay methods and was actively developing ligands that had been identified as potential therapeutics. CADUS was collaborating to develop high throughput pharmacological systems based on a multiwell plate format, and had entered into an agreement with MIT to perform similar assays on a biochip. CADUS had also received a $5M milestone payment from Bristol Myers Squibb for drug screening work, and had signed a $50M deal with SOLVAY to identify adenosine receptor antagonists. In addition to the assay, it had a library of 30,000 yeast strains, and assets in functional genomics.
Drained by the $18M bond, CADUS ceased operations about 1 year before its appeal of the SIBIA infringement ruling was decided. In a deal that can best be described as a firesale, OSI Pharmaceuticals acquired the CADUS technologies, including the SOLVAY collaboration in progress, certain intellectual property, and the 47 members of its scientific staff – for $2.2M. OSI also took over the lease and equipment at the CADUS research facility. Parenthetically, OSI’s lead drug, Tarceva™, now in Phase III trials for cancer is currently under development in partnership with Pfizer, but the CADUS technology has not been developed further.
As stated earlier, the CADUS appeal was taken on contingency, so the litigation continued in spite of the business setbacks. As planned by Pennie & Edmonds, the primary CADUS defense on Appeal was based on the new argument that the SIBIA claims were invalid on the grounds that they were obvious upon combination of two references in the prior art. It was argued that this was a bar against patentability under USC 35 §103a.
The two references chosen were Stumpo 1988 and Lester 1988. Stumpo described the effects of insulin binding in a co-transfected CHO cell containing the human muscarinic receptor and a reporter gene construct using c-fos as the transcriptional control element and CAT as the reporter gene, essentially as described subsequently in the SIBIA patent, examples I – III. The Stumpo reference, however, did not demonstrate or even conceptualize the utility of this co-transfected cell in a drug screening assay. Stumpo’s academic focus had been on the biology of ligand binding and transductional control under c-fos. There had to be a 'motivation' to modify the Stumpo reference into an assay.
CADUS argued that Lester 1988 supplied the missing link -- more precisely, that the Lester reference supplied the motivation to modify Stumpo so as to perfect a rapid drug-screening assay. Given a motivation to modify the Stumpo reference, and stipulating that the Stumpo composition almost identically mirrored the SIBIA composition, then under US patent law the assay was presumably already in the public domain because it was obvious as of the time of filing, and could not properly have been patented. Hence there is no infringement. Surprisingly, the court was persuaded by this argument—ruling that the SIBIA patent was obvious and invalid in light of the combination of Stumpo and Lester, and reversing the lower court’s decision that awarded SIBIA $18M dollars.
CADUS won. CADUS got its $18M bond back, plus interest, and paid off the lawyers in the summer of 2000, about 1 year after it had ceased operations and let go its staff.
So, justice was finally served. Or was it? The SIBIA decision on non-obviousness has a collateral effect of limiting what is a patentable biotechnological invention in an important but troublesome way. The problem goes back to the question of what level of inventiveness must a patent meet, a distinctly American thicket long known to be seeded with nettles. The decision is a concern to all inventors because it can be extended to invalidate patents based on the size of the inventive step, a torturous standard that has a history of problems in litigation.
For starters, the decision written by the Circuit Judge Gajarsa was controversial and split the panel 2 to 1. Chief Judge Mayer dissented, arguing that this was not a case about non-obviousness at all, and that the decision undermined the credibility and predictability of the judicial system on which all must rely, and indeed the rule of law because it overturned against all precedent a reasoned jury decision affirming the patent’s non-obviousness.
Chief Justice Mayer argued that the jury decision with respect to obviousness must stand as reasoned at trial, and pointed out the grave error of reading obviousness into an invention through hindsight. If a patentee’s own disclosure is read into past publications, old phrases take on new significance and new meaning in the light of the patentee’s teachings, and the claims of any patent at issue can suddenly seem to have been common knowledge a year or years before it was filed. The perceived motivation to modify Stumpo was never actually present in Lester, Justice Mayer opined, and was the product of hindsight.
The insidious influence of hindsight cannot be understated here. The Lester reference actually takes on a wholely different meaning when read accusatively against the SIBIA patent. In such a light, one could easily imagine that “several investigators” were already in possession of the reporter gene constructs disclosed by SIBIA. But importantly, if the article is read just for what it says in its historical context pre-HTS, each and every sentence is seen to be directed to unrelated interests, with no bearing or influence on the HTS invention by SIBIA.
In fact, Lester’s article is focused on an entirely different problem, a problem regarding validation of therapeutic targets. Drugs such as phenothiazines, benzodiazepams and dihydropyridine calcium channel blockers, among others in Lester’s view, lack specificity of drug action, a major cause of side effects. Existing drugs act on families of common cell surface receptors, many of which are the products of gene duplication. Tissues and organs contain varying amounts and combinations of multiple, closely related receptor sub-types, and the action of a drug is apparently the cumulative sum of all its actions on all the different receptor sub-types, a problem that invalidates use of these mixed receptor populations for drug screening. Thus, Lester argues, the recent availability of heterologously cloned mammalian receptors offers the opportunity to study each receptor sub-type in isolation, by cloning it into a separate cell, and to design drugs that act specifically on only one sub-type, ie. one therapeutic target at a time. As far as functional assays, there were existing physiological and pharmacological methods, including electrophysiology and second messenger assays, that were satisfactory for screening without the need to invent new ones. The following year, without the benefit of a reporter gene construct, investigators carefully separated therapeutic effects of drugs in isolated muscarinic receptor subtypes, exactly as Lester suggested, without HTS.
Of critical importance to the SIBIA decision, we need to compare the motivation of Lester versus the motivation of the SIBIA inventors. Recall that it was the motivation to modify the Stumpo reference, ie. to convert it to a drug-screening assay, that Justice Gajarsa found in Lester. However, this was simply not the case. Lester offered no incentive or suggestion to develop faster, high throughput screening tools for drug discovery—his entire point was that adequate tools were already in place. He listed various physiological and pharmacological techniques, including electrophysiology and second messenger assays, suitable for detecting functional receptor binding and stressed the need for patience in the laborious task of screening for new drug candidates. Lester urged investigators to focus on the biology, not the screening methods, precisely because the research techiques and the experimental model required for that purpose had recently been developed in the work leading up to his publication.
In contrast, the motivation or problem addressed by the SIBIA inventors was the unmet need for a rapid method to screen thousands of potential ligands, in essence, the need for HTS. We clearly see that use of a reporter gene construct offers the prospect of rapid drug screening, for example with fluorescent proteins in a multiwell plate. Speed was the instant criterion that Lester ignored.
Moreover, a good patent attorney would want to point out that the Lester reference actually “teaches away” from the SIBIA invention. Lester discussed the same muscarinic receptors that Stumpo dissected in detail and wrote, “Ligand screening probably can be conducted in cells that express (or can be induced to express) the appropriate G protein, with little regard for subsequent steps.” Upon reading this, no investigator would have been motivated to re-examine the Stumpo reference for the keys in those subsequent steps to a rapid screening technique based on downstream transcriptional control of a synthetic reporter gene construct. Lester unmistakably relies on native, endogenous assay endpoints, with no recourse to a synthetic molecular device for speeding up the screening. The fear and downside of introducing synthetic elements into the assay was always that ligand:receptor activation would no longer be representative of the in vivo biology. A great deal of work had been required to demonstrate that heterologously expressed receptor proteins, with a few exceptions, were functionally equivalent to their native counterparts. In Lester's view, the heterologous receptor model adequately reflected the biology, he saw no need for investigation of surrogate endpoints, indeed he discouraged it.
The leap made in SIBIA is thus even less obvious given the disincentive Lester gave to look for it. The motivation missing in Lester, but central to the teachings of SIBIA, is the perception of a need to screen very large numbers of compounds in a rapid, robust assay. Lester instead was content to view drug screening as a “long process”; the SIBIA inventors were not.
The example is especially valuable and worthy of study because upon re-reading the Lester article in Science magazine with the SIBIA teachings for hindsight, the article suddenly appears to be laced with prescient or even anticipatory comments under every subheading. Even the CADUS yeast technology is seemingly anticipated. Yet this pregnancy of meaning entirely disappears when the article is read for what it teaches in and of itself. The fascination of the illusion of hindsight in patent examination can readily be experienced by the reader. Those readers familiar with HTS will immediately think of reporter gene constructs, microtiter plates and fluorescent tags when they read the following sentences taken from Lester's article, but a naive reader would make no such leap because the words are simply not there, even by inference.
Nowdays, we would instantly think of fuel injection when someone talks about race car engine combustion, but before fuel injection was invented, only carburators would come to mind.
But for Justice Gajarsa, this was the smoking gun, proof certain that the invention was obvious. Lester had been within a hair's breadth of stating exactly so. Here was the motivation to modify Stumpo, one simply had to apply the full range of modern techniques, including those described by SIBIA in their patent. Hindsight is a kind of circular reasoning, as protested Chief Justice Mayer in dissent, to no avail.
To expand on Chief Justice Mayer's views, the mere “motivation to modify” is not the legal test for obviousness, the test is the motivation to modify in a particular way. In this light, the Lester article is seen more as evidence of a “long felt and unsatisfied need” than as a roadmap to the invention. There had to be a connection that demonstrated how a reader at the time would have turned to Stumpo to solve the problem posed by Lester. The test is not whether a reader at the time would have turned to Lester to figure out why to modify Stumpo, which is what Justice Gajarsa did. Chief Justice Mayer argued that Justice Gajarsa had made that connection in his own mind through the illusion of hindsight. What is perhaps tragic here is that an otherwise innocent 3d party was most hurt by the flawed decision in a case that should have taken no more than a pretrial hearing to dismiss when it was first filed, as a matter of established law readily applied, coming down on all fours on top of the facts of the case.
There is grounds to argue, as Justice Gajarsa is perhaps doing, that in fast-moving technologies, the standard for inventive step needs to be raised, but there are two ways to achieve the same end, one is to embellish the difficult legal concept of non-obviousness, the other is to simply limit the claims properly to the scope of the disclosure as already the law of the land (§112 ¶1,2: written description, enablement and sufficiency). There is no doubt in the SIBIA case that enablement had not been demonstrated, and the inventor's later claim to have conceived a broader invention was weak and semantic at best. Too often a patent attorney will stretch the scope of the invention beyond what can reasonably be technically supported, a self-serving and destructive proposition in the long run, but one that has been unnecessarily condoned by the courts, most probably for lack of technical insight..
No matter the outcome, the reasoning behind the decision in the SIBIA case is not moot and has been used as citable precedent. The SIBIA decision is important because, to put it bluntly, the judicial hindsight exercised here is not an isolated instance, and has the effect of 'robbing' more than a few legitimate inventors of the fruits of their imagination. That instant when an inventor who looks at common and well known facts, but suddenly sees them anew with an unexpected and useful insight or combination, ceases to be an inventive act. And just as disturbingly, appellate litigation becomes a wholely unpredictable undertaking, dependent as much on which judges are drawn for the panel as on the merits of the case. This is a problem of enormous magnitude today, undermining the whole system, and it is not the first time it has been made an issue.
The court has long sought to eliminate the subjective element of hindsight from its judgements on obviousness as a factor in patent validity. Central to this effort has been a series of cases culminating in Graham v John Deere, a Supreme Court decision familiar to every patent attorney but less so to discovery scientists. The Graham decision lists “secondary indicia of non-obviousness,” criteria that form objective, fact-based tests designed for the sole reason of avoiding hindsight. Subsequent litigation has reinforced the Supreme Court’s pointed assertion that the Graham tests are to be the first and most important factors in making the decision, not merely relegated to a checklist of activities after the decision is made.
And yet the SIBIA decision on appeal seems to disregard this wisdom, echoing earlier cases decided by hindsight. Note in the written decision that the reasoning is based first on the content of the Lester article, and then, as a procedural footnote, the secondary indicia of non-obviousness that the jury heard in testimony and explicitly considered are cursorarily dismissed as blatantly misleading or insignificant. So clear is the perceived import of the Lester reference to Justice Gajarsa that both a Supreme Court decision and a Jury decision is pushed aside.
The Jury at trial had found objective and convincing evidence of a long-felt need and commercial success of the SIBIA patent, two or three of the secondary indicia of non-obviousness. There was testimony of experts as to the long-felt need. A long felt need is taken as evidence of non-obviousness because, the reasoning goes, if the need had existed for a long time then the mere fact that no one else had yet come up with the invention was evidence that the solution was not so obvious after all. Also, SIBIA could demonstrate at trial that it had commercial success with the invention. It had actually licensed the invention to more than one competitor and had then sold the patent to MERCK. And OSI in turn had licensed the technology from MERCK. This information also demonstrates that various competitors had in fact adopted the method in their own drug screening efforts. Adoption of the invention by competitors is taken as another of the secondary indicia of non-obviousness, ie. imitation is flattery, if your competitors license your patent then the invention must have impressed them, again pointing to its validity. These were facts not disputed at trial. Thus, several factual findings in support of non-obviousness were noted by the jury. As a matter of law, the appellate court has an obligation to respect a Jury’s decision when based on reasoned evidence. It should also be noted that the PTO, which is charged to evaluate any prima facie case for obviousness as a condition of patentability, and employs scientists familiar with the state of the art in each field, had also decided the claims were valid. Yet CAFC trivialized all these factors. So we see that the problem here has broader implications. No precedent is safe, no prediction is reliable, patent law in the United States under CAFC is in a state of chaos. If one citizen is denied justice, then are any of us safe in the hands of the law? There are actually a much larger number of cases that point to the present reign of chaos in decisions by CAFC. That's the larger issue.
Could anything have been done to avert this fate for CADUS? Prior to the first trial, strategic motion practice might have allowed CADUS to position the question regarding the definition and scope of “cell” in the form of a motion whereby CADUS as defendant would have made opening argument and could frame the question in the most favorable light. However, once the District Court held a Markman hearing, where the plaintiff goes first, and interpreted the word “cell” to apply equally to all eukaryotic cells, there is very little the defendant’s attorneys could have done to change the trial verdict. The unfavorable outcome at the Markman hearing was literally the kiss of death at trial. Nonetheless, on Appeal, CAFC was free to make a de novo determination of the meaning of the claims should it so choose, and was not bound by the District Court’s interpretation of the word "cell". Again, strategic motion practice could have framed for the appellate court how the disconnect between the expansive meaning of the claim and the diminutive scope of the disclosure turned on the meaning of a single word, citing a context of relevant statute and other precedent.
Our larger concern is that this pattern is being repeated over and over; the more complex the science, the more confused the decisions. If CAFC had acted according to its own precedent, rather than find for the defendant on grounds of invalidity under §103, and had instead first examined the breadth of the claims at its own Markman hearing, it could have either restricted the meaning of “cell” to “mammalian cell” and remanded the case to District Court (for summary judgement there) or simply invalidated the patent claim itself under USC 35 §112 ¶1 in its own summary judgement. These actions might have been speedy enough to prevent the demise of CADUS. And even if they did not come speedily, they would have better sustained the integrity of the system for the long haul. The fact is that if patent litigation were more predictable, then there would necessarily be less of it, an irony perhaps most acutely perceived by the defendant in this case, the victor and loser in the reversal of fortunes recounted here.
The state of the science at the time of the invention is often the most difficult part of the case to make in court, and forms a frequent basis for invalidity defense. Effort to hammer this out during the drafting and examination phase offers clearcut rewards for all parties, including for the courts and the public.
Legislation enacted in 2003 has started to address the problem by enhancing inter partes proceedings before the US PTO.. The European Patent Office (EPO) has set the standard; both pre-grant and post-grant inter partes adversarial proceedings in about 7% of all filed patents result in more stringent examination of claims . These proceedings are inexpensive, typically involve the presentation of scientific data and argument by opposing parties, and are judged by qualified senior scientists employed by the EPO. The opposing parties are not denied due process, an appeal can transition the matter to the courts, and the scientific testimony on record serves both sides in any ensuing patent litigation. The cautionary tale told here illustrates the frustrations with the existing US model, the lamentable state of patent law practice before the CAFC, but also the relative ease by which the purely scientific issues underlying many litigations could be more thoroughly addressed.
Please address any comments to KLambert@4iPT.com.
 Markman v Westview Instruments Inc, 52 F3d 967, 979; 34 USPQ2d 1321, 1329 (Fed Circ 1995 en banc), aff’d 517 US 370 (Supreme Court 1996); and Lee W. 1998. The ever confounding question of claim construction: Markman and its progeny 531 PLI/Pat 151, 175-76.
 Fujikawa v Wattanasin 93 F3d 1559; 39 USPQ2d 1895 (Fed Circ 1996).
 In re Wands, 858 F2d 731; 8 USPQ2d 1400 (Fed Circ 1988).
 Enzo Biochem Inc v Calgene Inc. 188 F3d 1362, 52 USPQ2d 1129 (Fed Circ 1999)
 427 F2d 833, 166 USPQ 18 (CCPA 1970)
 Grubb PW. 1999. Patents for Chemicals, Pharmaceuticals and Biotechnology. Oxford University Press, Oxford. Pp 311-12.
 Biogen v Medeva RPC 68 (CA 1995)
 Heller MA and RS Eisenberg. 1998. Can patents deter innovation? The anticommons in biomedical research. Science 698: [May 1]
 93 F3d 1559; 39 USPQ2d 1895 (Fed Circ 1996)
 BG Corp v Walter Kidde & Co, 79 F2d 20, 22,26; USPQ 288 (2nd Circ 1935).
 SIBIA Neurosciences Inc v Cadus Pharmaceutical Corp, 99-1381 (Fed Circ 2000)
 Bar-Shalom A and RC Deegan. 2002. Patents and innovation in cancer therapeutics: lessons from Cellpro. The Milbank Quarterly 80(4):637-76.
 Payette P et al. 1990. Expression and pharmacological characterization of the human M1 muscarinic receptor in Saccharomyces cerevisiae. FEBS Lett 266:21-5.
 Kand et al. 1990. Mol Cell Biol 10:2582- .
 Erlenbach I et al. 2001. Functional expression of M1, M3 and M5 muscarinic acetylcholine receptors in yeast. Neurochem 77:1327-37.
 Fowlkes DM et al. 1998. Yeast cells engineered to produce pheromone system protein surrogates, and uses therefor. US Patent 5,789,184.
 Manfredi JP et al. 1996. Yeast alpha mating factor structure-activity relationship derived from genetically selected peptide agonists and antagonists of Ste2p. Mol Cell Biol 16:4700-09.
 Klein C et al. 1998. Identification of surrogate agonists for the human FPRL-1 receptor by autocrine selection in yeast. Nature Biotech 16:1334-7.
 Erlenbach I et al. 2001. Functional expression of M1, M3 and M5 muscarinic acetylcholine receptors in yeast. Neurochem 77:1327-37.
 King et al. 1990. Science 250:121-23; US Patent 5,482,835.
 Price, LA et al. Functional coupling of a mammalian somatostatin receptor to the yeast pheromone response pathway. Mol Cell Biol 15:6188-95.
 Stumpo DJ et al. 1988. Indentification of c-fos sequences involved in induction by insulin and phorbol esters. J Biol Chem 263:1611-14.
 Lester HA. 1988. Heterologous expression of excitability proteins: route to more specific drugs? Science 241:1057-63.
 Buckley NJ et al. 1989. Antagonist binding properties of five cloned muscarinic receptors expressed in CHO-K1 cells. Mol Pharmacol 35:469-76.
 Graham v John Deere Co 383 US 1; 148 USPQ 459 (CPA 1966).
 Anderson Black-Rock
Ó Copyright K K Lambert 2001-2005. All rights reserved. For permissions, contact the author.