In re Kubin: The Obviousness of DNA

    By Kevin E. Noonan --

The Federal Circuit heard oral argument for In re Kubin last week, and a very hot bench (Judge Rader presiding, joined by Judges Linn and Friedman) sharply challenged the positions of both Kubin and the Patent Office.  While it is foolish to attempt to read the tea leaves of judicial intention when listening to oral argument, the questions from the bench point out once again that obviousness remains a legal issue intimately dependent on the underlying facts (and misapprehension of those facts).

Judge Rader immediately raised the "obvious to try" issue with Barbara Rudolph, representing Kubin.  After letting Ms. Rudolph get about 80 words into her presentation, Judge Rader asked her "If I use the O'Farrell standard of a reasonable expectation of success, don't I come out with the office?" which, he reminded her, preceded In re Deuel and thus would be controlling precedent.  She reminded the Court that the O'Farrell standard for "obvious to try" sets forth two situations where what is obvious to try is not obvious.  One of these is when all possible parameters are varied with no direction from the art on which parameter to vary or how.  That was the situation here, according to Ms. Rudolph. 

Judge Friedman then gave counsel the opportunity to take a breath and set forth some background information, asking for a brief synopsis of what the patent applicants were trying to do.  She responded by explaining the underlying immunology of natural killer (NK) cells.  In response to renewed questioning from Judge Rader, that the prior art Valiante reference "taught p38 which is NAIL" (the subject matter of the Kubin application was a cDNA encoding human NAIL), she reminded the Court that what Valiante disclosed was "a band on a gel," i.e., that the Valiante patent disclosed an antibody immunologically reactive to p38, and therefore, that Valiante did not disclose purified p38, but merely its detection on a Western blot (where it would be expected to be in the presence of a multiplicity of other proteins).

At this point the Court went off the factual rails a bit.  Judge Rader asked:

You're talking process to me, and frankly the only difference with the Valiante process and the Kubin process is the difference between a liquid immunoabsorption technique and a solid immunoabsorption technique, but we are not talking processes anyway.

From which Judge Rader came to the conclusion that:

We're talking p38 is NAIL isn't it.  . . .  So the prior art teaches what is claimed here, and you're trying to say:  "But they got there a different way."

Frankly, no.  Ms. Rudolph reminded the Court that Kubin was claiming a cDNA and the prior art taught the p38 protein, and that Kubin's claim was not to the entirety of the NAIL sequence but just to a fragment of it.  This led the Court to what seemed to be a greater misunderstanding of the underlying facts:

Judge Rader:  You're right, they actually claimed the probe that binds to NAIL, but once you [have] the probe that binds to NAIL, and the methods for producing the sequence, why hasn't the prior art shown a reasonable expectation they get to everything you claimed very quickly?

Rudolph:  Well, what they claim was the gene fragment encoding NAIL, but not all of NAIL, just a particular . . .

Judge Rader:  Yeah, they claimed that, but remember what they taught:  p38, which is NAIL.  They were smart; they claimed the probe.  The probe is what is of value anyway, so that's what you claim.  But they taught p38, which is NAIL.

Rudolph:  They taught only . . . they did not teach p38 in the sense of giving its . . . any information to identify . . .

Judge Rader:  No, they didn't give its sequence.  But again, that's where you can use the commonly used methods [that] anyone with skill in the art knows to get the sequence, and the only difference between the way they got the sequence and [the way] you got the sequence is the difference between the liquid and the solid immunoabsorption technique.

This colloquy seems to indicate that the Court is at least receptive to the Patent Office position that the cloning methods were "routine."  But this ignores, of course, the critical factual distinctions between the prophetic example from the prior art Valiante patent and what Kubin actually did.  Ms. Rudolph tried to get the Court back on track:

Rudolph:  No, because there is no reasonable expectation of success, because there are things that are missing in Valiante.  First is the NK cell library.  Valiante does not disclose the starting material, and what does Matthews teach us?  Mathew teaches us when you use a conventional method [to] isolate this gene, it does not work.  And what they used was a NK cell library with total RNA from NK cells, and that did not work.

Judge Rader seemed to continue to miss Ms. Rudolph's point that what was deficient in the art were not cloning methods, but a source for preparing a cDNA library:

Judge Rader:  We get the library out of Sambrook.  That is why they cite that part.  Matthew is kind of a reinforcing reference that shows [that] they did do everything you said they haven't done in the murine context, which shows again [that] it's easy for one of skill in the art to do it.  They can do it in mice.  They can do it in humans.  [They’re] closely related genomes.

As anyone familiar with the underlying science will recognize, you don't "get the library out of Sambrook."  You get generic cloning methods from Sambrook, but the predicate for using these methods is having a biological source, a cell or tissue, that expressed the protein encoded by a cDNA of interest, from which source Sambrook can be used to prepare a library.  It is ironic that one of the classic scientific limitations of gene cloning was ignorance of the source of such important cDNAs like blood clotting Factor VIII.

Judge Linn commented that, on the one hand, the Matthew reference "troubled" him as a teaching away, but also said he thought that example 12 of the Valiante reference was "of consequence" to the obviousness question.  In response to Judge Linn's question about the significance of Example 12 being a prophetic example, Ms. Rudolph responded that while its prophetic nature "didn't matter," looking at the evidence as a whole showed there was no reasonable expectation of success that practicing Example 12 would result in Kubin's invention.

Judge Rader continued to emphasize the existence of the antibody probe, combined with using "the Sambrook libraries," and then (voila!) "you find the sequence."  Ms. Rudolph responded succinctly:  "Probe it in what?  . . .  [Y]ou need a starting material from which to probe.  And if you have a NK cell library, that's not going to produce p38 or sufficient amounts [of it]."  Judge Rader then asserted that "Valiante had already produced p38," to which Ms. Rudolph reminded the Court that had been done for the protein and not for the cDNA.  Specifically, she correctly asserted that "you need a starting cell library that will have a sufficient amount of that particular gene that will produce that particular protein."  Unfazed, Judge Rader responded:

Well that is the kind of thing you assign to your lab assistant -- you assign your lab assistant to do it.  And yes, it's trial and error; and yes, you make lots of mistakes.  But remember, we have dealt with that in our case law:  Atlantic, Atlas Powder for instance.  You can have lots of mistakes and still produce it, and fully enable your invention.

Ms. Rudolph made one more attempt to clarify the issue:

Rudolph:  But the Board hasn't identified -- jumping to a different aspect of the claim -- the Board has not identified any basis for identifying a specific binding region -- what the structure of that binding region is -- and it hasn't identified any basis for finding CD48-binding obvious. And those are claim limitations that the Board would consistently like to ignore, or the PTO would like to consistently ignore.  And those are important limitations that could not have been predicted from Valiante, or any of the cited references, or Sambrook.  Which is really . . . Sambrook does not contain a recitation of NK cell library, Sambrook is just a general cloning . . .

Judge Rader:  No, it tells one of skill in the art how to produce those libraries, and when you [have] the probe it's not so hard to do.

Rudolph:  It does not tell one with skill in the art how to produce a NK cell library, and if you look at the methodology that's recited in the specification, what they used was a specific mixture of resting cells, resting NK cells and NK cells stimulated with a very specific cocktail of activators.  That's not disclosed in Sambrook.  That's not disclosed in Valiante.  That's not disclosed anywhere, and nor is any of the CD48-binding aspects of this claim disclosed anywhere.  And this is really getting down to the problem that we see with the Board's analysis -- [the Board] is looking at the methodology and that's not the proper way.  It's easy to say, sure . . . could someone have used a hammer and a nail to get there?  Could someone have used conventional tools to get there?  That's a different question from:  Is this claimed subject matter as a whole obvious than what came in the prior art?  And looking at this invention, as is should be looked at -- the genetic basis for the binding region for an very important interaction between NAIL and CD48 -- when that interaction was not known in the prior art and has significant biological consequences.

Janet Gongola, Associate Solicitor for the Office of the Solicitor, argued for the Patent Office.  Judge Rader started her off with Deuel, asking her the converse of his O'Farrell question to Kubin's counsel.  She distinguished Deuel based on the statement in Deuel that:

Identification of a structurally similar molecule in the prior art is normally where you would start to make a chemical obviousness finding.  Normally -- that word is key because it is descriptive of one way you can establish chemical obviousness.  It's not prescripted  . . . so you always have to do it that way.

A unique argument to anyone with a memory of the In re Bell / In re Deuel dyad of cases and the Federal Circuit's clear holding (followed de facto by the Patent Office ever since) overruling the same arguments the Office now makes against Kubin's claims.  Judge Rader persisted, asking Ms. Gongola to "distinguish the rule of Deuel, which says general methodologies are irrelevant.  [And if so,] you lose Sambrook, and if you don't have Sambrook, you can't have the libraries [and] you can't use example 12 to get to Kubin."  Her response:

Gongola:  Respectfully, your Honor, I disagree with that Deuel goes on to general methodologies for making an undefined cDNA molecule.  The Court there said that it was irrelevant to whether a specific molecule would have been obvious.  In the very next sentence, the Board explained what to do -- I'm sorry, the Court -- explains what to do with this specific method, and the Court said:  "A prior art disclosure of the process reciting a particular compound is another matter raising issues of anticipation and obviousness."  We are that other matter, and what the Board did here is [it] looked at the teaching of Valiante.  Valiante identified p38 cDNA, [which] gave [the skilled artisan] motivation to make it, and then provided a methodology of doing so.  So Deuel envisions the situation we have here, and says the result would be different.  We have a specific method, teaching how to make a specific DNA molecule, falling within the scope of the genus.

Judge Linn then asked her about the statements by Kubin's counsel that Valiante did not show a library.  Ms. Gongola responded:

If you listen to what exactly Ms. Rudolph said, she explained to you that Example 12 did identify starting materials -- a specific kind of NK cell library that a person of skill of the art would have known to use.  This kind of goes to the question of reasonable expectation of success -- Matthews [sic] provides that reasonable expectation of success.  Matthews followed the method of Valiante identically except [for] using mouse NK cells instead of human [cells] as a starting material, and instead of using a probe specific for human p38 cDNA.  Matthews used [a probe] for 2B4 cDNA.  But Matthews successfully followed the method otherwise, and obtained 2B4 cDNA, the mouse counterpart to human NAIL.

So, the Board made specific findings in this regard on page A9 of its decision.  It explained that Matthews is cumulative to Valiante and provides the reasonable expectation of success.  Kubin itself, before the Board -- and the Board made this finding at Finding of Fact 13, admits that conventional cloning methodologies were known in the art.  People knew how to clone.  The Board capitalized on this, found that Valiante taught a prior art species, that [this] species was obvious in light of the method used to make it, and then in turn said that the prior art disclosure of an obvious species under the line of Lilly renders the claimed genus obvious.

There are, of course, two errors of fact and one of law in this argument.  The first is that neither the Matthew reference nor the Valiante reference taught that mouse 2B4 was the murine homolog of human NAIL.  That understanding was evident only when Kubin succeeded in cloning NAIL cDNA and compared it to the mouse 2B4 reference.  Second, the reason that Matthew could use substantially the same method that Valiante prophetically taught for cloning the murine p38-encoding gene is that mouse NK cells make sufficient 2B4 mRNA that a conventional murine cDNA library will be capable of producing cDNA clones encoding the sought-after gene.  Kubin (and Matthew) show that this isn't the case for human NK cells.  Ms. Gongola, and the Board, continue to base their obviousness argument on the obviousness of the cloning methods -- "People knew how to clone" -- but as Judge Lourie realized in Deuel, that isn't the standard:

The PTO's focus on known methods for potentially isolating the claimed DNA molecules is also misplaced because the claims at issue define compounds, not methods.  See In re Bell, 991 F.2d 781, 785, 26 USPQ2d 1529, 1532 (Fed. Cir. 1993).  In Bell, the PTO asserted a rejection based upon the combination of a primary reference disclosing a protein (and its complete amino acid sequence) with a secondary reference describing a general method of gene cloning.  We reversed the rejection, holding in part that "the PTO's focus on Bell's method is misplaced.  Bell does not claim a method.  Bell claims compositions, and the issue is the obviousness of the claimed compositions, not of the method by which they are made."  Id.

And KSR does nothing to change that.

Interestingly, in view of the professed desire of the Office to have the Federal Circuit overturn Deuel, Ms. Gongola did not assert any deficiencies in Deuel per se occasioned by the Supreme Court's KSR decision:

Judge Rader:  Deuel, is it consistent with KSR?

Gongola:  Yes, your Honor, except for one small feature of KSR.  Deuel . . .

Judge Rader:  You don't seem -- the Board doesn't seem to agree with you.

Gongola:  I would disagree with that characterization.  The Board explained -- if we look at page A8 of their decision -- they don't say that KSR overturns Deuel or anything along those lines.  The Board pointed out that KSR may cast doubt on Deuel to the extent that the Federal Circuit has rejected an obvious to try test.  In KSR, the Supreme Court said that in certain circumstances, obvious to try may be the standard for obviousness.  And that's what the Board is saying here, that to the extent people are citing Deuel for the proposition [that] obvious to try can't be the standard, well . . . KSR cast doubt on that.  But otherwise, Deuel remains good law.

Judge Rader:  Well, KSR said rather persuasively on page 12 here [that] it might have been obvious to try the combination of Asano, and the big mistake, of course, was rejecting Asano, because obvious to try . . . and they cite Deuel for the mistaken methodology of not trying the combination of Asano.  Doesn't that cast a little doubt on Deuel?

Gongola:  Only to the extent that Deuel says obvious to try can't be the standard.  I think the Supreme Court went on to clarify that when there is a design need or a market identifies a problem, and there's a finite number of identified solutions, and those solutions are predictable, then obvious to try can be the standard.  And the Board here applied that principle in an alternate fashion, so we've got the obvious species grounds and separate obvious to try grounds for finding [that] Kubin's claimed invention would have been obvious.  Here the Board said that the problem was to identify or isolate NAIL cDNA, there were a limited number of methodologies to do so, and a person of skill in the art would have followed those methodologies and met with success.  The Board was correct.  Valeinte teaches one and one only one method for isolating p38 cDNA.  Matthews comes along and provides the reasonable expectation of success, showing, hey, they followed that same method, and they obtained 2B4 cDNA.

She also asserted, in response to Judge Linn's comment that KSR involved simple mechanical technology, that the Supreme Court in KSR envisioned that its principles would be applicable across other technologies.  Ms. Gongola based this assessment on dicta in KSR that the application of such principles to other technologies might not be as easy as it had been for the Court in KSR.  She also cited the Federal Circuit's Takeda Chem. Indus., Ltd. v. Alphapharm Pty., Ltd. decision as one "that helps us in this case," again asserting the "facts" before the Court in Kubin.  The Federal Circuit's decision in finding Takeda's claims non-obvious was based on the "hundreds of [related] compounds" in the prior art and the failure of the art to teach "compound b" as a desirable compound.  Here, according to Ms. Gongola:

Our facts stand in direct contrast to those facts in Takeda.  Here we have a limited number of methodologies -- we have one method being taught by Valiante's Example 12.  And Matthews evidences the expectation of success.  So, this is a classic case to apply obvious to try in a unpredictable art, and to demonstrate it will work.  That's what the Board meant at pages A8 and A9 of its decision when it applied obvious to try to these facts.

Both Kubin and amici had the opportunity to argue in their briefs the factual distinctions and misapplication of the facts in the Board's obviousness analysis.  However, unless one of the judges or their clerks digs down far enough to understand the limitations of the art, and the hindsight recognition that mouse 2B4 and NAIL are homologs, the Court may need to defer to the Office's factual findings under Dickinson v. Zurko.  And Ms. Gongola's protestations that the Office does not seek to overturn Deuel should preclude a sea change in Patent Office examination even if the court upholds the Kubin rejection. 

A decision is expected (but not guaranteed) within 90 days.

For information regarding this and other related topics, please see:
• "Kubin Panel Questions Motivation behind Reversal in New Written Description Training Materials," January 8, 2009
• "In re Kubin to Be Argued before the Federal Circuit on Thursday," January 7, 2009
• "Docs at BIO: Panel Discusses IP Strategies after KSR," June 26, 2008
• "Docs at BIO: 'Gotcha' Games Continue at USPTO," June 25, 2008
• "Docs at BIO: Representatives from JPO, EPO, SIPO, and USPTO Discuss Recent Developments in Japan, Europe, China, and the U.S.," June 22, 2008
• "Briefs for In re Kubin filed by Amgen and BIO," June 12, 2008
• "USPTO's Bruce Kisliuk Addresses ACI Conference," March 3, 2008
• "DNA Non-obviousness under Ex parte Kubin (It Gets Worse)," October 18, 2007
• "Ex parte Kubin (B.P.A.I. 2007)," July 18, 2007

Comments

[Prediction: FAIL]

*****Judge Rader then asserted that "Valiente had already produced p38," to which Ms. Rudolph reminded the Court that had been done for the protein and not for the cDNA. Specifically, she correctly asserted that "you need a starting cell library that will have a sufficient amount of that particular gene that will produce that particular protein." ****

This is all dust kicked up by Kubin. No serious biotech entity in the 21st century faces serious difficulties identifying a gene sequence for an isolated protein.

"GONGOLA: ... the Board has not identified any basis for identifying a specific binding region -- what the structure of that binding region is --"

That's inherent to the sequencing of the gene.

"...and it hasn't identified any basis for finding CD48-binding obvious."

Meaningless Rochester-style functional claiming that should be ignored. If CD48 binding was some awesome unexpected result (a la the alleged awesomeness of the isolated enantiomer in Sanofi v. Apotex) then it would matter. But not here. And particularly not when the claim covers a billion different sequences, none of which were investigated by Kubin.

What Kubin really needed to do was what Sanofi did with their entantiomer arguments: make a lot of noise about how terribly difficult it was to do ("millions of dollars, thousands of hours") and how nobody really wanted to do it until some poor guy stuck his neck sooooooooo far out to take it upon himself to do the oh-so-unlikely-to-succeed work of purifying the entantiomer according to methods that nobody ever thought would work because they had only been invented 125 years ago.

[Kevin E. Noonan]

Dear Prediction:

You have hit on the problem we cited in our discussion of the oral argument: the assumption of something that isn't a fact. Valiante (USP 5,688,690) did not purify p38, and did not teach any way to do so. Valiante produced monoclonal antibodies the old-fashioned way - by immunizing an animal with human NK cells, producing hybridomas and then testing each hybridoma for NK cell-specific cross-reactivity. He chose the antibody against p38 because it was the only one that consistently reacted with human NK cells, and then probes a Western blot of NK cell proteins to identify the putative molecular weight of the protein (hence, "p38").

Before you argue that it would have been trivial to isolate the protein once you have an antibody (may be, or maybe not, depending on the amount of p38 produced by the cells, whether this cell-surface protein could be solubilized, and if so whether a reliable N-terminal sequence (which, of course, would lack the leader sequence) could be obtained), first let's just acknowledge that Valiante did NOT possess an isolated p38 protein in the art. As Judge Rader said, Valiante provided a probe, nothing more.

Thanks for the comment.

[Dan Feigelson]

Kevin, thanks for the post. How disheartening for us prep/proc guys - even if you do everything right (hard enough in and of itself), it still may not matter b/c the judges who ultimately decide don't understand the technology. And I say this as someone who thinks that Judge Rader usually gets things right.

[Kevin E. Noonan]

Dear Dan:

Barbara Rudolph pointed out what was missing in the prior art, and if the judges were listening (or if the review the argument, or if they have dedicated clerks), then there is a chance they will realize why the PTO Emperor has no clothes.

Alternatively, there is a possibility that the coyness of Janet Gongola's presentation with regard to the status of Deuel in view of KSR might lead to a very narrow holding even if the court affirms. While this will not be the right result for Amgen, the rest of us might be able to live with a decision that a cDNA sequence is obvious if the prior art provides: an isolated and thoroughly characterized protein; a probe antibody that is immunospecific for the protein; a cell or tissue source unambiguously taught in the art to express the protein; and a prophetic method that the art establishes will predictably produce the cDNA when followed exactly.

Thanks for the comment.

[Joe Reisman]

Thanks for the helpful analysis.

Bear in mind the analytical link between Bell/Deuel (obviousness) and Lilly (written description), which seem to create a parrellel universe of sorts for nucleic acid claims. Judge Rader's concerns with Lilly and its progeny are well-documented. His questioning in this case may reflect that he sees restrictions on Bell/Deuel (even subtle ones) as a necessary step in the narrowing/elimination of Lilly. In that light, one wonders were the "industry" really wants this to come out.

[Prediction: FAIL]

Kevin:

"You have hit on the problem we cited in our discussion of the oral argument: the assumption of something that isn't a fact. Valiante (USP 5,688,690) did not purify p38"

The issue is whether enough protein was identified to move forward with the mundane steps of cloning the gene. I'm still bothered by the haze around this issue, though. You yourself wrote in the original post: ******Judge Rader then asserted that "Valiente had already produced p38," to which Ms. Rudolph reminded the Court that had been done for the protein and not for the cDNA.*****

Do you see the problem? Judge Rader asserted that Valiente **had already produced the p38**

One would expect Ms. Rudolph to reply, "No, your honor, that is a falsehood that the PTO keeps repeating. It is not true. I can not emphasize this enough. That is NOT true. Valiente had NOT produced p38."

But instead (according to your summary), Ms. Rudolph admits that "it had been done for the protein ..."

So I may ask to be pardoned for misunderstanding some of the facts in this case. But Kubin and his reps seem to be trying to keep aspects of the record less than crystal clear. There's often a good reason to do that but it can lead to lengthy disputes ... and comment threads. ;)

[Kevin E. Noonan]

Dear Prediction:

You raise a good point, although I'm not sure I think Kubin deserves the blame for covering anything up here. I hesitate to criticize counsel at oral argument, since they have to juggle what they want to say/emphasize and at the same time respond to the judges's questions. In this case, Judge Rader was coming at Kubin's counsel pretty hard for about 10 minutes with just a brief letup (so she could answer Judge Friedman's question), and I think it's understandable that you and I might be able to come up with a more succinct response in the comfort of our computer consoles.

I presume that this point was argued in the briefs. What I put in my comment and the post was based on my review of the Valiante reference.

There is an open question, however, about whether the combination of expression cloning methods + a specific antibody may make unnecessary the traditional protein purification/sequencing/degenerate probe screening methods of the past (while recognizing that the existence of databases of human genome sequence makes the older ways "easier" insofar as clear mistakes in protein sequencing should be more readily detected using in silico methods). But even if cloning has gotten more routine, in this case there was no known source of NAIL cDNA - based on the teaching away from human NK cells in Matthew (even in view of the failure of the art to recognize that 2B4 = NAIL), and the fact that if you did expression cloning with human NK cells just like Valiante said to do and the PTO maintains Matthew did with mouse cells, you would not be able to detect human NAIL cDNA-harboring clones because they were not made in "resting" human NK cells - and Matthew, Valiante and the rest of the art did not tell you that you needed to activate human NK cells (indeed, one of the characteristics of NK cells is that they are not MHC-restricted and thus are not "activated" in the same way as T cells). Despite Judge Rader's statements about how equivalent mice and humans are, they are not identical and there are many instances of surprising differences betwee the species.

Thanks for the comments.