Bristol-Myers Squibb Company v. Teva Pharmaceuticals USA, Inc

United States Court of Appeals
for the Federal Circuit
______________________

BRISTOL-MYERS SQUIBB COMPANY,
Plaintiff-Appellant,

v.

TEVA PHARMACEUTICALS USA, INC.,
Defendant-Appellee.
______________________

2013-1306
______________________

Appeal from the United States District Court for the
District of Delaware in No. 10-CV-0805, Magistrate Judge
Christopher J. Burke
______________________

ON PETITION FOR PANEL REHEARING AND
REHEARING EN BANC
______________________

WILLIAM F. LEE, Wilmer Cutler Pickering Hale and
Dorr, LLP, of Boston, Massachusetts, filed a petition for
panel rehearing and rehearing en banc for plaintiff-
appellant. With him on the petition were LAUREN B.
FLETCHER and ANDREW J. DANFORD, of Boston, Massachu-
setts, and AMY K. WIGMORE and THOMAS G. SAUNDERS, of
Washington, DC. Of counsel on the petition were PAUL H.
BERGHOFF, ALISON J. BALDWIN, and JOSHUA RICH,
McDonnell Boehnen Hulbert & Berghoff LLP, of Chicago,
Illinois.

GEORGE C. LOMBARDI, Winston & Strawn LLP, of Chi-
cago, Illinois, filed a response for defendant-appellee.
2 BRISTOL-MYERS SQUIBB COMPANY V.
TEVA PHARMACEUTICALS USA, INC.

With him on the response were LYNN MACDONALD
ULRICH, IVAN M. POULLAOS, JULIA MANO JOHNSON, and
WILLIAM P. FERRANTI.

JONATHAN E. SINGER, Fish & Richardson P.C., of Min-
neapolis, Minnesota, for amicus curiae Biotechnology
Industry Organization. With him on the brief was CRAIG
E. COUNTRYMAN, of San Diego, California.

HOWARD W. LEVINE, Finnegan, Henderson, Farabow,
Garrett & Dunner, LLP, of Washington, DC, for amicus
curiae Bay Area Bioscience Association. With him on the
brief was JENNIFER S. SWAN, of Palo Alto, California.

ROY F. WALDRON, Pfizer Inc., of New York, New York,
for amicus curiae Pfizer Inc. With him on the brief were
JEFFREY J. OELKE, LESLIE MORIOKA, and ROBERT E.
COUNIHAN, White & Case LLP, of New York, New York.

STEVEN P. CALTRIDER, Eli Lilly and Company, of Indi-
anapolis, Indiana, for amicus curiae Eli Lilly and Compa-
ny.

CARTER G. PHILLIPS, Sidley Austin LLP, of Washing-
ton, DC, for amicus curiae Pharmaceutical Research and
Manufacturers of America. With him on the brief were
JEFFREY P. KUSHAN, RYAN C. MORRIS, and JAMES A. HIGH,
JR.

ROBERT M. ISACKSON, Orrick, Herrington & Sutcliffe
LLP, of New York, New York, for amicus curiae Intellec-
tual Property Owners Association. With him on the brief
were ELIZABETH A. HOWARD and T. VANN PEARCE. Of
counsel on the brief were PHILIP S. JOHNSON and KEVIN H.
RHODES, Intellectual Property Owners Association, of
Washington, DC. Of counsel was HERBERT C. WAMSLEY,
JR.

NICHOLAS G. BARZOUKAS, Baker Botts L.L.P., of Hou-
ston, Texas, for amicus curiae Merck Sharp & Dohme
BRISTOL-MYERS SQUIBB COMPANY V. 3
TEVA PHARMACEUTICALS USA, INC.

Corp. With him on the brief was JOSHUA DAVIS. Of coun-
sel on the brief were WILLIAM KROVATIN and GERARD M.
DEVLIN, Merck Sharp & Dohme Corp., of Rahway, New
Jersey.
______________________
Before PROST, Chief Judge, NEWMAN, PLAGER 1, LOURIE,
DYK, MOORE, O’MALLEY, REYNA, WALLACH, TARANTO,
CHEN, and HUGHES, Circuit Judges.
DYK, Circuit Judge, with whom WALLACH, Circuit
Judge, joins, concurs in the denial of the petition for
rehearing en banc.
O’MALLEY, Circuit Judge, concurs in the denial of the
petition for rehearing en banc.
NEWMAN, Circuit Judge, with whom LOURIE and
REYNA, Circuit Judges, join, dissents from the denial of
the petition for rehearing en banc.
TARANTO, Circuit Judge, with whom LOURIE and
REYNA, Circuit Judges, join, dissents from the denial of
the petition for rehearing en banc.
PER CURIAM.
ORDER
A combined petition for panel rehearing and rehear-
ing en banc was filed by plaintiff-appellant Bristol-Meyers
Squibb Company, and a response thereto was invited by
the court and filed by defendant-appellee Teva Pharma-
ceuticals USA, Inc. The petition for rehearing and re-
sponse were referred to the panel that heard the appeal,
and thereafter, the petition for rehearing en banc and
response were referred to the circuit judges who are
authorized to request a poll of whether to rehear the
appeal en banc. A poll was requested, taken, and failed.

Upon consideration thereof,

IT IS ORDERED THAT:
4 BRISTOL-MYERS SQUIBB COMPANY V.
TEVA PHARMACEUTICALS USA, INC.

(1) The petition for panel rehearing is denied.

(2) The petition for rehearing en banc is denied.

(3) The mandate of the court will issue on October 27,
2014.

FOR THE COURT

October 20, 2014 /s/ Daniel E. O’Toole
Date Daniel E. O’Toole
Clerk of Court

1 Circuit Judge Plager participated only in the deci-
sion on the petition for panel rehearing.
 United States Court of Appeals
for the Federal Circuit
______________________

BRISTOL-MYERS SQUIBB COMPANY,
Plaintiff-Appellant,

v.

TEVA PHARMACEUTICALS USA, INC.,
Defendant-Appellee.
______________________

2013-1306
______________________

Appeal from the United States District Court for the
District of Delaware in No. 10-CV-0805, Magistrate Judge
Christopher J. Burke.
______________________
DYK, Circuit Judge, with whom WALLACH, Circuit Judge,
joins, concurring in the denial of the petition for rehearing
en banc.
This case presents a question of obviousness, in par-
ticular whether evidence postdating the invention can be
used to establish unexpected results. The panel holds
that it cannot be considered in the circumstances of this
case. That position is correct. It is mandated by the
statute, which provides that an invention is not patenta-
ble if it “would have been obvious before the effective filing
date of the claimed invention to a person having ordinary
skill in the art to which the claimed invention pertains.”
35 U.S.C. § 103 (emphasis added).
2 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

The patent applicant’s discovery of unexpected results
at the time of the invention can help to establish that the
invention would not have been obvious to another skilled
person. But hindsight bias must be avoided in determin-
ing obviousness. And under longstanding Supreme Court
authority, the pertinent knowledge is that possessed at
the time of the invention. See KSR Int’l Co. v. Teleflex
Inc., 550 U.S. 398, 416 (2007) (focusing on “[w]hen Adams
designed his battery” and noting that “[t]he fact that the
elements worked together in an unexpected and fruitful
manner supported the conclusion that Adams’ design was
not obvious to those skilled in the art.”) (citing United
States v. Adams, 383 U.S. 39 (1966)); Ball & Socket Fas-
tener Co. v. Kraetzer, 150 U.S. 111, 116–17 (1893) (dis-
counting an advantage of a patented invention that “was
not originally within the contemplation of the patentee,
but is an afterthought”); see also Genetics Inst., LLC v.
Novartis Vaccines and Diagnostics, Inc., 655 F.3d 1291,
1315 (Fed. Cir. 2011) (Dyk, J., dissenting). This decision
properly does not allow consideration of post-invention
evidence in the circumstances of this case. There is no
basis for rehearing en banc.
 United States Court of Appeals
for the Federal Circuit
______________________

BRISTOL-MYERS SQUIBB COMPANY,
Plaintiff-Appellant,

v.

TEVA PHARMACEUTICALS USA, INC.,
Defendant-Appellee.
______________________

2013-1306
______________________

Appeal from the United States District Court for the
District of Delaware in No. 10-CV-0805, Magistrate Judge
Christopher J. Burke.
______________________

O’MALLEY, Circuit Judge, concurring in the denial of the
petition for rehearing en banc.
I concur in the court’s denial of the petition for rehear-
ing en banc. I write to assuage Bristol-Myers Squibb Co.’s
(“BMS”) and the amici’s 1 fears that this panel decision
has rewritten the test for obviousness for pharmaceutical
patents. In my view, the concerns expressed are unjusti-

1 Merck Sharp & Dohme Corp., Intellectual Proper-
ty Owners Association, Pharmaceutical Research and
Manufacturers of America, Eli Lilly & Co., Pfizer Inc.,
Biotechnology Industry Organization, and Bay Area
Bioscience Association (collectively, “the amici”).
2 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

fied and mischaracterize the opinion. This case does not
forge new ground or set down immutable principles. It
simply decides that, on the record before it, the district
court did not err in finding the asserted claim of the ’244
Patent invalid as obvious.
As the panel opinion explains, an invention is un-
patentable when “the differences between the subject
matter sought to be patented and the prior art are such
that the subject matter as a whole would have been
obvious at the time the invention was made to a person
having ordinary skill in the art to which said subject
matter pertains.” 35 U.S.C. § 103(a) (2006). 2 Obviousness
is a question of law based on the following underlying
factual findings: (1) the level of ordinary skill in the art;
(2) the scope and content of the prior art; (3) the differ-
ences between the claims and the prior art; and (4) objec-
tive indicia of nonobviousness, such as commercial
success, long-felt but unmet needs, failure of others, and
unexpected results. KSR Int’l Co. v. Teleflex, Inc., 550
U.S. 398, 406 (2007); Graham v. John Deere Co., 383 U.S.
1, 17–18 (1966).
In this case, the panel affirmed the district court’s
conclusion that entecavir, BMS’s antiviral compound used
to treat hepatitis B, was invalid as obvious. 3 The panel
found the record supported the selection of 2’-CDG as a
lead compound and the conclusion that one of ordinary
skill in the art would have been motivated to modify 2’-
CDG in such a way to arrive at the patented compound,

2 Because this invention was filed before the adop-
tion of the America Invents Act, the prior version of § 103
governs.
3 Specifically, the district court found that claim 8
of U.S. Patent No. 5,206,244 was invalid as obvious. See
Bristol-Myers Squibb Co. v. Teva Pharm. USA, Inc., 923
F. Supp. 2d 602, 686 (D. Del. 2013).
BRISTOL-MYERS SQUIBB COMPANY v. 3
TEVA PHARMACEUTICALS USA, INC.

entecavir. Bristol-Myers Squibb Co. v. Teva Pharm. USA,
Inc., 752 F.3d 967, 975 (Fed. Cir. 2014). The panel then
agreed with the district court that, despite some evidence
of objective indicia demonstrating non-obviousness, the
totality of the evidence supported the conclusion that
entecavir was obvious. Id. at 979.
In the petition for rehearing en banc and the support-
ing amicus briefs, BMS and the amici claim that, in
reaching its judgment, the panel dramatically altered the
jurisprudential landscape governing obviousness claims
in pharmaceutical cases. And they predict that dire
consequences will flow therefrom. They express concern
about (1) the panel’s treatment of post-invention evidence
regarding the differences between the prior art and the
invention, specifically when determining if a skilled
artisan would have been motivated to make the claimed
compound with a reasonable expectation of success for its
therapeutic use; (2) the panel’s description of what consti-
tutes an unexpected result in the pharmaceutical context;
(3) the party upon whom it placed the burden of proof at
certain stages of its obviousness inquiry; and (4) the way
in which the panel weighed the evidence of objective
indicia of non-obviousness.
BMS and the amici first contend that the panel im-
properly limits consideration of evidence regarding the
properties of the invention and the prior art to those
known at the time of the invention. Specifically, BMS
and the amici argue the panel forecloses the possibility of
reviewing later-discovered differences between the prior
art and the claimed invention by requiring these differ-
ences to be unexpected “by one of ordinary skill in the art
at the time of the invention.” Id. at 977 (emphasis added).
BMS and the amici allege that the panel erred by not
considering later-discovered unexpected results, and now
closes the door to all reference to such evidence. I disa-
gree.
4 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

Our case law clearly allows the consideration of later-
discovered differences between the prior art and the
invention. See Sanofi-Aventis Deutschland GmbH v.
Glenmark Pharm., Inc., 748 F.3d 1354, 1360 (Fed. Cir.
2014) (“Glenmark also argues that later-discovered bene-
fits cannot be considered in an obviousness analysis . . . .
That is incorrect; patentability may consider all of the
characteristics possessed by the claimed invention, when-
ever those characteristics become manifest.”); Genetics
Inst., LLC v. Novartis Vaccines & Diagnostics, Inc., 655
F.3d 1291, 1307 (Fed. Cir. 2011) (“[E]vidence of unex-
pected results may be [considered] . . . even if that evi-
dence was obtained after the patent’s filing or issue
date.”); Knoll Pharm. Co. v. Teva Pharm. USA, Inc., 367
F.3d 1381, 1385 (Fed. Cir. 2004) (“Evidence developed
after the patent grant is not excluded from consideration,
for understanding of the full range of an invention is not
always achieved at the time of filing the patent applica-
tion.”). These differences inform the obviousness analysis
and thus can be considered when assessing what was
understood by one of skill in the art at the time of the
invention and what expectations may have been reasona-
ble.
Like all evidence of objective indicia, the point of con-
sidering later-understood evidence regarding the proper-
ties of the invention is to guard against hindsight bias by
assessing claims of a motivation to combine as of the time
of invention in light of later surprises or developments.
KSR Int’l Co., 550 U.S. at 421; see Graham, 383 U.S. at
36; cf. Sinclair Refining Co. v. Jenkins Petroleum Process
Co., 289 U.S. 689, 697–98 (1932) (“The law will make the
best appraisal that it can, summoning to its services
whatever aids it can command [to assess a claimed
knowledge base or expectation]. . . . [I]f years have gone
by before the evidence is offered[,] [later acquired
e]xperience is then available to correct uncertain prophe-
cy. Here is a book of wisdom that courts may not neglect.
BRISTOL-MYERS SQUIBB COMPANY v. 5
TEVA PHARMACEUTICALS USA, INC.

We find no rule of law that sets a clasp upon its pages and
forbids us to look within.”). The panel opinion could not
rewrite this precedent even if it wanted to; in this case, I
see no evidence it sought to do so.
The line of the opinion to which BMS and the amici
refer simply notes that the inquiry into what one of skill
in the art understood and reasonably expected must be
fixed as of the time of the invention. It does not say only
properties of the invention known at the time of the
invention can be considered for purposes of informing that
inquiry. Indeed, as we have said repeatedly over the
years, post-issuance evidence regarding objective indicia
of non-obviousness may often be the most probative and
cogent evidence in the record. Stratoflex, Inc. v. Aeroquip
Corp., 713 F.2d 1530, 1538 (Fed. Cir. 1983). This is
especially true where the post-issuance evidence relates to
unexpected results. See Sanofi-Aventis, 748 F.3d at 1360.
Apparently recognizing that the panel opinion does
not expressly purport to change the law, BMS and the
amici argue that, by adopting the district court’s finding
that 2’-CDG was considered safe and non-toxic at the time
of the invention, despite evidence that it was later deter-
mined to be toxic, the panel implicitly condoned the
exclusion of evidence regarding later-discovered proper-
ties, in this and future cases. There is a distinction
between limiting the obviousness inquiry to pre-invention
evidence and finding post-invention evidence unpersua-
sive, however. See Allergan Inc. v. Sandoz Inc., 726 F.3d
1286, 1293 (Fed. Cir. 2013) (“We agree with the court’s
finding that this result was unexpected. However, we do
not find that these unexpected results are sufficient to
outweigh the other evidence of obviousness.”).
What the district court found was that the later evi-
dence of 2’-CDG’s toxicity was insufficient to overcome the
strong evidence that researchers at the time had a moti-
vation to start with 2’-CDG as the lead compound and
6 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

modify it in such a way as to make entecavir. The court
then cited a host of evidence in the record to support that
conclusion. BMS and the amici say that the trial court’s
conclusion cannot have been correct because 2’-CDG was
later shown to be toxic. They argue that no medicinal
chemist could have had a reasonable expectation of suc-
cess from use of 2’-CDG as a lead compound because they
could not have known if any modification to it would be
safe for human use.
As the district court pointed out, BMS did not ques-
tion the reasonableness of a skilled artisan’s expectation
on these grounds until its reply brief before the trial court
and, thus, arguably waived that argument. Bristol-Myers
Squibb Co., 923 F. Supp. 2d at 674 n.36. And, the district
court rejected the merits of BMS’s argument, finding that
the level of 2’-CDG’s cytotoxicity was not known at the
time of the invention and that tentative concerns about
the toxicity did not stop researchers from using 2’-CDG as
a starting point. As the district court said, “the best
indication that any such tentative references to possible
toxicity did not stop the medicinal chemist from selecting
2’-CDG as a lead compound in the late 1980s and 1990, in
light of its positive benefits, is the fact that researchers
were actually treating and using 2’-CDG as a lead com-
pound during the relevant time period.” Id. at 662.
Indeed, BMS’s own expert, Dr. Bud Tennant, testified
that, during his experiments investigating the effects of
2’-CDG against the woodchuck hepatitis virus—which
occurred after entecavir’s invention—he was surprised to
find 2’-CDG was toxic. Id. at 623–24.
While the later findings regarding 2’-CDG’s toxicity
certainly make claims regarding the reasonableness of
any expectation of success less credible, on this record, the
panel did not act beyond the pale in concluding that the
district court’s factual conclusion regarding the existence
BRISTOL-MYERS SQUIBB COMPANY v. 7
TEVA PHARMACEUTICALS USA, INC.

of a reasonable expectation of success was not clearly
erroneous. 4 The panel’s decision to affirm the district
court’s findings does not foreclose the possibility that
post-invention evidence regarding the properties of either
the invention or the prior art might be persuasive in the
appropriate case. BMS simply did not make a record
which would support the conclusion that they were in this
case.
BMS and the amici next contend that the panel inap-
propriately discounted the significance of unexpected
results when the panel parsed unexpected results into
“differences in kind” and “differences in degree.” They
argue that the panel’s treatment of entecavir’s unexpected
properties as mere differences in degree from 2’-CDG’s
properties diminishes the potentially meaningful distinc-
tions between two compounds by reducing the nuanced
unexpected results inquiry to a question of degree versus
kind. According to BMS and the amici, this characteriza-
tion of what may be considered unexpected results creates
impossible hurdles for the pharmaceutical industry to
overcome, where slight differences between compounds
can translate into life or death for a patient.
In its discussion of unexpected results, the panel ex-
plained that “[w]hen assessing unexpected properties . . .
we must evaluate the significance and ‘kind’ of expected
results along with the unexpected results.” Bristol-Myers
Squibb Co., 752 F.3d at 977. This statement is consistent
with our precedent that one should consider the substan-

4 While I agree with all of the concerns thoughtfully
expressed by Judge Taranto, as he acknowledges, the
current record does not permit us to reach those concerns.
BMS did not argue that there was insufficient evidence to
indicate that a modification of 2’-CDG would be therapeu-
tically effective, and there is no evidence in the record
that skilled artisans at the time doubted that it would be.
8 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

tiality of the differences between the properties of the
prior art and those of the invention to determine the
significance of those differences. See In re Soni, 54 F.3d
746, 751 (Fed. Cir. 1995) (“Mere improvement in proper-
ties does not always suffice to show unexpected results. . .
. [W]hen an applicant demonstrates substantially im-
proved results . . . and states that the results were unex-
pected, this should suffice to establish unexpected results
in the absence of evidence to the contrary.”); In re Chupp,
816 F.2d 643, 646 (Fed. Cir. 1987) (“[T]he mere submis-
sion of some evidence that a new compound possesses
some unpredictable properties does not require an auto-
matic conclusion of nonobviousness in every case.”); In re
Merck, 800 F.2d 1091, 1099 (Fed. Cir. 1986) (“In the
absence of evidence to show that the properties of the
compounds differed in such an appreciable degree that
the difference was really unexpected, we do not think that
the Board erred in its determination.”); In re Corkill, 771
F.2d 1496, 1501 (Fed. Cir. 1985) (“A greater than expected
result is an evidentiary factor pertinent to the legal
conclusion of [] obviousness.”).
While reading the panel’s statement that a “‘mere dif-
ference in degree’ is insufficient” to render a compound
patentable out of context admittedly could lead to some
confusion, the panel’s entire discussion of unexpected
results makes clear that one must consider the extent of
the differences between properties of the prior art and the
invention to determine the weight such evidence should
be given in the obviousness analysis. The reference to
differences in kind versus differences in degree was
merely illustrative of how one can assess unexpected
properties—it was not essential to the panel’s finding that
it would defer to the district court’s factual finding that
the results upon which BMS relied were not truly unex-
pected or substantial. Bristol-Myers Squibb Co., 752 F.3d
at 978. Accordingly, I do not believe the panel’s mere use
of the phrasing to which BMS objects inappropriately
BRISTOL-MYERS SQUIBB COMPANY v. 9
TEVA PHARMACEUTICALS USA, INC.

reduces the question of unexpected results to a purely
mechanical application of degree versus kind.
BMS and the amici next contend that the panel en-
dorsed the use of a burden-shifting framework, wherein
the burden shifted to the patentee once the alleged in-
fringer established a prima facie case of obviousness. The
panel neither used nor endorsed a burden-shifting analy-
sis, however; it said explicitly that it was employing a
holistic approach to obviousness. Id. at 976–77 (consider-
ing Teva’s strong evidence of obviousness alongside BMS’s
arguments relating to secondary considerations of nonob-
viousness); id. at 977 (explaining that “[s]econdary con-
siderations of nonobviousness ‘must always when present
be considered’’’ (quoting In re Cyclobenzaprine Hydrochlo-
ride Extended-Release Capsule Patent Litig., 676 F.3d
1063, 1075–76, 1079 (Fed. Cir. 2012))). 5 Again, the amici
and BMS see ghosts that are simply not there.
Lastly, BMS and the amici argue that the panel deci-
sion supports comparing the objective indicia of non-
obviousness against one another, allowing some to offset
others. As the panel itself stated, “[it] under[stood] the
district court to be noting that some categories of [objec-
tive indicia] evidence simply were not as helpful to BMS’s
case as others. [The panel did] not read the opinion as
suggesting that unhelpful evidence somehow diminished

5 Indeed, the district court made clear that it too
understood the importance of objective considerations in
the obviousness inquiry. Bristol-Myers Squibb Co., 923 F.
Supp. 2d at 675 (noting that objective evidence of non-
obviousness must be “considered collectively” with evi-
dence of obviousness, may not be “after-the-fact consider-
ations” and may not be “relegated to ‘secondary status’”)
(citing In re Cyclobenzaprine Hydrochloride, 676 F.3d at
1078). It just disagreed with the weight BMS asked that
they be given in this case.
10 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

the strength of the more persuasive forms of evidence.”
Id. at 979. Here, BMS and the amici simply mischarac-
terize the panel opinion.
Ultimately, a case is won or lost on the record. At the
district court, BMS’s own expert, Dr. Schneller, conceded
that 2’-CDG would have been considered as a lead com-
pound by one skilled in the art and acknowledged that 2’-
CDG was actually being used as a lead compound at the
time of entecavir’s invention. Bristol-Myers Squibb Co.,
923 F. Supp. 2d at 663 (“BMS’s expert, Dr. Schneller, . . .
repeatedly testified at trial that [SRI and Glaxo] chemists
were, in fact, treating and using 2’-CDG as a lead com-
pound” during the time of the invention.); id. at 664
(“[T]he testimony of [BMS’s] own expert at trial repeated-
ly and conclusively established that researchers were, in
fact, treating and using 2’-CDG as a lead compound in the
relevant time period.”). Additionally, Dr. Schneller ad-
mitted that a skilled artisan could have been led to modify
2’-CDG in such a way to arrive at the claimed invention.
Id. at 665 (“Dr. Schneller . . . agreed that when a lead
compound is selected, a chemist would seek to make
conservative changes to that structure.”); id. at 670–71
(Dr. Schneller stated in his expert report and on cross
examination that in light of the prior art a skilled artisan
could have been led to substitute an exocyclic methylene
group at the 5 prime position of 2’-CDG). In light of these
admissions, the district court concluded that, in “almost
every significant portion of [its] case, Teva’s position was
not only bolstered by the opinion of its expert, Dr. Heath-
cock, but also by the testimony of BMS’s expert, Dr.
Schneller.” Id. at 686.
Despite the testimony of its own expert below, on ap-
peal, BMS originally focused on the district court’s alleged
error in concluding that a skilled artisan would have
selected 2’-CDG as a lead compound and modified it to
arrive at the claimed invention. Reference to evidence of
entecavir’s unexpected properties was raised almost as an
BRISTOL-MYERS SQUIBB COMPANY v. 11
TEVA PHARMACEUTICALS USA, INC.

afterthought in BMS’s opening brief, as was its focus on
the reasonable expectation of success prong of the motiva-
tion to combine inquiry before the district court. Now,
BMS and the amici adopt a “sky is falling” approach to
what is simply a fact dependent opinion. The opinion
makes no dramatic changes to the law, closes no doors on
what evidence may be considered in undertaking an
obviousness inquiry, establishes no hard and fast tests for
what results might be considered unexpected in a case
involving a pharmaceutical compound, and does not
improperly shift the burden of proof or denigrate the
importance of objective indicia of non-obviousness. 6 On
this record and upon a fair reading of the panel opinion, I
do not believe en banc consideration is warranted.

6 I do not discount the fact that, in dissent from this
denial of en banc, Judges Newman, Lourie, and Reyna are
concerned that the panel opinion did go too far and that
Judge Taranto at least believes the opinion can be read—
fairly or not—as having done so. We all agree on the law,
we simply disagree whether this opinion is sufficiently at
odds with it to warrant en banc consideration.
 United States Court of Appeals
for the Federal Circuit
______________________

BRISTOL-MYERS SQUIBB COMPANY,
Plaintiff-Appellant,

v.

TEVA PHARMACEUTICALS USA, INC.,
Defendant-Appellee.
______________________

2013-1306
______________________

Appeal from the United States District Court for the
District of Delaware in No. 10-CV-0805, Magistrate Judge
Christopher J. Burke.
______________________

NEWMAN, Circuit Judge, with whom LOURIE and REYNA,
Circuit Judges, join, dissenting from the denial of the
petition for rehearing en banc.
This appeal concerns a patent owned by Bristol-Myers
Squibb Company on the product entecavir, a medicinal
product for treatment of hepatitis B. Litigation arose
under the provisions of the Hatch-Waxman Act, upon the
filing by Teva Pharmaceuticals USA of an Abbreviated
New Drug Application Paragraph IV Certification. A
2 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

panel of this court held the patent invalid, 1 creating
several new standards for determination of obviousness.
For example, the court deemed it irrelevant to the obvi-
ousness determination that the prior art “lead compound,”
the carbocyclic analog of 2′-deoxyguanosine (2′-CDG), is
highly toxic to humans, whereas the new product
entecavir is non-toxic. Bristol-Myers requests rehearing
en banc, arguing that the court has misapplied statute
and precedent.
This case has aroused extensive commentary, particu-
larly in the chemical and pharmaceutical fields; the
request for rehearing en banc is supported by much of the
nation’s research-based industry, which has filed briefs as
amici curiae to point out the disincentives and uncertain-
ties flowing from the court’s rulings. As summarized by
amicus Intellectual Property Owners Association:
This decision introduces substantial uncertainty
into what appeared to be a clear legal standard;
allowing this uncertainty to fester would affect
countless pending and future cases. Obviousness
is an issue in most patent examinations, litiga-
tions, and administrative proceedings. Particular-
ly in unpredictable chemical and pharmaceutical
fields, unexpected results evidencing differences
and objective considerations can tip the balance
between obviousness and non-obviousness. Pa-
tent owners would benefit from the certainty of an
en banc ruling on when and how later-discovered
differences between an invention and prior art
may be considered in the obviousness analysis.
Brief of Amicus Curiae Intellectual Property Owners
Association in Support of Rehearing En Banc at 6-7.

1 Bristol-Myers Squibb Co. v. Teva Pharm. USA,
Inc., 752 F.3d 967 (Fed. Cir. 2014).
BRISTOL-MYERS SQUIBB COMPANY v. 3
TEVA PHARMACEUTICALS USA, INC.

I outline the several conflicts with precedent here
produced; for until this case, inventors could confidently
establish patentability of a new product or a new use by
showing that the new property or use was unexpected in
light of the prior art.
1. Restriction on comparative data showing unex-
pected properties
The court held that entecavir’s unexpected properties
did not render it nonobvious in patent terms because
“additional unexpected properties, however, did not upset
an already established motivation to modify a prior art
compound based on the expected properties of the result-
ing compound.” Bristol-Myers, 752 F.3d at 976. The
court’s hindsight decision that Bristol-Myers merely
“ma[de] the minor modification to arrive at entecavir,” id.
at 973, while ignoring the unexpected differences in
properties between entecavir and the prior art compound,
conflicts with the entirety of precedent on the law of
obviousness.
When a new product (or device or method) is discov-
ered, its nonobviousness in patent terms often is demon-
strated by evidence of whether the new product (or device
or method) possesses properties not possessed by similar
products. The mechanism for providing this evidence is
the submission of comparative data in affidavits or decla-
rations filed pursuant to USPTO Rule 132, 37 C.F.R.
§1.132. Such data may involve new experiments per-
formed on the invention and the prior art for purposes of
comparison, and information already known although not
in comparative form. See, e.g., In re Chupp, 816 F.2d 643,
644 (Fed. Cir. 1987) (“To rebut the prima facie case of
obviousness, Chupp submitted a declaration discussing
the results of tests comparing the herbicidal activity of
the claimed compound with that of the closest prior art
compounds and with two commercial herbicides. . . . It is
undisputed that the claimed compound gave superior
4 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

results . . . .”); In re Payne, 606 F.2d 303, 306, 316 (CCPA
1979) (in response to an obviousness rejection based on
prior art, the inventor provided data that “purportedly
establishes an unexpectedly superior scope and level of
pesticidal activity of the claimed compounds in a compari-
son of the most representative compound of” the prior
art).
Such comparative data need not have been previously
available or known to the art at the time of the invention.
In In re Miller, 197 F.2d 340, 342 (CCPA 1952), the court
called for the “making of comparative tests” if needed to
support unexpected results. This is established practice.
See, e.g., In re Orfeo, 440 F.2d 439, 441 (CCPA 1971)
(applicants “have the right to have considered the Rule
132 affidavit which allegedly shows new and unexpected
results”); In re Khelghatian, 364 F.2d 870, 872 (CCPA
1966):
There appears to be agreement of the parties that
essential to the proper resolution of this issue is a
consideration of all the record evidence, including
an affidavit filed under Rule 132. Such has been
the law in this court for several years, and that
regardless of whether any “doubt” as to patenta-
bility exists upon an examination of the prior art
alone.
Precedent is clear that the information and comparative
data presented as evidence of nonobviousness need not
have existed before the patent application was filed. See,
e.g., Genetics Inst., LLC v. Novartis Vaccines & Diagnos-
tics, Inc., 655 F.3d 1291, 1307 (Fed. Cir. 2011):
[I]t would be error to prohibit a patent applicant
or patentee from presenting relevant indicia of
nonobviousness, whether or not this evidence was
available or expressly contemplated at the filing of
the patent application.
BRISTOL-MYERS SQUIBB COMPANY v. 5
TEVA PHARMACEUTICALS USA, INC.

See also Knoll Pharm. Co., Inc. v. Teva Pharm. USA, Inc.,
367 F.3d 1381, 1385 (Fed. Cir. 2004):
There is no requirement that an invention’s prop-
erties and advantages were fully known before the
patent application was filed, or that the patent
application contains all of the work done in study-
ing the invention, in order for that work to be in-
troduced into evidence in response to litigation
attack. Nor is it improper to conduct additional
experiments and provide later-obtained data in
support of patent validity.
Information learned after the patent application was filed
may provide evidence of unexpected or unpredicted prop-
erties. E.g., In re Zenitz, 333 F.2d 924, 925, 927 (CCPA
1964) (later discovered hypotensive and tranquilizing
properties that were not described in the specification
could render the claimed compounds nonobvious and thus
patentable); Sanofi-Aventis Deutschland GmbH v. Glen-
mark Pharm. Inc., USA, 748 F.3d 1354, 1360 (Fed. Cir.
2014):
Glenmark also argues that later-discovered bene-
fits cannot be considered in an obviousness analy-
sis, here referring to the improved kidney and
blood vessel function that were observed after the
patent application was filed. That is incorrect; pa-
tentability may consider all of the characteristics
possessed by the claimed invention, whenever
those characteristics become manifest.
Comparisons of newly found properties of both the inven-
tion and the prior art are routinely presented as evidence
in determinations of obviousness. In Leo Pharmaceutical
Products, LTD. v. Kappos, 726 F.3d 1346 (Fed. Cir. 2013),
the patentee during reexamination conducted tests of the
prior art and showed that the reference formulations
resulted in significant degradation of the vitamin D
analog and corticosteroid. In considering this post-
6 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

invention testing of the prior art, this court stated “[t]hese
test results are a strong indication that the ’013 patent’s
combination of known elements yields more than just
predictable results,” and reversed the Board’s obviousness
determination. Id. at 1358.
The provision of comparative data, whether or not the
data were available before the patent application was
filed, is long-established practice. See In re Payne, 606
F.2d at 315-16 (“A prima facie case of obviousness based
on structural similarity is rebuttable by proof that the
claimed compounds possess unexpectedly advantageous
or superior properties. Direct or indirect comparative
testing between the claimed compounds and the closest
prior art may be necessary.” (citing In re Papesch, 315
F.2d 381, 386-87 (CCPA 1963))); In re Merchant, 575 F.2d
865, 869 (CCPA 1978) (“An applicant relying upon a
comparative showing to rebut a prima facie case must
compare his claimed invention with the closest prior
art.”); In re Miller, 197 F.2d 340, 342 (CCPA 1952)
(“Where, as here, results superior to those produced by
the references of the prior art, or public knowledge and
use, constitute the basis for the claim of invention, the
making of comparative tests and the establishment of the
unexpected and superior results never before attained
must be established by a proper showing.”).
Despite this overwhelming precedent, the court de-
clined to weigh that the designated lead compound 2′-
CDG is highly toxic and concededly is useless in treating
hepatitis B. Such information cannot be ignored, alt-
hough it was not observed until the prior art compound
was tested in mammals, after the entecavir patent appli-
cation was filed. This does not render this unexpected
difference irrelevant to patentability, as a matter of law
or logic. In In re Papesch the court explained:
From the standpoint of patent law, a compound
and all of its properties are inseparable; they are
BRISTOL-MYERS SQUIBB COMPANY v. 7
TEVA PHARMACEUTICALS USA, INC.

one and the same thing. . . . [T]he thing that is
patented is not the formula but the compound
identified by it. And the patentability of the thing
does not depend on the similarity of its formula to
that of another compound but of the similarity of
the former compound to the latter. There is no
basis in law for ignoring any property in making
such a comparison.
315 F.2d at 391.
Our colleagues in concurrence disregard the entirety
of precedent and practice. The amici curiae protest the
court’s changes of law and understanding as confusing,
unnecessary, and contrary to the public interest in devel-
opment of useful and beneficial new products. Amicus
curiae the Biotechnology Industry Organization reminds
the court of the new statutory pressures for early filing,
now that the patent goes to the first inventor to file, not
the first to invent:
The AIA’s new first-to-file system puts pressure
on companies to file early, lest they lose priority.
But, under the panel’s approach, innovators might
be better off waiting, in case new, unexpected dif-
ferences between the invention and prior art come
to light during clinical testing. There is no reason
to put innovators to that difficult choice.
Brief of Amicus Curiae Biotechnology Industry Organiza-
tion in Support of Rehearing En Banc at 7.
If there is now to be a major restriction on the evi-
dence that can be adduced in support of patentability of
new and improved products, such change of law should be
determined en banc.
2. The misapplication of “secondary considerations”
Information about the “secondary considerations” of
nonobviousness, see Graham v. John Deere Co., 383 U.S. 1
8 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

(1966), is often based on post-filing knowledge and data.
Such information includes “commercial success, long felt
but unsolved needs, failure of others, etc.,” id. at 17, and
tends to become manifest after the patent application is
filed and the invention is used. “Evidence of secondary
considerations may often be the most probative and
cogent evidence in the record.” Stratoflex, Inc. v. Aeroquip
Corp., 713 F.2d 1530, 1538 (Fed. Cir. 1983).
Emphasizing the characteristics of medicinal and bio-
logical products, amicus curiae Pharmaceutical Research
and Manufacturers of America observes that the objective
indicia of nonobviousness do not come into existence until
after the patent application was filed. If there is now to
be some restriction on reliance on such information, it
should be clearly stated and contrary precedent should be
overruled.
3. The holding that an unexpected property is insuf-
ficient “by itself” to show nonobviousness
The court stated that “unexpected results do not per
se defeat, or prevent, the finding that a modification to a
lead compound will yield expected, beneficial properties.”
Bristol-Myers, 752 F.3d at 976. The court further stated
that “an unexpected result or property does not by itself
support a finding of nonobviousness.” Id. To the contra-
ry, an unexpected result or property is the touchstone of
nonobviousness.
Although the court recognized that entecavir has the
“unexpected properties [of]: (1) high potency against
hepatitis B, (2) a larger than expected therapeutic win-
dow, and (3) a high genetic barrier to resistance,” id. at
977, the court held that these unexpected properties were
expected because the prior art had these properties to
some failed extent. The court postulated that entecavir’s
non-toxicity was “not unexpected in light of the structur-
ally similar 2′-CDG,” id. at 978, although the toxicity of 2′-
BRISTOL-MYERS SQUIBB COMPANY v. 9
TEVA PHARMACEUTICALS USA, INC.

CDG was so high that it was abandoned as a potential
treatment for hepatitis B.
Precedent directly contradicts the court’s position. An
unexpected property “by itself” can, indeed, support a
finding of nonobviousness. This court stated in In re Soni:
One way for a patent applicant to rebut a
prima facie case of obviousness is to make a show-
ing of “unexpected results,” i.e., to show that the
claimed invention exhibits some superior property
or advantage that a person of ordinary skill in the
relevant art would have found surprising or unex-
pected. The basic principle behind this rule is
straightforward—that which would have been
surprising to a person of ordinary skill in a par-
ticular art would not have been obvious. . . .
[W]hen an applicant demonstrates substantially
improved results, as Soni did here, and states that
the results were unexpected, this should suffice to
establish unexpected results in the absence of evi-
dence to the contrary.
54 F.3d 746, 750-51 (Fed. Cir. 1995).
This principle is—or was—beyond dispute, as illus-
trated on a vast variety of facts. See, e.g., Procter &
Gamble Co. v. Teva Pharm. USA, Inc., 566 F.3d 989, 994
(Fed. Cir. 2009) (“If a patent challenger makes a prima
facie showing of obviousness, the owner may rebut based
on ‘unexpected results’ by demonstrating ‘that the
claimed invention exhibits some superior property or
advantage that a person of ordinary skill in the relevant
art would have found surprising or unexpected.’ (quoting
In re Soni, 54 F.3d at 750)); Kao Corp. v. Unilever U.S.,
Inc., 441 F.3d 963, 969-70 (Fed. Cir. 2006) (affirming
nonobviousness over prima facie case based solely on
evidence of unexpected results); In re Geisler, 116 F.3d
1465, 1469 (Fed. Cir. 1997) (observing that “a prima facie
case of obviousness can be rebutted if the applicant (1)
10 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

can establish ‘the existence of unexpected properties in
the range claimed’ or (2) can show ‘that the art in any
material respect taught away’ from the claimed inven-
tion.” (quoting In re Malagari, 499 F.2d 1297, 1303
(CCPA 1974))); In re Corkill, 771 F.2d 1496, 1501 (Fed.
Cir. 1985) (“A greater than expected result is an eviden-
tiary factor pertinent to the legal conclusion of the obvi-
ousness vel non of the claims at issue.” (citing United
States v. Adams, 383 U.S. 39, 51-52 (1966))); In re De
Blauwe, 736 F.2d 699, 706 n.8 (Fed. Cir. 1984) (“A proper
showing of unexpected results will rebut a prima facie
case of obviousness.” (citing In re Fenn, 639 F.2d 762
(CCPA 1981); In re Murch, 464 F.2d 1051 (CCPA 1972)));
In re Klosak, 455 F.2d 1077, 1080 (CCPA 1972) (“The fact
that an invention provides results which would not have
been expected by those skilled in the art is strong evi-
dence in rebuttal of an assertion that the invention would
have been obvious.”).
The court’s apparent departure from this principle,
and its holding that not all properties need be considered
in determining obviousness, is a primary focus of the
concerns stated by the amici curiae. Amicus curiae Bay
Area Bioscience Association writes that:
For many newly discovered pharmaceuticals, their
truly innovative and life-saving properties are of-
ten not discovered until well-controlled clinical
trials or even post-marketing studies have been
conducted—events that occur well after the filing
date of the patent-in-suit. Indeed, this is particu-
larly so in the area of personalized medicine,
where novel therapeutic treatments are tailored
to a particular patient’s genetic makeup.
Brief of Amicus Curiae Bay Area Bioscience Association
in Support of Rehearing En Banc at 7. Again, if the law is
to be changed and precedent discarded, en banc attention
BRISTOL-MYERS SQUIBB COMPANY v. 11
TEVA PHARMACEUTICALS USA, INC.

is required, rather than discordant conflict with prece-
dent.
4. The court’s oversimplified distinction between
“difference in degree” and “difference in kind”
The court held that a “mere difference in degree” is
“insufficient” to render a compound patentable. 752 F.3d
at 977. The flaw in this generalization is apparent from
its application here, where the court held that a new and
effective non-toxic treatment for hepatitis B is merely a
difference in degree from a highly toxic and useless
treatment for hepatitis B.
Precedent has placed the usage “difference in degree”
in a more useful context:
Whether the difference between the claimed in-
vention and the prior art is a difference “in kind”
or a difference “in degree” is not mentioned in sec-
tion 103. Section 103 simply requires a determi-
nation as to whether the invention as a whole
would have been obvious to one of ordinary skill in
the art at the time of appellant’s invention. An
unexpected increase in physiological activity may
be persuasive evidence of unobviousness.
In re Wagner, 371 F.2d 877, 885 (CCPA 1967) (citing In re
Grier, 342 F.2d 120 (CCPA 1965)); see also In re Chupp,
816 F.2d at 644, 646-47 (“selectivity factors (crop safety
combined with weed-killing activity) at least five times
greater than those of closest prior art compounds” were
sufficient evidence of unexpected difference in properties
to rebut prima facie case of obviousness); In re Wiechert,
370 F.2d 927, 932 (CCPA 1967) (“Appellant contends that
obviousness of a novel compound is to be decided not only
from a comparison of its structural formula with that of
the prior art compound, but from all properties of the
compounds. . . . We think appellant’s contentions have
merit . . . . As we indicated in In re Lohr, 317 F.2d 388
12 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

(CCPA 1963), it is possible to obtain a patent where the
showing proves substantially greater effectiveness. . . . In
the case at bar, we are impressed by the 7-fold improve-
ment in activity and, in the absence of valid countervail-
ing evidence, we find the claimed compounds to be
unobvious.”).
The stage at which an obvious difference in degree be-
comes an unobvious difference in kind is based on the
particular subject matter. On the undisputed facts here-
in, the prior art compound 2′-CDG was found to be toxic to
mammals, whereas entecavir is non-toxic to mammals.
This cannot be reasonably viewed as a “mere difference in
degree.”
All of the amici curiae expressed concern about the
negative impact on development of new and improved
products flowing from the court’s fresh uncertainty on the
availability of reliable patent rights. No policy reason has
been offered by the court, for its further restrictions on
access to patenting. From my colleagues’ refusal to
review this ruling en banc, I respectfully dissent.
 United States Court of Appeals
for the Federal Circuit
______________________

BRISTOL-MYERS SQUIBB COMPANY,
Plaintiff-Appellant,

v.

TEVA PHARMACEUTICALS USA, INC.,
Defendant-Appellee.
______________________

2013-1306
______________________

Appeal from the United States District Court for the
District of Delaware in No. 10-CV-0805, Magistrate Judge
Christopher J. Burke.
______________________

TARANTO, Circuit Judge, with whom LOURIE and REYNA,
Circuit Judges, join, dissenting from the denial of the
petition for rehearing en banc.
Bristol-Myers created a new chemical compound:
entecavir. Its structure is a modification of a prior-art
compound, apparently a common phenomenon in phar-
maceutical chemistry, where small changes can make
large differences. The prior-art compound (2′-CDG) was
described in published papers. It had shown excellent
activity against certain viruses, including the hepatitis B
virus, in tests on cell lines in in vitro experiments; but it
had never been tested even in animals, let alone humans,
whether for efficacy or toxicity. Bristol-Myers’s novel
compound proved to be effective and nontoxic (safe) for
2 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

the treatment of hepatitis B, and it was patented and
came to be approved and widely used for that purpose. In
contrast, the prior-art compound never came to have any
human-therapeutic use, because, just after Bristol-Myers
sought its patent, 2′-CDG was tested in animals for the
first time, and the tests so conclusively showed it to be
toxic that it has never been used in humans. In short, the
Bristol-Myers compound, which is a novel molecule, is
dramatically different from the prior-art compound in
providing practical human benefits: one provides such
benefits, the other does not. But that difference was
identified only after Bristol-Myers filed for its patent,
because the prior-art compound, not having been tested in
animals or humans, was not then known to be toxic.
After a trial, the district court invalidated the assert-
ed claim of the Bristol-Myers patent under 35 U.S.C.
§ 103(a) (2006), holding that the new compound would
have been obvious to a person of ordinary skill in the
pertinent art at the time of the invention (here, the filing
date of the patent application). In so holding, the court
determined that such a hypothetical skilled artisan would
have had a reasonable expectation of success in achieving
therapeutic usefulness in humans and would not have
found the favorable safety/efficacy profile of entecavir to
be unexpected, because, at that time, the prior-art com-
pound, never having been tested even in animals, was not
known to be toxic. Bristol-Myers Squibb Co. v. Teva
Pharm. USA, Inc., 923 F. Supp. 2d 602, 630–33 (D. Del.
2013). On appeal, the panel in this case affirmed those
determinations and the ultimate holding of obviousness.
Bristol-Myers Squibb Co. v. Teva Pharm. USA, Inc., 752
F.3d 967 (Fed. Cir. 2014).
In considering whether en banc review is warranted, I
focus on the doctrinal significance of the panel decision
from two perspectives—looking at what the decision says,
and at what it seems to decide on the facts. As to the
first: It is a well-recognized principle, and one essential to
BRISTOL-MYERS SQUIBB COMPANY v. 3
TEVA PHARMACEUTICALS USA, INC.

our system of precedent, that statements in opinions must
be read in context, considering their role in the decision
and the facts of the case. 1 Nevertheless, advocates often
ignore that principle, relying on phrases and sentences
found through database word searches without reading
the whole opinion, and arguing for a precedential effect
that is unwarranted. I here point out why the panel
opinion should not be taken to stand for certain proposi-
tions for which advocates are likely to cite it.
As to the second: Although it is not certain, the panel,
in what it actually decided in affirming invalidity for
obviousness on the recited facts, may have dismissed post-
filing discoveries of prior-art compounds’ true properties

1 See, e.g., R.A.V. v. City of St. Paul, 505 U.S. 377,
386–87 n.5 (1992); Armour & Co. v. Wantock, 323 U.S.
126, 132–33 (1944); Sterling v. Constantin, 287 U.S. 378,
400 (1932); Nat’l Am. Ins. Co. v. United States, 498 F.3d
1301, 1306 (Fed. Cir. 2007); Perez v. Dep’t of Justice, 480
F.3d 1309, 1312 (Fed. Cir. 2007); N. States Power Co. v.
United States, 224 F.3d 1361, 1367 (Fed. Cir. 2000);
Fromson v. W. Litho Plate & Supply Co., 853 F.2d 1568,
1578 (Fed. Cir. 1988) (“Cases should not be cited for mere
words. What counts is what the court did in a cited
case.”), overruled on other grounds by Knorr-Bremse
Systeme Fuer Nutzfahrzeuge GmbH v. Dana Corp., 383
F.3d 1337 (Fed. Cir. 2004); In re Hounsfield, 699 F.2d
1320, 1323 (Fed. Cir. 1983) (“Although some of the forego-
ing judicial statements standing alone could be read to
support the principle the Board here applied, those
statements must be read in the light of the facts of the
cases, the precise issues to be resolved therein, and the
courts’ holdings.”); In re Van Ornum, 686 F.2d 937, 946
(CCPA 1982) (“Precedents are of value for what they
decide, not for every sentence they contain.”); In re Rus-
cetta, 255 F.2d 687, 689 (CCPA 1958).
4 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

as categorically irrelevant to the statutory inquiry. Or it
may have more narrowly deemed insufficient the evidence
here—that, the first time the prior-art compound was
tested in animals, it proved so toxic that it had to be
abandoned as a candidate for human-therapeutic use.
Even if the panel merely rejected the particular post-filing
evidence here as insufficient, it is significant (for how the
decision will be invoked as precedent) that the panel did
not give any case-specific reasons for doing so except
timing: the discovery of the prior-art compound’s toxicity
post-dated the invention. The panel decision seems
highly likely to be viewed as addressing the timing-of-
evidence question—whether generally or in this context.
And that question is worthy of further attention.
Although I am not confident of the answers, I think
that the ruling raises questions about core aspects of the
widely used approach to obviousness analysis—
particularly, the proper meaning of the related elements,
“reasonable expectation of success” and “unexpected
results.” Those questions would benefit from plenary
consideration. In panel review, case-specific applications
on complex facts necessarily consume almost all of the
space of parties’ briefs, and attention is focused almost
exclusively on this court’s own precedents. En banc
review would allow a focus on and full analysis of the
doctrinal issues, considering the language of section 103
(what it resolves and what it leaves open); the role of
section 103 in the statute as a whole (which places a
premium on early filing); Supreme Court precedents
elaborating on the policy of section 103; our own prece-
dents; congressional actions in light of those precedents;
and pertinent, reliable information that may bear on
assessing the real-world consequences of one answer or
another in an industry where research is especially ex-
pensive and uncertain. The widened inquiry seems to me
worthwhile.
BRISTOL-MYERS SQUIBB COMPANY v. 5
TEVA PHARMACEUTICALS USA, INC.

A
1. The panel stated that this court’s en banc decision
in In re Dillon, 919 F.2d 688 (Fed. Cir. 1990), “explain[ed]
that an unexpected result or property does not by itself
support a finding of nonobviousness.” Bristol-Myers, 752
F.3d at 976 (citing Dillon, 919 F.2d at 693, 697). That
statement must not be read out of context to declare that
evidence of unexpected results cannot by itself support an
ultimate finding that a challenger has failed to demon-
strate obviousness by clear and convincing evidence.
First: The panel made the statement only in discuss-
ing whether to uphold the determination about a key
component of the traditional prima facie case in an obvi-
ousness challenge—that the hypothetical skilled artisan
would have had not only a reason to create the new chem-
ical compound (the claimed invention here) but also “a
reasonable expectation of success” concerning its favora-
ble human-therapeutic profile. Bristol-Myers, 752 F.3d at
976–77. The panel was not discussing whether, even if
there were a sustainable finding of a reasonable expecta-
tion of that success, evidence of particular unexpected
results—e.g., unexpectedly great efficacy or safety in the
expected use, or efficacy and safety for an additional,
unexpected use—could nevertheless support an ultimate
finding of non-obviousness. Indeed, reading the state-
ment to draw that conclusion would render immaterial
the extensive discussion of unexpected results that comes
next in the opinion. The panel did not introduce that
discussion by suggesting that it was an “even if” analysis
unnecessary to the bottom-line conclusion.
Second: Dillon itself does not establish that evidence
of unexpected results cannot support rejection of an
obviousness challenge despite supported findings of the
elements of a prima facie case. The issue addressed and
decided in Dillon was only what was needed to establish
the prima facie case in the first place; and Dillon took
6 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

care to stress that it was only that issue it was deciding,
not the ultimate determination of obviousness. 919 F.2d
at 697 (distinguishing In re Papesch, 315 F.2d 381, 391
(CCPA 1963), on ground that Papesch “did not deal with
the requirements for establishing a prima facie case,”
stating: “Papesch is irrelevant to the question of the
requirements for a prima facie case, which is the question
we have here”). Dillon did not need to consider whether
unexpected results could support rejection of a section 103
challenge despite a supported finding of the prima facie
case elements, because Dillon concluded that the PTO
properly found no unexpected results: “[Applicant] did not
present any showing of data to the effect that her compo-
sitions had properties not possessed by the prior art
compositions or that they possessed them to an unexpect-
edly greater degree.” 919 F.2d at 693.
Third: The panel in this case, like the court in Dillon,
had no occasion to rule on the doctrinal relationship
between a finding of unexpected results and a finding of
the prima facie case elements. The panel upheld the
district court’s determination that there were no appre-
ciable unexpected results. Bristol-Myers, 752 F.3d at
977–78; see id. at 978 (“[T]he district court’s findings
reflect that one of skill in the art would have expected
entecavir’s hepatitis B’s efficacy, safety, and therapeutic
window based on one’s knowledge of 2′-CDG.”); Bristol-
Myers, 923 F. Supp. 2d at 686 (“No witness testified that
the [low toxicity] of the drug would have been ‘unex-
pected.’”). On that premise, there were no unexpected
results whose relationship to the prima facie case the
panel had to consider. 2

2 The panel likewise had no occasion to address
broader issues concerning the familiar use of a “prima
facie case” as a sequence-of-presentation, issue-organizing
tool in a challenge to an issued patent, for which invalidi-
BRISTOL-MYERS SQUIBB COMPANY v. 7
TEVA PHARMACEUTICALS USA, INC.

2. The panel decision also does not establish a prece-
dent for the proposition that, putting aside the post-filing
evidence, the proof of “reasonable expectation of suc-
cess”—based entirely on in vitro experiments with the
lead compound—was adequate. Bristol-Myers never
argued otherwise to the court; it argued only that, once
the post-filing evidence of 2′-CDG is considered, the proof
of reasonable expectation of success was inadequate and
the proof of unexpected results in any event compelling.
Inadequacy apart from the post-filing evidence not having
been argued, the panel opinion is not precedent for deem-
ing the pre-filing evidence inadequate. 3
This is worth noting because it seems to me a serious
question whether, in this case and perhaps more general-
ly, the purely in vitro experiments on the lead compound
should be deemed to establish a “reasonable” expectation
of success. The success that must be reasonably expected
in this case would, I think, have to be success in what
motivated the investment in the research—an acceptable
safety/efficacy profile for human-therapeutic use. 4 Thus,

ty requires clear and convincing proof, Microsoft Corp. v.
i4i Ltd. P’ship, 131 S. Ct. 2238, 2252 (2011). Cf. U.S.
Postal Serv. Bd. of Governors v. Aikens, 460 U.S. 711,
714–16 (1983) (discussing burden of persuasion and
presentation-of-proof scheme in discrimination cases);
Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133,
147–49 (2000) (same).
3 See, e.g., District of Columbia v. Heller, 554 U.S.
570, 625 n.25 (2008); United States v. Verdugo-Urquidez,
494 U.S. 259, 272 (1990); United States v. L.A. Tucker
Truck Lines, Inc., 344 U.S. 33, 38 (1952); JVC Co. of Am.
v. United States, 234 F.3d 1348, 1353–54 (Fed. Cir. 2000)
(citing earlier cases).
4 See Leo Pharm. Prods., Ltd. v. Rea, 726 F.3d 1346,
1357 (Fed. Cir. 2013); In re Cyclobenzaprine Hydrochlo-
8 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

whatever the precise meaning of “reasonable expecta-
tion”—a matter worth clarifying, as discussed infra—Teva
had to show that a hypothetical skilled artisan would
have had a reasonable expectation of acceptable safety of
entecavir in humans in October 1990, when Bristol-Myers
filed for its patent. And such an expectation, it appears
undisputed, depended entirely on showing such an arti-
san’s reasonable expectation, at the time, that the lead
compound, 2′-CDG, would be acceptably safe in humans.
There is a serious question whether any such expecta-
tion was reasonable, given that 2′-CDG had been tested
only in in vitro experiments—never even in animals, let
alone humans. As a general matter, it may be that in
vitro tests are not reliably predictive of human safety. 5

ride Extended-Release Capsule Patent Litig., 676 F.3d
1063, 1070 (Fed. Cir. 2012); Pfizer, Inc. v. Apotex, Inc., 480
F.3d 1348, 1364 (Fed. Cir. 2007); Yamanouchi Pharm. Co.
v. Danbury Pharmacal, Inc., 231 F.3d 1339, 1345 (Fed.
Cir. 2000).
5 See Fed. Judicial Ctr., Reference Manual on Scien-
tific Evidence 645 (3d ed. 2011) (“Relatively few [in vitro
toxicity tests] have been validated by replication in many
different laboratories or by comparison with outcomes in
animal studies to determine if they are predictive of whole
animal or human toxicity.”); In re Gangadharam, 1989
WL 127023, at *2 (Fed. Cir. Oct. 27, 1989) (noting that a
prior art’s “remarking that the positive in vitro results
‘favored’ use in vivo does not meet the statutory standard”
of obviousness); see also Anna Astashkina et al., A Critical
Evaluation of In Vitro Cell Culture Models for High-
Throughput Drug Screening and Toxicity, 134 Pharmacol-
ogy & Therapeutics 82, 82, 94 (2012) (noting “strong
evidence that in vitro cell-based assays and [even] subse-
quent preclinical in vivo studies do not yet provide suffi-
cient pharmacological and toxicity data or reliable
BRISTOL-MYERS SQUIBB COMPANY v. 9
TEVA PHARMACEUTICALS USA, INC.

Although statistics require careful examination to be used
responsibly, I note that amici have pointed us to litera-
ture indicating that only small percentages of compounds
that start in the laboratory make it out the other end of
the drug-development process. 6
It may well be, of course, that in vitro testing sup-
ports a sound expectation about probable human safety
for certain compounds even if it does not do so generally.
But that is a matter to be addressed by scientific evidence
about the particular compounds at issue in a given case.
Here, with Bristol-Myers not having contested the point,
the panel had no occasion to scrutinize the record to
determine if there was evidence of a reliable basis for any
prediction of human safety for 2′-CDG in October 1990.
Optimism about the compound is not the same as a rea-
sonably grounded prediction. And testimony that 2′-CDG
had a “very good therapeutic window” in an in vitro test—
which is all that the quote refers to, because 2′-CDG had
never been given as “therapy” or even put into animals or
humans—does not support safety in humans without
sound evidence allowing the inference, none of which is
apparent. Bristol-Myers, 752 F.3d at 971, 974, 978. The
issue not having been contested, the panel decision cannot
be taken to have resolved the issue.

predictive capacity for understanding drug candidate
performance in vivo”).
6 See, e.g., Henry Grabowski, Patents, Innovation,
and Access to New Pharmaceuticals, 5 J. Int’l Econ. L.
849, 849–51 (2002) (“[F]ewer than 1% of the compounds
examined in the pre-clinical period make it into human
testing.”); Michael Hay et al., Clinical Development Suc-
cess Rates for Investigational Drugs, Nature Biotechnolo-
gy Jan. 2014, at 41–42, 47 (10-15% of drugs entering
human testing emerge as marketed drugs).
10 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

B
As already noted, the panel opinion may be read by
future litigants to suggest that any evaluation of prior art
must focus exclusively on what was known about the prior
art’s properties, and on that basis expected about
entecavir, at the time of the Bristol-Myers invention. See
Bristol-Myers, 752 F.3d at 974, 977, 978. The panel
opinion ultimately approves the district court’s decision to
excise from its analysis any consideration of 2′-CDG’s
later-discovered, severe toxicity. Id. at 978 (“The district
court ultimately made the correct direct comparison of the
patented compound to 2′–CDG, noting that prior art
compounds, ‘including 2′–CDG,’ ‘showed effectiveness
against hepatitis B without known toxicity issues.’”)
(emphasis added) (quoting Bristol-Myers, 923 F. Supp. 2d
at 685). The timing-of-evidence reasoning seems at the
heart of the obviousness invalidation. It raises questions
that I think warrant further exploration.
1. Judge Newman identifies ways in which the pan-
el’s approach to the timing-of-evidence question seems in
tension with this court’s precedents. It appears that, at
least since our predecessor court’s decision in In re
Papesch, 315 F.2d 381 (CCPA 1963), the analysis of
obviousness of new chemical inventions has involved
“liberal consideration of post-invention evidence.” Rebec-
ca S. Eisenberg, Pharma’s Nonobvious Problem, 12 Lewis
& Clark L. Rev. 375, 395 (2008). The reason seems clear:
“often it takes time to determine the properties of a new
chemical through testing and observation that cannot
take place until after the chemical is in hand,” id. at 396,
and the statute has always provided an incentive to file
early once the chemical is in hand (lest priority be lost),
an incentive now enhanced by the 2011 adoption of a first-
inventor-to-file system. Moreover, Judge Newman notes
that the post-filing experiments comparing properties of
the invention and prior-art compounds would seem often
to have developed new information about both the inven-
BRISTOL-MYERS SQUIBB COMPANY v. 11
TEVA PHARMACEUTICALS USA, INC.

tion and the prior-art compound. These precedents and
past practices raise questions about the panel’s ruling. 7
2. The statutory language does not itself provide an
answer to the question of post-filing evidence. 8 It is true

7 In still other ways, obviousness analysis routinely
considers relevant facts not in existence at the time of
patent filing, e.g., commercial success and proven meeting
of a long-felt need. See KSR International Co. v. Teleflex
Inc., 550 U.S. 398, 406 (2007); Galderma Labs., L.P. v.
Tolmar, Inc., 737 F.3d 731, 740 (Fed. Cir. 2013); Perfect
Web Techs., Inc. v. InfoUSA, Inc., 587 F.3d 1324, 1332–33
(Fed. Cir. 2009).
8 Section 103, reflecting the first-inventor-to-file
system adopted in 2011, now reads: “A patent for a
claimed invention may not be obtained, notwithstanding
that the claimed invention is not identically disclosed as
set forth in section 102, if the differences between the
claimed invention and the prior art are such that the
claimed invention as a whole would have been obvious
before the effective filing date of the claimed invention to
a person having ordinary skill in the art to which the
claimed invention pertains. Patentability shall not be
negated by the manner in which the invention was made.”
35 U.S.C. § 103.
35 U.S.C. § 103(a) (2006) was similar but reflected the
pre-2011 first-to-invent system: “A patent may not be
obtained though the invention is not identically disclosed
or described as set forth in section 102 of this title, if the
differences between the subject matter sought to be
patented and the prior art are such that the subject
matter as a whole would have been obvious at the time
the invention was made to a person having ordinary skill
in the art to which said subject matter pertains. Patenta-
bility shall not be negatived by the manner in which the
invention was made.”
12 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

that the language directs courts (and the PTO) to ask a
question about the time “before the effective filing date of
the claimed invention” (the “time the invention was
made” in the pre-2011 version). But the question is not
what certain people in the field in fact thought at that
time. Rather—to note three aspects of the language
(others may be relevant too)—the question is whether (a)
the claimed invention “as a whole” (b) “would have been”
obvious to a (c) “person having ordinary skill in the art.”
As to the first of those elements, the Papesch doctrine
has long treated all properties, including later discovered
ones, as part of the invention “as a whole.” As to the
third, the “person having ordinary skill in the art” is not a
real-world person, but a hypothetical person, constructed
in applying the provision to create a standard of patenta-
bility that effectuates the provision’s policy. 9 As to the
second, the statute uses the verb phrase “would have
been.” Grammatically, that formulation invokes a hypo-
thetical situation dependent on some “if” condition (would
have been obvious if “x” had been true). In section 103,
however, the required condition is not stated; there is no
“if” clause.
As a result, the statutory language itself requires
courts to fill in the conditions for the hypothetical inquiry
by an analysis of the provision’s history, role in the stat-
ute, and purpose, always considering workability of any
approach. It is common to hypothesize knowledge of all
pertinent prior art. See In re Moreton, 288 F.2d 940, 941
(CCPA 1961); In re Citron, 251 F.2d 619, 620 (CCPA
1958). It has been suggested, too, that the hypothetical
inquiries should not take as a given the current amassing

9 E.g., Bristol-Myers, 752 F.3d at 978; Norgren, Inc.
v. Int’l Trade Comm’n, 699 F.3d 1317, 1327 (Fed. Cir.
2012); Kimberly-Clark Corp. v. Johnson & Johnson, 745
F.2d 1437, 1454 (Fed. Cir. 1984).
BRISTOL-MYERS SQUIBB COMPANY v. 13
TEVA PHARMACEUTICALS USA, INC.

and organizing of resources and talent into firms that
undertake risky, expensive research, which might not
exist without patent protection. See Michael Abramowicz
& John F. Duffy, The Inducement Standard of Patentabil-
ity, 120 Yale L.J. 1590, 1614–16 (2011). Whatever the
proper approach, however, it is one that must be devel-
oped by looking at more than the effective-date-of-filing
(previously, invention-date) phrase in section 103, whose
terms as a whole call for a hypothetical inquiry requiring
judicial definition.
3. The proper analysis of the post-filing evidence re-
garding 2′-CDG would seem to focus on two closely related
phrases that identify standard parts of our obviousness
analysis: “reasonable expectation of success” and “unex-
pected results.” Both phrases evidently bear several
different potential meanings and so would benefit from
clarification. And the identification of the proper mean-
ings seems to have a strong bearing on whether the post-
filing evidence here is material.
The phrase “reasonable expectation of success” on its
face requires that any expectation of success be “reasona-
ble.” The same reasonableness requirement would seem
implicit, too, in the hypothetical character of the skilled
artisan whose expectations count. The hypothetical
character of the person doing any expecting seemingly
also should mean that “unexpected results” contains a
reasonableness requirement. What must be “reasonable”
are (hypothetical) “expectations.” But “expectation” is a
term that covers different ground in different circum-
stances.
Clarifying these concepts seems important here.
Should a reasonable expectation mean a mere educated
guess or surmise or plausible possibility? Should it mean
an affirmative well-grounded prediction, using a 50% or
other probability, based on the standards that scientists
would use professionally to assert such predictions—
14 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

whether in a scientific journal or in making a decision
about how to allocate scarce research funding? Depend-
ing on the meaning of “expectation,” should the reasona-
bleness of the expectation consider not just what evidence
has been developed but also what evidence could easily be
developed but has not yet been—so that, for example, it
may be irresponsible to assert an expectation in the
absence of such available but not-yet-secured evidence?
We have tied the “reasonable expectation of success”
standard to the Supreme Court’s use of “predictable” in
KSR, 550 U.S. at 416, 417, 421. See PharmaStem Thera-
peutics, Inc. v. ViaCell, Inc., 491 F.3d 1342, 1360 (Fed.
Cir. 2007). That precedent suggests a higher rather than
lower standard for “reasonable expectation.”
Perhaps the statutory policy of section 103 does as
well. The Supreme Court has suggested the policy (not
for case-by-case application but to inform doctrinal stand-
ards): to deny patent protection for a new invention only
when the invention would have been forthcoming (at
about the same time) even without patent protection, i.e.,
when patent protection was not needed to induce its
emergence. 10 Moreover, like the word “obvious” in one
understanding, what protection may be needed to induce
the invention plausibly depends on the costs and uncer-

10 See Graham v. John Deere Co., 383 U.S. 1, 11
(1966) (“The inherent problem was to develop some means
of weeding out those inventions which would not be
disclosed or devised but for the inducement of a patent.”);
KSR, 550 U.S. at 419 (“Granting patent protection to
advances that would occur in the ordinary course without
real innovation retards progress. . . .”); see generally
Abramowicz & Duffy, supra. Statutorily, this approach
amounts to adding something like “if patent protection
had been unavailable” as the missing “if” clause for the
“would have been” phrase.
BRISTOL-MYERS SQUIBB COMPANY v. 15
TEVA PHARMACEUTICALS USA, INC.

tainties of the work required for success. See, e.g.,
Abramowicz & Duffy, supra, at 1613–14, 1655; William
M. Landes & Richard A. Posner, The Economic Structure
of Intellectual Property Law 304 (2003).
4. The definitional questions seem to bear materially
on one issue central to Bristol-Myers’ argument—whether
the post-filing evidence of 2′-CDG’s immediate and con-
clusive failure in animal testing is significant to assessing
whether, before such testing, there truly was a reasonable
expectation of relevant (human-therapeutic) success of 2′-
CDG (and hence of entecavir). As a general evidentiary
matter, it seems relevant to determining the reasonable-
ness of any expectation before conducting a readily avail-
able animal test that the very first animal test
immediately showed such toxicity that 2′-CDG has never
since been tried in humans. 11 Even in the arena of busi-

11 Cf., e.g., 22 Charles Alan Wright & Kenneth W.
Graham, Jr., Federal Practice & Procedure § 5171, at
752–57 (2012) (experiments conducted after the event in
question may be admitted to show what could have been
done, what might have happened, to reveal the character-
istics of a product or the dangers arising from it); Burke v.
Deere & Co., 6 F.3d 497, 505–06 (8th Cir. 1993) (accidents
taking place after a manufacturer sold a specific item to
the plaintiff admissible to determine the actual risk of
harm posed by the defective product); Rocky Mountain
Helicopters, Inc. v. Bell Helicopters Textron, 805 F.2d 907,
918 (10th Cir. 1986) (results of a post-accident stress test
admissible in products liability trial, provided study
redacted any evidence of a subsequent redesign); Bailey v.
Kawasaki-Kisen, K.K., 455 F.2d 392, 397–98 (5th Cir.
1972) (where individual was injured by a falling boom,
evidence of the boom falling a second time was admissible
to prove that the boom was in fact defective), abrogated in
distinct respect (subsequent remedial measures) by Fed.
16 BRISTOL-MYERS SQUIBB COMPANY v.
TEVA PHARMACEUTICALS USA, INC.

ness forecasts—where changes in the world over time can
dramatically affect results—courts temper a great caution
about hindsight bias with a recognition that “a gross
disparity between prediction and fact” may be relevant to
assessing the reasonableness of the prediction. 12 All the
more so in the present context, which involves a general
biological property (toxicity of a particular compound)
that should be the same today as it will be next year. But
whether this is a sensible analysis may well depend on
precisely what “reasonable expectation” means in the
present context.
I would grant rehearing en banc to enable a full ex-
ploration of these questions.

R. Evid. 407, as stated in Rutledge v. Harley-Davidson
Motor Co., 364 F. App’x 103, 106 (5th Cir. 2010).
12 Spitzberg v. Houston Am. Energy Corp., 758 F.3d
676, 691 (5th Cir. 2014) (a “gross disparity between
prediction and fact” may form the basis for 10b-5 liability)
(quoting Lormand v. US Unwired, Inc., 565 F.3d 228, 248
n.13 (5th Cir. 2009)); Marx v. Computer Scis. Corp., 507
F.2d 485, 489 (9th Cir. 1974); G & M, Inc. v. Newbern, 488
F.2d 742, 745–46 (9th Cir. 1973).