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

752 F.3d 967, 111 U.S.P.Q. 2d (BNA) 1293, 2014 WL 2609924, 2014 U.S. App. LEXIS 10911

• Court: Court of Appeals for the Federal Circuit
• Decided: June 12, 2014
• Docket: 2013-1306
• Status: Published

Opinion

CHEN, Circuit Judge.

This patent infringement case concerns a drug for the treatment of hepatitis B. After a four-day bench trial, the United States District Court for the District of Delaware found claim 8 of U.S. Patent No. 5,206,244 (’244 patent) invalid as obvious. We affirm the district court’s invalidity judgment for the reasons that follow.

I.

Appellant Bristol-Myers Squibb Co. (BMS) owns the '244 patent. Claim 8 of the '244 patent is directed to a nucleoside analog composed of two regions: a carbo-cyclic ring and a guanine base. Nucleo-side analogs are manmade compounds designed to mimic the activity of natural nucleosides, the building blocks of DNA and RNA. These compounds are modified slightly from their natural counterparts to interfere with the replication of viral DNA — which means that they can serve as possible antiviral compounds. Claim 8 covers one such compound, entecavir. BMS markets entecavir as a treatment for hepatitis B under the trade name Bara-clude®.

Entecavir is a modified version of the natural nucleoside 2'-deoxyguanosine (de-oxyguanosine). Entecavir is structurally identical to deoxyguanosine except for one difference: it has a carbon-carbon double bond (also known as an exocyclic methylene group) at the 5' position of the carbo-cyclic ring where deoxyguanosine has an oxygen atom.

The chemical structures of entecavir and deoxyguanosine are illustrated below:

[[Image here]]

Appellant’s Br. 8; see also J.A. 11.

The structures referenced throughout this opinion include a “carbocyclic ring” of carbon atoms, which is illustrated above as the pentagonal structure at the left of each diagram, and a nucleoside base, which is illustrated above as the double ring structure to the right. In both figures above, the nucleoside base is guanine.

Entecavir is an effective treatment for hepatitis B. The drug is generally accepted as a safe drug, with a broad therapeutic window for treatment, providing for a wide gap between low doses of the drug that are effective against disease and the high doses that could cause .unwanted toxicity. It also has a high genetic barrier to resistance such that, if they have not previously received a nucleoside-based treatment, few patients treated with entecavir develop drug resistance to it.

The appellee, Teva Pharmaceuticals USA, Inc. (Teva), filed an abbreviated new drug application (ANDA) for a generic version of entecavir. In support of its ANDA, Teva filed “Paragraph IV” certifications, alleging that its generic products would not infringe the '244 patent, and/or that the patent was invalid or unenforceable. See 21 U.S.C. § 355<j)(2)(A)(vii)(IV).

BMS sued Teva for patent infringement, claiming that Teva’s ANDA filing infringed the '244 patent. See 35 U.S.C. § 271(e)(2). At trial, the parties narrowed the issues to obviousness and inequitable conduct. 1 Teva’s obviousness argument focused on the selection of 2'-CDG as a lead compound from the prior art.

A.

2'-CDG is a potent antiviral carbocyclic nucleoside analog that is structurally similar to the natural nucleoside deoxyguano-sine, differing only in that it replaces an oxygen atom with a carbon atom at the 5' position. The following illustrations compare the chemical structures of 2'-CDG and deoxyguanosine.

[[Image here]]

Appellant’s Br. 14; see also J.A. 11, 23.

The earliest priority date for the '244 patent is the date that BMS filed the application, October 18, 1990. 2'-CDG’s synthesis was first published in the Journal of Medicinal Chemistry in 1984 by Dr. Y. Fulmer Shealy (the Shealy reference) of the Southern Research Institute (the SRI). The Shealy reference taught that 2'-CDG exhibited better in vitro antiviral activity against the herpes virus than the FDA-approved best-selling drug at the time, Ara-A. Dr. Shealy obtained a patent for 2'-CDG and other related compounds, stating that they were useful in the treatment of viral infections. Subsequent research on 2'-CDG by Dr. Shealy showed in vivo activity against herpes viruses.

After the Shealy reference was published, other researchers soon began working with 2'-CDG as an antiviral, including scientists at SRI, Mount Sinai School of Medicine, GlaxoSmithKline (Glaxo), and other institutions. In 1989, Dr. J.A. Montgomery of SRI published an article summarizing the state of antiviral research at the time and reported that 2'-CDG was “[b]y far the most promising” antiviral against herpes. J.A. 2148. The Montgomery reference also taught that 2'-CDG was five to six times more potent than one of the leading drugs on the market, acyclovir. Teva’s expert, Dr. Heathcock, stated that the Montgomery reference was a “lamp post that really illuminate[d] 2'-CDG as ... a very exciting lead compound to work from,” and other chemists, during the relevant time period, were using 2'-CDG as a lead compound. J.A. 27. BMS’s expert, Dr. Schneller, conceded that he did not “completely disagree” with Dr. Heath-cock’s opinion. J.A. 27-28. In fact, Dr. Schneller himself published research investigating antiviral activity of carbocycl-ics, including 2'-CDG, and cited to Dr. Sheal/s article, noting the significant antiviral activity of 2'-CDG.

Also in 1989, Peter M. Price and other researchers with the Mount Sinai School of Medicine published the results of testing 2'-CDG against hepatitis (the Price reference). The Price reference disclosed that 2'~CDG showed “excellent activity” against the hepatitis B virus “with as little as 25 ng of 2'-CDG per ml” resulting in the “almost complete disappearance of replicating” hepatitis B virus. J.A.2086. The Price reference also taught that 2'-CDG was “nontoxic in concentrations up to 200 times the minimum effective inhibitory concentration.” Id. According to Teva’s expert, Dr. Heathcock, the Price reference demonstrated that 2'-CDG “had a very good therapeutic window [because] [i]t was effective at a level, much lower than its toxic level.” J.A. 1048.

While Shealy, Montgomery, Price, and others studied 2'-CDG, a group of medicinal chemists at Syntex published antiviral studies on another nucleoside analog composed of a carbocyclic ring and adenosine base. J.A.2001. The Madhavan reference disclosed that the substitution of an exo-cyclic methylene (carbon-to-carbon double bond) for the oxygen atom at the 5' posi tion on the carbocyclic ring on nucleoside aristeroinycin led to a nucleoside analog (Madhavan 30) with significantly superior antiviral properties. This exocyclic methylene substitution is the same modification at the same location made to 2'-CDG to form entecavir. The following illustrations compare the chemical structures of 2'-CDG, entecavir, and Madhavan 30.

[[Image here]]

Appellant’s Br. 8,14,15.

According to BMS’s inventor, Dr. Zah-ler, he was aware of the Madhavan reference before conceiving entecavir and testified that BMS submitted it to the USPTO as the “most relevant” piece of prior art in the '244 patent application. J.A. 1078, 1215, 1223. But BMS did not submit 2'-CDG as prior art to the USPTO.

B.

After a bench trial, the district court found that at the time of BMS’s invention, 2'-CDG was a lead compound for the development of antiviral drugs. Bristol-Myers Squibb Co. v. Teva Pharms. USA, Inc., 923 F.Supp.2d 602, 665 (D.Del.2013). Based on (1) the structural similarity between entecavir and 2-CDG, (2) the teachings of the Madhavan reference, (3) the finding that the exocyclic methylene substitution would be a “small, conservative change[ ]” and (4) the “totality of the prior art” on 2'-CDG, the district court found that a skilled artisan would have been motivated to substitute an exocyclic methylene group at the 5' position of 2'-CDG, with a reasonable expectation of success of creating a compound with beneficial antiviral properties. Id. at 669, 675.

The district court also analyzed secondary considerations of nonobviousness. Although the court found that some of these considerations — commercial success, long-felt need, and evidence of unexpected properties — cut in favor of non-obviousness, the court ultimately concluded that Teva proved by clear and convincing evidence that claim 8 would have been obvious. Id. at 696. As a result, the district court entered judgment in favor of Teva. Id. BMS appeals the district court’s invalidity finding. We have jurisdiction under 28 U.S.C. § 1295(a)(1).

II.

The only issue for our review is the district court’s obviousness ruling. Obviousness is a question of law with underlying factual findings. Honeywell Int'l, Inc. v. United States, 609 F.3d 1292, 1297 (Fed.Cir.2010). We review the conclusion of obviousness de novo, and the trial court’s factual findings for clear error. Id.

Obviousness requires assessing (1) the “level of ordinary skill in the pertinent art,” (2) the “scope and content of the prior art.” (3) the “differences between the prior art and the claims at issue,” and (4) “secondary considerations” of non-obviousness such as “commercial success, long-felt but unsolved needs, failure of others, etc.” KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406, 127 S.Ct. 1727, 167 L.Ed.2d 705 (2007) (quoting Graham v. John Deere Co., 383 U.S. 1, 17-18, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966)).

A party seeking to invalidate a patent as obvious must demonstrate “ ‘by clear and convincing evidence that a skilled artisan would have been motivated to combine the teachings of the prior art references to achieve the claimed invention, and that the skilled artisan would have had a reasonable expectation of success from doing so.’ ” Procter & Gamble Co. v. Teva Pharms. USA, Inc., 566 F.3d 989, 994 (Fed.Cir.2009) (quoting Pfizer, Inc. v. Apotex, Inc., 480 F.3d 1348, 1361 (Fed.Cir.2007)).

To establish obviousness in eases involving new chemical compounds, the accused infringer must identify some reason that would have led a chemist to modify a known compound. Takeda Chem. Indus., Ltd. v. Alphapharm Pty., Ltd., 492 F.3d 1350, 1357 (Fed.Cir.2007). Generally, an obviousness inquiry concerning such “known compounds” focuses on the identity of a “lead compound.” Eisai Co. Ltd. v. Dr. Reddy’s Labs., Ltd., 533 F.3d 1353, 1359 (Fed.Cir.2008).

A lead compound is a compound in the prior art that would be “a natural choice for further development efforts.” Altana Pharma AG v. Teva Pharms. USA, Inc., 566 F.3d 999, 1008 (Fed.Cir.2009). The motivation to modify that lead compound can come from any number of sources and need not necessarily be explicit in the art. “[I]t is sufficient to show that the claimed and prior art compounds possess a ‘sufficiently close relationship ... to create an expectation,’ in light of the totality of the prior art, that the new compound will have ‘similar properties’ to the old.” Otsuka Pharm. Co., Ltd. v. San-doz, Inc., 678 F.3d 1280, 1293 (Fed.Cir. 2012) (quoting In re Dillon, 919 F.2d 688, 692 (Fed.Cir.1990) (en banc)). Whether a lead compound and a claimed compound have a sufficiently close relationship frequently turns on their “structural similarities and differences.” Daiichi Sankyo Co. v. Matrix Labs., Ltd., 619 F.3d 1346, 1352 (Fed.Cir.2010).

Based on the prior art and testimony, the district court properly found strong evidence of obviousness, because the record shows that a skilled artisan would have selected 2'-CDG as a lead compound and made the minor modification to arrive at entecavir. Moreover, we see no clear error in the district court’s fact findings regarding evidence of secondary considerations of non-obviousness.

BMS attacks the lower court’s obviousness determination by contending that a skilled artisan would have had to make too many decisions to arrive at entecavir. Those decisions include selecting (1) the class of nucleoside analog compounds, (2) 2'-CDG as a lead compound from the class of earbocyclics, (3) the carbocyclic ring or guanine base of 2'-CDG for modification, (4) the 2' or 5' position on the carbocyclic ring, (5) the specific chemical element on the 5' position (carbon), and (6) the type of carbon to carbon bond (single or double). We conclude that the district court’s analysis is well supported.

During the relevant time period in the late 1980s, carbocyclic analogs were generating a great deal of interest among researchers searching for compounds with antiviral activity. Both parties’ experts agree on that point. Several research institutions — including Glaxo, Syntex, Abbott Laboratories, and SRI — investigated and published on antiviral activity of carbocyclic nucleosides. Thus, the district court had sufficient evidence to conclude that one of ordinary skill in the art during the relevant time period would have studied carbocyclic analogs “as a promising area” for antiviral drug discovery. J.A. 97.

Of the carbocyclic analogs on which researchers focused in the late 1980s, 2'-CDG was a “natural choice for further development.” Altana, 566 F.3d at 1008. The district court observed, based on BMS’s expert’s own testimony, that “medicinal chemists during the relevant time frame were actually treating and using 2'-CDG as a lead compound” in the search for new antivirals at the time. J.A. 97 (emphasis in original). The Shealy, Montgomery, and Price references collectively reinforce that understanding of 2'-CDG.

BMS challenges the selection of 2'-CDG as a lead compound because it was discovered to be toxic in the 1990s. However, at the time of entecavir’s invention, the Price reference showed that 2'-CDG was generally understood to be safe and nontoxic, and other researchers were already using it as a lead compound. As the district court points out, in “October 1990, 2'-CDG was not yet known to have high toxicity,” and BMS’s expert, Dr. Schneller, agreed that researchers at the time treated 2'-CDG as a “promising compound.” J.A. 104, 111 (emphasis in original). Therefore, we see no error in the selection of 2'-CDG as the lead compound here. See Velander v. Garner, 348 F.3d 1359, 1377 (Fed.Cir.2003) (“Obviousness, and expectation of success, are evaluated from the perspective of a person having ordinary skill in the art at the time of invention.” (emphasis added) (citation omitted)); see also Eisai, 533 F.3d at 1359.

Accordingly, we therefore agree with the district court that those of ordinary skill in the art would have selected 2'-CDG, a carbocyclic analog, as a lead compound for further development efforts before BMS applied for the ’244 patent in October 1990. See Daiichi, 619 F.3d at 1354 (explaining that more than mere structural similarity must be identified as a reason to select a compound as a lead compound; “knowledge in the art of the functional properties and limitations of the prior art compounds” are also important to the analysis). 2

After selecting a lead compound, both experts (Dr. Schneller and Dr. Heathcock) agreed that a chemist in drug development would seek to make small, conservative changes to that structure. In drug development, it is common to modify a lead compound in an effort to “obtain a compound with better activity.” Otsuka, 678 F.3d at 1291 (quotation omitted).

With 2'-CDG as a lead compound, the record here amply supports the conclusion that one of ordinary skill in the art would have had a motivation to modify 2'-CDG’s carbocyclic ring by substituting an exocyclic methylene group at the 5' position to make the patented compound, entecavir. See Otsuka, 678 F.3d at 1292; see also Takeda, 492 F.3d at 1357.

In choosing whether to modify 2'-CDG’s carbocyclic ring or its guanine base, BMS’s expert, Dr. Schneller, initially testified that he would “retain the [carbocyclic] portion,” J.A. 114, but he acknowledged on cross-examination that other chemists were making changes to the carbocyclic portion. JA 1307-08. Teva’s expert, Dr. Heath-cock, also testified that changing the car-bocyclic portion resulted in greater activity than changes to the guanine ring. Accordingly, this was a natural decision because the goal was to develop antivirals with improved activity.

Unrefuted expert testimony also explained how the next obvious choice for modification would have been either the 2' or 5' position on the carbocyclic ring, because only at these locations could small changes easily be made to the molecule. Both experts agreed that a skilled artisan would focus on the smallest elements on the top row of the periodic table, including carbon and fluorine. For a specific element, BMS’s expert, Dr. Schneller, testified in his deposition that he would “rule out everything but the carbon” and that carbon was “the only one that sticks out.” J.A. 1304. Teva’s expert, Dr. Heathcock, also explained that the choice to have an exocyclic methylene (carbon-to-carbon double bond) over a methyl group (carbon-to-carbon single bond) would be a more conservative choice, because a methyl group is bigger and longer than an exo-cyclie methylene group, and the easiest way to synthesize a methyl group would be to make methylene first. J.A. 117-118, 1053, 1306.

Futhermore, the Madhavan reference demonstrated that adding an exocyclic methylene group to a carbocyclic nucleoside analog can result in a lead compound with improved antiviral activity. Specifically, it teaches that aristeromycin and Madhavan 30 were two compounds that differed only in the presence of an exocyclic methylene substitution at the 5' position; that modification resulted in the formation of a much more potent antiviral.

Teva’s expert, Dr. Heathcock, testified that the substitution was an “obvious modification” in light of the prior art “because there were other [antiviral] compounds like that that had already been made” and a chemist would expect the nucleoside analog “to have similar biological properties of [2'-]CDG itself, which were good properties.” J.A. 44. During prosecution of the '244 patent, BMS presented the Madhavan reference as containing the closest piece of prior art to entecavir. BMS’s expert, Dr. Schneller, does not directly state that ente-cavir was obvious. But, on cross examination, Dr. Schneller stated that the Madha-van reference “would not dissuade” a chemist from adding the exocyclic methylene to form an antiviral drug. And even though the Madhavan compound was more toxic than the underlying lead compound, aristeromycin, (which itself was cytotoxic), Dr. Schneller admitted that the combination of features reported in the Madhavan and Shealy references “could have led” a person of skill in the art to seek new antivirals. J.A. 1311-12.

Based on the record, we see no clear error in the district court’s finding that the modification required to transform 2'-CDG into the structurally similar entecavir is a minor one: the addition of a single carbon atom to form an exocyclic methylene with the already-present carbon atom at the 5' position of the carbocyclic ring of 2'-CDG to create entecavir. See supra at 971-72. Upon selecting 2'-CDG as the lead compound, the steps of deciding which bond to modify and how to modify that bond “equate to a small, finite number of changes to try to [arrive at] the lead compound.” J.A. 117. See In re Cyclobenzap-rine Hydrochloride Patent Litig., 676 F.3d 1063, 1072 (Fed.Cir.2012) (“Evidence of obviousness ... is insufficient unless it indicates that the possible options skilled artisans would have encountered were ‘finite,’ ‘small,’ or ‘easily traversed,’ and that skilled artisans would have had a reason to select the route that produced the claimed invention.”) (citations omitted).

Further, the skilled artisan’s reasonable expectation of success is measured “as of the date of the invention.” Amgen Inc. v. Hoffman-La Roche, 580 F.3d 1340, 1362 (Fed.Cir.2009). The Madhavan reference’s teachings of improved potency with an exocyclic methylene at the 5' position supports the expectation that the same substitution to 2'-CDG would reasonably lead to similar properties. As explained above, 2'-CDG and entecavir are very structurally similar, and it is well settled that structurally similar compounds often have similar properties. See Takeda, 492 F.3d at 1356; Altana, 566 F.3d at 1007; In re Soni, 54 F.3d 746, 750 (Fed.Cir.1995) (“[T]he presumption [is] that similar compositions have similar properties.”).

BMS fails to establish any clear error in the district court’s factual determinations, which are based on the prior art and expert testimony. In light of those factual findings, we agree with the district court that Teva provided strong evidence of obviousness, given the use of 2'-CDG as a lead compound during the relevant time period, the “totality of the prior art,” and the structural similarity between entecavir and 2'-CDG suggesting similar properties. J.A. 128,131.

BMS also argues that a new chemical entity, as a matter of law, cannot be obvious when the claimed invention possesses unexpected properties. Specifically, BMS argues that the existence of unexpected properties forecloses a finding of a reasonable expectation of success. We have already rejected this argument en banc in Dillon, explaining that an unexpected result or property does not by itself support a finding of nonobviousness. Dillon, 919 F.2d at 693, 697. In Dillon, we held that the expected properties of a claimed compound may be sufficient to lead to a reasonable expectation of success in modifying a prior art compound to make that claimed compound. Id. at 697.

As here, Dillon’s claimed compound demonstrated both expected and additional, unexpected properties. Those additional unexpected properties, however, did not upset an already established motivation to modify a prior art compound based on the expected properties of the resulting compound. Id. at 693. We therefore upheld the finding of obviousness despite Dillon’s arguments of unexpected properties.

Contrary to BMS’s argument, unexpected results do not per se defeat, or prevent, the finding that a modification to a lead compound will yield expected, beneficial properties. 3 Rather, as secondary considerations of nonobviousness, they come into play in determining “the ultimate question of patentability.” Dillon, 919 F.2d at 692-93; Procter & Gamble, 566 F.3d at 997-98 (citing to Dillon and finding evidence of superior properties such as potency and safety that were “unexpected and could not have been predicted” could outweigh evidence of obviousness).

Secondary considerations of nonobviousness “must always when present be considered,” and can serve as an important check against hindsight bias. See Cyclobenzaprine, 676 F.3d at 1075-76, 1079 (quoting Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1538-39 (Fed.Cir.1983)). While secondary considerations must be taken into account, they do not necessarily control the obviousness determination. Pfizer, 480 F.3d at 1372; see also KSR, 550 U.S. at 426, 127 S.Ct. 1727. Here, the district court found evidence of some secondary considerations of nonobviousness, including commercial success, long-felt need, and unexpected results. On appeal, BMS focuses primarily on unexpected results.

To be particularly probative, evidence of unexpected results must establish that there is a difference between the results obtained and those of the closest prior art, and that the difference would not have been expected by one of ordinary skill in the art at the time of the invention. Kao Corp. v. Unilever U.S., Inc., 441 F.3d 963, 970 (Fed.Cir.2006); see also Pfizer, 480 F.3d at 1371. Unexpected properties, however, do not necessarily guarantee that a new compound is nonobvious. While a “marked superiority” in an expected property may be enough in some circumstances to render a compound patentable, a “mere difference in degree” is insufficient. In re Papesch, 50 CCPA 1084, 315 F.2d 381, 392 (CCPA 1963); In re Hoch, 57 CCPA 1292, 428 F.2d 1341, 1344 n. 5 (CCPA 1970) (explaining that unexpected “differences in properties” can mean “significant difference in degree of the same property” amounting to a “marked superiority” for purposes of evaluating unexpected results) (quotation omitted).

And “differences in degree” of a known and expected property are not as persuasive in rebutting obviousness as differences in “kind” — i.e., a new property dissimilar to the known property. Compare In re Merck, 800 F.2d 1091, 1099 (Fed.Cir.1986) (finding evidence that the new drug was a more potent sedative and stronger anticholinergic effect than the pri- or art was insufficient to outweigh the evidence of obviousness), with In re Albrecht, 514 F.2d 1389, 1396 (CCPA 1975) (reversing an obviousness rejection based on evidence of additional antiviral activity “totally dissimilar to any activity previously disclosed for prior art”). When assessing unexpected properties, therefore, we must evaluate the significance and “kind” of expected results along with the unéxpected results. See Hoffmann-La Roche Inc. v. Apotex Inc., 748 F.3d 1326, 1333-34 (Fed.Cir.2014) (“The evidence of superior efficacy does nothing to undercut the showing that there was a reasonable expectation of success with the 150 mg monthly dose, even if the level of success may have turned out to be somewhat greater than would have been expected.”); In re Eli Lilly & Co., 902 F.2d 943, 948 (Fed.Cir.1990) (finding claims obvious when “[patentee] has not shown that a significant aspect of his claimed invention is unexpected in light of the prior art”).

BMS primarily relies on three contentions for unexpected properties: (1) high potency against hepatitis B, (2) a larger than expected therapeutic window, and (3) a high genetic barrier to resistance. J.A. 150-51. The antiviral activity of enteeavir, however, was not entirely unexpected because, as the district court found, it was already known in the prior art that 2'-CDG was effective against hepatitis B. J.A. 150. Specifically, the Price reference suggested that the structurally similar enteeavir would likely have excellent antiviral activity against hepatitis B because 2-CDG already demonstrated “excellent activity” against the virus. J.A. 2086. Moreover, the Price reference also suggested — and Teva’s expert, Dr. Heath-cock, testified — that 2'-CDG was known to have a good therapeutic window. Thus, while the district court found that enteca-vir’s degree of effectiveness was unexpected, it also noted that entecavir’s “effectiveness against hepatitis B without known toxicity issues” was “not unexpected” in light of the structurally similar 2'-CDG. J.A. 150 (emphasis added). As for the high genetic barrier to resistance, the district court properly credited this attribute as an unexpected property. All taken together, the district court found that the proffered evidence of unexpected-properties provided “some support to BMS’s argument as to nonobviousness,” but did not find it sufficient. J.A. 151-53.

We give deference to a lower court’s factual findings regarding evidence of secondary considerations. See In re Inland Steel Co., 265 F.3d 1354, 1366 (Fed.Cir.2001) (“An examination for unexpected results ‘is a factual, evidentiary inquiry,’ ... and we give the [fact-finding tribunal] broad deference in its weighing of the evidence before it.”); Santarus, Inc. v. Par Pharma., Inc., 694 F.3d 1344, 1358 (Fed.Cir.2012) (“The district court’s findings of fact are entitled to deference, and [Paten-tee] failed to show that they are clearly erroneous.”) (citation omitted). Accordingly, we defer to the district court’s finding on unexpected results.

BMS’s remaining arguments regarding unexpected results are that the district court committed legal error by (1) comparing enteeavir to another hepatitis B drug on the market instead of the closest prior art, 2'-CDG; and (2) inappropriately looked to what the inventor knew at the time of the invention — instead of one of ordinary skill in the art — to determine what was expected. BMS is correct on both counts. See Kao, 441 F.3d at 970 (explaining that when unexpected results are used as evidence of patent’s nonobviousness, results must be shown to be unexpected compared with closest prior art); Standard Oil Co. v. Am. Cyanamid Co., 774 F.2d 448, 454 (Fed.Cir.1985) (holding that “obviousness is determined entirely with reference to a hypothetical ‘person having ordinary skill in the art’ ” and the “actual inventor’s skill is irrelevant” to the obviousness inquiry) (emphasis omitted). However, both errors were harmless. 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.” J.A. 150 (emphasis added). And regardless of what the district court determined the inventor may have known, the prior art, the trial record, and the 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.

Hi

As for BMS’s arguments regarding evidence of commercial success and long-felt need, we find no clear error with the district court’s factual findings. The district court found that Baraclude® achieved commercial success based on sales and market share, but it was “less dynamic” than BMS represented. J.A. 138. Bara- clude®’s market share built up gradually over four years and it ultimately held onto the top spot for less than a year. Evidence also showed that two other competitors were able to gain market share more quickly at launch than Baraclude®. J.A. 54, 137. Even BMS’s internal documents viewed Baraclude®’s market performance as “sub optimal.” J.A. 136-38.

On long-felt need, three other drugs for treating hepatitis B were invented before the filing date of entecavir. J.A. 147. These three drugs also gained FDA approval before entecavir. Finally, enteca-vir’s inventors did not know about its hepatitis B properties until four years after the filing date, and by then the first FDA-approved hepatitis B treatment was launched. J.A. 147-48. Therefore, we agree with the district court that the evidence of long-felt need is of limited value to BMS.

Finally, BMS argues that the district court improperly compared the weight of the different secondary considerations against each other when it summarized that the factors were “mixed.” J.A. 152. We understand the district court to be noting that some categories of evidence simply were not as helpful to BMS’s case as others. We do not read the opinion as suggesting that unhelpful evidence somehow diminished the strength of the more persuasive forms of evidence.

III.

We agree with the factual findings on secondary considerations and find no clear error. As stated previously, we also agree with the district court’s finding that the record demonstrates strong evidence of obviousness. After considering all of the findings for and against obviousness, as well as Teva’s burden of proof, we see no basis to disturb the district court’s ultimate legal conclusion, and we affirm the judgment that claim 8 of the '244 patent is invalid as obvious.

AFFIRMED

Costs

No costs.

1 . The district court found that Teva did not demonstrate by clear and convincing evidence that the inventor and prosecuting attorneys committed inequitable conduct. Teva does not raise the inequitable conduct issue on appeal.

2 . BMS asserts that the district court erroneously failed to consider other nucleoside analogs as potential lead compounds. We disagree. The district court assessed all three classes of nucleosides (furanosides, acyclics, and carbocyclics) but explained that the field for furanosides and acyclics, compared to car-bocyclics, was "crowded” and "fairly well developed” to the point that one would have a "hard time finding [a furanoside or acyclic that] someone else hadn't already tried.” J.A. 19, 94. This is in marked contrast to the "fertile” field of research in carbocyclics. Id.

3 . We have held an invention to be obvious despite findings of unexpected results. See, e.g., Allergan, Inc. v. Sandoz Inc., 726 F.3d 1286, 1293 (Fed.Cir.2013); Alcon Research, Ltd. v. Apotex, Inc., 687 F.3d 1362, 1365, 1369-70 (Fed.Cir.2012); Pfizer, 480 F.3d at 1372.