dissenting.
I respectfully dissent from the court’s opinion. There is no legal basis for the granting of a preliminary injunction, and its issuance is an abuse of discretion. Although generally the denial or issuance of a preliminary injunction is within the broad discretion of the district court, the decision of the district court must be reversed when it abuses its discretion. See Cybor Corp. v. FAS Technologies, Inc., 138 F.3d 1448, 1460 (Fed.Cir.1998) (en banc) (“A district court abuses its discretion when its decision is based on clearly erroneous findings of fact, is based on erroneous interpretations of the law, or is clearly unreasonable, arbitrary or fanciful.”).
“A preliminary injunction requires the movant to show four factors: (1) a reasonable likelihood of success on the merits, (2) the prospect of irreparable harm, (3) a balance of the parties’ hardships in favor of injunction, and (4) no potential injury to an important public interest.” See Jeneric/Pentron, Inc. v. Dillon Co., 205 F.3d 1377, 1380 (Fed.Cir.2000). When the district court considers the four factors, “the likelihood of success factor plays a key role,” id., and that is the factor I will focus on in my dissent. Because of “the extraordinary nature of the relief, the patentee carries the burden of showing likelihood of success on the merits,” in light of the presumptions and burdens that will inhere at trial, with respect to the patent’s validity, enforceability, and infringement. Nutrition 21 v. United States, 930 F.2d 867, 869 (Fed.Cir.1991) (emphasis in original); see also Amazon.com, Inc. v. Barnesand-noble.com, Inc., 239 F.3d 1343, 1350 (Fed. Cir.2001). If the defendant “raises a substantial question concerning either infringement or validity, i.e., asserts an infringement or invalidity defense that the patentee cannot prove ‘lacks substantial merit,’ the preliminary injunction should not issue.” Amazon.com, 239 F.3d at 1350-51. This court has explained that:
In resisting a preliminary injunction, however, one need not make out a case *1372of actual invalidity. Vulnerability is the issue at the preliminary injunction stages, while validity is the issue at trial. The showing of a substantial question as to invalidity thus requires less proof than the clear and convincing showing necessary to establish invalidity itself.
Abbott Labs. v. Andrx Pharms., Inc., 452 F.3d 1331, 1335 (Fed.Cir.2006) (herein “Andrx”) (quoting Amazon.com, 239 F.3d at 1359).
The majority opinion postulates that the findings of the district court are correct. It is error to so conclude because the district court failed to properly consider and weigh the ample evidence produced by Sandoz that clearly established a substantial question of invalidity and rendered the patent vulnerable to an invalidity challenge at trial. Instead, the district court erroneously required proof of clear and convincing evidence of invalidity at the preliminary stages of the proceedings. As I explain below, this conflicts with our clearly established precedent.
Under our precedent, the likelihood of success factor is properly analyzed by considering whether the alleged infringer raises a substantial question as to validity. See, e.g., E.I. du Pont de Nemours & Co. v. MacDermid Printing Solutions, L.L.C., 525 F.3d 1353, 1358 (Fed.Cir.2008) (“[I]f the accused infringer raises a substantial question regarding validity, the district court should find that the patentee has not shown a likelihood of success on the merits.” (internal quotation marks omitted)). Indeed, this court has consistently held that an alleged infringer can negate the patentee’s likelihood of success on the merits — and thus defeat a preliminary injunction — by raising a substantial question as to validity. For example, in Genentech, this court explained:
In order to demonstrate that it has a likelihood of success, [the patentee] must show that, in light of the presumptions and burdens that will inhere at trial on the merits, (1) it will likely prove that [the alleged infringer] infringes the [] patent and (2) its infringement claim will likely withstand [the alleged infringer’s] challenges to the validity and enforceability of the [ ] patent. In other words, if [the alleged infringer] raises a “substantial question” concerning validity, enforceability, or infringement (i.e., asserts a defense that [the patentee] cannot show “lacks substantial merit”) the preliminary injunction should not issue. More specifically, with regard to [the alleged infringer’s] validity defenses, the question on appeal is whether there is substantial merit to [the alleged infringer’s] assertion that the [] patent claim [is invalid].
Genentech, Inc. v. Novo Nordisk A/S, 108 F.3d 1361, 1364 (Fed.Cir.1997). Our subsequent cases consistently applied the law as it was explained in Genentech. See, e.g., Tate Access Floors v. Interface Architectural Res., 279 F.3d 1357, 1365 (Fed.Cir.2002) (“In order to demonstrate likely success on the merits, [the patentee] must show that, in light of the presumptions and burdens applicable at trial, it will likely prove that [the alleged infringer] infringes the asserted claims of the [ ] patent and that the patent will likely withstand [the alleged infringer’s] challenges to its validity. If [the alleged infringer] raises a substantial question concerning infringement or validity, meaning that it asserts a defense that [the patentee] cannot prove ‘lacks substantial merit,’ the preliminary injunction issued improperly.” (internal citations omitted; citing Genentech, 108 F.3d at 1364 and Amazon.com, 239 F.3d at 1350-51)).
Our most recent cases continue to adhere to the law as it was explained in Genentech. See, e.g., Erico Int’l Corp. v. Vutec Corp., 516 F.3d 1350, 1352, 1354 *1373(Fed.Cir.2008) (stating that “[the alleged infringer] must show a substantial question of invalidity to avoid a showing of likelihood of success” and vacating the preliminary injunction “[b]ecause this court finds that [the alleged infringer] has raised a substantial question as to the validity of the patent at issue”); PHG Techs., LLC v. St. John Cos., 469 F.3d 1361, 1365, 1369 (Fed.Cir.2006) (explaining that “in order to defeat the injunction on grounds of potential invalidity, [the alleged infringer], as the party bearing the burden of proof on the issue at trial, must establish a substantial question of invalidity” and holding the district court clearly erred in finding the patentee was likely to succeed “because [the alleged infringer] has satisfied its burden of raising a substantial question of invalidity”). Thus, under our clearly established precedent, when the alleged in-fringer raises a substantial question regarding validity, a preliminary injunction cannot issue because the patentee has failed to demonstrate a likelihood of success on the merits.
While Section VI of the opinion contains a superfluity of citations, it does not state the law relevant to this case. It is a pleasant, ambulatory, and meandering discussion; but it is not required to decide this case, is not part of the majority opinion, and is clearly dicta. Although Section VI discusses the relevant four-factor test and properly emphasizes the likelihood of success factor, it ignores the way this court has consistently analyzed whether or not a patentee has demonstrated it will likely succeed at trial. The real question before us in this case, as our precedent clearly explains, is whether the district court erred in finding that Sandoz had not established a substantial question as to the obviousness of the '718 patent. See, e.g., Genentech, 108 F.3d at 1364 (“[T]he question on appeal is whether there is substantial merit to [the alleged infringer’s] assertion that the [ ] patent claim [is invalid].”). Sandoz has, in fact, raised and substantially established that the validity of the '718 patent is vulnerable, and on the record before us, Abbott failed to prove the invalidity defense “lacks substantial merit.” See Amazon.com, 239 F.3d at 1350-51 (“If [the alleged infringer] raises a substantial question concerning either infringement or validity, i.e., asserts an infringement or invalidity defense that the patentee cannot prove ‘lacks substantial merit,’ the preliminary injunction should not issue.”). Thus, the district court committed reversible error when it analyzed the likelihood of success factor and determined that Abbott had established it would likely succeed on the merits. In light of that error, I would vacate the preliminary injunction and remand for reconsideration and reweighing of the injunctive factors. Moreover, various additional legal errors taint the district court’s decision.
I.
The district court’s grant of a preliminary injunction rested on only two claims, claims 1 and 4 of the '718 patent. Claim 1 reads:
a pharmaceutical composition for extended release of an erythromycin derivative in the gastrointestinal environment, comprising
an erythromycin derivative and
from about 5[%] to about 50% by weight of a pharmaceutically acceptable polymer [1], so that when ingested orally, *1374the composition induces statistically significantly lower mean fluctuation index [DFL] in the plasma than an immediate release composition of the erythromycin derivative while maintaining bioavailability substantially equivalent to that of the immediate release composition of the er-ythromycin derivative.
'718 Patent, col.11 ll.28-38. Claim 4 similarly claims an erythromycin derivative and “from about 5[%] to about 50% by weight of a pharmaceutically acceptable polymer” but has different PK parameters. Id. col. 11 ll.48-58. Claim 2 and claim 3 are dependant claims of claim 1. Claim 2 claims “the pharmaceutical composition of claim 1, wherein the polymer is a hydro-philic water-soluble polymer.” Id. col.11 ll.39-40. Claim 3 claims “the pharmaceutical composition of claim 2, wherein the polymer is selected from the group consisting of polyvinylpyrrolidine, hydroxypropyl cellulose, hydroxypropylmethyl cellulose, methyl cellulose, vinyl acetate/crotonic acid copolymers, methacrylic acid copolymers, maleic anhydride/methyl vinyl ether copo-lymers and derivatives and mixtures thereof.” Id. col. 1111.41 — 47.
Claims 1 and 4 of the '718 patent have three basic limitations: (a) an erythromy-cin derivative; (b) 5% to 50% by weight of a pharmaceutically acceptable polymer; and (c) various PK parameters. In the preferred embodiment, the erythromycin derivative is clarithromycin and the phar-maceutically acceptable polymer is HPMC at 10% to 30% by weight of the composition. Claim 4 requires PK parameters be such that “upon oral ingestion, maximum peak concentrations of the erythromycin derivative are lower than those produced by an immediate release pharmaceutical composition, and [AUC] and the minimum plasma concentration are substantially equivalent to that of the immediate release pharmaceutical composition.” Id. col.ll 11.52-58. Claim 1 achieves similar results with slightly different parameters. For claim 1, the composition must have a “statistically significantly lower mean fluctuation index,” DFL, which is defined in the specification as DFL = (Cmax - Cmin / CAv ), id. col.3 11.29-30, and substantially equivalent bioavailability, which the district court found meant that the “[ER] AUC values must be between 80% to 125% within a 90% confidence level as compared to the immediate release composition AUC values.”2 Abbott Labs. v. Sandoz, Inc., 500 F.Supp.2d 807, 831 (N.D.Ill.2007) (herein “Sandoz I ”).
II.
Sandoz based its obviousness arguments primarily on three prior art references, which it argues combined with common sense and the ordinary skill of the art at the time make the '718 patent anticipated or obvious. First, Sandoz argues that the PTC Application WO 95/30422 (“the '422 publication”) filed by Pfizer, discloses a controlled release dosage form of azithro-mycin, which like clarithromycin is an er-ythromycin derivative.3 According to the disclosure, these controlled released compositions operate to release the drug substantially slower than the immediate re*1375lease versions to reduce GI side effects. And, as Sandoz points out, the controlled release compositions disclosed in the '422 publication include a hydrophilic polymer composition of azithromycin, with a preferred embodiment being a matrix tablet containing 15% to 35% HPMC. Second, Sandoz noted that the '190 patent owned by Abbott discloses and claims controlled release compositions of clarithromycin in an (non-polymer) alginate matrix which are administered once a day and have slowed absorption such that they are bioe-quivalent with the current immediate release twice-a-day compositions and maintain therapeutic levels at 24 hours after ingestion. Claim 14 of the '190 patent also claims other marolides including azithro-mycin.4 Third, Sandoz argues that the '571 publication, filed by Eli Lilly, discloses sustained release formulations for antimicrobial agents including clarithromy-cin, which contain an active agent, namely a hydrophilic polymer such as HPMC, and an acrylic polymer. According to the specification, these formulations differ from the prior art that uses just hydrophilic polymers in that they are designed to allow a constant rate of release throughout the GI tract. In particular, the '571 publication disclosed using from about 5% to about 29% by weight hydrophilic polymer, and about 0.5% to about 25% by weight acrylic polymer, with the total weight of the two polymers not exceeding 30% by weight.
In addition to challenging the validity of the '718 patent based on the '190 patent, the '422 publication and the '571 publication, Sandoz also relies on various evidence that the PK parameters specified in the '718 patent were well known in the art and would have been sought by anyone designing a controlled release formulation. Most strikingly, the testimony of one of the inventors named on the '718 patent, Linda Gustavson, an Abbott employee, supports the Sandoz position. In particular, Gus-tavson testified as follows:
Q: Did you tell [the formulations department] what pharmacokinetic parameters there should be?
A: I mean, not specific numbers, but relative to the IR, yes. I told at least Sue that what we needed was *1376a lower Cmax, an AUC that met FDA requirements for bioequiva-lence and a Cmin that was at least comparable to the IR.
Q: And where did you get these parameters?
A: A few years of experience, I guess. They’re the — I mean certainly the Cmax and AUC are very basic PK parameters determined in virtually every study that has pharmacoki-netics. Cmin [ ] might or might not be important depending upon the drug you were talking about and what part of the pharmacokinetics you though might be associated with efficacy or safety. For clari-thromycin, there was some thought that keeping the concentrations above some minimum level might be at least in part important to maintaining effectiveness efficacy.
Q: Would you say that these PK parameters were pretty much known in the art?
A: Absolutely, yes.
Furthermore, Sandoz submitted references from 1983 (over a decade before the '718 application was submitted), which explained, inter alia, that the “objectives and possible advantages of controlled release dosage” forms included “maintaining] therapeutic drug levels,” “reducing] dosing frequency,” “reducing] fluctuations in drug levels,” and “reducing] side effects.” And the reference explained that the “essence of controlled drug release” was to “obtain prolonged circulating drug levels with less fluctuation compared to conventional dosage forms, and to achieve these with less frequent drug administration.” They also submitted references showing that HPMC was considered the “controlled release agent of choice” in the field.
The district court, writing prior to KSR, found that, despite this Court’s decision in Andrx to the contrary, claims 1 and 4 of the '718 patent were not obvious (at least based on the preliminary record). The trial court’s finding rested on the fact that, contrary to this court’s conclusion in Andrx, new evidence established that the '190 patent, the '571 publication, and the WO '422 publication did not disclose the specific PK limitations of the '718 patent. The court recognized that “[generally, a showing that there is an established structural relationship between a prior art composition and the claimed composition demonstrates a prima facie case of obviousness.” Sandoz I, 500 F.Supp.2d at 840. See In re Dillon, 919 F.2d 688, 692 (Fed.Cir.1990) (en banc) (“[S]tructural similarity between claimed and prior art subject matter, proved by combining references or otherwise, where the prior art gives reason or motivation to make the claimed compositions, creates a prima fa-cie case of obviousness.”). Still, the court concluded that Abbott had preliminarily rebutted this showing by showing the specific PK properties embodied in the claims were unobvious. The trial court found that “[t]o succeed on its obvious[ness] claim, Sandoz must produce evidence indicating that the PK limitations were disclosed in the prior art or were at the very least inherent to the structural limitations of the prior art compositions.” Sandoz I, 500 F.Supp.2d at 840. Sandoz, the court found, had not done so. Id. Moreover, the trial court found that “because the '190 prior art does not disclose the [specific] PK profile of the '718 patent, a person skilled in the art would not be motivated to look at the WO '422 publication and interchange clarithromycin for azithromy-cin.” Id. at 841.
Subsequently, the district court denied Sandoz’s motion for a stay of the preliminary injunction pending appellate review in light of the just issued KSR opinion. The court held that under KSR it was still *1377necessary to “demonstrate the presence of all claim limitations in the prior art” and that Sandoz had not produced evidence indicating that the PK limitations were disclosed in the prior art or inherent to the structural limitations of the prior art compositions. Abbott Labs. v. Sandoz, 500 F.Supp.2d 846, 851-53 (N.D.Ill.2007). According to the district court, it thus had not and did not need to reach the TSM test (or any change in the application of this test brought on by KSR). Id. at 853.
On appeal, there is no real dispute that the '571 publication expressly discloses a “sustained release matrix formulation in tablet form comprising ... erythromycin” and containing from about 5% by weight to about 29% by weight of a hydrophilic polymer, thus meeting all of the structural limitations of the '718 claims. Moreover, the prior art clearly disclosed sustained release versions of clarithromycin and creating extended release formulations of er-ythromycin derivatives using polymers, preferably HPMC. And evidence shows that the desirability of the PK parameters claimed were well known in the art.
In light of this evidence, Sandoz raised a substantial question as to the obviousness of the '718 patent. The district court’s decision to the contrary constituted an abuse of discretion. First, it was clearly error to find, as a matter of law, that since none of the prior art references cited by Sandoz explicitly disclosed a composition that had the PK limitations of the '718 patent, it had failed to demonstrate “the presence of all claim limitations in the prior art,” and therefore that the '718 invention could not be obvious. Contrary to the majority, this holding relies on an improperly limited view of what types of references can be combined to show obviousness and an impermissibly cramped view of the Supreme Court’s holding in KSR. There is no absolute requirement that each claim limitation be disclosed in a prior art reference. See, e.g., Takeda Chem. Indus. v. Alphapharm Pty., Ltd., 492 F.3d 1350, 1356 (Fed.Cir.2007) (“We have held that structural similarity between claimed and prior art subject matter, proved by combining references or otherwise, where the prior art gives reason or motivation to make the claimed compositions, creates a prima facie case of obviousness.” (emphasis added)); Tegal Corp. v. Tokyo Electron Am., Inc., 257 F.3d 1331, 1349 (Fed.Cir.2001) (acknowledging that a claim could be obvious over a single prior art reference that does not disclose one of the limitations in the claim). Rather, in all cases, the touchstone of the analysis is whether 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; KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 127 S.Ct. 1727, 1734, 167 L.Ed.2d 705 (2007); see also Takeda, 492 F.3d at 1357 (explaining that “in cases involving new chemical compounds” to show a prima facie case of obviousness one must “identify some reason that would have led a chemist to modify a known compound in a particular manner” (emphasis added)). Thus, a given claim limitation may be obvious over the prior art even if no single reference had specifically disclosed that limitation. Moreover, even assuming an absolute rule that to be obvious a claim must be a combination of elements disclosed in the prior art, that standard was met here. As the Supreme Court reiterated in KSR, “inventions in most, if not all, instances rely upon building blocks long since uncovered, and claimed discoveries almost of necessity will be combinations of what, in some sense, is already known.” 127 S.Ct. at 1741. In other words, it is the rare invention that is not a *1378combination of prior art elements.5 And this is not one of such rare cases. Whether or not the prior art disclosed compounds displaying the particular PK parameters in the '718 patent, Sandoz did provide evidence suggesting that the PK parameters disclosed in the '718 patent were absolutely known in the art and that the prior art established that they were desirable in an extended release formula (and indeed, that at least the AUC equivalence and lower Cmax were most likely essential to an extended release formula, at least one that would be approved by the FDA). This is sufficient to show that the claims might be a combination of elements previously known in the art. The prior art on record disclosing the PK limitations is of course further removed from the invention than, for example, a patent that disclosed a related drug formulation with the same PK limitations as the '718 patent. But while this may well make the former less likely than the latter to make the '718 patent claims obvious, there is nothing as a matter of law that prevents the invention from being considered an obvious combination of the prior art teachings contained in the current preliminary record. Cf Aventis Pharma Deutschland GmbH v. Lupin, Ltd., 499 F.3d 1293, 1301 (Fed.Cir.2007) (explaining that while it is necessary for there to be “some articulated reasoning with some rational underpinning to support the legal conclusion of obviousness ... such reasoning need not seek out precise teachings directed to the specific subject matter of the challenged claim” (internal quotation marks omitted)).6
Second, it is not dispositive that Abbott was not absolutely certain that using the formulations disclosed in the '422 patent would create a formulation with the desired PK parameters. Even before KSR, this court’s “case law [was] clear that obviousness cannot be avoided simply by a showing of some degree of unpredictability in the art so long as there was a reasonable probability of success.” Pfizer, 480 F.3d at 1364. And as the Supreme Court stated in KSR, “[w]hen there is a design need or market pressure to solve a problem and there are a finite number of identified, predictable solutions, a person of ordinary skill has good reason to pursue the known options within his or her technical grasp. If this leads to the anticipated success, it is likely the product not of innovation but of ordinary skill and common sense.” 127 S.Ct. at 1742 (“One of the ways in which a patent’s subject matter can be proved obvious is by noting that there existed at the time of invention a known problem for which there was an obvious solution encompassed by the patent claims.”).
*1379Third, the long standing precedent of this court and our predecessor, recently highlighted and relied upon in Pfizer is that “discovery of an optimum value of a variable” in a known process or composition is “usually obvious.” 480 F.3d at 1368 (citing In re Peterson, 315 F.3d 1325, 1330 (Fed.Cir.2003); In re Boesch, 617 F.2d 272, 276 (CCPA 1980); In re Alter, 42 C.C.P.A. 824, 220 F.2d 454, 456 (1955)). Accordingly, in Pfizer, for example, the court found that the optimization of a pharmaceutical compound to determine which acid salt was best was obvious, where “the prior art heavily suggests the particular anion used to form the salt.” Id. Similarly, here, if the PK parameters claimed were well known, and only routine experimentation by someone skilled in the art would have been necessary, in light of the HPMC formulations disclosed by the '422 publication and '571 publication, to create an ER clarithromycin-HPMC formulation with the claimed PK parameters, this would be sufficient to create a prima facie case of obviousness.
Accordingly, the district court erred in concluding that the fact that the '422 publication and '190 patent did not disclose the PK limitations of the asserted claims precluded a finding of obviousness. This legal error constitutes an abuse of discretion. And contrary to the holding of the district court, Sandoz has raised a “substantial question concerning” the obviousness of the asserted claims. On this basis, I would reverse the decision of the district court.
III.
Sandoz also argues that several acts by Abbott during the prosecution of its patent application constitute inequitable conduct and, thus, show that the district court abused its discretion in not rejecting the motion for a preliminary injunction based on the likelihood that the patents would be declared unenforceable.
The Patent Examiner initially rejected the claims of the '718 application and requested that Abbott show that one of its prior art compositions of clarithromycin, which was described as an immediate release pediatric suspension formula, did not have the same extended release properties as Abbott’s claimed invention. In response, Abbott submitted a declaration by Linda Gustavson stating that the Cmax of the ER clarithromycin “is statistically significantly lower than that for IR formulation given twice daily.” J.A. 10015. Abbott now admits that this statement was incorrect — that the data Gustavson relied on did not show a statistically significant lowering of the Cmax — but only a non-statistically significant apparent lowering, and Gustavson herself admits that she never performed any statistical analysis of the data and would not have known how to do it.
The district court found that the con-cededly false statement was immaterial since all the Examiner asked was whether the two products have the same PK properties. According to the district court, “[g]iven the accuracy of the ultimate conclusion — that the extended release formulation was indeed different from the immediate release suspension formulation, Gustavson’s declaration of a ‘statistically significantly lower’ Cmax is immaterial despite the fact that it satisfies the definition of ‘material’ provided by 37 C.F.R. § 1.56(b).” Sandoz I, 500 F.Supp.2d at 822. The district court reasoned that despite meeting the standard for materiality of § 1.56(b) a reasonable examiner would not have considered the statement important. Moreover, the district court emphasized that no claim of the '718 patent requires the extended release formulation to have a statistically significant lower Cmax than the immediate release formulation.
*1380The Gustavson statement was material, or more to the point, there is substantial likelihood that Sandoz would be able to so establish at the merits stage. First, contrary to the erroneous conclusion of the district court, we have held that “all misstatements or admissions that satisfy [37 C.F.R. § 1.56(b) ] are considered material.” Monsanto Co. v. Bayer Bioscience N.V., 514 F.3d 1229, 1237 n. 11 (Fed.Cir.2008). In addition, while no claim element in the '718 patent specifically states that the Cmax value must be statistically significantly lower, it does require Cmax values that are “lower” than in the immediate release formulation. Despite the fact that in other claim elements Abbott uses the term “statistically significantly lower,” it is far from clear that one can establish that the Cmax value is lower if the data does not show a statistically significant difference, which by definition, as ordinarily understood, means that the data cannot conclusively establish that there is a real difference. This court previously recognized that “there is little in the ['718] patent itself that establishes the differences (if any) between parameters that are simply ‘lower’ rather than ‘statistically significantly lower.’ ” Andrx, 452 F.3d at 1339 n. 4. Moreover, regardless of the claim construction, it would be important (if not dispositive) to a reasonable examiner to know that Abbott did not have data which showed a lower Cmax to any statistical significance over the structurally similar prior art suspension formulas in deciding whether to allow the claim over this prior art.
On this basis alone, the district court abused its discretion because it created such a high bar for materiality that in essence no statement or withholding of information would be material if it would not change the ultimate outcome of allowing the patent. This is inconsistent with our precedent. See, e.g., Hoffmann-La Roche, Inc. v. Promega Corp., 323 F.3d 1354, 1368 (Fed.Cir.2003) (“The fact that the examiner did not have to rely on the purity representations in issuing the patent is not inconsistent with a finding of materiality. Although the inventors’ statements regarding purity were not the principal focus of the office action response, they were clearly an important aspect of it. Under the circumstances, a reasonable examiner would have wanted to know that the patentability argument based on purity was unsupported by the experimental results cited by the inventors.” (internal citation omitted)); Merck & Co. v. Danbury Pharmacal, Inc., 873 F.2d 1418, 1421 (Fed.Cir.1989) (rejecting a “but for” standard of materiality).
In addition, while the district court did not reach the issue of intent, the fact that Gustavson submitted a declaration to the PTO in which she claimed to have found a statistically significant lowering of Cmax despite now admitting to never having done any statistical analysis is sufficient circumstantial evidence of intent to raise a substantial question of inequitable conduct, if not necessarily to prove inequitable conduct on the merits.
The district court also found that the failure by Abbott to disclose a new study7 *1381was immaterial because Abbott simply “chose to rely on the results of several other studies that showed differing mean DFL values” and the totality of the evidence demonstrates that the prior art formulation did not have the “same broad PK properties as those claimed for the ER formulation.” Sandoz I, 500 F.Supp.2d at 824. The test is not whether the Examiner would have refused to allow the patent to issue without the information, but just whether it would have been “important” to her consideration. Here the extent the PK parameters of the ER formulation differed from the clarithromycin formulations in the prior art was the primary focus of the examiner’s concerns regarding patent-ability, and Abbott’s ability to establish sufficient differences was the basis for allowing the claims. It was not for Abbott to decide unilaterally that it preferred the results of one set of studies that supported patentability and therefore could ignore studies reaching the opposite result. Cf. Paragon Podiatry Lab. v. KLM Labs., 984 F.2d 1182, 1193 (Fed.Cir.1993) (finding inequitable conduct for failure to disclose sales data and noting that “where the decision of whether or not to disclose sales before the critical date is close, the case should be resolved by disclosure, not by the applicant’s unilateral decision.”).
Accordingly, I would also vacate the preliminary injunction based on the allegations of inequitable conduct. The evidence raises a substantial question of unenforce-ability that makes the patents vulnerable to being found unenforceable at trial. Thus, the district court erred when it concluded that Abbott had shown it would likely succeed on the merits.
Because of the reasons stated above, I would reverse the district court on the basis that there are substantial questions of both validity and enforceability of the '718 patent preventing a finding of likelihood of success on the merits.
. The specification of the '718 patent states, in a list of definitions, that " 'pharmaceutically acceptable' as used herein, means those compounds which are, within the scope of sound medical judgment, suitable for use in contact with the tissues of humans and lower animals without undue toxicity, irritation, allergic response, and the like, in keeping with a reasonable benefiVrisk ratio, and effective for their intended use in the chemotherapy *1374and prophylaxis of antimicrobial infections.” In my judgment the district court's claim construction is ambiguous as to whether the pharmaceutically acceptable polymer must extend release or whether it can be part of a matrix in which other components extend the release. Sandoz is correct that there needs to be some showing that the polymer acts to extend release.
. These figures are based on FDA definitions.
. Abbott specifically carved out azithromycin from its definition of an erythromycin derivative in the '718 patent. See Andrx, 452 F.3d at 1337.
. In Andrx, this court relied primarily on the '190 patent, as combined with the '422 publication to find that there was a substantial question as to the obviousness of the '718 patent claims. 452 F.3d at 1340-41. First, we concluded that “Teva makes substantial arguments that the '190 patent discloses a clarithromycin composition ... that arguably has the pharmacokinetic parameters required in claim 4 of the '718 patent.” Id. at 1340. And we explained:
Because the '190 patent explicitly discloses only clarithromycin controlled release compositions, yet claims azithromycin compositions, ... Abbott has represented to the [PTO] that the differences between clari-thromycin and azithromycin were such that azithromycin could be substituted into a controlled release clarithromycin composition by a person of ordinary skill in the art without undue experimentation.... As a result, based on Abbott’s own '190 patent, there exists a substantial argument that a person of ordinary skill in the art would be motivated to combine the '422 publication, namely the use of HPMC in extended release macrolide compositions, with the ' 190 patent with a reasonable expectation of success.
Id. at 1341. In this case, Abbott presented evidence at trial suggesting that this court was scientifically incorrect to find that the '190 patent disclosed compounds that arguably had the same PK values as the asserted claims. Sandoz, based on this new evidence, disclaimed any reliance on the scientific evidence of the '190 patent disclosing compounds with the same PK values as the '718 patent. However, contrary to the majority opinion, Sandoz can rely on Andrx's conclusion that it would have been obvious for a person skilled in the art to substitute clari-thromycin for azithromycin in an extended release formula with the anticipation of success without undue experimentation.
. Given KSR’s broad understanding of nearly all or perhaps all inventions being combinations of elements in the prior art, the parties' dispute about whether KSR should be limited to such inventions becomes largely irrelevant. In any event, while KSR's holding is directed particularly at the TSM test, it certainly appears that the Court intended to expound principles of obviousness jurisprudence that were generally applicable. And particularly relevant to the case at bar, this court has already applied KSR's teachings to the question of whether new chemical compositions are obvious in light of the fact that chemists of ordinary skill would attempt to modify known substances in certain ways to “obtain compounds with improved properties.” Takeda, 492 F.3d at 1356. Accordingly, I think the district court clearly erred in concluding that KSR was not relevant to the question of obviousness here.
. See also Pfizer, Inc. v. Apotex, Inc., 480 F.3d 1348, 1362 (Fed.Cir.2007) ("[It] is irrelevant [to the question of obviousness] that none of the anions specifically listed in the '909 patent have a cyclic structure, because the motivation to make amlodipine besylate here is gleaned not only from the prior art as a whole rather than the '909 patent alone, but also from the nature of the problems encountered with the amlodipine maleate tablet formulations sought to be solved by the inventors of the '303 patent.”).
. The written description of the '718 patent states that "The mean DFL values for the controlled release formulation [another Abbott prior art reference disclosing a clari-thromycin formulation] and for the IR are substantially equal in value....” '718 Patent, col. 11 11.18-19. And it explains that lower DFL values for the ER formulation of the '718 patent show that it provides "less variable clarithromycin concentrations throughout the day than the IR and the sustained release compositions.” '718 Patent, col.ll 11.25-26. These statements were correct based on three studies that had been done prior to filing the application. However, a new study W98-268, which Gustavson had knowledge of, and which issued while the application *1381was pending, found that there was a statistically significant lower DFL value for the sustained release formulas than the IR formulas. Gustavson, however, failed to disclose the results of this new study.