concurring in part, dissenting in part.
I agree that neither Dr. Sturis nor Dr. Bork engaged in inequitable conduct, and concur in the judgment reversing the district court’s ruling in that respect. However, Novo’s discovery of the synergistic combination of metformin and repaglinide meets the criteria of patentability, and was incorrectly held to be unpatentable on the ground of obviousness, 35 U.S.C. § 103.
The section 103 determination in this case relates to a synergistic combination of two diabetes drugs. The combination described and claimed in the patent in suit, U.S. Patent No. 6,677,358 (“the '358 patent”), is eight-fold more effective than the additive properties, and is now apparently a treatment of choice for persons whose Type II diabetes had previously been untreatable. It is a lifesaving combination for such persons, and is valuable to other diabetics, for it permits a more flexible treatment regimen than prior products. The Novo inventors pursued this combination despite the advice of other “experts” that they were wasting time and money. Nonetheless the district court, and now my colleagues on this panel, find the combination obvious to them, and invalidate the patent. I respectfully dissent.
Disoussion
“Real world considerations provide ... a solid evidentiary foundation on which to rest a nonobviousness determination.” Minn. Mining & Mfg. Co. v. Johnson & Johnson Orthopaedics, Inc., 976 F.2d 1559, 1575 (Fed.Cir.1992). These “real world considerations” include the realities and challenges of discovering a new medicinal product. The panel majority discards this principle in concluding that the synergistic combination of metformin and repaglinide would have been obvious to a person of ordinary skill.
The question is not whether it would have been obvious to look for synergistic combinations; the question is whether it was obvious that the combination of met-formin and repaglinide would exhibit synergism and that the combination would be 800% more effective than the additive effect of the components separately.
My colleagues reason that because synergism is unpredictable, then if it is found, it is obvious. Maj. Op. at 1351. (“It was apparently well-known in the art that two drugs having different mechanisms for attacking diabetes may be more effective than one, and so drugs were often tested in combination.”). That is not the meaning of “obvious to try.” A new composition is “obvious to try” when it is reasonable to expect that the trial will produce a predictable result. See KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 421, 127 S.Ct. 1727, 167 L.Ed.2d 705 (2007) (“[T]he fact that a combination was obvious to try might show that it was obvious under § 103” if, among other things, “there are a finite number of identified, predictable solutions”). That situation did not here exist. See Eisai Co. v. Dr. Reddy’s Labs., Ltd., 533 F.3d 1353, 1359 (Fed.Cir.2008) (“To the extent an art is unpredictable, as the chemical arts often are, KSR’s focus on these ‘identified, predictable solutions’ may present a difficult hurdle because potential solutions are less likely to be genuinely predictable.”).
It was known that a combination of met-formin and a sulfonylurea — a class of compounds that does not include repaglinide— may or may not have a synergistic effect on blood sugar control, for only some sul-fonylureas showed such effect. Repagli-*1361nide is not chemically similar to the sulfo-. nylureas. It was known at the time of Novo’s invention that repaglinide “differs from the sulfonylureas in its molecular structure, profile of action, and excretion mechanism.” B.H.R. Wolffenbuttel et air, Effects of a new oral hypoglycaemic agent, repaglinide, on metabolic control in sulfo-nylurea-treated patients with NIDDM, Eur. J. Clin. Pharmacol. 45, 1993, at 113. The existence of synergy in some metfor-min-sulfonylurea combinations is not predictive of synergy in the combination of metformin with repaglinide.
The defendants deposed the inventors, who explained their thought processes in experimenting with this combination. The district court, and now my colleagues on this panel, cite the testimony of the inventors, who successfully pursued this unpromising combination, and hold that since the inventors pursued this combination and found the observed synergism, the synergism was obvious. The court uses the inventors’ exceptional intellect against them, rather than the knowledge of the person of ordinary skill.
The district court held that since these inventors pursued this combination, it was obvious to do so. The court stated that the PTO examiner, in granting the patent on the basis of unpredictable synergy, “did not have the benefit of the testimony of Müller [the inventor] and Damsbo [his colleague] as to the results they expected.” Novo Nordisk A/S v. Caraco Pharm. Labs., 775 F.Supp.2d 985, 1009 (E.D.Mich.2011). This is a misunderstanding of the law, for “[o]bviousness may not be established using hindsight or in view of the teachings or suggestions of the inventor.” Para-Ordnance Mfg., Inc. v. SGS Importers Int'l, Inc., 73 F.3d 1085, 1087 (Fed.Cir.1995); see Otsuka Pharm. Co. v. Sandoz, Inc., 678 F.3d 1280, 1296 (Fed. Cir.2012) (“The inventor’s own path itself never leads to a conclusion of obviousness; that is hindsight. What matters is the path that the person of ordinary skill in the art would have followed.”).
My colleagues adopt the district court’s reasoning, ignoring the wisdom counseled by precedent. My colleagues appear to hold that because these Novo scientists studied this combination, it was obvious to try this combination. Such a thesis would expunge patentability for all except random observations. All scientific experiments are conducted with a purpose of inquiry, and all experimenters have a theory of possible outcomes. Such experiments may partake of varying degrees of vision, hope, or expectation on the part of the experimenter, but these are not criteria of patentability.
Patentability is determined not from the position of the inventor, but from the knowledge of the person of ordinary skill. See Standard Oil Co. v. Am. Cyanamid Co., 774 F.2d 448, 454 (Fed.Cir.1985) (“Inventors, as a class, according to the concepts underlying the Constitution and the statutes that have created the patent system, possess something ... which sets them apart from the workers of ordinary skill, and one should not go about determining obviousness under § 103 by inquiring into what patentees (i.e., inventors) would have known or would likely have done.”) (emphases in original).
For questions of biological synergism, predictability is notoriously difficult. In re Luvisi, 52 C.C.P.A. 1063, 342 F.2d 102, 109-10 (1965) (“We do not accept the notion that every suggestion of synergism in the art coupled with a finding of synergism in the practice of the invention automatically compels a conclusion of obviousness .... [S]ome prior art compositions may show little synergism and others show considerable synergistic effects, with the net result that predictability is impossible save the fact that a synergistic re-*1362suit of some kind will probably be found.”); see also Allergan, Inc. v. Sandoz, Inc., — F.3d-,-, No. 2011-1619, 2013 WL 1810852, at *7 (Fed.Cir. May 1, 2013) (holding combination unobvious because there is “no reason why the success of unrelated drugs would make it obvious to one of ordinary skill that a fixed combination of brimonidine and ti-molol could be dosed twice per day without loss of efficacy”). My colleagues contravene precedent, and hold that because some synergism has been observed in some combinations with metformin, any discovery of a unique synergistic combination with unusual properties would have been obvious.
The PTO granted this patent based on the synergistic effect that these inventors discovered and established. This activity was not suggested in the prior art, was not predictable, and was not obvious. The court errs in holding otherwise.
A. The Evidence
There was evidence at trial that repagli-nide was not successful as an antidiabetic drug:
Most companies believed that there was no commercial use or value for repagli-nide given its pharmacodynamic profile when compared to longer-acting sulfony-lureas already on the market or other antidiabetics in development. More generally, no one was willing to go to the expense of doing clinical trials for regulatory approval if the drug had no commercial value.
Trial Tr. vol. 9, 14, Aug. 9, 2010 (testimony of Dr. Michael Mark, Novo scientist who conducted research on repaglinide).
Dr. Peter Müller, the inventor of the patent in suit, pressed for clinical trials despite the skepticism of the clinical investigators. Dr. John Miller, Medical Director of Novo in Australia, who coordinated and supervised the Australian study, testified as follows:
Given that repaglinide was such a short-acting compound that had no effect on FPG [fasting plasma glucose], the investigators tried to explain to the clinical development staff that it made no sense to use fasting glucose as a measurement [because FPG measures] the amount of glucose in blood plasma after the patient has 'not eaten for about eight hours (i.e., overnight).
Trial Tr. vol. 8,121-22, Aug. 5, 2010.
The Australian study was conducted with patients with poorly controlled diabetes, who were treated with metformin alone, repaglinide alone, or a combination of metformin and repaglinide. '358 patent col.7 1.61-col.lO 1.40. The patients receiving the combination exhibited markedly better control of blood sugar than either the patients on metformin monotherapy or those on repaglinide monotherapy. Id. col.9 11.37-58. The claimed drug combination’s ability to control blood sugar, according to fasting plasma glucose (FPG) and glycosy-lated hemoglobin (HbA lc) measurements, was substantially better than the additive effects of metformin and repaglinide. Id. For example, the combination lowered FPG by 2.18 mmol/1, which was over eight times the efficacy observed in patients on metformin alone (0.25 mmol/1). Id. col.9 11.45-58. Repaglinide monotherapy actually increased the patients’ FPG levels by 0.49 mmol/1, contributing to the problem instead of curing it. Id.
Novo scientists were concerned that the Australian study did not prove synergy to a statistical certainty because the study lacked a placebo group. The district court understood that “ethical reasons precluded removing the sick patients from all therapy.” Novo, 775 F.Supp.2d at 1010. When the patent examiner criticized Novo’s quantification of synergy, Novo conducted *1363the rat study to include untreated controls; the results showed a statistically significant synergistic effect between metformin and repaglinide, corroborating the results of the Australian human study.
The PTO issued the '358 patent on the basis of these studies, stating that “[t] he combined administration of repaglinide and metformin resulted in an unexpected synergistic effect on blood glucose levels.” '358 patent “Reasons for Allowance” (May 31, 2003). See Quad Envtl. Techs. Corp. v. Union Sanitary Dist., 946 F.2d 870, 876 (Fed.Cir.1991) (“[C]ourts may take cognizance of, and benefit from, the proceedings before the patent examiner,” although “the question [of validity] is ultimately for the courts to decide, without deference to the rulings of the patent examiner”).
The record shows that defendant Sun Pharmaceutical Industries1 advertises its repaglinide product as synergistic when combined with metformin:
Repaglinide in combination with metfor-min ...,
• Produced a greater improvement in glycemic control than that seen by the sum of the changes with the two agents alone.
Undated Advertisement for Rapilin, Sun’s Repaglinide Product. This is the benefit that Sun told the district court did not exist. .This benefit was unknown until discovered by Novo scientists, and unavailable to public benefit until federally approved at Novo’s initiative and expense.
B. Analysis
My colleagues misunderstand and misapply the “obvious to try” criterion of obviousness. The motivation to develop a new pharmaceutical “is not abstract, but practical, and is always related to the properties or uses one skilled in the art would expect the compound to have, if made.” In re Gyurik, 596 F.2d 1012, 1018 (CCPA 1979). This expectation must be rooted in the prior art and in the person of ordinary skill, not in the ingenuity or creativity of the inventor.
The district court reasoned that since metformin was known to form synergistic combinations with some sulfonylureas, it would be obvious to expect synergism of metformin and repaglinide, for although repaglinide is not a sulfonylurea, it was believed to have a similar mechanism of action as a secretagogue. This analysis is supported only by hindsight, for the record reflects a more complicated reality. It was known that synergy is not exhibited by all sulfonylurea secretagogues, and the district court agreed that one skilled in the art would “perhaps” have expected synergistic results from the metformin-repagli-nide combination. “Perhaps” is not clear and convincing evidence of obviousness of the unusually efficacious results that were obtained. See Microsoft Corp. v. i4i Ltd. P’ship, — U.S.-, 131 S.Ct. 2238, 2242, 180 L.Ed.2d 131 (2011) (holding that invalidity must be proved by clear and convincing evidence).
The panel majority offers the generalization that “earlier metformin/sulfonylu-rea combinations were generally understood to yield synergy.” Maj. Op. at 1354. This is inaccurate, for only some sulfonylu-reas formed synergistic combinations. Synergism was not a general property of such combinations. The particular metfor-min-sulfonylurea combination relied on by the district court to invalidate the '358 patent is metformin-glyburide.2 The district court observed that both glyburide and repaglinide are “insulin secreta-gogues” because they stimulate the secretion of insulin. Indeed, this is how *1364most diabetes treatments work. However, it was well-known that not all insulin stimulants form synergistic combinations with metformin.
There are significant differences between glyburide and repaglinide, in structure and in properties. The compounds are structurally quite different:
[[Image here]]
Glyburide
[[Image here]]
Repaglinide
Glyburide is “long-acting,” having a biological half-life of between ten and twenty hours. Repaglinide is “short-acting,” with a half-life of about one hour. The record repeatedly states that the short life of repaglinide deterred interest in this compound for treatment of diabetes.
The only prior art of record comparing repaglinide to glyburide, the 1993 Wolffen-buttel article cited supra, described the dissimilar effects of the two drugs on blood sugar control in diabetics:
After 12 weeks glibenclamide [glyburide] had reduced fasting blood glucose levels without any effect on postprandial blood glucose, whereas repaglinide had significantly lowered postprandial blood glucose, but with no effect on fasting blood glucose.
Eur. J. Clin. Pharmacol. 45, at 115. This observation was explained by repaglinide’s “mode of action and short plasma half-life.” Id.
Caraco’s expert Dr. Accili conceded at trial that repaglinide would have been expected to have “at best a small impact on fasting plasma glucose.” Trial Tr. vol. 4, 52-53, June 7, 2010. This is in contrast to long-acting sulfonylureas such as glyburide, which were known to reduce fasting plasma glucose even absent combination with metformin.
The structural and functional disparities between repaglinide and glyburide render it unreasonable to expect the repaglinide-metformin combination to have synergistic properties superior to the prior art combination of glyburide and metformin. See In re Lain, 747 F.2d 703, 707 (Fed.Cir.1984) (“[A] relevant property of a compound cannot be ignored in the determination of nonobviousness”). As Dr. Accili acknowledged, “[a]ny time multiple things are used [in combination therapy], the potential for error increases.” Trial Tr. vol. 3, 37, June 3, 2010.
The district court did not address the known differences between repaglinide and the sulfonylureas, including glyburide. The court did not mention the Wolffenbut-tel article, although it was the only refer*1365ence to compare repaglinide and a sulfony-lurea. In the search for scientific truth “[o]ne' cannot ... pick and choose among isolated disclosures in the prior art to deprecate the claimed invention.” In re Fine, 837 F.2d 1071, 1075 (Fed.Cir.1988); it is necessary to consider prior art that supports unobviousness of the claimed invention, as well as that which weighs against it. In re Young, 927 F.2d 588, 591 (Fed. Cir.1991).
The district court’s obviousness determination was based on the court’s finding that two prior art studies on the metfor-min-glyburide combination “reported greater reductions in HbAlc and FPG than those of the [Australian] Study.” The district court stated that these studies represented the “closest prior art,” and invalidated the '358 patent because “[t]he evidence does not establish that the claimed combination therapy produces clinical results superior to those produced by the closest prior art.” Novo, 775 F.Supp.2d at 1011-12. That is, the district court held that because the metformin-glyburide combination appeared to control diabetes as well as the metformin-repagli-nide combination, the patent was invalid for obviousness. That is not the law of obviousness. See In re Chupp, 816 F.2d 643, 646 (Fed.Cir.1987) (“To be patentable, a compound need not excel over prior art compounds in all common properties.”). Further, the district court did not discuss the evidence of effectiveness in difficult-to-treat cases of Type II diabetes, or other differences that were not predictable from prior art.
The question is whether it would have been obvious that this particular combination would produce results superior to the additive effect of the components separately. It was not shown that because glyburide was an effective synergist, the different compound repaglinide would be expected to be an effective synergist.
The expert witnesses for both sides testified as to the uncertainties of predicting synergistic action. It was not shown that the prior art metformin-glyburide combination predicted the claimed invention, for the differences between glyburide and re-paglinide were well-recognized. Glyburide is one of the longer-acting sulfonylureas, a poor comparator for short-acting repagli-nide. A more reasonable analysis would consider metformin in combination with nateglinide, which is structurally similar to repaglinide, or a shorter-acting sulfonylu-rea such as glipizide.
My colleagues state that “the prior art taught that metformin could be combined with certain sulfonylureas which were, like repaglinide, short-acting secretagogues.” Maj. Op. at 1355. However, the record discussing these short-acting sulfonylurea combinations does not state that their synergy with metformin was known or existed. The “closest prior art” is the reference having the most “in common” with the claimed invention, not the reference that happens to describe the most impressive results. In re Merchant, 575 F.2d 865, 868-69 (CCPA 1978); see KSR, 550 U.S. at 421, 127 S.Ct. 1727 (“A factfinder should be aware, of course, of the distortion caused by hindsight bias and must be cautious of arguments reliant upon ex post reasoning.”).
The purpose of our patent system is a practical one: “To promote the Progress of Science and useful Arts.” U.S. Const, art. I, § 8, cl. 8. Consistent with this “constitutional command,” Graham v. John Deere Co., 383 U.S. 1, 6, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966), the Supreme Court, and this court, have recognized that the statutory requirement of non-obviousness is a “practical test of patentability,” id. at 17, 86 S.Ct. 684. Section 103 must be “followed realistically,” id., if the law is to support innovation as it is manifested in *1366the pragmatic world of technologic advance and commercial investment. See KSR, 550 U.S. at 415, 127 S.Ct. 1727 (summarizing the Court’s “functional approach” to obviousness); In re Kahn, 441 F.3d 977, 986 (Fed.Cir.2006) (in enacting § 103, Congress created “a more practical ... test for patentability”); Rosemount, Inc. v. Beckman Instruments, Inc., 727 F.2d 1540, 1546 (Fed.Cir.1984) (“the facts of real-world experience” inform the obviousness analysis); In re Lunsford, 53 C.C.P.A. 1011, 357 F.2d 385, 391-92 (1966) (“The provisions of section 103 must be followed realistically to develop the factual background against which the section 103 determination must be made.”).
The synergy demonstrated by Novo for the metformin-repaglinide combination therapy was not predicted or predictable, and was not obvious.
C. Claim Scope
The district court faulted the Australian study as too narrow to support claims that state that the metformin-repaglinide combination is administered “to a patient in need,” a broader class of diabetics than patients whose diabetes is poorly controlled on metformin monotherapy. The district court stated that “Novo presented no evidence that the claimed combination therapy produced unexpected or synergistic results in drug-naive patients.”3 The court concluded that the Australian clinical study and Dr. Sturis’ rat study were not probative of synergy “in ‘all instances.’ ” Novo, 775 F.Supp.2d at 1015-17.
Novo’s experimentation was objective and substantial. The Australian study was conducted with over eighty persons whose diabetes was poorly controlled on metfor-min alone, and thus were not “drug naive.” Diabetes is typically treated with a single drug first; if monotherapy does not work, or stops working after a period of time, combination therapy is prescribed. It was reasonable for Novo to test the metformin-repaglinide combination on patients who responded poorly to metformin monothera-py-
Dr. Sturis’ rat study corroborated the observed synergism. The district court acknowledged that the Zucker obese rats studied by Dr. Sturis “are an accepted animal model with excellent predictive capabilities for humans with Type II diabetes.” Novo, 775 F.Supp.2d at 1013. Novo “demonstrate^] that an embodiment has an unexpected result and provide[d] an adequate basis to support the conclusion that other embodiments falling within the claim will behave in the same manner, [which] will generally establish that the evidence is commensurate with [the] scope of the claims.” In re Kao, 639 F.3d 1057, 1068 (Fed.Cir.2011).
The Australian study tested, and demonstrated, the efficacy of the metformin-repaglinide combination in a patient population with difficult-to-control Type II diabetes, those who would most benefit from the combination. It was not necessary for Novo to prove synergy in patients for whom combination therapy is not needed, in order to claim administering the combination of metformin and repaglinide to “a patient in need.” See In re Chupp, 816 F.2d at 646 (to be patentable, a compound need not “produce superior results in every environment in which the compound may be used”); In re Kao, 639 F.3d at 1068 (a patentee is not obligated “to test every embodiment within the scope of the claims”).
From my colleagues’ erroneous view of the evidence and incorrect application of *1367the law of obviousness, I respectfully dissent.
. Sun is the corporate parent of codefendant Caraco Pharmaceutical Laboratories.
. Glyburide is also known as glibenclamide.
. In this context, "drug naive” refers to a person whose diabetes has not yet been treated pharmaceutically.