Opinion for the court filed by Circuit Judge PROST. Opinion concurring in part and dissenting in part filed by Circuit Judge NEWMAN.
PROST, Circuit Judge.Novo Nordisk A/S and Novo Nordisk Inc. (“Novo”) appeal a decision of the United States District Court for the Eastern District of Michigan which held that claim 4 of U.S. Patent No. 6,677,358 (“'358 patent”) was invalid as obvious and that the '358 patent was unenforceable due to inequitable conduct. See Novo Nordisk A/S v. Caraco Pharm. Labs., 775 F.Supp.2d 985, 1025 (E.D.Mich.2011). For the reasons set forth below, we affirm in part and reverse in part.
I. BACKGROUND
This case, now before us for a third time,1 concerns pharmaceuticals used in the treatment of non-insulin dependent diabetes mellitus (“NIDDM” or “Type II diabetes”), a disease where the body secretes insufficient levels of the hormone insulin, and/or the body is resistant to the effects of insulin. Id. at 997. Type II diabetes can be treated with orally administered antidiabetic drugs (“OADs”) in the *1349form of monotherapy (a single OAD) or combination therapy (more than one OAD). Id. Before the filing date for the '358 patent, there were several known classes of OADs having different chemical qualities and varying mechanisms of action. This appeal concerns two of these OAD classes: insulin secretagogues and insulin sensitizers.
Insulin secretagogues work by stimulating insulin release from pancreatic beta cells, and they fall into two subclasses: meglitinides and sulfonylureas. Id. at 997-98. As of the '358 patent’s filing date, there were five known meglitinides and fifteen known secretagogues, but only a handful of these were generally prescribed for treating Type II diabetes. Id. at 997-98, 1004-06. The drug repaglinide, which is a meglitinide, is the primary focus of this appeal.
Insulin sensitizers reduce insulin resistance by acting on the liver to reduce glucose production and thereby improving insulin sensitivity in muscle and fat tissues. Id. Of the known sensitizers in the relevant time frame, the most widely-used and successful was a drug called metformin. Id. at 1006.
This case involves a claim for treating Type II diabetes with combination therapy using repaglinide and metformin, specifically: “[a] method for treating non-insulin dependent diabetes mellitus (NIDDM) comprising administering to a patient in need of such treatment repaglinide in combination with metformin.” Id. at 989.
A
Novo, a large pharmaceutical manufacturer, began experimenting in 1990 with repaglinide’s efficacy in monotherapy for treating Type II diabetes. Id. at 998. It found repaglinide to be a rapid and short-acting insulin seeretagogue that was quickly eliminated from the body, findings which corresponded to what was known about the drug in the art. See id. at 1004-05. Shortly thereafter, a team of Novo investigators conducted a study on Australian patients to determine whether repaglinide might be more effective when administered in combination therapy with metformin (“Moses Study”).
In the Moses Study, patients failing on metformin alone were given repagli-nide/metformin combination therapy. Id. at 1010. One of Dr. Moses’s test parameters, HbAlc or glycosylated hemoglobin, measured the patient’s average glucose level in the recent past. Id. The repagli-nide/metformin combination reduced that level by 1.41%, or roughly twice what repa-glinide and metformin separately yielded in monotherapy. Id.
The Moses Study also measured fasting plasma glucose (“FPG”) levels, which is the glucose level after the patient has not eaten for about eight hours. Id. Although repaglinide was previously thought to have no effect upon FPG due to its short-acting tendencies, the Moses Study found that repaglinide/metformin reduced FPG to levels more than eight times lower than what was typically achieved by metformin alone. Id.
Armed with the results from its Moses Study, Novo filed a provisional patent application for the repaglinide/metformin combination on October 29, 1997. Id. at 999. The examiner rejected this initial application, reasoning that it was obvious to try combining repaglinide with metfor-min, and it was also predictable that the combination would yield, at a minimum, a benefit equal to the drugs taken separately, i.e., an “additive” effect.2 Id. at 999-1000.
*1350Novo responded by directing the examiner’s attention to Example 3 of its application, which contained the data from the Moses Study, and arguing that this study yielded synergistic results which no artisan would have expected. Id. at 1000. The examiner disagreed, and continued her rejection. Id. Novo’s third and fourth office action responses, each of which asked the examiner to reconsider her position on the Moses Study, were both rejected. Id.
Novo then filed a fifth response, this time presenting via declaration the results of an additional study conducted by Novo scientist Dr. Sturis (“Sturis Declaration”). Id. at 1000-01. Dr. Sturis’s study tested the effects of metformin and repaglinide on the glucose levels of “Zucker obese rats,” which are rats bred specifically for use in pharmaceutical studies as models for obesity, diabetes and heart disease. Id. at 1000 n. 10. Dr. Sturis divided twenty rats into four groups: the first group was given a placebo, the second was given only metformin, the third was given only repa-glinide, and the fourth was given the repa-glinide/metformin combination therapy. J.A. 17167. Dr. Sturis measured the blood glucose levels of these rats at 30, 60, and 120 minutes, and then calculated the combination’s “hypothetical additive effect” by adding the glucose reduction found in the metformin-only rats to that of the repagli-nide-only rats. J.A. 17168. Finally, he compared the “hypothetical additive effect” to the actual glucose reduction found in the repaglinide/metformin group to calculate the probability (as a “p-value”3) that any glucose reduction caused by the combination therapy might be attributable to synergistic rather than additive effects. Id.
Dr. Sturis reported his study’s results in two ways. First, he plotted a chart showing glucose levels across time and then calculated the “area under the curve” for each line, thus expressing the average glucose reduction found across the entire study for each group of rats. J.A. 17168. According to this calculation, the rats who received the repaglinide/metformin treatment experienced a greater average reduction in blood glucose levels than the other three groups, and the combination therapy also proved moré effective than the “hypothetical additive effect” of the two drugs. Id. The p-value for this finding was 0.061. Id. Second, Dr. Sturis isolated just the glucose measurements taken at the 120-minute mark, where he found the largest disparity between repaglinide/metformin and the other three groups. J.A. 18169. Using this data, he calculated a p-value of 0.02. Id.
Dr. Sturis opined that the 0.061 p-value “indicate[d]” that repaglinide/metformin had a synergistic effect upon blood glucose levels over the entire two-hour span, and that the 0.02 p-value at the 120-minute mark in particular demonstrated “significant synergy.” J.A. 17168-69. He thus concluded that repaglinide/metformin combination therapy had synergistic effects in Zucker obese rats, and that his study together with the Moses Study “strongly suggested] that the combination of repa-glinide and metformin has synergistic properties in type 2 diabetic [human] patients.” Novo Nordisk, 775 F.Supp.2d at 1019.
Novo submitted the Sturis Declaration to the examiner, along with assertions from its counsel Dr. Richard Bork that the *1351declaration “provide[d] clear evidence of synergy ... in the treatment of type II diabetes,” and that any prima facie case”of obviousness was “rebutted by the evidence of synergistic and surprising results achieved by the claimed combined therapy in humans.” Id. at 1001.
The examiner withdrew her rejection, explaining that her decision was “[biased solely upon the Declaration submitted by Dr. Sturis and reconsideration of the synergistic effects demonstrated in Example 3.” Id. at 1001. After additional proceedings not relevant here, the '358 patent issued on January 13, 2004. Id. Claim 4 of the patent, which is the sole claim at issue here, was not amended at any point during prosecution. Id.
B
The present litigation arose under the Drug Price Competition and Patent Term Restoration Act of 1984, Pub.L. No. 98-417, 98 Stat. 1585 (1984) (codified at 21 U.S.C. §§ 355, 360cc; 35 U.S.C. §§ 156, 271), as amended by the Medicare Prescription Drug Improvement and Modernization Act of 2003, Pub.L. No. 108-173, 117 Stat.2066 (2003) (collectively, “Hatch-Waxman Act”). Specifically, this case stems from an Abbreviated New Drug Application that Caraco filed in 2005 requesting approval from the Food and Drug Administration to sell a generic version of repaglinide, which the Orange Book associated with Novo’s '358 patent.4 Caraco certified in the filing that the '358 patent was invalid or would not be infringed by the sales of the generic repaglinide, and Novo responded with a patent infringement lawsuit claiming that Caraco infringed claim 4 of the '358 patent. Caraco counterclaimed asserting, inter alia, obviousness and unenforceability. Novo Nordisk, 775 F.Supp.2d at 989.
Following a bench trial, the district court held that claim 4 of the '358 patent was invalid because of obviousness and that the patent was not enforceable because of inequitable conduct. Novo appeals these rulings, and we have jurisdiction under 28 U.S.C. §§ 1292(c)(2) and 1295(a).
II. Obviousness
We turn first to the district court’s obviousness ruling. The district court found, and the parties do not dispute, that Caraco set forth a prima facie case that it was obvious to try combination therapy using metformin and repaglinide to treat Type II diabetes. Novo Nordisk, 775 F.Supp.2d at 1007. 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 therapy after demonstrating effectiveness in monotherapy. Id. at 1002. Combination therapy using insulin sensitizers and insulin secreta-gogues was common at the time, and met-formin was the most widely-used insulin sensitizer as of the '358 patent’s filing date. Id.
Having thus established the existence of a prima facie case, but “[bjefore reaching the ultimate conclusion on the issue of obviousness,” the district court undertook a thorough analysis of “the evidence and assertions of unexpected and surprising results, as well as synergistic results.” Id. at 1007. The key question was whether repaglinide/metformin proved more effective than what would have been expected in view of the prior art, i.e., whether the *1352combination yielded an unexpected synergistic effect.
To undermine the examiner’s finding that the Moses and Sturis studies had demonstrated synergy, Caraco introduced new prior art and evidence which the examiner had never considered, such as testimony from expert witnesses and Novo scientists. Id. at 1009. After reviewing Caraco’s evidence, the district court determined that an artisan would have expected repaglinide/metformin would yield some synergy. Id. at 1010. The court then considered, and rejected, the premise that Novo’s studies had yielded an unexpected or superior level of synergy, finding instead that Novo’s results were entirely expected in view of the state of the art at that time. Id. at 1017. The court thus concluded that Caraco had shown, by clear and convincing evidence, that claim 4 of the patent was invalid as obvious. Id. at 1018.
Novo attacks the district court’s obviousness ruling on three grounds. First, it asserts that the district court misallocated the burden of persuasion in this case by forcing Novo to “overcome” Caraco’s “pri-ma facie” case of obviousness with evidence of unexpected results. Second, Novo argues that even if the burdens were properly allocated in this case, Caraco’s evidence insufficiently supported the court’s ultimate obviousness findings. Finally, Novo believes that the district court should have deferred to the examiner’s original finding that the Sturis and Moses studies demonstrated unexpected synergy.
For the reasons set forth below, we disagree with all three of Novo’s arguments, and we therefore affirm the district court’s conclusion that claim 4 of the '358 patent is invalid as obvious.
A
As its first ground for reversal, Novo contends that the district court mis-allocated the burden of persuasion during its obviousness analysis. It is black-letter law that a patent is presumed valid under 35 U.S.C. § 282, that a party challenging its validity bears the burden of proving the factual elements of invalidity by clear and convincing evidence, and that “because the presumption of validity remains intact ... throughout the litigation,” the burden of persuasion never shifts to the patentee during the course of a district court obviousness challenge. Pfizer v. Apotex, 480 F.3d 1348, 1359-60 (Fed.Cir.2007).
In this case, the district court recited the correct legal standard for obviousness, stating that:
[o]nce the challenger has presented a prima facie case of invalidity, the patent owner has the burden of going forward with rebuttal evidence. This requirement “does not in substance shift the burden of persuasion, because the presumption of validity remains intact and the ultimate burden of proving invalidity remains with the challenger throughout the litigation.”
Novo Nordisk, 775 F.Supp.2d at 992 (emphasis added) (quoting Pfizer, 480 F.3d at 1359-60).
Novo concedes that the district court correctly summarized the law of obviousness, but insists that the court misapplied this law in practice. Novo focuses upon language the court uses throughout its opinion, for instance, the court’s ultimate obviousness conclusion that “[Caraco’s] strong prima facie case of obviousness has not been overcome by Novo’s attempt to prove unexpected results and commercial success.” Id. at 1018 (emphases added). Novo cites this language as evidence that the court adopted an erroneous burden-shifting approach, similar to the one we recently rejected in In re Cyclobenzaprine Hydrochloride Extendedr-Release Capsule *1353Patent Litig., 676 F.3d 1063 (Fed.Cir.2012).
In Cyclobenzaprine, we reversed and vacated a district court decision where the court reached its ultimate conclusion on obviousness based solely upon the prima facie evidence. Id. at 1075 (the prematurity of the district court’s obviousness conclusion was apparent because “[i]t was not until after the district court found the asserted claims obvious that it proceeded to analyze the objective considerations”). In so doing, we reaffirmed our longstanding precedent that it is error to find a claim obvious “before ... considering] the objective considerations,” or to shift the burden of persuasion to the patentee at any point during its obviousness analysis. Id. at 1075.
Novo argues that just as in Cycloben-zaprine, the district court here reached its ultimate obviousness conclusion based solely upon Caraco’s prima facie evidence, then shifted the burden of persuasion onto Novo. It believes that this must have been so, because the court considered whether Novo had “overcome” Caraco’s evidence by “attempting] to prove unexpected results.” Novo Nordisk, 775 F.Supp.2d at 1018.
Novo misinterprets the role of the burden of persuasion in patent litigation. As noted above, the burden of persuasion remains with the challenger during litigation because every issued patent is entitled to a presumption of validity. However, the presumption of validity does not relieve the patentee of any responsibility to set forth evidence in opposition to a challenger’s prima facie case which, if left un-rebutted, would be sufficient to establish obviousness. Rather, the presumption of validity conveys two distinct advantages upon a patentee in the litigation context.
The patentee’s first advantage is a procedural one — he is required to come forward with evidence of nonobviousness only after the challenger has successfully made his prima facie case demonstrating that the patent might be obvious. See, e.g., Pfizer, 480 F.3d at 1360 (“[0]nce a challenger has presented a prima facie case of invalidity, the patentee has the burden of going forward with rebuttal evidence.”); Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1534 (Fed.Cir.1983). This benefit relates to the burden of production, which initially lies with the challenger, then shifts to the patentee during the course of the litigation.
The patentee’s second advantage is a substantive one — he prevails on the issue of validity unless the challenger proves to the decisionmaker by a clear and convincing standard that, after all of the evidence has been placed on the table for consideration, the claim is invalid. See, e.g., Pfizer, 480 F.3d at 1360; Stratoflex, 713 F.2d at 1534. This benefit relates to the burden of persuasion, and as we have often held (most recently in Cyclobenzaprine), this burden never shifts during the course of the litigation.
In this case, the court found that Caraco’s prima facie evidence, if unrebut-ted, would be sufficient to establish that the repaglinide/metformin combination was obvious to try, and that a person of ordinary skill in the art would have reasonably expected the combination would yield success in the form of beneficial, and even synergistic, results. Novo Nordisk, 775 F.Supp.2d at 1006-07, 1010. Having so found, it was entirely appropriate for the court to next consider whether Novo’s countervailing secondary consideration evidence of unexpected synergy (i.e., its “attempt to prove unexpected results”) was sufficient to “overcome” Caraco’s prima facie case. The mere fact that the court conducted this analysis using terms such as “overcome” and “prima facie” does not necessarily imply that it shifted the burden *1354of persuasion onto Novo. See KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 426, 127 S.Ct. 1727, 167 L.Ed.2d 705 (2007) (“Like the District Court, finally, we conclude Te-leflex has shown no secondary factors to dislodge the determination that claim 4 is obvious.”) (emphasis added) (affirming Teleflex Inc. v. KSR Int’l Co., 298 F.Supp.2d 581, 596 (E.D.Mich.2003) (“[T]he Court finds the evidence of commercial success insufficient to overcome Defendant’s clear and convincing evidence of obviousness.”) (emphasis added)); Wm. Wrigley Jr. Co. v. Cadbury Adams USA LLC, 683 F.3d 1356, 1365 n. 5 (Fed.Cir.2012) (“[U]se of the terms ‘prima facie’ and ‘rebuttal’ in addressing an invalidity challenge does not constitute reversible error as long as the court ‘consider[s] all evidence of obviousness and nonobviousness before reaching a determination’ and does not shift the burden from the patent challenger.”) (citing In re Cyclobenzaprine, 676 F.3d at 1077). Rather, as long as the court reserved its ultimate conclusion on validity until after it considered the evidence from both sides, this language simply reflects the court’s shift of the burden of production once the court determined that the challenger has established a prima facie case of obviousness.
Nothing in the court’s opinion in this case indicates that it reached a premature conclusion on obviousness. To the contrary, after considering the prima facie evidence but “fbjefore reaching the ultimate conclusion on the issue of obviousness,” the court thoroughly evaluated all evidence of unexpected synergy and commercial success. See Novo Nordisk, 775 F.Supp.2d at 1007 (emphasis added); id. at 1007-1017. It then concluded that, in view of all of this evidence, Caraco had shown by clear and convincing evidence that the combination was obvious. Id. at 1018. Therefore, the court’s analysis was entirely appropriate under Cyclobenzaprine and the rest of our obviousness law.
Furthermore, and in any event, the district court did not invalidate claim 4 due to Novo’s failure to prove unexpected results, as Novo alleges. Instead, it found that Caraco had established by “[c]lear and convincing evidence ... that the results of the claimed combination therapy said by Novo to be unexpected and unexplainable were, to the contrary, expected and explainable in light of the state of the art as of the critical date.” Id. (emphasis added). The nature of this finding further undercuts Novo’s claim that the burden of persuasion shifted during the course of the opinion.
We therefore reject Novo’s contention that the district court misallocated the burden of persuasion, and decline to reverse on this basis.
B
Novo next argues that even if the district court correctly allocated the burden of persuasion for obviousness, the record evidence did not support a conclusion of expected results. On appeal from a bench trial on obviousness, we review de novo the court’s the ultimate legal conclusion of whether a claimed invention would have been obvious, and review the underlying findings of fact for clear error. See Golden Blount, Inc. v. Robert H. Peterson Co., 365 F.3d 1054, 1058 (Fed.Cir.2004).
At trial, Caraco contended that an artisan seeking to predict the performance of repaglinide combined with metformin would have considered metformin’s history in combination therapy with other insulin secretagogues, particularly those in the sulfonylurea class. Caraco presented evidence, including some prior art and testimony never before the examiner, which indicated that earlier metformin/sulfonylu-rea combinations were generally understood to yield synergy. Based upon this evidence, Caraco argued that artisans *1355would have expected repaglinide to be likewise synergistic when combined with met-formin.
Novo countered that expectations for the repaglinide/metformin combination would have been instead based primarily, if not exclusively, upon repaglinide’s known efficacy in monotherapy. Novo’s evidence indicated that repaglinide was known to be a short-acting insulin secreta-gogue, different from the longer-acting sulfonylureas in Caraco’s prior art. Novo specifically relied upon a study by Wolffen-buttel which showed that repaglinide in monotherapy had no impact upon patient FPG. Based upon Wolffenbuttel, Novo insisted that an artisan would have been very surprised when repaglinide/metfor-min proved to be eight times more effective in reducing FPG levels than metfor-min alone.
The district court agreed with Caraco, applying the following three-step reasoning:
(1) the closest prior art [to the repagli-nide/metformin combination] was combination therapy using metformin and a sulfonylurea;
(2) combination therapy using metfor-min and one of the sulfonylurea class of secretagogues was well known in the art to produce beneficial and even synergistic results in controlling glucose levels in Type II diabetes patients; [and]
(3) repaglinide was known as an insulin seeretagogue having a similar mechanism of action to the sulfonylurea class of secretagogues.
Novo Nordisk, 775 F.Supp.2d at 1010. We see no clear error in these findings. Repa-glinide and sulfonylureas are both insulin secretagogues, and they therefore have a “similar mechanism of action” in that they both treat diabetes by stimulating the pancreas to release insulin. Id. at 997. Met-formin is an insulin sensitizer, which treats diabetes patients using a different mechanism, i.e., by reducing their resistance to insulin. Id. Furthermore, the prior art taught that metformin could be combined with certain sulfonylureas which were, like repaglinide, short-acting secretagogues. Id. at 1005. It is reasonable that an artisan seeking to combine a known insulin sensitizer (like metformin) with a new insulin seeretagogue (like repaglinide) would base his expectations upon prior art sensi-tizer/secretagogue combinations.
In view of these findings, it was not erroneous for the court to conclude that the prior art'predicted the results found in the Moses Study.’ For example, the' near-term and long-term benefits which Dr. Moses observed in his repagli-nide/metformin study were generally inferior to the results found by prior art studies involving metformin combined with sulfonylureas. See id. at 1011-12. Novo argues that these studies tell only half the tale, because they fail to account for the differences between sulfonylureas and re-paglinide in monotherapy. But other trial evidence also supports the district court’s conclusion — for instance, Dr. Sturis testified that he originally declined to sign a declaration supporting Novo’s application because he felt that Dr. Moses had not mathematically or scientifically proven the existence of synergy. See id. at 1019.
Dr. Sturis did, of course, eventually submit a declaration on Novo’s behalf. But he only did so after conducting his own study, and even then, he went only so far as to state that the evidence “strongly suggested]” synergy. Id. It was not clearly erroneous for the district court to find that Dr. Sturis’s results were expected, given that his conclusions mirror the conclusions drawn in Caraco’s new sulfonylurea prior art, which also suggested the existence of synergy.. Id. at 1009 (citing prior art reports that metformin/sulfonylurea combinations yielded an “apparent synergistic *1356effect” and “appear[ed] to have a synergistic effect”).
The only other study that supposedly demonstrated unexpected results was a study conducted by Pfeiffer, which compared the insulin sensitivity of eleven patients taking only metformin with that of the same patients after taking the repagli-nide/metfoi'min combination. Id. at 1014. Novo argued that no ordinary artisan would have expected the results that Pfeif-fer observed, namely, a 35% improvement in insulin sensitivity in combination therapy over what was seen when the patients took metformin alone. Id. But Caraco’s expert questioned that study’s reliability due to its small sample size, and also pointed out that Pfeiffer had himself explained away his results in a contemporaneous report as predictable in view of the prior art. Id.
The court further noted that certain tests conducted upon different classes of patients yielded results that contradicted those found by Moses, Sturis, and Pfeiffer. For instance, in one test where half of the patients were “drug-naive” (i.e., they had never before used any OADs), “the synergistic effect of combination therapy observed by Moses et al[.] was not consistently seen.” Id. at 1016. In another study, where all of the patients involved were drug-naive, the combination therapy did not show statistically better results than the drugs used in monotherapy. Id.
In view of all of these findings, few of which are challenged by Novo as clearly erroneous,5 Caraco proved by clear and convincing evidence that an artisan would have expected the level of synergy Novo found when it combined metformin and repaglinide. We therefore decline to reverse the district court’s obviousness determination on this basis.
C
Lastly, Novo contends that the district court should have deferred to the examiner’s finding that the Moses and Sturis studies demonstrated synergy. Novo’s theory cites the recent case Kappos v. Hyatt, where the Supreme Court held that, in cases involving district court review of U.S. Patent and Trademark Office (“PTO”) rejections under 35 U.S.C. § 145, new evidence may be considered and that “it makes little sense for the district court to apply a deferential standard of review to PTO factual findings that are contradicted by the new evidence.” Kappos v. Hyatt, — U.S.-, 132 S.Ct. 1690, 1696, 182 L.Ed.2d 704 (2012). Novo inverts this statement, and argues that if evidence presented at trial is not new evidence then the district court must defer to the findings of the examiner. Because Novo believes that all of Caraco’s prior art was merely cumulative of what was already before the examiner, it concludes that de novo fact-finding was not justified in this case.
Hyatt has no relevance here. Hyatt concerned 35 U.S.C. § 145, which provides for optional review in the Eastern District *1357of Virginia of decisions from the PTO rejecting patent applications in the first instance. 35 U.S.C. § 145; Hyatt, 132 S.Ct. at 1694. The present case is a district court challenge to an issued patent brought under the Hatch-Waxman Act, not a challenge to a PTO rejection brought under § 145, and Hyatt is therefore irrelevant. But in any event, in cases such as this we do not review the PTÓ’s decision. The- initial determinations by the PTO in determining to grant the application are entitled to no deference as they would be in an appeal to this court under 28 U.S.C. § 1295(4)(A) or (absent new evidence) in a district court proceeding under 35 U.S.C. § 145. Rather, we treat the issued patent as having a presumption of validity that must be overcome by clear and convincing evidence. No decision of the Supreme-Court or this court has ever suggested that there is an added burden to overcome PTO findings in district court infringement proceedings, and we reject Novo’s contrary assertion. Neither are we persuaded that the presence or absence of PTO findings on particular issues affects the basic presumption of validity.
III. Inequitable Conduct
We next address the district court’s determination that the '358 patent was unenforceable due to inequitable conduct. We review the district court’s ultimate finding of inequitable conduct for abuse of discretion, and review the underlying findings of materiality and intent for clear error. Therasense, Inc. v. Becton, Dickinson & Co., 649 F.3d 1276, 1291 (Fed.Cir.2011) (en banc); In re Omeprazole Patent Litig., 483 F.3d 1364, 1374-76 (Fed.Cir.2007).
At trial, Caraco alleged that the Sturis Declaration, as well as the representations made by Dr. Bork during prosecution of the '358 patent, constituted inequitable conduct. In particular, Caraco challenged: (a) Dr. Sturis’s omission of certain opinions regarding his own study and the Moses Study, as well as his failure to tell the PTO that certain of his reported results had not been part of his original test protocol; and (b) Dr. Bork’s assertion that Dr. Sturis’s data provided “clear evidence of synergy,” and his failure to disclose certain e-mails that allegedly refuted this statement. Novo Nordisk, 775 F.Supp.2d at 1019, 1021. These actions were particularly troubling, Caraco contended, because the examiner withdrew her rejection “[bjased solely upon the Declaration submitted by Dr. Sturis and reconsideration of the synergistic effects.” Id. at 1001.
The district court issued its opinion after we agreed to hear Therasense en banc, but before we reached our decision in that case. Therefore, it applied both the pre-Therasense and post-Therasense tests for materiality and intent, and found that under either standard, Sturis and Bork had intentionally withheld material information from the PTO. See id. at 995, 1021-22. Accordingly, the court concluded that the '358 patent was unenforceable due to inequitable conduct. Id. at 1024.
Novo argues that the district court clearly erred in finding that Sturis’s and Bork’s representations and omissions were material and intentional under Therasense. We agree with Novo on the issue of materiality, and so for the reasons outlined below, we reverse the district court’s conclusion on inequitable conduct.
A
Caraco’s inequitable conduct case against Dr. Sturis was based upon his trial testimony, wherein he conceded certain facts that were never submitted to the PTO. For instance, Dr. Sturis testified that when he first conducted his study, he had planned to calculate only one p-value based upon the “area under the curve” data, which resulted in a p-value of 0.061. But *1358after conducting his test, he decided to calculate and submit the second p-value (i.e., the 0.02 p-value) using the isolated data from the 120-minute interval. See id. at .1019-20. Dr. Sturis’s Declaration did not indicate to the PTO that his original test protocol called for calculating only the less favorable of the two p-values he ultimately presented.
Dr. Sturis also testified that his rat study, having been conducted on animals, could not alone establish that, the combination had a synergistic effect on humans. Id. at 1021. He further told the court that he held reservations, both before and after submitting his Declaration, about whether the Moses Study alone, or even taken together with the results of his rat study, could affirmatively prove synergy in humans. Id. at 1020-21.
Based upon this testimony, the district court found that Dr. Sturis had omitted material information from his Declaration, namely:
the facts that the two-hour data point was not part of the test protocol, and that a correction factor had not been applied to that p-value. That data point was the only one in his rat test that appeared to produce a statistically significant p-value of less than 0.05. Also undisclosed were Sturis’ opinions ... that, by his own standards requiring mathematical proof, neither his rat study nor the Moses Study alone proved synergy in humans.
Id. at 1020. The omissions were material under the pre-Therasense standard, the court found, “because they refuted or were inconsistent with the opinions expressed in his Declaration in support of patentability” and because “[a] reasonable examiner, fo-' cused on the issue of synergism as was the examiner here, would have wanted to consider any qualifications or reservations held by Sturis concerning the conclusions he expressed in his Declaration.” Id. at 1021. The court further found that “the examiner’s explicit reliance on the Sturis Declaration warrants the conclusion that the Declaration satisfied the [post-27mrosense] ‘but for’ materiality test.” Id.
We reject the district court’s materiality finding as clearly erroneous, because we fail to see how Dr. Sturis’s omissions qualify as “but for” material. For instance, any reasonable examiner would have understood that Dr. Sturis’s rat study was conducted on ánimals, and therefore could not definitively prove synergy in humans. Moreover, Dr. Sturis’s declaration, which stated only that his results “indicated” and “strongly suggest[ed]” synergy, was generally consistent with his trial testimony that synergy was not affirmatively proven and that “neither his rat study nor the Moses Study alone proved synergy in humans.” Id. at 1020.
While Dr. Sturis’s decision to omit his original test protocol from the Declaration is slightly more troubling, it similarly fails the “but for” materiality test. This is not a case where a declarant hid adverse test results from the PTO in favor of more promising data selected post hoc. Here, Dr. Sturis disclosed the results of his original protocol to the examiner, allowing her the opportunity to weigh the significance of the different p-values he calculated. Nor is this a case where the declarant’s omission expressly undermined his stated opinion. To the contrary, even after taking the omitted test protocol into account, the court specifically found that Dr. Stu-ris’s conclusions on synergy had not been shown to be false. See id. (“[Dr. Sturis’s ‘strongly suggests’ qualified conclusion] has not been shown by clear and convincing evidence to be .false.”)
Instead, Dr. Sturis stands accused of inequitable conduct because he failed to notify the PTO that the 0.02 p-value calculated at 120 minutes, while probative of *1359synergy, was not called for in his original test plan. Although this information ideally would have been disclosed to the PTO, it is nevertheless a non-material omission because it can “be rendered irrelevant in light of subsequent argument or explanation by the patentee.” See Therasense, 649 F.3d at 1294. For instance, at trial, Dr. Sturis justified his deviation from protocol by testifying that he saw surprisingly high levels of glucose reduction at 120 minutes, and he felt it would have been scientifically irresponsible for him not to investigate and report those findings. In view of this reasonable explanation, and the fact that he disclosed the results of his original test protocol to the PTO, we do not believe that his omitted test protocol was “but for” material.
B
Caraco’s inequitable conduct case against Dr. Bork focused primarily upon the following statements he made in support of the Sturis Declaration:
the data presented in the Declaration of Dr. Sturis, provides dear evidence of synergy for the use of the claimed combination of repaglinide and metformin in the treatment of type II diabetes.
... prima facie case [of obviousness] is rebutted by the evidence of synergistic and surprising results achieved by the claimed combination therapy in humans (Example application) and in Zucker obese rats (Sturis’ Declaration).
Novo Nordisk, 775 F.Supp.2d at 1021 (alteration in original). The district court found that this statement was material because it went beyond Sturis’s “strongly suggests” language by indicating that “clear evidence” of “synergistic and surprising results” in humans had been “achieved.”
Caraco also alleged that Dr. Bork should have provided the PTO with an e-mail he received from Dr. Sturis after making the above representations, wherein Dr. Sturis stated that “[t] he presence of greater-than-additive effects may be of relevance to the clinical efficacy of the [repagli-nide/metformin] combination.” Id. at 1022 (emphasis added). The court found that Bork should have understood that the word “may” contradicted his prior representation that “clear evidence” of synergy in humans had been “achieved,” and that this triggered his duty to disclose the email.
The court deemed Bork’s representations material under the pre-Therasense standard because his unqualified statements contradicted with what he knew about the Sturis Declaration, i.e., that it did not definitively prove synergy. Id. The district court addressed the post-Therasense test only briefly, stating that “[a]s in the case of the Sturis Declaration, [Dr. Bork’s representations] also satisfy the alternative ‘but for’ materiality test.” Id.
As with Dr. Sturis’s omissions, we believe that the statements and omissions by Dr. Bork are troubling, but not material. Dr. Bork’s characterization of the Sturis Declaration employed carefully-chosen language which tracked the qualified nature of Dr. Sturis’s opinions. For instance, whereas Dr. Sturis said his results “indicated” and “strongly suggested]” synergy, Dr. Bork referred to Sturis’s test results as “evidence” rather than “proof’ of synergy. These statements are also generally consistent with Dr. Sturis’s e-mail, and its use of the word “may.”
We therefore reverse the district court’s materiality and inequitable conduct findings as to both Dr. Sturis and Dr. Bork. We need not reach the issue of intent.
IV. ConClusion
For the reasons set forth above, we affirm the district court’s determination *1360that claim 4 of the '358 patent was invalid as obvious, but reverse the district court’s determination that the '358 patent was unenforceable due to inequitable conduct.
AFFIRMED IN PART AND REVERSED IN PART
. See Novo Nordisk A/S v. Caraco Pharm. Labs., Ltd., 601 F.3d 1359 (Fed.Cir.2010), rev’d. - U.S. -. 132 S.Ct. 1670, 182 L.Ed.2d 678 (2012), remanded to, 688 F.3d 766 (Fed.Cir.2012).
. Combination therapy yields an "additive” effect when the total effect of the combination equals the sum of the effect of each drug taken separately, and a "synergistic” effect *1350when the combination's effect exceeds the sum of the separately administered effects. Novo Nordisk, 775 F.Supp.2d at 998.
. The p-value is a value that statisticians use to show the level of uncertainty in a study’s results. Novo Nordisk, 775 F.Supp.2d at 1018 n. 20. A p-value is “statistically significant” if it is 0.05 or less, which indicates that there is 5% or less likelihood that the outcome was the result of pure chance. Id.
. For more details about the pharmaceutical applications that inspired this case, or about the history, language, and scope of the Hatch-Waxman Act, see Caraco, 132 S.Ct. at 1675-80.
. Novo does challenge as clearly erroneous the district court's finding that repagli-nide/metformin combination therapy had not been commercially successful, a conclusion the court reached based upon evidence and testimony that doctors seldom prescribe the combination to treat Type II diabetes. Novo Nordisk, 775 F.Supp.2d at 1017. Novo argues that the repaglinide/metformin combination was much more commercially successful than repaglinide alone, and that most repaglinide sales today are for use in combination therapy with metformin. However, the most probative evidence of commercial success is not overall sales, but whether those sales represent "a substantial quantity in th[e] market.” In re Applied Materials, Inc., 692 F.3d 1289, 1300 (Fed.Cir.2012) (citing In re Huang, 100 F.3d 135, 140 (Fed.Cir.1996)). Thus, the district court’s finding was not clearly erroneous.