concurring in part, dissenting in part, and concurring in the result.
I agree with Judge Prost that the district court erred in its construction of “clinically significant electrolyte shifts” and that the correct construction of that term is “alterations in blood chemistry that are outside the normal upper or lower limits of their normal range or other untoward effects.” Op. of Prost, J., at 9. Although it seems to me that Novel established non-infringement as a matter of law under the correct “clinically significant electrolyte shifts” construction, I join the remand to the district court for further proceedings on that issue.1 I also agree that the asserted claims of the '149 patent *1361are not invalid. Thus I join parts I — II, 111(A)(2), III(B) (except the second paragraph) and III(C).
However, I respectfully dissent from the majority’s conclusion that Novel’s ANDA meets the volume limitation of the asserted claims. Under the proper interpretation of the volume limitation, Novel established non-infringement as a matter of law. In my view, the majority’s contrary conclusion is inconsistent with established authority under the Hatch-Waxman Act.
I
I agree with the majority that the district court’s construction of “purgation” was correct. But even under that construction, Novel’s ANDA does not meet the claimed volume limitation. The asserted claims include a volume limitation that requires that the solution be between 100— 500 mL, as seen in representative claim 15:
A composition for inducing purgation of the colon of a patient, the composition comprising from about 100 ml to about 500 ml of an aqueous hypertonic solution ... wherein the composition does not produce any clinically significant electrolyte shifts....
Ex Parte Reexamination Certificate No. 6,946, 149C1 col. 2 ll. 23-30 (emphasis added). The asserted claims also include claims 18-20, and 23. Claims 19-20 and 23 are method of use claims.
In ANDA litigation, because a generic company seeking FDA approval does not yet “make[], use[], offer[] to sell, or sell[ ]” the product and therefore cannot infringe under 35 U.S.C. § 271(a), pharmaceutical companies must rely on 35 U.S.C. § 271(e)(2)(A), which creates an “artificial” act of infringement when the generic company submits an ANDA seeking approval “ ‘for a drug claimed in a patent or the use of which is claimed in a patent.’” Eli Lilly & Co. v. Medtronic, Inc., 496 U.S. 661, 675, 676, 110 S.Ct. 2683, 110 L.Ed.2d 605 (1990) (quoting § 271(e)(2)(A)). The infringement occurs only if the generic company seeks approval for a patented composition or use that has been approved by the FDA. Bayer Schering Pharma AG v. Lupin, Ltd., 676 F.3d 1316, 1319 (Fed.Cir.2012) (citing Warner-Lambert Co. v. Apotex Corp., 316 F.3d 1348, 1356 (Fed.Cir.2003) (“‘[T]he use’ in § 271(e)(2)(A) refers to the use for which the FDA has granted an NDA.”)); Glaxo, Inc. v. Novopharm, Ltd., 110 F.3d 1562, 1567-68 (Fed.Cir.1997) (composition is determined by the composition in the ANDA).2
Under the Hatch-Waxman Act, the infringement inquiry “must focus on what the ANDA applicant will likely market if its application is approved.” Id. at 1569. Novel’s ANDA does not cover administration of the drug for purgation sufficient for “evacuation of a copious amount of stool from the bowels.” Claim Construction Order, at *6. Rather, Novel’s ANDA states specifically that the drug “is indicated for cleansing of the colon as a preparation for colonoscopy in adults.” J.A. 5280. The ANDA limits its dose regimen (the SU-PREP regimen) as follows: “The dose for colon cleansing requires administration of two bottles of [oral solution]. Each bottle is administered as 16 oz [i.e., 473 mL] of [oral solution] .... ” J.A. 5280 (emphases added). Therefore, the volume of the dose in Novel’s ANDA is 946 mL.3 Braintree’s *1362infringement argument, adopted by the district court and by the majority here, is the “one bottle” infringement theory, which asserts that because one (half-dose) bottle of Novel’s proposed generic falls within the 100-500 mL volume range and would cause a “purgation,” the ANDA infringes.
Here, “one bottle” consisting of 473 mL is not the dose approved by the FDA. The approved NDA for SUPREP does not permit a dose regimen of only one 16 oz bottle (473 mL), but requires both administrations of the “split-dose” regimen of two bottles (946 mL). Nor does the NDA recommend a one bottle dose to induce purgation as a possible alternative. Since no clinical studies have been conducted to support a one bottle regimen for purgation, the FDA has not approved such a use. See 21 U.S.C. § 355(d)(1). Because there is no underlying NDA approval for a one bottle regimen for purgation, Novel is precluded from seeking an ANDA for such a regimen.4 Instead, Novel’s ANDA properly seeks approval for a 946 mL, two bottle regimen to induce colon cleansing, based on Braintree’s approved NDA. Therefore, Novel’s ANDA does not infringe the '149 patent, which claims a composition with a volume of 100-500 mL for inducing purgation and a method of using the same.
The majority’s contrary interpretation is inconsistent with this court’s longstanding precedent. An ANDA cannot infringe an asserted patent when the FDA-approved dose is not the dose claimed in the patent. As we held in Allergan, Inc. v. Alcon Labs., Inc., 324 F.3d 1322, 1332 (Fed.Cir.2003), “a method of use patent holder may not sue an ANDA applicant for induced infringement of its patent, if the ANDA applicant is not seeking FDA approval for the use claimed in the patent and if the use claimed in the patent is not FDA-approved.” (emphasis added) (citing Warner-Lambert, 316 F.3d at 1354-55). While the patent here includes both composition and method of use claims, all claims are limited to the use of a specified volume of the composition.
In particular, the majority’s decision is contrary to our decision in Bayer Sobering, where we held that because “the FDA has not approved [the use claimed in the patent] ... the defendants cannot be held liable for infringement of the patent.” 676 F.3d at 1326. There, the patent claimed the use of a particular drug to achieve three effects simultaneously, “a contraceptive effect, an antiandrogenic effect, and an antialdosterone effect.” Id. at 1319 (quoting U.S. Patent No. 5,569,652). But the FDA-approved NDA and label indicated the use of the branded drug only “for oral contraception.” Id. (internal quotation marks omitted). We held that the accused ANDAs did not infringe because they did not seek approval for two of the claimed uses. Id. at 1326. There was no infringement under § 271(e)(2)(A) because “the defendants’ ANDAs [sought] approval to market the generic [drug] for contraceptive use, and there is no valid patent on *1363the use of the drug for that purpose alone.” Id.
Crucially, we found it irrelevant that the drug did, in fact, cause the three claimed effects simultaneously — simply because a patient could practice the claims by using the drug to achieve all three effects did not mean that the corresponding ANDA infringed. Id. at 1325-26. “[Nothing] on the [approved] label provide[d] any safety or efficacy information associated with the possible use of [the drug] in treating patients who are in need of those [claimed] effects.” Id. at 1322. As we noted there:
[T]he point is not simply that the method of use [claimed in the patent] was not described in the Indications and Usage section that shows lack of FDA approval; the point is that the label, taken in its entirety, fails to recommend or suggest to a physician that [the drug] is safe and effective for inducing the claimed combination of effects in patients in need thereof.
Id. at 1324. So too here — -just because a patient could ingest only one bottle of Novel’s generic to achieve a purgation does not make Novel’s ANDA infringing. It is undisputed that Novel’s proposed label fails to recommend or suggest to a physician that the solution should be taken to achieve purgation, rather than cleansing, or that it suggests taking a dosage below the 946 mL level for any purpose.
Similarly, in Warner-Lambert, 316 F.3d at 1356, we held that “because an ANDA may not seek approval for an unapproved or off-label use of a drug ... it necessarily follows that 35 U.S.C. 271(e)(2)(A) does not apply to a use patent claiming only-such [an unapproved or off-label] use.” See also AstraZeneca LP v. Apotex, Inc., 633 F.3d 1042, 1059-60 (Fed.Cir.2010) (assuming that an ANDA seeking approval for a twice-daily dose regimen could not infringe a patent claiming a once-daily dose regimen). In short, Braintree’s patent claims a composition and a method, both of which require using a 100-500 mL solution. Novel’s ANDA seeking FDA approval for a 946 mL solution cannot infringe the '149 patent.
II
Even if there were no Hatch-Waxman issue, Braintree’s theory would be unsupportable. Braintree’s “one bottle” theory rests on a demonstrably incorrect claim construction. Here, “volume” refers to the total volume of the recommended dosage and not some portion of that total volume — whether or not the recommended dosage is given in one bottle or two bottles.
The '149 patent specification describes the two categories of bowel cleansing options in the prior art to the '149 patent, each with its own disadvantages. The first were “large volume” isotonic solutions that did not cause electrolyte shifts in patients. '149 patent col. 1 11. 31, 48-53. The second were “small volume” hypertonic preparations that caused severe electrolyte shifts in many patients, often with life-threatening results. Id. col. 2 11. 41-44, col. 4 11. 23-26. The inventors of the '149 patent further recognized that although “[c]linical trials have shown use of the 4 liter [ie., large volume] [isotonic] solution to be safe and efficacious [ie., did not cause electrolyte shifts],” that patient “compliance [with ingesting such solutions] is poor because of the large volume of solution.” Id. col. 4 11. 1-5. The '149 patent attempted to address both issues by creating a small volume solution for patient ingestion that did not cause clinically significant electrolyte shifts. The volume limitation was critical to distinguish the '149 patent from the prior art. The specification explains that the prior art solutions were customarily ingested in multiple administrations, such that a solution with a total volume of 4 liters would be divided into administrations of 500 mL or less.5 The specification sug*1364gested the same method of dividing the total volume for ingesting the claimed solution:
Optimally, the effective dose may be divided and administered, to the patient in two, or more administrations over an appropriate time period. Generally, 2 doses administered of equal portions of the effective dose, separated by 6 to 24 hours produce satisfactory purgation.
Id. col. 5 ll. 19-24 (emphasis added). The specification thus makes clear that the total volume is the combination of the divided doses. Under the correct construction as envisioned by drafters of the '149 patent, “volume” referred to the total claimed 100-500 mL volume of solution ingested by the patient as the recommended dose to induce purgation, even if this total volume has been divided into smaller administrations of unspecified size. This ensured that the claimed volume limitation did not run afoul of the prior art solutions such as Hechter, which suggested dividing a “large volume” solution of 4 liters into individual administrations of about eight ounces or 237 milliliters. Hechter col. 4 ll. 2-12, col. 6 ll.14-35.
Indeed, Braintree’s own arguments that Hechter does not anticipate the '149 patent focus on the total volume of the disclosed Hechter solution and are inconsistent with its “one bottle” infringement theory. Hechter disclosed a 4 liter isotonic solution that would be created from a 1 liter hypertonic stock solution. Hechter col. 3 11. 55-62. Novel argued that because Hechter taught that only 40% of the 4 liter isotonic solution must be ingested to achieve colon-cleansing, it inherently disclosed a 400 mL hypertonic stock solution capable of colon-cleansing. Appellant’s Br. 47-49. But Braintree’s expert witness in this case disagreed, objecting that there could be no anticipation from consuming only a portion of the recommended dose. See J.A. 21624 (Peura direct testimony) at 962:7-13 (explaining that he disagreed with Novel’s inherent disclosure theory, because “[w]ell, a liter is a liter. You know, there may be two 500 milliliters components to a liter or four 250, or ten 100; there’s no teaching in Hechter that would suggest that ... this liter would ever be administered to anybody, and there’s surely no teaching that you can divide this liter into component parts.”). Braintree’s own argument that Hechter does not anticipate the claimed invention depends, in part, on an assumption that the relevant volume is the total amount of a disclosed solution, not the volume of individual administrations or separate components of that solution.
In short, the specification of the '149 patent and Braintree’s own expert made clear that the invention was directed at “small volume” solutions of 100-500 mL total — not large volume solutions that could be divided into 100-500 mL administrations. Braintree’s “one bottle” theory of infringement therefore leads to the surprising result that although large volume solutions of 4 liters are derogated in the specification, if administered in 500 mL increments, such solutions would infringe the '149 patent. Therefore, the entire premise of the '149 patent — that a recommended dosage of a small volume would improve patient compliance — is ignored by Braintree’s one bottle theory.
Because Novel’s ANDA does not meet the volume limitation of the asserted claims, under this court’s ANDA infringe*1365ment precedent and under the language of the specification, it cannot infringe the '149 patent. I respectfully dissent from the contrary conclusion reached by the majority.
. The sole evidence in the summary judgment record concerning the percentage of individuals experiencing such electrolyte shifts is data from two clinical studies of SUPREP, and those data are sufficient to show that the claim limitation was not met. The first, BLI800-202, a Phase II study of eighteen patients, showed that a majority of patients experienced clinically significant electrolyte shifts. J.A. 5059-69; 6642, 6683; 6756, 6759; 6831; 6912; 7000, 7002; 7094, 7100; 7386; Def.’s Br. in Supp. of Mot. for Summ. J. 8, Braintree Labs., Inc. v. Novel Labs., Inc., No. 3:11-cv-1341 (D.N.J. Oct 9, 2012), ECF No. 159-1. The second, BLI800-303, a Phase III study of sixty-three patients, showed that a significant minority or about one-third of patients experienced clinically significant electrolyte shifts. J.A. 9679 (expert report of Dr. Goldfarb (Novel’s expert witness) summarizing BLI800-303 data). Braintree's expert witness Dr. Peura also testified that "abnormal electrolyte values have been observed in *1361some patients taking SUPREP.” J.A. 10322 (emphasis removed).
. Section 271(e)(2)(A) reads as follows:
(2) It shall be an act of infringement to submit—
(A) an application under [21 U.S.C. § 3550] or described in [21 U.S.C. § 355(b)(2) ] for a drug claimed in a patent or the use of which is claimed in a patent.
35 U.S.C. § 271(e)(2)(A).
. See also J.A. 22172 (Braintree's Patent Term Extension Request) ("The SUPREP product ... contains only 2x16 ounces of solution (i.e., approximately 2 x 0.47 L *1362= .94L of solution), as indicated in the approved label_”); J.A. 4262 (Jack A. Di Palma et al, A Randomized Clinical Study Evaluating the Safety and Efficacy of a New, Reduced-Volume, Oral Sulfate Colon-Cleansing Preparation for Colonoscopy, 104 Am. J. Gastroenterology 2275, 2276 (2009)) ("The total preparation volume [of SUPREP] is 960 ml.”).
. A generic may not file an ANDA that differs from the approved NDA in "route of administration, dosage form, or strength” unless it has obtained special permission from the FDA to do so, 21 U.S.C. § 355(j)(2)(C), and Novel neither sought nor received such permission here. A generic may not seek approval for a use that differs from the approved NDA under any circumstances. See Warner-Lambert, 316 F.3d at 1356 (citing § 355(j)(2)(A)(i)).
. See, e.g., '149 patent col. 1 ll. 53-56 ("Because the [large] solutions are isotonic, pa*1364tients are required to ingest a significant amount of volume of these solutions, up to one eight ounce glass [i.e., 237 mL] every ten minutes for a total of one gallon of fluid [i.e., about 4 L] to achieve effective purging.” (emphasis added)); id. col. 2 11. 32-34 ("Patients are typically required to take two (2) three ounce doses of [Fleets Phospho-Soda] separated by a three to 12 hour interval for a total of six ounces (180 ml).”).