“The outcome of this issue may yet prove controversial, as both the district court and the Federal Circuit arguably neglected established precedent regarding the doctrine of inherency in ruling that the claimed crystalline polymorph invention was obvious. . . .”
In Salix Pharmaceuticals, Ltd., Salix Pharmaceuticals, Inc., Bausch Health Ireland Ltd., Alfasigma S.P.A. v. Norwich Pharmaceuticals Inc., Nos. 2022-2153, 2023-1952 (Fed. Cir. April 11, 2024) (“Decision”), the Federal Circuit decided the parties’ cross-appeals arising from a bench trial in patent litigation brought by plaintiffs (collectively “Salix”) to enjoin approval of Norwich’s Abbreviated New Drug Application (“ANDA”) to sell a generic version of Salix’s branded drug Xifaxan.
While the original patents covering the branded drug’s active ingredient, rifaximin, have expired, this litigation involved later-issued Salix patents claiming a crystalline polymorph of rifaximin and methods of using rifaximin to treat irritable bowel syndrome with diarrhea (IBS-D) and hepatic encephalopathy (HE). Salix appealed the district court’s judgment invalidating as obvious the asserted patent claims directed to the treatment of IBS-D and to a specific crystalline polymorph of rifaximin. Concurrently, Norwich cross-appealed the district court’s determination that its judgment should not be modified to permit the non-infringing aspects of Norwich’s ANDA to be approved immediately by the FDA. The Federal Circuit affirmed the district court’s judgment in its entirety.
The appeal of the district court’s obviousness ruling regarding the IBS-D patents attracted two amicus briefs cautioning against the use of a patentee’s published clinical trial protocols as invalidating prior art and also resulted in one panel member dissenting from the majority opinion on the issue. The Federal Circuit’s affirmance of the obviousness of the polymorph patent claims was an unexpected departure from prior precedent that recognized the widely accepted unpredictability of the underlying chemistry and neglected the court’s strict prior precedent regarding the doctrine of inherency in analyzing obviousness under Section 103 when a patent claim element is not expressly disclosed in the prior art. Finally, the district court’s rulings on the HE patents presented the Federal Circuit with a question of first impression regarding the scope of the remedy required under 35 U.S.C. § 271(e)(4)(A)—whether any part of an ANDA that does not infringe a valid patent claim is immediately approvable by the FDA.
The IBS-D Patents
Salix’s IBS-D patents, U.S. Patent Nos. 8,309,569 and 10,765,667, disclose a method of treating IBS-D, that involves administering 550 mg of rifaximin three times daily, for a total of 1650 mg/day, for a 14-day treatment period. The primary prior art references considered by the district court included a Phase II clinical trial protocol posted by co-plaintiff Bausch on a government website pursuant to regulatory requirements and a journal article (“the Pimentel reference”) published in 2006.
The clinical trial protocol proposed a study of twice-daily doses of 550 mg (1100 mg/day) and 1100 mg (2200 mg/day) for durations of 14 and 28 days for the treatment of IBS-D but did not explicitly disclose the three-times-daily regimen of 550 mg (1650 mg/day) claimed in the Salix patents. Moreover, the Pimentel reference discussed the administration of 400 mg three times daily (1200 mg/day), while suggesting that “‘the optimal dosage of rifaximin may, in fact, be higher than that used in our study.’” Decision at 6.
Salix did not appear to dispute the district court’s conclusion that when combined, the two references disclosed all elements of its patents’ claims. Decision at 7. Instead, Salix contended that even if the 1650 mg/day dosage could be derived from the combined references, there was “insufficient evidence to support a finding of a reasonable expectation of success in using that particular dosage amount.” Decision at 7. Salix argued that the clinical trial protocol merely presented a plan for further study and therefore did not provide efficacy data or other information upon which to base a reasonable expectation of success. Id.
In response, the Federal Circuit noted that its prior decision in OSI Pharms., LLC v. Apotex Inc., 939 F.3d 1375, 1385 (Fed. Cir. 2019) permits a finding of reasonable expectation of success even in the absence of efficacy data. Id. at 7. However, the Federal Circuit expressed its hesitancy “to conclude as a general matter that the disclosure of a Phase II clinical trial plan, standing alone, provides an expectation of success sufficient to render obvious a dosage that was not included within the planned clinical trial.” Id. at 7-8. The court’s expressed hesitancy was likely inspired by amicus briefs cautioning against relying on a patentee’s published clinical trial protocols as invalidating references, especially where publication of the protocols was mandated by government regulations. See Appeal No. 22-2153 at Dkt. Nos. 43 & 44 (amicus briefs submitted by Vanda Pharmaceuticals Inc., Ocular Therapeutix Inc., and Regeneron Pharmaceuticals Inc.).
The court therefore looked to the Pimentel reference which teaches that the “optimal dosage . . . may in fact be higher than” the 400 mg three-times-daily (1200 mg/day) used in the published study. Id. According to the Federal Circuit, “the next higher” dosage unit in the prior art greater than the 400 mg dosage in the Pimental reference was the clinical trial’s 550 mg dosage (albeit disclosed only as a twice daily dosage). Id. The Federal Circuit, therefore, commented that it found no clear error in the district court’s conclusion that a skilled artisan would have had a reasonable expectation of success in using the protocol’s 550 mg dose in the Pimentel reference’s three-times-daily dosage plan, especially in view of other “background” prior art showing success with various dosing regimens of rifaximin. This conclusion, the court said, was consistent with its prior holdings that “certainty and absolute predictability are not required to establish a reasonable expectation of success.” Decision at 8 (citing Almirall, LLC v. Amneal Pharms. LLC, 28 F.4th 265, 275 (Fed. Cir. 2022); Acorda Therapeutics, Inc. v. Roxane Lab’ys, Inc., 903 F.3d 1310, 1333 (Fed. Cir. 2018)).
The dissent took issue with the majority’s reliance on the Pimentel reference’s statement that the “‘optimal dosage of rifaximin may, in fact, be higher than that used in our study.’” Dissent at 5. According to the dissent, the statement that the “optimal dosage . . . may . . . be higher” does little to create a reasonable expectation of success. Id. (emphasis added). The dissent noted that the district court relied on this statement from the Pimentel reference only as evidence of a motivation to combine the two references—not as evidence of a reasonable expectation of success. Id. In addition, the dissent noted that the other “background” prior art showing success with different dosing regimens relied on by the district court and the majority on appeal would tend to teach away from the expressly claimed three-times-daily 550 mg (1650 mg/day) dosage. For these reasons, the dissenting judge would have vacated the district court’s judgment of invalidity and remanded for further consideration of the background prior art’s tendency to teach away from the claimed invention. Id. at 6-7.
The Polymorph Form β Patents
Salix’s polymorph patents, U.S. Patent Nos. 7,612,199 and 7,902,206, claim a specific crystalline polymorph of rifaximin, referred to in the patents’ disclosures as “Form β.” The crystalline structure of Form β is specifically disclosed in the patents by reference to its three unique x-ray diffraction peaks. Decision at 12.
There was no dispute that the Salix polymorph patents were the first to characterize and disclose the crystalline structure of the rifaximin Form β polymorph. Nevertheless, the district court held that the Salix Form β polymorph patent claims were obvious in view of a prior art patent issued to the named inventor, Cannata. The Cannata reference disclosed that “rifaximin exists in crystalline form” but did not characterize or otherwise describe its crystalline structure. Id. At most, Cannata disclosed various rifaximin synthesis protocols that “include solvents for crystallization.” Id.
In ruling that Form β was obvious in view of Cannata, the district court adopted the opinions of Norwich’s expert that (1) “polymorph β is a commonly produced polymorph and the most stable form of rifaximin,” (2) a skilled artisan could have used the crystallization solvents disclosed in Cannata to make Form β, and (3) the work required to identify and characterize any resulting crystalline forms of rifaximin, including Form β, was merely routine. Id. at 15.
In affirming the district court, the Federal Circuit concluded that “Salix has done no more than combine known elements of the prior art to verify readily accessible information concerning a compound already in the hands of those of ordinary skill in the art, and such routine efforts do not justify removing this polymorph from the public domain.” Id. at 16 (quoting KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 427 (2007)). In doing so, the Federal Circuit rejected Salix’s contention that the court’s only two prior decisions addressing the obviousness of crystal polymorphs recognized the widely accepted unpredictability of polymorph chemistry and therefore weighed against the district court’s obviousness ruling. Decision at 13-14 (referencing Grunenthal GMBH v. Alkem Laboratories Ltd., 919 F.3d 1333 (Fed. Cir. 2019) and Pharmacyclics LLC v. Alvogen, Inc., No. 2021-2270, 2022 WL 16943006 (Fed. Cir. Nov. 15, 2022)). In response, the Federal Circuit explained that Grunenthal and Pharmacyclics merely held that the district court in each case did not clearly err in rejecting the obviousness challenges to the polymorph patent claims and that the two cases therefore were distinguishable on the basis of their unique facts. Id.
Finally, Salix argued that even if Cannata disclosed crystalline rifaximin, the physical structure of Form β and its unique x-ray diffraction pattern were unknown, could not have been predicted, and therefore could not have been obvious to a skilled artisan based on the crystallization solvents disclosed in Cannata. Id. at 16. At most, one might predict the appearance of some crystalline form of rifaximin but not Form β itself. Id. The Federal Circuit observed, without citation, that “Salix’s framing of the issue suggests that no unknown entity could ever be obvious, as one cannot reasonably expect what was hitherto unknown, which is incorrect.” Id.
The outcome of this issue may yet prove controversial, as both the district court and the Federal Circuit arguably neglected established precedent regarding the doctrine of inherency in ruling that the claimed crystalline polymorph invention was obvious. The issue presented here was that a specific claim element, the physical structure of Form β by reference to its x-ray diffraction pattern or otherwise, was not expressly disclosed in the Cannata prior art reference.
The doctrine of inherency, however, excuses this lack of express disclosure so long as the undisclosed claim element is a property inherent within the subject matter disclosed in the prior art. See Continental Can Co. v. Monsanto Co., 948 F.2d 1264, 1268 (Fed. Cir. 1991) (“evidence must make clear that the missing descriptive matter is necessarily present in the thing described in the reference”). Applying the doctrine of inherency in the context of an anticipation analysis under Section 102 is ordinarily straightforward. See Schering Corp. v. Geneva Pharm., 339 F.3d 1373, 1377 (Fed. Cir. 2003) (“a prior art reference may anticipate without disclosing a feature of the claimed invention if that missing characteristic is necessarily present, or inherent, in the single anticipating reference” even when the missing feature was unrecognized in the prior art).
However, the Federal Circuit has emphasized that “the use of inherency, a doctrine originally rooted in anticipation, must be carefully circumscribed in the context of obviousness.” Par Pharm., Inc. v. TWi Pharm., Inc., 773 F.3d 1186, 1195 (Fed. Cir. 2014) (citing In re Rijckaert, 9 F.3d 1531, 1533-34 (Fed. Cir. 1993)). “A party must, therefore, meet a high standard in order to rely on inherency to establish the existence of a claim limitation in the prior art in an obviousness analysis—the limitation at issue necessarily must be present, or the natural result of the combination of elements explicitly disclosed by the prior art.” Par Pharm., 773 F.3d at 1196.
The district court’s factual findings appear to satisfy this “necessarily must be present, or the natural result” test for inherency of the undisclosed claim element—here the inherent presence of the Form β polymorph. The district court specifically found, “‘[T]he as-synthesized form of rifaximin reported by Examples 1, 6, 7, and 9 [of Cannata] were necessarily rifaximin form Beta, because of the methods used, the solvent system used, and it was later confirmed by later work, including work from the named inventors.’” Decision at 15 (emphasis added).
Because the Cannata reference was asserted alone and not in combination with other prior art references, this finding suggests that Cannata inherently anticipated Salix’s polymorph patent claims under Section 102. See Schering, 339 F.3d at 1377. Thus, the district court and Federal Circuit likely reached the correct result but for the wrong reasons. Indeed, had the Federal Circuit found inherent anticipation, it could have avoided the unsatisfying struggle of explaining a skilled artisan’s reasonable expectation of success in producing an unknown outcome. See Decision at 16 (“Salix’s framing of the issue suggests that no unknown entity could ever be obvious, as one cannot reasonably expect what was hitherto unknown, which is incorrect”).
According to prior Federal Circuit precedent, “[T]he use of inherency in the context of obviousness must be carefully circumscribed because ‘[t]hat which may be inherent is not necessarily known’ and that which is unknown cannot be obvious.” Honeywell, Int’l Inc. v. Mexichem Amanco Holding S.A. de C.V., 865 F.3d 1348, (Fed. Cir. 1993) (quoting In re Spormann, 363 F.2d 444, 448 (C.C.P.A. 1966) (“[T]he inherency of an advantage and its obviousness are entirely different questions. . . . Obviousness cannot be predicated on what is unknown”)). Indeed, “[w]hat is important regarding properties that may be inherent, but unknown, is whether they are unexpected.” Honeywell, 865 F.3d at 1355 (emphasis added) (“All properties of a composition are inherent in that composition, but unexpected properties may cause what may appear to be an obvious composition to be nonobvious”); see also Par Pharm., 773 F.3d at 1195 (distinguishing prior case finding obviousness based on inherent properties where “neither party disputed that the [inherent properties] were expected in light of the dosages disclosed in the prior art”) (emphasis added).
Neither the district court nor the Federal Circuit found that, based on a reading of Cannata, a person of ordinary skill in the art would have expected to produce a crystalline polymorph having the physical structure of Form β with its attendant x-ray diffraction pattern by following the solvent crystallization methods disclosed in the reference. In fact, Norwich’s expert offered no such opinion. Instead, the Federal Circuit sidestepped the reasonable expectation issue by concluding that a skilled artisan “would have [had] a motivation to explore potential polymorphic forms of rifaximin” and that “the methods for characterizing the resulting crystalline rifaximin were well known and readily available to the skilled artisan.” Decision at 15.
According to this analysis, the burden of proving a reasonable expectation of success can alternatively be met by demonstrating a motivation to experiment using routine methods. Indeed, the Federal Circuit emphasized that the methods used to characterize crystalline structures are so routine that they “could have been performed ‘in one day.’” Decision at 13 (emphasis added). This suggests that a court’s obviousness determination perversely depends on whether the routine experimentation motivated by the prior art produces—not a result that a skilled artisan would expect from reading the prior art—but instead a result that was revealed in hindsight for the first time in the challenged patents.
This analysis arguably conflicts with the Patent Act and prior Federal Circuit precedent which preclude reliance on the routine nature of testing: “A further point regarding so-called ‘routine testing’ is that § 103 provides that ‘[p]atentability shall not be negated by the manner in which the invention was made . . . [a provision which] was enacted to ensure that routine experimentation does not necessarily preclude patentability.” Honeywell, 865 F.3d at 1356 (quoting 35 U.S.C. § 103). The “routine” nature of characterizing crystalline structures therefore was arguably (and statutorily) irrelevant to the court’s obviousness analysis.
Laboratory techniques evolve in ways that previously time-consuming and complicated bench assays eventually become hands-off procedures performed by automated equipment. Thus, the Federal Circuit’s obviousness analysis here establishes a standard that depends not on what can be expected from reading the prior art, but rather on whether a particular laboratory procedure tangentially related to producing or characterizing the claimed invention had become “routine” as of the priority date.
By focusing on the routine nature of testing, the district court and Federal Circuit arguably failed to address the key inquiry required by Honeywell and other Federal Circuit precedent—whether a skilled artisan would have expected the crystalline rifaximin disclosed in Cannata to have the structure and properties of the Form β polymorph based on Cannata’s disclosure and other teachings in the prior art. Given the widely accepted unpredictability of physical chemistry as discussed in Grunenthal and Pharmacyclics, it is fair to say that Form β could not reasonably have been expected from the prior art teachings and that the crystalline structure of the polymorph was not obvious, regardless of the experimental methods used. See Honeywell, 865 F.3d at 1356 (“[u]npredictability of results equate more with nonobviousness rather than obviousness”).
The HE Patents
Following the bench trial, the district court ruled that the asserted claims of Salix’s HE patents, U.S. Patent Nos. 8,624,573; 9,421,195; and 10,335,397, were both valid and infringed. Norwich chose not to appeal either ruling. Instead, Norwich cross-appealed from (1) the district court’s entry of judgment delaying approval of its ANDA until the expiration of the HE patents and (2) the district court’s denial of Norwich’s motion under Fed. R. Civ. P. 60(b) to modify the judgment based on its post-judgment amendments to the ANDA removing the infringing HE indications from the product label.
As to the first issue, Norwich contended that the district court misapplied 35 U.S.C. § 271(e)(4)(A) by imposing a blanket stay of final FDA approval of the entire ANDA until the expiration of the HE patents. Because the district court held that Salix’s IBS-D method of treatment and polymorph patent claims were invalid, Norwich argued that the district court’s judgment should have permitted the FDA to immediately grant final approval of the ANDA’s non-infringing IBS-D indications while delaying final approval of the infringing HE indications until the HE patents expired.
In addressing this issue, the Federal Circuit applied a strict reading of the statute, which requires that “‘the court shall order the effective date of any approval of the drug . . . to be a date which is not earlier than the date of the expiration of the patent which has been infringed.’” Decision at 18 (quoting Section 271(e)(4)(A)). The court made clear that if any part of an ANDA is found to infringe, the delayed approval required by the statute applies to the entire ANDA. This result follows naturally from the language of the statute which “describes delaying the approval of ‘the drug . . . involved in the infringement.’” Decision at 19 (quoting Section 271(e)(4)(A)).
Moreover, because Section 271(e)(2)(A) defines the act of infringement as the submission of an ANDA “for a drug claimed in a patent or the use of which is claimed in a patent,” the fact “[t]hat the ANDA further recited a non-patent-protected indication does not negate the infringement resulting from the ANDA’s submission.” Decision at 19. Conspicuously, the Federal Circuit noted that the district court’s judgment “appropriately delayed the effective final approval date of ‘this infringing ANDA’ submission” and likewise “appropriately said nothing that would prevent approval of a new non-infringing ANDA.” Id.
The Federal Circuit thus left open the possibility that Norwich could submit a new “skinny-labeled” ANDA without the infringing HE treatment indications. Id. In doing so, the Federal Circuit recognized the different procedural posture of a newly submitted “skinny labeled” ANDA compared to Norwich’s post-judgment ANDA amendments that removed the infringing HE indications. In its Rule 60(b) motion to modify the judgment in light of its amended ANDA, Norwich argued that “the amended ANDA satisfies the judgment by not seeking approval for the infringing use and that, in view of the amendment, it is no longer equitable to apply the judgment prospectively.” Decision at 21.
The district court disagreed and denied Norwich’s motion, explaining that “‘[i]t is not a simple matter to determine whether an ANDA applicant has successfully carved out language from a label to turn infringement into non-infringement’ and that what Norwich sought in its Rule 60(b) motion ‘would essentially be a second litigation’ following final judgment.’” Id. The district court noted further “that, other than simply asserting that it carved out the HE indication and providing the court with the amended label, Norwich ‘ha[d] presented no evidence in support of its assertion’ that the amended ANDA would no longer infringe the HE patents.” Id.
The Federal Circuit affirmed, finding no reason to conclude that the district court abused its discretion in refusing to modify the judgment in response to the ANDA amendments. Id.; see Ferring B.V. v. Watson Labs., Inc. Fla., 764 F.3d 1382, 1391 (Fed. Cir. 2014) (“[a] district court may reconsider its own finding of infringement in light of an amended ANDA” but need not do so). Implicit in the Federal Circuit’s affirmance is the notion that a newly submitted Norwich ANDA would present the appropriate context in which to adjudicate whether, despite the skinny label, no other safety, clinical, or treatment data remaining in the product label together with other information originating from Norwich such as promotional materials would support a finding of induced infringement of the HE method of treatment patent claims. These issues would be adjudicated appropriately in a newly filed lawsuit based on the subsequent ANDA submission, with the benefit of a new automatic 30-month stay of FDA approval to accommodate this judicial process.
Conclusion
The Federal Circuit's decision in Salix Pharmaceuticals Ltd. et al. v. Norwich Pharmaceuticals Inc. illustrates the complex and often unpredictable nature of Hatch-Waxman litigation. The courts’ increasing reliance on published clinical trial protocols as primary references in obviousness challenges to method-of-treatment patents arising from that clinical work presents a unique challenge to drug innovators who must publish the protocols to comply with regulatory requirements. Although the amicus filers cautioned the court against overreliance on clinical protocols, any lasting relief from the growing practice will likely have to come from Congress. The Federal Circuit’s reliance on the use of “routine” experimentation as a proxy for proof of a reasonable expectation of success is likewise troubling, especially in cases like this where the court effectively admitted that the allegedly “obvious” endpoint of experimentation was unknown given the unpredictable field of science. Finally, the decision complicates the strategic “skinny labeling” decisions made by generic challengers. The Federal Circuit’s strict interpretation of Section 271(e)(4)(A) means that the generic challenger cannot expect to easily conform its product labeling to the results of the litigation. In this situation, Norwich would have been much better off if its originally submitted ANDA did not include treatment indications for HE. Otherwise, Norwich would have had a clear path to launch and sell its generic drug product for the treatment of IBS-D.