Acorda Therapeutics, Inc. v. Roxane Laboratories, Inc.

United States Court of Appeals for the Federal Circuit ______________________ ACORDA THERAPEUTICS, INC., Plaintiff-Appellant ALKERMES PHARMA IRELAND LIMITED, Plaintiff-Appellee v. ROXANE LABORATORIES, INC., MYLAN PHARMACEUTICALS INC., TEVA PHARMACEUTICALS USA, INC., Defendants-Cross-Appellants ______________________ 2017-2078, 2017-2134 ______________________ Appeals from the United States District Court for the District of Delaware in Nos. 1:14-cv-00882-LPS, 1:14-cv- 00922-LPS, 1:14-cv-00935-LPS, 1:14-cv-00941-LPS, Chief Judge Leonard P. Stark. ______________________ Decided: September 10, 2018 ______________________ BRUCE M. WEXLER, Paul Hastings LLP, New York, NY, argued for plaintiff-appellant. Also represented by STEPHEN BLAKE KINNAIRD, IGOR VICTOR TIMOFEYEV, Washington, DC; GARRARD R. BEENEY, WENYING ANGELA CHANG, STEPHEN J. ELLIOTT, Sullivan & Cromwell LLP, 2 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. New York, NY; ANTHONY MICHAEL, JANE G. WASMAN, Acorda Therapeutics, Inc., Ardsley, NY. MARYELLEN NOREIKA, Morris, Nichols, Arsht & Tun- nell LLP, Wilmington, DE, for plaintiff-appellee. Also represented by JEREMY A. TIGAN. CHARLES B. KLEIN, Winston & Strawn LLP, Washing- ton, DC, argued for defendants-cross-appellants. Defend- ants-cross-appellants Roxane Laboratories, Inc., Teva Pharmaceuticals USA, Inc. also represented by ANDREW CURTIS NICHOLS; BRYCE COOPER, GEORGE C. LOMBARDI, REID SMITH, Chicago, IL. ROBERT FLORENCE, Parker Poe Adams & Bernstein LLP, Atlanta, GA, for defendant-cross-appellant Mylan Pharmaceuticals Inc. Also represented by MICHEAL L. BINNS, KAREN L. CARROLL. SARAH ANNE KAGAN, Banner and Witcoff, Ltd., Wash- ington, DC, for amicus curiae Biotechnology Innovation Organization. Also represented by MELISSA A. BRAND, LISA MEREDITH HEMMENDINGER; HANSJORG SAUER, Bio- technology Innovation Organization, Washington, DC. SCOTT E. KAMHOLZ, Covington & Burling LLP, Wash- ington, DC, for amicus curiae Pharmaceutical Research and Manufacturers of America. Also represented by BRIANNE BHARKHDA; DAVID EVAN KORN, Pharmaceutical Research and Manufacturers Association of America, Washington, DC. ______________________ Before NEWMAN, DYK, and TARANTO, Circuit Judges. Opinion for the court filed by Circuit Judge TARANTO. Opinion dissenting filed by Circuit Judge NEWMAN. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 3 TARANTO, Circuit Judge. Before us are patents that claim the administration of a medication containing the active ingredient 4- aminopyridine (4-AP) to improve walking in individuals with multiple sclerosis. Acorda Therapeutics, Inc., holds New Drug Application No. 022250, approved by the U.S. Food and Drug Administration (FDA). Pursuant to that approval, Acorda markets, under the name “Ampyra®,” 10 milligram 4-AP sustained-release tablets for twice- daily oral administration. In the FDA’s Approved Drug Products with Therapeutic Equivalence Evaluations, or Orange Book, Acorda has listed, as claiming methods of using Ampyra, four patents that Acorda owns: U.S. Pa- tent No. 8,007,826; No. 8,663,685; No. 8,354,437; and No. 8,440,703. Those patents (“the Acorda patents”) are the main patents at issue on appeal. One additional patent is before us. Acorda holds an exclusive license to an earlier, broader patent, U.S. Patent No. 5,540,938, referred to as “the Elan patent” because it was originally assigned to Elan Corporation, plc (whose successor in interest is Alkermes Pharma Ireland Ltd.). The Elan patent, listed in the Orange Book for Ampyra along with the Acorda patents, claims methods of treating patients having certain conditions, including multiple sclerosis, by administering a drug containing a sustained- release formulation of any of certain agents, one of them 4-AP. The later Acorda patents claim species of the Elan patent’s genus claims by adding further, more specific requirements to the Elan patent’s claimed methods. While the Elan patent’s claims broadly cover administer- ing a sustained-release formulation of 4-AP to individuals with multiple sclerosis, the Acorda patents’ claims further specify that such a drug must be administered (1) in a 10 mg dose twice a day (2) at that stable dose for the entire treatment period of at least two weeks (3) to achieve 4-AP serum levels of 15–35 ng/ml and (4) to improve walking. 4 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Roxane Laboratories, Inc.; Mylan Pharmaceuticals, Inc.; and Teva Pharmaceuticals USA, Inc., have submit- ted Abbreviated New Drug Applications seeking FDA approval to market generic versions of Ampyra. In July 2014, Acorda and Alkermes sued those entities (“defend- ants”) in the District of Delaware, alleging infringement of several claims in each of the Elan and Acorda patents. The defendants stipulated to infringement but challenged the validity of the asserted claims. The district court held that the asserted claims in the Acorda patents are invalid for obviousness. But the court upheld the asserted claims of the Elan patent against invalidity challenges and enjoined the defendants from activity infringing that patent until it expired on July 30, 2018. Acorda appealed the invalidity ruling regarding the Acorda patents. The defendants cross-appealed the validity ruling regarding the Elan patent and the result- ing injunction. We now affirm the judgment that the asserted Acorda patent claims are invalid. We dismiss the cross-appeal as moot. I A In view of our decision that the issues concerning the Elan patent are moot, we focus on the background of the Acorda patents. Essential to understanding the obvious- ness issue is an understanding of the prior art. 4-AP, also called “dalfampridine” and “fampridine,” was first identified in 1902. Acorda Therapeutics, Inc. v. Roxane Labs., Inc., No. 1:14-cv-00882-LPS, 2017 WL 1199767, at *3, *5 (Mar. 31, 2017) (Dist. Ct. Op.). Belong- ing to a class of compounds that function as potassium- channel blockers, 4-AP “has been found to slow the potas- sium flow in nerve impulse transmission” and, by doing so, help “restor[e] conduction in blocked and demyelinated nerves,” ’826 patent, col. 2, lines 5–11, i.e., nerves whose ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 5 myelin insulation has been damaged. 4-AP was first used in human studies in the 1970s to investigate its effect on neurological diseases resulting in muscle weakness. Dist. Ct. Op. at *5. For several decades, 4-AP has been the focus of research regarding the treatment of multiple sclerosis in particular. See, e.g., id. at *5–7 (reciting studies); J.A. 6697 (paper published in 1987 describing study of the effect of 4-AP on subjects with multiple sclerosis). Multiple sclerosis causes the demyelination, or loss of myelin, of nerves in the central nervous system and results in a wide variety of symptoms, including walking impairment, tingling or pain, brain scarring, cognitive changes, visual impairments, and fatigue. See ’826 patent, col. 1, lines 36–42; Dist. Ct. Op. at *2. Even- tually, 4-AP research led to the development, patenting, and FDA approval of Ampyra. 1 In the 1980s, researchers at the Rush Medical School conducted a study on 12 patients with multiple sclerosis, and 5 without, to determine whether intravenous admin- istration of 7 to 35 mg of 4-AP had any therapeutic effect on multiple sclerosis. J.A. 6697 (Dusan Stefoski et al., 4- Aminopyridine Improves Clinical Signs in Multiple Scle- rosis, 21 Annals of Neurology 71 (1987)). According to the published paper reporting that study (Stefoski), 10 of the 12 patients with multiple sclerosis “showed mild to marked improvement”; “[v]ision improved in 7 patients, oculomotor function in 5, and motor function (power, coordination, gait) in 5.” J.A. 6697. Improvements were seen at doses as low as 2 mg: In one patient, gait im- provement occurred within 25 minutes of administration of a total dose of 2 mg. J.A. 6699. Stefoski also reported: [W]e observed no serious or bothersome side ef- fects at total doses below 30 to 35 mg injected not less than 20 minutes apart for aliquots up to 3 mg. Moreover, the clinical improvements in 6 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. many of our patients were of sufficient magnitude to represent a functionally noteworthy therapeutic benefit. Studies are currently in progress to de- termine the clinical usefulness of oral 4-AP as a symptomatic treatment. J.A. 6701; accord J.A. 6697. In 1990, an overlapping group of researchers pub- lished a paper (Davis) reporting another study on 4-AP’s effect on symptoms of multiple sclerosis. J.A. 6327 (Floyd A. Davis et al., Orally Administered 4-Aminopyridine Improves Clinical Signs in Multiple Sclerosis, 27 Annals of Neurology 186 (1990)). In that study, 20 patients with multiple sclerosis were given either a single oral dose of 4- AP (15 patients) or a placebo (5 patients). J.A. 6327. Of those in the active treatment group, 4 patients were given a 10 mg dose of 4-AP, 2 were given 12.5 mg, 4 were given 15 mg, 4 were given 20 mg, and 1 was given 25 mg. Davis at 187 tbl.1. Davis states that “[m]ild to marked im- provements occurred in all of the 15 [multiple sclerosis] patients given 4-AP.” J.A. 6329; accord J.A. 6327. “Im- provements developed gradually with doses as low as 10 mg 4-AP, usually beginning within 60 minutes after drug administration.” J.A. 6329. Motor function im- proved in 9 of 13 patients in the active treatment group (motor function was not measured in 2). Davis at 187 tbl.1; J.A. 6329. The improvements were “most striking[] with respect to power and coordination” and “were appar- ent with both simple function tests and the performance of complex motor tasks such as gait and repetitive move- ments.” J.A. 6329. Finally, Davis notes, no “serious or bothersome side effects,” including seizures, were ob- served at single oral doses up to 25 mg. J.A. 6332. A few years later, researchers at a university hospital in the Netherlands published a paper (Van Diemen) reporting a randomized, double-blind, placebo-controlled crossover study that “demonstrated efficacy of [4-AP] in ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 7 improving disability of patients with multiple sclerosis.” J.A. 7037 (Harriët A. M. Van Diemen et al., 4- Aminopyridine in Patients with Multiple Sclerosis: Dosage and Serum Level Related to Efficacy and Safety, 16 Clini- cal Neuropharmacology 195 (1993)). In the second phase of the study lasting 12 weeks, 69 patients were orally administered 10–20 mg 4-AP per day, split into two or three doses. J.A. 7038, 7042. The doses were escalated during the second week, and again during the sixth week, by 5–15 mg. J.A. 7038–39. The paper reports improve- ments in certain measures of eye functioning. J.A. 7042. And it reports that “side effects were mild” for those patients given oral doses of 4-AP (versus intravenous 4- AP). J.A. 7045; see also Van Diemen at 200–01 (no sei- zures). Soon thereafter, some of the same researchers pub- lished a second paper (Polman) about the long-term efficacy and safety of 4-AP given to patients with multiple sclerosis. J.A. 6654 (Chris H. Polman et al., 4- Aminopyridine in the Treatment of Patients with Multiple Sclerosis, 51 Archives of Neurology 292 (1994)). Polman reports a study of 31 patients with multiple sclerosis, 19 of whom took a stable dose of 4-AP between 10–50 mg per day (the exact dose for each patient is unknown), and 12 of whom initially took 10–15 mg per day and then took increasing doses in 4 to 8 weeks. J.A. 6655; see J.A. 7042. In the first group, 18 of the 19 patients “had a favorable response to the medication” and “reported a subjective improvement in the ability to perform the activities of normal daily life, which was mainly owing to improved ambulation and reduction in severity of fatigue.” J.A. 6655. In 3 patients, the subjective improvement was significant on the Expanded Disability Status Scale (EDSS), id.—a composite measure of function in multiple sclerosis patients, including a walking component, that is “widely accepted in the [multiple sclerosis] community,” Dist. Ct. Op. at *8; see id. at *30; J.A. 6681. In the second 8 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. group, 6 patients reported a “favorable response” to 4-AP treatment, “as defined by the ability to perform activities of normal daily life.” J.A. 6655–56. One patient demon- strated a significant improvement in EDSS score. J.A. 6656. Overall, 23 patients (17 in the first group; 6 in the second group) continued active treatment for 6 to 32 months, with daily doses ranging from 15–40 mg. J.A. 6655–56. Those patients “indicated the drug to be beneficial because, by improving several neurologic func- tions, it increased their capability to perform the activi- ties of normal daily life,” including—for 13 of the 23 patients—a reported improvement in ambulation and fatigue. J.A. 6656 & tbl.1; see J.A. 6654. 1 The paper states: Although a placebo effect cannot be excluded, the dynamics of the response in relation to the intake of the medication and the deterioration and sub- sequent improvement in functioning during a drug-free interval and subsequent restarting of the therapy are, in our view, highly suggestive of a real effect being induced by the 4-[AP]. Im- provements in fatigue and ambulation were men- tioned quite often by the patients as being responsible for the favorable overall effect . . . . J.A. 6657. The paper thus reports improvements in specific measures, while few patients experienced a significant change in EDSS, the overall composite meas- ure. Id. As for adverse effects, two patients experienced a 1 By comparison, only 5 reported an improvement in visual function; 4 in cognitive function/concentration; and 1 in diplopia (double vision), speech, spasticity, and urinary and fecal incontinence. J.A. 6656 tbl.1; see J.A. 6654. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 9 seizure—one on the second day of treatment and the other after 18 months of treatment. J.A. 6656–57. 2 Otherwise, the subjective side effects reported by the patients “never were reported to be very troublesome.” J.A. 6657. Polman states several conclusions and suggestions for further research. First, the study “demonstrates that 4- [AP] therapy, in the majority of patients who favorably respond to it, results in responses that can continue for periods of up to 32 months or more without interfering with the course of the disease.” Polman at 296. Second, the fact that “three major, though not life-threatening, side effects” occurred (including 2 seizures) “indicates that careful medical supervision is warranted during 4-[AP] therapy.” Id. Third, based on the study data, the authors “suggest that approximately 30% of patients with [multi- ple sclerosis] will report a significant clinical response when they begin treatment with 4-[AP] and that 80% to 90% of these responders will benefit from long-term administration. More studies are needed for further elaboration of the exact value of 4-[AP] in the long-term treatment of patients with [multiple sclerosis].” Id. Around the same time, researchers at the University of Maryland, the Baltimore VA Medical Center, and Elan published a paper (Bever I) reporting the results of a randomized, placebo-controlled, double-blind, concentra- tion-controlled, crossover trial in 8 patients with multiple sclerosis. Christopher T. Bever, Jr., et al., The effects of 4-aminopyridine in multiple sclerosis patients: Results of a randomized, placebo-controlled, double-blind, concentra- tion-controlled, crossover trial, 44 Neurology 1054 (1994); see J.A. 6180 (excerpt of Bever I). Noting that 4-AP has a “narrow toxic-to-therapeutic range[],” the study aimed to evaluate the toxicity and efficacy of 4-AP when the result- 2 A third patient was presumptively diagnosed with a case of 4-AP-induced hepatitis. J.A. 6657. 10 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. ing peak concentration in blood was low (30–59 ng/ml) versus when it was high (60–100 ng/ml). Bever I at 1055. Regarding toxicity, the report states that “[a]ll patients experienced side effects” when serum concentration was high, with two serious adverse events: a seizure when serum 4-AP peaked at 104 ng/ml, and an episode of en- cephalopathy when serum 4-AP peaked at 114 ng/ml. Id. at 1054, 1056. Regarding efficacy, “[i]mprovements were seen in lower extremity strength,” including significant improvement in mean videotape scores of lower extremity strength (scoring muscle strength, reflexes, and ambula- tion) in both the low- and high-serum concentration ranges, although no significant changes were seen in EDSS scores or ambulation index (AI) scores. 3 Id. at 1056–57 & tbl.4; but see id. at 1058 (commenting that the increased side effects from the short treatment duration “may have contributed to the lack of improvement in overall function (EDSS and AI scores)”). Bever I concludes that the therapeutic response was not concentration-related as between the two ranges tested and, therefore, that “[t]he lower serum concentra- tion range of 30 to 59 ng/ml may . . . be adequate for inducing improvement of some neurologic deficits.” Bever I at 1058; see id. (“Because the high-serum- concentration arm produced much greater toxicity than the low without any obvious therapeutic advantage, it seems likely that clinically useful serum concentrations would be in the 30 to 59 ng/ml range.”). Bever I also states that the “rates of treatment-related improvements in visual and lower extremity motor function . . . were 3 See Stephen L. Hauser et al., Intensive immuno- suppression in progressive multiple sclerosis, 308 New Eng. J. Med. 173, 174, 180 (1983) (ambulation index is a rating scale to assess mobility by measuring the time and degree of assistance needed to walk 25 feet). ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 11 similar to those reported in similar short-term trials of [4- AP],” including Stefoski and Davis. Bever I at 1057–58. The article notes the limitations of the earlier trials’ designs, including “questions about blinding, failure to randomize treatment, and failure to either use prospec- tively defined neurologic deficits or adjust significance levels to compensate for multiple comparisons.” Id. at 1058. Bever I then observes that another study “ad- dressed some of the design weaknesses in earlier studies and suggested that not only can AP treatment improve specific residual deficits, but it can also improve overall function.” Id. The same year as Bever I appeared, Dr. Bever, with the University of Maryland and the Baltimore VA Medical Center, published a review article on studies of the effect of 4-AP on multiple sclerosis (Bever II). Christopher T. Bever, Jr., The Current Status of Studies of Amino- pyridines in Patients with Multiple Sclerosis, 36 Annals of Neurology S118 (1994); see J.A. 6172 (excerpt of Bever II). The article states: “Recently completed randomized, double-blind, placebo-controlled trials show that treat- ment with the potassium channel blockers 4- aminopyridine (AP) or 3,4-diaminopyridine (DAP) can improve residual neurological deficits in some multiple sclerosis (MS) patients.” Bever II at S118; accord id. at S120. As to efficacy, “[t]hese studies suggest that amino- pyridines may provide a new approach to the symptomat- ic treatment of [multiple sclerosis].” Id. at S118. 4 As to 4 Although criticizing a few 4-AP studies as involv- ing a small sample size or lacking a double-blinded or randomized design, Bever II also looked at “[l]arger randomized, double-blind, placebo-controlled crossover trials of” 4-AP with treatment periods as long as three months. J.A. 6172; accord Bever II at S118 (in the article abstract, stating that “[p]reliminary studies of [4-]AP 12 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. toxicity, “seizures are common at higher doses,” but 4-AP “rarely cause[s] seizures at the doses used in [multiple sclerosis] trials.” Id. at S120; see also id. at S118 (“Both agents [4-AP and DAP] have rarely caused seizures.”). The paper notes that one 4-AP study “showed that side effects correlated with peak serum concentrations, while efficacy correlated with total drug exposure, suggesting that controlled release formulations may be useful in minimizing toxicity.” Id. at S120. 2 The foregoing studies involved immediate-release, ra- ther than sustained-release, formulations of 4-AP. See Dist. Ct. Op. at *4; J.A. 761, 763, 767, 769, 774 (testimony of Acorda’s expert, Dr. Andrew Goodman). By 1990, Elan, which was known for its work on sustained-release formu- lations, entered into an agreement with the researchers at Rush Medical School to obtain their work on 4-AP phar- maceutical formulations. Dist. Ct. Op. at *4. According to Dr. Michael Myers, who worked at Elan at that time and is a named inventor on the Elan patent, Elan was inter- ested in developing a sustained-release formulation of 4- AP to “potentially reduce or eliminate some of th[e] side effects” associated with the immediate-release formula- tion. Sept. 19, 2016 Trial Tr. at 149, 155–56, Acorda Therapeutics, Inc. v. Alkem Labs. Ltd., No. 1:14-cv-00882- LPS (D. Del. Oct. 21, 2016), ECF No. 266. Elan developed a 4-AP sustained-release formulation in approximately a month’s time. Dist. Ct. Op. at *4. The inventors then filed for what became the Elan patent, demonstrated benefit in many temperature-sensitive patients with [multiple sclerosis], and improvement of function was found in a large randomized double-blind, placebo-controlled crossover trial of 3 months of oral treatment in 68 patients with [multiple sclerosis]”). ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 13 which claims, among other things, administration of a sustained-release formulation of 4-AP once or twice daily for the treatment of neurological diseases, including multiple sclerosis. Elan patent, col. 22, lines 16–25, 29– 30, 50–51 (independent claim 1 and dependent claims 3 and 8). The Elan patent has a priority date of Novem- ber 1, 1991; issuance date of July 30, 1996; and expiration date of July 30, 2018. In 1994, Elan conducted a double-blind, randomized, placebo-controlled clinical trial involving 161 patients with multiple sclerosis to study the safety and efficacy of the sustained-release 4-AP formulation. Dist. Ct. Op. at *8. Patients were administered 12.5 mg 4-AP twice a day, which was later increased to 17.5 mg twice a day and finally to 22.5 mg twice a day. Id. One of the primary endpoints measured was the EDSS composite measure of function. See id. For the primary endpoints and most of the secondary endpoints, including ambulation, the trial revealed no statistically significant improvements for 4- AP versus placebo. Id. But it did show a statistically significant improvement in the secondary outcome of lower extremity motor score, a measure of muscle strength in the legs. Id. The 1994 Elan study was not published. Elan also sponsored a smaller, double-blind, placebo- controlled, crossover study in ten patients with multiple sclerosis. That study was reported in a paper published in 1997 (Schwid), on which Dr. Goodman, Acorda’s expert at trial, was the senior author. J.A. 6681–84 (Steven R. Schwid et al., Quantitative assessment of sustained-release 4-aminopyridine for symptomatic treatment of multiple sclerosis, 48 Neurology 817 (1997)). In the background section, Schwid reports that an earlier, 161-patient study had been conducted to test improvement in EDSS for multiple sclerosis patients (the unpublished 1994 Elan study), but that it did not detect a significant improve- ment in that measure. J.A. 6681. Schwid notes, however, 14 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. that the EDSS “may have been an inadequate outcome variable for [the 1994 Elan] trial.” Id. The paper ex- plains: [The EDSS] is imprecise due to substantial intra- rater and inter-rater variability, and relatively in- sensitive to change due to its ordinal nature. For example, a patient who needed a cane to walk 100 meters would need to improve enough to walk without the cane before the EDSS score would change. Lesser improvements in gait would not be reflected by the EDSS, and notable changes in strength or other deficits could also be overlooked. We planned the present pilot study to assess the effect of 4AP [sustained release] on more sensi- tive, quantitative measures of function in [multi- ple sclerosis]. Id. (internal references omitted). In the Schwid study, ten patients were each given 17.5 mg sustained-release 4-AP twice a day for a week and placebo for a week. Id. The study measured (1) time to walk 8 meters (timed gait), (2) time to climb four stairs, (3) maximum voluntary isometric contraction measured quantitatively, (4) manual muscle testing, (5) grip strength, (6) EDSS, and (7) the patient’s global impres- sion. Id. Schwid reports that the administered drug demonstrated a statistically significant improvement over placebo for timed gait in 9 of 10 patients, with p = 0.02. Id. 5 In addition to that result, Schwid observes that “most of the other outcomes showed trends favoring 4AP 5 Dr. Goodman testified at trial (for Acorda) that the p-value would be 0.14 (greater than the customary 0.05 ceiling for “statistical significance”) if adjusted for the fact that there were multiple outcome measures (7 total). J.A. 878; see Dist. Ct. Op. at *13 n.10. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 15 [sustained-release].” J.A. 6684. Schwid concludes that, in the reported study, “4AP [sustained-release] improved motor function in [multiple sclerosis] patients.” J.A. 6681. The article notes that the results of the Schwid study are consistent with “[p]revious double-blind, placebo- controlled studies” using an immediate-release formula- tion of 4-AP, including another study reported by Stefoski (13 of 17 patients “showed ‘clinically important’ improve- ments”), Bever I (reporting that 4-AP “improved lower- extremity strength” and “a composite score of leg strength, spasticity, and ambulation”), and another study reported by Van Diemen (improvement in neurologic deficits, as measured by the EDSS). J.A. 6684. Schwid also states: “The quantitative outcomes used in this study permit more sensitive evaluation of the therapeutic effect and promise to be useful in future trials of symptomatic treatments for [multiple sclerosis].” J.A. 6681. It notes particularly that timed gait showed improvement where the EDSS did not. Id.; J.A. 6684. Schwid advises that future studies evaluate the more sensitive outcome measures, “establish[] efficacy in larger trials,” and “examine long-term efficacy and tolerability as well as further refine dosing regimens to optimize delivery despite a relatively narrow therapeutic window.” J.A. 6684. 3 While Elan was conducting those studies, Acorda was exploring the use of 4-AP in patients with spinal cord injuries. Dist. Ct. Op. at *8. In 1997, Elan granted Acor- da an exclusive license to the Elan patent for the use of Elan’s sustained-release formulation of 4-AP in patients with spinal cord injuries. Id. Acorda conducted two studies to evaluate the pharmacokinetic and safety profile of the sustained-release formulation, and the results of both studies are reported in a paper published in 2003 (Hayes). J.A. 6433–40 (Keith C. Hayes et al., Pharmaco- 16 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. kinetic Studies of Single and Multiple Oral Doses of Fampridine-SR (Sustained-Release 4-Aminopyridine) in Patients With Chronic Spinal Cord Injury, 26 Clinical Neuropharmacology 185 (2003)). In the second study, Acorda tested doses of 10 mg, 15 mg, 20 mg, and 25 mg of the sustained-release formulation of 4-AP administered twice daily in patients with spinal cord injuries. J.A. 6434. The average serum concentration level (at steady state) for the 10 mg twice-daily dose was 20.8 ± 5.7 ng/ml. J.A. 6439; accord ’826 patent, col. 25, lines 1–28 (Table 7); ’685 patent, col. 25, lines 5–32 (Table 7). Acor- da also conducted clinical trials to evaluate the efficacy of that sustained-release formulation of 4-AP in patients with spinal cord injuries, but those studies failed. Soon after, Acorda learned that Elan was “no longer interested in pursuing or supporting” research into use of Elan’s sustained-release formulation of 4-AP for treat- ment of multiple sclerosis. J.A. 596 (testimony of Dr. Ron Cohen, Acorda founder). Acorda told Elan that it wished to take over that research. Id. In 1998, Elan agreed to expand the earlier license to Acorda; it granted Acorda exclusive rights over the 4-AP sustained-release formula- tion for use in the treatment of multiple sclerosis. Dist. Ct. Op. at *8. Acorda reviewed Elan’s research, including Elan’s pharmacokinetic data and clinical study reports of the 1994 Elan study. Acorda then conducted its own clinical trials. Id. at *9. a In 2000 and 2001, Acorda ran a study—the MS-F201 study—which involved 36 patients with multiple sclerosis and whose results were published only in part. Id. 6 After 6 This was a Phase II study within the meaning of the FDA’s classification of certain studies as Phase I, II, ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 17 one week of a placebo lead-in, a group of 25 patients received 10 mg 4-AP twice daily for a week, then higher dosages, which increased weekly in 5 mg increments up to 40 mg twice daily at week 7. Id. The rest of the patients consistently received a placebo. See id. The outcome measures included fatigue, a lower extremity muscle test, a multiple sclerosis functional composite (timed 25-foot walk; nine-hole peg test; cognitive test), and subjective measures. Id. Only the lower extremity muscle test showed a statistically significant difference—“when comparing the seven week range [4-AP] group against placebo.” J.A. 604–05. The results were not statistically significant for the timed 25-foot walk for any particular dose of 4-AP; and in 3 of the 7 weeks, the placebo group did better in the timed walk than the 4-AP group taking 10 mg twice daily. Dist. Ct. Op. at *9. 7 After the study or III. See J.A. 870; U.S. Dep’t of Health & Human Servs., U.S. Food & Drug Admin., The FDA’s Drug Review Process: Ensuring Drugs Are Safe and Effective (2017), https://www.fda.gov/drugs/resourcesforyou/consumers/uc m143534.htm. 7 During oral argument, counsel for Acorda repeat- edly noted the result that the placebo group actually outperformed the 10 mg twice-daily group in 3 of the 7 weeks. E.g., Oral Arg. at 8:57–9:20; id. at 10:05–20. But Acorda has not shown where that result was pub- lished in the prior art. See Sept. 23, 2016 Trial Tr. at 785, Acorda Therapeutics, Inc. v. Alkem Labs. Ltd., No. 1:14- cv-00882-LPS (D. Del. Oct. 21, 2016), ECF No. 269 (coun- sel for Acorda stating at trial that the MS-F201 data was not publicly available prior art, other than the data reported in the Goodman references). On this record, that result could not have informed the legally relevant person of skill in the art about whether to expect (or, as Acorda argues, not to expect) the 10 mg twice-daily dose to suc- ceed in improving walking. 18 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. was completed, Acorda conducted a post-hoc analysis of the data on walking speed—which, unlike timed 25-foot walk, was not an endpoint the study was designed to test—and identified a statistically significant difference between the placebo and 4-AP groups considering all doses in the aggregate. Id. Most but not all of the just-described results of the MS-F201 study were published. Dr. Goodman published two nearly identical abstracts in early 2003 (Goodman I, J.A. 6371–72, and Goodman II, J.A. 6370) and presented a poster in connection with those abstracts in late 2002 (Goodman Poster, J.A. 6497–504). Goodman I explains that “[t]he primary aim” of the randomized, placebo- controlled, double-blinded Phase II dose-ranging study was to “determine the safety and tolerability of escalating doses of a sustained release (SR) formulation [of 4-AP], given orally to patients with [multiple sclerosis],” and that “[t]he secondary aim was to explore efficacy over a broad dose range using measures of fatigue and motor function.” J.A. 6371; see Dist. Ct. Op. at *14. The ab- stract discloses that the study involved 36 patients, 25 in the active-treatment and 11 in the placebo group, and that the active-treatment group received 20 mg/day 4-AP, with doses escalating 10 mg/day to reach a maximum of 80 mg/day during week 8 of the study. J.A. 6371–72; see Dist. Ct. Op. at *14. In the “Results” section, Goodman I reports that five subjects withdrew as a result of adverse effects, including two seizures, and that adverse effects were “more severe at doses of 50 mg/day and higher,” including the two seizures that occurred at doses of 60 and 70 mg/day. J.A. 6372; see Dist. Ct. Op. at *14. An- other reported result is that the 4-AP sustained-release treatment “group showed statistically significant im- provement from baseline compared to placebo in function- al measures of mobility (timed 25 walking speed; p=0.04) and lower extremity strength (manual muscle testing; p=0.01). Dose-response curves showed increasing benefit ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 19 in both measures in the 20 to 50 mg/day range.” J.A. 6372; see Dist. Ct. Op. at *14. The abstract clarifies that “[n]o other measures showed significant treatment effects.” J.A. 6372; see Dist. Ct. Op. at *14. The “Conclu- sions” section reads: The safety profile of [4-AP sustained-release] was consistent with previous experience. Doses above 50 mg [per day] added little benefit and increased adverse effects. There was significant improve- ment in measures of mobility and muscle strength. J.A. 6372. The Goodman Poster is similar. It reproduces almost all of the material in Goodman I in the “Abstract” section at the upper-left-hand corner of the poster. J.A. 6502 (capitalization altered). The Poster contains more detail in the “Background” section, which notes that “[r]ecent clinical studies have indicated that [4-AP] promotes improvement in motor strength, walking, fatigue, and endurance in people with [multiple sclerosis]”; that ob- served adverse events, including seizures, were associated with higher peak plasma concentrations and rapid plasma concentration changes caused by immediate-release 4-AP; and that sustained-released formulations were developed to address those problems. Id. (capitalization altered). The study objectives were defined as: (1) “[d]etermine safety of multiple doses of [sustained-release 4-AP] (one week each of 20 mg/day, 30 mg/day, 40 mg/day, 50 mg/day, 60 mg/day, 70 mg/day, and 80mg/day)”; and (2) “[o]btain evidence of efficacy and dose-response using several outcome measures.” Id.; accord id. (Methods section). The Goodman Poster notes that, because indi- viduals taking 4-AP “frequently report” improvements in activity and fatigue levels, the study focused on outcomes associated with such effects—namely, timed ambulation, manual muscle testing, and patients’ self-reports of 20 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. fatigue—rather than the EDSS, because “it was not clear whether” the EDSS “would adequately reflect this type of improvement.” Id. (Methods section). As to the study’s results concerning safety, the Good- man Poster provides, in the “Results Summary,” that “more severe adverse events,” including seizures, occurred “[a]t doses above 40 mg/day.” J.A. 6504 (capitalization altered). The Poster states that “the risk of seizure re- quires further study and characterization[,] particularly in the anticipated dose range.” Id. As to the results concerning efficacy, the Goodman Poster includes a graph of a dose-response curve for the 25-foot walk: J.A. 6503. The graph shows that the total time for the walk decreased significantly between the placebo dose (run-in) and the 20 mg/day dose. Id. The total time seems to have plateaued at higher doses. Id. (total time remained between approximately 12.5 and 14 seconds as doses increased from 20 mg/day to 80 mg/day); see also Sept. 19, 2016 Trial Tr. at 102–03, 137 (testimony of defendants’ expert Dr. Peroutka, observing a walk time ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 21 between 12 and 14 seconds for a “stable clinical effect at 20 to 40” mg/day in the “flat part of the dose response curve”). The results section also provides bar graphs showing changes in individual patients’ speed on the 25-foot walk. J.A. 6503. The upper bar graph shows, on average, im- provements in speed for patients in the active-treatment group, aggregated for doses ranging from 20–50 mg/day. Id.; see J.A. 416. It appears that a few of those patients’ speed decreased by approximately 0–10%, while more than a dozen patients’ speed increased by more than 10%—nine by more than 20%, four by more than 40%, and one by more than 60%. J.A. 6503. The lower bar graph shows, on average, zero or slight improvement in speed for patients in the placebo group, with no patient’s speed having improved by more than 20% and one pa- tient’s speed having decreased by more than 20%. Id. 22 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. The results for improvements in leg strength between the active-treatment group (aggregating the doses of 20– 50 mg/day) and placebo group showed a similar trend: Id. In the “Results Summary,” the Goodman Poster states that “[s]ignificant improvement in walking speed was observed in the [4-AP sustained-release] treated group (p=0.04*),” where the p-value reflects a “*repeated meas- ure ANOVA (weeks 1–7)”—i.e., the walking speed for the active-treatment group, aggregating the dose levels. J.A. 6504; see Dist. Ct. Op. at *14 n.11 (noting that Dr. Goodman explained that the p-value reflects “the aggre- gated value for the treatment group as a whole, including all dosages, and did not reflect the results associated with any single dosage” (emphasis omitted)). More specifical- ly, the Goodman Poster reports that (1) “[t]he average improvement in walking speed [in the 25-foot walk] during the low dose period (20–50 mg/day) included > 20% increase for 9 of the 25 subjects” and (2) “[c]hanges in the placebo-treated group were equally distributed between ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 23 increases and decreases in walking speed and none of the 11 subjects showed increases > 18% during the low dose period.” J.A. 6504. The Poster also reports, for the lower extremity manual muscle test (LEMMT), a “[s]tatistically significant improvement in the [4-AP sustained-release] treated group (p=0.01*).” Id. The Conclusions section contains six bullet points. The first states that the “[s]afety profile [is] consistent with previous experience.” J.A. 6503. The next few bullet points report a “[s]ignificant benefit on timed walking,” “[s]ignificant benefit on lower extremity strength,” “[n]o evidence of benefit on overall fatigue—susceptibility of fatigue to placebo effect,” and “[e]vidence of dose-response in 20–40 mg/day range.” Id. Finally, there was “[l]ittle added benefit, and increased [adverse events,] at doses above 50 mg/day.” Id. This Goodman prior art—which post-dates Elan’s transfer of the research project to Acorda and which added significantly to the teachings of the earlier prior art—became the most important prior art in the obvious- ness analysis in this case. b In 2003, after completion of the MS-F201 study, Acor- da conducted another placebo-controlled Phase II study (MS-F202 study) to test 4-AP’s effect on walking speed. Dist. Ct. Op. at *9. After a two-week up-titration period beginning with a 10 mg dose, patients were administered a stable dose of 10 mg, 15 mg, or 20 mg sustained-release 4-AP twice daily for twelve weeks. Id. Although none of the 4-AP groups demonstrated a statistically significant improvement in walking speed relative to placebo, anoth- er post-hoc analysis showed that responders were in the 4-AP group (p < 0.0001) and that there was no meaningful difference in efficacy among the tested 4-AP doses. Id.; see also J.A. 612–14 (Acorda founder Dr. Cohen explain- ing that isolating responders in the study—those patients 24 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. with improved walking—showed that responders were overwhelmingly in the active treatment groups and that there was no meaningful difference in efficacy among the responders in those treatment groups taking 10 mg, 15 mg, or 20 mg twice daily). Acorda then conducted two Phase III studies to evalu- ate the effect of 10 mg sustained-release 4-AP twice daily, with walking improvement responder analysis as the primary outcome measure. Id. Both studies were suc- cessful, with p < 0.0001. Id. Neither the results of the MS-F202 study nor the re- sults of the Phase III studies constitute publicly available prior art to the Acorda patents in this case. 4 On April 9, 2004, Acorda employees filed a provisional patent application; that date is undisputedly the priority date of the Acorda patents. Id. at *9 n.8. The Acorda patents issued between August 2011 and March 2014. The parties treat the Acorda patents’ claims, for pur- poses of the invalidity issue on appeal, as involving meth- ods of administering to a patient with multiple sclerosis a sustained-release 4-AP formulation (1) in a 10 mg dose twice daily, (2) at that stable dose for the entire treatment period of at least two weeks, (3) maintaining 4-AP serum levels of 15–35 ng/ml, (4) with walking improved. The parties treat claim 7 of the ’826 patent and claim 22 of the ’437 patent as representative. Claim 7 of the ’826 patent depends on claim 6, which reads: 6. A dosing regimen method for providing a 4- aminopyridine at a therapeutically effective con- centration in order to improve walking in a hu- man with multiple sclerosis in need thereof, said method comprising: ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 25 initiating administration of 4-aminopyridine by orally administering to said human a sustained release composition of 10 milligrams of 4- aminopyridine twice daily for a day without a pri- or period of 4-aminopyridine titration, and then, maintaining administration of 4- aminopyridine by orally administering to said human a sustained release composition of 10 mil- ligrams of 4-aminopyridine twice daily; without a subsequent period of 4-aminopyridine titration, whereby an in vivo CmaxSS:CminSS ratio of 1.0 to 3.5 and a CavSS of 15 ng/ml to 35 ng/ml are main- tained in the human. ’826 patent, col. 27, lines 41–57. Claim 7 covers “[t]he method of claim 6, whereby an increase in walking speed is obtained in said human.” Id., col. 27, lines 58–59. Claim 22 of the ’437 patent depends on claim 18, which depends on claim 1. Claim 1 of the ’437 patent reads: 1. A method of increasing walking speed in a human multiple sclerosis patient in need thereof comprising orally administering to said patient a sustained release composition of 10 milligrams of 4-aminopyridine twice daily for a time period of at least two weeks, wherein said 10 milligrams of 4- aminopyridine twice daily are the only doses of 4- aminopyridine administered to said patient dur- ing said time period. ’437 patent, col. 27, lines 55–61. Claim 18 requires that the sustained release composition in claim 1 be “a tablet,” id., col. 28, lines 47–48; and claim 22 requires that the tablet of claim 18 “exhibit[] a release profile to obtain a CavSS of about 15 ng/ml to about 35 ng/ml,” id., col. 28, 26 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. lines 55–57. The parties have not distinguished the claims for purposes of the invalidity issue before us. 8 5 Acorda submitted New Drug Application No. 022250 to the FDA for the use of 10 mg 4-AP extended-release tablets (Ampyra). The FDA granted priority review to that application and approved it on January 22, 2010. According to the approved FDA label, Ampyra “‘is in- dicated as a treatment to improve walking in patients with multiple sclerosis (MS). This was demonstrated by an increase in walking speed.’” Dist. Ct. Op. at *4 (cita- tion omitted). “Improvement in walking in MS patients is [the] only approved use” of Ampyra. Id. The “Descrip- tion” section of the label states that “‘Ampyra (dal- fampridine) is a potassium channel blocker, available in a 10 mg tablet strength . . . , formulated as an extended release tablet for twice-daily oral administration.’” Id. (capitalization altered). The “Dosage and Administration” section explains that “‘[t]he maximum recommended dose of Ampyra is one 10 mg tablet twice daily, taken with or without food, and should not be exceeded. . . . No addi- tional benefit was demonstrated at doses greater than 10 mg twice daily and adverse reactions and discontinua- tions because of adverse reactions were more frequent at higher doses.’” Id. (capitalization altered). Between the time of FDA approval in 2010 and the end of 2015, total sales of Ampyra were $1.7 billion and net income was $998.7 million. Id. at *16. Net sales of 8 Although the ’826 patent’s claim 7 does not re- quire a regimen of at least two weeks, asserted claim 39 does (claim 39 requires 12 weeks), as do the ’437 patent’s asserted claims 1, 2, 5, 22, 32, 36, and 37; the ’685 pa- tent’s asserted claims 3 and 5; and the ’703 patent’s asserted claims 36, 38, and 45. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 27 Ampyra, in dollars, increased at an average rate of 20% per year, and the volume of tablets sold increased at an average rate of 8% per year, despite an increasing price per tablet over that period (2010 to 2015). Id. Acorda also receives royalty payments from licenses to sell Ampy- ra outside the United States; it has collected at least $135 million from those licenses. Id. Commercial opportunity, however, is constrained be- cause Ampyra is indicated only for improvement of walk- ing. Id. at *16–17. Ampyra sales revenue is approximately 2–3% of the total sales revenue from the top ten multiple sclerosis drugs. Id. at *17. Not all multiple sclerosis patients respond to Ampyra. Among multiple sclerosis patients who experience walking diffi- culties, 15–20% of those patients are prescribed Ampyra. Id. On the other hand, Ampyra is the first and only drug approved for improving walking in multiple sclerosis patients. Id. When Sanofi-Aventis in 2008 conducted a Phase III study to test whether a different potassium- channel blocker, nerispirdine, would improve walking in patients with multiple sclerosis, it did not find evidence of a “specific significant difference between the responders [and] non-responders that received nerispirdine or place- bo” in a timed 25-foot walk. J.A. 726–28 (testimony of Acorda’s expert Dr. Fred Lublin); see Dist. Ct. Op. at *17. B In 2014, the defendants notified Acorda and Alkermes of the defendants’ submission of Abbreviated New Drug Applications seeking FDA approval to market generic versions of Ampyra. In mid-July 2014, Acorda and Alk- ermes filed suits against Roxane, Mylan, and Teva, among others, in the District of Delaware for the alleged infringement of several claims in each of the Elan and Acorda patents under 35 U.S.C. § 271(e). The cases were consolidated in 2015. 28 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. The defendants stipulated to infringement of the as- serted claims—claims 3 and 8 of the Elan patent; claims 1, 7, 38, and 39 of the ’826 patent; claims 3 and 5 of the ’685 patent; claims 1, 2, 5, 22, 32, 36, and 37 of the ’437 patent; and claims 36, 38, and 45 of the ’703 patent. Dist. Ct. Op. at *9–12, *18. The defendants, however, chal- lenged the validity of the asserted claims of all five pa- tents for obviousness under 35 U.S.C. § 103. 9 The defendants also challenged the validity of the asserted claims of the Elan patent for insufficient written descrip- tion and enablement under 35 U.S.C. § 112, ¶ 1. After a bench trial held in September 2016, the district court determined that the defendants had not proven invalidity of the Elan patent. Dist. Ct. Op. at *20–29. But the court held that the defendants had proven that the asserted claims of the Acorda patents are invalid for obviousness. Id. at *29–41. As to the Acorda patents: Based on the publications discussed above, as well as expert testimony, the court found that, as of 2004 (the priority date), a relevant skilled artisan would have been motivated to administer a stable dose of 10 mg of 4-AP twice daily and had a reasonable expectation of success in the objective of improving the walking ability of multiple sclerosis patients. Id. at *30–35. The court also found that the Acorda patents’ claim limitations regarding serum levels (the pharmacokinetic limitations) were inherent in the dosing claimed. Id. at *35–36. Finally, the court, while finding certain facts in Acorda’s favor regarding objective indicia of obviousness, ultimately discounted such indicia, relying on the fact that the Elan 9 Because the effective filing date of the claims of the Acorda patents are before March 16, 2013, the version of § 103 preceding the enactment of the Leahy-Smith America Invents Act, Pub. L. No. 112-29, 125 Stat. 284 (2011), governs this case. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 29 patent was a “blocking patent” for the claimed methods of the Acorda patents: any marketer of a drug for uses practicing those methods would need a license to the Elan patent—to which Acorda, for years preceding the 2004 priority date, had an exclusive license from Elan. Id. at *36–40. 10 On April 25, 2017, the court entered final judgment in favor of the defendants as to the Acorda patents and in favor of Acorda as to the Elan patent. The court set the effective date of any final FDA approval of the defendants’ Abbreviated New Drug Applications no earlier than the expiration date of the Elan patent—July 30, 2018—and enjoined the defendants from any infringing activity before that date. Acorda and the defendants timely appealed and cross- appealed, respectively. We have jurisdiction under 28 U.S.C. § 1295(a)(1). II Acorda makes essentially three arguments on appeal regarding the district court’s ruling that the Acorda patent claims are invalid for obviousness. First, Acorda contends, on a number of grounds, that the district court 10 In inter partes reviews initiated by a petitioner not included among the defendants here, the Patent Trial and Appeal Board considered challenges to the Acorda patents that did not involve Schwid or the Goodman references but, instead, depended on whether a particular filing with the Securities and Exchange Commission was prior art to the patents. The Board concluded that it was not. Coalition for Affordable Drugs (ADROCA) LLC v. Acorda Therapeutics, Inc., Nos. IPR2015-01850, -01853, - 01857, -01858, 2017 WL 950736, at *9–20 (P.T.A.B. Mar. 9, 2017). That ruling does not change the analysis in this case. 30 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. erred in finding that a person of skill would have had a motivation to combine the prior art to arrive at the Acor- da invention and a reasonable expectation of success in doing so. Second, Acorda challenges the court’s determi- nation that the claim limitations relating to pharmacoki- netics—i.e., achieving 4-AP serum levels of 15–35 ng/ml— are inherent in the claimed invention and therefore obvious. Third, Acorda argues that the court improperly applied a categorical rule that a blocking patent (the Elan patent) negates any findings in favor of Acorda on the objective indicia of commercial success, failure of others, and long felt but unmet need. 11 Under 35 U.S.C. § 103(a), obviousness is a question of law based on underlying questions of fact, including the level of ordinary skill in the art, the scope and content of the prior art, the differences between the claims and the prior art, motivation to modify or combine with a reason- able expectation of success, and objective indicia of non- obviousness. See KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406 (2007); In re Stepan, 868 F.3d 1342, 1345–46 (Fed. Cir. 2017); PAR Pharm., Inc. v. TWI Pharm., Inc., 773 F.3d 1186, 1193–94, 1196–97 (Fed. Cir. 2014). We review the district court’s determination of obviousness de 11 Acorda also argues that the district court failed to analyze the claimed inventions as a whole. We see no methodological error. The court did nothing other than follow the parties’ own breakdown of what aspects of the claimed inventions, alone or together, a skilled artisan at the priority date would have been motivated to adopt with a reasonable expectation of success and, more generally, would have found obvious. The court did not overlook any meaningful argument by Acorda that certain aggregations of claim elements, including the whole, required analysis beyond the analysis of the walking-benefit, dosage, stabil- ity, and serum-level aspects of the claims. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 31 novo and its underlying factual findings for clear error. In re Cyclobenzaprine Hydrochloride Extended-Release Capsule Patent Litig., 676 F.3d 1063, 1069 (Fed. Cir. 2012). A Acorda challenges the district court’s findings about the relevant skilled artisan’s motivations and expecta- tions regarding the administration of a stable 10 mg 4-AP dose twice daily to improve walking. It presents two relatively focused arguments: that Schwid teaches away from the claimed invention; and that the prior art teaches the administration of sustained-release 4-AP in a titrated- dosing regimen rather than a stable-dosing regimen. More broadly, Acorda argues that neither the Goodman Poster nor the prior art collectively teaches the efficacy of a stable 10 mg twice-daily dose or indicates that such a dose is among the small number of options that a skilled artisan would have been motivated to test with a reason- able expectation of success to improve walking. We reject these challenges. 1 Acorda contends that Schwid “affirmatively teaches away from Acorda’s invention.” Acorda Br. 36. The district court considered Schwid, as Acorda urged, among the teachings of the overall art available at the 2004 priority date, and it made findings as to the motivation and expectations of a relevant skilled artisan at that date regarding a stable 10 mg dosage of 4-AP to improve walking. Dist. Ct. Op. at *30–31. Acorda has not shown that Schwid renders the court’s findings on those issues clearly erroneous. Schwid supports a motivation to test, with a reasona- ble expectation of success, a 10 mg twice-daily dose of sustained-release 4-AP to improve walking in multiple sclerosis patients. Schwid itself used a 17.5 mg twice- 32 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. daily dose, but it found success with that dosage: as stated in Schwid, “[t]he results of this double-blind crossover study provide evidence that 4AP [sustained release] had a therapeutic effect on neurologic deficits from [multiple sclerosis].” J.A. 6684. In particular, there was a statisti- cally significant improvement for the 17.5 mg 4-AP versus placebo in timed gait (i.e., in walking ability); and the improvements in other outcomes, while not statistically significant, “showed trends favoring 4AP [sustained release].” J.A. 6681, 6684. Schwid expressly concludes that the study shows “4AP [sustained release] improved motor function in [multiple sclerosis] patients.” J.A. 6681. And, stressing toxicity concerns with high doses, Schwid provides affirmative reason to investigate low doses. See J.A. 6681 (“4AP can provoke seizures and acute encepha- lopathy”—episodes that “tend to occur when serum 4AP levels peak, suggesting that lower peak levels may in- crease safety.”); J.A. 6684 (“[F]uture studies of 4AP [sus- tained release] will need to examine long-term efficacy and tolerability as well as further refine dosing regimens to optimize delivery despite a relatively narrow therapeu- tic window.”). Schwid makes certain observations that its study showed favorable results in some outcome measures at high serum levels of 4-AP (60 ng/ml)—levels that, accord- ing to evidence emphasized by Acorda, may require the administration of 4-AP doses higher than 10 mg twice a day. See J.A. 445–48 (defendant’s expert’s testimony that 17.5 mg twice-daily or 25 mg twice-daily could result in serum levels at or above 60 ng/ml); J.A. 823 (Acorda’s expert’s testimony: similar). But Acorda overstates the significance of this serum-level observation to the issue of a reasonable expectation of success for walking improve- ment. Schwid found no statistically significant difference be- tween the 4-AP and placebo groups as to patients’ subjec- tive global impression of their condition, one of seven ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 33 outcome measures in the Schwid study. J.A. 6683. As to that outcome measure, Schwid states that “[n]one of the patients with a serum level less than 60 ng/mL felt better (according to their global impressions) on 4AP [sustained release] than placebo.” Id. But efficacy in patients’ global impression is not the issue—efficacy in timed gait is. Schwid made no such finding as to timed gait. Schwid also observes, as a general matter, that “[t]reatment [with 4AP sustained release] appeared particularly efficacious in subjects who achieved serum 4AP levels above 60 ng/mL, with everyone improving in timed-gait testing and grip strength, and five of six improving by MVICT [maxi- mum voluntary isometric contraction, measured quantita- tively] and their own subjective assessment [global impression].” J.A. 6684. But Schwid’s measured im- provement in timed gait was not limited to patients with high serum levels. See J.A. 6683 (9 of 10 patients im- proved in timed gait, and only 6 patients achieved serum levels greater than 60 ng/ml). In short, high serum levels were not required, and a dose of 17.5 mg sustained-release 4-AP twice-daily was sufficient, for improvement in timed gait in Schwid. Meanwhile, Acorda has pointed to nothing in Schwid declaring that doses lower than 17.5 mg twice-daily would not be effective in improving walking. Schwid therefore supports a finding that a person of skill would have had a reasonable expectation of success regarding the admin- istration of 17.5 mg of 4-AP twice-daily—or perhaps even a lower dose since 17.5 mg was sufficient—to improve walking in multiple sclerosis patients. And in light of Schwid’s warning that seizures may occur at higher doses, the district court did not clearly err in finding that a person of skill would look to lower doses rather than higher ones. See Dist. Ct. Op. at *32 (“While the prior art may have generally suggested that 4-AP would be more effective in higher doses, the art also reduced the set of 34 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. plausible doses because it suggested that higher doses of 4-AP were more likely to cause adverse events.”). 2 Acorda’s second argument is that the prior art teaches administering sustained-release 4-AP only in a titrated- dosing regimen to avoid the risk of seizure, and therefore that the district court could not properly find that a person of skill would have been motivated to pursue, or had a reasonable expectation of success concerning, a stable-dosing regimen. We reject this argument. The prior art is not limited to titrated dosing (where doses start low and move higher) but rather contains evidence of stable dosing (where the dose starts and stays at the claimed level). As the district court noted, Polman is evidence of safe and effective long-term oral admin- istration of a stable dose of immediate-release 4-AP. Dist. Ct. Op. at *34; see J.A. 6655. Schwid also provides evi- dence of a stable-dosing regimen of 4-AP, if only for a week. As for the studies that used escalating doses, some of those studies began with 10 mg as the lowest dose before titrating upwards to doses that may increase the risk of seizure. E.g., Davis at 187 tbl.1; see also Dist. Ct. Op. at *8 (1994 Elan study began with 12.5 mg 4-AP twice daily); id. at *9 (10 mg twice daily was the lowest dose used in the Acorda MS-F202 study); cf. J.A. 6647 (trial in patients with other conditions began with dose of 10 mg 4- AP twice daily and titrated up to 200 mg daily); J.A. 6434 (Acorda’s trial in patients with spinal cord injury began with 10 mg twice daily as the lowest dose). Significantly, the most important prior art, the Goodman references, report a start dose of 10 mg twice daily. J.A. 6370, 6372, 6502. Even if many earlier studies used a titrated-dosing scheme to avoid adverse effects caused by starting at higher doses, those studies do not, as the district court found, undermine the other evidence in the prior art that ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 35 a person of skill would have a reasonable expectation of success for a stable-dosing scheme at low doses. Dist. Ct. Op. at *34. The Bever II prior-art review article reports that while “seizures are common at higher doses,” 4-AP “rarely cause[s] seizures at the doses used in [multiple sclerosis] trials.” Bever II at S120. Other published studies say the same: seizures were seen at higher doses, but not lower ones like 10 mg. E.g., J.A. 6651 (trial in patients with Eaton-Lambert syndrome, congenital myas- thenia, and myasthenia gravis starting at dose of 10 mg 4-AP twice daily and escalating to 200 mg daily found that all of the patients who experienced seizures during the study “were receiving 80 mg or more of 4-AP daily”); J.A. 6504 (Goodman Poster “Results Summary”: “At doses above 40 mg/day, more severe adverse events were re- ported, including two cases of seizure (at 60 and 70 mg/day)”). And in Schwid, the authors advise that future studies pursue lower doses for long-term tolerability. See J.A. 6681 (“4AP can provoke seizures and acute encepha- lopathy,” but those episodes “tend to occur when serum 4AP levels peak, suggesting that lower peak levels may increase safety.”); J.A. 6684 (“[F]uture studies of 4AP [sustained release] will need to examine long-term effica- cy and tolerability as well as further refine dosing regi- mens to optimize delivery despite a relatively narrow therapeutic window.”). Expert testimony supports the district court’s finding that a person of ordinary skill in the art would have been motivated to pursue, and had a reasonable expectation of success in pursuing, a stable-dosing regimen of 10 mg 4- AP twice daily. According to Dr. Peroutka, “the general goal of drug development [is] to provide a stable dosing regimen.” J.A. 414. He testified that stable dosing was particularly desirable for treating multiple sclerosis because, as a chronic disease that requires long-term treatment, a stable oral dose is much easier to administer. See Sept. 19, 2016 Trial Tr. 110 (“Obviously, it’s a lot 36 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. easier simply to take one pill, the same pill twice a day than to have to figure out, well, this morning I need this much, that much. But with pills, it is almost impossible to titrate easily.”). Even Dr. Goodman conceded that “it would be desirable” to have a stable-dosing regimen where “the patient would be prescribed [some dose] to take on a regular basis.” J.A. 868. And titration was not required given such a low starting dose: Acorda founder Dr. Cohen testified that, upon recognizing the efficacy of the 10 mg twice-daily dose, “we realized we didn’t have to titrate anymore.” J.A. 614. Finally, Dr. Peroutka ex- plained that nothing in the prior art suggested that 4-AP could not be used for long-term treatment for a chronic condition. Sept. 19, 2016 Trial Tr. 104. 3 Acorda’s most general argument is that the district court improperly found that a relevant skilled artisan “would have formed a reasonable expectation of success based on Schwid and Goodman [in particular, the Good- man Poster], in light of the totality of the prior art,” regarding a 10 mg twice-daily dose of 4-AP to improve walking. Dist. Ct. Op. at *31. We reject Acorda’s argu- ment. As described above, Schwid reports a statistically sig- nificant improvement in timed gait for patients given 17.5 mg 4-AP twice-daily versus placebo. Also as de- scribed above, the Goodman Poster reports a statistically significant improvement in walking speed and in lower extremity strength for patients given 10–40 mg 4-AP twice daily versus placebo; an average improvement in walking speed during the low-dose period (10–25 mg 4-AP twice daily) of more than 20% for 9 of 25 subjects; and “more severe adverse events,” including seizures, at doses above 20 mg 4-AP twice daily. J.A. 6504. The Goodman Poster also reports a dose response in the timed walk at doses in the range of 10–20 mg 4-AP twice daily. See Dist. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 37 Ct. Op. at *33 (“Goodman states that the results showed ‘evidence of a dose response in the 20 to 40 milligram per day range,’ indicating that patients taking these dosages of 4-AP demonstrated a greater response to treatment than did patients receiving placebo.”). The district court did not clearly err in finding that a person of skill would have looked to both of those refer- ences, considered their limits, and had a reasonable expectation of success as to the efficacy of 10–20 mg 4-AP twice daily to improve walking. Despite certain identified “shortcomings” in the principal references, “the combined message a [person of skill in the art] would have dis- cerned from Schwid together with the Goodman refer- ences was a reasonable expectation of success in treating walking with 4-AP.” Id. at *31. Other prior art was consistent with that message. Id. As to dosages, the disclosures of Schwid and the Goodman Poster regarding relevant benefits at doses including or near to the Acorda- claimed range (recounted above), together with the re- ported concerns about high doses, support the further finding that a relevant skilled artisan would have “con- sider[ed] 10 mg/twice daily to be among the finite group of doses of sustained-release 4-AP that could reasonably be expected to improve walking in MS patients.” Id. at *33 (footnote attached citing further partial support from testimony of Acorda’s Dr. Goodman). In a finding reflect- ing both motivation and reasonable expectation of suc- cess, the district court stated: “As the lowest of the range of encouraging doses, 10mg/twice daily would have been an attractive starting point for a [person of skill in the art].” Id. These findings not only have adequate eviden- tiary support but comport with the guidance of KSR to “take account of the inferences and creative steps that a 38 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. person of ordinary skill in the art would employ.” 550 U.S. at 418. 12 Expert testimony further supports the district court’s findings. The defendants’ expert Dr. Peroutka explained that Schwid showed that the claimed formulation was effective at a 17.5 mg twice-daily dose and that the result was statistically significant. J.A. 406–07, 410. Dr. Peroutka also stated that the Goodman abstracts “said that dose response curves showed an increasing benefit in both measures in the 20 to 50 milligram a day range [10– 25 mg twice-daily range], meaning timed walking or lower extremity strength.” J.A. 414. According to Dr. Peroutka, the study presented in the Goodman abstracts was a dose- ranging study where “the goal” is “to find the most effica- cious dose without adverse events.” Id.; accord J.A. 869 (Acorda’s expert Dr. Goodman: “[W]hat we really want to find is the most effective dose that can be given safely.”). The additional information provided in the bar graph in the Goodman Poster showed that people taking 10–25 mg twice daily did better in walking speed than placebo, and the dose-response curve showed improvement in walking speed at the 10 mg twice-daily dose—a level of improve- ment that was maintained at higher doses. See J.A. 416; Sept. 19, 2016 Trial Tr. 102 (“They got the 10 milligrams to work at this level and that level of efficacy was main- tained through the dose ranges.”); id. at 103 (“[I]t’s cer- tain stable clinical effect at 20 to 40” milligrams per day 12 In its formulation describing the narrow set of choices facing the relevant artisan in 2004 in this case, the district court quoted KSR’s discussion of obviousness where the claimed invention was “obvious to try.” Dist. Ct. Op. at *32 (quoting 550 U.S. at 421). But the court fully applied the familiar standards focused on the rele- vant artisan’s motivation to make the claim-required combinations with a reasonable expectation of success. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 39 (doses of 10 mg, 15 mg, and 20 mg twice-daily).). Dr. Peroutka testified that he would have included the 10 mg dose in a Phase III study because there are “very serious” side effects at higher doses so “you would take the lowest effective dose that was safe.” Id. at 104. He also testified that a person of skill might even want to try a lower dose, but “based on the [Goodman] data, 10 is the lowest effec- tive dose.” Id. Acorda’s expert Dr. Goodman himself stated that the Goodman Poster “suggest[s]” “that the range for further testing would be the 20 to 40 milligrams per day [10 to 20 mg twice-daily] range.” J.A. 844–45; see also J.A. 874 (Dr. Goodman stating during his deposition that “a person of ordinary skill in the art in December 2003 would have been motivated based on the 201 study to design a study along the lines of what became the 202 study,” which tested the 10 mg twice-daily dose). Ulti- mately, the court found, based on the prior art and expert testimony, that a person of skill before the 2004 priority date would have looked (1) to the 10–20 mg twice-daily dose range for effective doses that would be reasonably expected to improve walking in multiple sclerosis patients and (2) to the low end of that range to avoid adverse effects. Dist. Ct. Op. at *32–33. Acorda’s core argument appears not to be that the ev- idence fails to support the finding of a motivation to combine. Rather, it appears to be that the evidence cannot support a finding of a reasonable expectation of success (in 2004) in the absence of publications showing a statistically significant difference in walking tests be- tween the specific dose of 10 mg 4-AP taken twice daily versus placebo. See Acorda Br. 41–42; Acorda Reply Br. 20–21; Oral Arg. at 6:10–30. We reject this contention. To the extent that Acorda’s contention is a legal one, asserting a law-required minimum for what can support a “reasonable” expectation of success, Acorda has offered no support for the contention. This court has long rejected a requirement of “[c]onclusive proof of efficacy” for obvious- 40 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. ness. See, e.g., Hoffman-La Roche Inc. v. Apotex Inc., 748 F.3d 1326, 1331 (Fed. Cir. 2014); PharmaStem Therapeu- tics, Inc. v. ViaCell, Inc., 491 F.3d 1342, 1364 (Fed. Cir. 2007); Pfizer, Inc. v. Apotex, Inc., 480 F.3d 1348, 1364, 1367–68 (Fed. Cir. 2007) (reasoning that “the expectation of success need only be reasonable, not absolute”). And Acorda has cited no authority from the Supreme Court or this court requiring as a matter of law, for reasonableness of an expectation of success, testing of specific doses versus placebo that shows the relevant result with statis- tical significance. Acorda has furnished no basis for treating the question in this case as anything but one of context-specific fact based on evidence. In some cases, of course, the evidentiary basis for an inference of reasonable expectation of success may be inadequate. See, e.g., In re Cyclobenzaprine, 676 F.3d at 1070–71. Here, though, as we have discussed, expert and other evidence indicates that a person of skill in the present context can draw reasonable inferences about the likelihood of success even without a perfectly designed clinical trial showing a statistically significant difference in efficacy between a specific dose and placebo. See also J.A. 6657 (Polman: “Although a placebo effect cannot be excluded, the dynamics of the response in relation to the intake of the medication and the deterioration and subse- quent improvement in functioning during a drug-free interval and subsequent restarting of the therapy are, in our view, highly suggestive of a real effect being induced by the 4-[AP]. Improvements in fatigue and ambulation were mentioned quite often by the patients as being responsible for the favorable overall effect.”). We see no clear error in the district court’s finding to that effect. We are not persuaded by Acorda’s reasons for a con- trary finding. To begin with, “Elan’s failure in the only large-scale and properly statistically powered trial of sustained-release 4-AP that deflated expectations for the drug,” Acorda Reply Br. 28, is not particularly relevant to ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 41 the expectations of success for the Acorda invention. The record shows that the Elan trial was unpublished and is only cursorily discussed in the introduction in Schwid, limiting any “deflat[ing]” effect on expectations in the field. Sept. 19, 2016 Trial Tr. 143–44 (Dr. Peroutka noting that, even in the short discussion of the 1994 Elan study in Schwid, there is very little detail and no mention of the dose of 4-AP that was used). Moreover, the abbre- viated discussion of that trial in Schwid distinguishes the aggregate outcome measure (EDSS) and results in the Elan study from the Schwid study’s measure of particular functionalities (e.g., timed gait). J.A. 6681 (noting the failure of the Elan study but stating that “[t]he EDSS . . . may have been an inadequate outcome variable for this trial,” as EDSS measures several outcomes and could “overlook” significant but lesser improvements in walk- ing). And the 1994 Elan study preceded the successes reported later in Schwid and the Goodman references, which were a sound basis for altering earlier expectations. Similarly, the “inconclusiveness of the exploratory studies of 4-AP, a 102-year old drug,” Acorda Reply Br. 28, does not speak to the more recent research relied on by the district court—namely, Schwid and the Good- man references. And “the rigorous 2003 Solari review of the field dispelling any confidence in using am[ino]pyridines to treat [multiple sclerosis],” id. at 29, does not dispel confidence in a walking improvement; rather, Solari, a prior-art literature review, reports a statistically significant improvement in walking, J.A. 7208 (reviewing three studies that “assessed the efficacy of aminopyridines on ambulation” and reporting that patients who received 4-AP showed a statistically significant improvement in ambulation compared to 42 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. placebo (p<0.0001)). 13 When Acorda asserts that the “prior art’s [Schwid’s] teaching that 4-AP had a narrow therapeutic window where high doses and high blood serum levels were necessary for any meaningful thera- peutic effect,” Acorda Reply Br. 29, Acorda is incorrect, as discussed previously: Schwid reports that a relatively low (17.5 mg twice a day) dose showed a statistically signifi- cant improvement in walking and that high serum levels were not required for improvements in timed gait. Schwid, which reports success and no seizure events with a stable dose of 17.5 mg twice daily, also undermines Acorda’s argument that “the prior art’s consistent use of titration to achieve a therapeutic dose because of seizure risk” conclusively precludes a reasonable expectation of success even for a low dose like 10 mg twice daily that avoids high peak serum levels. Id. In the end, Schwid, Goodman as a whole, and expert testimony supply a sufficient basis for the district court’s finding of a reason- able expectation of success in this case. In light of the record evidence, the district court did not clearly err in finding that a person of skill at the time of the invention would have had a motivation to combine, and a reasonable expectation of success in combining, the teachings of the prior art to arrive at the Acorda invention of a stable regimen of 10 mg twice-daily sustained-release 4-AP to improve walking in multiple sclerosis patients. B Acorda nevertheless contends that a skilled artisan would not have a reasonable expectation of success re- garding the invention of the Acorda patents because the prior art did not teach or suggest a final limitation of the 13 Alessandra Solari et al., Aminopyridines for symp- tomatic treatment in multiple sclerosis (Review), Cochrane Database of Systematic Reviews, Issue 4 (2002). ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 43 asserted claims—the pharmacokinetic limitation, which requires 4-AP serum levels in the 15–35 ng/ml range. E.g., ’826 patent, col. 27, line 29. We disagree. The district court found that the prior art taught that a dose of 10 mg sustained-release 4-AP twice daily would result in serum levels within the range claimed in the Acorda patents. Dist. Ct. Op. at *35. Hayes discloses that when a sustained-release formulation of 4-AP is adminis- tered in a 10 mg dose twice daily, and steady-state condi- tions are reached, the result is a 4-AP average serum level of 20.8 ± 5.7 ng/ml (15.1–26.5 ng/ml, which is within, and in fact covers most of, the Acorda patents’ claimed range). J.A. 6436, 6439 tbl.3. The Hayes study is sum- marized—and Hayes’s table listing the pharmacokinetic results is replicated—in the specifications of two of the Acorda patents. ’826 patent, col. 24, line 25 through col. 25, line 50 (Example 7 and Table 7); ’685 patent, col. 24, line 30 through col. 25, line 54 (Example 7 and Table 7). The district court noted that the parties did not dispute either of two propositions: the Hayes researchers used the Elan formulation that is claimed in the Acorda patents and is now marketed as Ampyra; and the phar- macokinetic results reported in Hayes are inherent prop- erties of that formulation. Dist. Ct. Op. at *35. As discussed in the previous subsections, the district court also found that a person of skill would have been motivat- ed, with a reasonable expectation of success, to administer a dose of 10 mg sustained-release 4-AP twice daily to improve walking in multiple sclerosis patients. Id. at *35–36. Based on those findings, the court invoked the principle that “an obvious formulation cannot become nonobvious simply by administering it to a patient and claiming the resulting serum concentrations,” Santarus, Inc. v. Par Pharm., Inc., 694 F.3d 1344, 1354 (Fed. Cir. 2012), and concluded that the pharmacokinetic limitation could not alter the obviousness analysis. 44 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. On appeal, Acorda does not directly object to the dis- trict court’s inherency finding about Hayes, but Acorda suggests that a person of skill would expect that the inherent pharmacokinetic profiles would differ between patients with spinal cord injury (as in Hayes) and pa- tients with multiple sclerosis (as in the Acorda patents). But Acorda cites no support for that assumption, and Acorda appears to have made the opposite assumption by including the Hayes pharmacokinetic data in its own patents on using 4-AP to treat multiple sclerosis. Acor- da’s expert also admitted at trial that Hayes “may cer- tainly show the pharmacokinetic profile that’s analogous to what would be found in MS [multiple sclerosis] pa- tients. I don’t have any dispute with that.” J.A. 825. The defendants’ expert agreed, testifying that a person of skill would expect the same pharmacokinetic profile in pa- tients with either condition. J.A. 539–40. And while Acorda argues that a person of skill in the art “would have no basis to connect Hayes with [multiple sclerosis] prior art,” Acorda Br. 54, Hayes’s introduction explicitly makes that connection, stating that “[4-AP] is the first compound shown to restore some neurologic function in patients with chronic [spinal cord injury] or other demye- linating conditions such as multiple sclerosis.” J.A. 6433 (internal references omitted). 14 Even if the pharmacokinetic profile is inherent in the 10 mg twice-daily administration of sustained-release 4- AP in Hayes, Acorda complains that a person of skill may not have known the details of the formulation used in 14 Hayes also discloses that the reported study on the pharmacokinetics of sustained-release 4-AP was sponsored by Acorda. J.A. 6433. That disclosure links Hayes to the Goodman references, which also disclose an association with Acorda in a sustained-release 4-AP study. J.A. 6370, 6372, 6498. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 45 Hayes (Ampyra) and therefore would not have known whether the formulation claimed in the Acorda patents would produce the same pharmacokinetic profile. Cf. In re Cyclobenzaprine, 676 F.3d at 1069–71 (obviousness analysis of patent claims to a “therapeutically effective plasma concentration” and to particular pharmacokinetic parameters required a factual finding regarding what a skilled artisan would know about the serum levels needed to produce a therapeutic effect). But Acorda, in response to the district court’s question as to whether the pharma- cokinetic limitation would have been obvious, conceded at trial that a skilled artisan in 2003 would know the phar- macokinetic data for a 10 mg twice-daily dose of sus- tained-release 4-AP. J.A. 1108–09 (counsel for Acorda: “It was known in the art that a sustained-release formulation of 10 [mg] [twice daily] could achieve that PK [pharmaco- kinetic result], not that that PK would yield any efficacy for walking.”). Acorda itself therefore assumed that a person of skill would know that a regimen of 10 mg twice- daily dosing of sustained-release 4-AP—regardless of the specifics of the rest of the formulation—would achieve that pharmacokinetic profile. And, again, Acorda has not pointed to any evidence to contradict that assumption, such as evidence showing that a person of skill would expect another sustained-release formulation containing the same dose of 4-AP to produce a different pharmacoki- netic profile, how that formulation would differ, or how the associated profile would differ. C Acorda’s remaining argument on appeal concerns the proper analysis of objective indicia of nonobviousness in this case. Acorda focuses on the district court’s reliance on the Elan patent as a blocking patent for the Acorda patents’ claimed inventions, in determining that commer- cial success, failure of others, and long-felt but unmet need did not “support” or “militate in favor of” nonobvi- ousness. Dist. Ct. Op. at *39, *40. Acorda characterizes 46 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. the district court as having applied a categorical rule that a blocking patent defeats the significance of such objective indicia to the obviousness determination. We think, however, that the district court’s opinion is best read not as invoking a categorical rule, but as drawing conclusions on the limited factual record created in this case bearing on the effect of a blocking patent. In any event, the court did not err in concluding that the defendants proved obviousness, considering the evidence on objective indicia. 1 A patent has been called a “blocking patent” where practice of a later invention would infringe the earlier patent. The existence of such a blocking patent may deter non-owners and non-licensees from investing the re- sources needed to make, develop, and market such a later, “blocked” invention, because of the risk of infringement liability and associated monetary or injunctive remedies. If the later invention is eventually patented by an owner or licensee of the blocking patent, that potential deterrent effect is relevant to understanding why others had not made, developed, or marketed that “blocked” invention and, hence, to evaluating objective indicia of the obvious- ness of the later patent. See Note, Subtests of “Nonobvi- ousness”: A Nontechnical Approach to Patent Validity, 112 U. Pa. L. Rev. 1169, 1177 (1964) (Regarding commercial success, “a court must be assured that the patentee’s market domination is not attributable to monopoly power or other economic coercion, or to other factors unrelated to patent validity.”) (cited in Graham v. John Deere Co. of Kansas City, 383 U.S. 1, 18, 36 (1966)). We briefly discussed blocking patents in Merck & Co. v. Teva Pharmaceuticals USA, Inc., 395 F.3d 1364 (Fed. Cir. 2005) (Merck I). The Merck patent at issue, applied for in 1998, was for the weekly administration of alendro- nate monosodium trihydrate (Fosamax). Id. at 1366–67. That patent was preceded by Merck’s earlier patent ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 47 (issued in 1986) covering a method of administering an effective amount of Fosamax to treat osteoporosis, as well as Merck’s statutory right, since obtaining FDA approval in 1995, to the exclusive marketing of any dosage strength of Fosamax for the next five years. 395 F.3d at 1367, 1377; Br. for Def.-Appellant Teva Pharm. USA, Inc., Merck I, No. 04-1005, 2003 WL 24307848, at *62–63 (Fed. Cir. Dec. 17, 2003). We ruled that the district court had erred in its analysis of commercial success because the earlier patent and FDA regulatory approval depressed incentives for others to invent the weekly-dosing scheme. 395 F.3d at 1377 (“Because market entry by others was precluded on those bases, the inference of non- obviousness of weekly-dosing, from evidence of commer- cial success, is weak.”). In that context, we said, the evidence of commercial success was “not enough to show the claims at bar are patentably distinct from the weekly- dosing ideas in the [invalidating prior art].” Id. In Galderma Laboratories, L.P. v. Tolmar, Inc., 737 F.3d 731 (Fed. Cir. 2013), we considered the district court’s finding, in support of commercial success, that the FDA-approved product “quickly gained and maintained market share.” Id. at 740. Because earlier patents owned by Galderma may have “blocked” competition to market the FDA-approved product by any entity other than Galderma, we reasoned that the commercial success of the product was “of ‘minimal probative value’” and not suffi- cient to justify a conclusion of nonobviousness in light of the other evidence supporting obviousness. Id. at 741 (quoting Merck I, 395 F.3d at 1376). Recently, in Merck Sharp & Dohme Corp. v. Hospira, Inc., 874 F.3d 724 (Fed. Cir. 2017) (Merck II), we conclud- ed that Merck’s exclusive license to a blocking patent did not, all by itself, justify discounting evidence of commer- cial success. Id. at 730–31. We explained that commer- cial success is “a fact-specific inquiry” that may involve considering the operation of specific blocking patents on 48 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. possible competition. Id. at 731. But the mere existence or sheer number of blocking patents does not, without more, “necessarily detract from evidence of commercial success of a product or process.” Id. Nevertheless, “even giving the evidence of commercial success its full and proper weight,” we affirmed the judgment invalidating the claims at issue for obviousness in light of “the evi- dence that the claimed process was substantially de- scribed in the prior art” and that “merely ordinary experimentation was required to arrive at the [patent at issue].” Id. Merck II’s reasoning reflects a common-sense recogni- tion that, as a theoretical matter, a blocking patent may or may not deter innovation in the blocked space by commercially motivated potential innovators other than the owners or licensees of the blocking patent. 15 Where the owner of the blocking patent or exclusive licensee is different from the owner of the patent in suit, the grant- ing of a license may be a realistic possibility. Even where, as here, the owner of the patent in suit and the exclusive licensee of the blocking patent are the same, such a potential innovator might or might not think it could successfully challenge the blocking patent. And such a potential innovator might or might not be willing to research in the blocked space without a license to a block- ing patent—even if the research itself is within the safe harbor provided by 35 U.S.C. § 271(e)(1)—and wait until it has already developed and patented its aimed-at im- 15 We use the term “blocked space” to refer to what would infringe given the “boundaries,” Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722, 730 (2002), or “metes and bounds,” Brenner v. Manson, 383 U.S. 519, 534 (1966), set by the blocking patent’s claims. See Andrew Corp. v. Gabriel Electronics, Inc., 847 F.2d 819, 823 (Fed. Cir. 1988). ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 49 provement to negotiate for a cross-license with the block- ing patent’s owner to share the profits from the improve- ment. Besides the assessment of whether the blocking patent can be successfully challenged, a number of varia- bles appear generally relevant to the calculus, including: the costliness of the project; the risk of research failure; the nature of improvements that might arise from the project, and whether such improvements will be entirely covered by the blocking patent; the size of the market opportunities anticipated for such improvements; the costs of arriving at the improvements and getting them to market; the risk of losing the invention race to a blocking- patent owner or licensee; the risk that the blocking-patent owner (making its own economic calculations, perhaps in light of its own other products or research activities) will altogether refuse to grant a license to the improvement or will demand so large a share of profits that the whole project is not worthwhile for the potential innovator—all evaluated in light of other investment opportunities. For such reasons, it is clear that, if all other variables are held constant, a blocking patent diminishes possible rewards from a non-owner’s or non-licensee’s investment activity aimed at an invention whose commercial exploita- tion would be infringing, therefore reducing incentives for innovations in the blocked space by non-owners and non- licensees of the blocking patent. Such a blocking patent therefore can be evidence that can discount the signifi- cance of evidence that nobody but the blocking patent’s owners or licensees arrived at, developed, and marketed the invention covered by the later patent at issue in litigation. But the magnitude of the diminution in incen- tive in any context—in particular, whether it was great enough to have actually deterred activity that otherwise would have occurred—is “a fact-specific inquiry.” Merck II, 874 F.3d at 731. That inquiry, conducted within the framework under which the challengers always retain the burden of persuasion on obviousness, may be a difficult 50 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. one as a practical matter. In a particular case, a court may ultimately be left, for its evaluation, with the solid premise of diminished incentives, plus some evidence (possibly weak or ambiguous) about the significance of the deterrence, together with a background sense of the general realities in the area at issue that can affect the weight to be given to the evidence in the specific case. 2 Against this background, we review the district court’s consideration of objective indicia of nonobvious- ness in light of the Elan patent. Acorda licensed the Elan patent in the late 1990s, before the period of commercial success alleged by Acorda and found by the district court. Here, Acorda bore the burden of producing evidence of objective indicia, but the “ultimate burden of proving obviousness” at all times remained with the defendants. Galderma, 737 F.3d at 736–38. We conclude that the district court did not err in viewing the Elan patent, among other evidence, as evidence that discounted the weight of Acorda’s evidence of commercial success, failure of others, and long-felt but unmet need so that “the evi- dence as a whole” in the case “prove[d] clearly and con- vincingly that the Acorda Patents are invalid due to obviousness.” Dist. Ct. Op. at *41. The parties presented evidence on the objective indi- cia of commercial success, failure of others, and long-felt but unmet need. 16 In particular, the defendants present- ed evidence of blocking by the Elan patent. See Dist. Ct. Op. at *38 & n.43 (undisputed that invention of Acorda patents practice the Elan patent). 16 Acorda also presented evidence of unexpected re- sults, but the district court found the evidence unpersua- sive. See Dist. Ct. Op. at *39. Acorda does not appeal that finding. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 51 As to commercial success, the district court found that “no one other than the Elan patentees and their licensees could have practiced the invention of the Acorda patents without facing liability for patent infringement. The risk of such liability would have provided an independent incentive for a patentee not to develop the invention of the Acorda patents, even if those inventions were obvious.” Id. at *38. The district court therefore found that the evidence of commercial success did not support the con- clusion that the Acorda patent claims were non-obvious. Id. at *39. We will interpret the district court’s statements to- gether as referring to domestic marketing of a product. As discussed below, the Elan patent would not preclude practice of the Elan invention outside the United States or under the safe harbor provision of 35 U.S.C. § 271(e)(1) for specified FDA-related activities. The district court’s key finding, therefore, is that “[t]he risk of [infringement] liability” for marketing in the United States “would have provided an independent incentive for a patentee not to develop the invention of the Acorda patents, even if those inventions were obvious.” Dist. Ct. Op. at *38. That finding is supported by the record. The defend- ants offered unrebutted testimony from an expert in economics and pharmaceuticals that the Elan patent acted as a blocking patent for entities other than Acorda (the exclusive licensee to the Elan patent) that wanted to pursue commercial opportunities like Ampyra. J.A. 965– 66 (“[O]ther entities that might want to pursue commer- cial opportunity like Ampyra . . . would not have access to [the sustained-release 4-AP formulation claimed in the Elan patent] because Acorda has that exclusive license.”). The Elan patent issued in 1996 and was licensed exclu- sively to Acorda in 1997 for spinal cord injury and in 1998 for multiple sclerosis treatment. J.A. 965. After that, the exclusive license blocked others from domestic marketing without risk of infringement. 52 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Other evidence supports a finding that the Elan pa- tent would have deterred entities other than Elan (holder of the Elan patent) and Acorda (exclusive licensee) from investing in research whose reward depended on market- ing a drug like Ampyra. After more than a decade of research by different groups and then issuance of the Elan patent in 1996, clinical trial research into sustained- release 4-AP treatment for multiple sclerosis appears, based on the prior art introduced at trial, to have been limited to Elan and Acorda. When seeking to use 4-AP for multiple sclerosis, Acorda itself sought and obtained a license to the Elan patent. There is no evidence that Elan sought to license the Elan patent to any entity other than Acorda, or that Acorda sought to sublicense the Elan patent, either of which would dilute the power of the blocking patent. J.A. 966. And what Elan granted Acor- da was an exclusive license, suggesting the significance of the Elan patent’s blocking power. Acorda notes that U.S. patents do not block sales out- side the United States. That observation is relevant, but it is not shown to be weighty in this case by any concrete evidence about the particular inventions at issue. Indeed, the two international studies that Acorda highlights were both conducted before issuance of the Elan patent in 1996. See J.A. 6654 (1994 Polman study); J.A. 7037 (1993 Van Diemen study). Acorda also notes that potential innovators would not have been blocked from practicing the Elan patent in the ways covered by the safe harbor provision of 35 U.S.C. § 271(e)(1), which declares specified activities to be non- infringing if undertaken “solely for uses reasonably relat- ed to the development and submission of information” to the FDA. See Merck KGaA v. Integra Lifesciences I, Ltd., 545 U.S. 193, 205–08 (2005). That safe harbor is certainly relevant, but it does not eliminate infringement liability for the eventual reward-collecting activity of generally marketing the product. We have no basis for finding clear ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 53 error in the district court’s finding about the explanatory significance of the risk of such liability. Acorda did not supply evidence to make unreasonable the implicit find- ing that securing freedom from blocking patents in ad- vance is likely important to pharmaceutical research investments. 17 And amici appearing in this court on appeal have not supplied such evidence either. 18 17 Without contrary evidence, we see nothing inher- ently unreasonable about the implicit finding to that effect. See Stoyan A. Radkov, Freedom to Operate (FTO) from a large company’s perspective 3, 5, Royal Society of Chemistry (Oct. 11, 2010), http://www.rsc.org/images/ StoyanRadkov_tcm18-192425.pdf (in a presentation by an attorney for Novartis Pharma AG an FTO analysis of “[t]he ability to perform a particular commercial activity (e.g. commercialize a product, provide a service, perform a manufacturing process or use a product) without ‘infring- ing’ 3rd party’s valid IP [intellectual property] rights,” explaining that “[i]dentifying possible 3rd party IP rights posing risks as soon as possible is essential”); Saharsh Davuluri, Generic Drugs – The Freedom to Operate, Neutland Labs. Ltd. (Aug. 2, 2014), https://www.neuland labs.com/blog/2014/08/02/generic-drugs-the-freedom-to-op erate/ (“A Freedom to Operate analysis is crucial – and is best performed before embarking down the product devel- opment path.”). In so stating, we do not prejudge what evidence in another case might demonstrate. 18 Amici point out that pharmaceutical improve- ments (new formulations, new combinations, and new indications of previously marketed drugs) are not un- common: 23 were approved by the FDA and launched in 2016. Biotech. Innovation Org. Br. at 20 (citing A.I. Graul et al., The year’s new drugs & biologics 2016: Part I, 53 Drugs of Today 27, 28 (2017)). But amici do not specify whether the approved applications for those improve- 54 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Acorda offers no more persuasive basis for challenging the district court’s findings of the weakness of Acorda’s evidence of the failure of others and long-felt but unmet ments are held by the owners (or licensees) of any original blocking patents or by competing entities. See Chie Hoon Song & Jeung-Whan Han, Patent cliff and strategic switch: exploring strategic design possibilities in the pharmaceutical industry, 5 SpringerPlus 692, 698–99 (2016) (noting that some of the best ways for a pharma- ceutical company to avoid the “patent cliff” of losing the monopoly on its brand-name drug from patent expiration is through a product-line extension (new formulations, new combinations), new indications, or a follow-on prod- uct). For example, among the examples from 2016 listed in the Graul article are Ilaris, Ezetrol, and Inegy, see Graul, The year’s new drugs & biologics, 53 Drugs of Today at 56, 57, which involve improvements (new indica- tions) on drugs previously approved for other indications for marketing by the same company that submitted the application for the new indication. See Product Update: New indication for Inegy, The Pharmaceutical Journal (Mar. 1, 2016), https://www.pharmaceutical-journal.com/ news-and-analysis/notice-board/new-indication-for-inegy/ 20200796.article?firstPass=false (Merck sells the drug Inegy (ezetimibe/simvastatin) for both old and new indica- tions); U.S. Food & Drug Admin., U.S. Dep’t of Health & Human Servs., FDA News Release: FDA approves expand- ed indications for Ilaris for three rare diseases (Sept. 23, 2016), https://www.fda.gov/newsevents/newsroom/press announcements/ucm522283.htm (Ilaris (canakinumab) sold by Novartis for old and new indications); Joel Levy, MHRA approves new indication for MSD’s Ezetrol, Phar- mafile (Feb. 26, 2016), http://www.pharmafile.com/news/ 503098/mhra-approves-new-indication-msd-s-ezetrol (Merck (MSD) sells Ezetrol (ezetimibe) for both old and new indications). ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 55 need as evidence of non-obviousness. Dist. Ct. Op. at *39– 40. As to the former, the district court found that Sanofi- Aventis experimented with another potassium-channel blocker and was unsuccessful, and “Sanofi-Avenits likely did not use 4-AP because” of the blocking effect of the Elan patent. Id. at *39. Acorda has not shown clear error in that finding. Acorda also points to the failure of Elan’s 1994 study. But the district court reasonably found that “Elan’s failure is not particularly probative” because the Elan study preceded publications that would render the invention obvious to those of skill in the art (Schwid and Goodman) as of the 2004 priority date. Dist. Ct. Op. at *40; see Graham, 383 U.S. at 36 (“The [1956] Scoggin invention . . . rests upon exceedingly small and quite non- technical mechanical differences in a device which was old in the art. At the latest, those differences were rendered apparent in 1953 by the appearance of the Livingstone patent [invalidating prior art], and unsuccessful attempts to reach a solution to the problems confronting Scoggin made before that time became wholly irrelevant.”); see also Note, Subtests of “Nonobviousness,” 112 U. Pa. L. Rev. at 1174 (“In receiving evidence of unsuccessful research, courts must take care that such research was conducted under the same state of the art as that which confronted the patentee. It may be that an intervening innovation made that which the patentee accomplished obvious even though it was not obvious to prior unsuccess- ful researchers.” (internal reference omitted)). By 1997, the art expressly explained why improvement of multiple sclerosis symptoms with 4-AP was promising despite the failed 1994 Elan study. See, e.g., J.A. 6681 (1997 Schwid article states that the EDSS score was “an inadequate outcome variable” for the Elan study, reports a significant improvement in timed gait, and concludes that “4AP [sustained-release] improved motor function in [multiple sclerosis] patients.”). 56 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. As to long-felt but unmet need, the district court discounted its finding of such need in light of the evidence of blocking by the Elan patent. Dist. Ct. Op. at *40. We see no clear error. While not dispositive, the evidence of blocking we have discussed is pertinent, in this case, to the factual question of long-felt but unmet need—at least as to the period after the issuance of the Elan patent in 1996. III The defendants cross-appealed the district court’s ruling that the Elan patent is not invalid and the result- ing injunction. Because the injunction terminated by its terms on the date of expiration of the Elan patent (July 30, 2018), and no retrospective liability is at issue, the cross-appeal is dismissed as moot. See Fed. R. App. P. 41(b), (c); 16AA Charles A. Wright & Arthur R. Miller, Federal Practice and Procedure § 3987 (4th ed. 2018); cf. Defs.’ Br. 61 (“the Court need not reach the cross-appeal unless the Court intends to issue a decision before August 2018”). IV We affirm the district court’s ruling that the asserted claims of the Acorda patents are invalid and dismiss the defendants’ cross-appeal as moot. AFFIRMED United States Court of Appeals for the Federal Circuit ______________________ ACORDA THERAPEUTICS, INC., Plaintiff-Appellant ALKERMES PHARMA IRELAND LIMITED, Plaintiff-Appellee v. ROXANE LABORATORIES, INC., MYLAN PHARMACEUTICALS INC., TEVA PHARMACEUTICALS USA, INC., Defendants-Cross-Appellants ______________________ 2017-2078, 2017-2134 ______________________ Appeals from the United States District Court for the District of Delaware in Nos. 1:14-cv-00882-LPS, 1:14-cv- 00922-LPS, 1:14-cv-00935-LPS, 1:14-cv-00941-LPS, Chief Judge Leonard P. Stark. ______________________ NEWMAN, Circuit Judge, dissenting. The court today holds that the new Acorda treatment for multiple sclerosis, Ampyra®, achieved after decades of failed research, was obvious. For this discovery, where a relatively small pharmacological difference produced long- sought medical benefits, it is essential that the correct law and analysis of obviousness are applied. 2 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. The district court observed that the objective indicia, viz. commercial success, long-felt but unmet need, failure of others, and copying, could change the result, yet dis- counted its weight on the theory that the patentee had a “blocking” patent. Adopting this flawed reasoning, my colleagues hold that this new treatment for multiple sclerosis was obvious. However, it is apparent that there is not clear and convincing evidence of obviousness. The consequences of this new legal theory are large, as the amici curiae advise. Had the court’s approach to the law of obviousness been in effect when Acorda took up the study of 4-aminopyridine after decades of failures by others, it is questionable whether this new treatment for multiple sclerosis would have been discovered and pur- sued. The loser is the afflicted public. 1 From my colleagues’ continuation of this error, and their erroneous conclusions, I respectfully dissent. I The Decades of Failures As the court reports, 4-AP has “for several decades” been the “focus of research regarding the treatment of multiple sclerosis.” Maj. Op. at 5. Starting in the 1980s or earlier, scientists in several countries tried and failed to provide safe and effective application of 4-AP. My colleagues agree, as do the Defendants who initiated these Hatch-Waxman proceedings, that the Acorda Pa- tents describe novel technology, and that a safe and effective formulation for 4-AP was not previously known. The Acorda inventors succeeded where many others had 1 The FDA gave the Acorda product expedited ap- proval, in view of the public need for relief of multiple sclerosis. Appellant’s Br. at 23. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 3 failed. The panel majority treats these past failures simply as invalidating prior art. The court recognizes that the Acorda Patents are di- rected to a new, effective treatment to relieve the “walk- ing impairment” of multiple sclerosis. 2 However, the court holds that Acorda merely “add[ed] further, more specific requirements to the Elan Patent’s claimed meth- ods.” Maj. Op. at 3. The court does not mention that Elan, after years of failures, abandoned its attempts to use 4-AP to treat multiple sclerosis and licensed the sustained-release patent to Acorda. The record shows that many scientists in many insti- tutions studied and eventually abandoned 4-AP as a treatment prospect for multiple sclerosis. These aban- doned studies constitute the prior art on which the dis- trict court and my colleagues rely for obviousness of the Acorda Patents. However, the experimentation with 4-AP shows just the opposite – it shows that work with 4-AP was abandoned due to the inability to balance the com- pound’s potential effectiveness with its toxicity. To review obviousness of the Acorda Patents, I start with the cited references, whose chronology illustrates the initial encouragement followed by failed attempts to apply the neurological properties of 4-aminopyridine, and the 2 The symptoms of multiple sclerosis include “walk- ing impairment, visual difficulty, fatigue, bladder dys- function, tingling or pain, sexual dysfunctions, balance problems, and cognitive changes,” with “weakness in the legs and/or alterations in walking among the most com- mon symptoms.” Acorda Therapeutics, Inc. v. Roxane Labs., Inc., No. 1:14-cv-00882-LPS, 2017 WL 1199767 (D. Del. Mar. 31, 2017) (Dist. Ct. Op.) at *2. 4 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. eventual abandonment of this product despite some positive observations. A. The Stefoski Study In 1987, Stefoski et al. reported a one-day test of the effects of 4-AP on vision and gait in twelve multiple sclerosis patients. 3 They reported that, following intrave- nous injection of 7 to 35 mg of 4-AP, in 1 to 5 mg doses every ten to sixty minutes, “[v]ision improved in 7 pa- tients, oculomotor function in 5, and motor function (power, coordination, gait) in 5,” stating that there were “no serious side effects,” and “transient therapeutic bene- fit in selected patients.” Stefoski et al. at 71. My col- leagues rely on this publication for rendering obvious Acorda’s improvement in walking, while downplaying the “serious side effects” including seizures reported by Bev- er 4 and others, and the criticism of the small sample size and the brief duration of these one-day tests. B. The Davis Study In 1990, Davis and Stefoski reported a study of fifteen patients using an orally-administered formulation of 4- AP. 5 They concluded that the results “suggest a safe and effective therapeutic window for orally administered 4- AP,” but they cautioned that similar studies had found 3 Dusan Stefoski et al., 4-Aminopyridine Improves Clinical Signs in Multiple Sclerosis, 21 Annals of Neurol- ogy 71 (1987), J.A. 6697. 4 Christopher T. Bever, Jr., The Current Status of Studies of Aminopyridines in Patients with Multiple Sclerosis, 36 Annals of Neurology S118 (1994) (“Bever II”), J.A. 6172. 5 Floyd A. Davis et al., Orally Administered 4- Aminopyridine Improves Clinical Signs in Multiple Scle- rosis, 27 Annals of Neurology 186 (1990), J.A. 6327. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 5 that side effects of 4-AP “precluded its clinical use,” and that “MS patients have an increased risk of seizures.” Davis et al. at 191. These studies were criticized by Bever as “limited be- cause they did not use a randomized treatment design, were not double blinded, and relied on outcome measures that were not widely accepted.” Bever II at S119. Alt- hough my colleagues cite Davis’ reports of “mild to marked improvements,” Maj. Op. at 5, they do not men- tion the risk of seizures as warned by Davis, or Bever’s criticisms. While the panel majority states that Davis reported “no serious or bothersome side effects, including seizures” at doses up to 25 mg, id., Elan, which relied on Davis’ research team, Dist. Ct. Op. at *4, terminated its devel- opment of 4-AP based on toxicity and seizures, and li- censed its sustained release patent to Acorda. Nonetheless, my colleagues hold that the Davis studies contributed to the obviousness of the Acorda Patents, ignoring the problems that were reported, and the aban- donment of 4-AP by these researchers. C. The Van Diemen study The panel majority also relies on a study conducted in the Netherlands and published in 1993 by Van Diemen. 6 The publication reports the effect of escalating doses of 4- AP, measured by the Kurtzke expanded disability status scale (EDSS) that is frequently used as a benchmark to measure symptoms in multiple sclerosis patients. The 6 Harriët A. M. Van Diemen et al., 4-Aminopyridine in Patients with Multiple Sclerosis: Dosage and Serum Level Related to Efficacy and Safety, 16 Clinical Neuro- pharmacology 195 (1993), J.A. 7037. 6 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. study examined the effect on eye function of intravenous and oral administration of 4-AP for up to 12 weeks. My colleagues report that eye functioning was bene- fited, but ignore the report of side effects, including nau- sea and dizziness, at the “escalated” dosages needed to produce improvement in eye function. Van Diemen et al. at 200, 203. D. The Polman study Polman7 describes an unblinded study of the treat- ment with 4-AP of thirty-one multiple sclerosis patients, some of whom had been involved in an earlier study. Twenty-three patients were treated with 4-AP for longer than six months. The new patients were given an upward titration dosing plan in accordance with the tolerability by the patient, up to a maximum dose (based on patient weight) over four to eight weeks. Polman measured efficacy based on subjective reports from the patients during clinic visits. The Van Diemen and Polman references were relied on by the district court as teaching “stable dosing,” but they involve stable dosing only after titration to the highest tolerable dose for each individual patient. Both Van Diemen and Polman describe using a titration scheme up to the maximum amount based on the patient’s weight. Dist. Ct. Op. at *12–13. These references only teach stable dosing after the maximum tolerable dose has been determined for each patient, after upward titration. Goodman, post, also reports an “increasing benefit” for doses up to 50 mg/day if such doses can be tolerated. These sources all show the understood need to target 7 Chris H. Polman et al., 4-Aminopyridine in the Treatment of Patients with Multiple Sclerosis, 51 Archives of Neurology 292 (1994), J.A. 6654. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 7 higher doses to the extent they can be tolerated. See Goodman Poster (reporting increasing benefit as dosage was increased from 20mg to 50mg). 8 Polman reported that “[i]mprovements in fatigue and ambulation were mentioned quite often by the patients.” Polman et al. at 295. However, two patients in the Pol- man study experienced seizures and discontinued partici- pation. Id. at 294–5. My colleagues cite Polman’s report of “favorable response to the medication,” Maj. Op. at 7 (citing id. at 293), but downplay Polman’s conclusion that there was little quantifiable benefit of the therapy using the primary EDSS benchmark, my colleagues stating that the side effects were not troublesome, despite the reports of seizures. Maj. Op. at 8–9. E. Additional studies reported by Bever The Bever II reference reports additional studies, as follows: Two double-blind, placebo-controlled crossover trials of DAP have recently been completed. Carter and associates, using 3-week treatment pe- riods and doses up to 80 mg/day, found subjective improvement in 48% of patients on DAP but only 24% on placebo. Although this difference was not statistically significant, treatment-related differ- ences were found in sensitivity to thermal chal- lenge. 8 Dist. Ct. Op. at *14 (“The Goodman Poster is a poster presented at the September 2002 annual meeting of the America Committee for Treatment and Research in Multiple Sclerosis, held in Baltimore, Maryland.”), J.A. 6497–504. 8 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Bever II at S120 (citing JL Carter et al., A double-blind, placebo-controlled crossover trial of 3,4-diaminopyridine in the symptomatic treatment of multiple sclerosis, 34 Annals of Neurology 272 (1993)). These studies further illustrate the uncertain state of the art at that time, and the “differences” and “sensitivi- ty” that led to abandonment of development of 4-AP. These studies did not lead to any proposed treatment of multiple sclerosis, despite the accumulating knowledge concerning 4-AP. My colleagues mention the toxic effects including seizures, encephalopathy, and hepatitis, but skip over their importance. However, it is apparent that others did not ignore their importance, for no proposed product, no proposed treatment, resulted from these studies. F. The abandoned Elan studies The manifestations and miseries of multiple sclerosis are powerful, and Elan Corporation entered the field to pursue the idea that sustained-release formulations of 4- AP might relieve the toxic effects and provide “therapeu- tically effective blood levels throughout a given treatment period.” U.S. Patent No. 5,540,938 (the “Elan Patent”) at col.2, l.15. In 1991 Elan filed the patent application leading to the Elan Patent, which described and claimed sustained-release formulations of 4-AP. Elan undertook major efforts to develop a treatment for multiple sclerosis using sustained-release formulations. Reports of these unsuccessful efforts were published. Schwid 9 reports a failed clinical trial in 1994, de- scribed as a six-week, 161-patient placebo-controlled 9 Steven R. Schwid et al., Quantitative assessment of sustained-release 4-aminopyridine for symptomatic ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 9 study of the administration of sustained-release 4-AP to multiple sclerosis patients. The results were measured using the EDSS benchmark, and included measures of walking disability including gait and speed. The conclu- sion was that there was no improvement over the placebo. Schwid et al. at 817. Another Elan study of ten patients, also reported by Schwid, stated that nine of these patients showed an improvement in speed of walking. Schwid discussed that the mean serum level of 4-AP during the study was “65±25 ng/ml (range, 34-99)” and that the treatment “appeared particularly efficacious in subjects who achieved serum 4AP levels above 60 ng/ml.” Id. at 819– 20. The study reported that “[n]one of the patients with a serum level less than 60 ng/ml felt better (according to their global impressions) on 4AP SR [sustained-release] than placebo,” while all patients with serum levels above 60 ng/ml demonstrated improvement in timed gait, grip strength, and five of six improving by their own subjective impression. Id. at 819–20. In contrast, the Acorda Pa- tents are directed to a serum range of about 15-35 ng/ml, which Schwid described as unlikely to produce therapeu- tic effect. The 17.5 mg dose used by Schwid was stated to be in- effective in a number of respects, including the EDSS benchmark. Schwid et al. at 817. Schwid suggested that further research should be conducted, but this does not convert Schwid’s reported failures into a teaching of the path to success. My colleagues state that Schwid reported “promising” results, Maj. Op. at 55, but do not mention Schwid’s conclusion that 4-AP was not effective at the doses that were necessary to limit toxicity, or the lack of treatment of multiple sclerosis, 48 Neurology 817 (1997), J.A. 6681-84. 10 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. improvement over placebo. Instead, my colleagues sug- gest that Schwid contributed to obviousness because Schwid suggested that, since the EDSS benchmark had failed, it might be useful to look at “more sensitive, quan- titative measures.” Maj. Op. at 13–14 (quoting Schwid, J.A. 6681). Thus the panel majority concludes that these studies rendered obvious the Acorda success that had eluded Schwid. Elan also sponsored studies at the University of Mar- yland, published by Bever et al. 10 Bever summarized that the “lower serum concentration range of 30 to 59 ng/ml may . . . be adequate for inducing improvement of some neurologic deficits,” Bever I at 1058, quoted at Maj. Op. at 10; but the panel majority ignores that the study did not show any improvement on the EDSS benchmark or on an ambulation benchmark, id. at 1056–57, and treats the Bever report of “increased side effects,” including a grand mal seizure, as a throwaway, Maj. Op. at 10. These studies surely added to the body of knowledge, but they did not produce a usable product. Although these studies used Elan’s sustained-release formulations, the effort was eventually abandoned. The record is con- sistent in showing that Elan, like the others who had studied 4-AP, had been unable to achieve an effective product free of toxicity and serious side effects. 10 Christopher T. Bever, Jr. et al., The effects of 4- aminopyridine in multiple sclerosis patients: Results of a randomized, placebo-controlled, double-blind, concentra- tion-controlled, crossover trial, 44 Neurology 1054 (1994) (“Bever I”), J.A. 6180. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 11 G. The Hayes report of early Acorda studies Hayes 11 reports Acorda’s activity, starting in 1993 and investigating use of 4-AP for treatment of spinal cord injury. The first of these studies evaluated single doses of sustained-release 4-AP in fourteen patients with spinal cord injury, and the second study examined multiple doses of sustained-release 4-AP in sixteen patients with spinal cord injury. Dist. Ct. Op. at *15 (citing Hayes et al. at 186). The Hayes publication stated that all patients in both studies experienced at least one adverse event, such as dizziness, hypotension, or nausea. Hayes et al. at 188, 191. H. The Solari review article Solari 12 is a review of medical knowledge related to 4- AP, including reports on clinical trials conducted with MS patients. From the studies in its analysis, Solari tabulat- ed that 54% of the multiple sclerosis patients taking 4-AP or diaminopyridine experienced improved motor func- tions, compared to 7% of placebo. Solari et al., J.A. 7204. Solari concluded that its “review of trials found there is not enough evidence about the safety of these drugs or whether benefits are certain.” Solari et al., J.A. 7218. 11 Keith C. Hayes et al., Pharmacokinetic Studies of Single and Multiple Oral Doses of Fampridine-SR (Sus- tained-Release 4-Aminopyridine) in Patients With Chronic Spinal Cord Injury, 26 Clinical Neuropharmacology 185 (2003), J.A. 6433. 12 Alessandra Solari et al., Aminopyridines for symp- tomatic treatment in multiple sclerosis (Review), Cochrane Database of Systematic Reviews, Issue 4 (2002), J.A. 7204. 12 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. II The Acorda Studies As outlined supra, Acorda began research with 4-AP in 1993 for treatment of spinal cord injury. As reported by Hayes, successful results were not obtained. Dr. Ron Cohen, the founder of Acorda, turned to study of multiple sclerosis. Dr. Cohen testified that he took on the “daunt- ing challenges” of seeking an effective treatment for multiple sclerosis, with knowledge of the failures of Elan and others. Appellant’s Br. at 13 (citing J.A. 596–97). Acorda scientists conducted research over the ensuing six years, and published their results as experience accu- mulated and knowledge evolved. These publications are treated as prior art to the Acorda Patents. A. Acorda’s initial failures Acorda’s initial publications reported that the multi- ple sclerosis population receiving various experimental 4- AP treatments showed some improvement in walking speed and lower extremity muscle strength, but “did not show that any individual dosage had a statistically signif- icant effect versus placebo.” Appellant’s Br. at 15; see Goodman Poster, n.9 ante. Dr. Goodman was the lead clinical investigator for Acorda, and the lead author for the published results of Acorda’s MS-F201 study, 13 a randomized double-blind placebo-controlled study with the aim of “determin[ing] the safety and tolerability of escalating doses of a sustained-release (‘SR’) formulation given orally to patients with MS.” Goodman I at S116. 13 Andrew Goodman et. al., Placebo-Controlled Dou- ble-blinded Dose Ranging Study of Fampridine-SR in Multiple Sclerosis, 8 Multiple Sclerosis S116 (P308) (July 2002), (“Goodman I”) J.A. 6370. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 13 Goodman I states that the MS-F201 data “showed sta- tistically significant improvement from baseline compared to placebo in functional measures of mobility (timed 25 walking speed; p=0.04) and lower extremity strength (manual muscle testing; p=0.01).” Id. at S117. It further states that “[d]ose response curves showed increasing benefit in both measures in the 20 to 50 mg/day range.” Id. However, two participants withdrew due to seizures. Id. The Goodman Poster reported that the MS-F201 study demonstrated “statistically significant improve- ments in the timed 25-foot walk and manual muscle test relative to placebo.” Dist. Ct. Op. at *15. However, the Poster also stated that a greater improvement in fatigue was reported by the placebo group as compared to the 4- AP treated group, and referred to the withdrawal of two subjects due to seizure. Goodman Poster, J.A. 6502. Dr. Goodman testified at trial that “[a]ll of the prespecified analyses failed except for the lower extremity manual muscle test.” J.A. 604 (289:24–5). He stated that the result of the timed walk “was not at all significant,” and was consistent with the failed Elan study. J.A. 605 (290:5). The district court found that the Goodman Poster es- tablished that “the use of a 10 mg sustained-release dose of 4-AP twice per day to treat walking in MS patients would have been obvious to a POSA at the priority date of the Acorda Patents.” Dist. Ct. Op. at *33. Acorda states that “the district court’s conception that the Goodman Poster teaches anything about a 10 mg BID dose of 4-AP as the sole individual dose of an MS treatment protocol— as opposed to merely the starting point of an escalating dosing scheme—is impermissible hindsight.” Appellant’s Br. at 38. Acorda is correct that the Goodman Poster does not suggest this low-dose formulation with a reasonable expectation of success, but reports increasing benefit as dosage was increased from 20 to 50 mg. 14 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Acorda correctly states that the Elan work and these initial Acorda studies show, if anything, that 4-AP treat- ment requires upward titration to determine the maxi- mum tolerable dose for individual patients since efficacy can only be achieved at higher doses, and that these studies do not provide any reason to believe that a low dose would be effective. Goodman I reported an “increas- ing benefit in both measures in the 20-50 mg/day range,” referring to mobility and lower extremity strength. Goodman I at S117. In 2003, Acorda conducted a 206-patient clinical study, designated MS-F202. The study employed upward titration to successively higher doses, starting at dosages of 10 mg of sustained-release 4-AP twice-daily. The highest tolerable dose was then continued for 12 weeks. It was concluded that no treatment group showed im- provement over placebo, over the 12-week testing period. Dist. Ct. Op. at *9. The low dose protocol developed by Acorda is not sug- gested in the prior art. Although the goal was a stable dose without individual titration, no study, no reference reported successful results using the low dose of the Acorda Patents, or even suggested that it should be tried. The panel majority’s contrary theory is devoid of support. B. Acorda’s analytical breakthrough Acorda analyzed the MS-F202 results, focusing on “patients in the study who, after treatment, showed a ‘meaningful difference’ from their before-treatment base- line—i.e. the ‘responders,’” and learned that the therapeu- tic effect of 4-AP did not increase with increase in dosage, as prior reports and Acorda’s own research had suggested. Appellant’s Br. at 19. Dr. Cohen testified that they “were extremely surprised” because “[e]verything that we had come to expect throughout the program told us that we should be seeing more and more efficacy the higher the dose went as long as the patients were tolerating it and ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 15 that turned out not to be the case.” J.A. 614 (299:5–9). This contradicted the teachings of all of the earlier stud- ies. Only the courts find it obvious. Acorda then conducted additional clinical studies at the lower dosages, and established that a twice-daily sustained-release 10 mg dose produced improvement in walking gait and speed, while avoiding the toxicity and seizures of higher dosages. Acorda filed a provisional patent application on April 9, 2004, directed to this treatment. Acorda continued its studies, and after a total of twelve years of investigation and development, Acorda in 2010 obtained FDA approval for a product for improv- ing the walking impairment in multiple sclerosis patients. This product has the brand name Ampyra®. The Acorda Patents are directed to and limited to the twice-daily administration of 10 mg doses of sustained-release 4-AP formulation. The district court, affirmed by my colleagues, held the Acorda Patents invalid on the ground of obviousness. The district court ruled that the evidence of long-felt need, failure of others, unexpected results, and commercial success are irrelevant because the Elan Patent was a “blocking” patent. However, the Elan Patent did not block research on 4-AP, did not block other possible treatments for multiple sclerosis, and did not affect the Defendants’ development and copying and Hatch- Waxman challenge to the Elan and Acorda patents. The court’s theory of “blocking” is unrelated to whether the Acorda product meets a long-felt need in treating multiple sclerosis, for the Elan and Acorda patents do not block the Defendants from developing a competitive treatment for multiple sclerosis. The patents that support Acorda’s eventual success do not block others from using and learning from Acorda’s teachings, experimenting with and comparing with Acorda’s product, and engaging in com- petitive activity. 16 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. III The District Court’s Analysis The Defendants conceded infringement, and the dis- trict court found the Acorda Patents invalid on the ground of obviousness. The district court determined that four claim elements were common to the Acorda Patents, then found that each of these elements is present in a separate reference, and held that a person of ordinary skill in this field would obviously have selected and combined these elements to produce the Acorda product and method. The district court did not find any motivation or sug- gestion in the prior art as to which elements to select and combine, and did not find any teaching or suggestion that such selection and combination would be likely to succeed in treating the walking impairment of multiple sclerosis. Acorda attributes the district court’s rulings to “hindsight bias” and incorrect statements of law by the Defendants. Indeed, without the hindsight knowledge of Acorda’s success, there is no teaching or suggestion of this selection and combination or its likelihood of success. A. The selected claim elements The district court selected four aspects of the Acorda claims, as follows: (1) the use of 4-AP to improve walking in multiple sclerosis patients; (2) the use of a 10 mg twice- daily sustained release dose; (3) the use of stable dosing without upward titration; and (4) the specific pharmaco- kinetic parameters achieved. The court concluded that “a POSA would have been motivated to combine these limi- tations with a reasonable expectation of success.” Dist. Ct. Op. at *29. However, the question is not whether these four ele- ments, if combined, would produce a successful treatment. The question is whether the prior art contains a sugges- tion or motivation to select these four elements from the decades of inconclusive prior art, with a reasonable expec- ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 17 tation that the selection would eliminate the failures of the prior art. See, e.g., In re Cyclobenzaprine Hydrochlo- ride Extended-Release Capsule Patent Litig., 676 F.3d 1063, 1068–69 (Fed. Cir. 2012) (“a party seeking to invali- date a patent as obvious must ‘demonstrate “by clear and convincing evidence that a skilled artisan would have had reason to combine the teaching of the prior art references to achieve the claimed invention, and that the skilled artisan would have had a reasonable expectation of suc- cess from doing so.”’” (quoting Procter & Gamble Co. v. Teva Pharms. USA, Inc., 566 F.3d 989, 994 (Fed. Cir. 2009))); In re Kubin, 561 F.3d 1351, 1359 (Fed. Cir. 2009) (prior art does not provide a reasonable expectation of success where the art may suggest “vary[ing] all parame- ters or try[ing] each of numerous possible choices until one possibly arrived at a successful result, where the prior art gives either no indication of which parameters were critical or no direction as to which of many possible choic- es is likely to be successful.” (quoting In re O'Farrell, 853 F.2d 894, 903 (Fed. Cir. 1988)). The years of studies and failures weigh heavily against the simplistic post hoc predictability accepted by the court. The district court analyzed the purported obviousness of each of the four limitations, as follows. 1. Improvement in walking The district court found that several references showed improved walking upon treatment with 4-AP. The court framed the question as whether a POSA would have “a reasonable expectation that 4-AP could be suc- cessfully used as claimed to treat (i.e., achieve therapeuti- cally-effective blood levels in) even a single patient.” Dist. Ct. Op. at *30. The court referred to Schwid’s analysis of the early Acorda studies as showing “a statistically signif- icant improvement in . . . timed gait, which was found to be improved in nine out of 10 patients, in comparison to the placebo group.” Id. 18 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. However, the early Acorda studies all stated concern about toxicity, particularly seizures, at the dosages that these studies showed were needed to obtain relief. No witness suggested that these early studies taught or suggested that a low dosage formulation would be effec- tive. 2. The dosage of 10 mg twice daily The district court concluded that this dosage was an obvious choice, because the prior art evaluated doses ranging from 10 mg to 80 mg. Dist. Ct. Op. at *32. How- ever, the prior art contains no suggestion, indeed no hint, that a 10 mg twice-daily sustained-release formulation would be effective. All of the early references demon- strated the need for upward titration, showing that high- er doses are needed for efficacy, with individual titration to determine the highest tolerable dose before seizures occurred. The district court cited the Goodman Poster as showing that toxicity increased at higher dosages, and as providing “[e]vidence of dose-response in [the] 20-40 mg/day range.” Dist. Ct. Op. at *32. However, the stud- ies reported by Goodman did not provide a safe and efficacious product, but depended on individual titration to establish individual dosages at the highest tolerable level. The district court held that it was obvious to use the 10 mg dose, despite the general showing of ineffectiveness of the 10 mg dose. Dist. Ct. Op. at *33. It is not disputed that the general teaching was that doses higher than 10 mg were needed for therapeutic effect. It cannot reasona- bly be viewed as obvious that a dosage that was described in the prior art as ineffective, is in fact the optimum dosage. 3. Stable dosing without upward titration The district court found that the prior art, particular- ly the Van Diemen and Polman references, taught the use ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 19 of uniform dosing of 4-AP, and “included reports of safe and effective long-term use of stable dosing of immediate- release 4-AP.” Dist. Ct. Op. at *34. The district court further found that the prior art’s “consistent use of titra- tion . . . did not undermine the other evidence in the prior art that supports finding that a POSA would have had a reasonable expectation of success with stable dosing.” Id. Only hindsight can construct the Acorda formulation from these inapt teachings, for the references cited by the district court require upward titration to select the high- est tolerable dose, for low stable doses were ineffective. The panel majority, seeking to fill this gap, asserts that “[t]he prior art is not limited to titrated dosing,” Maj. Op. at 34, citing Polman and Schwid. However, Polman involved titration, and reported that therapeutic doses required in excess of 40 mg for minimal quantifiable benefit. See Polman et al. at 295 (stating that the report- ed improvements generally did not result in significant changes to the EDSS benchmark). In addition, Schwid suggested the need for a far higher dose, only maintained stable dosing for a week, and did not report meaningful success in treating multiple sclerosis. Schwid et al. at 817. 4. Pharmacokinetic limitations For the fourth limitation, the district court found that the claimed pharmacokinetic serum levels were disclosed by Hayes, for “[i]t is undisputed that the Hayes research- ers used the Ampyra® formulation in their study.” Dist. Ct. Op. at *35. The district court considered Acorda’s argument that “there is nothing in the prior art identify- ing the pharmacokinetic values recited in the claims as being effective to improve walking or increase walking speed in MS patients,” id., and found that “a POSA would have been aware that a sustained-release dosage form achieving the pharmacokinetic parameters disclosed in 20 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. Hayes III would have been associated with an improve- ment in walking in MS patients.” Dist. Ct. Op. at *36. The Defendants argue that even if the serum level in the Acorda Patents is not obvious based on the Hayes reference, the claimed range is inherent in the dosage of 4-AP, citing Santarus, Inc. v. Par Pharm., Inc., 694 F.3d 1344, 1354 (Fed. Cir. 2012), where the court held that reciting the blood serum concentration resulting from a dosage form did not impart patentability to known dosage forms. Acorda responds that the prior art did not teach or suggest that any specific blood serum levels would im- prove walking in multiple sclerosis patients. No such teaching or suggestion appears anywhere in the record. Hayes does not relate its serum analysis to efficacy in improving gait or walking speed in persons afflicted with multiple sclerosis. The district court referred to Acorda’s statement at trial, that “[i]t was known in the art that a sustained release formulation of 10 megs BID could achieve” the claimed pharmacokinetic values. Dist. Ct. Op. at *35 n.39 (citing J.A. 1108–1109). The district court found that there was a reasonable expectation of success with regard to the pharmacokinetic parameters because these param- eters are inherent in the claimed dosing. Id. The court did not find, and the prior art does not establish, that this pharmacokinetic range was known to have a beneficial effect on walking speed and gait in persons afflicted with multiple sclerosis. B. The combination of elements The district court found a reasonable expectation of success on combination of the four claim elements, stating that “a POSA would consider 10 mg/twice daily to be among the finite group of doses of sustained-release 4-AP that could reasonably be expected to improve walking in MS patients,” Dist. Ct. Op. at *33. The court concluded that: ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 21 Defendants have adduced clear and convincing ev- idence that a POSA at the priority date would have been motivated and would have had a rea- sonable expectation of success to practice and combine each of the limitations of the asserted claims of the Acorda Patents. Dist. Ct. Op. at *40. Acorda is correct that there was no suggestion in the prior art that the claimed combination should be tried, and there is no hint of a reasonable expectation of success. Acorda points to the decades of failure of others to develop a safe and effective treatment for multiple sclerosis using 4-AP, despite its known toxicity. The district court’s selection of separate limitations from separate sources, and retrospectively fitting them into the Acorda template, is achieved only with the hindsight knowledge of Acorda’s eventual success. See Sanofi-Synthelabo v. Apotex, Inc., 550 F.3d 1075, 1086 (Fed. Cir. 2008) (“The determination of obviousness is made with respect to the subject matter as a whole, not separate pieces of the claim.”). Here, only the Acorda Patents teach the combination that successful- ly treats this multiple sclerosis impairment while avoid- ing toxicity and seizures. Acorda’s path to successfully harness the neurological benefits of 4-AP eluded the many scientists studying multiple sclerosis. Although the district court acknowl- edged the known adverse effects of 4-AP including sei- zures, Dist. Ct. Op. at *41 (stating that “the Court agrees with Plaintiffs that, at the priority date of the Acorda Patents, the risk of seizures loomed over the work of exploring the use of 4-AP in MS”), nonetheless the court found that a person of ordinary skill would have had a reasonable expectation of success with the Acorda prod- uct. The recognized need for a stable, non-toxic dosage protocol does not render the solution obvious if it is even- tually discovered. The record does not show any teaching 22 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. or suggestion of success of the formulation in the Acorda Patents. Nor does the record support a finding of “obvious to try.” Such a finding requires that a person of ordinary skill would not only have selected these specific elements from various discarded experiments, but also “would have had a reasonable expectation of success in doing so.” Pfizer, Inc. v. Apotex, Inc., 480 F.3d 1348, 1361 (Fed. Cir. 2007). It is clear that the prior art does not provide a reasonable expectation of success of the Acorda Patents’ specific dosage and protocol. IV The Objective Indicia of Unobviousness The objective indicia “may often be the most probative and cogent evidence in the record . . . . It is to be consid- ered as part of all the evidence, not just when the deci- sionmaker remains in doubt after reviewing the art.” Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1538–39 (Fed. Cir. 1983). The district court, affirmed by the panel majority, err in discounting the undisputed evidence of commercial success, long-felt need, failure of others, unexpected results, and copying. The district court discussed the objective indicia, and concluded that they did not “outweigh” the conclusion of obviousness. The district court found that Ampyra® could be considered a commercial success “[g]iven the strength of Ampyra®’s sales, and the absence of any evidence that its sales are disappointing given its limited indication and patient population.” Dist. Ct. Op. at *38. However, the court concluded that this commercial success did not weigh heavily because “no one other than the Elan pa- tentees and their licensees could have practiced the invention of the Acorda Patents without facing liability for patent infringement.” Id. ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. 23 Commercial success is measured against the products available for the same purpose, not against infringing copies of the patented product. Defendants do not con- tend that they are precluded from providing or developing other treatments for multiple sclerosis. The Acorda product met a long-felt need, for which the failure of others, despite decades of experimenting with the neuro- logical properties of 4-AP, is evidence of the unobvious- ness of the Acorda achievement. Such evidence is an important aid to a court that is attempting to divine whether the patentee’s discovery was obvious in accord- ance with law. Concerning failure of others, the panel majority states that Elan’s failure “is not particularly relevant to the expectation of success.” Maj. Op. at 40–41. This is a peculiar conclusion, for Elan had undertaken an immense investment, including clinical trials, in the hope that its extended-release concept would solve the problems en- countered by others. Elan eventually gave up. Nonethe- less, my colleagues find that Acorda’s success was obvious to them. The district court and my colleagues also misapply the concept of “blocking patent,” and hold that because a patent provides the right to exclude infringers, the indicia of commercial success, long-felt need, failure of others, and copying are diminished. However, as the Pharmaceu- tical Research and Manufacturers of America, as amicus curiae, reminds us, “a prior patent would not have cate- gorically precluded others from further developing the technology,” pointing to the statutory safe harbor of § 271(e)(1), the knowledge provided in the patents, and the right to conduct research on patented subject matter. Br. of Amicus Curiae at 4. The objective indicia of unobviousness are measured against the state of the science and in the commercial context. Here the unexpected success and its human 24 ACORDA THERAPEUTICS, INC. v. ROXANE LABORATORIES, INC. benefits are not disputed. The district court was advised that the Patent Trial and Appeal Board sustained the validity of the Acorda Patents in inter partes review, at Coalition for Affordable Drugs (ADROCA), LLC v. Acorda Therapeutics, Inc., 2017 WL 950736 (P.T.A.B. Mar. 9, 2017). Although the majority reports this event, as did the district court, its consequences are not explored, including issues of privity, estoppel, and finality. CONCLUSION Obviousness of the Acorda Patents was not estab- lished by clear and convincing evidence. The prior art did not provide a suggestion to select the specific elements and limitations of the Acorda formulation, and did not suggest that such selection and combination would have a reasonable expectation of success in relieving the walking impairment of multiple sclerosis. From my colleagues’ contrary holding, I respectfully dissent.