In Re Glumetza Antitrust Litigation

Opinion

1 2 3 4 5 6 UNITED STATES DISTRICT COURT 7 NORTHERN DISTRICT OF CALIFORNIA 8

10 In re No. C 19-05822 WHA 11 GLUMETZA ANTITRUST No. C 19-06138 WHA LITIGATION. No. C 19-06839 WHA 12 No. C 19-07843 WHA No. C 19-08155 WHA 13 No. C 20-01198 WHA This Document Relates to: No. C 20-05251 WHA 14 ALL ACTIONS. (Consolidated) 15 ORDER RE SUMMARY 16 JUDGMENT

17 18 INTRODUCTION 19 In a secret pharmaceutical-patent infringement settlement agreement, concealed from the 20 district judge, brand and generic manufacturers of the type 2 diabetes drug Glumetza allegedly 21 pledged not to compete with each other by agreeing to not introduce a generic version of the 22 drug for several years. So, while generic competition should have driven drug prices down, the 23 brand manufacturer instead hiked prices up. The generic manufacturer belatedly entered the 24 market at premium pricing, and, as a result, the conspiring manufacturers allegedly extracted 25 huge sums of money from consumers. In this resulting antitrust action, a certified class of 26 direct purchasers move for partial summary judgment that our defendant brand and generic 27 manufacturers wielded market power. For their part, defendants move for summary judgment, 1 STATEMENT 2 An estimated thirty million Americans suffer from type 2 diabetes. The condition, 3 “caused by the combination of insulin resistance . . . and deficient insulin secretion,” causes an 4 alarming array of complications. Those with diabetes face nearly twice the average risk of 5 stroke and heart disease. They may suffer foot ulcers that take months or years to heal, or may 6 require amputation. Around forty percent suffer some degree of kidney disease, with one 7 percent suffering kidney failure. In fact, “[p]ersons with diabetes make up the fastest growing 8 group of kidney dialysis and transplant recipients in the United States.” “Diabetes is the 9 leading cause of new cases of blindness among adults aged 18–64 years.” And, in 2017, 10 diabetes was estimated to be the seventh leading cause of death in the United States. CTRS. 11 DISEASE CTRL. & PREV., NAT’L DIABETES STATS. RPT. 12 (2020), https://www.cdc.gov/ 12 diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf; NAT’L INST. OF DIABETES & 13 DIGESTIVE & KIDNEY DISEASES, DIABETES IN AMERICA, ch. 1, pp. 2–3, ch. 22, p. 1, fact sheet 14 (3d ed. 2018), https://www.niddk.nih.gov/about-niddk/strategic-plans-reports/diabetes-in- 15 america-3rd-edition. 16 Fortunately, modern medicine offers a battery of remedial medications. One such drug, 17 metformin hydrochloride, helps control blood-sugar levels. Following approval by the Food 18 and Drug Administration in June 2005, defendant Depomed, Inc. launched a new extended- 19 release version of metformin in late 2006. Glumetza, introduced first in 500 mg and later in 20 1000 mg tablets, extended the release of metformin over a prolonged period by disbursing the 21 active drug into a polymeric matrix. In the stomach, the metformin would more slowly diffuse 22 out of the matrix and ensure its smooth delivery into the bloodstream over time, without the 23 usual initial spike and later lull in drug level, to offer consistent blood-sugar control. 24 Depomed obtained several patents covering the developments embodied in Glumetza, 25 United States Patent Nos. 6,340,475 and 6,635,280, which expired in September 2016, No. 26 6,488,962, which expired in June 2020, and No. 6,723,340, which will expire in October 2021. 27 Depomed listed these patents as covering Glumetza in the FDA’s “Orange Book.” U.S. DEP’T 1 HEATH & HUM. SERVS., FDA, APPROVED DRUG PRODUCTS WITH THERAPEUTIC EQUIVALENCE 2 EVALUATIONS (41st ed. 2021), https://www.fda.gov/media/71474/download. 3 In 2009, defendants Lupin Pharmaceuticals, Inc. and Lupin Limited filed an Abbreviated 4 New Drug Application with the FDA, seeking to manufacture and market generic versions of 5 both 500 mg and 1000 mg Glumetza. Lupin certified to the FDA that Depomed’s patents were 6 either invalid or not infringed, yet Depomed sued anyway in November 2009, asserting several 7 claims from the ’475, ’280, and ’962 patents (it also asserted but later dropped the ’340 patent). 8 The case came before the Honorable Phyllis J. Hamilton of our district. Following a year of 9 litigation, Judge Hamilton held a Markman hearing in January 2011 and issued a claim 10 construction order in May. The parties fought for the rest of that year, but, before filing 11 dispositive motions, settled in February 2012. As far as the parties told the judge, they would 12 simply go their separate ways and Lupin would launch its generic on February 1, 2016. 13 Depomed, Inc. v. Lupin Pharms., Inc., No. C 09-05587 PJH, Dkt. No. 152 (N.D. Cal. Mar. 27, 14 2012). This, our plaintiffs now allege, concealed an underlying unlawful conspiracy. 15 * * * 16 Congress enacted the Drug Price Competition and Patent Term Restoration Act of 1984, 17 more commonly known as the Hatch-Waxman Act, to lower pharmaceutical drug prices by 18 speeding generics to market. In general, a pharmaceutical drug must undergo a grueling — 19 and costly — testing regimen to obtain FDA approval to market. To ease generic entry, 20 however, the Act permits a generic drug manufacturer to file an Abbreviated New Drug 21 Application (ANDA) to piggyback on the approval efforts of the underlying brand-name drug 22 that uses the same active ingredient and to which the proposed generic is biologically 23 equivalent. The Act then implements a network of interlocking incentives to encourage faster 24 introduction of these low-cost generics into the pharmaceutical market, yet still drive new drug 25 development. See FTC v. Actavis, 570 U.S. 136, 142 (2013); 21 U.S.C. §§ 355(j)(2)(A)(ii), 26 (iv). 27 The ANDA relieves generic drug manufacturers of a significant burden by, in practical 1 aboard and profit from a brand manufacturers’ costly investment, the ANDA scheme relies on, 2 among others, patents to temporarily exclude generic manufacturers from use of the novel drug 3 ingredient itself or its new formulation, delivery mechanism, or form of treatment. H. 4 Hovenkamp, Anticompetitive Patent Settlements & the Supreme Court’s Actavis Decision, 15 5 MINN. J.L. SCI. & TECH. 3, 10–11 (2014). Recognizing that the long FDA approval process 6 would otherwise eat into a substantial portion of a patent’s twenty-year term, the Act provides 7 for up to a five-year extension of a drug patent’s term. New York ex rel. Schneiderman v. 8 Actavis PLC, 787 F.3d 638, 644 (2d Cir. 2015). The Act adds a further layer to drug-patent 9 protection, requiring an ANDA application to certify that the relevant patents have either 10 expired or will expire before the generic reaches the market, or that the patents are either 11 invalid or not infringed. Actavis, 570 U.S. at 143; 21 U.S.C. § 355(j)(2)(A)(vii)(I)–(IV). 12 Balancing this patent protection with its overall goal of expediting generic entry, the Act 13 encourages generic manufacturers to challenge weak validity or infringement cases. See King 14 Drug Co. v. SmithKline Beecham Corp., 791 F.3d 388, 394 (3d Cir. 2015). The first generic to 15 file an ANDA including a certification of invalidity or noninfringement (the “Paragraph IV” 16 approach) wins 180 days of generic exclusivity, meaning that once it gains FDA approval and 17 enters the market, the FDA can’t approve any other generics during that time. This, however, 18 doesn’t stop the brand from launching its own “authorized generic” to win some market share 19 back from the first filer, and the first filer can lose the exclusivity if fails to market promptly 20 after a later generic filer gains FDA approval. But the exclusivity period can be “worth several 21 hundred million dollars” to the first-filer generic and outweigh the risk of infringement suit. 22 Actavis, 570 U.S. at 143–44; In re Wellbutrin XL Antitrust Litig., 868 F.3d 132, 144 fn. 7 (3d 23 Cir. 2017); See Teva Pharm. v. Crawford, 410 F.3d 51, 55 (D.C. Cir. 2005); 21 U.S.C. 24 § 355(j)(5)(B)(iv), (D). 25 To ease timely judicial resolution of ANDA patent challenges, a Paragraph IV 26 certification operates as a technical act of patent infringement. And, to encourage prompt 27 commencement, the Hatch-Waxman Act imposes a thirty-month stay against FDA approval of 1 resolves the infringement suit during that time, the FDA follows that result. If not, the FDA 2 will then be free to approve the ANDA at the expiration of the statutory stay. Actavis, 570 3 U.S. at 143; 21 U.S.C. § 355(j)(5)(B)(iii); 35 U.S.C. § 271(e)(2)(A). 4 With FDA approval, a generic drug may enter the market. Though, if the infringement 5 suit remains ongoing, the launch is commonly referred to as “at risk,” given the brand 6 manufacturer might still win a judgment of infringement, damages (perhaps enhanced for 7 willfulness), and, where appropriate, an injunction. Wellbutrin, 868 F.3d at 168 fn. 59; 35 8 U.S.C. §§ 283–84. But once a generic drug makes it to market, the FDA’s Orange Book 9 broadcasts its availability to the public as an alternative to the underlying brand drug. All fifty 10 states and the District of Columbia have some form of mandatory or permissive drug 11 substitution law to help shift consumers from expensive brand drugs to their cheaper generics, 12 without another visit to the doctor. See Schneiderman, 787 F.3d at 644–45. 13 This scheme is supposed to expedite our access to cheaper generics. But the 14 pharmaceutical industry found a way around it. In some infringement cases, the brand 15 manufacturer (the patent owner) pays the generic (the accused infringer) to settle. This may 16 seem topsy-turvy but the patent owner actually comes out ahead. In exchange, the first-filer 17 generic manufacturer agrees to stay off the market for a few years. So, instead of expedited 18 generic entry, the brand maintains its monopoly and cuts the supposed-generic a share of the 19 profits. 20 In 2013, the United States Supreme Court found that these “pay for delay” schemes can 21 violate federal antitrust law. True, a patent gives a brand drug the power to exclude 22 competition within the bounds of the claims for its twenty-year term. But by paying the 23 accused infringer to stay out of the market, the brand in essence concedes the weakness of the 24 patent’s validity or infringement case against the ANDA. Thus, the settlement suppresses 25 generic competition and does so outside the bounds of the brand manufacturer’s patent 26 protection. See Actavis, 570 U.S. at 147–49, 153–58. 27 Our plaintiffs allege just a scheme here. The stipulated dismissal in 2012, which the 1 generic market entry, in fact concealed from the judge an extensive side agreement which 2 preserved the brand monopoly and kept Lupin’s generic Glumetza off the shelves for years. 3 Per the secret quid pro quo, dated February 22, 2012, Lupin agreed to walk away, leave 4 the patents alone, and delay market entry of its generic Glumetza for four years, until February 5 1, 2016. Instead of a simple cash payment to stay away, Depomed gave Lupin something else 6 of value — a guarantee that Depomed would not launch an authorized-generic to compete with 7 Lupin for one year, atop the FDA-granted 180 days of market exclusivity from other generic 8 competition. The settlement also included two clauses to protect Lupin from other generic 9 competition, providing, first, that if any other manufacturer succeeded in marketing a generic 10 Glumetza before February 2016, Lupin could market immediately, and, second, that Depomed 11 would not license any other generic Glumetza manufacturers until at least 180 days following 12 Lupin’s market entry. These provisions undercut the incentive for any other generic 13 manufacturer to enter the market before Lupin. But, as noted above, only the first of these 14 terms, Lupin’s February 2016 market entry, made it into the parties’ stipulated dismissal. 15 Crucially, as alleged, defendants did not disclose the no-authorized generic (no-AG) provision 16 until a February 5, 2016, quarterly earnings call with Lupin executives. 17 This artificially-prolonged monopoly and synthesized duopoly proved a windfall to our 18 defendants. Depomed had already sold its marketing rights in Glumetza to defendant Santarus, 19 Inc., but then sold its royalty rights, estimated between $140 million and $180 million, to PDL 20 Biopharma in October 2013 (Dkt. No. 464-1). The juiced Glumetza portfolio aided Santarus’ 21 sale to defendant Salix Pharmaceuticals for $2.6 billion in November 2013, and Salix’s sale to 22 defendant Valeant Pharmaceuticals for $14.5 billion in April 2015. 23 Then, in the summer of 2015, Valeant hiked the price of Glumetza on the order of 800%. 24 The wholesale acquisition cost of a 500 mg pill spiked from $5.72 in May to $51.48 by the end 25 of July. In the same time, the cost of a 1000 mg pill spiked from $12.37 to $113.36. Glumetza 26 accounted for a three hundred fifty million dollar swell in Valeant’s profits in the second half 27 of 2015. When Lupin joined the market in February, the supposed affordable generic did so at 1 around $45 for each 500 mg tablet. All in all, Lupin reaped to the tune of six hundred million 2 dollars from Glumetza over the next few years. 3 $10,000 $10,020 July 2015 4 $8,000 5 $6,680 June 2015 w 56,000 6 = $5,148 July 2015 7 $4,000 $3,432 July 2015 8 $2,000 $1,113 May 2015 9 50 $572 May 2015 Jan-10 Jan-11 Jan-12 Jan-13) Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 10 —Glumetza S0OMG WAC Price —-Glumetza 1000MG WAC Price i qa 12 Only adding to their spoils, defendants’ scheme apparently dissuaded other generic

13 manufacturers from marketing their own generic versions of Glumetza until after Lupin did.

v 14 Sun Pharmaceuticals filed its ANDA in March 2011. But Depomed and marketing-partner © 15 Santarus promptly sued, and a January 2013 settlement kept Sun’s generic off the market until 16 August 2016. Watson Pharmaceuticals then filed its ANDA in March 2012. Another prompt

= 17 lawsuit resulted in a November 2013 settlement keeping Watson’s generic also off the market

O Z 18 until August 2016. In the end, Watson (by then Teva Pharmaceutical Industries Ltd.) did not 19 market its generic Glumetza until May 2017, and Sun not until July 2018. 20 Glumetza prices have to some extent slowly recovered, but not to pre-hike estimates. 21 The most immediate fallout for our defendants? Bad press. So, in 2018, Depomed changed its 22 name to Assertio Therapeutics, Inc., and Valeant changed its name to Bausch Health 23 Companies Inc. 24 Plaintiffs started suing in August 2019. Ultimately, twelve cases by both direct and 25 indirect purchasers arrived before the undersigned. An order dated March 5, 2020, largely 26 denied defendants’ motions to dismiss the direct purchasers but granted the motions in part 27 against the indirect purchasers, who all subsequently dismissed. In re Glumetza Antitrust 28 Litig., No. C 19-05822 WHA, 2020 WL 1066934 (N.D. Cal. Mar. 5, 2020). The remaining

1 direct purchasers, several who sued as assignees of absent direct purchasers, have proceeded in 2 two groups: (1) the direct-purchaser class, and (2) the retailer plaintiffs. An August 15 order 3 certified the Rule 23(b)(3) direct-purchaser class comprising:

4 All persons or entities in the United States and its territories who directly purchased Glumetza or generic Glumetza from a defendant 5 from May 6, 2012 until the date of [the] order. 6 In re Glumetza Antitrust Litig., 336 F.R.D. 468 (N.D. Cal. 2020). Days prior, plaintiff Humana 7 Inc. attempted to resurrect the slew of indirect-purchaser claims. Orders dated December 5 and 8 February 2, 2021, rejected that ploy. 9 Over the past three months, our parties have filed ten Daubert motions and four motions 10 for summary judgment. This order resolves the motions for summary judgment, reserves the 11 Daubert motions for later, and follows full briefing and oral argument held via 12 videoconference due to COVID-19. 13 ANALYSIS 14 Section 1 of the Sherman Act prohibits “[e]very contract, combination in the form of 15 trust or otherwise, or conspiracy, in restraint of trade or commerce among the several States.” 16 Put plainly, it bars (1) concerted conduct which (2) unreasonably restrains trade. F.T.C. v. 17 Qualcomm Inc., 969 F.3d 974, 989 (9th Cir. 2020). An unreasonable restraint comes either as 18 a per se violation (“when surrounding circumstances make the likelihood of anticompetitive 19 conduct so great as to render unjustified further examination”) or a violation of the rule of 20 reason, meaning the “restraint’s harm to competition outweighs its procompetitive effects” in 21 light of the peculiarities of the business, the nature and history of the restraint, and its reason 22 for being. NCAA v. Bd. of Regents of Univ. Okla., 468 U.S. 85, 103–04 (1984); Tanaka v. 23 Univ. of S. Cal., 252 F.3d 1059, 1063 (9th Cir. 2001); United States v. Topco Assocs., 405 U.S. 24 596, 607 (1972). In pharmaceutical reverse-payment cases, such as ours, we apply the rule of 25 reason. Actavis, 570 U.S. at 158–59. 26 Section 2 condemns “[e]very person who shall monopolize, or attempt to 27 monopolize . . . any part of the trade or commerce among the several States.” Unlike Section 1 conduct, but only that which threatens to or actually monopolizes. Alaska Airlines v. United 2 Airlines, 948 F.2d 536, 541 (9th Cir. 1991). Such violation involves the (1) possession of 3 power in the relevant market following (2) its willful acquisition or maintenance and (3) an 4 injury resulting from abuse or leverage of that market power. This causal-harm element 5 emphasizes the distinction between, on the one hand, the unlawful acquisition of market power 6 and, on the other, its “development as a consequence of a superior product, business acumen, 7 or historic accident.” Qualcomm, 969 F.3d at 990 (quoting United States v. Grinnell Corp., 8 384 U.S. 563, 570–71 (1966)). 9 Our threshold inquiry asks whether defendants wielded power over the relevant market, 10 “the area of effective competition.” Section 2 carries this demand on its face; a monopolist 11 unifies the market for a given good and wields the resulting anticompetitive power. Though 12 less obvious a requirement of Section 1, “courts usually cannot properly apply the rule of 13 reason without an accurate definition of the relevant market,” as the analysis requires our 14 location of defendants within the market structure to gauge the results of their anticompetitive 15 conduct. Ohio v. Am. Express Co., 585 U.S. ___, 138 S. Ct. 2274, 2285 (2018); Qualcomm, 16 974 F.3d at 989–92; Rebel Oil Co. v. ARCO, 51 F.3d 1421, 1434 (9th Cir. 1995). 17 Plaintiffs establish the antitrust violation itself via a tripart burden shift, beginning with a 18 prima facie concerted restraint under Section 1 or willful or attempted monopoly under Section 19 2. The defendant must then offer a nonpretextual procompetitive rationale for the restraint or 20 the monopolization, which plaintiffs must either rebut or show, under Section 1, that a less 21 anticompetitive means would achieve the same ends, or under Section 2, that the 22 anticompetitive harm outweighs the procompetitive benefit. See Qualcomm, 974 F.3d at 991; 23 Image Tech. Serv. v. Eastman Kodak Co., 903 F.2d 612, 618–19 (9th Cir. 1990), aff’d, 504 24 U.S. 451 (1992). 25 If plaintiffs establish the antitrust violation, they must then prove their causal antitrust 26 injury, i.e., an injury both of the kind the antitrust laws seek to prevent and which flows from 27 the unlawfulness of defendants’ conduct, and offer a fair estimate of their damages. Brunswick 1 U.S. 251, 264–65 (1946); Qualcomm, 974 F.3d at 989–92; In re New Motor Vehicles Canadian 2 Export Antitrust Litig., 522 F.3d 6, 19 fn. 18 (1st Cir. 2008). 3 This order denies plaintiffs’ motion for partial summary judgment. True, a reasonable 4 trier of fact could find that defendants wielded market power during the relevant time period. 5 Compelling evidence illuminates classic indicators of power in the market for brand and 6 generic Glumetza. In 2015, Bausch profitably hiked prices and reduced consumption. Lupin 7 profitably maintained those supracompetitive prices for several years before competition with 8 the authorized generic, Teva, and Sun finally brought prices back down. But a reasonable trier 9 could reject these considerations in light of defendants’ evidence of the cross-elasticity of 10 demand between Glumetza and the many, and more prevalent, variants of metformin 11 hydrochloride available to consumers. 12 This order also denies defendants’ motions for summary judgment. A reasonable trier of 13 fact could conclude that defendants violated the rule of reason. The challenged settlement 14 included cross-covenants not to compete, one from Lupin to pause its generic Glumetza entry 15 for four years, and the other from Assertio and Santarus to forego an authorized generic for one 16 year. These horizontal restraints on competition mirrored classic per se antitrust violations. 17 Defendants, however, contend these restraints fell within the scope of their patent rights. But 18 given evidence calling into question the strength of defendants’ infringement case, a 19 reasonable trier of fact could find defendants’ covenants exceeded those bounds. 20 A reasonable trier of fact could further conclude that defendants’ restraint of the market 21 delayed generic entry, stifled competition, and caused plaintiffs to pay more for brand and 22 generic Glumetza than they would have otherwise. In the absence of the unlawful scheme, 23 Lupin might have marketed its generic earlier via either an at-risk launch, by prevailing in the 24 underlying suit, or by negotiating an earlier market entry date in the absence of the no-AG 25 agreement, and other generics might have followed quickly thereafter. This earlier generic 26 competition would have prevented or mitigated defendants’ price gouging. A reasonable trier 27 of fact could also find plaintiffs’ suit timely and reject both Assertio and Lupin’s solo defenses. 1 1. DEFENDANTS’ MARKET POWER. 2 A reasonable trier of fact could find that defendants’ wielded power in the market for 3 brand and generic Glumetza. It could also find otherwise. Up front, no one doubts that the 4 market for brand Glumetza included both its authorized generic and those by Lupin, Sun, and 5 Teva. Drug substitution laws nationwide permit or even require pharmacies to swap these 6 cheaper generics in place of a brand drug. Schneiderman, 787 F.3d at 644–45, 657; ORANGE 7 BOOK at vii–xvi, 3-284. If our relevant market stopped there, then defendants’ market power 8 would likely follow. Until the authorized generic joined in February 2017, our defendants 9 retained 100% market share, the FDA approval regime posed a significant barrier to market 10 entry, and others like Sun and Teva proved unable to ramp up production until mid-2017 and 11 2018. Rebel Oil Co. v. ARCO, 51 F.3d 1421, 1434 (9th Cir. 1995). 12 Our parties, however, dispute whether the relevant market also encompassed other 13 variants of metformin hydrochloride, such as Glucophage, Glucophage XR, and Fortamet 14 (including their generics). This question of fact involves a broad inquiry into “the particular 15 structure and circumstances” of the pharmaceutical context and into the cross-elasticity of 16 demand between Glumetza and its proposed substitutes, including empirical study, to 17 determine which, if any, would have been adequate and available enough to constrain prices. 18 Rebel, 51 F.3d at 1434–35; Verizon Comm’ns Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 19 U.S. 398, 411 (2004); Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962); Times- 20 Picayune Publ. Co. v. United States, 345 U.S. 594, 612 fn. 31 (1953). 21 In the dysfunctional pharmaceutical market, the doctor who chooses the drug does not 22 pay for it, and the consumer who pays does not choose. Insurance obfuscates the price the 23 consumer sees. The prescription requirement, by corollary, limits alternatives. And, consumer 24 “demand” for pharmaceutical drugs often manifests as degrees of desperation, because we 25 consume these drugs to prevent illness, ease suffering, and even forestall death. See 26 Schneiderman, 787 F.3d at 645–47; Mylan Pharms. v. Warner Chilcot Pub. Ltd. Co., 838 F.3d 27 421, 429–30 (3d Cir. 2016); F.M. Scherer, The Pharmaceutical Industry, in I HANDBOOK OF 1 This all shifts bargaining power to the drug companies, but it only tells part of the story. 2 Defendants’ expert testimony indicates that physicians both account for price in prescribing 3 and view Glumetza as therapeutically interchangeable with the numerous, and more prevalent, 4 metformin variants (Glumetza has only ever accounted for about a two and a half percent 5 share). Drug manufacturers then compete behind the scenes for preferred placement on health 6 plan formularies which govern drug insurance coverage. 7 Both sides’ experts also attack cross-elasticity empirically. Plaintiffs’ expert submits 8 studies indicating little to no cross-elasticity between Glumetza and other metformin variants. 9 Defendants’ expert presents the contrary. Plaintiffs’ Daubert motion may narrow this battle of 10 the experts, but the breadth of this inquiry ensures that sufficient, though perhaps lopsided, 11 disputes will remain. Ultimately, the balancing of this evidence must be left to our trier of fact. 12 See Primiano v. Cook, 598 F.3d 558, 564–65 (9th Cir. 2010); Provenz v. Miller, 102 F.3d 13 1478, 1490 (9th Cir. 1996). 14 Seeking to bypass this maze, plaintiffs offer direct evidence of defendants’ market power. 15 True enough, market power asks “whether an arrangement has the potential for genuine 16 adverse effects on competition.” Under either Section 1 or 2, “proof of actual detrimental 17 effects . . . can obviate the need for an inquiry into market power.” F.T.C. v. Ind. Fed’n 18 Dentists, 476 U.S. 447, 460–61 (1986); United States v. Microsoft, 253 F.3d 34, 51 (D.C. Cir. 19 2001) (en banc). 20 Plaintiffs point to the “hallmarks of market power” — defendants’ profitable 21 maintenance of supracompetitive pricing and reduced output for several years. Bausch’s 2015 22 price hike, which plaintiffs characterize as an 800% increase, though defendants would place 23 the number around 275%, drove thousands of consumers from Glumetza to (we infer in 24 defendants’ favor) Glucophage and Fortamet, whose producers ably supplied the new demand. 25 Meanwhile, Bausch reaped three hundred fifty million dollars more in profit over the latter half 26 of 2015. Lupin joined the market at a similar premium in February 2016, quickly captured the 27 brand consumer base, and maintained that profitable supracompetitive pricing for several years 1 1988); NCAA, 468 U.S. at 85-86; William M. Landes & Richard A. Posner, Market Power in 2 Antitrust Cases, 94 HARV. L. REV. 937 (1981). AG 3 $100 Price Per Lupin Entry, Teva Sun 4 $90 Unit Entry; Entry Entry 5 $80 s-+>-,Glumetza WAC nin 3 i” awe

6 $70 nee Glumetza Brand Price $60 \ 7 $50 8 Glumetza)GenericPrice $40 $30 coeeeeeeeeeel weneest 10 $20 i veda OV $10 □□ aces ' ! 12 op pam ! & 13 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | Prices & WAC of Extended-Release and 14 Immediate-Release Metformin 2010-2019

15 Though the Supreme Court approved of direct proof of market power in Indiana 16 Federation, there remains a broad gap between affirmance of a direct factual finding of market

3 17 power versus plaintiffs’ request here to take that question from the jury. Plaintiffs cite, and this

5 18 order is aware of, no appellate decision stemming from Indiana Federation affirming the leap 19 plaintiffs’ seek. Additionally, Indiana Federation did, in fact, roughly define the market, 20 emphasizing the locality of dentistry services and defendants’ majority share in two specific 21 cities. Defendants’ evidence of the cross-elasticity of demand between Glumetza and other 22 metformin variants, particularly given Glumetza’s small role in the broader metformin array, 23 precludes our reliance on even a rough market definition here. 476 U.S. at 460-61; Oltz, 861 24 F.2d at 1448. 25 A reasonable trier of fact could exclude other versions of metformin and include only 26 Glumetza and its generics in our market. The reasonable trier could also find that defendants’ 27 have already wielded anticompetitive market power. But as our court of appeals has observed: 28

The ultimate issue in a rule of reason case is whether a challenged 1 practice will produce adverse effects on price or output. The only direct way to answer that question is to introduce evidence of 2 actual price increases or reductions in output after the challenged practice. But even if a plaintiff is lucky enough to gather such 3 evidence, he will face the momentous task of proving that the observed price or output effects were not attributable to any one of 4 an infinite number of independent causes: exhaustion of raw materials, increases in labor costs, increases in the price of 5 substitute goods, tax hikes, etc. Situations will arise where a plaintiff is able to meet this burden. But doubtless those occasions 6 will be rare. 7 E.W. French & Sons, Inc. v. General Portland Inc., 885 F.2d 1392, 1404 (9th Cir. 1989) 8 (Farris, J., concurring) (cleaned up). On this record, material questions of fact remain. 9 2. DEFENDANTS VIOLATED THE RULE OF REASON. 10 Armed with a finding that our relevant market encompassed Glumetza and its generics, 11 and that defendants had power therein, a reasonable trier of fact could conclude that 12 defendants’ concealed settlement violated the rule of reason. In February 2012, faced with the 13 prospect of Lupin’s generic Glumetza entry as early as that May, our defendants conspired to 14 allocate the market for brand and generic Glumetza between them for the foreseeable future. 15 They first agreed to maintain Assertio and Santarus as the sole producer for four more years, 16 delaying Lupin’s generic entry until February 1, 2016. And, our brand defendants secured 17 Lupin’s assent to this delay with a covenant not to compete, the promise not to launch an 18 authorized-generic version of Glumetza until February 1, 2017. 19 Following discovery, we’ve got the actual agreement. The terms of this exchange appear 20 right there on its face. The delay:

21 Effective upon the License Effective Date [February 1, 2016], Depomed hereby grants to Lupin, and Santarus hereby consents to 22 and authorizes the grant to Lupin of, a fully paid up, royalty free, non-exclusive (except as otherwise provided herein) license with 23 respect to the Depomed Patents. 24 And the payment:

25 Depomed, Santarus and their Affiliates shall not, and shall not authorize any Third Party to, Market an Authorized Generic any 26 earlier than twelve (12) months after the License Effective Date. 27 (Dkt. No. 493-12, §§ 3.1, 3.5). Or, in more classic antitrust terms, we have one covenant not to 1 not to compete, one year without marketing an authorized generic. In most cases, we could 2 stop right here. Such naked horizontal restraints — an agreement to allocate a market, such as 3 in time or space, between a select few competitors at the same level of the market — are a per 4 se Section 1 violation. See Topco, 405 U.S. at 607–08. The Court of Appeals for the Sixth 5 Circuit held as much in one of the earliest appellate treatments of a pharmaceutical reverse- 6 payment settlement. See In re Cardizem CD Antitrust Litig., 332 F.3d 896, 907–08 (6th Cir. 7 2003), abrogated by Actavis, 570 U.S. at 158–60. Even defendants do not attempt to convince 8 us that their agreement was not such a horizontal allocation. 9 But, as defendants take care to note, we have a twist. By the patent laws, Congress has 10 expressly sanctioned certain conduct that would otherwise constitute even a per se violation. 11 A patent grants its owner the right to exclude others from practicing the claimed invention. 12 Now, a patent owner may let the world make, use, or sell the invention. It may, in exchange 13 for royalty payments, license to as many producers as necessary to create a competitive market. 14 But the patent owner may also monopolize or restrain the market for the invention. It may 15 license only a few producers. Or just one. Indeed, a patent owner may exclude all others from 16 practicing the invention, and may even refrain from practice itself, thereby suppressing the 17 invention from the market entirely. 35 U.S.C. §§ 271, 283. As the Supreme Court recognized 18 in NCAA, certain contexts require courts, even those presented with otherwise per se 19 violations, to step back to the rule of reason. That is why here, even faced with a naked 20 horizontal restraint, we must step back to evaluate the restraint in context to then determine 21 whether, as defendants urge, it fell within the bounds of their patents. Actavis, 570 U.S. at 22 147–51, 158–59; NCAA, 468 U.S. at 100–01. 23 In arguing that the settlement agreement toed the line their patents drew, defendants 24 would have us presume infringement and place the burden on plaintiffs to show otherwise. 25 This argument runs afoul of patent fundamentals. True, we presume a patent’s validity. See 26 Microsoft Corp. v. i4i Ltd. Partnership, 564 U.S. 91, 100 (2011). But we also presume 27 noninfringement. The patent owner always bears that burden. In other words, we presume a 1 producer’s right to market a good unless and until shown otherwise. Medtronic v. Mirowski 2 Family Ventures, 571 U.S. 191, 198–99, 203 (2014). 3 The ANDA framework doesn’t change this baseline; it illustrates it. The Paragraph IV 4 certification just permits suit. It tells us nothing about whether the proposed generic, in fact, 5 infringes the listed patent. Eli Lilly & Co. v. Medtronic, Inc., 496 U.S. 661, 678 (1990); Glaxo, 6 Inc. v. Novopharm, Ltd., 110 F.3d 1562, 1569 (Fed. Cir. 1997). The FDA is free to approve a 7 Paragraph IV ANDA one hundred eighty days after filing, and the generic manufacturer will 8 be free to market the drug unless the brand manufacturer sues within the forty five days and 9 triggers the thirty-month stay. But, if the stay expires without a judgment of infringement, the 10 FDA again becomes free to approve the ANDA, and the generic manufacturer would be free to 11 market. See Actavis, 570 U.S. at 143; 21 U.S.C. § 355(j)(5). At that point, it would take a 12 preliminary injunction in the shorter term or a judgment of infringement (or the parties’ joint 13 assessment of infringement) in the longer term to take the generic off the market. In other 14 words, the brand patents alone bar nothing. The stay does at first. But after that, a judgment of 15 (or settlement admitting to) infringement must do it. 16 These fundamentals fit within the rule of reason’s burden-shifting framework. After 17 plaintiffs establish an antitrust violation, the burden falls to the defendants to offer a 18 procompetitive rationale, which, if non-pretextual, plaintiffs must rebut. It follows, then, that 19 to traverse a motion for summary judgment raising defendants’ patents as the procompetitive 20 justification, plaintiffs, as several courts have held, must show “some evidence” of 21 noninfringement or invalidity. See Wellbutrin, 868 F.3d at 165; Nexium, 842 F.3d at 63; see 22 also Teikoku Pharma, 296 F. Supp. 3d at 1152–55. But noninfringement remains our baseline. 23 Some anticompetitive conduct will fall within the scope of the patent; some will not. This is 24 why the Supreme Court explained in Actavis that “[w]hether a particular restraint lies beyond 25 the limits of the patent monopoly is a conclusion that flows from [the antitrust] analysis and 26 not, as [both our defendants and] the Chief Justice suggest[ed], its starting point.” See 570 27 U.S. at 149 (quotes omitted); Medtronic, 571 U.S. at 198–99; Qualcomm, 969 F.3d at 991. 1 As already stated, plaintiffs have carried their initial burden. Defendants’ pay-for-delay 2 settlement agreement contained two cross-covenants not to compete within the market for 3 Glumetza. These horizontal restraints place the burden upon defendants to show they go no 4 further than their patents authorized. Our parties appear to agree that direct evidence of the 5 parties’ mental states will suffice, given that neither offers a limitation-by-limitation analysis 6 of Lupin’s ANDA in their briefs. This, of course, follows from the Supreme Court’s 7 recognition that “it is normally not necessary to litigate patent validity” — and infringement, 8 by context — “to answer the antitrust question.” Actavis, 570 U.S. at 157. It also follows from 9 the pivotal role that intent plays in antitrust violations. A multiparty concerted restraint 10 certainly requires some degree of intent. Cf. Alaska Airlines, 948 F.2d at 540–41. And, as the 11 Supreme Court has noted, “no monopolist monopolizes unconscious of what he is doing.” 12 Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 602 (1985). 13 This may seem a difficult logical leap, but patents already provide a common example of 14 an antitrust analysis which avoids any limitation-by-limitation invalidity or infringement 15 analysis, patent licensing. The license to market a good on payment of a royalty is a horizontal 16 restraint on production of that good. But no one would challenge ordinary patent licensing as a 17 violation of the antitrust laws because we understand that the parties’ good-faith belief of 18 infringement (or their belief that’s a reasonable enough possibility) places the agreement 19 within the scope of the patent. The royalty payment from the licensee to the patent owner 20 flows as expected and, in contrast to most restraints, a patent license brings a product to 21 market. This abbreviated rule of reason justifies ordinary patent licensing without a limitation- 22 by-limitation infringement analysis. 23 So too here, defendants contend, arguing that both parties believed Lupin’s ANDA 24 infringed the relevant patents. At first glance, they tell a compelling story. Lupin, defendants 25 say, knew its ANDA infringed the patents early on. Outside counsel said such a result was 26 likely from the beginning, and once Judge Hamilton adopted claim constructions which both 27 undermined Lupin’s invalidity argument and made a judgment of infringement very likely, 1 challenged settlement agreement manifested an intent to accelerate Lupin’s generic entry and 2 ran afoul of no antitrust concerns. 3 Our jury would be free to reject this narrative as a pretext. Questions of fact appear on its 4 face, starting with the glaring oddity that only one of our three defendants involved in the 5 underlying suit, Lupin, has waived attorney-client privilege to tell its tale of certain 6 infringement. Besides, if the patents were so valid and infringed, why did Assertio and 7 Santarus let Lupin practice them royalty free for, depending on the patent, several months to 8 four years? And, if both parties so strongly believed Lupin’s ANDA infringed, why did they 9 stipulate that there remained “significant risk to each of them” given Judge Hamilton “ha[d] 10 not adjudicated Depomed’s charges of patent infringement or Lupin’s defenses and 11 counterclaims[?]” Depomed, No. C 09-05587, Dkt. No. 152. 12 More important, though, defendants’ telling forgets a key fact — their actual settlement 13 agreement. Defendants would have us believe that both parties knew Lupin infringed and 14 would be barred from marketing a generic Glumetza for several years. And yet, to settle the 15 suit, our brand defendants agreed to expedite Lupin’s generic market entry, cede a year’s worth 16 of revenue that an authorized generic could recoup from Lupin (recall, the Supreme Court has 17 recognized that even 180 days of generic exclusivity can be worth several hundred million 18 dollars), and pay three million dollars in Lupin’s legal fees (Dkt. No. 493-12, §§ 2.5, 3.1, 3.5). 19 In other words, defendants would have us believe that they all knew Lupin was going to lose 20 — yet Lupin emerged far and away the settlement victor? Our jury could reject these facial 21 contradictions outright. See Kodak, 903 F.2d at 618–19. 22 Contrary to Actavis, defendants contend that here the large and otherwise unexplained 23 payment to Lupin cannot be taken as suggesting that Assertio and Santarus doubted their 24 infringement case. Their first basis for this argument, that the FTC labors under a lower 25 causation standard than does a private plaintiff, fails to explain how a different burden of 26 antitrust causation would change our analysis of whether defendants violated the antitrust 27 laws. Indeed, nothing in the rationale of Actavis turned on the identity of the FTC as the 1 Food & Commercial Workers Local 1776 & Participating Employers Health & Welfare Fund 2 v. Teikoku Pharma USA, 296 F. Supp. 3d 1142, 1154 (N.D. Cal. 2017) (Judge William H. 3 Orrick). In a similar vein, defendants say that direct evidence of Lupin’s confidence (recall, 4 Assertio and Santarus did not also waive attorney-client privilege here) in the infringement 5 case displaces the use of the payment as “proxy” for their doubt. This, however, ignores the 6 plain language of Actavis. An unexplained payment “suggest[s] that the patentee has serious 7 doubts” about its infringement case. Such evidence probes defendants’ states of mind just as 8 their later, direct testimony does. 570 U.S. at 157. 9 Even if our jury does not go so far as to find defendants’ justification a pretext, plaintiffs 10 do present evidence that defendants doubted their infringement case. At bottom, we cannot 11 forget defendants’ own explicit admission of that doubt to Judge Hamilton, acknowledging in 12 their consent decree that “there [was] significant risk to each of them associated with the 13 continued prosecution of [the case.]” Depomed, No. C 09-05587 PHJ, Dkt. No. 152. Our jury 14 will be entitled to place a certain amount of weight upon defendants’ contemporaneous 15 admissions to a federal judge. 16 Plaintiffs offer several batches of evidence, but sufficient here is the noninfringement 17 testimony of Lupin’s expert in the underlying litigation. In a summer 2011 draft 18 noninfringement report, never submitted due to the settlement, Dr. Kinam Park applied Judge 19 Hamilton’s claim construction and explained the two different methods of controlling the 20 extended release of metformin. Assertio’s patents claimed the “matrix” form, where dispersion 21 of the active drug within a polymeric matrix controls its release. Lupin’s ANDA used the 22 other form, a “reservoir,” which surrounded the active drug in a membrane or coating to 23 control its release. Indeed, plaintiffs say, evidence of Lupin’s settlement negotiation 24 preparations shows Lupin believed and acted upon Dr. Park’s noninfringement report. Earlier 25 in the summer Lupin had drafted an opening settlement offer with a May 2014 entry date in 26 exchange for a royalty to Assertio. After Dr. Park’s initial draft report circulated, Lupin 27 revised its draft settlement offer, deleting the royalty payments and moving up its proposed 1 Instead of responding to the merits of Dr. Park’s noninfringement report, defendants 2 attack its weight and admissibility. That Dr. Park’s report was unsigned and never tested in the 3 underlying litigation says little. None of the infringement or noninfringement arguments were 4 tested in the underlying litigation, as the parties admitted in their settlement stipulation. 5 Defendants’ frustrations with Dr. Park’s memory of the circumstances surrounding his report 6 go to the weight of his testimony, not its admissibility. And, plaintiffs will be more than 7 welcome to cite Lupin’s prior inconsistent statements against it at trial. Fed. R. Evid. 801. 8 Defendants also contend that plaintiffs misclassified Dr. Park as a Rule 26(a)(2)(C) 9 nonreporting expert. Not so. Rule 26(a)(2)(C) permits certain non-retained experts to testify 10 as percipient witnesses. This usually arises in the context of a “treating physician” who may 11 testify to opinions formed during the course of medical treatment, in contrast to a medical 12 expert retained to review and opine on medical records provided by counsel and who must 13 provide a written report. Here, Dr. Park will act more as a percipient witness, testifying to the 14 circumstances surrounding and the opinions contained in his previous report. Plaintiffs have 15 not retained Dr. Park, and he will not opine on our present record. Rule 26 requires no new 16 report. And, defendants cannot claim prejudice, given they have had Dr. Park’s report for 17 nearly a decade. See Goodman v. Staples, 644 F.3d 817, 824–26 (9th Cir. 2011). 18 It’s also worth noting the ridiculousness of defendants’ attempts to exclude Dr. Park’s 19 report. Defendants tout Lupin’s decision to waive attorney-client privilege in the underlying 20 litigation ad nauseum. Aside from the fact that this ploy presents a facially incomplete record 21 (Assertio and Santarus did not also waive attorney-client privilege), defendants cannot both 22 proclaim Lupin’s transparency and bury inconvenient portions of Lupin’s underlying record. 23 In sum, on this record a reasonable trier of fact could conclude that defendants 24 unreasonably restrained the production of Glumetza via cross-covenants not to compete and 25 that that these restraints did not fall within the scope of their patents. 26 3. PLAINTIFFS’ CAUSAL ANTITRUST INJURY. 27 Armed with the findings that defendants’ settlement violated the rule of reason and that 1 settlement caused plaintiffs’ antitrust injury. “It is generally sufficient to show with 2 reasonable probability some causal connection between the antitrust violation and [the claimed 3 injury.]” Catlin v. Wash. Ener. Co., 791 F.2d 1343, 1347 (9th Cir. 1986). The Supreme Court 4 has long held that a “jury [may] conclude as a matter of just and reasonable inference from the 5 proof of defendants’ wrongful acts and their tendency to injure plaintiffs’ business, and from 6 the evidence of the [effects], not shown to be attributable to other causes, that defendants’ 7 wrongful acts had caused damage to the plaintiffs.” Bigelow, 327 U.S. at 264. In other words, 8 plaintiffs need not rule out “all possible alternative sources of injury,” but satisfy the inquiry by 9 showing the illegality “to be a material cause of the injury.” Zenith Rad. Corp. v. Hazeltine 10 Rsch., Inc., 395 U.S. 100, 114 fn. 9 (1969). 11 Plaintiffs’ claimed injury is the overcharge they actually paid compared to the price they 12 would have paid absent the unlawful settlement. As detailed at class certification:

13 Generic drugs crash brand monopolies, entering the market at significant discounts. Studies show this discount increases over 14 time (and as more generics enter the market) and quickly pervades the market as the generics capture the prescription base through 15 permissive or (often, state or locally) mandated generic- substitution at the pharmacy. And, this generic market penetration 16 increases over time. 17 336 F.R.D. at 476. Delayed generic entry delays the resultant price cuts. So plaintiffs’ theory 18 of causation distills to this: defendants delayed Lupin’s generic entry. Absent that delay, Lupin 19 would have marketed a generic version of Glumetza earlier. The price-suppressing 20 competition which (as seen above) actually played out over the last several years would have 21 arrived sooner, cutting prices earlier. Thus, at each point in time, plaintiffs paid more for brand 22 or generic Glumetza than they should have. 23 This certainly counts as a valid antitrust injury; “price cutting is a practice the antitrust 24 laws aim to promote.” Cascade Health Solutions v. PeaceHealth, 515 F.3d 883, 896 (9th Cir. 25 2008). Actavis itself recognized the overcharge as the relevant harm. 570 U.S. at 157. 26 Defendants (with one exception, detailed below) concede this cognizable injury; instead they 27 challenge the causation itself, arguing that generic competition would have come to the market 1 Plaintiffs offer three scenarios of earlier generic competition, positing that either Lupin 2 would have prevailed in the underlying suit, entered the market at-risk, that is, upon FDA 3 approval following the expiration of the 30-month stay even as the underlying litigation 4 continued, or that Lupin and our brand defendants would have agreed to an earlier entry date if 5 their settlement agreement had not included the unlawful no-AG provision. Plaintiffs also 6 argue that both Sun and Watson (now Teva) would have entered the market earlier on Lupin’s 7 heels, further speeding generic competition and reduced prices. 8 Defendants’ running argument on this front is that plaintiffs can point to no evidence that 9 defendants, or their officers, or any third parties contemplated any of plaintiffs’ but-for 10 scenarios. In other words, defendants would premise causation here (and thus their liability) 11 on the notion that they must have contemplated a lawful course of action before abandoning it 12 for the unlawful. This is both wrong on the law — “[t]he most elementary conceptions of 13 justice and public policy require that the wrongdoer shall bear the risk of the uncertainty which 14 his own wrong has created” — and in reasoning — our actors have no reason to contemplate 15 action they might have taken in the but-for world which never came to pass. Our jury will 16 craft the but-for world, in which defendants acted rationally and lawfully, based on measured 17 deviation from the record. See Bigelow, 327 U.S. at 265; Dolphin Tours, Inc. v. Pacifico 18 Creative Service, Inc., 773 F.2d 1506, 1511 (9th Cir. 1985). 19 First, plaintiffs offer cogent evidence that Lupin may have marketed their generic 20 Glumetza at risk or upon prevailing in the underlying suit. As above, defendants’ patents may 21 justify their agreement. But noninfringement remains our baseline. See, e.g., Nexium, 842 22 F.3d at 63; Medtronic, 571 U.S. 198–99. With this understanding, then, this order agrees with 23 defendants that “[t]he clear import of Nexium and Wellbutrin is that a plaintiff must offer some 24 evidence of non-infringement or patent invalidity in order to proceed on an at-risk launch 25 theory of causation.” Apotex, Inc. v. Cephalon, Inc., 255 F. Supp. 3d 604, 614 (E.D. Pa. 2017) 26 (Judge Mitchell S. Goldberg). 27 Again, the circumstances of the settlement itself betray defendants’ doubts in the 1 Park’s noninfringement analysis, this without even addressing plaintiffs’ expert evaluation of 2 Lupin’s noninfringement case. Whether this and plaintiffs’ other evidence of noninfringement 3 would have supported a judgment of noninfringement, traversed a preliminary injunction, or 4 encouraged a rational actor to market at risk presents a question of evidence weighing and 5 balancing reserved for the jury. 6 We also have evidence that Lupin, if it could have outlasted Assertio and Santarus, would 7 have pressed that advantage. Congress designed the ANDA scheme to encourage generic 8 entry, especially if Lupin believed in its noninfringement case. The 180 days of generic 9 exclusivity, often worth several hundred million dollars, would have paid Lupin’s legal fees 10 and then some. Indeed, here Lupin’s actual 12-month exclusivity period generated a gross 11 margin of over one hundred twenty million dollars (Dkt. No. 493-79 at 2). 12 More important, in thirty seven times as the first filer, plaintiffs count four times that 13 Lupin has launched at risk, including another generic version of metformin, our active drug 14 here. Plaintiffs also highlight that Lupin, to this point, had never quit a patent suit when it was 15 the first ANDA filer. And, as late as nine days before the challenged settlement, Lupin had 16 still been internally circulating production calendars listing a generic Glumetza launch for as 17 early as November 2012. Evidence shows that Lupin may have been able to ramp-up its 18 production in time for such a launch. Between October and December 2011, Lupin made three 19 commercial-size batches of generic Glumetza, and only needed three months to produce the 20 validation batches for its February 2016 launch (Dkt. No. 493-1 at ¶¶ 45–51). 21 Defendants contest this evidence, in their telling Lupin only launched at risk once, and 22 minimize the proffered production calendars as mere placeholders carrying little weigh. But as 23 above, plaintiffs need not show that the Lupin of our world actually contemplated a given 24 scenario within the but-for world. In the end, whether Lupin, looking ahead to those green 25 pastures of profit, would have pressed its contemporaneous noninfringement theory to victory, 26 or at least stood firm behind it long enough to dissuade our brand defendants from further 27 litigation, is a question for our jury. 1 Second, plaintiffs offer competent evidence that if their settlement had not included the 2 unlawful no-AG provision, defendants would have agreed to an earlier entry date for Lupin. 3 Defendants’ attempt to reject this hypothetical bargain as speculative fails. Contrary to 4 defendants’ assertions, the California Supreme Court’s discussion of the interplay between an 5 antitrust violation under the state’s Cartwright Act and federal patent law does not counsel 6 otherwise and, indeed, defendants’ oft-cited Wellbutrin agreed that juries frequently predict the 7 outcomes of hypothetical scenarios by assessing individual or corporate motive, thought 8 processes, and behavior. 61 Cal. 4th 149–50; 868 F.3d at 167 fn. 57; Teikoku, 296 F. Supp. 3d 9 at 1161–62. At bottom, this method requires no speculation. It merely asks the jury to project 10 how rational actors in defendants’ shoes would have proceeded in the absence of the unlawful 11 settlement term. Dolphin, 773 F.2d at 1511. 12 Simply put, the February 1, 2016, generic entry date for Lupin, is not, and never was, set 13 in stone. It was always mobile based on (among other factors) defendants’ competing 14 evaluations of the patent merits and their projection of the revenue they could extract from the 15 market as Lupin’s entry date shifted. If, as our jury could conclude based on the above, doubts 16 about infringement animated our defendants’ settlement, then our brand defendants paid 17 Lupin’s attorney’s fees plus the amount their authorized generic could recoup in the first 18 twelve months of generic entry in exchange for Lupin staying off the market for some amount 19 of time. The value to Lupin of those twelve months without an authorized generic 20 corresponded to the value to Assertio and Santarus of some number of months of their 21 preserved monopoly. Remove the pay, remove the delay. 22 Though this order doubts whether plaintiffs need to provide evidence that our defendants 23 contemplated or signaled willingness to entertain an earlier entry date (again, given plaintiffs 24 have provided sufficient evidence for a jury to conclude defendants reached an unlawful 25 agreement, it makes little sense to condition recovery on a showing that defendants first 26 contemplated a lawful settlement before pursing their ultimate course) plaintiffs do offer 27 evidence on this point. Lupin’s original August 2011 settlement offer sought a January 1, 1 countered with a January 1, 2018, entry date. In October, Lupin responded with the twelve 2 month no-AG period and a January 1, 2016, entry, remarkably close to the ultimate agreement. 3 Taken in plaintiffs favor, as we must on summary judgment, this provides direct evidence that 4 the increase in the no-AG period extended Lupin’s delay (Dkt. No. 493-1 at ¶¶ 23–24). 5 Plaintiffs also provide expert testimony to nail down the generic entry date our 6 defendants would have agreed to absent the no-AG provision. Defendants seek to exclude this 7 testimony, and we need not address it here. It suffices that plaintiffs have provided evidence 8 above that defendants could have agreed to an earlier date absent their unlawful conduct; any 9 more detail at this point goes to the measure of damages. See Zenith, 395 U.S. at 114, fn. 9. 10 Third, our record offers adequate evidence that other generics would have followed 11 Lupin to market earlier absent the unlawful settlement. Ordinarily, the ANDA scheme 12 encourages a first-filer generic with the prospect of 180-days of market exclusivity, and the 13 later-filer generics follow. A delay by the first-filer creates a bottleneck, discouraging later- 14 filers from leading the charge, as each faces a near certain thirty months of litigation with no 15 exclusivity at the end to defray the risk. See Actavis, 570 U.S. at 142–44; Wellbutrin, 868 F.3d 16 at 144–45 fn. 7; Nexium, 842 F.3d 34 at 41; 21 U.S.C. § 355(j)(5)(B)(iii)–(iv), (D); Cf. 17 Hovenkamp, Anticompetitive Patent Settlements, 15 MINN. J.L. SCI. & TECH. at 7 fn. 19. 18 Lupin filed the first ANDA to market a generic form of Glumetza in July 2009. 19 Tentative FDA approval came in January 2012 and Lupin could have expected final approval 20 in (or shortly after) May 2012 on expiration of the thirty-month stay. Instead Lupin settled 21 with our brand defendants and did not market until February 1, 2016. In that time, Sun and 22 Teva had filed their Paragraph IV ANDAs around March of 2011 and 2012, respectively. But 23 after Lupin settled, the two follow-ons promptly settled themselves in January and November 24 2013, agreeing to August 2016 launches. Teva did not launch until May 2017, Sun not until 25 July 2018. “[A] reverse payment settlement with the first filer . . . removes from consideration 26 the most motivated challenger, and the one closest to introducing [generic] competition.” 27 Actavis, 570 U.S. at 155 (quotes omitted). The just and reasonable inference being, or so our 1 Defendants point to manufacturing and regulatory difficulties that delayed Teva and 2 Sun’s market entries. But the facial defect in this argument is that if Lupin marketed when it 3 should have, and Sun and Teva followed when they should have, resolution of their difficulties 4 would move up accordingly. Following up on this point, plaintiffs add that Sun’s regulatory 5 issues with the FDA in December 2015 would not have mattered for our purposes if it had 6 succeeded in launching its generic Glumetza before then. 7 Plaintiffs further offer evidence (including expert testimony) of Sun’s strong 8 noninfringement case, arguing that Sun might have attempted to jump the line by winning a 9 judgment of noninfringement and triggering Lupin’s 75-day forfeiture clock. Now, the 10 settlement would have let Lupin market immediately to maintain its exclusivity period, posing 11 a major threat to Sun’s early market share and ability to recoup its litigation costs. But, as 12 defendants point out, “[p]laintiffs do not and cannot dispute that there is no evidence that either 13 Sun or [Teva] knew about the acceleration terms” (Dkt. No. 500 at 19). 14 To each of these theories, defendants again fault plaintiffs for not pointing to evidence 15 from Sun or Teva to the effect that either contemplated such action. But as described above, 16 this argument rests upon the befuddling notion that our actors must have contemplated what 17 they would have done in a but-for world which, due to defendants’ antitrust violation, never 18 happened. Not so. We construct the but-for world not out of whole cloth, but as a matter of 19 reasoned deviation from the actual world based upon our changed assumption of defendants’ 20 lawful conduct. Whether those changed circumstances, within the interlocking-incentive 21 scheme of the pharmaceutical-patent regulatory apparatus, would merely have driven rational 22 actors in Sun and Teva’s positions through the same course of conduct, just shifted forward in 23 time, or driven one in Sun’s position to pursue Lupin more aggressively, remains a question for 24 our jury. See Bigelow, 327 U.S. at 264–65. 25 4. PLAINTIFFS’ SUIT REMAINS TIMELY. 26 A reasonable trier of fact could also find plaintiffs’ suit timely. Our antitrust laws impose 27 a four-year statute of limitations. 15 U.S.C. § 15b. The March 5 order resolving defendants’ 1 violations of the antitrust laws, permitting plaintiffs to recover (on the pleadings at least) for all 2 Glumetza purchases from August 29, 2015, and defendants’ fraudulent concealment of the 3 scheme, permitting recovery back to the beginning. Together, defendants assail only the 4 second finding here. 5 A limitations period will be tolled where defendants affirmatively concealed their 6 unlawful conduct and plaintiffs, acting reasonably, had neither actual nor constructive 7 knowledge. Hexcel Corp. v. Ineos Polymers, Inc., 681 F.3d 1055, 1060 (9th Cir. 2012). 8 Defendants do not contest their affirmative concealment, but contend that facts available to 9 plaintiffs in mid-2015 would have excited the inquiry of a reasonable person and prompted 10 diligent investigation. Conmar Corp. v. Mitsui & Co. (U.S.A.), Inc., 858 F.2d 499, 504 (9th 11 Cir. 1988). 12 Specifically, defendants argue that Lupin’s launch calendars and an earnings call with 13 executives in mid-2015 informed plaintiffs of Lupin’s expectation that Bausch would not 14 launch an authorized generic for some time, exciting the line of inquiry in a reasonable person 15 that, if diligently pursued, would have led plaintiffs to discovery of the pay-for-delay scheme. 16 “The determination of whether a plaintiff knew or should have known of a cause of action 17 presents a question for the trier of fact,” though “diligence is tested by an objective standard.” 18 Volk v. D.A. Davidson & Co., 816 F.2d 1406, 1417 (9th Cir. 1987). Summary judgment here 19 will be appropriate “only if uncontroverted evidence irrefutably demonstrates that a plaintiff 20 discovered or should have discovered [the cause of action] but failed to file a timely 21 complaint.” Hexcel, 681 F.3d at 1063 (quotes omitted). Questions remain. 22 As early as February 2015, Lupin began distributing launch calendars to several plaintiffs 23 (including at least H-E-B, Kroger, Walgreen, AmerisourceBergen, CVS, Cardinal Health, and 24 Humana), indicating both Lupin’s status as the first-ANDA filer on Glumetza, entitling it to the 25 180-days of generic exclusivity, and the expectation that there would be “no AGx [authorized 26 generic] expected” at launch. Then, on July 5, Vinita Gupta, Lupin’s CEO, announced that 27 “Glumetza launches in February and we have a sole exclusivity for 6-months on Glumetza” 1 Defendants say this information provided constructive notice of the no-AG provision 2 because plaintiffs admitted that Lupin’s expectation that Bausch would not launch an 3 authorized generic for some time was an unusual piece of knowledge. Indeed, defendants note, 4 plaintiff Kroger did inquire, asking Bausch about its plans for an authorized generic, albeit 5 within the four year limitations period in December 2015 (Dkt. No. 464-29). No matter, 6 defendants assure us. If the information prompted Kroger to ask later, it should have had the 7 same effect earlier. 8 A reasonable jury could reject this argument and find that plaintiffs still lacked 9 constructive notice of their claims. Lupin knew good and well that Bausch would not launch 10 an authorized generic for at least one year after Lupin’s own generic launch because both had 11 signed the secret side-settlement agreement. It remains difficult to overemphasize that our 12 entire inquiry rests on the baseline that on March 26, 2012, our defendants concealed the 13 existence of the no-AG provision from Judge Hamilton. Depomed, No. C 09-05587 PJH, Dkt. 14 No. 151 (N.D. Cal. Mar. 26, 2012). “The duty of candor is not simply an obligation to answer 15 honestly when asked a direct question by the trial court. It includes an affirmative duty to 16 inform the court when a material statement of fact or law has become false or misleading in 17 light of subsequent events.” Levine v. Berschneider, 56 Cal. App. 5th 916, 921 (2020). 18 Defendants’ omission of the material no-AG provision from their disclosure to a federal judge 19 meant that the term did not exist. And, from that angle, the evidence that Lupin only began 20 hinting of the no-AG provision in mid-2015 means that our defendants continued to suppress 21 public knowledge of the no-AG provision for three years. The conceptual leap defendants 22 propose, from “no-AG expected” to “we should investigate whether Bausch secured the Lupin 23 settlement with a no-AG provision,” would have required our plaintiffs to conclude that their 24 regular business partners had both misled a federal judge and spent three years covering it up. 25 This hazy record does not mandate that jump and, in fact, permits a jury to conclude plaintiffs 26 did not have constructive notice of the no-AG provision.

27 1 5. ASSERTIO’S INDIVIDUAL DEFENSES FAIL. 2 Assertio, on its own, proclaims that it washed its hands of the alleged unlawful scheme 3 long ago, arguing that it withdrew from the conspiracy in 2013, that it committed none of the 4 overt acts contributing to defendants’ continuing violation, and that it bears no culpability for 5 Bausch’s unforeseeable price hikes. 6 The story goes that Assertio granted Santarus control of most marketing for Glumetza in 7 July 2008, though the two shared price control. In August 2011, Assertio transferred all 8 Glumetza commercialization rights to Santarus, including sole price control, though Assertio 9 retained a royalty on sales. But that link severed in October 2013, when Assertio sold 10 “substantially all its royalty interests for Glumetza” to PDL Biopharma (Dkt. No. 464 at 9–10). 11 In other words, as Assertio tells us, the last thing it did in service of the conspiracy was sign 12 the settlement agreement in February 2012. 13 On its face, this story leaves a curious omission, one highlighted by defendants’ repeating 14 chorus that the challenged settlement agreement effected no more than the patent rights granted 15 to, who else but, Assertio. Who lists the patents as covering Glumetza in the Orange Book? 16 Who must be present in any infringement suit to trigger the thirty-month stay and exclude new 17 generics? Who grants or withholds licenses? Who holds the rights which might exculpate 18 defendants’ alleged scheme? The patent owner, Assertio. See Lone Star Silicon Innovs. v. 19 Nanya Techs., 925 F.3d 1225, 1228–29 (Fed. Cir. 2019); In re Actos End-Payor Antitrust 20 Litig., 848 F.3d 89, 94 (2d Cir. 2017); Nexium, 842 F.3d at 63. If Assertio maintained the 21 patents throughout the conspiracy, and as far as defendants have told us it did, then Assertio 22 maintained both the primary regulatory scarecrow against generic competition and the primary 23 (indeed only, thus far) justification for defendants’ scheme. In for a penny, in for a pound. 24 First, actions in furtherance of a conspiracy are imputed to all involved “until its purpose 25 has been achieved or abandoned.” U.S. v. Inryco, Inc., 642 F.2d 290, 293 (9th Cir. 1981); 26 Arandell Corp. v. Centerpoint Ener. Servs., 900 F.3d 623, 634 (9th Cir. 2018). Withdrawal 27 from a conspiracy requires disavowal of the unlawful goal, an affirmative act to defeat its 1 States v. Lothian, 976 F.2d 1257, 1261 (9th Cir. 1992) (quotes omitted). While Assertio 2 covered its coconspirators’ exits, even from afar, it cannot be said to have withdrawn, and will 3 answer alongside them. 4 Second, a continuing violation accrues with each overt act, which is both new and 5 independent, not a mere reaffirmation of previous conduct, and which inflicts new and 6 accumulating injury. Samsung, 747 F.3d at 1202–03. True, as Assertio notes, the Honorable 7 Lucy H. Koh, relying on our court of appeals’ language in Pace Industries v. Three Phoenix 8 Company that “an overt act by the defendant is required to restart the statute of limitations,” 9 concluded that the continuing violation doctrine does not impute acts between coconspirators. 10 Garrison v. Oracle Corp., 159 F. Supp. 3d 1044, 1072 (quoting 813 F.2d 234, 237 (9th Cir. 11 1987)). But this appears an unduly literal interpretation of the language in Pace given, first, 12 that our court of appeals did not raise the distinction that Judge Koh later raised and, second, 13 because the Supreme Court does appear to approve of imputing a continuing violation across a 14 conspiracy, explaining famously that “[i]n the context of a continuing conspiracy to violate the 15 antitrust laws . . . each time a plaintiff is injured by an act of the defendants[,] a cause of action 16 accrues to him to recover the damages caused by that act.” Zenith Radio Corp. v. Hazeltine 17 Research, Inc., 401 U.S. 321, 338 (1971) (emphasis added). This order thus respectfully 18 disagrees with Judge Koh’s ruling, though, regardless, we have such an overt act here. 19 Assertio provided, and maintained control of throughout, the patents, the regulatory leverage 20 that enabled defendants’ entire alleged scheme. Samsung Elects. v. Panasonic Corp., 747 F.3d 21 1199, 1203 (9th Cir. 2014). 22 Third, “a defendant cannot be held liable for substantive offenses committed before 23 joining or after withdrawing from a conspiracy.” Lothian, 976 F.2d at 1262. But our record 24 supports the conclusion that Assertio, by maintaining the regulatory levers permitting and 25 perhaps justifying defendants’ conspiracy, never withdrew from it. It’s also worth noting that 26 if Assertio retained the patents, it too retained, potential contract breaches aside, the power to 27 stop Valeant’s apparently unforeseeable price hikes by pulling the patent license or permitting 1 6. LUPIN’S INDIVIDUAL DEFENSES FAIL. 2 Alone, Lupin argues that its distribution contracts insulated distributors 3 AmerisourceBergen, Cardinal Health, and McKesson, whose assignees comprise a large 4 portion of our direct purchasers, from actual injury. According to Lupin, it sold drugs to the 5 distributors at the wholesale acquisition price and the distributers primarily derived revenue not 6 from any markup they might have added but from nonrefundable service fees Lupin paid. The 7 distributors made more money as the percentage fees grew with drug prices. 8 This argument mistakes plaintiffs’ cognizable injury, which encompasses the immediate 9 extent of the overcharge without regard to plaintiffs’ later profit. Hanover Shoe’s cut-and-dry 10 treatment of the overcharge works no injustice here. What Hanover Shoe takes, Illinois Brick 11 has given. And, to this point Lupin and its codefendants have enjoyed Illinois Brick’s shield 12 against indirect-purchaser liability several times over. Hanover Shoe v. United Shoe Mach. 13 Corp., 392 U.S. 481, 489–93 (1968); Illinois Brick v. Illinois, 431 U.S. 720, 749 (1977) 14 (Brennan, J., dissenting); Glumetza, 336 F.R.D. at 480–81 (quoting In re Skelaxin 15 (Metaxalone) Antitrust Litigation, No. MD 12-02343 CLC, 2014 WL 2002887, *5 (E.D. Tenn. 16 May 15, 2014) (Judge Curtis L. Collier)). 17 Lupin also argues that Bausch would have raised the price of Glumetza as it did in 2015 18 even if generics had already entered the market. Even taking this as true, the presence of 19 cheaper generic alternatives in 2015 would have both shielded consumers from the full brunt of 20 the price hikes and helped bring the prices back down to a competitive level sooner. Yet 21 plaintiffs also offer evidence that Bausch had never raised prices to the extent it did with 22 Glumetza on any drug already subject to generic competition. Of the 206 drugs for which 23 Bausch raised the price over 100%, it did so following generic entry only 15 times, and none of 24 those reached anywhere close to the nearly 800% increases seen in our case (Dkt. No. 493-1 at 25 ¶¶ 67–69). A reasonable trier of fact could conclude that generic competition would have 26 suppressed the 2015 price hikes. 27 1 CONCLUSION 2 All motions for summary judgment are DENIED. This order has trodden carefully to 3 resolve the parties’ four dispositive motions without requiring full resolution of the parties’ ten 4 pending Daubert motions. Argument on the Daubert motions will be set midsummer. A 5 future order will resolve those motions. The Court trusts that if resolution of the Daubert 6 motions fundamentally undercuts a party’s ability to make a certain argument, the party will 7 own up to it and not waste our time at trial. For that matter, trial remains set for and voir dire 8 will commence on OCTOBER 4 AT 7:30 A.M. The Court is looking forward to it. 9 IT IS SO ORDERED. 10 11 Dated: May 6, 2021. 12

= WILLIAM ALSUP 14 UNITED STATES DISTRICT JUDGE 15 16

= 17

Z 18 19 20 21 22 23 24 25 26 27 28