In Re: Plavix Marketing v.

• Court: Court of Appeals for the Third Circuit
• Decided: September 1, 2020
• Docket: 18-2472
• Status: Published

Opinion

PRECEDENTIAL

UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT _______________

No. 18-2472 _______________

IN RE: PLAVIX MARKETING, SALES PRACTICES AND PRODUCTS LIABILITY LITIGATION (NO. II)

UNITED STATES OF AMERICA; STATE OF CALIFORNIA; STATE OF COLORADO; STATE OF CONNECTICUT; STATE OF DELAWARE; DISTRICT OF COLUMBIA; STATE OF FLORIDA; STATE OF GEORGIA; STATE OF ILLINOIS; STATE OF INDIANA; STATE OF IOWA; STATE OF LOUISIANA; COMMONWEALTH OF MASSACHUSETTS; STATE OF MICHIGAN; STATE OF MINNESOTA; STATE OF MONTANA; STATE OF NEVADA; STATE OF NEW JERSEY; STATE OF NEW YORK; STATE OF NORTH CAROLINA; STATE OF OKLAHOMA; STATE OF RHODE ISLAND; STATE OF TENNESSEE; COMMONWEALTH OF VIRGINIA; STATE OF WASHINGTON; STATE OF WISCONSIN, ex rel. JKJ PARTNERSHIP 2011 LP

v.

SANOFI-AVENTIS U.S. LLC; SANOFI-AVENTIS U.S. SERVICES, INC.; AVENTIS, INC.; AVENTIS PHARMACEUTICALS, INC.; BRISTOL-MYERS SQUIBB COMPANY; BRISTOL-MYERS SQUIBB SANOFI PHARMACEUTICALS HOLDING PARTNERSHIP

JKJ PARTNERSHIP 2011 LLP, Appellant _______________

On Appeal from the United States District Court for the District of New Jersey (D.C. Nos. 3-11-cv-06476 & 3-13-cv-02418) Chief District Judge: Honorable Freda L. Wolfson _______________

Argued: June 25, 2020

Before: JORDAN, BIBAS, and NYGAARD, Circuit Judges

(Filed: September 1, 2020) _______________

William H. Narwold Mathew P. Jasinski [ARGUED] Motley Rice 20 Church Street One Corporate Center, 17th Floor Hartford, CT 06103

W. Scott Simmer Baron & Budd 600 New Hampshire Avenue, Suite 10-A Washington, DC 20037 Counsel for Appellant

2 Anand Agneshwar [ARGUED] Arnold & Porter Kaye Scholer 250 West 55th Street New York, NY 10019

Murad Hussain Kirk Ogrosky Arnold & Porter Kaye Scholer 601 Massachusetts Avenue, N.W. Washington, DC 20001

Gavin J. Rooney Lowenstein Sandler One Lowenstein Drive Roseland, NJ 07068 Counsel for Appellees _______________

OPINION OF THE COURT _______________

BIBAS, Circuit Judge. To intervene is to butt in, not to be dragged in or to replace an existing party to a lawsuit. But no one butted in here. Three people formed a partnership to sue several pharmaceutical companies as a qui tam relator under the False Claims Act. When one of them left the partnership and was replaced, that change amounted to forming a new partnership. The compa- nies moved to dismiss because the Act’s first-to-file bar stops a new “person” from “interven[ing] or bring[ing] a related ac- tion based on the [same] facts.” 31 U.S.C. § 3730(b)(5). But the verb “intervene” means to inject oneself between two existing

3 parties, as under Federal Rule of Civil Procedure 24. The new partnership did not do that, but instead came in as the relator. Still, the District Court ruled for the companies, based mainly on a dictum from a Supreme Court case. But that case involved a very different issue, and the Court’s opinion never considered the issue here. Because the Act’s plain text bars only intervention or bringing a related suit, we will vacate and remand. I. BACKGROUND A. The partnership and the complaints In 2011, Jeffrey Stahl, Kelly Evans, and John Venditto formed a Delaware limited liability partnership named JKJ (af- ter their three first initials). Venditto and Stahl were doctors, and Evans was a former sales representative at Sanofi-Aventis. The partnership’s sole purpose was to prosecute a qui tam False Claims Act suit against Sanofi-Aventis and Bristol-My- ers Squibb, two pharmaceutical companies that developed and marketed the anti-clotting drug Plavix. So JKJ filed this suit in federal district court. The gist of the suit was that Sanofi and Bristol had pro- moted Plavix to treat a broad range of patients, even though they knew that many of them would reap little if any benefit. Sanofi and Bristol’s marketing, JKJ alleged, caused many false claims to be submitted for federal and state healthcare reim- bursement. JKJ alleged False Claims Act claims on behalf of the United States, as well as claims on behalf of dozens of states under their own qui tam statutes.

4 The complaints named JKJ as the sole plaintiff-relator. While the early complaints discussed the three partners’ back- grounds, they did not name them, using pseudonyms instead. The United States declined to intervene in the suit. By 2016, the partners’ relations had apparently soured. Venditto left and was replaced by Dr. Paul Gurbel. The second amended complaint, the one at issue, was filed in 2017 and names all three partners. But it still names JKJ as the sole rela- tor. The partners viewed the old JKJ partnership as the same entity as the new one. That theory would be tested in this suit’s pinball journey among three different state and federal courts. B. District Court proceedings Sanofi and Bristol moved to dismiss, invoking the False Claims Act’s first-to-file bar. 31 U.S.C. § 3730(b)(5); see United States ex rel. JKJ P’ship 2011, LLP v. Sanofi Aventis, U.S., LLC (In re Plavix Mktg., Sales Practices & Prods. Liab. Litig. (No. II)), 315 F. Supp. 3d 817, 821–22 (D.N.J. 2018) (In re Plavix I). Sanofi and Bristol’s theory was that New JKJ was a different legal entity from Old JKJ, so its effort to pursue the suit as the relator amounted to an “interven[tion]” banned by the first-to-file bar. See In re Plavix I, 315 F. Supp. 3d at 830. The District Court agreed and dismissed the suit. In a care- ful and comprehensive opinion, it thoroughly analyzed Dela- ware partnership law, concluding that Old JKJ and New JKJ were distinct legal entities. In re Plavix I, 315 F. Supp. 3d at 830–34. Because the two entities were different, the District Court held that New JKJ’s presence violated the first-to-file

5 bar. Id. at 834–35. It relied heavily on a Supreme Court deci- sion, United States ex rel. Eisenstein v. City of New York, 556 U.S. 928, 933–35 (2009), which it read as holding that when- ever a nonparty tries to join a qui tam False Claims Act suit, it is “interven[ing]” within the meaning of the first-to-file bar. See Plavix I, 315 F. Supp. 3d at 834–35. It also relied on a re- cent Tenth Circuit decision that read Eisenstein the same way. Id. at 835–36 (discussing United States ex rel. Little v. Triumph Gear Sys., Inc., 870 F.3d 1242 (10th Cir. 2017)). Because the District Court read Eisenstein’s rule as barring a new relator from replacing an old one in an amended complaint, it dis- missed JKJ’s suit. Id. C. Delaware Supreme Court proceedings The partnership appealed to us. Recognizing the important role that Delaware partnership law plays here, we certified three questions to the Delaware Supreme Court. The Delaware Supreme Court accepted the certification and assisted us by an- swering them. First, it agreed with the District Court that Old JKJ and New JKJ were distinct partnerships. United States ex rel. JKJ P’ship 2011 LLP v. Sanofi-Aventis U.S. LLC, 226 A.3d 1117, 1123 (Del. 2020) (In re Plavix II). Second, it could not answer whether Old JKJ survived long enough to file the com- plaint now before us. Id. at 1133. But it noted that the second amended complaint named the three partners as Stahl, Evans, and Gurbel, so that complaint must have been filed by New JKJ. Id. Third, it explained that Old JKJ could not keep prose- cuting the litigation as part of the winding-up process. Id. at 1135–36. It did not see how liquidation could be “[ ]consistent with continuing with carrying on the business for which the

6 Partnership was established.” Id. at 1135. With these state-law issues addressed, the case is before us again. II. THE FIRST-TO-FILE BAR IS NOT JURISDICTIONAL First off, the parties dispute whether the first-to-file bar is jurisdictional. Sanofi and Bristol say that it is and thus that their motion to dismiss falls under Federal Rule of Civil Procedure 12(b)(1). But JKJ suggests that a qui tam complaint that vio- lates § 3730(b)(5) merely fails to state a claim and thus that the motion falls under Rule 12(b)(6). The distinction sometimes matters. First, the plaintiff bears the burden of persuasion on jurisdiction, while the burden of showing that a complaint fails to state a claim falls on the de- fendant. Davis v. Wells Fargo, 824 F.3d 333, 349 (3d Cir. 2016). Second, a defendant challenging subject-matter juris- diction may sometimes submit evidence, while on Rule 12(b)(6) we must take the complaint’s well-pleaded allegations as true. Id. Finally, jurisdiction may be challenged at any time, even sua sponte by the court. See Fed. R. Civ. P. 12(h)(3). But the failure-to-state-a-claim defense is waived if not raised “be- fore the close of trial.” 5C Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure § 1392, at 530 (3d ed. 2004). Our sister circuits are split on this question. The lower- numbered circuits treat the first-to-file bar as not jurisdictional. See United States ex rel. Heath v. AT&T, Inc., 791 F.3d 112, 120–21 (D.C. Cir. 2015); United States ex rel. McGuire v. Mil- lennium Labs., Inc., 923 F.3d 240, 250–51 (1st Cir. 2019); United States ex rel. Hayes v. Allstate Ins. Co., 853 F.3d 80,

7 85–86 (2d Cir. 2017) (per curiam). The higher-numbered cir- cuits take the opposite view, mainly in older opinions. See United State ex rel. Carter v. Halliburton Co., 710 F.3d 171, 181 (4th Cir. 2013); United States ex rel. Branch Consultants v. Allstate Ins. Co., 560 F.3d 371, 376–77 (5th Cir. 2009); Wal- burn v. Lockheed Martin Corp., 431 F.3d 966, 970 (6th Cir. 2005); United States ex rel. Lujan v. Hughes Aircraft Co., 243 F.3d 1181, 1187–89 (9th Cir. 2001); Grynberg v. Koch Gate- way Pipeline Co., 390 F.3d 1276, 1278 (10th Cir. 2004). As the Third Circuit, our number falls in the middle. We hold that the first-to-file bar is not jurisdictional. As the Supreme Court has recently instructed, unless Congress states clearly that a rule is jurisdictional, we will treat it as nonjuris- dictional. Sebelius v. Auburn Reg’l Med. Ctr., 568 U.S. 145, 153 (2013); see, e.g., Fort Bend County v. Davis, 139 S. Ct. 1843, 1850 (2019). The contrary circuit cases mostly predate these Supreme Court cases and do not apply the Court’s clear- statement rule. Sanofi and Bristol point to no language in § 3730(b)(5), nor do we see any, that “plainly show[s] that Con- gress imbued [the first-to-file] bar with jurisdictional conse- quences.” United States v. Kwai Fun Wong, 575 U.S. 402, 410 (2015). The bar could have referred to jurisdiction or to a dis- trict court’s power to review claims directly. Heath, 791 F.3d at 120. It did not. On the contrary, the bar on intervention kicks in only after a party has filed a suit over which the court al- ready has jurisdiction. If Congress had meant to make the first-to-file bar jurisdic- tional, it would have logically placed the bar in one of two other sections that mention jurisdiction and were added at the

8 same time as it. 31 U.S.C. §§ 3730(e)(1), (2)(A), 3732 (cap- tion); False Claims Amendments Act of 1986, Pub. L. No. 99- 562, §§ 3, 6(a), 100 Stat. 3153, 3155, 3157–58. Instead, Con- gress put it in § 3730(b), alongside other run-of-the-mill proce- dural rules. See, e.g., 31 U.S.C. § 3730(b)(2) (service and fil- ing). Lacking a clear statement, Sanofi and Bristol rely on prin- ciples of Article III standing. They note that the Article III in- jury is suffered not by the relator, but by the Government. See Vt. Agency of Nat. Res. v. United States ex rel. Stevens, 529 U.S. 765, 771, 773–74 (2000). So, they argue, the first-to-file bar shows which parties have standing to sue on behalf of the United States. But the first-to-file bar is a matter of statutory authorization, not constitutional standing. The first-to-file bar asks only “whether [the relator] falls within the class of plain- tiffs whom Congress has authorized to sue,” which is another way to ask whether the statute gives it a cause of action. Lexmark Int’l, Inc. v. Static Control Components, Inc., 572 U.S. 118, 128 (2014). It is not about “the court’s statutory or constitutional power to adjudicate the case.” Id. at 128 n.4 (quoting Verizon Md. Inc. v. Pub. Serv. Comm’n, 535 U.S. 635, 643 (2002)). Because the first-to-file bar is not jurisdictional, Sanofi and Bristol’s motion to dismiss falls under Rule 12(b)(6) for failure to state a claim. On to the merits.

9 III. THE FALSE CLAIMS ACT DOES NOT BAR NEW RELATORS FROM ENTERING A QUI TAM SUIT

Parties may sue under the False Claims Act in two ways. First, the Attorney General may sue anyone who violates the Act. 31 U.S.C. § 3730(a). Second, any person may bring a qui tam suit in the Government’s name. Id. § 3730(b)(1). When a private person (the relator) brings a qui tam suit, the Govern- ment may choose to intervene. Id. § 3730(b)(2). Whether the Government intervenes or not, it gets the bulk of the recovery. See id. § 3730(d). If it chooses to intervene, it takes over the lead as plaintiff; if not, the relator keeps prosecuting the suit. Id. § 3730(c)(1), (3). The False Claims Act’s first-to-file bar provides: When a person brings an action under this sub- section, no person other than the Government may intervene or bring a related action based on the facts underlying the pending action.

31 U.S.C. § 3730(b)(5). This text, we hold, has a straightforward meaning. In nor- mal civil litigation, there are three ways for nonparties with in- terests relevant to a suit to become parties to a suit. They can intervene in the existing suit. They can file their own related suits based on the same facts. Or they can be added to the ex- isting suit by the court or the existing parties. In False Claims Act qui tam suits, the first-to-file bar precludes the first two options, but not the third. So long as the new party is named in the (likely amended) complaint, there is no problem. This read- ing follows from the plain meaning of the term “intervene.”

10 And nothing in the Supreme Court’s opinion in Eisenstein is to the contrary. A. The first-to-file bar’s plain meaning We start with the word’s plain meaning. “Intervene” comes from the Latin venire, “to come,” plus the prefix inter-, “be- tween.” Intervene, Oxford English Dictionary (2d ed. 1989). To “intervene” is “[t]o come between in action; to interfere, interpose.” Id. (def. 3a). An entity that intervenes does not be- come one of two sides; it “come[s] between” existing sides. Intervene, Webster’s Third New International Dictionary (1966). Often, intervention is not done at the behest or even with the acquiescence of the existing parties. See, e.g., id. (def. 1) (“to enter or appear as an irrelevant or extraneous feature or circumstance”); id. (def. 3) (“to come in or between by way of hindrance or modification”). In short, a third party intervenes when he injects himself between two existing sides, not when he is drawn in or becomes one side or another. The legal sense of intervention reflects the word’s etymol- ogy and plain English meaning. “Intervention” is “[t]he entry into a lawsuit by a third party who, despite not being named a party to the action, has a personal stake in the outcome.” Inter- vention, Black’s Law Dictionary (11th ed. 2019) (def. 1). So in law, a party who “intervene[s]” invokes “[t]he legal procedure by which such a third party is allowed to become a party to the litigation.” See id. (def. 2). The choice to intervene is made not by the existing parties, but by the intervenor. An “intervenor” is “[o]ne who voluntarily enters a pending lawsuit because of a personal stake in it.” Intervenor, in id. (emphasis added). In

11 the federal courts, this process is governed by Federal Rule of Civil Procedure 24. Intervention is distinct from joinder, a way that existing parties can bring a third party into a lawsuit. See Joinder, Black’s Law Dictionary, supra. Required joinder ensures the presence of a necessary party. Fed. R. Civ. P. 19. Permissive joinder consolidates parties who might otherwise be involved in related but distinct actions. Id. R. 20. Either way, joinder is begun by the existing parties or the court. Intervention, by con- trast, involves a nonparty’s motion to enter a suit to protect its interests, whether or not the existing parties want its company. Id. R. 24(a)–(b). And there are other ways for parties to enter a lawsuit. For instance, a defendant can implead a nonparty who might be li- able to it for part of the claim. Fed. R. Civ. P. 14(a)(1). A party facing potentially duplicative liability can interplead all the parties to whom it might be liable. Id. R. 22(a). Or if a party dies, the court can substitute its successor or representative. Id. R. 25(a)(1). What distinguishes intervention from these other methods of adding new parties is that it requires action by an outside party who seeks a seat at the table. See 7C Charles Alan Wright et al., Federal Practice and Procedure § 1901, at 257–60 (3d ed. 2007). Given that distinction, Congress would not have chosen the specific verb “intervene” to express the general con- cept of entry into a suit, no matter who initiated it or how. See Mississippi ex. rel. Hood v. AU Optronics Corp., 571 U.S. 161, 169–70 (2014) (presuming “Congress is aware of existing law when it passes legislation,” including “Federal Rule of Civil

12 Procedure 20, governing party joinder” (internal quotation marks omitted)). By contrast, when Congress meant to refer to a broader class of parties entering suits, it covered “claims that involve the joinder or intervention of additional parties.” 28 U.S.C. § 1367(a) (emphasis added) (defining federal courts’ supple- mental jurisdiction). Congress chose to draft that statute more broadly, but the first-to-file bar more narrowly. If Congress had wanted the first-to-file bar to reach more broadly, it would have said so. But it chose a “narrower” term (intervention), and we must “respect, not disregard,” that choice. Wis. Cent. Ltd. v. United States, 138 S. Ct. 2067, 2072 (2018). To be sure, other provisions of the False Claims Act limit who can be a proper plaintiff. For instance, when the allega- tions underlying the suit have already been disclosed publicly, no private party can sue unless it is “an original source of the information.” 31 U.S.C. § 3730(e)(4)(A). But beyond barring intervention, the first-to-file bar itself says nothing about other ways in which third parties may enter an existing suit. A final note: The False Claims Act does provide for a spe- cial kind of intervention by the Government. When the Gov- ernment chooses to intervene, it does not remain a third party between two existing parties. See 31 U.S.C. § 3730(b)(1)–(2), (c)(3) (describing Government intervention). Rather, even though the relator is still a party to the case, the Government takes over the relator’s “primary” role in pursuing it. Id. § 3730(c)(1). But this special kind of intervention occurs only when the Government intervenes, not when a private party does. It works differently not because of the first-to-file bar,

13 but because of the special Government-control provision in § 3730(c)(1). And the text of the first-to-file bar limits its scope to intervention by private parties, not the Government. Id. § 3730(b)(5). Thus, we see no reason to give “intervene” in that bar a unique meaning specific to the False Claims Act. B. Eisenstein is silent on the question before us Resisting this reading of the statute, Sanofi and Bristol rely on the Supreme Court’s decision in Eisenstein. But Eisenstein addressed a very different question: the time limit for filing a notice of appeal in False Claims Act suits. See 556 U.S. at 929. Typically, parties to civil cases have 30 days, but when the United States is a party, the limit is 60 days. See Fed. R. App. P. 4(a)(1)(A)–(B). The Court held that even though False Claims Act suits are always on behalf of the United States, the Government is not a “party” to the litigation for purposes of appellate procedure when it has declined to intervene. Id. at 937. In reaching its conclusion, the Court discussed the relation- ship between the terms “party” and “intervention.” See Eisen- stein, 556 U.S. at 932–34. “[W]hen the term [to intervene] is used in reference to legal proceedings, it covers the right of one to interpose in, or become a party to, a proceeding already in- stituted.” Id. at 933 (emphasis and second alteration in original) (quoting Rocca v. Thompson, 223 U.S. 317, 330 (1912)). Be- cause intervening is how the United States becomes a party to a qui tam suit, the Court held, it is not a party unless it inter- venes. Id.

14 The first-to-file bar was not before the Eisenstein Court. Nor did Eisenstein turn on the meaning of “intervene.” So Sanofi and Bristol’s reliance on Eisenstein hinges on a single sentence from the discussion referred to above: “The Court has further indicated that intervention is the requisite method for a non-party to become a party to a lawsuit.” 556 U.S. at 933. According to Bristol and Sanofi, this sentence proves that in- tervention is not just one way, but the only way, for a party to enter a lawsuit. The Tenth Circuit has likewise taken that broad a view of Eisenstein’s dictum. United States ex rel. Little v. Triumph Gear, 870 F.3d at 1246–48. Put another way, any time a party enters (or becomes a party in) a lawsuit, it must have intervened. We reject this overreading. We must read Eisenstein, like any opinion, in context. Judicial opinions are not statutes, from which we squeeze all we can out of every last word. Rather, we try to understand the Court’s language against the backdrop of the particular controversy that the Court was resolving. See, e.g., Illinois v. Lidster, 540 U.S. 419, 424 (2004); Cohens v. Virginia, 19 U.S. (6 Wheat.) 264, 399 (1821) (Marshall, C.J.). Eisenstein was about appellate timing and Government inter- vention, not the first-to-file bar for private parties. The parties neither briefed nor argued about how private parties intervene. The bar’s plain text, not a stray dictum about that text, resolves its meaning. Thus, the first-to-file bar reaches intervention under Rule 24 or the bringing of a new action on the same facts, but not

15 other methods of joining an existing case like joinder, substi- tution of parties, or amendment of a complaint. So it does not bar New JKJ’s participation as a relator here. IV. WE DECLINE TO RESOLVE MOST OF THE RESIDUAL ISSUES

Just because New JKJ can take part without violating the first-to-file bar does not mean that it is taking part as a matter of partnership law and ordinary civil procedure. Deciding whether New JKJ is properly a relator depends on record ma- terials and recent developments that the District Court never had the chance to consider. Rather than hazard answers, we will instead send the case back to let the District Court answer those lingering questions first. A. Important considerations for remand To guide the District Court on remand, we flag several is- sues that bear on whether New JKJ is a proper relator. First, Sanofi and Bristol argue that the second amended complaint’s allegations are so different from those that ap- peared earlier that they make the amendment improper. As noted, even wholesale changes in a complaint do not implicate the first-to-file bar, as long as the relator does not bring a new “action.” 31 U.S.C. § 3730(b)(5). But we offer no view on whether, as a matter of generic civil procedure, this amendment might exceed the bounds of Rule 15. Second, the Delaware Supreme Court found that both New and Old JKJ are aggregate partnerships. In re Plavix II, 226 A.3d at 1132–33. A key feature of aggregate partnerships is

16 that they cannot sue or be sued in their own names. See id. at 1123. So one could question whether either partnership was ever a proper relator. The District Court should decide whether that is true and, if so, whether to allow amendment so that the partners can name themselves as the real relators in interest. See Fed. R. Civ. P. 17(a)(3). Third, the District Court will need to decide whether Old JKJ or New JKJ owns this lawsuit. While this appeal was pend- ing, the partnership executed an agreement purporting to trans- fer this litigation asset. Perhaps the agreement was unneces- sary: perhaps the change in membership and concomitant dis- solution of Old JKJ and the formation of New JKJ automati- cally transferred the asset from the former to the latter. If not, the District Court should determine whether the suit could be transferred and whether Old JKJ had the power to transfer it as part of its winding-up process. This list of issues for remand is not exhaustive. The court is free to take up other matters as it sees fit. B. Property stakes in qui tam suits are transferrable Though we leave many of these issues for remand, we will settle one last False Claims Act issue. Sanofi and Bristol argue that the Anti-Assignment Act bars a qui tam relator from ever assigning its interest in a suit. 31 U.S.C. § 3727. A qui tam re- lator is, of course, entitled to a share of the Government’s re- covery. So, they assert, transferring the relator’s interest in the action would amount to a forbidden “transfer or assignment of any part of a claim against the United States Government or an interest in the claim.” Id. § 3727(a)(1).

17 That argument fails. Whether a plaintiff sues for itself or for another, its claim is “against” the defendant, not “against” any real party in interest who might get a share of the recovery. The relator might eventually have a claim against the Govern- ment if the Government recovered a judgment and refused to give the relator its share. See 31 U.S.C. § 3730(d)(1). But the first suit does not assert a “claim against the United States Government.” Id. § 3727(a)(1) (emphasis added). * * * * * The False Claims Act’s first-to-file bar stops new relators from intervening in other parties’ suits or bringing their own separate suits based on the same facts. Yet it does not bar par- ties from amending a complaint to add, remove, or swap rela- tors. While some courts have overread the Supreme Court’s dictum in Eisenstein to suggest otherwise, our touchstone is the statutory text. The District Court’s otherwise thoughtful opin- ion missed this point. So we will vacate and remand for further proceedings.