`United States Court of Appeals
`FOR THE DISTRICT OF COLUMBIA CIRCUIT
`
`
`
`Decided April 16, 2021
`
`Argued December 10, 2020
`
`
`No. 20-5026
`
`GENUS MEDICAL TECHNOLOGIES LLC,
`APPELLEE
`
`v.
`
`UNITED STATES FOOD AND DRUG ADMINISTRATION,
`APPELLANT
`
`
`Appeal from the United States District Court
`for the District of Columbia
`(No. 1:19-cv-00544)
`
`
`Daniel Winik, Attorney, U.S. Department of Justice,
`argued the cause for appellant. With him on the briefs were
`Jeffrey Bossert Clark, Acting Assistant Attorney General, Scott
`R. McIntosh, Attorney, Robert P. Charrow, General Counsel,
`U.S. Department of Health and Human Services, AnnaMarie
`Kempic, Deputy Chief Counsel for Litigation.
`
`
`Noam B. Fischman was on the brief for amicus curiae
`Bracco Diagnostics Inc. in support of appellant.
`
`
`James A. Boiani was on the brief for amicus curiae Giskit
`B.V. in support of appellant.
`
`
`
`
`

`

`2
`Douglas B. Farquhar argued the cause and filed the brief
`for appellee.
`
`
`Before: HENDERSON, PILLARD and KATSAS, Circuit
`Judges.
`
`Opinion for the Court filed by Circuit Judge HENDERSON.
`
`Opinion concurring in the judgment filed by Circuit Judge
`PILLARD.
`
`KAREN LECRAFT HENDERSON, Circuit Judge: The Federal
`Food, Drug, and Cosmetic Act (FDCA or Act), 21 U.S.C.
`§§ 301 et seq., sets forth separate and detailed regimes for the
`regulation of medical products classified, inter alia, as drugs or
`devices. The question before us is whether the U.S. Food and
`Drug Administration (FDA) enjoys discretion to classify as a
`“drug” a product that meets the statutory definition of a
`“device.” The FDA claims that, if a medical product satisfies
`the statutory definitions of both a “drug” and a “device,” the
`Act’s overlapping definitions grant by implication the FDA
`broad discretion to regulate the product under either regime.
`Since 2017 the FDA has exercised its claimed discretion to
`classify Genus Medical Technologies’ (Genus) “Vanilla SilQ”
`line of diagnostic contrast agents as drugs, notwithstanding the
`FDA’s recognition that the products “appear” to satisfy the
`statutory definition for devices. Genus subsequently filed suit,
`challenging the FDA’s classification decision as inconsistent
`with the Administrative Procedure Act (APA), 5 U.S.C.
`§ 706(2), and
`the FDCA.
` Finding
`that
`the FDCA
`unambiguously forecloses the FDA’s interpretation, the district
`court granted summary judgment in Genus’s favor and vacated
`the FDA decision to classify Genus’s products as drugs. We
`agree with the district court that the text, statutory structure and
`legislative history of the Act make plain that the Congress did
`
`
`
`

`

`3
`not grant the FDA such sweeping discretion. Accordingly, we
`affirm the district court’s grant of summary judgment.
`
`I. BACKGROUND
`
`A. Statutory & Regulatory Framework
`
`The FDCA grants the FDA the authority to regulate certain
`categories of medical products, including drugs, devices,
`biologics and dietary supplements. Relevant here are the
`statutory definitions for “drug” and “device.” The Act, in
`relevant part, defines “drugs” to include:
`
`articles intended for use in the diagnosis, cure,
`mitigation, treatment, or prevention of disease
`in man or other animals . . . .
`
`21 U.S.C. § 321(g)(1)(B). “Devices” are defined to include:
`
`an instrument, apparatus, implement, machine,
`contrivance, implant, in vitro reagent, or other
`similar or
`related article,
`including any
`component, part, or accessory, which is . . .
`intended for use in the diagnosis of disease or
`other conditions, or in the cure, mitigation,
`treatment, or prevention of disease, in man or
`other animals, . . . and which does not achieve
`its primary intended purposes through chemical
`action within or on the body of man or other
`animals and which is not dependent upon being
`metabolized for the achievement of its primary
`intended purposes.
`
`
`
`

`

`4
`Id. § 321(h)(1).1 Because the two definitions share a common
`“intended-use clause”—that is, both definitions include articles
`intended for use in the diagnosis, cure, mitigation, treatment or
`prevention of disease—and because the drug definition
`features no other relevant limitations, it is apparent that any
`product that satisfies the “device” definition also satisfies the
`definition of a “drug.” The converse, however, is not true.
`Because a device must be “an
`instrument, apparatus,
`implement, machine, contrivance, implant, in vitro reagent, or
`other similar or related article,” and further, because it may
`neither “achieve its primary intended purposes through
`chemical action within or on the body of man” nor be
`“dependent upon being metabolized for the achievement of its
`primary intended purposes,”2 the set of products that satisfy the
`device definition is necessarily encompassed by, but narrower
`than, the set of products that satisfy the drug definition.
`
`Drugs and devices are subject to distinct regulatory
`regimes. To begin, separate divisions of the FDA are primarily
`responsible for each product category. Whereas drugs are
`
`1 At the time of the FDA’s decision, the device definition was
`located at 21 U.S.C. § 321(h). The Congress later relocated the
`amendment
`to 21 U.S.C. § 321(h)(1).
` See Safeguarding
`Therapeutics Act, Pub. L. No. 116-304, § 2(b), 134 Stat. 4915, 4916
`(2011) (codified at 21 U.S.C. § 321(h)(1)).
`2 Although FDA guidance refers to these “primary intended
`purpose[]” limitations as the device definition’s “exclusionary
`clause” or exclusionary clauses, Classification of Products as Drugs
`and Devices & Additional Product Classification Issues: Guidance
`for Industry and FDA Staff, U.S. Dep’t of Health and Human Servs.,
`FDA,
`6
`&
`n.11
`(Sept.
`2017),
`https://www.fda.gov/media/80384/download, we refer to them as the
`“mode-of-action clauses” in order to distinguish them from the
`“instrument clause,” which also has the effect of “excluding” certain
`products that would otherwise satisfy the device definition. See
`supra n.1.
`
`
`
`

`

`5
`generally regulated by the FDA’s Center for Drug Evaluation
`and Research, devices are within the purview of the FDA’s
`Center for Devices and Radiological Health.
`
`The FDA holds new drugs to a high standard of pre-market
`review and approval. To market a new prescription drug, the
`sponsor (typically the manufacturer) must submit a new-drug
`application and demonstrate through clinical trials that the drug
`is safe and effective for its proposed use. 21 U.S.C. § 355(a)–
`(b). Sponsors may, however, be able to take advantage of an
`abbreviated new-drug application if their drug is sufficiently
`similar to drugs that the FDA has previously approved. Id.
`§ 355(j).
`
`The FDA’s pre-market review of devices is more varied.
`Devices are assessed by the FDA and, with the assistance of
`expert “classification panels,” classified into one of three
`categories based on the risks they pose. Id. § 360c. First are
`Class I devices, which are “subject only to minimal regulation
`by ‘general controls’” because they “present no unreasonable
`risk of illness or injury . . . .” Medtronic, Inc. v. Lohr, 518 U.S.
`470, 476–77 (1996) (quoting 21 U.S.C. § 360c(a)(1)(A)).
`Class II devices include “[d]evices that are potentially more
`harmful” and, “although they may be marketed without
`advance approval, manufacturers of such devices must comply
`with federal performance regulations known as ‘special
`controls.’” Id. at 477 (quoting 21 U.S.C. § 360c(a)(1)(B)).
`Finally, devices that, inter alia, “‘presen[t] a potential
`unreasonable risk of illness or injury,’ or which are ‘purported
`or represented to be for a use in supporting or sustaining human
`life or for a use which is of substantial importance in preventing
`impairment of human health’” are generally classified as Class
`III and, like drugs, subject to pre-market approval. Id.
`(alteration in original) (quoting 21 U.S.C. § 360c(a)(1)(C)); 21
`U.S.C. § 360e. To introduce a new Class III device into the
`
`
`
`

`

`6
`market, the sponsor must provide the FDA with “detailed
`information regarding the safety and efficacy” of the device
`and the FDA must have “‘reasonable assurance’ that the device
`is both safe and effective.” Medtronic, 518 U.S. at 477
`(quoting 21 U.S.C. § 360e(d)(2)).
`
`The regulatory differences do not end at the product
`approval stage. Throughout the lifecycle of a medical product,
`its treatment by the FDA depends upon its classification as
`either a drug or a device. The FDCA sets forth separate rules
`for, inter alia, annual manufacturer registration, compare 21
`U.S.C. § 360(b)(1)
`(registration
`requirements
`for drug
`manufacturers), with id. § 360(b)(2) (registration requirements
`for device manufacturers); routine manufacturer inspections,
`compare 21 U.S.C. § 360(h)(3)
`(risk-based
`inspection
`schedules for drug manufacturers), with id. § 360(h)(2) (risk-
`based inspection schedules for device manufacturers); routine
`product reporting, see id. § 356i (reporting of marketing status
`for drugs only); and adverse-event reporting, compare id.
`§ 355b (reporting of adverse drug events), with id. § 360i
`(records and reports on devices, including reporting of adverse
`device events).
`
`The result is that, on average, it is more costly for a sponsor
`to develop and market a product as a drug than it would be to
`develop and market an otherwise identical product as a device.
`Genus maintains that its cost would be approximately $60,000
`to seek device clearance for Vanilla SilQ—the product line in
`question here. Genus Med. Techs., LLC v. FDA, 427 F. Supp.
`3d 74, 78 (D.D.C. 2019). If, however, the same product line
`were classified as drugs, Genus estimates that it would cost
`them more than $500,000 to obtain pre-market approval in
`addition to a recurring cost of more than $186,000 per year to
`continue marketing their products as drugs. Id.
`
`
`
`

`

`7
`Fortunately for sponsors, the FDCA contemplates at least
`a limited role for sponsor input in the course of the product
`classification process. Specifically, if the classification of a
`product is unclear, a product sponsor may file a request for
`designation (RFD) to obtain a formal, binding determination
`from the FDA as to the “classification of the product . . . or . . .
`the component of the [FDA] that will regulate the product.” 21
`U.S.C. § 360bbb-2(a). A sponsor submits its RFD—including
`a recommended classification—to the FDA’s Office of
`Combination Products (OCP) and the OCP must respond
`thereto no later than 60 days after the RFD’s filing. Id.
`§ 360bbb-2(b). If the OCP fails to respond, the sponsor’s
`recommended classification becomes final. Id. § 360bbb-2(c).
`A classification made through the RFD process cannot be
`changed “except with the written consent of the [sponsor], or
`for public health reasons based on scientific evidence.” Id.
`§ 360bbb-2(b)–(c).
`
`B. Factual and Procedural History
`
`Genus has manufactured its Vanilla SilQ product line
`since 2015. Compl. ¶ 25. Vanilla SilQ belongs to a category
`of products known as contrast agents. Contrast agents are used
`in medical imaging to improve the visualization of tissues,
`organs and physiological processes. According to Genus,
`Vanilla SilQ is an oral solution used in combination with X-ray
`examinations or other radiologic procedures to enhance the
`visualization of the gastrointestinal tract for diagnostic
`purposes. The product’s key ingredient is an inert metal salt
`known as barium sulfate. When swallowed, the barium sulfate
`coats the inside of the individual’s gastrointestinal tract and
`facilitates the absorption of X-rays. Subsequently, the X-ray
`examination will appear lighter for areas coated with barium
`sulfate and darker for the surrounding tissues that are not
`coated. Although some contrast agents cannot be classified as
`
`
`
`

`

`8
`devices because they achieve their primary intended purpose
`through metabolization or chemical action within or on the
`body of man, the FDA agrees that Genus’s Vanilla SilQ
`product line “appear[s] to meet the definition of ‘device’”
`insofar as it does not achieve its primary intended purposes
`through either of the excluded modes.3 Appellant’s Br. 12–13.
`
`Genus avers “that before and after it started producing
`Vanilla SilQ, it sought FDA clearance to distribute its
`products” as either devices or grandfathered drugs (which,
`unlike new drugs, do not require pre-market approval). Genus,
`427 F. Supp. 3d at 79. In June 2016, however, the FDA
`conducted a three-day inspection of Genus’s distribution
`facility. Id. The result of the inspection was a warning letter,
`issued on May 2, 2017, notifying Genus that its products
`constituted “drugs” within the meaning of the FDCA. Id.
`Genus, responding to the FDA in a letter dated May 19, 2017,
`asserted that its products are devices and that the FDA could
`not regulate them as drugs because they do not “achieve [their]
`
`3 We note that it is not immediately obvious to us how a contrast
`agent satisfies the device definition’s requirement that the regulated
`product be “an
`instrument, apparatus,
`implement, machine,
`contrivance, implant, in vitro reagent, or other similar or related
`article, including any component, part, or accessory . . . .” 21 U.S.C.
`§ 321(h)(1). Nor is it altogether settled that Vanilla SilQ satisfies the
`device definition’s mode-of-action clauses. Compare, e.g., Amicus
`Bracco Br. 5–6 (arguing that Vanilla SilQ may not be regulated as a
`device because it achieves its primary intended purpose through
`chemical action), with Appellee’s Br. 52 (arguing that Vanilla SilQ
`does not achieve its primary intended purpose through chemical
`action). Because neither question is part of the administrative
`decision now under review—the FDA found only that Genus’s
`products “appear to meet” the device definition, see Joint Appendix
`(J.A.) 122, 152, and both parties continue to agree that they do—we
`reserve the question whether Vanilla SilQ satisfies the device
`definition’s instrument and mode-of-action clauses.
`
`
`
`

`

`9
`primary intended purposes through chemical action within or
`on the body” or through “metaboliz[ation].” J.A. 157–61
`(quoting 21 U.S.C. § 321(h)(1)). On September 6, 2018, the
`FDA responded that, “[a]lthough [the Vanilla SilQ products]
`appear to meet the definition of ‘device’ . . . they also meet the
`definition of ‘drug’ [under the FDCA] because they are articles
`intended for use in the diagnosis of disease.” Id. at 152. The
`FDA stated that “[w]hile [it] generally regulates products that
`meet the definition of a device under the device authorities,
`there are certain exceptions” and “[b]ecause not all contrast
`agents meet the definition of a device, but all of them do meet
`the definition of a drug, [it] has for many years regulated these
`products as drugs in order to regulate them consistently under
`the same authority . . . .” Id.
`
`its
`through
`the FDA
`to convince
`Having failed
`correspondence, Genus next submitted an RFD, in which it
`formally requested that the OCP classify its Vanilla SilQ
`products as devices under the FDCA. Genus, 427 F. Supp. 3d
`at 79–80. The OCP responded with an official Designation
`Letter in which it echoed the FDA’s previous reasoning that,
`although the Vanilla SilQ products appeared to meet the
`definitions for both a device and a drug, it was nonetheless
`appropriate to regulate uniformly all contrast agents as drugs.
`Id. at 80.
`
`On February 28, 2019 Genus filed suit in district court. In
`addition to certain claims not relevant here, Genus claimed that
`the FDA’s decision to regulate Vanilla SilQ as a drug rather
`than as a device was arbitrary and capricious and in excess of
`statutory authority under the FDCA and the APA. In a decision
`filed December 6, 2019, the district court granted summary
`judgment to Genus, concluding that the plain language of the
`FDCA unambiguously requires that “a product that meets the
`device definition must be regulated as such” and that the court
`
`
`
`

`

`10
`must therefore “end[] its analysis at Chevron step one.” Genus,
`427 F. Supp. 3d at 84. The district court vacated the FDA’s
`classification of Vanilla SilQ as a drug and remanded the
`matter to the FDA for further proceedings. Id. at 87.
`
`II. ANALYSIS
`
`Our review of a summary judgment grant is de novo,
`affirming only if “there is no genuine issue as to any material
`fact [and] the moving party is entitled to judgment as a matter
`of law.” Mylan Labs., Inc. v. Thompson, 389 F.3d 1272, 1278–
`79 (D.C. Cir. 2004) (alteration in original) (quoting Trans
`Union LLC v. Fed. Trade Comm’n, 295 F.3d 42, 48 (D.C. Cir.
`2002)). In a case like this one, in which the district court
`reviewed an agency action under the APA, “[w]e review the
`administrative record and give no particular deference to the
`District Court’s views.” Eagle Pharms., Inc. v. Azar, 952 F.3d
`323, 329–30 (D.C. Cir. 2020) (quoting Am. Bankers Ass’n v.
`Nat’l Credit Union Admin., 934 F.3d 649, 662 (D.C. Cir.
`2019)). We review the FDA decision to classify Genus’s
`products,
`then, under
`the
`familiar standards of
`the
`Administrative Procedure Act, which require that we uphold
`the FDA decision unless it is “arbitrary, capricious, an abuse of
`discretion, or otherwise not in accordance with law” or “in
`excess of statutory jurisdiction, authority, or limitations . . . .”
`5 U.S.C. § 706(2). We defer to the FDA’s interpretation of the
`FDCA “so long as the Congress has not unambiguously
`forbidden it and it is otherwise permissible.” Cal. Metro
`Mobile Commc’ns, Inc. v. FCC, 365 F.3d 38, 43 (D.C. Cir.
`2004) (citing Chevron U.S.A., Inc. v. Nat. Res. Def. Council,
`Inc., 467 U.S. 837, 842–43 (1984)); see also Teva Pharms.
`USA, Inc. v. Sebelius, 595 F.3d 1303, 1315 (D.C. Cir. 2010)
`(applying Chevron framework to FDA interpretations of
`FDCA contained in letter rulings); Mylan Labs., 389 F.3d at
`1279–80 (same). Our task requires that “[w]e examine the
`
`
`
`

`

`11
`statute’s text, structure, purpose, and legislative history to
`determine
`if
`the Congress has expressed
`its
`intent
`unambiguously.” Eagle Pharms., 952 F.3d at 330 (alteration
`omitted) (quoting U.S. Sugar Corp. v. EPA, 830 F.3d 579, 605
`(D.C. Cir. 2016) (per curiam)).
`
`A. FDCA’s Text
`The question before us is a purely legal one: whether the
`FDCA grants the FDA discretion to classify as a “drug” a
`product that satisfies the statutory definitions of both a “drug”
`and a “device.” In answering the question, “[w]e begin ‘where
`all such inquiries must begin: with the language of the statute
`itself.’” Caraco Pharm. Labs., Ltd. v. Novo Nordisk A/S, 566
`U.S. 399, 412 (2012) (quoting United States v. Ron Pair
`Enters., Inc., 489 U.S. 235, 241 (1989)). We are mindful,
`however, that if the text alone is insufficient to end the inquiry,
`we may turn to other “customary statutory interpretation tools,”
`including “‘structure, purpose, and legislative history.’” Cal.
`Metro Mobile, 365 F.3d at 44–45 (quoting Consumer Elecs.
`Ass’n v. FCC, 347 F.3d 291, 297 (D.C. Cir. 2003)); see also
`Chevron, 467 U.S. at 843 n.9 (“If a court, employing traditional
`tools of statutory construction, ascertains that Congress had an
`intention on the precise question at issue, that intention is the
`law and must be given effect.”) (emphasis added). We
`conclude that the FDCA’s text unambiguously forecloses the
`FDA’s interpretation.
`
`The parties’ dispute is purely legal. Genus contends that
`when a product satisfies both the drug and device definitions of
`the FDCA, the product is a device. Conversely, the FDA
`argues that it can choose whether to treat products that satisfy
`both definitions as drugs or devices. Because the FDA’s legal
`theory did not require it to do so, it made no factual findings
`about whether the Vanilla SilQ products satisfied the particular
`
`
`
`

`

`12
`requirements of the FDCA’s device definition. Instead, it
`found only that the products fell within the drug definition and
`remarked that they “appear” to also satisfy the device
`definition.
`
`Beginning with the statute’s text, the FDA argues that,
`because it is possible for a product to simultaneously satisfy the
`linguistic demands of both the drug and device definitions, the
`Congress must have granted the FDA discretion in such
`instance to choose a classification. Simply put, any product
`meeting the device definition may be classified as a device, any
`product meeting the drug definition may be classified as a drug
`and, according to the FDA’s reading, any product meeting both
`definitions may be classified as either. To the extent the FDCA
`is silent on how to treat products that meet both definitions, the
`FDA argues that we should read it as a sign of statutory
`ambiguity and defer to the FDA’s purportedly reasonable
`interpretation.
`
`Whereas the FDA draws our attention to the definitional
`overlap, Genus urges us to focus on the elements of the device
`definition that set it apart, including, most prominently, its
`mode-of-action clauses. Genus argues that, because the drug
`and device definitions are broadly similar except for the device
`definition’s mode-of-action clauses—excluding products that
`achieve their primary intended purposes through “chemical
`action within or on the body of man” or “metaboliz[ation],” 21
`U.S.C. § 321(h)(1)—products that do not achieve their primary
`intended purposes through either excluded mode (and that
`otherwise satisfy both definitions) must be regulated as devices
`and devices alone.
` According
`to Genus, any other
`interpretation would “effectively read[] the Mode of Action
`Clause[s] out of the statute.” Appellee’s Br. 23.
`
`
`
`

`

`13
`Genus also urges us to rely on two traditional canons of
`statutory construction. First, the “old and familiar rule” that
`“the specific governs the general.” RadLAX Gateway Hotel,
`LLC v. Amalgamated Bank, 566 U.S. 639, 645–46 (2012)
`(quoting United States v. Chase, 135 U.S. 255, 260 (1890);
`Morales v. Trans World Airlines, Inc., 504 U.S. 374, 384
`(1992)). And second (and relatedly), the basic interpretive
`canon that a “statute should be construed [to give effect] to all
`its provisions, so that no part will be inoperative or superfluous,
`void or insignificant.” Corley v. United States, 556 U.S. 303,
`314 (2009) (quoting Hibbs v. Winn, 542 U.S. 88, 101 (2004)).
`Applying these canons, Genus argues that the FDA’s
`interpretation would render the device definition’s mode-of-
`action clauses inoperative and allow the device definition to be
`“swallowed by the more general drug definition.” Appellee’s
`Br. 30–31 (quoting Genus, 427 F. Supp. 3d at 83).
`
`the FDA’s
`that
`Although we are unpersuaded
`interpretation would render
`the mode-of-action clauses
`completely inoperative—under the FDA’s theory, the mode-
`of-action clauses would still be necessary for differentiating
`products that may be regulated as devices from those that may
`not—we nonetheless agree with Genus that this is a case where
`the specific must govern the general. The FDA does not
`dispute that the FDCA’s definition of a “device” is drawn more
`narrowly than its definition of a “drug.” Indeed, as we
`discussed above, supra Section I.A, the set of products that
`satisfy the device definition is necessarily encompassed by, but
`narrower than, the set of products that satisfy the drug
` Moreover,
`the general-specific canon
`is
`definition.4
`
`4 The concurring opinion contends that recognizing complete
`overlap in the definitions would render the instrument clause
`surplusage. Concurring Op. 7–8. But the instrument clause, like the
`mode-of-action clauses, necessarily restricts which medical products
`are devices. It does clear work in determining which medical
`
`
`
`

`

`14
`particularly appropriate where, as here, the provisions at issue
`are “interrelated and closely positioned” as “parts of the same
`statutory scheme.” RadLAX Gateway Hotel, 566 U.S. at 645
`(alteration adopted) (quoting HCSC-Laundry v. United States,
`450 U.S. 1, 6 (1981) (per curiam)). Thus, the device
`definition’s instrument and mode-of-action clauses make it a
`classic candidate for application of the canon that the specific
`governs the general, and to the extent the drug and device
`definitions conflict, it is the narrower definition—the device
`definition—to which we must give effect. See D. Ginsberg &
`Sons, Inc. v. Popkin, 285 U.S. 204, 208 (1932) (“Specific terms
`prevail over the general in the same or another statute which
`otherwise might be controlling.”).
`
`The only question, then, is whether the two definitions are
`truly in conflict. The FDA claims they are not. More
`specifically, according to the FDA, the general-specific canon
`is inapplicable here because it is “most frequently applied to
`statutes in which a general permission or prohibition is
`contradicted by a specific prohibition or permission” or where
`“a general authorization and a more
`limited, specific
`authorization exist side-by-side.” RadLAX Gateway Hotel, 566
`U.S. at 645. The FDA argues that there is no such contradiction
`here because the provisions in question are definitions as
`opposed to authorizations or prohibitions and both definitions
`can be given simultaneous effect. There is “no reason,”
`according to the FDA’s opening brief, that “the statute must be
`read so that a given product qualifies as either a ‘drug’ or a
`‘device,’ but not both.” Appellant’s Br. 24 (emphasis in
`original).
`
`On this point the FDA is mistaken. In theory, it may be
`possible for a product to satisfy both definitions at once. What
`
`products, from among those that satisfy the broader drug definition,
`also satisfy the narrower device definition.
`
`
`
`

`

`15
`the FDA omits, however, is that the FDCA’s statutory
`definitions are meaningful only insofar as they carry concrete
`regulatory consequences. As discussed, the FDCA elaborates
`distinct regulatory regimes for drugs and devices. And each
`scheme is mandatory: The FDCA prohibits the sale of “any
`new drug” not approved under the regime for drug approvals.
`21 U.S.C. § 355(a) (emphasis added). Similarly, all new Class
`III devices are “required” to satisfy the pre-market review
`regime for devices, id. § 360e(a), and Class I and Class II
`devices must meet other distinct requirements, see id. § 360c.
`Nor can the Secretary circumvent these requirements. Id.
`§ 355(c)(1) (Secretary “shall” either approve new-drug
`application pursuant to drug regime or deny application);
`§ 360c(b)(1) (Secretary “shall” classify “all” new devices
`intended for human use into three device classes). In short, it
`is not textually possible to say that an item is a drug (or device)
`but need not be regulated as such. And no one suggests that
`the FDCA requires products meeting both definitions to be
`regulated both as drugs and devices, which would create a
`breathtaking example of statutory redundancy. The statute,
`then, is clear: a product may be regulated as a drug or a device,
`but not both, and while a single product may simultaneously
`satisfy the linguistic elements of two definitions, it is not
`possible for the FDA to give simultaneous effect to both. Thus,
`this is precisely the sort of setting in which we must give effect
`to the specific over the general. To do otherwise would be in
`violation of the “settled” principle that “[h]owever inclusive
`may be the general language of a statute, it will not be held to
`apply to a matter specifically dealt with in another part of the
`same enactment.” Fourco Glass Co. v. Transmirra Prods.
`Corp., 353 U.S. 222, 228 (1957) (quoting Clifford F. MacEvoy
`Co. v. U.S. ex rel. Calvin Tomkins Co., 322 U.S. 102, 107
`(1944)).
`
`
`
`

`

`16
`Before proceeding to the parties’ structural claims, we
`briefly dispatch with the FDA’s argument that we should be
`guided by a 1990 amendment to the FDCA’s drug definition.
`Specifically, the FDA argues that interpreting the drug and
`device definitions as mutually exclusive would be
`to
`“effectively read[] back into the statute” an old version of the
`drug definition the Congress affirmatively abandoned when it
`adopted the Safe Medical Devices Act of 1990 (SMDA), Pub.
`L. No. 101-629, 104 Stat. 4511. Appellant’s Br. 21–22. Before
`1990, the FDCA definition of a drug specifically excluded
`“devices or their components, parts, or accessories.” FDCA,
`Pub. L. No. 75-717, § 201(g), 52 Stat. 1040, 1041 (1938)
`(codified as amended at 21 U.S.C. § 321(g)(1)). The 1990
`SMDA struck this exclusionary language, thereby making it
`possible for a single product to satisfy—simultaneously—the
`terms of both definitions. SMDA § 16(b)(1), 104 Stat. at 4526.
`The FDA argues that, by eliminating the drug definition’s
`exclusionary language, the Congress granted it authority to
`regulate certain products as either drugs or devices.
`
`This argument presumes that the Congress dramatically
`expanded the FDA’s authority by deleting a phrase from a
`statutory definition. As the Supreme Court has counseled,
`“[f]undamental changes in the scope of a statute are not
`typically accomplished with so subtle a move.” Kellogg Brown
`& Root Servs., Inc. v. U.S. ex rel. Carter, 575 U.S. 650, 661
`(2015). Instead, we conclude that “the removal of the
`. . . provision was more plausibly driven by” a narrower
`concern. Id. The change occurred in a section of the statute
`authorizing the FDA to “regulate products that constitute a
`combination of a drug, device, or biological product,”
`depending on “the primary mode of action of the combination
`product.” SMDA § 16(a), 104 Stat. at 4526 (codified as
`amended at 21 U.S.C. § 353(g)) (authorizing the regulation of
`“combination products”). These new provisions thus created a
`
`
`
`

`

`17
`distinct regulatory regime that gave the Secretary flexibility to
`determine the standards for pre-market review of these
`combination
`products.
`
`e.g.,
`21 U.S.C.
`See,
`§ 353(g)(2)(A)(ii)(I)
`(entitling sponsors of combination
`products to meet with the Secretary to “address the standards
`and requirements for market approval or clearance of the
`combination product”); id. § 353(g)(7) (“Nothing in this
`subsection shall prevent the Secretary from using any agency
`resources of the [FDA] necessary to ensure adequate review of
`the safety, effectiveness, or substantial equivalence of an
`article.”). The definitional change helped to implement the
`scheme by removing the previously categorical prohibition on
`ever treating a drug as a device and vice versa. See Miller v.
`Mylan Inc., 741 F.3d 674, 677 (6th Cir. 2014) (“The deletion
`reflected the replacement of the binary scheme with a tripartite
`scheme[ that included combination products].”). But the
`amended definition provides no affirmative support for the
`proposition that the FDA may treat drugs as devices—and vice
`versa—even absent any combination. As explained above, the
`FDCA’s basic textual architecture forecloses such reading of
`the statute.
`
`Legislative history confirms that the amendments seek
`only to facilitate the FDA’s regulation of the new category of
`“combination products.”5 See S. Rep. No. 101-513, at 43
`(1990) (“Section 19 [of the SMDA] alters the drug and device
`definitions in [21 U.S.C. § 321]. Language is removed from
`the drug definition that will permit an approval of a
`drug/device combination.”) (emphasis added); id. at 30 (“By
`deleting this language, a product whose primary mode of action
`
`5 We note that our analysis of the FDA’s argument regarding
`the 1990 SMDA depends upon the FDCA’s legislative history, which
`we further discuss in the following section, infra Section II.B.
`Because the FDA’s SMDA argument is primarily textual, we address
`it here.
`
`
`
`

`

`18
`is attributable to a drug, but has a device component, may be
`reviewed under this Act’s drug authority.”). Thus, we read the
`SMDA to facilitate the regulation of combination products, not
`to grant the FDA near-limitless discretion to categorize as
`drugs any product meeting the device definition.6
`
`B. FDCA’s Structure, Purpose and Legislative History
`We turn next to the FDCA’s structure, purpose and
`legislative history. See, e.g., Pharm. Rsch. & Mfrs. of Am. v.
`Thompson, 251 F.3d 219, 224 (D.C. Cir. 2001); see also
`Roberts v. Sea-Land Servs., Inc., 566 U.S. 93, 101 (2012)
`(“[Because s]tatutory language . . . ‘cannot be construed in a
`vacuum . . . [i]t is a fundamental canon of statutory construction
`that the words of a statute must be rea