836 F2d 135 Tri-Bio Laboratories Inc v. United States

836 F.2d 135

UNITED STATES of America and Food and Drug Administration, Appellees.

No. 87-5123.

United States Court of Appeals,
Third Circuit.

Submitted Pursuant To Third Circuit

Rule 12(6) Aug. 19, 1987.
Decided Dec. 29, 1987.
Rehearing and Rehearing En Banc Denied Feb. 1, 1988.

Delbert J. McQuaide, Steven S. Hurvitz, Darryl R. Slimak, McQuaide, Blasko, Schwartz, Fleming & Faulkner, Inc., State College, Pa., for appellant Tri-Bio Laboratories, Inc.

Richard K. Willard, Asst. Atty. Gen., James J. West, U.S. Atty., Margaret A. Cotter, Asst. Director, Office of Consumer Litigation, Gerald C. Kell, Office of Consumer Litigation, Civ. Div., U.S. Dept. of Justice, Washington, D.C. (Thomas Scarlett, Chief Counsel, Eric M. Blumberg, Associate Chief Counsel for Enforcement Food and Drug Admin., Rockville, of counsel), for appellees U.S. and Food and Drug Admin.

Before GIBBONS, Chief Judge, and WEIS, Circuit Judge, and KELLY,* District Judge.


WEIS, Circuit Judge.

view counter

A pharmaceutical manufacturer filed for Food and Drug Administration approval of a generic animal drug, incorporating in its application the research and testing data submitted by another manufacturer that earlier had obtained approval to market the predecessor brand name drug. Because of its policy treating such data as proprietary and confidential, the FDA refused to consider the previously filed material. The district court sustained the agency action and entered summary judgment against the generic manufacturer. We will affirm.


Plaintiff manufactures Gentaject, a veterinary pharmaceutical product recommended for use in one day old chicks to prevent early mortality caused by three species of bacteria.1 Gentaject, a generic drug,2 contains the same ingredients as a previously approved product manufactured by the Schering Corporation of Kenilworth, New Jersey and currently marketed under the brand name "Garasol."


Garasol was patented in 1963 and approved for commercial distribution by the FDA in 1978. The patent expired in 1980. A year later, plaintiff sought FDA approval to market its generic copy of Garasol under the brand name Gentaject. As required by statute and FDA regulations, plaintiff filed a new animal drug application (NADA) to demonstrate that its product was safe and effective for its intended use. However, instead of including the extensive test data ordinarily required in new applications, the plaintiff's application merely referred the FDA to Schering's test data in support of the original Garasol application. Because of the plaintiff's failure to submit original safety and effectiveness data, the FDA rejected the application as incomplete.


In October 1982 plaintiff renewed its request for FDA approval, asserting that Gentaject was not a "new drug" requiring the usual panoply of testing. Plaintiff equated its request to a remarketing of a previously FDA-approved drug (Garasol) under a new label (Gentaject). Plaintiff contended that Gentaject, as the bioequivalent3 of Garasol, in essence already had received FDA approval. But the FDA reaffirmed its denial of the plaintiff's request, citing longstanding agency policy that all new animal drug applications "must stand or fall on the basis of the submitted data."


After some preliminary, inconclusive legal skirmishing, plaintiff agreed to file a citizen's petition4 with the FDA, requesting the agency to declare that Gentaject was not a new animal drug and, thus, did not require a full application. In the petition plaintiff asserted that Gentaject was not a "new drug" within the meaning of the Food, Drug and Cosmetic Act, 21 U.S.C. Sec. 360b(a)(1), because the drug formulation--marketed under the Garasol label--was generally recognized as safe and effective for its intended use by experts qualified to make such judgments.


The Commissioner denied the citizen's petition. The agency concluded that Gentaject was a new animal drug within the meaning of the Act and that it had not become generally recognized as safe and effective.


Plaintiff then brought this action, asking the district court to declare that Gentaject was not a new drug, to direct the FDA to approve its application by incorporating Schering's data, or to reconsider the citizen's petition using the data previously submitted by Schering. Alternatively, plaintiff urged that the FDA approve Gentaject based on agency in-house research--conducted in defense of the litigation--through which the FDA had found Gentaject in fact safe and effective for its intended use. The parties filed cross-motions for summary judgment.

view counter

The district court granted the FDA's motion and entered judgment against plaintiff. The court concluded that the FDA properly interpreted its statutory authority in formulating the policy of rejecting applications which depended on data submitted by previous applicants. The court further determined that the FDA's rejection of the citizen's petition was not arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.


On appeal plaintiff argues that the FDA's approach is unreasonable because it compels expensive and indefensible duplicative testing of drugs already approved for commercial marketing. Plaintiff also asserts that the FDA did not adequately analyze all the submitted evidence, a review which plaintiff contends would have revealed that Gentaject was not a "new drug". In the absence of a proper factual hearing by the agency, plaintiff maintains that the district court should have examined the issue de novo.


The FDA responds that plaintiff failed to present the requisite safety and effectiveness data for its product and that, without such evidence, the Food, Drug and Cosmetic Act mandates rejection of the plaintiff's application.



The federal government began premarketing regulation of pharmaceuticals with the Food, Drug, and Cosmetic Act of 1938, Pub.L. No. 75-717, 52 Stat. 1040 (1938) (codified as amended at 21 U.S.C. Secs. 301-92). After passage of the 1938 Act, FDA approval was necessary before marketing any drug not generally recognized as safe by the scientific community. The Act assumes that a new drug is adulterated and thus unsafe for public distribution until it has been approved. See 21 U.S.C. Secs. 355(a), 360b(a)(1).


By 1962, however, the FDA had begun approving generic copies of established "pioneer" drugs5 without the usual full-scale application ordinarily required for premarketing approval. As to these generics, the FDA concluded that, because the pioneers were generally recognized as safe and were used "to a material extent or for a material time," the regular new drug application was unnecessary. See Flannery & Hutt, Balancing Competition and Patent Protection in the Drug Industry, 40 Food Drug Cosm. L.J. 269, 272 (1985).


In 1962 Congress amended the Act to require that a new drug applicant establish to the FDA's satisfaction not only the product's safety, but also its effectiveness for the intended use. Drug Amendments of 1962, Pub.L. No. 87-781, 76 Stat. 780 (1962). The retroactive amendments necessitated FDA-retesting of all drugs approved before 1962 for compliance with the new effectiveness standard. To carry out this enormous mandate, the FDA adopted abbreviated procedures. The active ingredients of 4,000 pioneer drug formulations were analyzed. The drugs that survived this scrutiny no longer were considered to be "new drugs" and, thus, were relieved of the arduous, formal reapplication process.


The FDA also decided that a generic drug applicant need not duplicate a pre-1962 pioneer drug's testing; instead, the copycat manufacturer could submit an abbreviated new drug application (ANDA). This abbreviated application need recount only bioequivalence and bioavailability,6 demonstrating that the generic was identical to the pioneer drug.


The FDA's policy differed, however, with respect to generics using formulations of pioneer products approved after 1962. The agency adopted the policy of treating as confidential and proprietary the safety and effectiveness data prepared and submitted by these pioneer drug manufacturers. In the FDA's view, use of this post-1962 research by the agency to review generic drug applications would constitute expropriation.


The FDA's restrictive policy on abbreviated applications thus required generic manufacturers to duplicate many of the pioneers' original testing procedures. The additional expense inherent in this policy necessarily increased the price to consumers of generic drugs. Nevertheless, the FDA feared that appropriation by "me-too" drug manufacturers of data gathered by the pioneer applicants at considerable expense would discourage the development of new products and new uses for existing ones.


Pharmaceuticals for humans and animals were regulated under the same statute until 1968 when Congress recodified the animal drug provisions. Animal Drug Amendments of 1968, Pub.L. No. 90-399, 82 Stat. 343 (1968). After this separation, the FDA continued to impose the same restrictions on abbreviated applications for new animal drugs as it had followed for human drugs.


In 1984 Congress adopted the Drug Price Competition and Patent Term Restoration Act, Pub.L. No. 98-417, 98 Stat. 1585 (1984) (codified at 21 U.S.C. Sec. 355). This statute attempts to balance the interests of the generic drug manufacturers, who sought to avoid unnecessary testing, against the research investments of the pioneer manufacturers, at the same time mindful of the public need for safe commercial drugs. The 1984 Amendments, which applied only to human drugs, reflect a statutory compromise of the competing concerns.


The 1984 Act provides an abbreviated application procedure for generic human drugs demonstrating bioequivalency with pioneer drugs approved either before or after 1962. In return, the Act grants pioneer drug manufacturers a statutorily defined period of market exclusivity during which no abbreviated application can be approved. See 21 U.S.C. Sec. 355(j)(4)(D)(i)-(v). This market exclusivity lengthens the pioneer's marketing monopoly beyond the expiration of the drug's patent. In addition, the Act also extends the term of the patent by a period equal to the distribution time lost during the FDA's premarketing testing and approval process. See 35 U.S.C. Sec. 156.


As noted earlier, the 1984 Amendments applied only to human pharmaceuticals. The FDA's no-abbreviated application policy remained in effect for animal drugs. Congress was aware of the 1984 Act's limited scope, and bills to similarly revise the animal drugs statute were introduced in both the Senate and House. See S. 2407, 99th Cong., 2d Sess. (1986); H.R. 5069, 99th Cong., 2d Sess. (1986). Congress adjourned before it could act on the legislation, and the animal drug bills were reintroduced in the 100th Congress. See H.R. 3120, 100th Cong., 1st Sess. (1987). In the absence of congressional accommodation of these conflicting interests of the generic and pioneer animal drug manufacturers, the FDA's policy continues to protect the pioneer's investment at substantial cost to generic manufacturers.


With this general background in mind, we turn to the merits of the plaintiff's appeal. First, plaintiff challenges the FDA's no-abbreviated application policy contained in 21 C.F.R. Sec. 514.1(a) as an invalid agency interpretation of the congressional mandate. Second, plaintiff asserts that the FDA erred in declaring Gentaject a "new drug" requiring the full panoply of premarket testing.



The plaintiff's objection to the application process rests on the premise that, because the results of the required retesting are known in advance, the current policy serves no valid scientific purpose. Plaintiff asserts that the FDA practice contravenes congressional intent and lacks a reasonable basis in law.


The principal rationale the agency offers in defense of its policy is that pioneer manufacturers possess a property interest in the test data they present to support their new drug applications. The FDA posits that this proprietary interest may not be appropriated by the government without just compensation. Because the statute indicates no evidence of congressional intent contemplating payment for that interest, no realistic alternative to the policy is evident.


In a statutory arrangement similar to the one considered here, the Federal Environmental Pesticide Control Act, 7 U.S.C. Secs. 136-137, requires submission of test data before approval for commercial marketing of a pesticide. In Chevron Chemical Co. v. Costle, 641 F.2d 104 (3d Cir.), cert. denied, 452 U.S. 961, 101 S.Ct. 3110, 69 L.Ed.2d 972 (1981), the applicant's property interest in the filed research material was at issue. We concluded there that the pioneer manufacturer had no proprietary right in the requested information because, during the pertinent period, federal law did not create in the submitter a legitimate expectation that the agency would not use the data. The manufacturer was unable to cite persuasive authority that state law on trade secrets would restrict the agency's right to review the pioneer's research data in processing later applications of generic producers.


The Supreme Court was confronted with an aspect of this problem not addressed in Costle. In Ruckelshaus v. Monsanto Co., 467 U.S. 986, 104 S.Ct. 2862, 81 L.Ed.2d 815 (1984), a pesticide manufacturer sued to challenge amendments to the federal pesticide law which permitted the agency to consider portions of the pioneer's marketing application when evaluating later submissions by other applicants. The pioneer asserted that the challenged provisions constituted a Fifth Amendment taking, entitling the manufacturer to just compensation.


At the outset, the Supreme Court found that the submitted data constituted a property interest under state law. See id. at 1001-04, 104 S.Ct. at 2872 (citing Restatement (First) of Torts Sec. 757 comment b (1939)). Next, to determine whether the agency's internal reference to one manufacturer's research material for review of another's amounts to a Fifth Amendment taking, the Court considered whether first the manufacturer had a basis for a "reasonable investment-backed expectation" of non-use by the agency. Id. at 1005-06, 104 S.Ct. at 2874. On this point the Court concluded that, in the absence of any "provision of law" proscribing internal agency use, the pioneer could not have had a reasonable expectation that the agency would not consult the submitted data to evaluate applications filed by other manufacturers. Id. at 1008. Thus, for data filed in the period during which no federal law protected the material from internal agency use, the Court ruled that such use would not constitute a "taking" under the Fifth Amendment.


A different result emerged, however, for the time between 1972 and 1978, during which Congress had provided protection against internal agency reference. The Court determined that, for data submitted during this period, Congress had expressly guaranteed pioneers "an extensive measure of confidentiality and exclusive use." Id. at 1011, 104 S.Ct. at 2877. This guarantee, the Court concluded, "formed the basis of a reasonable investment-backed expectation," and agency use for evaluating other applications constituted a taking exposing the government to Tucker Act liability for damages. Id.


The Monsanto holding is relevant to whether Schering had a protectible property interest here. We anticipated the Monsanto variation in Costle, commenting that the pioneer manufacturer had "shown us no federal statute preventing internal agency use of the contents of files compiled in the performance of the agency's statutory functions." Costle, 641 F.2d at 115. A footnote contrasted this statement to a regulation of the FDA, 21 C.F.R. Sec. 314.1(b) (1980), which prohibited use of agency files to register new human drug compounds without the original registrant's permission. Id. at 115 n. 26. The cited regulation parallels one adopted in 1971 for use in the animal drug application context. See 21 C.F.R. Sec. 514.1(a) (1987).


The regulation, which memorialized the FDA policy, included the following provision: "Any reference to information furnished by a person other than the applicant may not be considered unless its use is authorized in a written statement signed by the person who submitted it." 21 C.F.R. Sec. 514.1(a) (1987) (originally codified at 21 C.F.R. Sec. 135.4a(a), redesignated to 21 C.F.R. Sec. 514.1(a) by 40 Fed.Reg. 13802, 13825 (1975)). Although in Monsanto the governmental guarantee against internal agency use was grounded on the statute, we are convinced that this agency regulation provided pioneer animal drug manufacturers with an equally reasonable investment-backed expectation that the FDA would refrain from nonconsensual use of research material.


We assume at this point that this regulation was within the FDA's legitimate rulemaking authority, although that factor is not determinative here. As the Supreme Court noted, "the relevant consideration for our purposes is the nature of the expectations of the submitter at the time the data were submitted." Monsanto, 467 U.S. at 1013 n. 17, 104 S.Ct. at 2878 n. 17. At the time Schering sent its data to the FDA, the regulation had been in effect for some years and had received a consistent interpretation by the FDA, thus entitling Schering to rely on it. Accordingly, the FDA's regulation--as published and implemented--had created for Schering a property interest in its data. Use of that material in processing the plaintiff's Gentaject application, therefore, would constitute a Fifth Amendment taking, requiring payment of compensation by the government.


Whatever the merits of the argument that before 1971 the FDA had interpreted the Act unreasonably, clearly the agency routinely followed its regulation before 1978, when it approved Garasol. By that time, an investment-backed expectation, perhaps questionable before the 1971 promulgation of the regulation, had ripened into a concrete property right. Once that interest sprouted, Monsanto required the FDA to read the Food, Drug, and Cosmetic Act to support its policy interpretation.7


We find nothing in the Act evidencing any intent to have the government compensate the pioneer registrant for the use of its data to review a "me-too" application. Such a procedure would amount to a virtual government subsidy of the generic animal drug manufacturers. The statutory language does not envision such an outcome, nor has plaintiff referred us to any legislative history justifying that result.


In sum, in light of the 1971 regulation, the statute does not now permit the FDA's use of the pioneer manufacturer's data in the processing of the generic manufacturer's new drug application.



As a second argument in support of reversal, plaintiff contends that the FDA erred in classifying Gentaject as a new animal drug. Section 360b requires that all new animal drugs be evaluated on the filing of a complete application consisting of "full reports of investigations which have been made to show whether or not such drug is safe and effective for use." 21 U.S.C. Sec. 360b(b). Plaintiff argues that because Gentaject is an exact duplicate of a previously approved drug, Gentaject is not a "new drug" necessitating the same exhaustive testing as its pioneer.


The Act designates all animal drugs "new" unless they have been "generally recognized, among experts qualified by scientific training and experience to evaluate the safety and effectiveness of animal drugs, as safe and effective" and have "been used to a material extent or for a material time." 21 U.S.C. Sec. 321(w). In order to be "generally recognized" as safe and effective, an animal drug must have a general reputation in the scientific community for those properties. "[E]ither the unawareness of the drug product by experts generally or a genuine dispute among qualified experts regarding a drug product's safety and effectiveness preclude its qualifying for exclusion as 'generally recognized'." Premo Pharmaceutical Laboratories, Inc. v. United States, 629 F.2d 795, 803 (2d Cir.1980).


This exclusion for "generally recognized" drugs is a narrow one. Id. at 802. To establish general recognition, the applicant must produce "evidence consisting of adequate and well-controlled investigations, including field investigation, by experts qualified by scientific training and experience to evaluate the effectiveness of the drug...." 21 U.S.C. Sec. 360b(d)(3). In addition to this clinical experimentation, the investigation should be "backed by substantial support in scientific literature." Weinberger v. Bentex Pharmaceuticals, Inc., 412 U.S. 645, 652, 93 S.Ct. 2488, 2493, 37 L.Ed.2d 235 (1973). The ultimate hurdle of general recognition must be overcome by "substantial evidence." Weinberger v. Hynson, Wescott & Dunning, Inc., 412 U.S. 609, 629, 93 S.Ct. 2469, 2483, 37 L.Ed.2d 207 (1973).


The plaintiff's petition in this case depended heavily on Schering's Garasol data, yet it did not refer to any established body of published literature on Garasol. Moreover, the Commissioner found deficiencies in those materials submitted in support of Garasol's general recognition status. Although these shortcomings played a part in the FDA's denial, it is clear that once the agency properly determined that Schering's proprietary data could not be expropriated to support the plaintiff's petition, the foundation for the citizen's petition collapsed.


After a careful and searching review of the findings of the FDA in rejecting the plaintiff's petition, the district court refused to overturn the denial of the citizen's petition.


We are mindful that in evaluating scientific evidence in the drug field, the FDA possesses an expertise entitled to respectful consideration by this court. Bentex, 412 U.S. at 653-54, 93 S.Ct. at 2494. After a review of the record and the arguments supporting reversal, we are not persuaded to differ with the FDA's decision that the plaintiff's petition did not present adequate scientific evidence to sustain a finding of general recognition.



Finally, plaintiff points out that the FDA, following its own independent testing performed in connection with the litigation in the district court, has agreed that Gentaject is, in fact, safe and effective for its intended use. This concession, however, does not alter our conclusion.


By its terms, the Act provides no exclusion for new drugs the FDA independently has determined to be actually safe and effective. The only statutory exception available to plaintiff relates to drugs that the scientific community generally accepts as safe and effective. This omission might reflect Congress' desire to avoid burdening the FDA with extraordinary drug testing responsibilities, or simply Congress' preference for conclusions reached by the independent scientific community. Whatever the reason, the statute does not permit approval on the basis of FDA test results.8


Moreover, even if this court could permit plaintiff to rely on the FDA's testing, the petition was nonetheless properly denied. The agency's research was performed after litigation had begun. As a consequence, the drug application, at the time of filing, still lacked the essential prerequisites for approval.


Practical necessity also counsels rejection of the plaintiff's reliance on the agency's data. To hold otherwise would impose on the FDA a new and onerous responsibility. Such a ruling would enable a generic drug manufacturer to circumvent the clear requirements of section 360b by filing an incomplete application, which cites the pioneer's data, and afterward initiating litigation after FDA denial. The applicant would simply wait for the FDA to conclude its pre-litigation testing, and then point out to the court any favorable, government-financed results. The prohibitive costs of this scenario, both budgetary and in terms of agency personnel, mandate that this option be promptly rejected.


As the Court observed in Board of Governors of the Fed. Reserve Sys. v. Dimension Financial Corp., 474 U.S. 361, 374 n. 7, 106 S.Ct. 681, 689 n. 7, 88 L.Ed.2d 691 (1986), "[t]he process of effectuating congressional intent at times may yield anomalies," and "the explicit language of a statute" in application may produce "a curious result." Nevertheless, "[n]oting that nothing prohibited Congress from passing unwise legislation, we upheld the enforcement of the statute as Congress had written it." Id. The same reasoning applies here. Although the result may appear "curious," it is compelled nonetheless.



Although we will affirm the district court and thus will not set aside the FDA's action, we acknowledge the attractiveness of plaintiff's position. Undeniably, the FDA's administration of the Act imposes duplicative expense on generic animal drug manufacturers. We may not, however, overlook the fact that the "me-too" manufacturer seeks to enjoy, without remunerating the pioneer manufacturer, the benefit of the pioneer's substantial investment in research and testing. The consumer may suffer somewhat higher generic drug prices, but, in the long run, will avoid the risk of being denied the pioneer's scientific advances deferred by the prospect of generic manufacturers taking advantage of the developer's labor. Balancing these various factors is a task uniquely suited for legislative resolution, and it appears that the legislation pending in Congress will take up this task.


This case illustrates the classic allocation of power between the legislative and judicial bodies. "[I]n our constitutional system the commitment to the separation of powers is too fundamental for us to pre-empt congressional action by judicially decreeing what accords with 'common sense and the public weal.' Our Constitution vests such responsibilities in the political branches." Tennessee Valley Auth. v. Hill, 437 U.S. 153, 195, 98 S.Ct. 2279, 2302, 57 L.Ed.2d 117 (1978).


Because the district court properly carried out its function of applying existing law, we will affirm its order.


The Honorable James McGirr Kelly, United States District Judge for the Eastern District of Pennsylvania, sitting by designation


The drug is intended to combat Escherichia coli, Salmonella typhimurium, and Pseudomonas aeruginosa in day-old chicks


The term "generic drug" is used throughout this opinion in its broadest possible context: to denote a product not protected by trademark and the non-proprietary name of which is ordinarily descriptive of its chemical composition. See Dorland's Illustrated Medical Dictionary 639 (25th ed. 1974). Our definition, therefore, is more expansive than the one set out by the Supreme Court. Cf. United States v. Generix Drug Corp., 460 U.S. 453, 454-55, 103 S.Ct. 1298, 1299, 75 L.Ed.2d 198 (1983). We adopt our broader definition in light of the plaintiff's assertion that Gentaject is an exact duplicate of Garasol, identical to Garasol in every respect and in all active and inactive ingredients


Bioequivalence compares the action of two drugs that, "when administered to the same individual in the same dosage regimen, result in equivalent concentrations of drug in blood and tissue." Merck Manual 2241 (R. Berkow 14th ed. 1982). See also 21 C.F.R. Sec. 320.1(e) (1987)


A citizen's petition to the FDA to issue, amend or revoke an agency regulation or take other administrative action is authorized by 21 C.F.R. Sec. 10.30 (1987). The agency's final resolution of the citizen's petition is, with certain exceptions not relevant here, subject to judicial review


Pioneer drugs are original drugs which have been approved after compliance with the FDA's rigorous testing requirements. Generic drugs, also known as "copycat" or "me-too" drugs, are copies of the pioneer drugs and can be marketed at reduced prices because their manufacturers have not incurred the developmental and testing costs of the pioneers. See United States v. Generix Drug Corp., 460 U.S. 453, 455 n. 1, 103 S.Ct. 1298, 1299 n. 1, 75 L.Ed.2d 198 (1983)


Bioavailability is the rate at which and the extent to which the drug enters the general circulation. See Merck Manual, supra at 2241; 21 C.F.R. Sec. 320.1(a) (1987)


Because the Trade Secrets Act, 18 U.S.C. Sec. 1905, prohibits only public disclosure of application data, it does not bar internal agency use of submitted data. Ruckelshaus v. Monsanto, 467 U.S. 986, 1009, 104 S.Ct. 2862, 2876, 81 L.Ed.2d 815 (1984); Chevron Chemical Co. v. Costle, 641 F.2d 104, 115 (3d Cir.), cert. denied, 452 U.S. 961, 101 S.Ct. 3110, 69 L.Ed.2d 972 (1981). For the same reason, the plaintiff's reliance on 21 U.S.C. Sec. 331(j) and 21 C.F.R. Secs. 20.61, 514.11 is unavailing


Approval by the FDA does not transform a new drug into one that is "generally recognized." To be exempt under the general recognition exclusion, the drug must have been used "to a material extent or for a material time." 21 U.S.C. Sec. 321(w)(2). The Food, Drug, and Cosmetic Act "is designed so that drugs on the market ... will have mustered the requisite scientifically reliable evidence of effectiveness long before they are in a position to drop out of active regulation by ceasing to be a 'new drug.' " Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 631, 93 S.Ct. 2469, 2484, 37 L.Ed.2d 207 (1973)