E.R. Squibb and Sons, Inc. v. Otis R. Bowen, M.D., Secretary of Health & Human Services

Opinion for the Court filed by Circuit Judge D.H. GINSBURG. D.H. GINSBURG, Circuit Judge: This is a petition for review of an order of the Commissioner of Food and Drugs withdrawing approval of, and revoking certification for, four combination drugs manufactured by petitioner E.R. Squibb and Sons, Inc. We deny Squibb’s petition for review. I. Background Section 505 of the Federal Food, Drug, and Cosmetic Act, 21 U.S.C. § 301 et seq. (1982 & Supp. IV 1986) (the FDCA) prohibits interstate distribution of any new drug that has not been approved by the Food and Drug Administration (FDA). 21 U.S.C. § 355(a). In order to obtain FDA approval for a new drug prior to 1962, a manufacturer needed to show only that the drug was “safe for use.” Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 615, 93 S.Ct. 2469, 2476, 37 L.Ed.2d 207 (1973). During the 1950s, Squibb received FDA approval for, and began marketing, four oral combination drugs (referred to collectively as Mysteclin), each of which contained both the antibiotic tetracycline and one of two antifungal agents, nystatin or amphotericin B. Squibb included the anti-fungal agents on the theory that ingestion of tetracycline could lead to an overgrowth of the organism Candida albicans, a fungus that is commonly present in the human body but that can in some circumstances cause infection. In 1962, Congress amended the FDCA, 76 Stat. 780, to require proof that, in addition to being “safe for use,” a drug is “effective in use.” 21 U.S.C. § 355(b). Under the amended statute, the Secretary of Health and Human Services (who has delegated his authority to the Commissioner of Food and Drugs, see 21 C.F.R. § 5.10(a)(1)) must withdraw approval of a drug, including a drug approved under the pre-1962 standards, if he finds “on the basis of new information before him with respect to such drug, evaluated together with the evidence available to him when the application was approved, that there is a lack of substantial evidence that the drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling thereof_” 21 U.S.C. § 355(e). The 1962 amendments place on the drug manufacturer the burden of coming forward with substantial evidence to support its claim of effectiveness, see Hynson, 412 U.S. at 617, 93 S.Ct. at 2477, and define “substantial evidence” as: evidence consisting of adequate and well-controlled investigations, by experts qualified by scientific training and experience to evaluate the effectiveness of the drug involved, on the basis of which it could fairly and responsibly be concluded by such experts that the drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling or proposed labeling thereof. 21 U.S.C. § 355(d). In accordance with the 1962 amendments, the FDA undertook a review of drugs that it had approved under the pre-1962 standard. In order to accomplish this task, it called upon the National Academy of Science — National Research Council (NRC) to perform preliminary evaluations of the effectiveness of pre-1962 drugs. See Hynson, 412 U.S. at 614-15, 93 S.Ct. at 2475-76. In 1969, the FDA announced that it would delete from the list of certifiable drugs in its regulations those drugs containing the combination of antibiotic and antifungal agents found in the various forms of Mysteclin, and that it would withdraw approval of all certifications granted pursuant to those regulations. The agency *680 based this decision on the NRC’s conclusion that “substantial evidence is lacking that each of these combination drugs will have the effect it purports or is represented to have.” Tetracycline-Amphotericin B Combination Drugs, 34 Fed.Reg. 9,336 (1969). See also Certain Tetracycline-Nystatin ... Combination Drugs for Oral Use in Humans, 34 Fed.Reg. 18, 161 (1969). Squibb filed objections to this decision, which was then stayed pending the outcome of an evidentiary hearing in which manufacturers of the affected drugs were invited to demonstrate that their products met the statutory standards for effectiveness. Certain Combination Drugs Containing Antibiotics and Antifungal Agents; Notice of Hearing, 47 Fed.Reg. 23,564 (1982). As of the time of the hearing, Squibb used two types of labeling for Mysteclin, depending upon the antifungal agent involved: for Mysteclin products containing amphotericin B, the labeling claimed effectiveness in preventing candidal disease attributable to antibiotic therapy; for products containing nystatin, the labeling claimed effectiveness only in suppressing candidal overgrowth. During the hearing, Squibb introduced several alternative forms of labeling for Mysteclin products containing amphotericin B, apparently in order to conform its labeling to what it hoped the evidence would show. The new labeling, like the labeling for the products containing nystatin, claimed effectiveness only against the possibility of candidal overgrowth in the intestinal tract. Although Squibb introduced evidence in support of both the disease and the suppression claims, it limits its current petition for review to the FDA’s rejection of the suppression claim, and we tailor our discussion accordingly. Following the hearing, the Administrative Law Judge held that Squibb had failed to show that Mysteclin is “effective in use” and ordered its certification withdrawn. Squibb appealed that decision to the Commissioner, who affirmed the AU in all relevant respects. Specifically, the Commissioner held that (1) in order to show that a drug is “effective in use,” the manufacturer must demonstrate not only that it has the effect-claimed on its label, but also that the claimed effect is of some medical significance; (2) Squibb had failed to establish “that suppression of gastrointestinal candida is itself a medically significant effect”; (3) the record evidence in fact showed that “candida overgrowth is not itself a disease” and that its suppression “is a pharmacologic, not necessarily a therapeutic, effect, i.e., it is a physiologic effect of no proven benefit to the patient”; and (4) in any event, the studies submitted by Squibb failed to establish that Mysteclin is effective in suppressing candidal overgrowth. The FDA then announced its final decision to withdraw approval for Myste-clin. Certain Fixed-Combination Antibiotic/Antifungal Products; Withdrawal of Approval of New Drug Applications, 53 Fed.Reg. 5,044 (1988). Squibb challenges each of the Commissioner’s conclusions. II. The Proper Interpretation op the Statute Squibb’s primary contention is that the Commissioner erroneously required it to show that the suppression effect claimed for Mysteclin on its labeling is itself medically significant. As Squibb reads the 1962 amendments, a manufacturer, in order to satisfy the “effective in use” standard, need show only that a drug does what its label says it does, not that the claimed effect produces any particular medical benefit. Although no provision of the FDCA defines the term “effective in use,” Squibb argues that both the plain language and the legislative history of the amendments evince a clear congressional intent compelling its interpretation, and that we must therefore reject the Commissioner’s contrary reading. In Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), the Supreme Court set forth a two-step analysis for judicial review of an agency’s interpretation of the statute it is charged with implementing. First, the Court held that if “Congress has directly spoken to the precise question at issue,” such that “the *681 intent of Congress is clear,” the court must not defer to a contrary agency interpretation, since “the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” 467 U.S. at 842-43, 104 S.Ct. at 2781. Second, the Court held that if the statute is “silent or ambiguous with respect to the specific issue,” the reviewing court must ask only “whether the agency’s answer is based on a permissible construction of the statute”; if it answers that question in the affirmative, it must defer to the agency’s construction. Id. at 843, 104 S.Ct. at 2782. We turn now to the first step of the Chevron analysis. A. Squibb’s “Plain Language” Argument In support of its “plain language” argument, Squibb points to (1) the statutory provision for withdrawal of FDA approval if “there is a lack of substantial evidence that [a] drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling thereof,” 21 U.S.C. § 355(e); and (2) the definition of “substantial evidence” as “evidence ... that the drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in the labeling ... thereof,” 21 U.S.C. § 355(d). Squibb reads these provisions as clearly indicating that as long as the “effect” claimed for a drug in its labeling is not harmful (since, if it were, it would fail the Act’s requirement that a drug be proven “safe for use”) and as long as that claim is accurate, the Commissioner may not withdraw approval of the drug. In other words, Squibb interprets the Act as erecting, under the head of effectiveness, only a truth-in-labeling requirement. The statutory language does not compel Squibb’s reading, however. The cited provisions are directed primarily at the eviden-tiary standard that a manufacturer must meet in order to demonstrate that a drug is “effective in use”; they do not address the question whether any claimed “effect,” however innocuous, is sufficient. B. Squibb’s Arguments from the Legislative History of the 1962 Amendments Squibb argues, however, that the legislative history of the 1962 amendments reveals Congress’s clear intent that the statute be applied as the company suggests. Squibb points to a statement by Senator Kefauver, the principal sponsor of the amendments, to the effect that the standard for approval set forth in the then current version of the bill (which, as discussed below, used the term “efficacious” rather than “effective”) was intended to require only that the manufacturer “satisfy the [FDA] that [its drug is] efficacious for the use intended and claimed by the manufacturer, not ... [that] it is better than some other drug or poorer than some other drug, and it is not supposed to be limited to efficaciousness as to a particular disease.” Hearings before the Subcommittee on Antitrust and Monopoly of the Committee on the Judiciary, United States Senate, 87th Cong., Pursuant to S.Res. 52 on S. 1552 (1961) at 172. Senator Kefauver was not, however, “directly sp[eaking] to the precise question at issue” here. Chevron, 467 U.S. at 842, 104 S.Ct. at 2782. Read in context, it is apparent that his statement was directed at industry concerns that the FDA might be called upon to pass upon the relative efficacy of proposed drugs, i.e., to determine whether one drug is more likely than another to produce a desired effect or to cure a particular disease. Nowhere in the legislative history does Senator Kefauver, or anyone else, suggest that the FDA would be required to approve a drug whose claimed effect is of no medical significance. Squibb also notes that the “effective in use” standard that Congress enacted differs from that proposed in previous versions of the bill: the original version used the term “efficacious in use”; that term was changed to “intended or purported physiological effect,” which in turn was replaced with the current language. Squibb argues that by rejecting “efficacious” in favor of “effective,” Congress *682 meant to eliminate any requirement of efficacy in the sense of medical benefit. We cannot divine so much from these entrails of the legislative process: the various changes in the statute, which are simply not explained in the legislative history, are of uncertain import, and the scattered statements in the hearings and reports on the various versions of the statute, as it made its way through the legislative process, are inconclusive. It thus appears that Congress, in enacting the 1962 amendments, did not address “the precise question at issue” in this case — since nobody ever raised it — whether a drug must be approved even if its only proven effect is of no medical significance. In sum, we find nothing either in the plain language of the 1962 amendments or in their legislative history that compels the reading advanced by Squibb. C. The Statutory Definitions of “Drug” Although the 1962 amendments are silent as to the precise meaning of “effective in use,” the FDCA read as a whole would appear, if anything, to preclude the reading advanced by Squibb. The pre-marketing approval provisions here at issue apply, by their terms, only to “drugs,” which the FDCA defines, insofar as relevant here, as “articles intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man ...,” or “to affect the structure or any function of the body of man_” 21 U.S.C. §§ 321(g)(1)(B), (C). The former provision plainly applies only to drugs whose intended effect is in some way related to disease. It would appear to foreclose Squibb’s argument that Mysteclin is “effective in use” so long as it has only a safe use that is not disease-related. The latter provision, in contrast, appears to contemplate that a substance will be approved as a “drug” even though it has only a physiologic, rather than a therapeutic, effect. At no point in these proceedings has Squibb disputed that Mysteclin is a “drug” subject to the pre-marketing approval process of § 505. On this appeal, Squibb contends that because it has removed all therapeutic claims from its labeling for Myste-clin, the drug now falls solely within the “structure or ... function” definition, and thus must be approved if it is safe (which is not disputed) and it has the physiologic effect claimed for it. “[T]he required proof of effectiveness must be for those physiologic claims, not for some unstated (or in some cases absent) therapeutic claim,” it argues. In support of this view, Squibb notes that the FDA has approved several articles, most notably birth control pills and weight reduction remedies, as to which only physiologic claims are made and which have not been proven medically beneficial. Whatever the merits of Squibb’s argument with respect to drugs that are properly characterized as “structure or ... function” drugs, we do not think that Mysteclin is such a drug. First, it is questionable whether a drug that acts only upon non-human organisms that happen to reside within the human body can properly be understood as affecting the “body of man” (as opposed to the “prevention of disease in man”) within the meaning of the definition. Second, assuming that such organisms could be understood as part of the human body, a drug that suppresses their growth does not affect the “structure” or “function” of the human body as the courts have construed those terms. Congress added the “structure or ... function” definition to the FDCA in 1938 in order to bring within the regulatory framework certain drugs having only physiologic effects; it thereby gave the FDA authority over such articles as “anti-fat remedies,” which “had escaped regulation under the prior definition.” Nutrilab, Inc. v. Schweiker, 713 F.2d 335, 336 (7th Cir.1983). See also Action on Smoking and Health v. Harris, 655 F.2d 236, 238 (D.C.Cir.1980) (“structure or ... function” definition was intended “to enable the FDA to assert jurisdiction over slenderizing remedies ... ”). As we have previously noted, the “structure or ... function” definition, unlike the “disease in man” definition, is relatively narrow, and was not intended to encompass all articles that might have some remote physical effect upon the body. Action on Smoking and Health, 655 F.2d at 240 (not *683 ing that while any article which “comes into contact with the senses may be said to be an article ‘intended to affect the functions of the body of man,’ ” Congress did not mean for the “structure or ... function” definition to be read so broadly) (quoting Federal Trade Commission v. Liggett & Myers Tobacco Co., 108 F.Supp. 573 (S.D.N.Y.1952), aff'd, 203 F.2d 955 (2d Cir.1953)). Accordingly, courts construing the “structure or ... function” definition of “drug,” and the similar definition of “device,” see 21 U.S.C. § 321(h)(2), have confined those definitions to articles that purport literally to change the physical structure of the body, see e.g., Orthopedic Equipment Co. v. Eutsler, 276 F.2d 455, 459 (4th Cir.1960) (surgical nail inserted into broken bone is a “device” within “structure or ... function” definition); United States v. Article ... Consisting of 216 Cartoned Bottles, More or Less, Sudden Change, 409 F.2d 734, 741 (2d Cir.1969) (lotion claiming to effect a “face lift” purports to affect the structure of the human body and is thus a drug); United States v. Article Consisting of 36 Boxes, More or Less, Labeled “Line Away, Temporary Wrinkle Smoother, Coty,” 284 F.Supp. 107 (D.Del.1968), aff'd, 415 F.2d 369 (3d Cir.1969) (product intended to tighten the skin affects the structure of the skin and “falls within the literal definition of a drug” under the “structure or ... function” definition), or to alter its basic functions, such as sleep, see United States v. 23, More or Less, Articles, 192 F.2d 308, 309 (2d Cir.1951) (a phonograph record purporting to induce sleep is a “device,” since sleep is a function of the body), or digestion, see Nutrilab, 713 F.2d at 339 (a substance that inhibits digestion of starch is a “drug,” since digestion is a function of the body). See also United States v. 354 Bulk Cartons ... Trim Reducing-Aid Cigarettes, 178 F.Supp. 847, 851 (D.N.J.1959) (cigarettes purporting to reduce appetite affect both the structure (body weight) and a function (appetite) of the human body); United States v. Article Consisting of 46 Devices, “Dynatone”, 315 F.Supp. 588, 589 (D.Minn.1970) (“facial exerciser” purporting to stimulate muscle growth on the face and thereby reverse the process of wrinkling is intended to affect both the structure and a function of the human body). Cf. United States v. Article of Drug, 414 F.Supp. 660, 666 (D.N.J.1975) (in vitro home pregnancy test does not affect the structure or any function of the human body since it in no way comes into contact with the body); Action on Smoking and Health, 655 F.2d at 240 (although nicotine may stimulate the senses, it does not affect either the structure or any function of the body). Given Congress’s intent, in adding the “structure or ... function” definition, specifically to bring “anti-fat” remedies within the regulatory framework, there is no doubt that articles intended to reduce weight, thus affecting the structure of the human body, fall within that definition. Similarly, articles intended to prevent pregnancy, thus affecting the reproductive function of the human body, fall within that definition. The suppression effect Squibb claims for Mysteclin, in contrast, would simply reduce the number of non-human organisms residing within the intestinal tract. Candida organisms are hardly part of the physical “structure” of the human body, nor does their suppression affect any “function” of the body in the sense that articles that induce sleep or inhibit digestion do. Mysteclin’s undisputed status as a drug, therefore, appears to depend upon its being an article “intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man_” Cf. Immunology and Microbiology Devices; General Provisions and Classification of 162 Devices, 47 Fed.Reg. 50,814, 50,815 (1982) (devices “that have medical uses only are clearly intended for medical purposes and, therefore, will be regulated as medical devices whether or not medical claims are made for them.”). As such, the statutory mandate that it be shown “effective in use” would appear plainly to require a showing that, consistent with its identity as a “disease in man” drug, it has some disease-related effect. It is inconceivable that Congress meant for a drug to be approved *684 without some showing of effectiveness in the intended use (i.e., relating to disease) that made it subject to regulation in the first place. Nonetheless, we are reluctant for two reasons to rule, on the basis of the distinction between disease-related drugs and “structure or ... function” drugs, that the statute unambiguously precludes the reading advanced by Squibb. First, and most important, the FDA does not argue that anything turns on the definitional distinction discussed above. Rather, in response to Squibb’s argument that Mysteclin is a drug only by virtue of the “structure or ... function” definition of drug (and as such need not be shown effective in any medically significant way), which it raised for the first time in its reply brief, the agency, at oral argument, offered only the broad rejoinder that some showing of therapeutic significance is required for all drugs — even those falling within the "structure or ... function” definition. Absent an explicit reference to “disease in man” in the latter definition, however, it is not clear to us that the statutory language compels the agency’s broad reading. Moreover, in light of our conclusion that Mysteclin falls within the “disease in man” definition of drug, the agency’s broad argument need not be decided in this case. We therefore think it prudent to avoid either rejecting that argument as inconsistent with the statutory distinction between the two types of drugs, or holding, on the basis of a distinction not advanced by the agency, that the agency’s reading of the statute is compelled as to the particular category of drugs now before us. Second, were we to hold that the statute precludes Squibb’s interpretation of the statute as to all drugs, we would thereby foreclose the agency in future cases from itself assessing whether the two definitions of “drug” have different implications for the effectiveness inquiry. Although we think that the reference to “disease in man” in the definition of “drug” under which Mysteclin falls makes Squibb’s “plain language” argument inapposite on the facts of this case, the degree to which the “effective in use” standard for approval of a drug incorporates the intended uses that make the article a drug in the first place is not entirely free from doubt. We therefore think it wiser to allow the agency initially to address that question should it arise in a future proceeding. We therefore resolve this aspect of the case on the narrowest possible ground, by holding only that neither the FDCA nor its legislative history shows that Congress resolved the precise question presented in favor of Squibb’s position that a drug may be deemed “effective in use” even if its intended use is of no medical significance. We need not, and do not, decide here whether the statute compels the agency’s contrary reading. Rather, assuming ar-guendo that the statute is ambiguous as to the correctness of that reading, we turn directly to the question whether the agency’s interpretation, as applied to this case, is permissible under the second step of Chevron. D. The Reasonableness of the Agency’s Interpretation As noted above, the FDA’s reading of the 1962 amendments to require proof of some therapeutic effect finds strong support in the FDCA as a whole, at least for drugs falling within the “disease in man” definition. In this case, the Commissioner found that Squibb had failed to show that the suppression effect claimed for Mystesclin is of any therapeutic significance at all. Assuming for the moment the correctness of that finding for the purpose of evaluating the Commissioner’s legal determination, we hold that the agency’s reading of the statute as requiring it to withdraw approval for Mysteclin is a permissible interpretation of the statute. Squibb also points to the FDA’s approval of neomycin, a drug intended to reduce the number of ammonia-forming bacteria in the intestinal tract, in an effort to show that the agency’s position on the necessity of a therapeutic effect is inconsistent with its prior practice, and that as such an explanation is required. See Oil, Chemical and Atomic Workers Int’l Union v. National Labor Relations Board, 806 F.2d 269, 273- *685 74 (D.C.Cir.1986). Squibb notes that in its initial evaluation of neomycin, the NRC concluded that there was no evidence that it was effective in reducing infection, and argues that the PDA’s approval of the drug notwithstanding that finding constituted acceptance of mere suppression as an appropriate indication for a drug. In approving neomycin, however, the FDA specifically found that it was “effective as an adjunct for treatment of hepatic coma; and for treatment of diarrhea due to enteropa-thogenic K Coli” Neomycin Sulphate Oral Preparations, 35 Fed.Reg. 10,241 (1970). See also Neomycin Sulphate Oral Preparations, 36 Fed.Reg. 18,113, 18,114 (1971). Although Squibb challenges the FDA’s decision as inconsistent with that reached by the NRC, we cannot gainsay the agency’s final determination that neo-mycin had been proven effective for therapeutic uses, a conclusion it was unable to reach with respect to Mysteclin. III. Squibb’s Evidence of Disease-Related Effect Squibb next challenges the Commissioner’s conclusion that “[t]he record contains no evidence to support Squibb’s argument that suppression of gastrointestinal candida is itself a medically significant effect.” Squibb asserts that it met its evidentiary burden on this point with the testimony of experts who stated that, in their view, suppression of candidal overgrowth is of some medical significance. One of the FDA’s own witnesses, for example, testified that candidal overgrowth in the intestine is to be avoided because “the Candida can be an important cause of disease,” including “Candida infections or disease, thrush, of the mouth or the esophagus or of the vagina or of the bowel, or of the skin.” Squibb’s witnesses testified to like effect, one also noting that in his opinion, “a drug with this [suppression] effect is of value in the practice of medicine.” According to Squibb, the Commissioner must defer to the opinions of these experts as long as their opinions are based upon substantial evidence. Squibb does not challenge, however, the Commissioner’s conclusion that its studies failed to show that suppression of candidal overgrowth is of any disease-related significance. Squibb asserts that (1) both its studies and the testimony of its experts showed that Mysteclin does produce the claimed suppression effect; and (2) the Commissioner is required to defer to the experts on the medical significance of that effect. Because we reject Squibb’s second argument, we need not address its first. The 1962 amendments, as noted above, define “substantial evidence” as “adequate and well-controlled investigations, including clinical investigations, by experts qualified by scientific training and experience to evaluate the effectiveness of the drug involved....” 21 U.S.C. § 355(d). FDA regulations state that the agency will not consider “[¡Isolated case reports, random experience, and reports lacking the details which permit scientific evaluation.” 21 C.F.R. § 314.126(e). In Hynson, the Supreme Court observed that the FDA’s exclusion of “anecdotal evidence indicating that doctors ‘believe’ in the efficacy of a drug[] [is] amply justified by the legislative history” of the 1962 amendments, which indicates “a marked concern that impressions or beliefs of physicians, no matter how fervently held, are treacherous.” 412 U.S. at 619, 93 S.Ct. at 2477. In light of our holding that the Commissioner permissibly required Squibb to show some medical significance for its suppression claim, it follows that such a showing must meet the substantial evidence standard set forth in the 1962 amendments. Thus, Squibb was required to come forward with substantial evidence demonstrating not just that Mysteclin produces the claimed suppression effect, but also that that effect is of some medical value. Since Squibb has adduced only anecdotal evidence to support its claim of medical efficacy — evidence that cannot constitute “substantial evidence” under either the statutory standard or the FDA’s regulations— we find that the Commissioner’s withdrawal of approval for Mysteclin is supported by the record. *686 IV. Conclusion We conclude that nothing in the language or the legislative history of the 1962 amendments compels the FDA to approve a drug, such as Mysteclin, that is defined as such only by reference to its intended effect on a “disease in man,” 21 U.S.C. § 321(g)(1)(B), but that claims on its labeling an effect that is of no proven medical significance. We further hold that the Commissioner’s reading of the FDCA to require some showing of medical efficacy is a permissible construction of the statute, at least as applied to an article falling within § 321(g)(1)(B). Finally, we hold that the record supports the Commissioner’s conclusion that Squibb failed to make the required showing. For the foregoing reasons, Squibb’s petition for review of the Commissioner’s determination to withdraw approval of, and to revoke certification for, its Mysteclin drugs is DENIED.