United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued December 3, 1997 Decided October 27, 1998
No. 97-5188
Serono Laboratories, Inc.,
Appellee
v.
Donna E. Shalala, et al.,
and
Ferring Pharmaceuticals Inc.,
Appellants
Consolidated with
No. 97-5227
Appeals from the United States District Court
for the District of Columbia
(No. 97cv01227)
Christine N. Kohl, Attorney, U.S. Department of Justice,
argued the cause for the federal appellants, with whom Frank
W. Hunger, Assistant Attorney General, Mary Lou Leary,
U.S. Attorney, and Douglas N. Letter, Litigation Counsel,
U.S. Department of Justice, were on the briefs.
John R. Fleder, David F. Weeda and Arthur Y. Tsien were
on the briefs for appellant Ferring Pharmaceuticals Inc.
Bruce S. Manheim, Jr. argued the cause for appellee, with
whom Terry S. Coleman and Matthew D. Peterson were on
the brief. Michael D. Petty entered an appearance.
Before: Henderson, Rogers and Garland, Circuit Judges.
Opinion for the Court filed by Circuit Judge Garland.
Garland, Circuit Judge: In this case we consider the
validity of a district court order, preliminarily enjoining ap-
proval by the Food and Drug Administration ("FDA") of a
generic drug, that was issued at the behest of the manufac-
turer of the competing brand-name drug. We previously
stayed the preliminary injunction pending our resolution of
this appeal. Because we find plaintiff has not satisfied the
standards for a preliminary injunction, and in particular has
not shown a likelihood of success on the merits, we now
vacate the injunction.
I
The Food, Drug, and Cosmetic Act (the "Act") provides
that "[n]o person shall introduce or deliver for introduction
into interstate commerce any new drug" without first obtain-
ing FDA approval. 21 U.S.C. s 355(a). To obtain FDA
approval, the first applicant to market a drug, known as the
"pioneer," must submit a new drug application ("NDA")
containing, among other things, "full reports of investiga-
tions" made "to show whether or not such drug is safe for use
and whether such drug is effective in use." Id. s 355(b)(1).
Recognizing that the NDA process is costly and time-
consuming, and seeking "to make available more low cost
generic drugs," Congress amended the Act in 1984. H.R.
Rep. No. 98-857, pt. 1, at 14 (1984), reprinted in 1984
U.S.C.C.A.N. 2647, 2647. The Drug Price Competition and
Patent Term Restoration Act of 1984, Pub. L. No. 98-417, 98
Stat. 1585 (known as the "Hatch-Waxman Amendments"),
permits a manufacturer of a generic alternative to a pioneer
drug to seek FDA approval by submitting an abbreviated new
drug application ("ANDA") that need contain only the more
limited information specified in 21 U.S.C. s 355(j)(2).1
Two aspects of the ANDA process, corresponding to two
kinds of drug ingredients, are relevant to this case. First,
with respect to "active ingredients," the statute provides that
the Secretary of Health and Human Services shall approve an
application for a generic drug unless the Secretary finds,
among other things, that "information submitted with the
application is insufficient to show that the active ingredients
are the same as the active ingredients of the listed [pioneer]
drug...." 21 U.S.C. s 355(j)(3)(C)(ii). The FDA defines an
"active ingredient" as "any component that is intended to
furnish pharmacological activity or other direct effect in the
diagnosis, cure, mitigation, treatment, or prevention of dis-
ease, or to affect the structure or any function of the
body...." 21 C.F.R. s 210.3(b)(7).
Second, with respect to "inactive ingredients," the statute
provides that the Secretary shall approve an application
unless she finds that "information submitted in the applica-
tion or any other information available to the Secretary
shows" that "the inactive ingredients of the drug are unsafe"
or "the composition of the drug is unsafe ... because of the
type or quantity of inactive ingredients included or the man-
ner in which the inactive ingredients are included." 21 U.S.C.
s 355(j)(3)(H). The FDA defines an "inactive ingredient" as
__________
1 Prior to the Hatch-Waxman Amendments, the FDA had estab-
lished its own abbreviated procedures for generic copies of pioneer
drugs approved before 1962, but not of pioneer drugs approved
after 1962. The Amendments generally extended those procedures
to cover generic copies of post-1962 pioneer drugs. See H.R. Rep.
No. 98-857, pt. 1, at 14 (1984), reprinted in 1984 U.S.C.C.A.N. 2647,
2647; 130 Cong. Rec. 23,057 (1984) (statement of Rep. Waxman).
"any component other than an active ingredient." 21 C.F.R.
s 210.3(b)(8).
In 1969, the FDA approved an NDA submitted by plaintiff
Serono Laboratories, Inc. ("Serono") for Pergonal, a pioneer
drug. Pergonal is a "menotropins" product administered by
intramuscular injection and used to treat male and female
infertility. A menotropins product is extracted from the
urine of post-menopausal women, and contains two active
ingredients: follicle-stimulating hormone ("FSH") and lutein-
izing hormone ("LH"). See Joint Appendix ("J.A.") 473;
Dorland's Illustrated Medical Dictionary 1013 (28th ed.
1994). FSH and LH make up less than five percent of
Pergonal, with lactose and uncharacterized urinary proteins
("UUPs") constituting the remainder. See FDA Br. at 6;
Serono Br. at 4.
In 1990, Lederle Parenterals, Inc. ("Lederle") submitted an
ANDA to the FDA seeking approval of a generic version of
Pergonal, now known as Repronex. Defendant-intervenor
Ferring Pharmaceuticals Inc. ("Ferring") acquired the rights
to Lederle's ANDA while it was pending. In December 1992,
Serono filed a "citizen petition," pursuant to 21 C.F.R.
s 10.30, urging the FDA to withhold approval of the ANDA.
Serono argued, among other things, that the UUPs in the
proposed generic drug were "inactive ingredients" that dif-
fered from those in Pergonal and had not adequately been
demonstrated to be safe. J.A. 465-69.
In a subsequent meeting, and in a supplemental filing on
March 21, 1997, Serono also argued that the active ingredient
FSH in the proposed generic drug was not, as required by
statute, "the same as" the FSH in Pergonal because of
differences in "isoforms" of the two products. Id. at 481.
FSH is a protein-based hormone consisting of two protein
chains in a backbone-like configuration, with carbohydrate
side chains. Natural variation in the carbohydrate elements
leads to different isoforms of the hormone. See id. at 482-83.
Serono argued that this isoform variation in FSH rendered
Repronex different from Pergonal, and hence ineligible for an
ANDA. Id. at 481-82.
On January 30, 1997, the FDA approved the ANDA for
Repronex. Id. at 459-60. The FDA gave Repronex an "AB"
rating in its publication, Approved Drug Products with Ther-
apeutic Equivalence Evaluations (known as the "Orange
Book"), meaning that physicians and pharmacists could sub-
stitute Repronex for Pergonal. See FDA Br. at 7-8. In a
memorandum to the administrative record filed on that date,
Gordon Johnston, the Deputy Director of the FDA's Office of
Generic Drugs, addressed another issue that had surfaced
during the review--a difference in the concentration of the
inactive ingredient lactose in Repronex and Pergonal. John-
ston noted that although a 1992 regulation required an inac-
tive ingredient in a generic drug to be in the same concentra-
tion as in the pioneer drug, that regulation was not in effect
when the ANDA for Repronex was filed in 1990. Johnston
concluded that since FDA policy was to review an application
under the regulations in effect at the time of filing, the
different lactose concentrations did not preclude ANDA ap-
proval. He also found that they posed no safety concerns.
Id. at 457-58 (Memorandum to Record by G. Johnston, Jan.
30, 1997) (hereinafter "Johnston Memorandum").
On May 30, 1997, Serono sued the FDA in district court,
raising many of the same issues contained in its still-pending
citizen petition as well as the additional issue of the differing
lactose concentrations. See Complaint WW 21-23. On June 13,
1997, Serono moved for a preliminary injunction rescinding
the FDA's approval of Repronex, and Ferring intervened as a
defendant.
On June 17, 1997, the FDA issued its final decision denying
Serono's citizen petition. J.A. 472-86 (Letter from J. Wood-
cock to Serono, June 17, 1997) (hereinafter "Woodcock Let-
ter"). Dr. Janet Woodcock, Director of the FDA's Center for
Drug Evaluation and Research, rejected Serono's claim that
the isoform variation in the active ingredient FSH meant that
the FSH in Repronex was not the "same as" the FSH in
Pergonal. Dr. Woodcock acknowledged the isoform variation,
but concluded that it was not "clinically significant for the
product's intended uses" and therefore did not preclude a
"sameness" finding for purposes of 21 U.S.C. s 355(j). Id. at
484. Dr. Woodcock further concluded that the differing
lactose concentrations in the two products, as well as the
differing UUP profiles, did not affect the safety of Repronex.
Id. at 480-81 & n.12. She also rejected the characterization
of the UUPs as "inactive ingredients," classifying them in-
stead as "impurities." Id. at 479-80.
On July 28, 1997, the district court granted Serono's motion
for a preliminary injunction, barred the FDA "from approving
the Ferring ANDA," and ordered it to "rescind immediately
its designation of an 'AB' rating [for Repronex] in the Orange
Book." Serono Lab. v. Shalala, 974 F. Supp. 29, 37 (D.D.C.
1997). The district court found that Serono was likely to
prevail on the merits of its claims; that Serono would suffer
irreparable injury if interim relief were not granted; and that
both the balance of harms to Serono and Ferring, and the
public interest, favored granting injunctive relief. See id. at
32-37.
II
A court considering a plaintiff's request for a preliminary
injunction must examine whether: (1) there is a substantial
likelihood plaintiff will succeed on the merits; (2) plaintiff will
be irreparably injured if an injunction is not granted; (3) an
injunction will substantially injure the other party; and (4)
the public interest will be furthered by the injunction. See
Washington Metro. Area Transit Comm'n v. Holiday Tours,
Inc., 559 F.2d 841, 843 (D.C. Cir. 1977). These factors
interrelate on a sliding scale and must be balanced against
each other. "If the arguments for one factor are particularly
strong, an injunction may issue even if the arguments in
other areas are rather weak." CityFed Fin. Corp. v. Office of
Thrift Supervision, 58 F.3d 738, 746 (D.C. Cir. 1995); see
Holiday Tours, 559 F.2d at 843-45.
We review the district court's weighing of the preliminary
injunction factors under the "abuse of discretion" standard,
see Transohio Sav. Bank v. Director, Office of Thrift Supervi-
sion, 967 F.2d 598, 614 (D.C. Cir. 1992), and its findings of
fact under the "clearly erroneous" standard, see National
Wildlife Fed'n v. Burford, 835 F.2d 305, 319 (D.C. Cir. 1987).
"[T]o the extent the district court's decision hinges on ques-
tions of law," however, "our review is 'essentially de novo.' "
O'Hara v. District No.1-PCD, 56 F.3d 1514, 1522 (D.C. Cir.
1995) (quoting Transohio, 967 F.2d at 614).
III
Serono's argument that it is likely to succeed on the merits
depends upon the validity of its contentions regarding three
of Repronex's ingredients: FSH, lactose, and UUPs. We
consider each of these in turn.
A
As noted above, FSH is an active ingredient in both
Repronex and Pergonal. The Hatch-Waxman Amendments
provide that the FDA "shall approve" an ANDA for a generic
drug unless it finds, among other things, that the information
submitted "is insufficient to show that the active ingredients
are the same as the active ingredients of the listed drug." 21
U.S.C. s 355(j)(3)(C)(ii) (emphasis added). FDA regulations
also state that "[f]or determining the suitability of an abbrevi-
ated new drug application, the term 'same as' means identi-
cal in active ingredient(s)...." 21 C.F.R. s 314.92(a)(1) (em-
phasis added).
As we also have noted, the chemical structure of FSH
roughly consists of two components: (1) a protein backbone
with a specific amino acid sequence, and (2) carbohydrate side
chains. See J.A. 482-83 (Woodcock Letter). In concluding
that the FSH in Repronex is the "same as" or "identical" to
that in Pergonal, the FDA determined that their protein
backbones and amino acid sequences are identical. Id. at 483.
There are, however, slight natural variations in the configura-
tion of the carbohydrate side chains, a phenomenon known as
"microheterogeneity." See id. at 482-83. But, the FDA
observed, "complete chemical identification of all the carbohy-
drate variants in a protein product often is not possible or
feasible," id. at 482, a point Serono does not dispute. Indeed,
it usually is not even possible "to assure by chemical analysis
that different batches" of the same product "are identical at
the level of the carbohydrate side chains"--including different
batches of Pergonal itself. Id. at 482-83.
In light of the fact that "most glycoprotein products will
have microheterogeneity," the FDA determined that the rele-
vant "question is how much variation should be permitted."
Id. at 482. The agency answered that question as follows:
To be considered to have the same active ingredients as
the reference listed drug, generic FSH products based
on Pergonal as the reference listed drug must have the
same primary structure, i.e., the same protein backbone
and amino acid sequence as Pergonal (assured by using
the same natural source material), the same potency, and
the same degree of batch-to-batch uniformity. The
batch-to-batch uniformity of Pergonal is achieved using
in vivo rat potency tests, specified by the U.S. Pharma-
copeia [USP].... The bioactivity of each batch of ge-
neric Menotropins ... is also controlled using USP rat
bioassays, which provides the same assurance of potency
and batch-to-batch uniformity as is provided by Serono
for Pergonal.
Id. at 483. After reviewing additional clinical data, the FDA
found "that any potential variations in FSH isoforms between
the Ferring menotropins product and Pergonal appear not to
be clinically significant for the product's intended uses." Id.
at 484. The FDA concluded that such clinical identity ren-
ders menotropins products "the same for the purposes of 21
U.S.C. s 355(j)," id., as long as the protein backbone, amino
acid sequence, and potency are the same, and the degree of
batch-to-batch variation in isoforms is no different than that
in Pergonal itself, id. at 483.
Serono argues, and the district court agreed, that "same
as" under the statute, and "identical" under the regulation,
must mean absolute "chemical" identity. The court rejected
the FDA's view that "clinical" identity is sufficient for a
menotropins product as long as the above-described condi-
tions are met, and therefore concluded that Serono was likely
to prevail on the merits of its claim that the FSH in Repronex
and Pergonal is not the same. See Serono Lab., 974 F. Supp.
at 32-34. Since the district court's conclusion rests on issues
of statutory and regulatory interpretation, we review that
conclusion de novo.
Chevron U.S.A. Inc. v. Natural Resources Defense Council,
Inc., 467 U.S. 837 (1984), governs our analysis of the validity
of an agency's interpretation of a statute. Under Chevron,
we first ask "whether Congress has directly spoken to the
precise question at issue," in which case we "give effect to the
unambiguously expressed intent of Congress." Id. at 842-43.
But if Congress has been silent or ambiguous about the
meaning of the specific question at issue, we defer to the
agency's interpretation so long as it is "based on a permissi-
ble construction of the statute." Id. at 843.
In evaluating the first Chevron inquiry, we use "traditional
tools of statutory construction" to determine whether Con-
gress has unambiguously expressed its intent. Id. at 843 n.9.
In this case, the statute does not define the term "same as,"
and does not indicate whether chemical or clinical identity
was contemplated. We need to consider, therefore, what the
terms mean in context. See McCarthy v. Bronson, 500 U.S.
136, 139 (1991). What the statute requires to be the "same"
are the two drugs' "active ingredients," and FDA regulations
pre-dating the Hatch-Waxman Amendments define an "active
ingredient" as "any component that is intended to furnish a
pharmacological activity or other direct effect." 21 C.F.R.
s 210.3(b)(7) (1978). Hence, the ingredients that are to be
compared for "sameness" are themselves defined in terms of
pharmacological activity, adding credibility to the FDA's view
that chemical identity is not the only way to read the statuto-
ry language.
The district court rejected this reading, in part because in
its view, "[n]othing in [s 355(j)] permits an ANDA applicant
to substitute USP animal assays for information demonstrat-
ing that the active ingredients of the proposed generic prod-
uct are identical to those in the innovator product." Serono
Lab., 974 F. Supp. at 33. As noted above, the FDA permitted
Ferring to use such assays to ensure the potency and batch-
to-batch uniformity of Pergonal. But while the court was
correct in noting that nothing in the statute permits the use
of animal assays, the important point is that nothing in the
statute prohibits their use. Indeed, the statute says nothing
at all about the type of information an applicant must submit
to demonstrate "sameness," nor about the type of information
upon which the FDA may rely. It says only that the informa-
tion must not be "insufficient" to show that the active ingredi-
ents are the same. 21 U.S.C. s 355(j)(3)(C)(ii). If anything,
this broad grant of discretion to the agency with respect to
the information it may consider in making a finding of
"sameness" indicates that Congress did not have one precise
definition of the term in mind. Cf. Schering Corp. v. FDA, 51
F.3d 390, 399-400 (3d Cir. 1995) (holding that FDA's interpre-
tation of 21 U.S.C. s 355(j)(7)(B) "as not limiting its discre-
tion to determine what tests or studies would provide it with
appropriate information from which to determine bioequiva-
lence is a reasonable construction of the Act").
Moreover, the statutory phrase must be read in the context
of the kind of drug at issue. As Dr. Woodcock noted, "it is
usually not possible to assure by chemical analysis that
different batches of [a protein product like FSH] are identical
at the level of the carbohydrate side chains." J.A. 482. For
the same reason, "batch to batch variability in isoform pat-
terns" exist for Pergonal itself. Id. at 483. This means that
if absolute chemical identity were required, it would not be
possible to say any generic was the "same as" Pergonal,
because the "batch to batch variability" would make the
target of the comparison (not just Pergonal, but the specific
batch of Pergonal) indeterminate. Indeed, the Woodcock
Letter indicates that if absolute chemical identity were re-
quired, not only menotropins but other categories of protein
products would be excluded from the ANDA process as well.
See id. at 482; see also id. at 317 (internal FDA memorandum
noting that other products derived from natural sources
besides proteins, including lipids, phospholipids and oligosac-
charides, also "can not be fully characterized chemically").
Yet, it seems likely--although by no means certain--that if
Congress had intended to exclude entire categories of drugs
from the scope of the Hatch-Waxman Amendments, which
were passed to "facilitat[e] the approval of generic copies of
drugs," Mead Johnson Pharm. Group v. Bowen, 838 F.2d
1332, 1333 (D.C. Cir. 1988), there would be some mention of
that fact in the statute or legislative history. Instead, both
are wholly silent on the subject. We thus conclude that the
statute does not unambiguously require the term "same as"
to be defined as complete chemical identity.
Turning to the second Chevron inquiry, we ask whether the
agency's definition is "based on a permissible construction of
the statute," Chevron, 467 U.S. at 843, which requires only
that its construction be a "reasonable" one, id. at 844. Simi-
larly, we defer to an agency's reading of its own regulations,
here the regulation defining "same as" as "identical to,"
unless it is "plainly erroneous or inconsistent with the regula-
tion." Auer v. Robbins, 117 S. Ct. 905, 911 (1997) (internal
quotation omitted); Cassell v. FCC, No. 97-1005, slip op. at
10, 1998 WL 598099 at *6 (D.C. Cir. Sept. 11, 1998). We
conclude that the FDA's definition of "same as" and "identi-
cal," as applied to menotropins products, is reasonable.
The FDA concluded that "[t]o be considered to have the
same active ingredients as the reference listed drug, generic
FSH products based on Pergonal ... must have the same
primary structure, i.e., the same protein backbone and amino
acid sequence as Pergonal (assured by using the same natural
source material), the same potency, and the same degree of
batch-to-batch uniformity." J.A. 483. The agency thus en-
deavored to guarantee the greatest degree of "sameness"
possible for this kind of product, by ensuring an identical
chemical structure where possible (in the primary structure),
while reducing natural batch-to-batch variance (in the carbo-
hydrate side chains) to the same degree as that found in the
pioneer drug. To accomplish the latter, the FDA observed
that Serono controls the batch-to-batch uniformity of Pergo-
nal by using USP rat potency tests, and that Ferring does the
same for Repronex. Id. at 483-84. The agency concluded
that "it would be unreasonable to hold the generic menotro-
pins product to a higher standard of uniformity than the
standard used for Pergonal." Id. at 484 n.17.
Finally, Dr. Woodcock noted that there also were isoform
variations between Pergonal and another approved menotro-
pins product, Humegon, and that clinical trials and published
literature on those two drugs "demonstrated no differences in
safety and efficacy." Id. at 483. Those studies, the FDA
found, indicate that "any currently observed differences in
FSH isoforms do not have clinical significance." Id. at 484.
In light of the standards it set, and the evidence of clinical
equivalence, Dr. Woodcock concluded that "the active ingredi-
ents, FSH and LH, of the approved menotropins products are
the same for purposes of 21 U.S.C. s 355(j)." Id. (emphasis
added).
The FDA's determination of what is required to establish
"sameness" for purposes of the Act rests on the "agency's
evaluations of scientific data within its area of expertise," and
hence is entitled to a "high level of deference" from this court.
A.L. Pharma, Inc. v. Shalala, 62 F.3d 1484, 1490 (D.C. Cir.
1995); see Schering Corp., 51 F.3d at 399-400. The district
court appeared to grant the FDA's determination less than
this usual deference because internal FDA memoranda indi-
cated there was some disagreement among FDA chemists as
to whether the isoform variation rendered the active ingredi-
ents different. See Serono Lab., 974 F. Supp. at 33 & n.6.2
But Chevron deference is owed to the decisionmaker autho-
rized to speak on behalf of the agency, not to each individual
agency employee. See Michigan Citizens For An Indep.
Press v. Thornburgh, 868 F.2d 1285, 1293 (D.C. Cir. 1989)
(giving Chevron deference to Attorney General's statutory
interpretation over contrary view of Antitrust Division, be-
__________
2 The district court also read the minutes of a 1993 meeting
between FDA staff and Lederle (the original ANDA applicant) to
indicate that the staff "implicitly" rejected the use of "the USP
bioassay for menotropins" as a method for evaluating "pharmaceuti-
cal equivalence"--because the staff required Lederle to do addition-
al chemical testing. Serono Lab., 974 F. Supp. at 33 n.7. Whether
or not this was the implication of the staff's actions, the views of
FDA staff do not bind the agency's final decisionmaker. See 21
C.F.R. s 10.65(a) (action at meetings with FDA staff does not
constitute final administrative action).
cause Congress "place[d] responsibility for reconciling the
conflicting policies and values called for in this type of case
[not] upon the Antitrust Division, but rather on the Attorney
General"); cf. San Luis Obispo Mothers For Peace v. United
States Nuclear Regulatory Comm'n, 789 F.2d 26, 33 (D.C.
Cir. 1986) (en banc) (holding that the "position of an agency's
staff, taken before the agency itself decided the point, does
not invalidate the agency's subsequent application and inter-
pretation of its own regulation"); Homemakers N. Shore, Inc.
v. Bowen, 832 F.2d 408, 413 (7th Cir. 1987) (" 'The Secretary's
position' is the position of the Department as an entity, and
the fact that people in the chain of command have expressed
divergent views does not diminish the effect of the agency's
resolution of those disputes."). Indeed, were we to hold
otherwise, we would effectively empower any individual em-
ployee not just to veto the views of the agency head, but to
preclude any deference to the agency at all, since we would
have no basis for deciding to whose view we should defer.
Dr. Woodcock was the authorized decisionmaker for the
agency on this matter, see 21 C.F.R. s 5.31(a)(2)(i) (Director
of Center for Drug Evaluation and Research authorized to
grant or deny citizen petition), and hers is the view to which
the courts owe deference.
Of course, differing views among an agency's staff may
indicate that there is more than one reasonable way to read a
statute. And there may well be more than one reasonable
way to read this one. But under Chevron, courts are bound
to uphold an agency interpretation as long as it is reason-
able--regardless whether there may be other reasonable, or
even more reasonable, views. Here, the FDA interpreted
"same as," in the context of menotropins products, to require:
clinical equivalence to the pioneer, chemical identity to the
extent possible, and limitations on inherent isoform variation
to the same extent as in the pioneer. This interpretation is a
reasonable, and hence permissible, reading of the statutory
term. Cf. Bristol-Myers Squibb Co. v. Shalala, 91 F.3d 1493,
1499-1500 (D.C. Cir. 1996) (upholding FDA determination
that statutory provision requiring that labeling of generic be
the "same as" labeling of pioneer, permitted FDA to approve
a generic even though its label would not include all of the
indications on the label of the pioneer). It is also a reason-
able interpretation of the word "identical" in the agency's own
regulation.3
We conclude that the district court erred as a matter of law
in ruling that the FDA's interpretation of the statute and
regulations was impermissible. As that ruling was the princi-
pal basis for the court's conclusion that Serono was likely to
succeed on the merits of its claim that the active ingredient
FSH in Repronex was not the "same as" that in Pergonal, the
court erred in that conclusion as well.
__________
3 The Federal Register notice accompanying 21 C.F.R.
s 314.92(a)(1), which defines the term "same as" to mean "identi-
cal," supports the FDA's view that the regulation does not require
complete chemical identity regardless of the kind of drug at issue.
The notice indicates the FDA decided against adopting a proposal
that would have required "applicants to demonstrate that their
active ingredients 'exhibit the same physical and chemical charac-
teristics[;] that no additional residues or impurities can result from
the different manufacture or synthesis process; and that the stereo-
chemistry characteristics and solid state forms of the drug have not
been altered.' " 57 Fed. Reg. 17,950, 17,958-59 (1992). Instead,
the notice indicates the FDA adopted a more flexible approach:
FDA will consider an active ingredient to be the same as that
of the reference listed drug if it meets the same standards for
identity. In most cases, these standards are described in the
U.S. Pharmacopeia [USP]. However, in some cases, FDA
may prescribe additional standards that are material to the
ingredient's sameness. For example, for some drug products,
standards for crystalline structure or stereoisomeric mixture
may be required. Should questions arise, an applicant should
contact the Office of Generic Drugs to determine what informa-
tion would be necessary to demonstrate that its active ingredi-
ent is the same as that in the reference listed drug.
Id. at 17,959 (emphasis added). As discussed in the text above, the
FDA followed this approach here, "rely[ing] on the USP test for its
determination of the sameness of the active ingredients" in Repro-
nex and Pergonal. J.A. 483 (Woodcock Letter).
B
Lactose is an inactive ingredient in both Repronex and
Pergonal. With regard to inactive ingredients, the Act di-
rects the FDA to approve an ANDA for a generic drug unless
the agency finds the inactive ingredients are "unsafe for use"
or the composition of the drug is unsafe "because of the type
or quantity" of the inactive ingredients. 21 U.S.C.
s 355(j)(3)(H). Although the statute itself contains no other
limitation, an FDA regulation that became effective in 1992
provides that the agency will not grant an ANDA for a
generic drug intended for parenteral (injectable) use, "unless
it contains the same inactive ingredients ... in the same
concentration as the listed drug...." 21 C.F.R.
s 314.127(a)(8)(ii)(B). Repronex is intended for parenteral
use and, although there is no dispute that the lactose in
Repronex and Pergonal is the same, it is conceded that the
concentration of lactose in the two drugs is different. Repro-
nex contains twice as many milligrams of lactose per vial as
Pergonal. See J.A. 457.
Deputy Director Johnston addressed this issue in his Janu-
ary 1997 memorandum, determining that because the ANDA
for Repronex was filed in 1990, the regulations that were in
effect in that year rather than those that went into effect in
1992 should apply. He explained his determination as follows:
OGD [the Office of Generic Drugs] has generally used
the filing and approval criteria in effect at the time of
submission as the basis for approval of applications. At
the time that [the Repronex] applications were submitted
in June 1990, the regulations implementing the Waxman-
Hatch amendments were not in effect. The regulations
in effect at that time did not require that
parenteral products contain the same inactive ingredients
at the same concentration. [See, e.g., 21 C.F.R.
ss 314.125(b)(2), (3), (4) (1990); 21 C.F.R. s 314.2
(1984).] Moreover, OGD did not have a specific policy
that addressed limitations on inactive ingredients in par-
enteral products. Thus, with regard to inactive ingredi-
ents, the generic menotropins application was approvable
under the regulations in effect at the time the application
was submitted.
J.A. 457.
The district court rejected Johnston's determination. The
court did not dispute the FDA's representation that its policy
has been to apply the regulations in effect at the time of the
submission of the ANDA. See, e.g., 48 Fed. Reg. 2751, 2753
(1983) (in pre-Hatch-Waxman period, applying regulations
only to ANDAs submitted after the regulations' effective
date). Instead, it pronounced itself "dumbfound[ed]" by the
contention that a new drug could "come to market on a more
lenient basis than required by existing law." Serono Lab.,
974 F. Supp. at 34-35. "While the court understands Grand-
father clauses," it said, "if one does exist in this case, they
have no place where the public safety is involved." Id. at 35.
We do not find the FDA's policy so dumbfounding. First,
the agency's decision not to apply the 1992 "same concentra-
tion" rule did not free the agency to disregard safety consid-
erations. The statute's bottom line--that the agency must be
satisfied that the lactose in the generic is not unsafe--still
holds. Second, as long as the agency continues to ensure an
ingredient's safety on a case-by-case basis, the decision not to
retroactively apply a per se rule regarding concentration is
not irrational. The application process for new drugs can be
a long one--even the "abbreviated" ANDA process utilized
here took more than six years for an agency decision. If
every pending application had to be revised each time the
FDA changed its regulations, the process would become much
more lengthy--even Sisyphean if the rules of the game
changed each time the application neared the finish line.
Indeed, if complete retroactivity were required, the unintend-
ed consequence might well be to force the agency to limit its
revision of regulations, in order to prevent the process from
becoming unworkable.
More important, however we or the district court may
appraise the reasonableness of grandfather clauses in drug
regulation, Congress itself plainly contemplated that the FDA
would follow a grandfather policy. Although the Hatch-
Waxman Amendments authorized the FDA to promulgate
regulations to implement its new ANDA provisions, and one
such regulation was the 1992 "same concentration" rule, the
Amendments also expressly stated that ANDAs "may be
submitted in accordance with" the FDA's existing regulations
until the new regulations "take effect." Pub. L. No. 98-417,
s 105(b), 98 Stat. 1585, 1597 (codified at 21 U.S.C. s 355
note). As the FDA rightly points out, for this provision to
have any meaning, the FDA must also be permitted to review
applications under the regulations in effect at the time of the
submission. Accordingly, the district court erred as a matter
of law in concluding that Serono was likely to succeed on the
merits because the FDA had failed to apply its 1992 regula-
tion to the Repronex ANDA.4
The district court further stated that even if the FDA did
not have to apply the 1992 regulation, the court nonetheless
was "uncertain" whether the bottom-line requirement that
the lactose in Repronex not be "unsafe" was satisfied. Sero-
no Lab., 974 F. Supp. at 35. The court's uncertainty derived,
it said, from the following sentence in the Johnston Memoran-
dum: " '[T]he difference in the amount of lactose present in
[Repronex] does not raise serious questions of safety.' " Id.
(citing J.A. 458 (Johnston Memorandum)) (emphasis added by
district court). The agency's use of the word "serious," the
district court suggested, indicated too much "tentativeness" to
give the court comfort. Id.
The FDA contends, and we agree, that the district court
misread the memorandum. Presumably the court thought
the use of the adjective "serious" indicated the FDA still
might harbor questions, even if not serious ones. In context,
however, it is apparent that Deputy Director Johnston did not
use "serious" to suggest that unresolved questions remained.
Rather, he used serious as a synonym for "reasonable." This
__________
4 Serono's citations to cases regarding an agency's duty to comply
with its own regulations are inapposite. The FDA did not "fail to
comply" with an applicable regulation. Rather, it found that under
its existing policy, the 1992 regulation was inapplicable to the
earlier-filed ANDA for Repronex.
usage is made clear by the language Johnston used to sum-
marize his analysis of the safety issue: "[T]here is no reason-
able basis to conclude that the lactose would have been or is a
safety concern." J.A. 458 (emphasis added); see also id. at
481 n.12 (Woodcock Letter) ("[T]he lactose concentration
variation between Ferring's product and Serono's product
does not pose safety concerns.").5
Nor does anything in the Johnston Memorandum suggest
the FDA reached its conclusion cavalierly. Deputy Director
Johnston made clear that although the Repronex ANDA was
not subject to the "same concentration" regulation, the agen-
cy had "assured that safety was not a problem." Id. at 458.
There was "no reasonable basis" for a safety concern arising
from the different concentrations of lactose, he said, for four
reasons. First, "lactose is a sugar obtained from milk" which
is commonly used as an inactive ingredient in many parenter-
al drug products, and which "has been found to be generally
recognized as safe (GRAS) in preclinical or animal studies by
the Center for Drug Evaluation and Research (CDER)." Id.
Second, "lactose has been used safely in amounts that far
exceed[ ] the amount present in [Repronex]." Id. Third,
"every lot of [Repronex] is checked for efficacy by [a recog-
nized] method." Id. And finally, "three safety studies were
performed on the product that showed no demonstrated
potential for an increase in the incidence and severity of
cardiovascular incidents or hypersensitivity and anaphylactic
reactions." Id.
Nothing in the Johnston Memorandum, then, suggests the
agency was left with any residual safety concerns. To the
contrary, Deputy Director Johnston concluded: "Thus, FDA
determines that there is not [a] safety concern (from the
__________
5 Johnston's use of the word "serious" appears to have been
nothing more than a parroting of the language in the FDA regula-
tion cited by Serono in its citizen petition. See J.A. 465 ("An
inactive ingredient ... will be considered unsafe if there is a
reasonable basis to conclude that the ingredient ... raises serious
questions of safety.") (emphasis added) (citing 21 C.F.R.
s 314.127(a)(8)(ii)(A)).
inactive ingredients or the impurity profile), efficacy concern,
or bioequivalence problem that would preclude approval of
the generic drug product." Id. To the extent the district
court concluded otherwise regarding the firmness of the
FDA's view, that conclusion was clearly erroneous.
Serono contends that however certain the agency may have
been about the safety of the lactose in Repronex, the district
court's decision was still justified because the three safety
studies referred to in the Johnston Memorandum were ani-
mal studies. According to Serono, the Hatch-Waxman
Amendments prohibit the use of such studies to analyze
safety. The district court did not explicitly rely on this
argument, but Serono presses it as an alternate ground for
affirmance. Serono Br. at 33-35.
The only provision of the Act to which Serono points for
support of its no-animal-studies proposition is one that states
the FDA "may not require that an abbreviated application
contain information in addition to that required by clauses (i)
through (viii)" of 21 U.S.C. s 355(j)(2)(A). Id. Because
nothing in those clauses mentions animal studies, Serono
contends they are barred. This provision, however, does not
bear the weight Serono applies.
First, the indicated clauses do not suggest that animal
studies are in any way disfavored. The clauses simply de-
scribe what the "information" in an application must "show."
They do not specify the kinds of studies that can or cannot be
used to satisfy the requirement. See, e.g., id.
s 355(j)(2)(A)(ii)(II) ("An abbreviated application for a new
drug shall contain ... information to show that the active
ingredients of the new drug are the same as those of the
listed drug.").
Moreover, the most the provision cited by Serono does is
bar the FDA from requiring an applicant to submit more
information than required by the statute. It does not bar an
applicant from voluntarily submitting additional informa-
tion--including animal studies--as part of its ANDA. Nor
does it bar the FDA from relying on animal studies to make
its findings. To the contrary, the statute expressly provides
that the FDA may make safety determinations on the basis of
information submitted in the ANDA "or any other informa-
tion available to the Secretary." Id. s 355(j)(3)(H). Accord-
ingly, we reject Serono's contention that the Act prohibits
reliance on animal studies to confirm the safety of Repronex's
inactive ingredients. See Schering Corp., 51 F.3d at 399
(holding that FDA's "judgments as to what is required to
ascertain the safety and efficacy of drugs fall squarely within
the ambit of the FDA's expertise and merit deference from
us").6
In sum, we conclude the district court erred in finding that
Serono was likely to succeed on the merits regarding its
lactose claim, because that finding was based on (1) the
legally erroneous conclusion that the FDA was bound to
apply its 1992 "same concentration" regulation to Repronex's
1990 ANDA, and (2) the clearly erroneous factual finding that
the agency was "tentative" in its views regarding Repronex's
safety. Serono's alternative rationale, that the FDA unlaw-
fully employed animal studies in this case, also fails.
C
Finally, Serono argues that the conceded differences in the
UUPs in Repronex and Pergonal render the former unfit for
ANDA approval. Serono regards the UUPs as inactive in-
gredients, and again cites the FDA's 1992 regulation, which
__________
6 We also reject Serono's contention that FDA policy bars the
agency's use of animal studies in the manner in which they were
used here. Serono relies on a Federal Register notice stating that
an ANDA is not an appropriate vehicle for approval of a drug if
animal or clinical studies are "necessary to show that the drug is
safe or effective." 57 Fed. Reg. at 17,958. The notice explains,
however, that such studies are permitted if they constitute "limited
confirmatory testing," i.e., "simple studies [that are] intended to
rule out unlikely problems" and that are not "necessary" to demon-
strate overall safety. Id. The FDA's determination that the
animal studies at issue here fall within that category is supported
by the fact that the studies were only one of four grounds upon
which the agency relied for its conclusion that the lactose in
Repronex is safe. See J.A. 458.
requires that an ANDA not be approved unless the generic
drug "contains the same inactive ingredients ... in the same
concentration as the listed drug...." 21 C.F.R.
s 314.127(a)(8)(ii)(B). The district court relied heavily on
what it characterized as the FDA's "efforts to skirt" this
regulation in concluding that Serono was likely to succeed on
the merits of this claim. See Serono Lab., 974 F. Supp. at 34.
As we have already held, however, the FDA appropriately
declined to apply its 1992 regulation to Ferring's 1990 ANDA,
leaving only the statutory (and similar 1990 regulatory) re-
quirement that available information not show the generic
drug's inactive ingredients are "unsafe." 21 U.S.C.
s 355(j)(3)(H). The Woodcock Letter adequately addressed
that requirement. Dr. Woodcock noted that Ferring "per-
formed three confirmatory safety studies to rule out the
unlikely possibility [asserted in Serono's citizen petition] that
the differences in impurity profiles between the Ferring and
Serono products might affect the safety of the generic drug
product." J.A. 480. Although the studies involved animals,
we have held above that the statute does not bar FDA from
relying on such studies for this purpose. Moreover, Wood-
cock further determined that the results of the animal studies
were "consistent with human clinical studies" submitted by
another menotropins manufacturer, Organon, in support of its
NDA for another menotropins product, Humegon, which "like
Ferring's product, contains urinary proteins that may be
different from Pergonal." Id. at 481. After reviewing those
studies, Woodcock concluded that "Ferring has adequately
demonstrated that the potential difference (from Pergonal) in
urinary proteins is not a safety concern." Id. Both this
court and the district court are bound to show deference to
the agency's fact-finding in this area of its technical expertise.
See, e.g., Schering Corp., 51 F.3d at 399.
Serono interposes one final argument. It contends we
should ignore the Woodcock Letter because it was a post hoc
rationalization of the agency's action. Although the letter
was the agency's response to Serono's citizen petition, Serono
labels it post hoc because it was issued after Serono had
already moved for injunctive relief in the district court. In
this case, however, timing isn't everything. Dr. Woodcock's
letter represents the considered views of the agency-
decisionmaker herself, announced at the usual point in the
agency's decision-making process (the end), rather than the
views of litigation counsel trying to come up with an explana-
tion after the fact. See Auer, 117 S. Ct. at 912 ("There is
simply no reason to suspect that the interpretation does not
reflect the agency's fair and considered judgment on the
matter in question."). The fact that Serono filed for prelimi-
nary injunctive relief before the agency ruled on its petition
does not change the analysis. Cf. Local 814, Int'l Bhd. of
Teamsters v. NLRB, 546 F.2d 989, 992 (D.C. Cir. 1976)
(" 'The post hoc rationalization' rule is not a time barrier
which freezes an agency's exercise of its judgment after an
initial decision has been made and bars it from further
articulation of its reasoning. It is a rule directed at reviewing
courts which forbids judges to uphold agency action on the
basis of rationales offered by anyone other than the proper
decisionmakers.").
In sum, the district court's conclusion that Serono was
likely to succeed on the merits of this claim was substantially
based on its determination that the law compelled the FDA to
apply its 1992 regulation requiring that the inactive ingredi-
ents in a generic drug be the same as those in the pioneer.
That determination was legally erroneous. On the other
hand, the record indicates that the FDA's treatment of the
UUPs is likely to satisfy the statutory safety requirement for
inactive ingredients. See J.A. 480-81 (evaluating safety of
UUPs as if they were inactive ingredients). Accordingly, we
need not consider the FDA's alternative argument that the
UUPs are not "inactive ingredients" at all, but rather are
merely "impurities" not subject to that requirement. See id.
at 479-80.
IV
In this case, our conclusion that Serono is not likely to
succeed on the merits effectively decides the preliminary
injunction issue. Here, the other preliminary injunction fac-
tors--injury to Serono, injury to Ferring, and the public
interest--either are a wash or are inextricably linked to the
merits.
Serono contends that it will be irreparably injured if the
FDA is not enjoined from approving Repronex, because it will
suffer an unrecoverable loss of sales to Ferring. But even if
such a loss does constitute irreparable injury, as the district
court found and defendants dispute, see Serono Lab., 974
F. Supp. at 35, that injury must be weighed against the next
factor--the extent to which an injunction will substantially
injure the other party, Ferring. And that balance of harms
results roughly in a draw. Whatever sales Serono will lose to
Ferring in the absence of an injunction, Ferring will lose to
Serono in the presence of one. See Serono Lab. v. Shalala,
Civ. No. 97-1227 (D.D.C. Aug. 19, 1997) (J.A. 622) (district
court order denying Ferring motion for stay pending appeal,
because "while Ferring may be harmed by the granting of the
preliminary injunction, Plaintiff would be equally harmed if
the injunction were to be stayed"). As a consequence, even
Serono concedes that the court should "ignore[ ] the injury to
both companies when balancing the harms since the lost
revenues at issue are offsetting." Serono Br. at 39-40;7 see
Delaware & Hudson Ry. Co. v. United Transp. Union, 450
F.2d 603, 620 (D.C. Cir. 1971) ("It often happens that ... one
party or the other will be injured whichever course is taken.
A sound disposition ... must [then] depend on a reflective
and attentive appraisal as to the outcome on the merits.").
The final preliminary injunction factor, the public interest,
also offers Serono no support because it is inextricably linked
with the merits of the case. If, as we have held, Serono is not
likely to establish that Ferring's ANDA was wrongly ap-
proved, then public interest considerations weigh against an
injunction. The purpose of the Hatch-Waxman Amendments
was, after all, "to increase competition in the drug industry by
facilitating the approval of generic copies of drugs." Mead
__________
7 Because Ferring plans to sell Repronex for less than Serono
sells Pergonal, whether the lost revenues are exactly offsetting
depends upon the elasticity of demand.
Johnson, 838 F.2d at 1333. Congress expected that competi-
tion "to make available more low cost generic drugs." H.R.
Rep. No. 98-857, pt. 1, at 14 (1984), reprinted in 1984
U.S.C.C.A.N. 2647, 2647. Congress' purpose is directly impli-
cated here, the FDA argues, because Ferring has priced
Repronex to sell at 40% below the price of Pergonal, and
because there has been a shortage of this type of fertility
drug. FDA Br. at 46 (citing J.A. 123, 133, 420, 458). As
Deputy Director Johnston put it, "the availability of a generic
menotropins injection ... will enable some patients to afford
the drug product that previously could not." J.A. 458.8
Of course if the ANDA should not have been granted in
this case, because the statute's standards--particularly its
safety standards--were not met, then the public interest
balance plainly would weigh in favor of an injunction. But on
the current record it appears likely that the ANDA was
properly granted, and the FDA has assured this court, in the
strongest possible terms, that there are no safety concerns.
See FDA Br. at 46; Oral Arg. Tr. at 34. Neither we, nor the
district judge, are scientists independently capable of assess-
ing the validity of the agency's determination--beyond hold-
ing it to the standards of rationality required by the Adminis-
trative Procedure Act, 5 U.S.C. s 706(2)(A). See Troy Corp.
v. Browner, 120 F.2d 277, 283 (D.C. Cir. 1997); Schering
Corp., 51 F.3d at 399. Indeed, not even Serono argues that
the evidence shows Repronex represents a safety concern.
At oral argument, Serono's counsel, choosing his words care-
fully, would say no more than that "we don't know whether
there are unresolved safety issues." Oral Arg. Tr. at 22.
Such agnosticism is too insubstantial a basis for us to rescind
the decision of the expert agency entrusted by Congress with
the authority to assess the safety of drugs.
__________
8 For this reason, one of the public interest considerations relied
upon by the district court--that "the reproductive potential of each
woman declines with age ... and any lost cycle (usually of a
month's duration) will contribute to the loss of reproductive poten-
tial"--cuts against rather than in favor of an injunction. Serono
Lab., 974 F. Supp. at 36.
V
For the foregoing reasons, we vacate the preliminary in-
junction entered by the district court and remand the case for
further proceedings consistent with this opinion. Our opinion
does not foreclose the possibility that at a trial on the merits,
and upon a fuller record, Serono may be able to establish that
there are grounds for overturning the grant of Repronex's
ANDA. We hold only that upon the current record, Serono
has failed to establish that it meets the criteria for the grant
of a preliminary injunction.